Chemical Modifications of Graphene for Biotechnology Applications

Home , Dispersion staining, Fluorescence microscope, Raman microscope, Sarfus

Chemical modifications of graphene

for biotechnology applications

A thesis submitted to the University of Manchester for the degree

Doctor of Philosophy

in the Faculty of Science and Engineering

2016

Andrea Francesco Verre

Schools of Materials

Index Figure captions………………………………………………………………………… 6

List of Abbreviations…...……………………………………………………………... 11

Abstract.………………………………………………………………………………...14

Declaration……………………………………………………………………………...15

Acknowledgments……………………………………………………………………...17

Chapter 1: Introduction… …………………………………………………………….. 20

1.1 Carbon nanomaterials………………………………….………………………….. 20 1.2 Graphene and related nanomaterials production………………………………….. 22 1.3 Characterization of Graphene and graphene related nanomaterials by microscopy…………………………………………………………..…………… .26 1.4 Characterization of Graphene and graphene related nanomaterials by spectroscopy…………………………………………………………..………… 28 1.5 Graphene Oxide Functionalization………………………………………………... 32 1.6 Lipid-Dip Pen Functionalization of Graphene… ………………………………….33 1.7 Graphene and related nanomaterials based substrates for stem cell culture and differentiation……………………………………………..… …………………….36 1.8 Conclusions………………………………………………………………………...47 References….…………………………………………………………………………..49

Chapter 2: Methodology………………………………………………………………. 62

2.1 GO synthesis and purification…………………………………………………….. 62 2.2 Biological functionalization of GO……………………………………………….. 62 2.3 Preparation of GO-based coatings… ………………………………………………64 2.4 Substrate Characterization… ………………………………………………………65 2.4.1 Atomic Force Microscopy… …………………………………………………… 65 2.4.2 Raman Spectroscopy……………………………………………………………. 66 2.4.3 X-Ray Photoelectron and Fourier Transform Infrared Spectroscopies………….,67 2.5 Biological Characterization of the Substrates…………………………………….. 69 2.5.1 Human Adipose Stem Cells Harvesting and Differentiation into dASC…...……69 2

2.5.2 MTS Assay………………………………………………………… 70 2.5.3 Live/Dead Assay…………………………………………………… 70 2.5.4 RNA extraction and gene expression studies…………………….... 71 2.5.5 Dorsal Root Ganglia (DRG) harvesting and immunohistochemistry 73 2.6 CVD graphene array fabrication……………………………………….74 2.6.1 CVD graphene transfers on silicon dioxide substrates……………....74 2.6.2 Photolithography on CVD graphene substrates……………………...74 2.7 Dip Pen Nanolithogrpahy Patterning and Membrane Protein Insertion.78

References………………………………………………………………….81

Introduction to Chapter 3… ……………………………………………….85

Chapter 3: Biochemical functionalization of graphene oxide for directing stem cell differentiation……………………………………………………………... 86

Abstract…………………………………………………………………… 86 3.1 Introduction.….………………………………………………………...87 3.2 Materials and methods………………………………………………... 88 3.2.1 Graphene based materials synthesis, substrates preparation and characterization… ……………………………………………………………………………...88 3.2.2 Human Adipose stem cells harvesting and culture…………………..89 3.2.3 Cell Proliferation and Live/Dead assays…………………………… 89 3.2.4 Quantitative real-time polymerase chain reaction (qRT-PCR)...……90 3.2.5 Rat DRG neurons harvesting, culture and immunohistochemistry… 90 3.2.6 Statistical analysis………………………………………………….. 91 3.3 Results and Discussion……………………………………………….. 91 3.3.1 Substrates Characterization… ………………………………………91 3.3.2 Effect on IKVAV functionalization on improving neuronal attachment and neurite outgrowth…………………………………………………………………. 93 3.3.3 Proliferation of ASCs on graphene substrates… ……………………94 3.3.4 Gene Expression Studies… …………………………………………95 3.4 Conclusions… ………………………………………………………...98 References……………………………………………………………….. 99 Supporting information………………………………………………….. 103

Introduction to Chapter 4…………………………………………………107

Chapter 4: Reduced GO substrates increase gene expression of neurotrophins and filament proteins by Schwann-like differentiated adipose stem cells… …. 108

Abstract……………………………………………………………………………….109 4.1 Introduction....…………………………………………………………………….109 4.2 Experimental… …………………………………………………………………..110 4.2.1 Graphene based materials synthesis, substrates preparation and characterization… ………………………………………………………………………………………..110 4.2.2 Human Adipose stem cells harvesting and differentiation……………………..110 4.2.3 Cell Proliferation and Live/Dead assays……………………………………….110 4.2.4 Quantitative real-time polymerase chain reaction (qRT-PCR)………………...110 4.2.5 Statistical analysis……………………………………………………………...111 4.3 Results and discussion…………………………………………………………...111 4.3.1 Substrates Characterization…………………………………………………….111 4.3.2 Proliferation of dASCs on graphene substrates………………………………...112 4.3.3 Gene Expression Studies……………………………………………………….112 4.4 Conclusions………………………………………………………………………115 References…………………………………………………………………………... 116

Introduction to Chapter 5… …………………………………………………………120

Chapter 5: Selective Immobilization of Membrane Proteins carried in Nanodiscs on Functionalized Graphene Oxide… ………………………………………………….121

Abstract…………………………………………………………………………….. 121 5.1 Introduction……………………………………………………………………...122 5.2 Methodology… …………………………………………………………………123 5.2.1 Preparation and functionalization of GO flakes……………………………….123 5.2.2 XPS and AFM Characterization…………………………………………...... 123 5.3 Results and Discussion……...…………………………………………………...123 5.3.1 XPS characterization of the carboxylation of graphene oxide………………...123 5.3.2 AFM studies of the protein binding on functionalized GO……………………126 5.4 Conclusions……………………………………………………………………...128 References…………………………………………………………………………...130

Introduction to Chapter 6…………………………………………………………....133

Chapter 6: Tail-anchoring proteins insertion into phospholipid biomimetic membranes on graphene………………………………………………………………………….134

Abstract…………………………………………………………………134 6.1 Introduction…………………………………………………………135 6.2 Materials and methods……………………………………………...136 6.2.1 Graphene-Array Fabrication……………………………………...137 6.2.2 Lipid Deposition by L-DPN……………………………………...137 6.2.3 Protein Insertion and Immunostaining…………………………...137 6.3 Results and Discussion……………………………………………. 138 6.4 Conclusions………………………………………………………...141 References……………………………………………………………...143

Chapter 7: Conclusions and future works……………………………...146

Intoduction to Appendix………………………………………………..149

Appendix 1: Graphene oxide selectively targets cancer stem cells, across multiple tumor types: Implications for non-toxic cancer treatment, via “differentiation-based nano- therapy”.

Appendix 2: Molecular Dynamics Simulations of Biomimetic Phospholipid Membrane Organization from Dip Pen Nanolithography

Figure captions Chapter 1

Figure 1.1: a) Optical microscopy image of micromechanical exfoliated graphene flakes ; b) and c) AFM topography image of graphene flakes, d) SEM image of graphene field effect transistor, e) Schematic representation of graphene FET device. Image taken from [7]

Figure 1.2: (a, b) SEM image of CVD-graphene films , (c, d) CVD-graphene films transferred on Si/SiO2 and glass substrates respectively. Image taken from reference [8]

Figure 1.3: (a-b) SEM of sieded graphite and from the sediment after centrifugation respectively; c-g) bright field TEM image of exfoliated graphene obtained after the liquid exfoliation process.; h) Histogram of visual observation of flakes as function of number of layers. Image taken from [17]

Figure 1.4: Structural models of GO proposed by Lerf-Klinowski. Image taken from [25]

Figure 1.5: AFM topography image of reduced graphene oxide flakes. On the right hand side height profile measurement of the flakes on the spot highlighted from the black lines. Image taken from [30]

Figure 1.6: Optical microscopy image of micromechanically exfoliated graphene flakes on arbitrary substrates. Image taken from [31]

Figure 1.7: A) GO and rGO films; B) GO and rGO dispersions; C) Optical microscopy image of GO flakes; D)Optical microsocpy image of rGO flakes. Image taken from [34]

Figure 1.8: AFM topography image of graphene flakes with height profile measurement. Image taken from [31]

Figure 1.9: Raman Spectrum of mechanically exfoliated graphene flakes (a), Variation of 2D peak as function of different number of layers in micromechanical exfoliated graphene flake. Image taken from [33]

Figure 1.10: Raman spectrum of functionalized graphene sheet (FGS), graphene oxide (GO) and graphite . Taken from reference [36]

Figure 1.11 : XPS C1s deconvolution in 6 components. XPS C1s spectrum of GO (a), Graphite oxide (b), diamond (c), HOPG (d), sodium terephthalate (e), sodium dodecanedioate (f), hydroqunone (g). Taken from [39].

Figure 1.12: Functionalization of GO by exploiting carboxylic functional groups. Image taken from [25]

Figure 1.13: Schematics of DPN patterning. Images taken from [72]

Figure 1.14: a) Relative expression of pluripotent stem cells markers; b) Fluorescence microscopy image after immunohistochemical analysis of Sox2 and Oct4. Image taken from [89]

Figure 1.15: Immunostaining for desmin and osteocalcin (OCN) in green and nuclear marker DAPI In blue to evaluate the osteogenic differentiation of hMSCs grown on PET ,

PDMS and Si/SiO2 substrates with and without CVD graphene. Image taken from [101]

Figure 1.16: Evaluation of osteogenic differentiation on PDMS, CVD graphene (G) and graphene oxide (GO) substrates by Alzarin Red staining. Image taken from [103].

Figure 1.17: Immunohistochemistry on hippocampal neurons grown on functionalized graphene oxide substrates. Image taken from [107]

Figure 1.18: Neurite lenght outgrowth, number of branches, number of neurites and cell body area measurements of hippocampal neurons on GO, GO-MPC, and GO-DMAEMA substrates. Image taken from [108]

Chapter 2

Figure 2.1: Schematics of Atomic Force Microscopy. Image taken from [8]. To the right hand side, an AFM topography image representing a glass substrate coated with GO flakes prepared as described previously.

Figure 2.2: Schematics of Raman microscopy [10]

Figure 2.3: Fluorescent microscope image of hASC cells grown on GO substrates after Live/Dead assay. Live cells are stained in green while dead cells are stained in red.

Figure 2.4: Fluorescence microscopy image of DRG neurons on GO-IKVAV substrates after immunohistochemistry analysis with anti -tubulin 3 antibody.

Figure 2.5: Optical Microscopy Image of transferred CVD-graphene on top of Silicon oxide wafer

Figure 2.6: Optical microscopy image of CVD graphene 20 micron x 20 micron squares array after photolithography

Figure 2.7: Optical microscopy image of CVD graphene 20 micron x 20 micron square array after oxygen plasma etching.

Figure 2.8: Optical microscopy image of CVD graphene 20 micron x 20 micron square array after the lift off of the resist polymers

Figure 2.9: Optical microscopy Image of CVD graphene square array in different sizes after the lift off of the resist polymers

Figure 2.10: Optical microscopy image of the CVD graphene square in air after lipid patterning by DPN

Chapter 3

Figure 3.1 (a, b, c) AFM characterization of the GO, rGO and GO IKVAV coverslips respectively. (d, e, f) Optical microscopy characterization of GO, rGO and GO IKVAV coverslips respectively. (g, h, i) XPS characterization of GO, rGO and GO IKVAV coverslips characterization.

Figure 3.2 (a) Immunohistochemical analysis of the attachment of DRGs on the different coverslips. (b) Neurite length measurement after immunohistochemical analysis on the different coverslips.

Figure 3.3 (a) MTS assay proliferation results on the different substrates (b) Live Dead quantification expressed as % of Live ASC Cells grown on the different substrates.

Figure 3.4 (a) Schematic representation of the differentiation pathways of ASCs studied here, and the marker genes corresponding to each lineage. (b - i) Quantitative Real-Time PCR studies of ASCs grown on the different substrates.

Figure S1: (a,b,c) Raman spectra of GO, rGO and GO-IKVAV substrates. (d,e,f) XPS wide scan spectra of GO, rGO and GO-IKVAV substrates. (g,h) XPS N1s spectra of rGO and GO-IKVAV respectively.

Figure S2: (a, b)AFM measurement of GO after exfoliation and height profile measurement. (c, d) AFM measurement of GO thin film and height profile measurement. (e, f) FT-IR measurement of GO, IKVAV and GO IKVAV.

Chapter 4

Figure 4.1: (a, b) AFM topography image of GO, rGO and substrates respectively; (c, d) Optical images of GO and rGO substrates respectively; (e, f) Raman spectra of GO and rGO substrates respectively. (g, h) XPS C1s spectra of GO and rGO coated substrates respectively

Figure 4.2 a) MTS assay of dASCs grown on rGO coverslips (purple line), GO coverslips (red line), and bare glass coverslips (blue line). b) Live/Dead assay results of dASCs grown on the different coverslips.

Figure 4.3: Gene expression of dASCs on the different substrates: a) GDNF; b) BDNF; c) NGF; d) Nestin; e) Vimentin; f) TrkC ; g) Ret; h) TrkB

Chapter 5

Figure 5.1 (a,) Schematic of the functionalization of GO performed in this study. b) Illustration of the protein immobilization strategy performed in this study.

Figure 5.2 (a) XPS characterization of GO before carboxylation, (b) XPS characterization of GO reacted with 0.5 M of chloroacetic acid and 4M of NaOH. (c) XPS characterization of GO reacted with 1M of chloroacetic acid and 4M of NaOH. (d) XPS characterization of GO reacted with 3M of chloroacetic acid and 4M of NaOH

Figure 5.3 (a) AFM characterization of GO as control reference, (b) AFM and height profile measurement of GO flakes after His-tagged Orco immobilization (c) AFM and height profile measurement of GO flakes after imidazole elution of the His-tagged Orco from the surface of the GO flakes. (A and B in the AFM height graphs stands for Height and Count percentage respectively.

Chapter 6

Figure 6.1. Illustration of the graphene-based array used for the biomimetic phospholipid membrane by DPN and the tail-anchor protein insertion performed in this study. Antibody directed towards the Cytochrome B5 –Synaptobrevin-2 protein is needed to study the protein binding by fluorescence microscopy.

Figure 6.2 (a) Optical characterization of the graphene-based microarray. (b) Average Raman spectrum acquired inside the graphene squares. (c) AFM topography image of one graphene square after the fabrication process. (d) Height profile measurement of graphene- based array after the fabrication process.

Figure 6.3 (a) Optical characterization of the graphene-based microarray after lipid patterning (b) Fluorescence microscopy of the lipid patterning in air. (c) Fluorescence microscopy image of lipid patterning in liquid

Figure 6.4 (a, b) Fluorescence microscopy of the lipid-protein interaction (lipid-rhodamine fluorescence in red- alexa 488 fluorescence from anti-synaptobrevin 2 after immunohistochemical assays.

List of abbreviations

AFM: Atomic Force Microscopy

ALP: Alkaline Phosphatase

ASC: Adipose Stem Cells dASC: Schwann-Like Differentiated Adipose Stem Cells

ATR: Attenuated Total Reflection

BDNF: Brain Derived Neurotrophic Factor

BM-MSC: Bone-Marrow Mesenchymal Stem Cells

BSA: Bovin Serum Albumin

CSI: Cell Shape Index

CVD: Chemical Vapour Deposition

DCC: N,N'-Dicyclohexylcarbodiimide

DDMAT: S-1-Dodecyl-S΄-(A,A΄-Dimethyl-A΄-Acetic Acid) Trithiocarbonate

DMA: N,N-Dimethylacetamide

DMAEMA: Dimethylamino Ethyl Methacrylate

DMEM: Dulbecco`S Modified Eagle Medium

DMEU: 1,3-Dimethyl-2-Imidazolidinone

DOPA: 1,2-Dioleoyl-Sn-Glycero-3-Phosphate

DOPC: 1,2-Dioleoyl-Sn-Glycero-3-Phosphocholine

DOPE: 1,2-Dioleoyl-Sn-Glycero-3-Phosphoethanolamine

DPN: Dip Pen Nanolithography

DRG: Dorsal Root Ganglia

EDC: 1-Ethyl-3-(3-Dimethylaminopropyl)Carbodiimide

EDTA: Ethylenediaminetetraacetic acid hESC: Embryonic Stem Cells

FACS: Fluorescence-Activated Cell Sorting 11

FET: Field Effect Transistors

FGF-2: Fibroblast growth factor

FTIR: Fourier Transformed Infrared

FWHM: Full Width at Half Maximum

GFP: Green Fluorescent Protein

GDNF: Glial Derived Neurotrophic Factor

GO: Graphene Oxide

GO IKVAV: Ikvav Functionalized Graphene Oxide

HATU: (7-Aza-1h-Benzotriazole-1-Yl)-1,1,3,3-Tetramethyluronium Hexafluorophosphate

His-TAG: Histidine-Tagged

HIV-1: Human Immunodeficiency virus type 1

HOPG: Highly Oriented Pyrolitic Graphite iPSC: Induced Pluripotent Stem Cells

MEM: Minimal Eagle Medium Alpha Modifications

MES: 2-(N-morpholino)ethanesulfonic acid

MHA: Mercaptohexadecanoic Acid

MPC: 2-methacryloyloxyethylphosphorylcholine

MTS: 3-(4,5-Dimethylthiazol-2-Yl)-5-(3-Carboxymethoxyphenyl)-2-(4-Sulfophenyl)-2h- Tetrazolium)

MTT: 3-(4,5-Dimethylthiazol-2-Yl)-2,5-Diphenyltetrazolium Bromide

NGF: Nerve Growth Factor

NMP: N-Methyl-2-pyrrolidone

NSC: Neural Stem Cells

NTA: Nitrilotriacetic Acid

NTA-NH2: Nα,Nα-Bis(Carboxymethyl)-L-Lysine Hydrate

ODT: 1-Octadecanethiol

PANi: polyaniline

PBS: Phosphate Buffered Saline

PCL: polycaprolactone

PDGF: Platelet-Derived Growth Factor

PDMS: Polydimethylsiloxane

PEDOT: Poly(3,4-ethylenedioxythiophene)

PEDOT/PSS: Poly(3,4-ethylenedioxythiophene) polystyrene sulfonate

PEI: Polyethyleneimine

PET: Polyethylene terephthalate

PLL: Polylysine

PMGI: Polymethylglutarimide

PMMA: Poly(methyl methacrylate)

PU: Polyurethane qPCR: Real Time Polymerase Chain Reaction

Rho-PE: 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) rGO: Reduced Graphene Oxide

SAM: Self-assembled monolayers

SOCl2: Thyonil Chloride

TCPS: Tissue Culture Polystirene

XPS: X-Ray Photoelectron Spectroscopy

Abstract

The aim of this thesis is to investigate different functionalization strategy of graphene nanomaterials for graphene-based different biotechnological applications such as graphene-directed stem cell growth and differentiation and graphene-based biosensors. Chemical functionalization of graphene is required in many biological applications; in this thesis we have focused on exploiting the carboxylic groups available on GO molecules and non-covalent functionalization of graphene.

GO has been a promising material for stem cell culture due to high specific surface area, ease of functionalization, its ability to support cell proliferation and to not cause cytotoxicity when stem cells are cultured on its substrate. The impact of biochemical functionalization on stem cell differentiation was not widely research, and many research groups worldwide have been focusing only on GO and rGO surfaces only. The approach of this thesis is to fabricate and characterize different graphene-based substrates to investigate the impact of biochemical functionalization of GO in directing adipose stem cell differentiation and to influence the gene expression pathways of Schwann-like differentiated adipose stem cells.

The fabrication of graphene based biosensors is still challenging as biological molecules need to be attached to graphene-based sensors to increase both the specificity and the selectivity of the biosensors.

In this thesis, two different chemical functionalization approaches were considered. Firstly, the covalent immobilization of membrane proteins embedded on a lipid nanodisc structure on GO was achieved. Secondly, the feasibility of using dip-pen nanolithography as a tool to locally functionalize graphene arrays with phospholipids was demonstrated. Phospholipid interface layer can act as bioactive layer which can be used for the protein insertion of tail-anchoring recombinant proteins as a new route for a non-covalent biological functionalization of graphene array.

Declaration No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

Copyright statement i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trade marks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property University IP Policy (see http://documents.manchester.ac.uk/display.aspx?DocID=24420), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses

Il ponte dell’autostrada Salerno-Reggio Calabria sorvola Via Degli Stadi mozzandola in perpendicolare. Incarna il confine tra la vita ordinaria e il nulla che ci avvolge quotidianamente. Tutte le volte che passo sotto l’autostrada la memoria mi fa rivivere ancora quel pomeriggio. Il ponte sembra voler dire che qui comincia un’altro mondo. La fontanella sulla destra introduce gli ultimi trecento metri del viale. In fondo, il San Vito!

Via Degli Stadi mi assorbe una domenica si e una no da quando ero bambino. E’ un viale lungo, lunghissimo se percorso a piedi e col fiato in gola, per arrivare piu presto al bar dietro la curva nostra. Tutte le volte che percorro questa strada, mi lascio alle spalle una parte di me, una buona metà, forse la peggiore, comunque la più presentabile. Di sicuro è il lato meno sincero che rimane a casa.

“Certo che deve essere stata una bella fregatura. Mentre dirigenti e politici facevano affari loschi e pappavano soldi, voi macinavate chilometri per il Cosenza, rischiavate la pelle per la bandiera”.

“Ma non per loro. A noi interessava la nostra storia, l’orgoglio della nostra curva”.

(Claudio Dionesalvi, B.D.D)

Acknowledgements

The accomplishment of a PhD thesis is a really tough journey where the success depends not only from the author but also from the people, which helped me in any possible way.

I would really like to thank my supervisor Dr Aravind Vijayaraghavan for accepting me in his research group and to provide all the resources to accomplish this thesis. I want to thank him for his supervision and I think he clearly put a great input in my scientific development, one of the most important skill that PhD. I am very grateful to him for giving me the opportunity to work in different collaboration projects, enlarging my knowledge. I am very grateful to Dr Alessandro Faroni and Dr Adam Reid for the great experience during the experimental work in the Blonde McIndoe Laboratories. I am very grateful to them for the technical and theoretical knowledge they have provided to me throughout these years. I am very grateful to Dr.Michael Hirtz from Karlsruhe Institute of Technology for welcoming me in his research group during the perforiming of the Dip- Pen Nanolithography experiments. I am very happy to work in a great environment such as the Nanofunctional Materials Group and I would like to thanks all the current and past members of the research group: Dr Maria Iliut, Michelle Vaqueiro-Contreras, Christian Berger, Daniel Melendrez Armada, Dr Claudio Silva, Cian Bartlam, Rory Phillips, Nick Clark, Wentao Zou, Richard Fu, Steffan Llewellyn, Jacek Wychowaniec, Dr Sebastian Heeg, Dr Antonios Oikonomou, Dr Patricia Gorgojo-Alonso and Monica Alberto. It was a pleasure to work with you these years.

Living in Manchester gave me the opportunity of meeting wonderful people which became great and invaluable friends such as Nicholas Black and Manuela Gutierrez. Their friendship is one of the experience I will always treasure from the PhD days. I want to thank also Dr Zsofia Toth, Cristina Inversi, Miles Ten Brinke , Dr Christos Begkos, Dr Dimitrios G. Papageorgiou, Dr Oana Istrate, for their friendship.

I would like to thank greatly four people which represented a lot for me in the last two years: Angela Leo (aka Angelozzi) for never getting upset for my jokes and pranks and for the great time in Salento this summer, Angelozzi you are great. Stefania Bisogno (aka L`ingegnere) has become a true and great friend. I was lucky to meet her during the PhD. Sharing great time together in Manchester with Stefania was amazing and put the basis for an enduring friendship.

I would like to thank Lugi Vespucci (aka The Owl) and Francesca Maggio (aka The Miss of Patience) for the great times together watching football matches (unfortunately for Francesca) and enjoying life in Whalley Range. Their friendship means a lot for me.

I want to thank my friend from Cosenza which I always kept in touch with. Three people I more grateful with and they are Attilio Apicella, Edoardo Cozza, Piergiorgio Valentino and Davide Carino. It is difficult to keep friendships when you live in different countries and I am lucky to have great friends like them.

I want to thank a lot my friends Marco Fiorillo, Assunta Fiore, Francesca Trotta, Bela Ozvari, Ernestina De Francesco which I shared great time and I enjoyed time together in Manchester during their stay in the CRUK center in Didsbury. Their continuous support and friendship was priceless.

More importantly, I would like to dedicate the thesis to my family: my mother, my father and my sister for always accepting my choice and always be supportive. I am and will be very pleased with them for always supporting me throughout these years. Without them, it would have been definitely a more difficult journey. This thesis is dedicated also to the memory of Francesca Romana Frasconi. She and what she learnt to me will always be part of me.

Outline of the Thesis

This thesis focuses on the chemical modifications for biotechnological applications of graphene and related nanomaterials particularly the production of graphene based substrates for stem cell culture and covalent and non-covalent strategies to integrate membrane proteins on graphene and related nanomaterials flakes. In Chapter 1, it is introduced the existing literature on graphene and related nanomaterials, their synthesis and characterization, covalent and non-covalent functionalization strategies and finally the production of graphene-based substrates for stem cells growth and differentiation. The Chapter 2 focuses on the methodology followed to build the experiments conducted in this thesis. Chapters 3-6 cover the results of this thesis in the form of peer reviewed journal articles. Chapter 7 focuses on the conclusions and future work of this thesis.

Chapter 1: Introduction

1.1 Carbon nanomaterials

Graphene and related nanomaterials belong to the family of carbon nanomaterials. Carbon is a chemical element, which is tetra-valent as it has four electrons available to make chemical bonds. These four electrons are called valence electrons and can be contained into two electron orbitals, s and p. Hybridization is the phenomenon that occur when the valence electrons change their position inside the orbitals. In fact, in the sp3 configuration, one orbital s is combined with 3 orbitals p. In the sp2 configuration, one orbital s is mixed with two orbital p while in the sp hybridization one orbital s is mixed with one orbital p. Hybridization is very important in the chemical and physical properties of carbon as this phenomenon is responsible for the formation of carbon allotropes. When an element can exist in two or more different forms, this property is called allotropy.

Carbon has three allotropes when carbon atoms are arranged in a three-dimensional structure. These allotropes are called: diamond, graphite and amorphous carbon. In diamond, carbon atoms are arranged in a cubic lattice and they have sp3 configuration. In graphite, carbon atoms are sp2 hybridized. Graphite structure is made of many layers of plane held together by van der Waals forces. Each plane is composed of carbon atoms arranged in a hexagonal crystal structure. Lastly amorphous carbon does not have a long- range precise crystalline structure even though in a short-range carbon atoms are bound together by covalent bonds in a sp2 and sp3 hybridization. Different hybridizations are responsible for different physical properties. For instance, it is common knowledge that 20 graphite is an opaque and soft material that conducts electricity while diamond is a transparent and hard material which has low electrically conductivity.

In the last decades, many researchers have worked on low-dimensional carbon allotropes and especially on sp2 low dimensionality carbon allotropes. In 1985, Kroto et al published the first report [1] that graphite powder can be vaporized by laser irradiation to form a polygon made of sixty carbon atoms, thirty of those are arranged in a hexagonal shape and the rest are arranged in a pentagonal one. This molecule is called C60 or buckminsterfullerene and is a 0-dimensional carbon allotrope. Kroto, Smalley and Curl won the Nobel Prize in Chemistry in 1996 for this discovery. In 1991 the Japanese researchers Iijima et al reported [2] the synthesis of graphitic carbon arranged in a tubular structure with a diameter ranging from 4 to 30 nm and 1 µm long. The synthesis was carried out at the negative end of a carbon electrode at the pressure of 100 torr. Carbon nanotubes are 1-dimensional allotropes of carbon atoms, which have remarkable mechanical [3,4], electrical [5] and optical [6] properties.

In 2004, the last low-dimensional carbon sp2 allotrope was isolated and characterized by a team lead by Sir Konstantin Novoselov and Sir Andre Geim [7]. They were able to succeed in exfoliating micromechanically highly oriented pyrolytic graphite (HOPG) down to a single layer of carbon atoms.The single layer is called graphene and it is a 2- dimensional crystal made of graphitic carbon arranged in a hexagonal lattice pattern. Graphene is characterized by unique properties and since 2004 a growing interest has been raising to exploit those properties in different applications.

Figure 1.1: a) Optical microscopy image of micromechanical exfoliated graphene flakes ; b) and c) AFM topography image of graphene flakes, d) SEM image of graphene field effect transistor, e) Schematic 21 representation of graphene FET device. Image taken from [7] 1.2 Graphene and related nanomaterials production

The method reported by Geim and Novoselov [7] is the micromechanical exfoliation of HOPG where the different layers are pulled apart mechanically until a single layer or few layers graphene is obtained as shown in Fig1.1. The advantage of this method is to allow obtaining very high quality graphene that is useful for high-speed electronics and fundamental research studies. The carrier mobility measured by Geim and Novoselov [7] ranged from 3,000 cm2/V∙s up to 10,000 cm2/V∙s. On the other hand is not possible to reach large-scale production of graphene flakes by this method. Other production routes can be exploited to obtain large-scale production of graphene flakes but with lower quality compared to micromechanical exfoliation.

Chemical vapour deposition (CVD) consists in heating hydrocarbons in gas phase at 800- 1000 ͦC inside a hot furnace under vacuum. Usually hydrocarbons are placed inside the furnace in a gas phase at low concentration. For this purpose, methane is often used as

Figure 1.2: A and B) SEM image of CVD-graphene films , C and D) CVD-graphene films transferred on Si/SiO2 and glass substrates respectively. Image taken from reference [8]

22 gas precursor. At that temperature, methane decomposes on a metal catalyst surface and carbon atoms diffuse on the metal film. During the cooling process, carbon diffuse out of the metal surface forming graphene layers. The graphene growth occurs as the metal surface has a lattice similarity with the hexagonal lattice of graphene. This technique was used exploiting different metals such as Cu [8,], Ni [9,10], Ru [11], Pd [12] , Pt [13] and Co [14]. Cu is the most used catalyst metal as carbon diffuses slower on copper foils compared to other metals and this improves the uniformity of the graphene film and also the monolayer yield. . Li et al [8] achieved the growth of large area uniform graphene film made for 95% of monolayer graphene as shown in Fig 1.2. Carrier mobility extracted from a field effect-transistor (FET) made using CVD was around 4,050 cm2/V∙s.

Epitaxial growth of graphene [15,16] consists in depositing graphite on H2-etched SiC wafers. The samples are heated up to 1450 ͦC and monolayer or few layer graphene are formed on the surface of epitaxially matched SiC surfaces. Epitaxial graphene can be used to fabricate FETs with a carrier mobility of about 2,000cm2/V∙s.

Liquid exfoliation of graphite powder can be achieved by dissolving graphite powder in organic solvents such as N-methyl-pyrrolidone (NMP), N,N-Dimethylacetamide (DMA), g-butyrolactone (GBL) and 1,3-dimethyl-2-imidazolidinone (DMEU) under sonication [17]. After sonication, the product is purified by centrifugation and macroscopic aggregates are removed from the solution. The conductivity measurement of liquid- exfoliated graphene films was found to be around 6,500 S/m.

Figure 1.3: a-b) SEM of sieved graphite and from the sediment after centrifugation respectively; c-g) bright field TEM image of exfoliated graphene obtained after the liquid exfoliation process.; h) Histogram of visual observation of flakes as function of number of layers. Image taken from [17]

Another liquid-phase route to produce graphene is production and the reduction of graphene oxide (GO). GO is the main derivative of graphene. GO like graphene is a two- dimensional carbon nanomaterial with a true monolayer flake shape. GO is the product of the oxidation and of the exfoliation of flake graphite. The oxidation of flake graphite can be achieved through different ways. Brodie [18], Hoffmann [19] and Staudenmeier [20] have used potassium cholate as oxidising agent. The reaction media in Brodie` s reaction is a mixture of nitric and sulphuric acid, Hoffmann`s method involves the use of non-fuming nitric acid and Staudenmeier` synthesis is performed in fuming nitric acid. Hummer`s method [21] uses a mixture of sodium permanganate and sodium nitrite as oxidising agents. The reaction media is composed by concentrated sulphuric acid. Lastly, the method developed by Tour [22] involves the use of sodium permanganate and a reaction media composed by a mixture of phosphoric and sulphuric acid. The exfoliation of graphite oxide down to a monolayer is achieved more often by stirring or sonicating in water or in hydrophilic solvents. A combination of stirring and sonication is often performed.

The exact structural model of GO is still under debate and there is not a model that is uniformly accepted from the scientific community. The model that has attracted the widest adoption is the Lerf–Klinowski [23], which describes GO as composed of two different regions.

Figure 1.4: Structural models of GO proposed by Lerf-Klinowski. Image taken from [25]

The basal plane is composed of carbon atoms mainly sp2-hybridized arranged in a graphene-like structure. The basal plane is also composed of hydroxyl and epoxy functional groups bound to carbon atoms, which are sp3-hybridized. On the edges of the graphene layers, carboxylic functional groups are the main functional groups in that molecular region. An alternative model has been proposed from Dekany and co-workers [24]. Their model denies the presence of carboxylic groups in GO molecule and the functionalities are mainly cyclohexyl species interspersed with tertiary alcohols and 1,3- ethers, and a corrugated network of keto/quinoidal species. Those two different models are the most accepted today from the scientific community.

GO has very different properties compared to graphene as the disruption of sp2-hydridized carbon atom networks has a profound impact on the properties of the materials. For instance, GO is an insulating material while graphene is a very conductive one. The restoration of π-network can restore electrical conductivity and therefore the reduction of GO is the most performed of the reaction of GO. Reduced GO (rGO) can be obtained chemically, thermally and electrochemically.

The chemical reducing agents employed in the chemical reduction of GO are mainly hydrazine monohydrate [26], sodium borohydrate [27], hydroquinone [28] and ascorbic acid [29]. Thermal reduction under vacuum occurs as high temperature creates an enormous increase in the pressure in the stacked layer. Functional groups are not stable at temperatures much higher than room temperature and the removal of oxygen functionalities occurs by formation of carbon oxides.

Figure 1.5: AFM topography image of reduced graphene oxide flakes. On the right hand side height profile measurement of the flakes on the spot highlighted from the black lines. Image taken from [30]

Reduction of GO can occur also at the surface of electrodes and when a linear sweep voltammetry is applied a reduction reaction takes places at voltages ranging from -0,60 to -0,90.

All these reduction methods are successful in deoxygenating the starting GO molecules and in restoring a graphene-like material. Electrical conductivity measurement confirms that rGO conductivity is restored compared to GO. Measurement performed on freestanding rGO papers at room temperature resulted in conductivity of 7,200 S/m [30]. This value is similar to the value measured for liquid-exfoliated graphene [17].

Regarding mechanically exfoliated graphene and CVD graphene, the most useful characterization techniques employed are optical microscopy [31], atomic-force microscopy (AFM) [32] and Raman spectroscopy [33]. X-ray photoelectron (XPS) and Fourier transform infra-red (FTIR) spectroscopies are techniques mostly used to analyse the presence of functional groups in GO samples, the extent of de-oxygenation in rGO samples and the residual presence of the organic solvents in liquid exfoliated graphene.

1.3 Characterization of graphene and graphene related nanomaterials by microscopy

Graphene flakes and CVD graphene can be seen when it is exfoliated on top of silicon/silicon dioxide wafers due to the interference effect. The thicknesses of silicon dioxide wafers for an optimal contrast are 90 nm and 290 nm. The monolayer graphene flake appears almost transparent while few-layers graphene become darker with the increasing number of layers [31]. GO is semi-transparent material as it is typically insulating. Therefore, the decrease in both the carrier mobility and carrier concentration is followed by a decrease in the reflection. On the other hand, rGO due to the restored conductivity increases its transparency and becomes clearly visible as graphene flakes [34]. A typical example of optical characterization of graphene samples on top of silicon oxide wafers with different oxide thicknesses is shown in Fig. 1.6 and Fig. 1.7.

Optical microscopy is a good technique for initially checking the thickness of graphene but it does not provide additional information such as surface roughness, thickness of contaminants deriving from the fabrication process.

Figure 1.6: Optical microscopy image of micromechanically exfoliated graphene flakes on top of silicon oxide substrates with different thicknesses. Image taken from [31]

AFM is a powerful technique to gain information on mechanical properties of graphene, the roughness, the thickness and the uniformity of graphene flakes and graphene thin films.

Usually a monolayer graphene flake has a thickness ranging from 0.4 to 1.7 nm [31, 35]. Increased step height can be caused by increased number of layers, polymers or biomolecules grafted on the surfaces. AFM due to the uncertainty on the thickness measurement is often accompanied through spectroscopy techniques such as Raman to

Figure 1.7: A) GO and rGO films; B) GO and rGO dispersions; C) Optical microscopy image 27 of GO flakes; D)Optical microsocpy image of rGO flakes. Image taken from [34] give additional information on thickness and presence of defects in the lattice and XPS, FTIR to give information on the chemical properties of these materials.

1.4 Characterization of graphene and graphene related nanomaterials by spectroscopy techniques

Raman spectroscopy is a powerful tool regarding carbon nanomaterials characterization. The Raman spectrum of sp2 carbon is composed by three main peaks: the G peak which lies in the range of around 1585 cm-1, the D peak which arizes at around 1350 cm-1 and the second order of the D peak which is found to lie in the range of 2500-2800 cm-1.

The G peak is caused by the vibration of the graphitic plane and corresponds to the E2g phonon. This peak belongs to all graphitic structures and therefore to all sp2 carbon nanomaterials. Splitting of the G peak can be caused by defect in the crystal or by applying strain to graphene. Electronic doping can be detected by this technique as in this condition the position of the peak will shift towards higher wavenumber.

The 2D peak is important to quantify the number of layers. In monolayer graphene, the 2D is made of a single sharp peak. In bilayer graphene, the 2D gets broader and can be fitted in 4 small components (2D1B, 2D1A, 2D2A, 2D2B.) and is also up-shifted compared to the monolayer graphene. When the number of layers is between 3 and 5, the 2D peak is very similar to the bilayer spectrum with the only difference that the relative intensity of the 2D1A and 2D 1B components decreases. When the number of layer is bigger than 5, the 2D peak is very similar to the bulk graphite and can be fitted in the same way. It is made up of two components 2D1 and 2D2 as shown in Fig. 1.9.

D peak only occurs when the lattice is defected and corresponds to the breathing mode of the six carbon atoms in the graphitic rings. Therefore it is widely used as a Raman fingerprint to identify defected sp2 carbon nanomaterials. Mechanical exfoliated graphene, CVD graphene and liquid exfoliated graphene who not have D peak due to the integrity of the lattice. D peak can be appreciated at the edges of the flakes, and when their lattices are defective. In the case of GO, the presence of the oxygen-containing functional groups creates defects in the flakes. Therefore, the D peak and the G peak compose the Raman spectrum of GO.

Figure 1.8: AFM topography image of monolayer (a,b), bilayer (c,d) and trilayer (e,f) graphene flakes with height profile measurement performed in repulsive (red arrows) and attractive regimes (green arrows). Image taken from [31]

The ratio between the intensities of the D peak and the intensity of G peak is used to characterize GO solutions. The range of ID/IG ratio lies between 0.85 and 0.9 and this variability depends on the different production routes. [26,36].

When GO is reduced to rGO, the ID/IG increases. This increase could be explained by that the average size of the sp2 domains is smaller after the reduction but the number of the domains gets bigger. As illustrated by Fig. 1.10, an increase of ID/IG ratio occurs also when GO is functionalized covalently as a result of further distorted materials due to the introduction of external molecules.

XPS is a powerful technique which gives useful information about the chemical content present on the analysed surface. XPS is very useful in GO and rGO samples as it allows analysing the level of oxygenation before and after the reduction process.

XPS can be useful in detecting the introduction of new chemical bonds after a functionalization reaction. XPS wide scan spectra of GO consists predominantly of two peaks. The C1s peak occurs at around 284.6 eV and the O1s at around 534 eV. The C/O ratio is used to analyse the level of oxidation in GO synthesis. The range of C/O ratio in GO is between 2. 47 and 1. 95 depending on the different routes of production. The Tour and the Hummers methods allow obtaining an increased level of oxidation. The deconvolution of C1s peak and the O1s enables us to understand which functional groups are present in the GO molecule. [37]

C1s peak for GO molecule was deconvoluted differently in the literature. Initially, many researchers adopted a four-component deconvolution of the C1s model. This C1s peak deconvolution model presented C-C bonds at 284.8 eV, C-O bonds at 286.2 eV, C=O bonds at 287.8 eV and carboxyl groups 289.0 eV.

Figure 1.10: Raman spectrum of functionalized graphene sheets, GO and graphite . Taken from reference [36]

This fitting model does not represent very well the amount of different chemical functionalities of GO. Others reported C1s peak of GO deconvoluted in 6 components [39]. C=C (sp2 carbon) is centred at 284.6 eV, C-C (sp3 carbon) at 285 eV, C-OH (hydroxyl groups) at 286.3 eV, C-O-C (epoxy rings) at 287.2 eV, C=O (carbonyl) at 287.8 eV and HO-C=O (carboxyl groups) at 288.8 eV.

When GO is reduced, obvious changes are observable both in the C/O ratio and in the C1s deconvolution. The C/O ratio increases after the reduction reaction. The C1s deconvolution shows an increased peak of C=C (sp2 carbon); C-C (sp3 carbon) and C-O- C (epoxy group) are decreased while C-OH (hydroxyl group) is increased. [39,40]

FTIR spectroscopy is often used in combination with XPS to analyse the sample chemical structure. In the case of GO, we have a strong peak at 3450 cm-1 which originates form

Figure 1.11: XPS C1s deconvolution in 6 components. XPS C1s spectrum of GO (a), Graphite oxide (b), diamond (c), HOPG (d), sodium terephthalate (e), sodium dodecanedioate (f), hydroqunone (g). Taken from [39].

31 the OH vibration stretching, the carboxyl C=O (1690 cm-1) and C-O at (1400 cm-1), the epoxy peak at 1250 cm-1 and finally the hydroxyl C-O at 1100 cm-1 [110]. After the reduction of GO, most of the peaks corresponding to the oxygen functionalities are strongly decreased.

1.5 Graphene Oxide Functionalization GO can be functionalized covalently and non-covalently. Functionalization can tune the properties of the materials such thermal conductivity, electrical conductivity, surface area and solubility. Covalent functionalization usually exploits the reactivity of carboxylic, epoxy and hydroxyl groups. Non-covalent functionalization is based on electrostatic interaction, van der Waals and π-π stacking. [41]

Carboxylic groups are not good leaving groups in nucleophilic addition reaction and therefore require an activation step. Activating agents often used are thionyl chloride

(SOCl2) [42,43,44], carbodiimides such as 1-ethyl-3-(3-dimethylaminopropyl)- carbodiimide (EDC) and 3 N,N΄-dicyclohexylcarbodiimide (DCC) [45,46,47]; and finally -(7-aza-1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HATU) [48]. After the activation step, nucleophilic species such alcohol and amines are able to react in the synthesis of amidated or esterified GO. Many research groups worldwide have described this covalent route to create amidated GO and their applications in optoelectronics, drug delivery, biological devices and polymer composites.

Figure 1.12: Functionalization of GO by exploiting carboxylic functional groups. Image taken from [25]

Opening of epoxy rings is another way of functionalising GO covalently. The reaction of the nucleophile with the α-carbon atom and the consequent opening of the ring can graft nucleophilic molecules with GO. An amine such as octadecylamine was grafted to GO by this reaction scheme. Amine can be used in their polymeric, aliphatic and aromatic species. [49,50].

Hydroxyl groups can be used in an esterification reaction. In this case, the hydroxyl groups present on the surface of GO can react with molecules such as S-1-dodecyl-S΄- (a,a΄-dimethyl-a΄-acetic acid) trithiocarbonate (DDMAT) [51] which present carboxylic groups . Alternatively siloxy linkages can be performed by the reaction of the hydroxyl groups and trialcoxysilanes or alkyltrichlorosilanes. [52,53,55]

GO and rGO can react non-covalently with polymers such as poly(methyl methacrylate) (PMMA) [54], polyaniline (PANi) [56], polycaprolactone (PCL) [57], and polyurethane (PU) [58] by simply solution mixing exploiting hydrophobic interaction between the sp2 carbon atom rings of both the polymers and GO/rGO. Composites made with epoxy- containing polymers and PANi are often synthesized by in situ polymerization, which means that the polymer is firstly mixed with GO/rGO solutions and after the initiator of the polymerization is added to the reaction solution.

GO can also be non-covalently functionalized with DNA and other biomolecules such as protein exploiting through non-covalent interaction [59]. GO-doxorubicin hybrid was synthesized exploiting hydrogen bonding between the polar groups of GO and the doxorubicin molecule together with hydrophobic interaction between the sp2 networks of GO and the quinone functionality of the antibiotic [60].

Graphene is more difficult to covalently functionalize as it lacks the presence of chemical functional groups inside its lattice structure. In the next paragraph it is discussed the use of Dip Pen Nanolithography (DPN) as a tool to non-covalently functionalized different surfaces with particular focuses on lipid functionalization of graphene surfaces.

1.6 DPN-assisted Functionalization of Graphene

DPN is a useful lithographic tool to chemically functionalize any surface with a spatial resolution of sub-100 nanometers. This technique is based on the functionalization of AFM tips with a molecular ink, which is deposited on the surface after the functionalized tips have contacted the surface. The group led by Chad Mirkin at the Northwestern

University developed this technique. They firstly reported geometrically ordered arrays of 1-octadecanethiol (ODT) self-assembled monolayers (SAM) on gold surface [61]. This technique has many advantages over other indirect lithography techniques; in fact there is no need to use denaturing conditions such as electron-beam, ultraviolet or ion irradiations, organic solvents. All these conditions can cause cross-contaminations and structural damages to biomolecules [62]. DPN greatly advanced the fabrication of biological nano-arrayed biosensors as it was proved that different biological soft matters like DNA [63,64], proteins [65,66], peptides [67.68], viruses [69,70] and bacteria [71] can be deposited on surfaces by using this technique without being damaged.

Lee et al [73] showed the feasibility of fabricating biological nanoarrays to screen the presence of human immunodeficiency HIV-1 virus (HIV-1) p24 antigen in human serum. Briefly, they formed arrays of mercaptohexadecanoic acid (MHA) nano-dots on gold substrates. These nano-dots were used as templates to immobilize antibodies on the surface of the array. The presence of HIV-1-p24 is evaluated by measuring the height profile of the anti p-24 antibody arrays. After the binding between the antibody and the antigen, increased height profile is detectable by AFM height measurement. Researchers in this study avoided non-specific interaction by passivation of the surfaces using bovine serum albumin (BSA).

By the same sensing mechanism, Kang et al [74] measured molecular interactions between printed integrin ανβ3 nano-arrays and cell-adhesion proteins.

Figure 1.13: Schematics of DPN patterning. Images taken from [72]

Lipid Patterning by DPN is a functional tool to fabricate biomimetic lipid membranes on solid substrates [75]. The formation of biomimetic membranes allows creating bioactive surfaces to be used in biosensing. Sekula et al [75] showed the formation of lipid biomimetic membranes by DPN. The phosphate groups of the lipids were functionalized with biotin and nickel chelating molecules. Binding between histidine-tagged and streptavidin-tagged proteins was detected by fluorescence microscopy. This study demonstrated the stability of the lipid patterns under cell culture conditions. The same group reported the successful fabrication of an allergen screening sensor based on the binding between amine functionalized lipid head groups and Ig-E antibodies. [76]

Graphene is a promising sensing surface due to its electrochemical, nano- electromechanical and optoelectronic properties. Previous reports showed functional graphene sensors for DNA sequencing [77], electrical nose [78], and electrochemical detection of biomarkers [79]. Chemical functionalization of graphene is required to increase sensitivity and selectivity. Covalent chemistry is not advisable as it disrupts the sp2 carbon bonds networks thus decreasing the electronic properties of the materials. DPN-assisted functionalization is an important technique to form biomimetic membranes on graphene surfaces, which allows chemical functionalization without disrupting the electronic properties of the materials.

Hirtz et al [80] realized the first study where functionalization of pristine graphene by lipid DPN was performed. They achieved the formation of biomimetic lipid patches on graphene flakes. The lipid showed a higher lateral mobility on graphene compared to the hydrophilic silicon oxide thus they spread forming uniform lipid biomimetic membranes due to the higher hydrophobicity of graphene surfaces. Experimental thickness of lipid patches in liquid environment has not been performed as this kind of experiment will be hindered by the blocking BSA layers. This configuration in liquid was proven by a successful biotin-streptavidin binding experiment. The binding was detected by fluoresce quenching assay [80]. Because graphene quenches the fluorescence from the lipid, no fluorescence is observed before the binding event. After the binding event, the fluorescence is restored due to the increased spatial distance between the graphene surfaces and the fluorophore. The binding can only occur if the biotin-containing phosphate groups are clustered towards the streptavidin-containing buffer proving the monolayer formation in liquid environment.

Hirtz et al [81] improved the sensor fabrication by fabricating multiplexed biomimetic membranes on large area graphene sensor arrays using a multiple DPN-pens array. They also showed that the spreading of the lipid membranes is well confined to the graphene without spreading into the surrounding silicon oxide wafer. The biological activity of the lipid interface was therefore demonstrated following the streptavidin-biotin binding array.

Next step in the fabrication of such graphene-based biosensor is binding functional proteins into the lipid layers. Example of functional proteins can be antibodies, ion channels, membrane receptors for a variety of biotechnological applications such as biosensing and drug screening for instance.

1.7 Graphene and related nanomaterials based substrates for stem cell culture and differentiation

Graphene and related nanomaterials have been attracted in the recent years a tremendous interest as biomaterials to be used in the fabrication of substrates for stem cell culture and differentiation. The ease of functionalization of GO with biomolecules such as growth factors, oligopeptides and proteins, nucleic acids make these makes this material very attractive for this kind of applications. On the contrary rGO and CVD graphene lacs the availability of functional groups, but it offers other properties such as electrical conductivity, mechanical stiffness, thermal conductivity which can be exploited as well in tissue engineering. All graphene and related materials have high specific surface area, which allows superior loading capacity and higher concentration of biomolecules on the surfaces of the cell culture substrates. In this paragraph, it is going to be discussed the biological classification of different types of stem cells and the existing literature on the fabrication of these substrates and the impact of these materials on the differentiation of different types of stem cells such as adult stem cells and pluripotent stem cells.

A profound difference in stem cell biology exists between adult stem cells and pluripotent stem cells. Adult stem cells can only being differentiated towards a limited amount of phenotypes while pluripotent stem cells such human embryonic stem cells (hESC) or induced pluripotent stem cells (iPSC) can be differentiated towards any cells or tissue of the human organism [82]. hESC are produced after the approval of ethics committees. After the approval of the ethical authorities, donated human embryos are cultured until they reach the blastocyst development stage and hESCs are developed from the blastocyst. hESC are characterized

36 by the overexpression of Oct-4, Oct3a/4, Nanog and Sox2, as well as other transcription factors such as the pluripotency markers Lef, Tcl1, Essrb. All these markers on one hand down-regulate the transcription of lineage-specific genes and on the other hand increase the transcription of all the genes involved in the maintenance of the pluripotency state. When hESCs are differentiated to a specific lineage, there is a decrease of expression of the pluripotency markers and an increase in the gene expression of tissue-specific markers [83].

Human iPSC are cell produced in vitro by treating the cells with the Yamanaka factors. Yamanaka won the Nobel Prize in Physiology and Medicine in 2012. His research team was able to transform mature cells such as mouse or human skin fibroblast cells into pluripotent stem cells. His team induced the expression of Oct3/4, Sox2, c-Myc, and Klf4 by using retroviral integration technique [86,87]. They noticed that after the fibroblasts start expressing the 4 genes they introduced, their cellular shape and their growth properties were similar to hESCs. Researchers noticed that these cells overexpressed many hESCs cells such as Oct3 /4, Sox2, Nanog, Stat3, E-Ras, c-myc and β-catenin.

Regarding adult stem cells, bone marrow is a source of adult stem cells, which can be divided, in two different categories such as bone marrow mesenchymal stem cells (BM- MSC) and haematopoietic stem cells [90]. BM-MSC can be differentiated into multiple mesenchymal lineages like bone, adipocytes and chondrocytes. [91].

Friedenstein et al [92] were able to selectively culture BM-MSC fraction into plastic substrates used in cell cultures due to the extremely high cellular adherence of those cells compared to the haematopoietic fraction. The biological characterization of BM-MSCs can be performed by analysing the expression of surface markers. The biological requirements for MSCs classifications are the presence of expression of CD105, CD73, CD90, CD44 and CD106 and the loss of expression of CD45, CD34, CD14, CD11b, CD79a, CD19 and HLA-DR. [93]

When the BM-MSC are differentiated the expression of the stemness markers decreases and the expression of tissue-specific markers increases.

Dezawa et al [94] were able to transdifferentiate for the first time BM-MSC population into Schwann-like cells by treating not differentiated BM-MSC with beta- mercaptoethanol, followed by treatment with all-trans-retinoic acid and lastly cells were cultured for 14 days with the presence of specific growth factors such as forskolin, bFGF,

PDGF and heregulin-1. The explanation of this protocol was given by Kuroda et al [95]. Briefly, beta-mercaptoethanol increases the production of glutathione therefore driving the differentiation towards neuronal lineages, all-trans-retinoic acid increases the biological reactivity of the cells to the action of neurotrophins. The presence of growth factors in the two weeks of cell cultures induces cell survival and increases the cell proliferation during the maturation of Schwann-like phenotype.

Adipose-derived Stem Cells (ASC) shares almost all phenotypic markers of BM-MSC, even though some differences in the protein expression were observed. ASC can be differentiated into mesenchymal lineages such as fat, bone and cartilage [96].

Compared to BM-MSC cells can be harvested from the patient in a less painful way simply using liposuction procedures which does not require general anaesthesia [97]. The adipose tissue harvested from the patient can then be digested by collagenase obtaining two different cell populations: the adipocytes which are not adherent to the plastic supports which are discarded by centrifugation and the ASC that adhere to the plastic supports and can be cultured and differentiated towards different cellular phenotypes. [98]

Several groups were able to transdifferentiate ASC into Schwann-Like Differentiated Adipose Stem Cells (dASC) following Dezawa`s protocol. dASC express different molecular marker of Schwann cells such S100, p75, GFAP and p0 [99]. Faroni et al [100] demonstrated that this process is not truly transdifferentiation of stem cells but the ASC are pushed towards glial lineages by the growth factors present in the cell media. When the growth factors are withdrawn from the cell media, the expression of GDNF and BDNF decreased while NGF and NT3 increased. Moreover, the expression of neurotrophin receptors decreased together with the neuregulin pathway. This study highlights a potential problem of using dASC obtained from ASC as a replacement of Schwann cells for the clinical use. A better protocol has to be defined to obtain a continuous delivery of growth factors to maintain a stable Schwann-like phenotype.

Graphene has been reported to enhance cardiomyogenic and hematopoietic differentiation of hESC.

Lee et al [84] reported the impact on graphene-based substrates in directing the differentiation of hESCs towards cardyomyocites. Their study compared graphene and glass substrates coated with the protein vinculin. Their results showed that graphene substrates increased the expression of markers such as cardiomyogenic transcriptional

38 factors (NKX2-5, GATA4, and MEF2C), cardiomyogenic contractile proteins (α-MHC, β-MHC, MLC2a, and cTnT), and gap junction proteins (CONNEXIN43) compared to glass substrates. Garcia-Alegria et al [85] showed that GO substrates induce higher haematopoietic differentiation of hESCs compared to gelatin control substrates. Their results showed that GO increased the extent of both primitive and definitive haematopoiesis compared to gelatin controls. Their study followed the amount of the expression of embryonic βH1 haemoglobin locus, a primitive haematopoiesis marker, and the expression of CD41, a definitive haematopoiesis marker. In both cases, GO induced higher expression compared to gelatin. Gene expression studies of haematopoietic genes Pu.1, MPO, CD45, and Gata1 showed that GO substrates induced higher expression compared to gelatin substrates accelerating the haematopoietic differentiation of hESCs.

The effect of graphene-based substrates on human iPSC growth and differentiation is a new field of research. Yoo et al. [88] reported a higher rate of cellular reprogramming into the iPSC state on CVD graphene substrates compared to glass substrates. Their results showed that graphene substrates induced higher gene expression of pluripotency markers such as Oct4, SSEA1 and Nanog compared to the control. Oct4 -tagged colonies with green fluorescent protein (GFP) were significantly more on CVD graphene substrates compared to glass substrates.

Chen et al. [89] used GO and rGO coated substrates to study the impact of these materials on the germ line differentiation of induced pluripotent stem cells (iPSC). iPSC were seeded on glass substrates, GO substrates and rGO substrates at the density of (1 x 104 cells/cm2). The authors selected a model where iPSC differentiation lead to a decrease in the expression of the GFP. Their results showed loss of pluripotency after 9 days as demonstrated by a decreased GFP expression on glass and GO substrates while no loss of pluripotency occurred on rGO substrates as the expression of GFP remained stable. Expression of lineage-specific genes was measured to assess which lineage differentiation was preferred on the different substrates. The results showed an increased expression of endodermal markers on GO substrates and reduced expression of these markers in the rGO substrates. Ectodermal and mesodermal differentiation of iPSC was not statistically different between the different groups of substrates studied.

In conclusion, it appears in particular that CVD graphene and rGO substrates are able to induce higher rate of cellular reprogramming into iPSC pluripotent state and GO

Figure 1.14: a) Relative expression of pluripotent stem cells markers; b) Fluorescence microscopy image after immunohistochemical analysis of Sox2 and Oct4. Image taken from [89] substrates induce increased expression of endodermal markers compared to glass and rGO substrates.

More research has been conducted on graphene-based substrates for the growth and the differentiation of BM-MSC and ASC.

Nayak et al [101] studied the effect of graphene substrates on mesenchymal stem cells attachment and proliferation. In their study glass, silicon oxide, polyethylene terephthalate (PET) and polydimethylsiloxane (PDMS) substrates were coated with CVD graphene. Their findings confirmed that the graphene had no effect on cell viability and cell morphology. When the MSC were grown under osteogenic media, uncoated substrates showed a CD+44 positive and osteocalcin negative staining confirming the lack of osteogenic differentiation while the opposite behaviour was observable on coated substrates which showed osteogenic differentiation and moreover calcium deposit increased notably with the presence of graphene coatings in all the different substrates enforcing the positive role of graphene in leading to osteogenic differentiation, even without adding osteogenic growth factors in the cell media. To assess whether this effect is due to the properties of graphene, the same experiments were performed with 40 amorphous carbon thin films and HOPG coatings. In all this group, signs of differentiation were not observable demonstrating that osteogenic differentiation is merely promoted by the physical properties of graphene.

Kim et al [102] tested the adhesion, proliferation and differentiation of human ASC on GO-coated substrates and they compared the results to bare glass coverslips. Human ASC were characterized by fluorescence-activated cell sorting (FACS) and they were expressing typical ASC markers such as CD105, CD190, CD29 and they lacked the expression of CD45. Results showed no statistically difference in cell adhesion and in the nucleus shape index between glass and GO substrates. Cell shape index (CSI) increased in the GO samples to 0.338 ± 0.138 compared to glass samples (0.249 ± 0.094,) highlighting that GO substrates induces circular shape to this stem cells. The attachment of human ASC on the different substrates is assessed by immunohistochemical analysis of vinculin expression. The expression of vinculin gives an indication of the number of

Figure 1.15: Immunostaining for desmin and osteocalcin (OCN) in green and nuclear marker DAPI In blue to evaluate the osteogenic differentiation of hMSCs grown on PET , PDMS and Si/SiO2 substrates with and without CVD graphene. Image taken from [101]

41 focal adhesions. Their results showed an increased number of focal adhesions on GO substrates compared to the glass controls, indicating a positive effect of GO on human ASC attachment. To assess the impact of GO on human ASC differentiation, tissue culture polystyrene (TCPS) substrates were 5selected as a control group. Osteogenic differentiation was promoted by GO substrates (P<0.05) as demonstrated by Alzarin Red S staining. Adipogenic differentiation was also promoted by GO substrates (P<0.05) as demonstrated by Oil Red O staining (P<0.01). Chondrogenic differentiation showed the opposite behaviour: staining with Alcian Blue revealed that this differentiation process was lower on GO substrates (P<0.05).

Cheng Lee et al [103] studied the osteogenic differentiation process on GO-coated and CVD-coated substrates. They used PDMS as control substrates. Cell morphology experiments proved that BM-MSC grown on CVD and GO substrates have a spindle shape which is typical for BM-MSC. DAPI staining also indicated that more cells were present on GO and CVD substrates compared to PDMS substrates. The cell morphology

Figure 1.16: Evaluation of osteogenic differentiation on PDMS, CVD graphene (G) and graphene oxide (GO) substrates by Alzarin Red staining. Image taken from [103].

42 differs completely on PDMS samples where the cells looked more circular and cellular protrusions were not observable. The difference in cell morphology however seemed less pronounced after one week from plating. GO and CVD substrates compared to PDMS substrates. The cell morphology differs completely on PDMS samples where the cells looked more circular and cellular protrusions were not observable. The difference in cell morphology however seemed less pronounced after one week from plating. Staining with Alizarin Red S revealed that osteogenic differentiation is more pronounced on CVD- coated substrates compared to GO and PDMS substrates (P<0.05). Staining with Oil Red O proved the opposite behaviour with adipogenic differentiation: GO substrates strongly enhanced adipogenesis with a value of 32.3 μm2/cell of lipid accumulation while CVD- coated substrates have a lipid accumulation of 6.29 μm2/cell. The authors postulate that the concentration of osteogenic growth factors such as dexamethasone and β- glycerolphosphate measured by UV-spectrophotometry was as higher on CVD-coated substrates.

This is due to the π-π stacking interactions between the carbon atoms of the graphene lattice and the aromatic biomolecules. The same trend happened with the adipogenic differentiation where the main growth factor involved is the hormone insulin. Insulin was greatly adsorbed on GO substrates (4800 mg of insulin per gram of GO) compared to CVD (1900 mg of insulin per gram of graphene) and PDMS (1580 mg of insulin per gram of PDMS) substrates. GO compared to CVD had higher affinity towards insulin because of the functional groups which allowed hydrogen-bonding and electrostatic interactions between the substrates and the biomolecule. However these results confirmed that graphene-related materials could act as a better platform due to its ability to concentrate biomolecules involved in the differentiation process.

Alzhavan et al [104] studied the osteogenic differentiation on graphene nanoribbon array on PDMS. Their results indicated that the osteogenic differentiation was higher in rGO nanoribbons substrates compared to the GO nanoribbons nanoribbons and PDMS control substrates. The results were in good agreement with the study conducted by Cheng Lee et al because osteogenic growth factors were concentrated more in the rGO nanoribbons sample compared to GO nanoribbons and PDMS controls. The authors also compared rGO and GO nanoribbons substrates with the GO and rGO flakes. Their results showed that rGO and GO nanoribbons substrates performed better than rGO and GO substrates. This is explained as nanoribbons contained some functional groups on the edge of the

43 ribbons, which allows for some covalent bonding between the growth factors and the substrates to take place.

Qi et al [105] studied the osteogenic differentiation in substrates made by composite films of GO and poly-L-lysine (PLL). BM-MSC were able to adhere and proliferate on the GO/PLL substrates and by fluorescence staining most of the cells were alive. Also the number of the cells at day 7 on the GO/PLL was 186.0%, 142.7%, 105.8% and 120.7% higher than glass, GO alone, PLL alone and TCPS substrates respectively. Osteogenic differentiation was evaluated by immunohistochemical staining for osteocalcin and for alkaline phosphatase (ALP), an enzyme expressed in high quantity in osteoblasts. The results confirmed that on the day 2 no staining was observable for ALP. Positive staining was observable from day 7 on GO substrates, GO/PLL and PLL substrates alone. More interestingly from day 7 onwards the cells grown on GO/PLL substrates had a more intense staining compared to GO, PLL and TCPS substrates. The increased expression of osteocalcin and the decreasing staining of CD44 is consistent with an improved performance of graphene-based cell culture substrates. UV-spectrophotometry analysis revealed that GO-PLL and GO substrates had the highest absorption rates for osteogenic growth factors furtherly confirming that graphene-based materials enhance stem cell differentiation by allowing higher growth factor concentration on the substrates.

Tang et al [106] compared the performance of bare TCPS substrates with TCPS substrates coated with CVD graphene on improving neural stem cells (NSC) differentiation into neuronal network and to enhance neurite outgrowth. Scanning electron microscopy analysis confirmed that graphene-coated substrates showed healthier adhesion to NSCs. This was proven by the formation of filopodia/graphene interaction which occurs only on graphene surfaces. Electrical current was applied to the graphene substrates, the intracellular influx of ions was evaluated by fluorescence microscopy with a calcium-dependant dye. Results confirmed that an increase of 30% of fluorescence intensity was observable after electrical stimulation with high K+ stimulation, implying that an increase in calcium ion influx inside the cytoplasm of the differentiated NSCs is happening. Moreover, electrical stimulation also increased the expression of C-Jun. An increased of calcium influx is observable on graphene substrates even without external stimuli. Spontaneous calcium oscillations were two times higher on graphene substrates compared to TCPS (P<0.05). This study also highlighted that CVD-coated substrates increased the number of MAP-2 positive cells as proven immunohistochemically and by Western Blot analysis (P<0. 05). 44

Tu et al [107] studied the influence of surface charges of chemically modified graphene oxide solutions on neuronal outgrowth and branching. GO was chemically modified with amino (–NH2), poly-m-aminobenzene sulfonic acid- (–NH2/–SO3H), carboxyl (-COOH) or methoxyl- (–OCH3) functionalities. Substrates were prepared by coating firstly glass coverslips with thin layer of polyethyleneimine (PEI). Afterwards, the coverslips were wetted with the differently charged GO solutions and GO films were formed after ethanol evaporation. Their results showed that chemical modifications of GO did not alter the biocompatibility of these substrates. In fact, quantification of cell viability analysis of primary hippocampal neurons showed that 96% of cultured cells were alive and well adhered on all the different functionalized GO coverslips. When the authors analysed the impact of functionalization on the neurite outgrowth, they found out that the total number of neurites per neuron did not show any statistically relevant difference between the different groups of coverslips. But GO-NH2 coverslips showed longer maximum neurite length and higher total outgrowth per neuron, pointing out that positively charged substrates can be better suited coverslips for neurite outgrowth.

Figure 1.17: Immunohistochemistry on hippocampal neurons grown on functionalized graphene oxide substrates. Image taken from [107]

The zwitterionic GO-PABS and the neutral GO-OCH3 performed better than negatively charged GO-COOH highlighting that chemical modification of GO can be used to modulate neurite outgrowth by changing the surface charge on the cell culture substrates.

Tu et al [108] used biomimetic GO composites to study the effect of these substrates in boosting neurite outgrowth. GO composites were prepared by covalently binding thiol- terminated cysteamine to the activated carboxylic groups of GO. The thiol groups of cysteamine were then used in the second step of the reaction to immobilize an acetylcholine-like unit (dimethylaminoethyl methacrylate, DMAEMA) and a phosphorylcholine-like unit (2-methacryloyloxyethylphosphorylcholine, MPC). Thiolated GO solution at the concentration of 1 mg/mL was mixed with 1 mg/mL of DMAEMA or MPC solutions. Substrates were then prepared by spin-coating the two different composites solutions on glass coverslips. Cell viability assays on rat hippocampal neurons confirmed that both unmodified GO and GO-DMAEMA and GO- MPC coverslips are biocompatible and the quantification of alive cells reach more than 96% (GO: 97.5%, GO−MPC: 97.1%, and GO−DMAEMA:96.9%). Immunohistochemical analysis confirmed that most of adhered cells were hippocampal neurons. The effect of the composite functionalization was obvious analysing the neurite outgrowth. Compared to unmodified GO coverslips, a marked cell body and longer neurites were observable in GO-DMAEMA and GO-MPC coated coverslips. The number of neurites per neuron, the area of cell body, both the maximum and mean neurite length was significantly higher in GO-DMAEMA and GO-MPC coated coverslips compared to GO. To further investigate the molecular changes induced by the different coatings on the coverslips, the expression of the neuronal development marker GAP-43 was studied. Western Blot analysis revealed that the expression of GAP-43 was higher in both GO- DMAEMA and GO-MPC compared to unmodified GO (p<0.05).

GO can also be functionalized with conductive polymers such as poly(3,4- ethylenedioxythiophene) (PEDOT) or poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT/PSS) to produce electrically conductive composite materials. The presence of many negatively charged functional groups on GO make this molecule a good candidate to act as a dopant in the electro-polymerization of conducting polymers. Luo et al [109] tested the biocompatibility of GO-PEDOT coated coverslips compared to PEDOT/PSS coated coverslips. Cell viability analysis confirmed that rat neurons grow well both in GO-PEDOT coated coverslips and in the PEDOT/PSS controls with 12.79 ± 5.0% of dead cells in GO/PEDOT group compared to 20.61 ± 3.78 % of dead cell in PEDOT 46

/PSS group. Neurite length was longer on GO/PEDOT compared to the control experiments (36.4 ± 2.0 µm vs 22.5 ± 1.8 µm, p < 0.01. A laminin fragment peptide, p20, was bound to the graphene oxide composite. The effect of functionalization changes the electrical properties as GO/PEDOT/p20 thin films exhibits an increase of impedance after the reaction due to the formation of a non-conductive peptide layer, but it remained lower than bare metal confirming this strategy as a possible way to build electrically active coatings for neuronal microelectrode. The functionalization with p20 also increased the average neurite length compared to unmodified GO/PEDOT and GO/PEDOT modified by physical absorption of p20. In fact, covalently modified GO/PEDOT/p20 showed an average neurite length of 20.48 ± 1.45 µm, which is longer than the value measured for GO/PEDOT (14.29 ± 0.63 µm) and physically absorbed GO/PEDOT/p20 (14.59 ± 1.72 µm , p<0.05).

1.8 Conclusions

As discussed a growing interest has emerged in graphene and related nanomaterials research since the seminal research article on graphene written by Sir Andre Geim and Sir Konstantin Novoselov in 2004.

Figure 1.18: Neurite lenght outgrowth, number of branches, number of neurites and cell body area measurements of hippocampal neurons on GO, GO-MPC, and GO-DMAEMA substrates. Image taken from [108]

Graphite can be oxidized and exfoliated to produce GO, a water soluble derivative of graphene due to the presence of different chemical functionalities available on the graphitic structure.

These materials can be characterized by different microscopy and spectroscopy techniques. The thickness, the level of contaminants and the amount of defects of these materials is characterized by a combination of optical microscopy, AFM and Raman spectroscopy. Other spectroscopies such as XPS and FTIR are very important to evaluate the amount of oxygen and which functional groups are present on the graphitic structure in GO and rGO samples.

In this introduction, much attention has been also paid to covalent and non-covalent functionalization strategies of these materials. In particular, biomolecules can be bound to GO by exploiting the presence of carboxylic groups. Another approach covered in this introduction is the exploitation of DPN as a tool to biochemical functionalization to fabricate biosensors, with particular emphasis on lipid-DPN assisted functionalization of CVD graphene surfaces.

Lastly, an overview of the most recent work on the impact of graphene and related nanomaterials in enhancing stem cell growth and differentiation. Up to date, these materials have been reported to enhance stem cell growth and differentiation compared to glass and plastic substrates. This introductory chapter provided the theoretical framework of the methodology and results chapters which will follow.

References: 1. Kroto, H. W.; Heath, J. R.; O'Brien, S. C.; Curl, R. F.; Smalley, R. E., C60: Buckminsterfullerene. Nature 1985, 318, 162-163.

2. Iijima, S., Helical microtubules of graphitic carbon. Nature 1991, 354, 56-58.

3. Yu, M.-F.; Lourie, O.; Dyer, M. J.; Moloni, K.; Kelly, T. F.; Ruoff, R. S., Strength and Breaking Mechanism of Multiwalled Carbon Nanotubes Under Tensile Load. Science 2000, 287, 637-640.

4. Ruoff, R. S.; Qian, D.; Liu, W. K., Mechanical properties of carbon nanotubes: theoretical predictions and experimental measurements. Comptes Rendus Physique 2003, 4, 993-1008.

5. Hong, S.; Myung, S., Nanotube Electronics: A flexible approach to mobility. Nat Nano 2007, 2, 207-208.

6. Misewich, J. A.; Martel, R.; Avouris, P.; Tsang, J. C.; Heinze, S.; Tersoff, J., Electrically Induced Optical Emission from a Carbon Nanotube FET. Science 2003, 300, 783-786.

7. Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A., Electric Field Effect in Atomically Thin Carbon Films. Science 2004, 306, 666-669.

8. Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S., Large-Area Synthesis of High-Quality and Uniform Graphene Films on Copper Foils. Science 2009, 324, 1312- 1314.

9. Reina, A.; Jia, X.; Ho, J.; Nezich, D.; Son, H.; Bulovic, V.; Dresselhaus, M. S.; Kong, J., Large Area, Few-Layer Graphene Films on Arbitrary Substrates by Chemical Vapor Deposition. Nano Letters 2009, 9, 30-35.

10. Liu, N.; Fu, L.; Dai, B.; Yan, K.; Liu, X.; Zhao, R.; Zhang, Y.; Liu, Z., Universal Segregation Growth Approach to Wafer-Size Graphene from Non-Noble Metals. Nano Letters 2011, 11, 297-303.

11. Sutter, P. W.; Flege, J.-I.; Sutter, E. A., Epitaxial graphene on ruthenium. Nat Mater 2008, 7, 406-411. 49

12. Kwon, S.-Y.; Ciobanu, C. V.; Petrova, V.; Shenoy, V. B.; Bareño, J.; Gambin, V.; Petrov, I.; Kodambaka, S., Growth of Semiconducting Graphene on Palladium. Nano Letters 2009, 9, 3985-3990.

13. Sutter, P.; Sadowski, J. T.; Sutter, E., Graphene on Pt(111): Growth and substrate interaction. Physical Review B 2009, 80, 245411.

14. Varykhalov, A.; Rader, O., Graphene grown on Co(0001) films and islands: Electronic structure and its precise magnetization dependence. Physical Review B 2009, 80, 035437.

15. Berger, C.; Song, Z.; Li, X.; Wu, X.; Brown, N.; Naud, C.; Mayou, D.; Li, T.; Hass, J.; Marchenkov, A. N.; Conrad, E. H.; First, P. N.; de Heer, W. A., Electronic Confinement and Coherence in Patterned Epitaxial Graphene. Science 2006, 312, 1191- 1196.

16. Orlita, M.; Faugeras, C.; Plochocka, P.; Neugebauer, P.; Martinez, G.; Maude, D. K.; Barra, A. L.; Sprinkle, M.; Berger, C.; de Heer, W. A.; Potemski, M., Approaching the Dirac Point in High-Mobility Multilayer Epitaxial Graphene. Physical Review Letters 2008, 101, 267601.

17. Hernandez, Y.; Nicolosi, V.; Lotya, M.; Blighe, F. M.; Sun, Z.; De, S.; McGovern, I. T.; Holland, B.; Byrne, M.; Gun'Ko, Y. K.; Boland, J. J.; Niraj, P.; Duesberg, G.; Krishnamurthy, S.; Goodhue, R.; Hutchison, J.; Scardaci, V.; Ferrari, A. C.; Coleman, J. N., High-yield production of graphene by liquid-phase exfoliation of graphite. Nat Nano 2008, 3, 563-568.

18. Brodie, B. C., On the Atomic Weight of Graphite. Philosophical Transactions of the Royal Society of London 1859, 149, 249-259.

19. Hofmann, U.; König, E., Untersuchungen über Graphitoxyd. Zeitschrift für anorganische und allgemeine Chemie 1937, 234, 311-336.

20. Staudenmaier, L., Verfahren zur Darstellung der Graphitsäure. Berichte der deutschen chemischen Gesellschaft 1898, 31, 1481-1487.

21. Hummers, W. S.; Offeman, R. E., Preparation of Graphitic Oxide. Journal of the American Chemical Society 1958, 80, 1339-1339.

22. Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M., Improved Synthesis of Graphene Oxide. ACS Nano 2010, 4, 4806-4814.

23. Lerf, A.; He, H.; Forster, M.; Klinowski, J., Structure of Graphite Oxide Revisited. The Journal of Physical Chemistry B 1998, 102, 4477-4482.

24. Szabó, T.; Berkesi, O.; Forgó, P.; Josepovits, K.; Sanakis, Y.; Petridis, D.; Dékány, I., Evolution of Surface Functional Groups in a Series of Progressively Oxidized Graphite Oxides. Chemistry of Materials 2006, 18, 2740-2749.

25. Dreyer, D. R.; Todd, A. D.; Bielawski, C. W., Harnessing the chemistry of graphene oxide. Chemical Society Reviews 2014, 43, 5288-5301.

26. Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S., Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 2007, 45, 1558-1565.

27. Shin, H.-J.; Kim, K. K.; Benayad, A.; Yoon, S.-M.; Park, H. K.; Jung, I.-S.; Jin, M. H.; Jeong, H.-K.; Kim, J. M.; Choi, J.-Y.; Lee, Y. H., Efficient Reduction of Graphite Oxide by Sodium Borohydride and Its Effect on Electrical Conductance. Advanced Functional Materials 2009, 19, 1987-1992.

28. Wang, G.; Yang, J.; Park, J.; Gou, X.; Wang, B.; Liu, H.; Yao, J., Facile Synthesis and Characterization of Graphene Nanosheets. The Journal of Physical Chemistry C 2008, 112, 8192-8195.

29. Zhang, J.; Yang, H.; Shen, G.; Cheng, P.; Zhang, J.; Guo, S., Reduction of graphene oxide vial-ascorbic acid. Chemical Communications 2010, 46, 1112-1114.

30. Li, D.; Muller, M. B.; Gilje, S.; Kaner, R. B.; Wallace, G. G., Processable aqueous dispersions of graphene nanosheets. Nat Nano 2008, 3, 101-105.

31. Blake, P.; Hill, E. W.; Castro Neto, A. H.; Novoselov, K. S.; Jiang, D.; Yang, R.; Booth, T. J.; Geim, A. K., Making graphene visible. Applied Physics Letters 2007, 91, 063124.

32. Nemes-Incze, P.; Osváth, Z.; Kamarás, K.; Biró, L. P., Anomalies in thickness measurements of graphene and few layer graphite crystals by tapping mode atomic force microscopy. Carbon 2008, 46, 1435-1442.

33. Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K., Raman Spectrum of Graphene and Graphene Layers. Physical Review Letters 2006, 97, 187401.

34. Pei, S.; Cheng, H.-M., The reduction of graphene oxide. Carbon 2012, 50, 3210- 3228.

35. Cameron, J. S.; Ashley, D. S.; Andrew, J. S.; Joseph, G. S.; Christopher, T. G., Accurate thickness measurement of graphene. Nanotechnology 2016, 27, 125704.

36. Kudin, K. N.; Ozbas, B.; Schniepp, H. C.; Prud'homme, R. K.; Aksay, I. A.; Car, R., Raman Spectra of Graphite Oxide and Functionalized Graphene Sheets. Nano Letters 2008, 8, 36-41.

37. Chua, C. K.; Sofer, Z.; Pumera, M., Graphite Oxides: Effects of Permanganate and Chlorate Oxidants on the Oxygen Composition. Chemistry – A European Journal 2012, 18, 13453-13459.

38. Yang, D.; Velamakanni, A.; Bozoklu, G.; Park, S.; Stoller, M.; Piner, R. D.; Stankovich, S.; Jung, I.; Field, D. A.; Ventrice Jr, C. A.; Ruoff, R. S., Chemical analysis of graphene oxide films after heat and chemical treatments by X-ray photoelectron and Micro-Raman spectroscopy. Carbon 2009, 47, 145-152.

39. Michio, K.; Hikaru, T.; Kazuto, H.; Shinsuke, M.; Chikako, O.; Asami, F.; Takaaki, T.; Yasumichi, M., Analysis of Reduced Graphene Oxides by X-ray Photoelectron Spectroscopy and Electrochemical Capacitance. Chemistry Letters 2013, 42, 924-926.

40. Ganguly, A.; Sharma, S.; Papakonstantinou, P.; Hamilton, J., Probing the Thermal Deoxygenation of Graphene Oxide Using High-Resolution In Situ X-ray-Based Spectroscopies. The Journal of Physical Chemistry C 2011, 115, 17009-17019.

41. Dreyer, D. R.; Park, S.; Bielawski, C. W.; Ruoff, R. S., The chemistry of graphene oxide. Chemical Society Reviews 2010, 39, 228-240.

42. Niyogi, S.; Bekyarova, E.; Itkis, M. E.; McWilliams, J. L.; Hamon, M. A.; Haddon, R. C., Solution Properties of Graphite and Graphene. Journal of the American Chemical Society 2006, 128, 7720-7721.

43. Liu, Z.-B.; Xu, Y.-F.; Zhang, X.-Y.; Zhang, X.-L.; Chen, Y.-S.; Tian, J.-G., Porphyrin and Fullerene Covalently Functionalized Graphene Hybrid Materials with Large Nonlinear Optical Properties. The Journal of Physical Chemistry B 2009, 113, 9681-9686.

44. Zhang, X.; Huang, Y.; Wang, Y.; Ma, Y.; Liu, Z.; Chen, Y., Synthesis and characterization of a graphene–C60 hybrid material. Carbon 2009, 47, 334-337.

45. Liu, Z.; Robinson, J. T.; Sun, X.; Dai, H., PEGylated Nanographene Oxide for Delivery of Water-Insoluble Cancer Drugs. Journal of the American Chemical Society 2008, 130, 10876-10877.

46. Veca, L. M.; Lu, F.; Meziani, M. J.; Cao, L.; Zhang, P.; Qi, G.; Qu, L.; Shrestha, M.; Sun, Y.-P., Polymer functionalization and solubilization of carbon nanosheets. Chemical Communications 2009, 2565-2567.

47. Shen, J.; Shi, M.; Yan, B.; Ma, H.; Li, N.; Hu, Y.; Ye, M., Covalent attaching protein to graphene oxide via diimide-activated amidation. Colloids and Surfaces B: Biointerfaces 2010, 81, 434-438.

48. Mohanty, N.; Berry, V., Graphene-Based Single-Bacterium Resolution Biodevice and DNA Transistor: Interfacing Graphene Derivatives with Nanoscale and Microscale Biocomponents. Nano Letters 2008, 8, 4469-4476.

49. Yang, H.; Shan, C.; Li, F.; Han, D.; Zhang, Q.; Niu, L., Covalent functionalization of polydisperse chemically-converted graphene sheets with amine-terminated ionic liquid. Chemical Communications 2009, 3880-3882.

50. Wang, S.; Chia, P.-J.; Chua, L.-L.; Zhao, L.-H.; Png, R.-Q.; Sivaramakrishnan, S.; Zhou, M.; Goh, R. G. S.; Friend, R. H.; Wee, A. T. S.; Ho, P. K. H., Band-like Transport in Surface-Functionalized Highly Solution-Processable Graphene Nanosheets. Advanced Materials 2008, 20, 3440-3446.

51. Zhang, B.; Chen, Y.; Xu, L.; Zeng, L.; He, Y.; Kang, E.-T.; Zhang, J., Growing poly(N-vinylcarbazole) from the surface of graphene oxide via RAFT polymerization. Journal of Polymer Science Part A: Polymer Chemistry 2011, 49, 2043-2050.

52. Hou, S.; Su, S.; Kasner, M. L.; Shah, P.; Patel, K.; Madarang, C. J., Formation of highly stable dispersions of silane-functionalized reduced graphene oxide. Chemical Physics Letters 2010, 501, 68-74.

53. Ou, X.; Jiang, L.; Chen, P.; Zhu, M.; Hu, W.; Liu, M.; Zhu, J.; Ju, H., Highly Stable Graphene-Based Multilayer Films Immobilized via Covalent Bonds and Their Applications in Organic Field-Effect Transistors. Advanced Functional Materials 2013, 23, 2422-2435.

54. Mohamadi, S.; Sharifi-Sanjani, N.; Mahdavi, H., Functionalization of Graphene Sheets via Chemically Grafting of PMMA Chains Through in-situ Polymerization. Journal of Macromolecular Science, Part A 2011, 48, 577-582.

55. Ou, J.; Wang, Y.; Wang, J.; Liu, S.; Li, Z.; Yang, S., Self-Assembly of Octadecyltrichlorosilane on Graphene Oxide and the Tribological Performances of the Resultant Film. The Journal of Physical Chemistry C 2011, 115, 10080-10086.

56. Bai, H.; Xu, Y.; Zhao, L.; Li, C.; Shi, G., Non-covalent functionalization of graphene sheets by sulfonated polyaniline. Chemical Communications 2009, 1667-1669.

57. Song, J.; Gao, H.; Zhu, G.; Cao, X.; Shi, X.; Wang, Y., The preparation and characterization of polycaprolactone/graphene oxide biocomposite nanofiber scaffolds and their application for directing cell behaviors. Carbon 2015, 95, 1039-1050.

58. Yousefi, N.; Gudarzi, M. M.; Zheng, Q.; Lin, X.; Shen, X.; Jia, J.; Sharif, F.; Kim, J.-K., Highly aligned, ultralarge-size reduced graphene oxide/polyurethane nanocomposites: Mechanical properties and moisture permeability. Composites Part A: Applied Science and Manufacturing 2013, 49, 42-50.

59. Lu, C.-H.; Yang, H.-H.; Zhu, C.-L.; Chen, X.; Chen, G.-N., A Graphene Platform for Sensing Biomolecules. Angewandte Chemie International Edition 2009, 48, 4785- 4787.

60. Yang, X.; Zhang, X.; Liu, Z.; Ma, Y.; Huang, Y.; Chen, Y., High-Efficiency Loading and Controlled Release of Doxorubicin Hydrochloride on Graphene Oxide. The Journal of Physical Chemistry C 2008, 112, 17554-17558.

61. Piner, R. D.; Zhu, J.; Xu, F.; Hong, S.; Mirkin, C. A., "Dip-Pen" Nanolithography. Science 1999, 283, 661-663.

62. Wu, C.-C.; Reinhoudt, D. N.; Otto, C.; Subramaniam, V.; Velders, A. H., Strategies for Patterning Biomolecules with Dip-Pen Nanolithography. Small 2011, 7, 989-1002.

63. Chung, S.-W.; Ginger, D. S.; Morales, M. W.; Zhang, Z.; Chandrasekhar, V.; Ratner, M. A.; Mirkin, C. A., Top-Down Meets Bottom-Up: Dip-Pen Nanolithography and DNA-Directed Assembly of Nanoscale Electrical Circuits. Small 2005, 1, 64-69.

64. Demers, L. M.; Ginger, D. S.; Park, S.-J.; Li, Z.; Chung, S.-W.; Mirkin, C. A., Direct Patterning of Modified Oligonucleotides on Metals and Insulators by Dip-Pen Nanolithography. Science 2002, 296, 1836-1838.

65. Lee, K.-B.; Lim, J.-H.; Mirkin, C. A., Protein Nanostructures Formed via Direct- Write Dip-Pen Nanolithography. Journal of the American Chemical Society 2003, 125, 5588-5589.

66. Lim, J.-H.; Ginger, D. S.; Lee, K.-B.; Heo, J.; Nam, J.-M.; Mirkin, C. A., Direct- Write Dip-Pen Nanolithography of Proteins on Modified Silicon Oxide Surfaces. Angewandte Chemie International Edition 2003, 42, 2309-2312.

67. Cho, Y.; Ivanisevic, A., TAT Peptide Immobilization on Gold Surfaces: A Comparison Study with a Thiolated Peptide and Alkylthiols Using AFM, XPS, and FT- IRRAS. The Journal of Physical Chemistry B 2005, 109, 6225-6232.

68. Jiang, H.; Stupp, S. I., Dip-Pen Patterning and Surface Assembly of Peptide Amphiphiles. Langmuir 2005, 21, 5242-5246.

69. Cheung, C. L.; Camarero, J. A.; Woods, B. W.; Lin, T.; Johnson, J. E.; De Yoreo, J. J., Fabrication of Assembled Virus Nanostructures on Templates of Chemoselective Linkers Formed by Scanning Probe Nanolithography. Journal of the American Chemical Society 2003, 125, 6848-6849.

70. Smith, J. C.; Lee, K.-B.; Wang, Q.; Finn, M. G.; Johnson, J. E.; Mrksich, M.; Mirkin, C. A., Nanopatterning the Chemospecific Immobilization of Cowpea Mosaic Virus Capsid. Nano Letters 2003, 3, 883-886.

71. Rozhok, S.; Shen, C. K. F.; Littler, P.-L. H.; Fan, Z.; Liu, C.; Mirkin, C. A.; Holz, R. C., Methods for Fabricating Microarrays of Motile Bacteria. Small 2005, 1, 445-451.

72. Salaita, K.; Wang, Y.; Mirkin, C. A., Applications of dip-pen nanolithography. Nat Nano 2007, 2, 145-155.

73. Lee, K.-B.; Kim, E.-Y.; Mirkin, C. A.; Wolinsky, S. M., The Use of Nanoarrays for Highly Sensitive and Selective Detection of Human Immunodeficiency Virus Type 1 in Plasma. Nano Letters 2004, 4, 1869-1872.

74. Lee, M.; Kang, D.-K.; Yang, H.-K.; Park, K.-H.; Choe, S. Y.; Kang, C.; Chang, S.-I.; Han, M. H.; Kang, I.-C., Protein nanoarray on Prolinker™ surface constructed by atomic force microscopy dip-pen nanolithography for analysis of protein interaction. PROTEOMICS 2006, 6, 1094-1103.

75. Sekula, S.; Fuchs, J.; Weg-Remers, S.; Nagel, P.; Schuppler, S.; Fragala, J.; Theilacker, N.; Franzreb, M.; Wingren, C.; Ellmark, P.; Borrebaeck, C. A. K.; Mirkin, C. A.; Fuchs, H.; Lenhert, S., Multiplexed Lipid Dip-Pen Nanolithography on Subcellular Scales for the Templating of Functional Proteins and Cell Culture. Small 2008, 4, 1785- 1793.

76. Sekula-Neuner, S.; Maier, J.; Oppong, E.; Cato, A. C. B.; Hirtz, M.; Fuchs, H., Allergen Arrays for Antibody Screening and Immune Cell Activation Profiling Generated by Parallel Lipid Dip-Pen Nanolithography. Small 2012, 8, 585-591.

77. Chaitanya, S.; Anuj, G.; Jean-Pierre, L.; Klaus, S., Electronic detection of dsDNA transition from helical to zipper conformation using graphene nanopores. Nanotechnology 2014, 25, 445105.

78. Park, S. J.; Kwon, O. S.; Lee, S. H.; Song, H. S.; Park, T. H.; Jang, J., Ultrasensitive Flexible Graphene Based Field-Effect Transistor (FET)-Type Bioelectronic Nose. Nano Letters 2012, 12, 5082-5090.

79. Junhua, W.; Jingjing, Q.; Li, L.; Liqiang, R.; Xianwen, Z.; Jharna, C.; Shiren, W., A reduced graphene oxide based electrochemical biosensor for tyrosine detection. Nanotechnology 2012, 23, 335707.

80. Hirtz, M.; Oikonomou, A.; Georgiou, T.; Fuchs, H.; Vijayaraghavan, A., Multiplexed biomimetic lipid membranes on graphene by dip-pen nanolithography. Nature Communications 2013, 4, 2591.

81. Hirtz, M.; Oikonomou, A.; Clark, N.; Kim, Y.-J.; Fuchs, H.; Vijayaraghavan, A., Self-limiting multiplexed assembly of lipid membranes on large-area graphene sensor arrays. Nanoscale 2016, 8, 15147-15151.

82. Tabar, V.; Studer, L., Pluripotent stem cells in regenerative medicine: challenges and recent progress. Nat Rev Genet 2014, 15, 82-92.

83. Characterization of human embryonic stem cell lines by the International Stem Cell Initiative. Nat Biotech 2007, 25, 803-816.

84. Lee, T.-J.; Park, S.; Bhang, S. H.; Yoon, J.-K.; Jo, I.; Jeong, G.-J.; Hong, B. H.; Kim, B.-S., Graphene enhances the cardiomyogenic differentiation of human embryonic stem cells. Biochemical and Biophysical Research Communications 2014, 452, 174-180.

85. Garcia-Alegria, E.; Iliut, M.; Stefanska, M.; Silva, C.; Heeg, S.; Kimber, S. J.; Kouskoff, V.; Lacaud, G.; Vijayaraghavan, A.; Batta, K., Graphene Oxide promotes embryonic stem cell differentiation to haematopoietic lineage. Scientific Reports 2016, 6, 25917.

86. Takahashi, K.; Yamanaka, S., Induction of Pluripotent Stem Cells from Mouse Embryonic and Adult Fibroblast Cultures by Defined Factors. Cell 2006, 126, 663-676.

87. Takahashi, K.; Tanabe, K.; Ohnuki, M.; Narita, M.; Ichisaka, T.; Tomoda, K.; Yamanaka, S., Induction of Pluripotent Stem Cells from Adult Human Fibroblasts by Defined Factors. Cell 2007, 131, 861-872.

88. Yoo, J.; Kim, J.; Baek, S.; Park, Y.; Im, H.; Kim, J., Cell reprogramming into the pluripotent state using graphene based substrates. Biomaterials 2014, 35, 8321-8329.

89. Chen, G. Y.; Pang, D. W. P.; Hwang, S. M.; Tuan, H. Y.; Hu, Y. C., A graphene- based platform for induced pluripotent stem cells culture and differentiation. Biomaterials 2012, 33, 418-427.

90. Raff, M., Adult Stem Cell Plasticity: Fact or Artifact? Annual Review of Cell and Developmental Biology 2003, 19, 1-22.

91. Pittenger, M. F.; Mackay, A. M.; Beck, S. C.; Jaiswal, R. K.; Douglas, R.; Mosca, J. D.; Moorman, M. A.; Simonetti, D. W.; Craig, S.; Marshak, D. R., Multilineage Potential of Adult Human Mesenchymal Stem Cells. Science 1999, 284, 143-147.

92. Friedenstein, A. J.; Chailakhyan, R. K.; Latsinik, N. V.; Panasyuk, A. F.; Keiliss- Borok, I. V., STROMAL CELLS RESPONSIBLE FOR TRANSFERRING THE MICROENVIRONMENT OF THE HEMOPOIETIC TISSUES: Cloning In Vitro and Retransplantation In Vivo. Transplantation 1974, 17, 331-340.

93. Dominici, M.; Le Blanc, K.; Mueller, I.; Slaper-Cortenbach, I.; Marini, F. C.; Krause, D. S.; Deans, R. J.; Keating, A.; Prockop, D. J.; Horwitz, E. M., Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy 2006, 8, 315-317.

94. Dezawa, M.; Kanno, H.; Hoshino, M.; Cho, H.; Matsumoto, N.; Itokazu, Y.; Tajima, N.; Yamada, H.; Sawada, H.; Ishikawa, H.; Mimura, T.; Kitada, M.; Suzuki, Y.; Ide, C., Specific induction of neuronal cells from bone marrow stromal cells and application for autologous transplantation. The Journal of Clinical Investigation 113, 1701-1710.

95. Kuroda Y, K. M., Wakao S, Dezawa M Mesenchymal Stem Cells and Umbilical Cord as Sources for Schwann Cell Differentiation: their Potential in Peripheral Nerve Repair. The Open Tissue Engineering and Regenerative Medicine Journal, 2011, 54-63.

96. Zuk, P. A.; Zhu, M.; Ashjian, P.; De Ugarte, D. A.; Huang, J. I.; Mizuno, H.; Alfonso, Z. C.; Fraser, J. K.; Benhaim, P.; Hedrick, M. H., Human Adipose Tissue Is a Source of Multipotent Stem Cells. Molecular Biology of the Cell 2002, 13, 4279-4295.

97. Philips, B. J.; Marra, K. G.; Rubin, J. P., Healing of grafted adipose tissue: Current clinical applications of adipose-derived stem cells for breast and face reconstruction. Wound Repair and Regeneration 2014, 22, 11-13.

98. Strem, B. M.; Hicok, K. C.; Zhu, M.; Wulur, I.; Alfonso, Z.; Schreiber, R. E.; Fraser, J. K.; Hedrick, M. H., Multipotential differentiation of adipose tissue-derived stem cells. The Keio Journal of Medicine 2005, 54, 132-141.

99. Kingham, P. J.; Kalbermatten, D. F.; Mahay, D.; Armstrong, S. J.; Wiberg, M.; Terenghi, G., Adipose-derived stem cells differentiate into a Schwann cell phenotype and promote neurite outgrowth in vitro. Experimental Neurology 2007, 207, 267-274.

100. Faroni, A.; Smith, R. J. P.; Lu, L.; Reid, A. J., Human Schwann-like cells derived from adipose-derived mesenchymal stem cells rapidly de-differentiate in the absence of stimulating medium. European Journal of Neuroscience 2016, 43, 417-430. 58

101. Nayak, T. R.; Andersen, H.; Makam, V. S.; Khaw, C.; Bae, S.; Xu, X.; Ee, P.-L. R.; Ahn, J.-H.; Hong, B. H.; Pastorin, G.; Özyilmaz, B., Graphene for Controlled and Accelerated Osteogenic Differentiation of Human Mesenchymal Stem Cells. ACS Nano 2011, 5, 4670-4678.

102. Kim, J.; Choi, K. S.; Kim, Y.; Lim, K.-T.; Seonwoo, H.; Park, Y.; Kim, D.-H.; Choung, P.-H.; Cho, C.-S.; Kim, S. Y.; Choung, Y.-H.; Chung, J. H., Bioactive effects of graphene oxide cell culture substratum on structure and function of human adipose- derived stem cells. Journal of Biomedical Materials Research Part A 2013, 101, 3520- 3530.

103. Lee, W. C.; Lim, C. H. Y. X.; Shi, H.; Tang, L. A. L.; Wang, Y.; Lim, C. T.; Loh, K. P., Origin of Enhanced Stem Cell Growth and Differentiation on Graphene and Graphene Oxide. ACS Nano 2011, 5, 7334-7341.

104. Akhavan, O.; Ghaderi, E.; Shahsavar, M., Graphene nanogrids for selective and fast osteogenic differentiation of human mesenchymal stem cells. Carbon 2013, 59, 200- 211.

105. Qi, W.; Yuan, W.; Yan, J.; Wang, H., Growth and accelerated differentiation of mesenchymal stem cells on graphene oxide/poly-l-lysine composite films. Journal of Materials Chemistry B 2014, 2, 5461-5467.

106. Li, N.; Zhang, X.; Song, Q.; Su, R.; Zhang, Q.; Kong, T.; Liu, L.; Jin, G.; Tang, M.; Cheng, G., The promotion of neurite sprouting and outgrowth of mouse hippocampal cells in culture by graphene substrates. Biomaterials 2011, 32, 9374-9382.

107. Tang, M.; Song, Q.; Li, N.; Jiang, Z.; Huang, R.; Cheng, G., Enhancement of electrical signaling in neural networks on graphene films. Biomaterials 2013, 34, 6402- 6411.

108. Tu, Q.; Pang, L.; Chen, Y.; Zhang, Y.; Zhang, R.; Lu, B.; Wang, J., Effects of surface charges of graphene oxide on neuronal outgrowth and branching. Analyst 2014, 139, 105-115.

109. Tu, Q.; Pang, L.; Wang, L.; Zhang, Y.; Zhang, R.; Wang, J., Biomimetic Choline- Like Graphene Oxide Composites for Neurite Sprouting and Outgrowth. ACS Applied Materials & Interfaces 2013, 5, 13188-13197.

110. Rattana; Chaiyakun, S.; Witit-anun, N.; Nuntawong, N.; Chindaudom, P.; Oaew, S.; Kedkeaw, C.; Limsuwan, P., Preparation and characterization of graphene oxide nanosheets. Procedia Engineering 2012, 32, 759-764.

Chapter 2: Methodology

2.1 GO synthesis and purification

Graphite flakes were prepared by using a modified Hummer’s method [1]. Briefly, 3 grams of graphite powder was dissolved in a 9:1 mixture of 360 mL of concentrated sulphuric acid (H2SO4) and 40 mL of phosphoric acid (H3PO4). 18 grams of potassium permanganate (KMnO4), a strong oxidising agent, were added in the reaction mix contained in a glass beaker. The reaction is ten heated up to reach a temperature of 50 C under magnetic stirring for 12 hours. After 12 hours, the reaction is cooled to room temperature and poured into an iced solution (400 mL) of 30% of hydrogen peroxide

(H2O2). The mixture is then initially sieved through a metal U.S. Standard testing sieve (W.S. Tyler, 300 m) and then finally filtered through polyester fiber (Carpenter Co.). The filtrate is poured in 250 mL plastic tube and centrifuged in a Sidewall Legend XTR centrifuge machine (Thermo Scientific) using a F14-6x250 (Thermo Scientific) rotor at the speed of 2451 relative centrifugal forces at maximum radius (RCF Rmax) for 40 minutes. After the centrifugation step, the supernatant is discarded and the solid pellet is resuspended in the washing buffer made of 200 mL of water (H2O), 30% of concentrated hydrochloric acid (HCl), 200 mL of ethanol. After the pellet is resuspended, it is sieved and filtered as described before and then the filtrate is centrifuged at 2451 RCF Rmax. The supernatant is discarded and the pellet is resuspended for 18 cycles of washing. The pellet is purified extensively in this way and this is very important for graphene oxide based biomedical applications. Biomedical applications requires high purity GO therefore extensive purification is the only way researcher can be sure of eliminating toxic contaminants needed in the synthetic protocols (concentrated acids for instance). After the purification steps, graphite oxide is then exfoliated to monolayer GO by a combination of 1 hour of ultrasonication and vigorous magnetic stirring. Longer ultrasonication cycle tends to decrease the lateral size of GO producing smaller GO flakes.

2.2 Biological functionalization of GO

100 mg of GO is reacted with 100 mM of chloroacetic acid and 250 mM of sodium hydroxide (NaOH) under magnetic stirrer for 3 hours. The mixed reaction is then purified by centrifugation at 15317 RCF Rmax for one hour using the same centrifuge setup and plastic tube reported in paragraph 2.1. The centrifugation step is repeated three times. The solid pellet is then suspended in 100 mL of deionized water. This treatment opens the

62 epoxy rings in GO and converts hydroxyl group into carboxyl groups. Increasing carboxyl groups is important as it allows increasing binding sites to attach biomolecules.

The esterification reaction followed a published procedure [2]. GO is reacted with 50 mL of N-hydroxysuccinimide (NHS) at the concentration of 50 mg/mL solubilized in 50mL of 2-N-morpholino-ethanesulfonic acid (MES) buffer at the concentration of 500mM (pH 6.1) for 15 minutes under magnetic stirring at room temperature. Then 60 mL of 1-Ethyl- 3-(3-dimethylaminopropyl)carbodiimide (EDC) at the concentration of 10 mg/ml is added to the reaction mix for 30 minutes under stirring at room temperature. Esterified GO is then purified by centrifugating at 18533 RCF Rmax for 60 minutes. The centrifugation is repeated three times. Esterification of GO is required as carboxylic groups are less reactive compared to esters groups.

After the last purification step, 50 mg of GO-ester is dissolved in 50 mL of the MES buffer (pH 6.1) and 10 mL at the concentration of 10 mg/ml of isoleucine-lysine-valine- alanine-valine (IKVAV) peptide was reacted for 60 minutes under magnetic stirrer at room temperature. IKVAV peptide was chosen as it is active peptide domain of laminin, a protein of the extracellular matrix. This peptide has been reported to have a role in improving neurite outgrowth and neural attachment. [34]

The product is then purified by centrifugation at 18533 RCF Rmax for one hour. The centrifugation step is repeated three times. Finally, the pellet was resuspended in 50 mL of deionized water.

Another functionalization strategy is performed in thesis following a previously published protocol [3]. Carboxylation of GO with NaOH and chloracetic acid is performed by reacting 100 mg of GO with 0.5M, 1M and 3M of chloroacetic acid and 4M of NaOH. The reason to try different concentrations of chloracetic acid is to verify which concentration of acid yields the highest amount of carboxylation.

Then 0.7 mg/mL carboxylated GO is spun down on the surface of 290 nm SiO2/Si chips by spin coating at the speed of 2,500 rpm, at acceleration of 250 rpm/sec for 2 minutes. The chips were incubated in a solution of 2mM EDC/5mM Sulfo/NHS in MES Buffer (500 mM , pH 6.0). The incubation lasted 15 minutes and afterwards the chips were washed with 500 mM MES buffer twice. After the washing steps, the chips were incubated with a solution of 11,3 mM of Nα,Nα-Bis(carboxymethyl)-L-lysine hydrate

(NTA-NH2) in 500 mM phosphate buffered saline (PBS, pH 7.5) for 2 hours. After 2 63 hours, the chips were washed with deionized water twice. Then, the chips were incubated and placed into a solution of 11.3 mM NiCl2 for 30 minutes and then washed twice with deionized water.

Defrosted His-tagged Orco-DMPG-Nanodisc were diluted 1:100 with 30 mM HEPES buffer, pH 7.3. A solution of 0.5 mL Orco-DMPG-Nanodisc was pipetted on each individual chip at room temperature for 30 minutes. After the reaction, the chips were washed with deionized water twice and blown dry using nitrogen gun for further characterization.

2.3 Preparation of GO-based substrates

Once the GO and functionalized GO dispersions were produced, the fabrication of the GO-based substrates for further chemical and biological characterization. Furthermore, even the reduction of GO is performed directly on the substrates. Thermal reduction on the substrates is preferred as reduction of GO in solution decreases water dispersability making more difficult the production of the coatings.

Glass and 290 nm silicon oxide substrates were used in this thesis. Glass coverslips were used as substrates for GO coatings in the biological experiments and for AFM and Raman characterization. Silicon oxide wafers were used as substrates for GO coatings in XPS characterization. Substrates are initially cleaned chemically and then by oxygen-plasma before the coating takes place.

Glass substrates were put in a glass beaker filled with Decon 90 for 15 minutes in a bath sonicator. The Decon 90 solution is then discarded and the glass substrates are immersed in deionized water for another cycle of sonication for 15 minutes. Deionized water is then discarded and glass substrates are immersed in isopropanol and another cycle of sonication was applied for 15 minutes. Silicon oxide wafers were cleaned in a bath sonicator in a similar manner. Firstly, the silicon oxide substrates were put in a glass beaker filled with acetone and sonicated for 10 minutes in a sonicator bath. Acetone solution is then discarded and the silicon oxide substrates were immersed in deionized water for a cycle of sonication for 15 minutes. Deionized water is then discarded, and silicon oxide substrates were lastly immersed in isopropanol for a cycle of sonication which lasted 15 minutes.

The substrates are then dried using a nitrogen gun. After being dried, substrates were treated with oxygen plasma. . Oxygen plasma is an efficient technique to fully eliminate contaminants from the surface. Plasma is formed by the use of gaseous species at low 64 pressure. Plasma generates ultra-violet energy, which is strong enough to cause the breaking of many covalent bonds, therefore eliminating any traces of contaminants from the surfaces [4]. Oxygen-based plasma produces the formation of oxygen species on the surface of the substrates, therefore increasing the wettability and the hydrophilicity of the substrates. [5]

Both glass and silicon oxide substrates were put in a Henniker Plasma HPT-100 plasma machine for 5 mins at 100 watts (full power) using air as gas supplier. After the oxygen- plasma treatment, spin-coating technique was employed to thesis to coat GO on glass and silicon oxide substrates. Inside the spin-coater, there is a sample holder where the substrate is hold under vacuum. The GO dispersions at different concentrations were deposited adjusting the volume to cover the whole surface of the substrate. Then there is an automated controller, which allows controlling the rotational speed of the rotor, the acceleration and the amount of the spinning time.

The basic principle of this technique is quite simple: when the rotor is spinning, the centrifugal forces will spread the liquid dispersions on the substrates. Spin coating allows very flat coatings with controlled thickness. The thickness of the film will depend on the rotational speed selected [6]. GO coatings were formed by using the spin coater WS-650- 23 (Laurell Technologies) with the following parameters: selected rotational speed was 2,500 rpm, and the acceleration was 250 rpm/sec for 2 minutes. Two different concentrations were used in this thesis: the lowest concentration was 0.7 mg/mL and the highest concentration was 2 mg/mL. The lower concentration dispersion produced coatings characterized by the presence of single GO flakes. The higher concentration dispersion produced coatings characterized by uniform thin films.

Thermal reduction of GO occurred directly on the substrates. GO substrates were prepared as explained previously. GO-coated substrates were placed inside a vacuum oven (Townson-Mercer EV 018 model) at the temperature of 180 °C for three days in vacuum. The combination of low-pressure and high temperature enables the detachment of the oxygen functional groups allowing the restoration of the graphitic basal plane on the surface of the substrate.

2.4 Substrate Characterization

2.4.1 Atomic Force Microscopy

AFM is a scanning-probe microscopy technique developed by IBM researchers in 1986 [7]. This technique allows to image samples at the nanoscale. Briefly, the image is created

65 by the measurement of forces that are created when a sharp tip is kept at the distance of 0.2-10 nm from the surface of the substrate. The tip raster scans the surface along the x and y directions. To create the image, the cantilever detection is measured by the feedback loop in the instrument. Different heights will cause different deflections of the cantilever and they are sensed by the AFM which will recreate a false colour image where every x and y grid forms a pixel and where each colour represents the recorded signal. [8,9]

In this thesis, tapping-mode AFM in air was performed. Tapping mode is an operating mode where the tip is not in direct contact with the substrates but it is oscillating at its resonant frequency. The amplitude of the tip oscillation is usually about tens of nanometers. Tapping mode AFM is better than other AFM operation modes to image soft matter substrates such as biological samples. 20 m x 20 m area were imaged using a Bruker FastScan Dimension AFM microscope. The operation mode was tapping mode in air with an amplitude set point of 11 nanometers, and a scanning frequency of 1 kHz. The images were acquired with a resolution 512 x 512 points.

To analyse the images, Nanoscope software was employed. First, the image is flattened so the surface of the substrate is considered the reference point. Then, it is possible to measure the thickness and the lateral sizes of the flakes.

2.4.2 Raman Spectroscopy

Raman spectroscopy operates by the inelastic scattering originated from a monochromatic light in either visible, infrared and ultraviolet wavelength range. The laser energy cause molecular vibrations on the substrates. These vibrations interact with the phonons generated in the system causing upshift or downshift in the laser energy [10,11]. In this thesis, the wavelength of the laser was selected to be 532 nanometers.

Figure 2.2: Schematics of Raman microscopy [10]

The power of the laser was below 1mW to avoid laser-induced defects on the substrates.

A 100% objective lens was used in the Raman characterization with 1800 grooves/mm.

Raman mapping measurements of 150 spectra were selected in order to investigate a larger area of the substrates. At the end of the measurements, Microcal Origin Pro 8.5 was used in the data processing. The Raman mapping data were averaged to use a representative spectrum of the measurement. Then, the baseline of the Raman spectrum was subtracted and ultimately a full Lorentzian function was used to fit the Raman peaks present in the spectrum. By fitting the G and the D peaks, it was possible to measure the full width at half maximum (FWHM), the height of the peak and the Raman shift position. In GO spectrum it is very important to measure the ratio between the intensities of the D and the G peak (ID/IG). The value of the ratio is obtained by dividing the heights of the two peaks.

2.4.3 X-Ray Photoelectron and Fourier Transform Infrared Spectroscopies

XPS is a spectroscopy technique where the sample is bombarded with monoergetic low- energy X-rays in vacuum (typically ~ 10-10 torr). The X-rays when they hit an electron in the sample cause the ejection of a photoelectron from the substrate which is analysed

67 from the detector. Different atoms eject photoelectrons at different kinetic energies therefore, the value of the kinetic energy is used to discriminate the atomic composition and the nature of functional groups present in the sample. Measurement in vacuum is required to maximize the transmission of the photo-emitted electron from the substrate to the analyser of the instrument and also to minimize the risk of contamination of the surface during the measurement. During the measurement, the substrates are irradiated with a monochromatic Al Kα source at low energy (1486.6 eV).

This technique is very sensitive as the photoelectron attenuation length ranges from 3 to 5 nanometers. Spin-coating technique was not employed to prepare substrates as the obtained thickness of the GO-based thin films was similar to the value of the attenuation length. To avoid measuring any oxygen signal arising from the glass or silicon oxide substrates, sample preparation was performed by drop-casting which allows the formation of much thicker films on the substrate. Specifically, GO and functionalized GO dispersions were drop-casted on top of silicon oxide wafers and allowed to dry in vacuum overnight. Once dried, they were kept in the desiccator until the XPS measurement. rGO samples were prepared by drop-casting GO dispersions on top of silicon oxide wafers and left to reduce in the vacuum oven following the same parameters explained in the paragraph 2.3. Once the reduction is performed, rGO substrates were kept in the desiccator until the XPS measurement.

XPS wide scan, C1s and N1s spectra were acquired to characterize the thin films used in this thesis. Wide scan spectrum is important in GO characterization as it measures the atomic ratio between carbon and oxygen on the substrates. C1s spectrum is important to analyse how the carbon atoms are bound in the structure of GO.

CasaXPS software was used to analyse the XPS spectra obtained. This software has a library tool which helps to assign to each atom the right peak in wide scan spectra. This software measures the area, the FWHM and the atomic percentage of each peak. By dividing, the atomic percentage of carbon and oxygen atom it is possible to measure the C/O ratio.

In C1s spectra, it is important to deconvolute the carbon peak into the individual component: to do so, the FHWM of the components was constrained between 0.2 and 0.9 eV and the position of the component was constrained between 284.6 and 292 eV. The software measures the atomic percentage of each component. To fit the peak, a Gaussian- Lorentzian product form (GL30) was used.

Attenuated total reflection (ATR)-FTIR techniques involves the deposition of the samples on top of an ATR crystal. When an infrared beam reflects from the surface of the crystal an evanescent wave is formed. The evanescent wave then is projected into the sample which is placed on top of the ATR crystal allowing the measurement of the FTIR spectrum. Inside the machine, a moving mirror changes the amount of infrared light passed through the interferometer. The recorded signal is the output light expressed as function of the position of the moving mirror. A processing technique called Fourier Transform will process the raw data (called also interferogram) into the desired spectrum expressed as function of infrared wavenumber. Sample preparation for FT-IR was performed by avoiding the presence of any solvent during the measurement. Dispersions of GO, GO-IKVAV and IKVAV peptide alone were deposited covering the whole surface of three glass Petri dishes. The Petri dishes were put under vacuum to dry in the vacuum oven overnight until the different powders were formed. Powders were analysed by FT- IR Avatar 360 ESP spectrometer in the range 500–2500 cm−1 in ATR mode and three scans were acquired for each samples. The spectra is then analysed using Microcal Origin Pro 8.5, which is used to locate and to properly assign the peaks to the right chemical covalent bonds. [13]

2.5 Biological Characterization of the Substrates

2.5.1 Human Adipose Stem Cells Harvesting and Differentiation into dASC

Human adipose stem cells were collected by liposuction on consenting patients undergoing reconstructive surgery at University Hospital South Manchester, UK. ASC were isolated according to a previously reported protocol [14]. Adipose tissue were minced by a razor blade and dissociated by an enzymatic treatment of 0,2 % (w/v) of collagenase (Life Technologies, Paisley, UK) for 60 minutes at 37°. The digested tissue was then filtered through a vacuum-assisted 100 µm nylon mesh (Merck Millipore, Watford, UK). An equal volume of stem cell growth medium containing a-minimum essential Eagle’s medium (aMEM) (Sigma-Aldrich, Poole,UK), 10% (v/v) fetal bovine serum (FBS) (LabTech, Uckfield, UK), 2 mM L-glutamine (GE Healthcare UK, Little Chalfont, UK), and 1% v/v) penicillin–streptomycin was added. The solution were centrifuged at 300 g for 10 minutes and the resulting pellet was suspended in 1 mL of Red Blood Cell Lysis Buffer (Sigma-Aldrich) for 1 min, and 20 mL of aMEM was added to arrest lysis. The mixture was centrifuged at 300 g for 10 min, and the resulting pellet was resuspended in aMEM and plated in T75 flasks for cell culture.

Figure 2.3: Fluorescent microscope image of hASC cells grown on GO substrates after Live/Dead assay. Live cells are stained in green while dead cells are stained in red.

To differentiate ASC into dASC, this protocol was followed [15} : ASC at the passage 2 at 30% of confluence were treated with 1 mM β-mercaptoethanol (sigma-Aldrich) for 24 hours, then with 35 ng/mL of all-trans-retinoic acid for 72 hours. After this initial treatment, ASC were treated by 5 ng/mL if platelet-derived growth factor (Peprotech EC, London,UK) , 10 ng/mL basic fibroblast growth factor (Peprotech EC), 14 mM of forskolin (Sigma-Aldrich) and 192 ng/mL glial growth factor (GGF-2) (Acorda Therapeutics, Ardsley,NY, USA). ASC were kept under these conditions for 2 weeks and after 72 hours the cell medium was refreshed.

2.5.2 MTS assays Tetrazolium reduction assays were employed to measure the proliferation rate of both ASC and dASC on the GO-based substrates. In this thesis, 3-(4,5-dimethylthiazol-2-yl)- 5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H tetrazolium (MTS) –based assays were employed. MTS is added in the cell culture media with the presence of phenazine methosulfate (PMS). MTS reduction in living cells produces a soluble formazan crystal with an absorbance peak at the wavelength of 490 nanometers. [16,17]

ASC and dASC cells were plated at the concentration of 50,000 cells per coverslip in triplicate on the sterilized functionalized coverslip. The coverslips were put in a 24-well cell plate under ultra-low attachment condition (Corning Costar®). At days 1, 4 and 7 the cell medium was aspirated and cells were washed in PBS. After the washing step, the cells were incubated in 20% (v/v) CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Southampton, UK), diluted in phenol-free DMEM (Sigma-Aldrich) for

90 minutes in the dark at 37°. After the incubation, the absorbance at 490 nm was recorded using an Asys UVM-340 microplate reader/spectrophotometer (Biochrom, Cambridge, UK).

2.5.3 Live/Dead Assay LIVE/DEAD™ assay is a fluorescent-based assay, which selectively quantify living cells and dead cells on the substrates. This assays is based on the use of two fluorescent markers such as calcein AM and ethidium homodimer-1. (Eth-D1). [18,19]

Calcein AM contains acetomethoxy esters group, which are cleaved by intracellular esterase. After the action of this group of enzymes, calcein binds to intracellular deposit emitting green fluorescence. This stains is selective as dead cells cannot cleave acetomethoxy ester groups and therefore there is not green fluorescence emission. Eth- D1 only binds to disrupted plasma membranes emitting red fluorescence. It is therefore a selective reporter of dead cells. ASC and dASC cells were plated using a 24-well ultra- low attachment cell plate (Corning Costar®) at the concentration of 25,000 cells per coverslip in triplicate in a cell medium containing α-MEM and 1% (v/v) P/S without growth factors. After 48 hours, the medium was aspirated and the cells were washed in PBS. Calcein-AM fluorescein and ethidium homodimer-1 (Eth-D1) purchased from LIVE/DEAD® Viability/Citoxicity Kit, Molecular Probes, Invitrogen, UK) were added at the concentration of 0.5 and 2 μg/ml respectively in PBS and left to react for 30 minutes at room temperature. After this step, a final wash in PBS was performed and images were taken using a fluorescence-inverted microscope (Olympus IX51, Japan) under 4% magnification. Percentage of Live Cells is measured dividing the average green stained area by the average of the whole (Live + Dead) stained area multiplied by 100. Percentage of Dead Cells is measured by dviding the average red stained area by the average of the whole (Live + Dead) stained area multiplied by 100.

2.5.4 RNA extraction and gene expression studies To study the gene expression of both ASC and dASC cells, the RNA was extracted after 48 hours of cellular growth on the different coverslips at the concentration of 50,000 cells x coverslip. The cells were removed from the substrates by treating with trypsin/EDTA and centrifuged for 5 mins at 900 rpm. The media was discarded and the cells were washed using PBS. Another centrifuge step with the same settings were performed. The media was discarded and the cellular pellet was used to extract the RNA by using the RNeasy Plus Mini Kit (Qiagen). The cellular pellet was suspended in RLT (Qiagen) lysis buffer enriched with 1% (v/v) β- mercaptoethanol (Sigma-Aldrich, UK). 71

After, the suspension was centrifuged at 10,000 g for 2 minutes in QIAshredder spin columns (QIAGEN, UK). The supernatant is mixed with 70% ethanol (v/v) and placed inside RNeasy mini column tubes at the volume of 1.5 mL and centrifuged at 10,000x g for 15 seconds. The RNA and DNA are retained inside the column tubes after the centrifuge. The column tubes are treated with DNase I solution (QIAGEN, UK) for 15 minutes at room temperature to degrade the DNA. The column tubes are then washed with RW1 and RPE (Qiagen) buffers as per manufacturer’s protocol, the RNA was eluted with RNAse-free water into a fresh RNAse- free 1.5ml Eppendorf tube. The concentration of RNA was measured with NanoDrop® ND-1000 UV-Vis Spectrophotometer (Labtech International Ltd., UK). The quality of RNA can be checked by measuring the ratio between the absorbance at 260 nanometers by the absorbance measured at 280

Figure 2.3: Fluorescent microscope image of hASC cells grown on GO substrates after Live/Dead assay. Live cells are stained in green while dead cells are stained in red. nanometers. When there are no protein contaminations in the sample, the absorbance ratio expected is 2.0. [20]

1 µg of each RNA sample were reverse transcribed using the RT2 First Strand Kit (Qiagen) following the instruction of the manufacturer. DNA elimination steps were included in either RNA extraction and cDNA synthesis to get rid of downstream genomic DNA amplification. qRT-PCR was performed with RT2 SYBR Green qPCR Mastermix (Qiagen) and a Corbett Rotor Gene 6000 Qiagen), by the use of the following protocol: hot start for 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C, annealing for 30 s at 55 °C, and extension for 30 s at 72 °C. To verify the specificity of the reactions, a melting curve was obtained with the following protocol: 95 °C for 1 min, 65 °C for 2 min, and a gradual temperature increase from 65 °C to 95 °C (2 °C/min)

Fold changes were measured by the comparative Ct method [21]. Ct can be defined as the PCR cycle at which the fluorescence signal of fluorescent reporter SYBR green oversteps an arbitrary placed threshold. In the Ct method, firstly the difference between the Ct of the gene of interest and the Ct housekeeping gene is calculated. This difference is calculated for each group studied. In this experiment, the selected housekeeping gene is S18, which is a gene stably expressed. Then the difference between the Ct values of the experimental groups and the control group is measured. The fold change in the expression of the gene of interest can be expressed as equal to 2-ΔΔCT. The term 2-ΔΔCT is equal to:

(퐶푡 표푓 푔푒푛푒 표푓 𝑖푛푡푒푟푒푠푡 − 퐶푡 표푓 푡ℎ푒 ℎ표푢푠푒푘푒푒푝𝑖푛푔 푔푒푛푒)푠푎푚푝푙푒 1 − (퐶푡 표푓 푔푒푛푒 표푓 𝑖푛푡푒푟푒푠푡 − 퐶푡 표푓 푡ℎ푒 ℎ표푢푠푒푘푒푒푝𝑖푛푔 푔푒푛푒)푠푎푚푝푙푒 2 72 where sample 1 and sample 2 are substrates of the same groups. Glass substrates were used as control and all the measurements were in triplicate.

2.5.5 Dorsal root ganglia (DRG) harvesting and immunohistochemistry. Dorsal root ganglia neurons (DRG) were harvested from adult male Sprague–Dawley rats as previously reported in the literature [24,25]. To enzymatically dissociate the neurons, two treatments of 0.125% w/v collagenase type IV (Worthington Biochemical, UK) were performed for 1 h at 37°C. After the treatments, the neurons were washed in a 0.25% w/v trypsin (Worthington Biochemical) solution for 30 min at 37°C. After this step, the action of trypsin was blocked by another wash with 33% FBS solution. Neurons were lastly washed by F12 (Sigma-Aldrich) medium and then transferred to a Falcon tube to be mechanically dissociated by a glass pipette. The cell suspension obtained was filtered and centrifuged, then the resuspended pellet was slowly pipetted down a gradient tail created using F12 cell medium with the addition of a 50% v/v BSA. After a final centrifugation step, the pellet was resuspended in Bottenstein and Sato’s medium [BS; 1% v/v N2 supplement (PAA, UK) in F12 medium, with 50 ng/ml nerve growth factor (NGF; Millipore, USA) and DRG neurons were grown on the different coverslips for 72 hours.

Neurons were fixed using 4% w/v paraformaldehyde (PFA) solution for 20 mins at room temperature. The neurons were then carefully washed in PBS, permeabilized in 0.2% v/v Triton-X/PBS and washed again in PBS. Non-specific antigens on the coverslips surfaces were blocked with 1% w/v BSA in antibody diluent for 1 h at room temperature and then the primary antibody (anti-β tubulin III mouse monoclonal, 1:500) was allowed to react overnight at 4°. On the following day, the neurons were washed in PBS and then the secondary antibody solution (Goat Anti-mouse IgG-AlexaFluor 488, 1:500 in antibody diluent) is left to incubate for 60 minutes in the dark. After the incubation step, the neurons on the substrates were washed using a PBS solution before being mounting the samples on microscopy glass slides containing Vectashield DAPI staining. Using a fluorescence microscope (Olympus BX60), 20–25 random images were captured for neurite length and cell attachment quantification. The neurite lengths were manually evaluated using the freehand line tool of ImageJ software [National Institutes of Health (NIH), USA]. Results were expressed as average lengths ± SE of the mean (n=3). Cell attachment was quantified counting each cell with a manual cell counter. Results were expressed as average of the number of cells ± SE of the mean (n=3) [26,27]

Figure 2.4: Fluorescence microscopy image of DRG neurons on GO-IKVAV substrates after immunohistochemistry analysis with anti -tubulin 3 antibody.

2.6 CVD graphene array fabrication 2.6.1 CVD graphene transfers on silicon dioxide substrates

CVD graphene was transferred on top of a fresh silicon oxide wafer through a well- established method [28,29]. Briefly, a layer of PMMA was spun on one side of the CVD foil and baked at 130 °C for 5 mins. After, the foil was plasma etched on the other side in a mixture of oxygen (10%) and argon (90%) for 30 seconds to remove the second layer of graphene and copper on the back of the foil. Then the sample was placed in a solution of ammonium persulfate (APS) at the concentration of 25g /L overnight to etch away the copper foil. The sample is the fished out with a clean glass coverslip from the etchant solution and washed several times in deionized water. Then the graphene-PMMA membrane is transferred to a fresh Si/SiO2 wafer and left to dry. Then the PMMA is removed with several washes in acetone, DI water and hexane.

The outcome of the CVD transfers is judged by optical microscopy. Once the CVD graphene is transferred on top of silicon oxide wafers, the next step on the fabrication of the array is to perform photolithography.

2.6.2 Photolithography on CVD graphene substrates

Photolithography is a long process composed of three different parts: sample preparation,

lithography and development, and lastly the lift off of the resists.

Sample were prepared by coating the CVD graphene with three layers of polymers: the first and the second layers were PMMA and polymethylglutarimide (PMGI) respectively. These layers served to avoid direct contact of the CVD graphene with the photoresist. The third layer is the photoresist S1805 which is going to be directly exposed to the laser.

Briefly, a solution of PMMA (950 molecular weight, Microchem) 1.5% (w/v) in anisol was spin-coated was spun down at 7,000 rpm for one minute by spin coating. Then, the substrates were place on top of a hot plate at the temperature of 150 C for 5 minutes. It is important not to heat at higher temperatures than 150 C to avoid the cross-linking of the polymer chains. Then, PMGI is spun down at 3,000 rpm for one minute by spin coating. The substrates were then placed into a hot plate at the temperature of 170 ° C for 5 minutes. Lastly, the photoresist S1805 was spun down at 3,000 rpm for one minute by spin coating. The substrates were then placed into a hot plate at the temperature of 115 C for one minute. After the sample preparation, the lithography part takes place. Two designs were created by using the software Clewin 4. One design is composed by a series of square arrays. The size of the square in one design were 20 m x 20 m with a pitch of 66 m. In a second design, the size of the square ranges from 20 m to 40 m keeping the pitch constant at 66 μm. Optical lithography is performed by using Micro Tech’s Laser Writer LW405. After loading the design into the software of the instrument, a rastered focused laser beam exposed the photoresist causing molecular changes of the resist making it soluble to MF319 developer.

Figure 2.5: Optical Microscopy Image of transferred CVD-graphene on top of Silicon oxide wafer

Figure 2.6: Optical microscopy image of CVD graphene 20 micron x 20 micron squares array after photolithography

These parameters were followed for the exposure: lens 5 was selected as focusing lens. The D-step selected was 1, the laser gain 8.1 and the selected filter was 10 %. After photolithography, the sample was developed by immersing the chip in a solution of MF319 developer for 30 seconds. After 30 seconds, the chip is immersed into a solution of deionized water for one minute and then imaged under an optical microscope. Fig. is an optical microscopy image showing the sample at this stage of the process. The next step to perform is the elimination of the CVD graphene surface still present outside the square pattern.

Oxygen plasma is employed for this purpose. The substrates are treated for 3 minutes of plasma at 10-6 mBar pressure in an atmosphere of 90% argon and 10% oxygen. After, the system is vented by using nitrogen gas; the chip is imaged back with an optical microscope. Fig. is an optical microscopy image showing the aspect of the sample after the oxygen plasma treatment.

The last step in the array fabrication is the lift off. The lift off is based on solubilizing of the polymers employed in the sample preparation in a layer-by-layer manner. Firstly, the photoresist S18105 was solubilized by immersing the chip in a solution of acetone for one minute. Then after one minute, the chip is washed by immersing for one minute in a solution of deionized water. The PMGI layer is dissolved by immersing the chip in a

Figure 2.7: Optical microscopy image of CVD graphene 20 micron x 20 micron square array after oxygen plasma etching. solution of MF319 developer for another minute. The chip is then washed in deionized water for another minute. The PMMA layer was dissolved by immersing the substrates with acetone for 10 minutes; after the 10 minutes the acetone solution was refreshed and

Figure 2.8: Optical microscopy image of CVD graphene 20 micron x 20 micron square array after the lift off of the resist polymers

the substrates were immersed for another 10 minutes in acetone. After this step, again the acetone was refreshed and the substrates were immersed in fresh acetone overnight The morning after, the sample were immersed into fresh solution of hexane for 10 minutes and blown dry by using nitrogen gun and imaged by optical microscope. Fig. are the optical images of the CVD graphene arrays after the whole photolithography process.

2.7 Dip Pen Nanolithography Patterning and Membrane Protein Insertion

DPN was used to perform the lipid patterning to create biomimetic lipid membranes on the graphene surface on the square. Multiple 12- pen arrays were used to perform simultaneous lipid DPN. The pitch which means the physical distance between each pen is 66 m. The value of the pitch of the pen array is matched with the graphene square array. Different lipid inks were used: DOPC + 5 mol% 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rho-PE) at the flanking regions of the array to fluorescently monitor the formation of the biomimetic lipid membrane.

DOPC + 5 mol% 1,2-dioleoyl-sn-glycero-3-phosphate (DOPA), DOPC + 5 mol% 1,2- dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) and DOPC + 5 mol% 1,2- dioleoyl-sn-glycero-3-{[N(5-amino-1-carboxypentyl)iminodiacetic acid]succinyl} (nickel salt)) were used to pattern the central region of the array. It was chosen to use

Figure 2.9: Optical microscopy Image of CVD graphene square array in different sizes after the lift off of the resist polymers

78 different lipid formulations to check whether different charges affected the lipid-protein interaction. [31,32,33]

The DPN tips contacted the graphene array squares at the relative humidity of 40% allowing the ink flow from the DPN tips to the surface of the square flowing a square design loaded into the software.

After the lipid patterning, the samples were imaged under optical microscope and a Zeiss Axio Imager set-up equipped with SARFUS Software (Nanolane) fluorescent microscope with a rhodamine filter.

The chips were washed by pipetting 50 L of PBS and then imaged by Zeiss Axio Imager set-up equipped with SARFUS Software (Nanolane) fluorescent microscope as after wetting lipid biomimetic membranes became too thin to be imaged by optical microscopy. The chips were then prepared for the protein-binding assay. Briefly, the chips were blocked to avoid non specific protein binding. The surface was blocked by immersing the chips in a 10/ (v/v) BSA solution in PBS for one hour. After the blocking, the chips were washed by immersion in PBS for 10 minutes. The chips were washed three times. After the wash, the protein binding assay was performed. 40 L (at the

Figure 2.10: Optical microscopy image of the CVD graphene square in air after lipid patterning by DPN

79 concentration of 1.7 mg/mL) of recombinant protein Cytochrome B5-Synaptobrevin kindly provided by research group led by Professor Stephen High was pipetted in each chip and left to react for 3 hours.

The full sequence, expressed in amino acids of the protein is the following:

G S S A T A A T V P P A A P A G E G G P P A P P P N L T S N R R L Q Q T Q A Q V D E V V D I M R V N V D K V L E R D Q K L S E L D D R A D A L Q A G A S Q F E T S A A K L K R K Y W W K N L K W W T N W V I P A I C A V A V A L M Y R L Y M A E D S R M N G T E G P N F Y V P F S N K T V D where the red letters indicate the synaptobrevin sequence , the blue ones the cytochrome B5 sequence and the black ones indicate the opsin tag at the carboxy terminal (closed to the cytochrome B5 end) and the thrombin cleavage at the amino terminal (close to the synaptobrevin-2 end).

After the binding, the chips were washed three times in PBS. Each wash lasted 10 minutes.

Then, for each chip 5 g of monoclonal mouse anti-synaptobrevin-2 (Synaptic Systems, Germany) were dliuted in 50 L of PBS. Each chip was then incubated in the 50 L and then left to incubate overnight. Then on the next day, the chips were washed three times in PBS for 10 minutes each.

Then, an Alexa-488 labelled Anti-mouse secondary antibody 1:500 dilution in PBS was used for the immunostaining of the chips.

Each chip was immersed in 50 μL of the secondary antibody dilution and left one hour to react. After one hour the samples were washed as explained previously. The chips were then mounted on a microscope glass coverslips and imaged on a Zeiss Axio Imager set- up equipped with SARFUS Software (Nanolane).

References:

1. Hummers, W. S.; Offeman, R. E., Preparation of Graphitic Oxide. Journal of the American Chemical Society 1958, 80, 1339-1339.

2. Shen, J.; Shi, M.; Yan, B.; Ma, H.; Li, N.; Hu, Y.; Ye, M., Covalent attaching protein to graphene oxide via diimide-activated amidation. Colloids and Surfaces B: Biointerfaces 2010, 81, 434-438.

3. Lu, Y.; Lerner, M. B.; John Qi, Z.; Mitala, J. J.; Hsien Lim, J.; Discher, B. M.; Charlie Johnson, A. T., Graphene-protein bioelectronic devices with wavelength- dependent photoresponse. Applied Physics Letters 2012, 100, 033110.

4. Chai, J.; Lu, F.; Li, B.; Kwok, D. Y., Wettability Interpretation of Oxygen Plasma Modified Poly(methyl methacrylate). Langmuir 2004, 20, 10919-10927.

5. Krüger, P.; Knes, R.; Friedrich, J., Surface cleaning by plasma-enhanced desorption of contaminants (PEDC). Surface and Coatings Technology 1999, 112, 240- 244.

6. Taylor, J. F., Spin coating: An overview. Metal Finishing 2001, 99, 16-21.

7. Binnig, G.; Quate, C. F.; Gerber, C., Atomic Force Microscope. Physical Review Letters 1986, 56, 930-933.

8. Shan, Y.; Wang, H., The structure and function of cell membranes examined by atomic force microscopy and single-molecule force spectroscopy. Chemical Society Reviews 2015, 44, 3617-3638.

9. Jalili, N.; Laxminarayana, K., A review of atomic force microscopy imaging systems: application to molecular metrology and biological sciences. Mechatronics 2004, 14, 907-945.

10. Biswas, A.; Wang, T.; Biris, A. S., Single metal nanoparticle spectroscopy: optical characterization of individual nanosystems for biomedical applications. Nanoscale 2010, 2, 1560-1572.

11. Tolles, W. M.; Nibler, J. W.; McDonald, J. R.; Harvey, A. B., A Review of the Theory and Application of Coherent Anti-Stokes Raman Spectroscopy (CARS). Appl. Spectrosc. 1977, 31, 253-271.

12. Hollander, J. M.; Jolly, W. L., X-ray photoelectron spectroscopy. Accounts of Chemical Research 1970, 3, 193-200.

13. Gaffney, J. S.; Marley, N. A.; Jones, D. E., Fourier Transform Infrared (FTIR) Spectroscopy. In Characterization of Materials, John Wiley & Sons, Inc.: 2002.

14. Kingham, P. J.; Kalbermatten, D. F.; Mahay, D.; Armstrong, S. J.; Wiberg, M.; Terenghi, G., Adipose-derived stem cells differentiate into a Schwann cell phenotype and promote neurite outgrowth in vitro. Experimental Neurology 2007, 207, 267-274.

15. Faroni, A.; Rothwell, S. W.; Grolla, A. A.; Terenghi, G.; Magnaghi, V.; Verkhratsky, A., Differentiation of adipose-derived stem cells into Schwann cell phenotype induces expression of P2X receptors that control cell death. Cell Death Dis 2013, 4, e743.

16. Berridge, M. V.; Herst, P. M.; Tan, A. S., Tetrazolium dyes as tools in cell biology: New insights into their cellular reduction. In Biotechnology Annual Review, Elsevier: 2005; Vol. Volume 11, pp 127-152.

17. Cory, A. H.; Owen, T. C.; Barltrop, J. A.; Cory, J. G., Use of an aqueous soluble tetrazolium/formazan assay for cell growth assays in culture. Cancer Commun 1991, 3, 207-212.

18. Decherchi, P.; Cochard, P.; Gauthier, P., Dual staining assessment of Schwann cell viability within whole peripheral nerves using calcein-AM and ethidium homodimer. Journal of Neuroscience Methods 1997, 71, 205-213.

19. Haugland, R. P.; MacCoubrey, I. C.; Moore, P. L., Dual-fluorescence cell viability assay using ethidium homodimer and calcein AM. Google Patents: 1994.

20. Ian, B. M., Extraction of RNA From Cells and Tissue. In Hypertension: Methods and Protocols, Fennell, J. P.; Baker, A. H., Eds. Humana Press: Totowa, NJ, 2005; pp 139-148.

21. Schmittgen, T. D.; Livak, K. J., Analyzing real-time PCR data by the comparative CT method. Nat. Protocols 2008, 3, 1101-1108.

22. Faroni, A.; Smith, R. J. P.; Lu, L.; Reid, A. J., Human Schwann-like cells derived from adipose-derived mesenchymal stem cells rapidly de-differentiate in the absence of stimulating medium. European Journal of Neuroscience 2016, 43, 417-430.

23. Smith, C. A.; Richardson, S. M.; Eagle, M. J.; Rooney, P.; Board, T.; Hoyland, J. A., The use of a novel bone allograft wash process to generate a biocompatible, 82 mechanically stable and osteoinductive biological scaffold for use in bone tissue engineering. Journal of Tissue Engineering and Regenerative Medicine 2015, 9, 595-604.

24. Mobasseri, A.; Faroni, A.; Minogue, B. M.; Downes, S.; Terenghi, G.; Reid, A. J., Polymer Scaffolds with Preferential Parallel Grooves Enhance Nerve Regeneration. Tissue Engineering Part A 2014, 21, 1152-1162.

25. Malin, S. A.; Davis, B. M.; Molliver, D. C., Production of dissociated sensory neuron cultures and considerations for their use in studying neuronal function and plasticity. Nat. Protocols 2007, 2, 152-160.

26. K. Hancock, M.; Kopp, L.; Kaur, N.; Hanson, B. J., A Facile Method for Simultaneously Measuring Neuronal Cell Viability and Neurite Outgrowth. Current Chemical Genomics and Translational Medicine 2015, 9, 6-16.

27. Silva, G. A.; Czeisler, C.; Niece, K. L.; Beniash, E.; Harrington, D. A.; Kessler, J. A.; Stupp, S. I., Selective Differentiation of Neural Progenitor Cells by High-Epitope Density Nanofibers. Science 2004, 303, 1352-1355.

28. Suk, J. W.; Kitt, A.; Magnuson, C. W.; Hao, Y.; Ahmed, S.; An, J.; Swan, A. K.; Goldberg, B. B.; Ruoff, R. S., Transfer of CVD-Grown Monolayer Graphene onto Arbitrary Substrates. ACS Nano 2011, 5, 6916-6924.

29. Li, X.; Zhu, Y.; Cai, W.; Borysiak, M.; Han, B.; Chen, D.; Piner, R. D.; Colombo, L.; Ruoff, R. S., Transfer of Large-Area Graphene Films for High-Performance Transparent Conductive Electrodes. Nano Letters 2009, 9, 4359-4363.

30. Hirtz, M.; Oikonomou, A.; Clark, N.; Kim, Y.-J.; Fuchs, H.; Vijayaraghavan, A., Self-limiting multiplexed assembly of lipid membranes on large-area graphene sensor arrays. Nanoscale 2016, 8, 15147-15151.

31. Hirtz, M.; Oikonomou, A.; Georgiou, T.; Fuchs, H.; Vijayaraghavan, A., Multiplexed biomimetic lipid membranes on graphene by dip-pen nanolithography. Nature Communications 2013, 4, 2591.

32. Sekula, S.; Fuchs, J.; Weg-Remers, S.; Nagel, P.; Schuppler, S.; Fragala, J.; Theilacker, N.; Franzreb, M.; Wingren, C.; Ellmark, P.; Borrebaeck, C. A. K.; Mirkin, C. A.; Fuchs, H.; Lenhert, S., Multiplexed Lipid Dip-Pen Nanolithography on Subcellular Scales for the Templating of Functional Proteins and Cell Culture. Small 2008, 4, 1785- 1793.

33. Sekula-Neuner, S.; Maier, J.; Oppong, E.; Cato, A. C. B.; Hirtz, M.; Fuchs, H., Allergen Arrays for Antibody Screening and Immune Cell Activation Profiling Generated by Parallel Lipid Dip-Pen Nanolithography. Small 2012, 8, 585-591.

34. Tashiro, K.; Sephel, G. C.; Weeks, B.; Sasaki, M.; Martin, G. R.; Kleinman, H. K.; Yamada, Y., A synthetic peptide containing the IKVAV sequence from the A chain of laminin mediates cell attachment, migration, and neurite outgrowth. Journal of Biological Chemistry 1989, 264, 16174-16182.

Introduction to Chapter 3

In this chapter, the fabrication of reduced graphene oxide, graphene oxide and functionalized graphene oxide based substrates is performed. Graphene oxide and functionalized graphene oxide thin films were produced on glass coverslips by spin- coating. Thermal reduction of GO was performed in vacuum oven on the substrate. Characterization of the substrates was performed by Raman and X-Ray photoelectron and Fourier Transform Infrared spectroscopies, atomic force and optical microscopies.

Adipose-derived mesenchymal stem cells (ASCs) and rat dorsal root ganglia were grown on the substrates and the biocompatibility was tested by MTS and Live/Dead assays. The IKVAV functionalization of graphene oxide increase the differentiation of ASCs towards neuroglial phenotypes while graphene oxide and reduced graphene oxide directed ASCs toward osteogenic and chondrogenic differentiation. IKVAV functionalization has a positive effect also in increasing the neuronal attachment on the substrates and in increasing the average neurite length.

The impact of graphene oxide functionalization in directing stem cells differentiation is demonstrated in this chapter. This opens up the feasibility of graphene oxide as coating material for cell culture substrates due to its biocompatibility and the ease of functionalization compared to other materials used in biotechnology sector currently.

Authors Contribution

Maria Iliut prepared and purified graphene oxide dispersions. Andrea Francesco Verre performed the IKVAV functionalization and the thermal of graphene oxide.

Andrea Francesco Verre fabricated the substrates and performed Raman and atomic force microscopy characterization of the substrates. Andrea Francesco Verre made the sample preparation for the XPS and FTIR measurements, which was performed by Christopher Muryn and Lei Gao respectively. Andrea Francesco Verre and Claudio Silva analysed jointly the XPS and FTIR data. Alessandro Faroni harvested the adipose stem cells and the rat dorsal root ganglia neurons. Andrea Francesco Verre and Alessandro Faroni performed MTS, Live/Dead assays and quantitative Real-Time PCR and immunohistochemical experiments.

Adam Reid, Alessandro Faroni and Aravind Vijayaraghavan helped Andrea Francesco Verre in the interpretation of the results and in the writing of the manuscript. 85

Biochemical functionalization of graphene oxide for directing stem cell differentiation

Andrea Francesco Verre,1 Alessandro Faroni,2 Maria Iliut,1 Claudio Silva,1 Cristopher Muryn,3 Lei Gao,3 Adam J Reid2 and Aravind Vijayaraghavan1,*

1 School of Materials and National Graphene Institute, University of Manchester, Manchester M13 9PL, UK

2 School of Chemistry, University of Manchester, Manchester M13 9PL, UK

3 Faculty of Life Science, University of Manchester, Manchester M13 9PL, UK

* Corresponding author: [email protected]

Abstract

The regeneration of peripheral nerve tissue is crucial in the treatment of peripheral neuropathies. In recent years, there has been an increased focus on biomaterial scaffolds which can deliver stem cells, direct differentiation to the desired lineage and guide tissue regeneration at the site of injury. Here, we report on a graphene oxide (GO) scaffold which has been chemically functionalized to direct the differentiation of adipose stem cells towards the neuroglial lineage and maintain the viability of differentiated cells. A comparison is made between substrates of GO, reduced GO and IKVAV-functionalized GO in comparison to standard tissue culture glass. Our results show that the graphene substrates are highly biocompatible, and the IKVAV functionalized substrates are more effective in directing stem cell differentiation towards neuroglial phenotypes compared to glass and other graphene substrates. Furthermore, GO IKVAV substrates showed increased neuronal attachment and neurons grown on GO-IKVAV sprouted longer neurites. These results suggest that functionalized GO could provide a viable scaffold in regenerative therapies of peripheral nerve injuries and disorders.

1. Introduction

Peripheral neuropathies represents a serious burden for our societies as for instance twenty millions of Americans are affected each year by this kind of diseases [32]. There is lack of pharmacological treatment and still today the clinical outcome is poor [33].

Cellular therapies can be a viable alternative to stimulate peripheral nerve regeneration. Schwann cells (SCs) are cells involved in promoting nerve regeneration by releasing of neurotrophins and extracellular matrix (ECM) proteins. The limitations of SC-based cellular therapies are based on the evidence of slow expansion rate and the difficulty of harvesting. [1] Adipose stem cells (ASCs) represent a clinical alternative to SC. In fact, ASCs can be diferentiated towards a SC-like phenotype (dASCs). dASCs express glial markers and nerotrophins, produce myelin, induce neurite outgrowth both in vitro and in vivo studies [2,3,4,5,6,]. To build a successful regenerative therapeutic approach, it is crucial to deliver dASCs to the site of injury.

Graphene and related nanomaterials have been extensively studied in the recent years in biomedical research due to their unique combination of properties including high specific surface area, ease of functionalization, mechanical strength, biocompatibility and electrical and thermal conductivity. Graphene materials are promising candidates for a range of biomedical applications such as vectors for drug and gene delivery, scaffold materials for regenerative medicine and functional surfaces for biosensors [16]. Graphene and its oxidized derivative graphene oxide (GO) have been extensively studied in the recent years as substrates to support stem cell growth and influence their differentiation. Both graphene and GO substrates were found to support induced pluripotent stem cells (iPSCs) growth [7] and promote mesoderm and ectoderm differentiation, but GO substrates promoted endoderm differentiation while graphene substrates suppressed this differentiation. Graphene substrates were found to improve the cell reprogramming into the pluripotent state compared to glass substrates [8]. Furthermore, neuronal, myocardial and haematopoietic differentiations of embryonic stem cells were improved by GO substrates [9,10,11]. GO based substrates were found to promote the osteogenic and adipogenic differentiation of ASCs and mesenchymal stem cells [12,13,14]. Interestingly graphene based substrates were found better substrates for osteogenic differentiation compared to GO, while the opposite behaviour was observed for adipogenic differentiation. Lastly, graphene oxide suspensions were able to eradicate cancer stem cells by inducing their differentiation. [15]

Thus far, the study of graphene based scaffold materials has been limited to the most common variations – graphene, GO and reduced (r)GO, which while influencing stem- cell differentiation, does not offer sufficient determinacy and specificity towards a particular lineage of choice. In this study, we show that graphene can be chemically modified to achieve far greater deterministic specificity in its effect on stem cell differentiation, in this case, functionalization with IKVAV to direct differentiation towards the neuroglial lineage. IKVAV is an oligo peptide sequence contained in the alpha1 chain of the ECM protein laminin. This peptide sequence is able to promote cell adhesion, neural development ad neurite outgrowth [17]. GO-IKVAV substrates are shown to be biocompatible, and increase the expression of neuronal markers in ASCs while GO and rGO substrates increased the expression of osteogenic and chondrogenic markers. GO IKVAV substrates also showed an increased cell attachment of rat dorsal root ganglia (DRG) neurons and increased neurite outgrowth length.

2. Materials and methods

2.1 Graphene based materials synthesis, substrates preparation and characterization

GO was synthesized by a modified Hummer`s method [18] and exfoliated to yield a dispersion of 100% monolayers of GO. GO was then used as a starting reagent for the production of GO-IKVAV solutions. GO was then reacted with 100 mM of chloroacetic acid and 250 mM of NaOH under magnetic stirrer for 3 hours to increase the yield of carboxyl groups in GO. Subsequently, a two-step reaction [19] where bovine serum albumin (BSA) was grafted to GO was followed to in order to bind IKVAV peptide to GO. Briefly, GO is reacted and mixed with 50 mL of N-Hydroxysuccinimide (NHS) at the concentration of 50 mg/mL solubilized in 50mL of 2-(N-morpholino)ethanesulfonic acid (MES) buffer 500mM (pH 6.1) for 15 minutes under magnetic stirring at room temperature. Then 60 mL of N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC) 10 mg/ml is added to the reaction mix for 30 minutes under stirring at room temperature. The esterified GO is then purified by three centrifugation steps at 11,000 rpm for 60 minutes. After, the esterified GO was dissolved in 50 mL of the MES buffer (pH 6.1) and 10 mL of 10 mg/ml of IKVAV peptide was reacted for 60 minutes under magnetic stirrer at room temperature. The product was then purified in the same way as described for the estered GO. Circular glass coverslips with a diameter of 13 millimeters were washed in a sonication bath with Decon® 90 for 15 mins followed by DI water for another 15 mins and finally with isopropanol for 15 mins. After the 88 coverslips were dried, they were treated for 5 mins in oxygen plasma to increase the hydrophilicity of the coverslips. GO and GO IKVAV solutions were then spin-coated on the glass coverslips at the concentration of 2 mg/ml at 2500 rpm, 250 rpm/sec acceleration for 2 mins. To obtain rGO coverslips, GO coverslips were annealed for three days at 180 °C in vacuum. Atomic force microscopy (AFM) measurements were carried using a Bruker microscope in tapping mode. Raman spectroscopy was performed using a Renishaw inVia Raman microscope with 532 nm laser excitation. X-ray photoelectron spectroscopy (XPS) spectra of drop-casted substrates were recorded with a SPECS NAP- XPS system employing a monochromatic Al Kα source (1486.6 eV). FT-IR analysis was carried out using a solid FT-IR Avatar 360 ESP spectrometer in the range 500–2500 cm−1 in ATR mode on sample powder.

2.2 Human Adipose stem cells harvesting and culture

Human adipose stem cells were collected by liposuction on consenting patients undergoing reconstructive surgery at University Hospital South Manchester, UK. ASC were isolated according to a previously reported protocol [20]. Adipose tissue were minced by a razor blade and dissociated by an enzymatic treatment of 0,2 % (w/v) of collagenase (Life Technologies, Paisley, UK) for 60 minutes at 37 °C. The digested tissue was then filtered through a vacuum-assisted 100 µm nylon mesh (Merck Millipore, Watford, UK). An equal volume of stem cell growth medium containing a-minimum essential Eagle’s medium (aMEM) (Sigma-Aldrich, Poole, UK), 10% (v/v) foetal bovine serum (FBS) (LabTech, Uckfield, UK), 2 mM L-glutamine (GE Healthcare UK, Little Chalfont, UK), and 1% v/v) penicillin–streptomycin was added. The solution were centrifuged at 300 g for 10 minutes and the resulting pellet was suspended in 1 mL of Red Blood Cell Lysis Buffer (Sigma-Aldrich) for 1 min, and 20 mL of aMEM was added to arrest lysis. The mixture was centrifuged at 300 g for 10 min, and the resulting pellet was suspended in aMEM and plated in T75 flasks for cell culture.

2.3 Cell Proliferation and Live/Dead assay

For the MTS assay, ASC cells were plated at the concentration of 50,000 per coverslip in triplicate on the sterilized functionalized coverslips. The coverslips were put on a 24-well ultra low attachment cell plate (Corning Costar®). At days 1, 4 and 7 the cell medium was aspirated and cells were washed in PBS. After the washing step, the cells were incubated incubated in 20% (v/v) CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Southampton, UK), diluted in phenol-free DMEM (Sigma-Aldrich) for 90 mins in the dark at 37 °C. After the incubation, the absorbance at 490 nm was recorded 89 using an Asys UVM-340 microplate reader/spectrophotometer (Biochrom, Cambridge, UK). For the Live/Dead assay, ASC cells were plated at the concentration of 25,000 cells per coverslip in triplicate in a cell medium containing -MEM and 1% (v/v) P/S without growth factors. After 48 hours, the medium was aspirated and the cells were washed in PBS. Calcein-AM fluorescein and ethidium homodimer-1 (Eth-D1) purchased from LIVE/DEAD Viability/Citoxicity Kit, Molecular Probes, Invitrogen, UK) were added at the concentration of 0.5 and 2 g/ml respectively in PBS and left to react for 30 minutes at room temperature. After this step, a final wash in PBS was performed and images were taken using a fluorescence inverted microscope (Olympus IX51, Japan) under 4% magnification. Data from MTS assay was expressed were expressed as absorbance at 490 nm ± SE of the mean (n=3) while the percentage of live and dead cells is expressed by dividing the average area of the fluorescein/ethidium fluorescence by the average of the whole area imaged (Live + Dead) stained area multiplied by 100.

2.4 Quantitative real-time polymerase chain reaction (qRT-PCR)

For gene expression studies, the RNA was extracted after 48 hours of cellular growth on the different coverslips at the concentration of 50,000 cells per coverslip. RNA was extracted using the RNeasy Plus Mini Kit (Qiagen) following the instruction of the manufacturer. The concentration of the RNA was quantified at the NanoDrop ND-100 (Thermo Fisher Scientific, Waltham, MA, USA) spectrophotometer. 1 µg of each sample were reverse transcribed using the RT2 First Strand Kit (Qiagen) following the instruction of the manufacturer. DNA elimination steps were included in either RNA extraction and cDNA synthesis to get rid of downstream genomic DNA amplification. qRT-PCR was performed with RT2 SYBR Green qPCR Mastermix (Qiagen) and a Corbett Rotor Gene 6000 Qiagen), by the use of the following protocol: hot start for 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C, annealing for 30 s at 55 °C, and extension for 30 s at 72 °C. To verify the specificity of the reactions, a melting curve was obtained with the following protocol: 95 °C for 1 min, 65 °C for 2 min, and a gradual temperature increase from 65 °C to 95 °C (2 °C/min). Data were normalized for the housekeeping gene, and the ΔΔCt method was used to determine the fold changes in gene expression with glass coverslips as controls. The primers were previously published in the literature [30,35].

2.5 Rat DRG neurons harvesting , culture and immunohistochemistry

DRG were harvested from adult male Sprague–Dawley rats as previously reported in the literature [34]. To enzymatically dissociate the neurons, two treatments of 0.125% w/v

90 collagenase type IV (Worthington Biochemical, UK) were performed for 1 h at 37 °C. After the treatments, the neurons were washed in a 0.25% w/v trypsin (Worthington Biochemical) solution for 30 min at 37 °C. After this step, the action of trypsin was blocked by another wash with 33% FBS solution. Neurons were lastly washed by F12 (Sigma-Aldrich) medium and then transferred to a 50 mL Falcon tube to be mechanically dissociate by a glass pipette. The cell suspension obtained was filtered and centrifuged, then the resuspended pellet was slowly pipetted down a gradient tail created with a 50% v/v bovine serum albumin (BSA)/F12 solution. After a final centrifugation step, the pellet was resuspended in Bottenstein and Sato’s medium [BS]; 1% v/v N2 supplement (PAA, UK) in F12 medium, with 50 ng/ml nerve growth factor (NGF; Millipore, USA) and DRG neurons were grown on the different coverslips for 72 hours. To assess the neuronal attachment and neurite length, the cells were fixed using 4% w/v PFA solution for 20 mins at room temperature. The cells were then carefully washed in PBS, permeabilized in 0.2% v/v Triton-X/PBS and washed again in PBS. Non-specific antigens on the coverslips surfaces were blocked with 1% w/v BSA in antibody diluent for 1 h at room temperature and then the primary antibody (anti-β tubulin III mouse monoclonal, 1:500) was allowed to react overnight at 4 °C. On the following day, the cells were washed in PBS and then the secondary antibody solution (Goat Anti-mouse IgG-AlexaFluor 488, 1:500 in antibody diluent) was incubated for 60mins in the dark. After the incubation step, the cells on the substrates were washed using a PBS solution before being mounting the samples on microscopy glass slides containing Vectashield DAPI staining. Using a fluorescence microscope (Olympus BX60), 20–25 random images were captured for neurite length and cell attachment quantification. The neurite lengths were manually evaluated using the freehand line tool of ImageJ software [National Institutes of Health (NIH), USA]. Results were expressed as average lengths ± SE of the mean (n=3). Cell attachment was quantified counting each cell with a manual cell counter. Results were expressed as average of the number of cells ± SE of the mean (n=3).

2.6 Statistical analysis

Statistical significance of the studies was evaluated by the use of GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA, USA) using a one-way ANOVA test followed by Dunnett’s multiple comparison test using glass as a control sample. Level of significance was expressed as P-values.

3. Results and Discussion

3.1 Substrates Characterization 91

Figure 3.1: (a, b, c) AFM characterization of the GO, rGO and GO IKVAV coverslips respectively. (d, e, f) Optical microscopy characterization of GO, rGO and GO IKVAV coverslips respectively. Scale bars in figures (d) and (e) are equal to 50 µm while scar bar in figure (f) is equal to 100 µm (g, h, i) XPS characterization of GO, rGO and GO IKVAV coverslips characterization. The XPS C1s spectra are fitted with 6 components: C-C (cyan) at 284.6 , C=C (red) at 285.1 eV, C-O and C-N (magenta) at 286.1 eV and 285.8 eV respectively, C-O-C (green) at 286.9 eV, C=O (blue) at 287.7 eV and lastly carboxylic groups (orange) at 288.8 eV

A combination of optical microscopy (OM) and AFM was used to investigate the thickness and the uniformity of graphene coverage on the glass coverslip surfaces. OM and AFM measurement confirmed the uniformity of the films and that uncoated areas. were not traceable as shown in Fig.3.1 a-c and Fig. S2. As shown in Fig S2, the thickness was measured by an AFM scan over a scratch in the film, created by gently scratching the centre of the substrate with a pair of cleanroom tweezers. Then the scratched area has the same thickness of the glass substrate and it can be used as a reference to measure the thickness of the film. The thickness measured ranged from 10 to 15 nm. The presence of graphene oxide on the substrates is also confirmed by Raman spectroscopy (supporting information). Two typical peaks are visible in the Raman spectrum of GO: the D peak at ~1350 cm-1 and the G peak at ~1586 cm-1. The intensity ratio between these two peaks

(ID/IG) of the GO coverslips is 0.92 while upon thermal reduction we noticed a decrease (0.84) in the intensity ratio. Upon IKVAV functionalization we noticed an increase in the

ID/IG to 1.14. This increase can be explained as the material became more defective during the functionalization process. XPS analysis confirmed the chemical functionalization and the thermal reduction of GO. XPS wide scan (Supporting Information) of GO shows a C/O ratio of 2.4 while after the thermal reduction, this ratio increases up to 5.5 confirming successful deoxygenation of GO. XPS wide scan of both rGO and GO IKVAV showed the presence of nitrogen peaks. All GO C1s (Fig.3.1 g-i) spectrum can be deconvoluted in 6 components: C=C (sp2 carbon) at 284.6 eV, C-C (sp3 carbon) at 285.1 eV, C-OH at 286 eV, C-O-C at 286.9 eV, C=O at 287.7 eV and HO-C=O at 288.8 eV. Upon reduction, it is noted an increase of the sp2 carbon, a decrease of sp3 carbon and all the other oxygen- containing functional groups with the exception of hydroxyl group [21]. N1s peak of rGO substrate is formed only by a single peak at 399.8 eV, linked to the presence of non- protonated nitrogen atoms. After the IKVAV functionalization, it is observed a down shift in the hydroxyl and carboxyl peaks (285.8 eV and 288 eV respectively) coherent with the forming of C-N and HN-C=O covalent bonds [22,23]. Also the percentage concentration of the two peaks increases compared to GO spectrum. N1s spectrum (Supporting Information) can be deconvoluted in two peaks: one at 399.8 eV is linked to the non- protonated nitrogen atoms of the peptidic bonds; the other peak at 402 eV is linked to the protonated amino groups of the side chain of lysine, one of the amino acid which composes the penta-peptide IKVAV [24].

3.2 Effect on IKVAV functionalization on improving neuronal attachment and neurite outgrowth

Figure 3.2: (a) Immunohistochemical analysis of the neurite length from DRGs grown on the different coverslips. (b) DRG attachment measurement on the different coverslips after immunohistochemical analysis .

The choice of IKVAV functionalization to direct stem cell differentiation was based in part on the observation that GO-IKVAV substrates showed an improved neuronal attachment and neurite outgrowth. Our results (Fig.3.2a) show that GO-IKVAV substrates have a significant increase of DRG neurons attached (mean of 209 ± 45 neurons, p<0,05) compared to glass substrates (mean of 105 ± 17 neurons). GO substrates shows a marginal increase in the DRG neurons attached compared to the glass controls (mean of 166 ± 12 DRG), even though this difference is not statistically significant. rGO substrates show a similar behaviour to the glass substrates. Studying two different parameters assessed Neurite outgrowth: the percentage of sprouting neurons and the length of neurites. DRG neurons grown on GO-IKVAV substrates show an average neurite length which is double than the average neurite length measured in DRG neurons grown on glass substrates (127.9 ± 17.4 µm vs 64.4 ± 7 µm, p<0, 001, Fig.3.2b). The average neurite length measured on both GO and rGO substrates follows the same trend of glass substrates with no statistically significant difference between the different groups studied. rGO substrates shows the same behaviour of GO-IKVAV substrates. The observations that GO-IKVAV substrates represent a better surface to grow DRG neurons compared to the other substrates influenced our decision to study the effect of this functionalization on ASCs to assess if the functionalization increases the neuro-glial development on these stem cell populations.

3.3 Proliferation of ASCs on graphene substrates.

MTS assay is employed to assess ASCs proliferation on the graphene substrates on three

Figure 3.3: (a) MTS assay proliferation results on the different substrates (b) Live Dead quantification expressed as % of Liv ASC Cells grown on the different substrates. time-points: day 1, day 4 and day 7. As it can be seen on Fig. 3.3a, results on day 1 and day 4 show no statistical difference between the groups. On day 7, we observe a statistically significant increase on GO and GO-IKVAV substrates compared to glass controls (p< 0.0001 and p<0, 001 respectively). Live/Dead assay is employed to measure the biocompatibility of these substrates. Results show that more than 99 % of cells were alive after being grown for 48 hours on all the substrates (Fig.3.3b). We can conclude that all the coated substrates are highly biocompatible and there is no statistical difference compared to the glass substrates. Moreover, GO and GO-IKVAV substrates show a proliferation rate compared to glass, positioning them as a better substrate for ASCs cell culture. ASC cells were cultured in FBS so it is not possible to exclude physical absorption of FBS to the substrates and a consequent effect played by FBS in ASC cell attachment and proliferation.

3.4 Gene Expression Studies

To evaluate the effect of the different substrates on the stem cell differentiation, qRT- PCR was employed to assess fold-changes in gene expression for osteogenic, chondrogenic and neuroglial markers (Fig. 3.4a).

RUNX2/CBFA1 is a transcription factor involved in osteoblast differentiation [25] while alkaline phosphatase (ALP) is an enzyme highly expressed in osteoblasts [26]; therefore we chose these two markers to assess the osteogenic differentiation on our substrates as previous reports highlighted the importance of increased expression of these two markers in the differentiation of mesenchymal stem cells towards osteoblasts. [27]

Interestingly, we noticed that GO and rGO substrates significantly increased the expression of ALP (0.5 fold increase, p<0.001 on both the substrates, Fig. 3.4b). The

95 levels of ALP mRNA expression were found highly similar between GO-IKVAV and glass substrates. GO and rGO substrates did not increase the expression of RUNX2/CBFA1 compared to glass, but we found out that GO-IKVAV significantly decreased the expression of this transcription factor (0.2 fold decrease, p<0.05, Fig. 3.4c).

The chondrogenic differentiation of mesenchymal stem cells was analysed by studying the expression of SOX9 and Col1a [28,29]. We noticed that GO and rGO substrates marginally increase the expression of both SOX9 and Col1a. This increase is not statistically significant (p>0.05, Fig.3.4d, e) due to the high variability between the substrates in both the GO and rGO substrates. In GO-IKVAV substrates, this marginal increase is not observed and the level of mRNA expression of both SOX9 and Col1a is highly similar to the glass controls. We can postulate that growing ASCs on GO and rGO substrates is directing the cells towards osteogenic and marginally chondrogenic differentiation. But the IKVAV functionalization is reversing this behaviour inhibiting all the effect promoted by both GO and rGO.

We then tested the effect of the substrates on neuroglial differentiation to check our hypothesis that the IKVAV functionalization directs ASCs towards a neuroglial differentiation. We studied the expression of neuronal differentiation markers such as GDNF, Erb3, TrkA and NT-3 (Fig. 3.4f - h). Upon neuroglial differentiation, increased expression of GDNF, TrkA and a decreased expression of NT-3 has been reported [30]. The role of Erb3 is more controversial. Faroni et al reported a decreased expression of Erb3 following neuroglial differentiation [30] while others reported an increased expression of this marker upon glial differentiation of mesenchymal stem cells [31]. Our results showed an increased expression of the neurotrophin GDNF only in the GO- IKVAV substrates and not in GO and rGO substrates. The expression of Erb3 receptor is significantly increased on GO-IKVAV substrates compared to the glass controls. This behaviour is not observed in GO and rGO substrates. GO-IKVAV substrates marginally increase the expression of TrkA receptor and marginally decrease the expression of NT3 substrates. We can observe a clear effect of GO-IKVAV substrates in preventing the osteogenic and chondrogenic differentiation of ASCs and in promoting the neuroglial differentiation. We conclude that on one hand the functionalization with IKVAV reverses the expression of osteogenic and chondrogenic markers, and on the other hand it directs the ASCs to differentiate towards a neuroglial lineage.

SOX9

Col1a

Figure 3.4: (a) Schematic representation of the differentiation pathways of ASCs studied here, and the marker genes corresponding to each lineage. (b - i) Quantitative Real-Time PCR studies of ASCs grown on the different substrates.

4. Conclusions This study highlights the chemical versatility of graphene as a scaffold material for stem- cell differentiation, in particular, that graphene can be biochemically functionalized in a deterministic way to direct stem cell differentiation along a predefined lineage. Graphene can be readily functionalized using the vast toolkit of organic chemistry, in particular, by using GO as a starting point that. We observed that the GO substrate functionalized with the penta-peptide IKVAV resulted in increased attachment of DRG neurons on the surface and sprouting longer neurites. This led us to explore the effect of IKVAV on differentiation pathways, and we report that GO-IKVAV substrates direct ASCs to differentiate along a neuroglial lineage, while suppressing chondrogenic and osteogenic differentiation. This result provides the basis for identifying further functional derivatizations of graphene for directing stem cell differentiation and maintaining differentiated lineages, providing a versatile platform for regenerative therapies.

References:

1. Tohill, M.; Terenghi, G., Stem-cell plasticity and therapy for injuries of the peripheral nervous system. Biotechnology and Applied Biochemistry 2004, 40, 17-24.

2. di Summa, P. G.; Kalbermatten, D. F.; Raffoul, W.; Terenghi, G.; Kingham, P. J., Extracellular Matrix Molecules Enhance the Neurotrophic Effect of Schwann Cell-Like Differentiated Adipose-Derived Stem Cells and Increase Cell Survival Under Stress Conditions. Tissue Engineering Part A 2012, 19, 368-379.

3. Faroni, A.; Terenghi, G.; Magnaghi, V., Expression of Functional γ-Aminobutyric Acid Type A Receptors in Schwann-Like Adult Stem Cells. Journal of Molecular Neuroscience 2012, 47, 619-630.

4. Faroni, A.; Rothwell, S. W.; Grolla, A. A.; Terenghi, G.; Magnaghi, V.; Verkhratsky, A., Differentiation of adipose-derived stem cells into Schwann cell phenotype induces expression of P2X receptors that control cell death. Cell Death Dis 2013, 4, e743.

5. Kaewkhaw, R.; Scutt, A. M.; Haycock, J. W., Anatomical site influences the differentiation of adipose-derived stem cells for Schwann-cell phenotype and function. Glia 2011, 59, 734-749.

6. Tomita, K.; Madura, T.; Sakai, Y.; Yano, K.; Terenghi, G.; Hosokawa, K., Glial differentiation of human adipose-derived stem cells: Implications for cell-based transplantation therapy. Neuroscience 2013, 236, 55-65.

7. Chen, G. Y.; Pang, D. W. P.; Hwang, S. M.; Tuan, H. Y.; Hu, Y. C., A graphene- based platform for induced pluripotent stem cells culture and differentiation. Biomaterials 2012, 33, 418-427.

8. Yoo, J.; Kim, J.; Baek, S.; Park, Y.; Im, H.; Kim, J., Cell reprogramming into the pluripotent state using graphene based substrates. Biomaterials 2014, 35, 8321-8329.

9. Lee, T.-J.; Park, S.; Bhang, S. H.; Yoon, J.-K.; Jo, I.; Jeong, G.-J.; Hong, B. H.; Kim, B.-S., Graphene enhances the cardiomyogenic differentiation of human embryonic stem cells. Biochemical and Biophysical Research Communications 2014, 452, 174-180.

10. Garcia-Alegria, E.; Iliut, M.; Stefanska, M.; Silva, C.; Heeg, S.; Kimber, S. J.; Kouskoff, V.; Lacaud, G.; Vijayaraghavan, A.; Batta, K., Graphene Oxide promotes embryonic stem cell differentiation to haematopoietic lineage. Scientific Reports 2016, 6, 25917. 99

11. Lv, M.; Zhang, Y.; Liang, L.; Wei, M.; Hu, W.; Li, X.; Huang, Q., Effect of graphene oxide on undifferentiated and retinoic acid-differentiated SH-SY5Y cells line. Nanoscale 2012, 4, 3861-3866.

12. Nayak, T. R.; Andersen, H.; Makam, V. S.; Khaw, C.; Bae, S.; Xu, X.; Ee, P.-L. R.; Ahn, J.-H.; Hong, B. H.; Pastorin, G.; Özyilmaz, B., Graphene for Controlled and Accelerated Osteogenic Differentiation of Human Mesenchymal Stem Cells. ACS Nano 2011, 5, 4670-4678.

13. Kim, J.; Choi, K. S.; Kim, Y.; Lim, K.-T.; Seonwoo, H.; Park, Y.; Kim, D.-H.; Choung, P.-H.; Cho, C.-S.; Kim, S. Y.; Choung, Y.-H.; Chung, J. H., Bioactive effects of graphene oxide cell culture substratum on structure and function of human adipose- derived stem cells. Journal of Biomedical Materials Research Part A 2013, 101, 3520- 3530.

14. Lee, W. C.; Lim, C. H. Y. X.; Shi, H.; Tang, L. A. L.; Wang, Y.; Lim, C. T.; Loh, K. P., Origin of Enhanced Stem Cell Growth and Differentiation on Graphene and Graphene Oxide. ACS Nano 2011, 5, 7334-7341.

15. Fiorillo, M.; Verre, A. F.; Iliut, M.; Peiris-Pagés, M.; Ozsvari, B.; Gandara, R.; Cappello, A. R.; Sotgia, F.; Vijayaraghavan, A.; Lisanti, M. P., Graphene oxide selectively targets cancer stem cells, across multiple tumor types: Implications for non- toxic cancer treatment, via “differentiation-based nano-therapy”. 2015.

16. Bitounis, D.; Ali-Boucetta, H.; Hong, B. H.; Min, D.-H.; Kostarelos, K., Prospects and Challenges of Graphene in Biomedical Applications. Advanced Materials 2013, 25, 2258-2268.

17. Tashiro, K.; Sephel, G. C.; Weeks, B.; Sasaki, M.; Martin, G. R.; Kleinman, H. K.; Yamada, Y., A synthetic peptide containing the IKVAV sequence from the A chain of laminin mediates cell attachment, migration, and neurite outgrowth. Journal of Biological Chemistry 1989, 264, 16174-16182.

18. Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M., Improved Synthesis of Graphene Oxide. ACS Nano 2010, 4, 4806-4814.

19. Shen, J.; Shi, M.; Yan, B.; Ma, H.; Li, N.; Hu, Y.; Ye, M., Covalent attaching protein to graphene oxide via diimide-activated amidation. Colloids and Surfaces B: Biointerfaces 2010, 81, 434-438.

100

20. Kingham, P. J.; Kalbermatten, D. F.; Mahay, D.; Armstrong, S. J.; Wiberg, M.; Terenghi, G., Adipose-derived stem cells differentiate into a Schwann cell phenotype and promote neurite outgrowth in vitro. Experimental Neurology 2007, 207, 267-274.

21. Michio, K.; Hikaru, T.; Kazuto, H.; Shinsuke, M.; Chikako, O.; Asami, F.; Takaaki, T.; Yasumichi, M., Analysis of Reduced Graphene Oxides by X-ray Photoelectron Spectroscopy and Electrochemical Capacitance. Chemistry Letters 2013, 42, 924-926.

22. Bao, H.; Pan, Y.; Ping, Y.; Sahoo, N. G.; Wu, T.; Li, L.; Li, J.; Gan, L. H., Chitosan-Functionalized Graphene Oxide as a Nanocarrier for Drug and Gene Delivery. Small 2011, 7, 1569-1578.

23. Tu, Q.; Pang, L.; Chen, Y.; Zhang, Y.; Zhang, R.; Lu, B.; Wang, J., Effects of surface charges of graphene oxide on neuronal outgrowth and branching. Analyst 2014, 139, 105-115.

24. Polzonetti, G.; Battocchio, C.; Dettin, M.; Gambaretto, R.; Di Bello, C.; Carravetta, V.; Monti, S.; Iucci, G., Self-assembling peptides: A combined XPS and NEXAFS investigation on the structure of two dipeptides Ala–Glu, Ala–Lys. Materials Science and Engineering: C 2008, 28, 309-315.

25. Gaur, T.; Lengner, C. J.; Hovhannisyan, H.; Bhat, R. A.; Bodine, P. V. N.; Komm, B. S.; Javed, A.; van Wijnen, A. J.; Stein, J. L.; Stein, G. S.; Lian, J. B., Canonical WNT Signaling Promotes Osteogenesis by Directly Stimulating Runx2 Gene Expression. Journal of Biological Chemistry 2005, 280, 33132-33140.

26. Zhang, W.; Yang, N.; Shi, X.-M., Regulation of Mesenchymal Stem Cell Osteogenic Differentiation by Glucocorticoid-induced Leucine Zipper (GILZ). Journal of Biological Chemistry 2008, 283, 4723-4729.

27. Granéli, C.; Thorfve, A.; Ruetschi, U.; Brisby, H.; Thomsen, P.; Lindahl, A.; Karlsson, C., Novel markers of osteogenic and adipogenic differentiation of human bone marrow stromal cells identified using a quantitative proteomics approach. Stem Cell Research 2014, 12, 153-165.

28. Akiyama, H.; Chaboissier, M.-C.; Martin, J. F.; Schedl, A.; de Crombrugghe, B., The transcription factor Sox9 has essential roles in successive steps of the chondrocyte differentiation pathway and is required for expression of Sox5 and Sox6. Genes & Development 2002, 16, 2813-2828.

101

29. Herlofsen, S. R.; Küchler, A. M.; Melvik, J. E.; Brinchmann, J. E., Chondrogenic Differentiation of Human Bone Marrow-Derived Mesenchymal Stem Cells in Self- Gelling Alginate Discs Reveals Novel Chondrogenic Signature Gene Clusters. Tissue Engineering Part A 2010, 17, 1003-1013.

30. Faroni, A.; Smith, R. J. P.; Lu, L.; Reid, A. J., Human Schwann-like cells derived from adipose-derived mesenchymal stem cells rapidly de-differentiate in the absence of stimulating medium. European Journal of Neuroscience 2016, 43, 417-430.

31. Brohlin, M.; Mahay, D.; Novikov, L. N.; Terenghi, G.; Wiberg, M.; Shawcross, S. G.; Novikova, L. N., Characterisation of human mesenchymal stem cells following differentiation into Schwann cell-like cells. Neuroscience Research 2009, 64, 41-49.

32. Noble, J.; Munro, C. A.; Prasad, V. S. S. V.; Midha, R., Analysis of Upper and Lower Extremity Peripheral Nerve Injuries in a Population of Patients with Multiple Injuries. Journal of Trauma and Acute Care Surgery 1998, 45, 116-122.

33. Faroni, A.; Mobasseri, S. A.; Kingham, P. J.; Reid, A. J., Peripheral nerve regeneration: Experimental strategies and future perspectives. Advanced Drug Delivery Reviews 2015, 82–83, 160-167.

34. Malin, S. A.; Davis, B. M.; Molliver, D. C., Production of dissociated sensory neuron cultures and considerations for their use in studying neuronal function and plasticity. Nat. Protocols 2007, 2, 152-160.

35. Smith, C. A.; Richardson, S. M.; Eagle, M. J.; Rooney, P.; Board, T.; Hoyland, J. A., The use of a novel bone allograft wash process to generate a biocompatible, mechanically stable and osteoinductive biological scaffold for use in bone tissue engineering. Journal of Tissue Engineering and Regenerative Medicine 2015, 9, 595-604.

102

Supporting Information

1. Substrate Characterization

Fig. S1 a shows the Raman spectrum of GO substrates with its characteristic D and G peak. The ID/IG ratio measured on GO substrates is 0.92. Upon thermal reduction the value of this ration decrease down to 0.84 (Fig. S1 b). After functionalization with IKVAV, the

ID/IG increases up to 1.14 due to the increase of structural defects in GO structure occurring during the functionalization process (Fig. S1 c). XPS wide scan graphs are used to analyse the level of oxygenation before and the after the reduction process and to appreciate the presence of nitrogen atoms after IKVAV functionalization.

Fig.S1 d shows the XPS wide scan of GO substrates where no nitrogen atom peak is observable. The main peaks observable are the carbon, oxygen, silicon and sodium peaks. The atomic C/O ratio is measured at 2.4. Fig.S1 e shows the XPS wide scan of rGO substrates where carbon, oxygen, nitrogen and silicon peaks are observable. The atomic C/O ration after the reduction process increases up to 5.5 confirming the successful outcome of the reduction process.

Fig.S1 f shows the XPS wide scan of GO IKVAV substrates where a stronger nitrogen peak is observable after the functionalization process. To discriminate the different nature of nitrogen peaks in rGO and GO IKVAV substrates we analysed the N1s peak of the two different substrates. N1s peak of GO IKVAV shows two different peaks: one at 399.8 eV is linked to the non-protonated nitrogen atoms of the peptidic bonds; the other peak at 402 eV is linked to the protonated amino groups of the side chain of lysine, one of the amino acid which composes the penta-peptide IKVAV (Fig.S1 h). N1s peak of rGO shows only one peak at 399.8 eV which is due to non protonated nitrogen atoms bound to the substrate. (Fig.S1 g). N1s peak analysis confirmed the different covalent bond in the different substrates further confirming the grafting of IKVAV pentapeptide on GO.

Fig.S2 a shows a picture of GO flakes after exfoliation with a measured thickness of 1.5- 2 nanometers. (Fig.S2 b).

Fig.S2 c-d show the thickness of GO thin films, obtained after spin-coating at the concentration of 2 mg/mL, measured on purposely scratched area of the substrate. The measured thickness ranges from 9-14 nanometers.

103

Figure S1: (a,b,c) Raman spectra of GO, rGO and GO-IKVAV substrates. (d,e,f) XPS wide scan spectra of GO, rGO and GO-IKVAV substrates. (g,h) XPS N1s spectra of rGO and GO-IKVAV respectively.

104

105

Fig. S2 e-f shows the FT-IR spectrum of GO, IKVAV and GO-IKVAV to further confirm the IKVAV functionalization of GO. Broad band at ca. 3380 cm-1 can be assigned to νN- H and νO-H vibrations of amine, amide and carboxyl groups of IKVAV. In GO spectrum a broader band is observed at this region due to hydrogen bonded hydroxyl, carbonyl and carboxyl groups and moisture content. The bands at 3000-2800 cm-1 range can be assigned to νC-H vibrations of methyl and methylene groups, which are not clearly observed in GO spectrum. Broad band at 1800-1700 cm-1 can be assigned to carbonyl and carboxyl νC=O vibrations. This band decreases in relative intensity when compared to GO-IKVAV, which is associated to GO amidation process. GO-IKVAV spectrum show a band at 1630 cm-1 which is similarly observed in IKVAV spectrum. This feature is attributed to overlapped bands due to νC=O and βN-H vibrations of amide groups or amine. Such results are consistent with GO amidation process by IKVAK pentapeptide. At 1300-900 cm-1 range, IKVAV and GO-IKVAV spectral profiles are very similar and present two sharp and intense bands at ca. 1185 and 1030 cm-1 (νC-N and νC-O, respectively), whereas GO presents an intense and very broad band at 1100-930 cm-1 (νC- O and βC-H). These features also indicate the IKVAV association with GO.

106

Introduction to Chapter 4

In this chapter, the fabrication of reduced graphene oxide and graphene oxide based substrates is performed. Graphene oxide thin films were produced on glass coverslips by spin-coating. Thermal reduction of GO was performed in vacuum oven on the substrate. Characterization of the substrates was performed by Raman and X-Ray photoelectron spectroscopies, atomic force and optical microscopies.

Schwann-like differentiated adipose stem cells (dASCs) were grown on the substrates and the biocompatibility was tested by MTS and Live/Dead assays. The impact of the different substrates on the gene expression of neurotrophins, intermediate filament proteins and receptors of neurotrophins was assessed by quantitative Real Time PCR.

Neurotrophins and intermediate filament proteins are key molecular players involved in the regeneration of peripheral nerve injury.

Reduced graphene oxide positively modulate the gene expression of neurotrophins and intermediate filament proteins therefore it is an interesting material for this application also as it conducts electricity, Reduced graphene oxide electrodes can be fabricated for the electrical stimulation of this cells in co-culture with damaged neurons to electrically stimulate nerve regeneration.

Authors Contribution

Maria Iliut prepared and purified graphene oxide dispersions. Andrea Francesco Verre fabricated the substrates and performed Raman and atomic force microscopy characterization of the substrates. Andrea Francesco Verre made the sample preparation for the XPS measurement, which was performed by Christopher Muryn. Andrea Francesco Verre and Claudio Silva analysed the XPS data. Alessandro Faroni harvested the adipose stem cells and differentiated them towards a Schwann-like cells phenotype. Andrea Francesco Verre and Alessandro Faroni performed MTS , Live/Dead assays and quantitative Real-Time PCR experiments.

Adam Reid, Alessandro Faroni and Aravind Vijayaraghavan helped Andrea Francesco Verre in the interpretation of the results and in the writing of the manuscript.

107

Reduced GO substrates increase gene expression of neurotrophins and filament proteins by Schwann-like differentiated adipose stem cells

Andrea Francesco Verre,1 Alessandro Faroni,2 Maria Iliut,1 Claudio Silva,1 Cristopher Muryn,3 Adam J Reid2 and Aravind Vijayaraghavan1,*

1 School of Materials and National Graphene Institute, University of Manchester, Manchester M13 9PL, UK

2 School of Chemistry, University of Manchester, Manchester M13 9PL, UK

3 Faculty of Life Science, University of Manchester, Manchester M13 9PL, UK

* Corresponding author: [email protected]

Abstract

There is urgent clinical need to improve the clinical outcome of peripheral neuropathies. Many efforts are directed towards the fabrication of bioengineered conduits, which delivers Schwann-like differentiated cells at the site of injury to assist and guide the peripheral nerve regeneration. Aim of this study is to assess if graphene and related nanomaterials can be useful in the fabrication of such conduits. A comparison is made between substrates of GO and reduced GO substrates in comparison to standard tissue culture glass. Our results show that the graphene substrates are highly biocompatible, and the reduced GO substrates are more effective in increasing the gene expression of the biomolecules involved in the regeneration process compared to the other substrates studied.

108

1. Introduction

Schwann cells (SC) are key cellular elements in assisting the regeneration of peripheral nerve after injury. SC switch from a myelinating to repair phenotype which results in increased expression ofWhen a peripheral nerve is cut or degenerated, the cell debris produced by the inflammation is eliminated by the presence of macrophages aided by SCs. Subsequently, SCs are fundamental players in repairing the site of injury. These cells start to express extracellular matrix (ECM) proteins, neurotrophins and growth factors;, and as well asfurthermore, SC undergoing profound morphological and phenotypical changes which result in changes of the expressionupregulation of filament cytoskeletal proteins such as nestin and, actin. All these molecules are secreted in the site of the injury and guide the neuronal regeneration process [1]. Despite a clear need for novel therapies, use of SC as a clinical intervention for peripheral nerve injury (PNI) is problematic due to the necessity of harvesting a functional nerve and the limited expansion capacity of SC. As a clinically viable alternative, mesenchymal stem cells and adipose-derived mesenchymal stem cells (ASCs) have been differentiated in vitro towards a Schwann-like cells phenotype [2,3]. These differentiated adipose stem cells (dASCs) express glial markers such as GFAP, S100 and p75 [3]; express myelin protein [4] and myelin structures when in co-culture with neurons [5]; and when implanted in bioengineered conduits to repair murine peripheral nerve gap in vivo, dASC have demonstrated promotion of nerve regeneration, reduction of muscle atrophy, increased nerve conduction velocity and higher rates of myelination [6-9]. Graphene and related nanomaterials can play an important role in bioengineered nerve conduits as they can act as functional electro-active surface to electrically stimulate the regeneration process. Although the biocompatibility of these materials in the liquid phase depends on many variables such as the thickness, the lateral size of the flakes, the level of hydrophilicity and the extent of functionalization [28], it can be stated that when these materials were used as coatings on surfaces to support stem cell growth, the extent of cytotoxicity was limited [29,30]. Graphene and related nanomaterials were found to be effective in positively modulating axonal outgrowth and nerve regeneration in vitro [10-13]. Thus far, researchers have been exploring the effect of graphene and related nanomaterials on the neurite outgrowth, but there have not been studies regarding the effect of these materials in supporting the growth of dASCs. The aim of this study is the biological characterization of graphene oxide (GO) and reduced GO (rGO) coated coverslips as substrates to grow dASCs. Our results confirm the biocompatibility of the graphene-

109 based substrates and show increased expression of neurotrophins and filament proteins mainly on rGO and GO substrates. These results strongly position graphene coatings to be used as electro-active functional surface in a co-culture model with DRGs under electrical stimulation.

2. Experimental

2.1 Graphene based materials synthesis, substrates preparation and characterization

The preparation of GO solution, the deposition of GO solution on freshly plasma-cleaned glass coverslips and the thermal reduction of GO into rGO were performed as explained in Chapter 3.

2.2 Human Adipose stem cells harvesting and differentiation

The harvest of ASC cells were performed as explained in Chapter 3. The differentiation of ASC towards dASCs was performed following a previously reported protocol [3]. Briefly, ASCs at the passage 2 at 30% of confluence were treated with 1 mM - mercaptoethanol (sigma-Aldrich) for 24 hours, then with 35 ng/mL of all-trans-retinoic acid for 72 hours. After this initial treatment, ASCs were treated by 5 ng/mL if platelet- derived growth factor (Peprotech EC, London,UK) , 10 ng/mL basic fibroblast growth factor (Peprotech EC), 14 mM of forskolin (Sigma-Aldrich) and 192 ng/mL glial growth factor (GGF-2) (Acorda Therapeutics, Ardsley,NY, USA). ASCs were kept under these conditions for 2 weeks and after 72 hours the cell medium was refreshed.

2.3 Cell Proliferation and Live/Dead assays

To assess cell proliferation, dASCs cells were plated at the concentration of 50,000 cells per coverslip in triplicate on the sterilized functionalized coverslips and punt on a 24- well cell plate (Corning Costar) under ultra-low attachment conditions microscope. MTS and Live/Dead® assays were performed as explained in Chapter 3.

2.4 Quantitative real-time polymerase chain reaction (qRT-PCR)

For gene expression studies, the RNA was extracted after 48 hours of dASC cellular growth on the different coverslips. RNA was extracted using the RNeasy Plus Mini Kit (Qiagen) following the instruction of the manufacturer and qRT-PCR was performed as explained in Chapter 3.

110

2.5 Statistical analysis

3. Results and Discussion

3.1 Substrates Characterization

OM and AFM showed the uniformity of thin films coverage on the substrate and no empty areas were observable on the substrates (Fig.4.1a-d) as explained in Chapter 3. Successful reduction was confirmed by the C1s spectrum of rGO: we noticed a clear decrease in all the oxygen functionalities with the exception of hydroxyl group, a decrease in sp3 carbon and increased sp2 carbon as shown in Chapter 3. Raman spectroscopy is a useful tool to detect the presence of graphene oxide on any surfaces. The typical spectrum of GO is composed by two peaks: the D peak at ~1350 cm-1 and the G peak at ~1586 cm1 The intensity ratio (ID/IG) between these two peaks is employed to characterize GO dispersions. The measured ID/IG of GO was 0.92 while upon thermal reduction we noticed a decreased ID/IG down to 0.84 as explained in Chapter 3.

Fig. 4.1:(a), (b) AFM topography image of GO and rGO uniform thin films respectively; (c),(d) Optical microscopy image of GO and rGO coverslips respectively

111

3.2 Proliferation of dASCs on graphene substrates.

To assess dASCs proliferation rate on the different substrates studied we used MTS assay. We selected three time-points on day 1, day 4 and day 7. As it can be seen on Fig. 4.2a, the proliferation on both GO and rGO coverslips is statistically reduced on all the three different time points compared to glass substrates. Here it is noted that while for ASC proliferation rate there is an increase in the latest time point in GO and GO-IKVAV, the opposite behaviour is noted for dASC where a slower proliferation rate takes places both on GO and rGO substrates. Live/Dead assay is employed to quantify the amount of cells, which were alive after being grown on the substrates. We measured that more than 99.7%, as shown in Fig.4.2b, of cells were alive in all the different groups studied. We can therefore conclude that although the proliferation rate slowed down on the rGO and GO coverslips, the amount of live cells still remained very good in all the different substrates studied. We then decided to study the gene expression of crucial proteins and growth factor being involved in the peripheral nerve regeneration process.

3.3 Gene Expression Studies

Key molecules involved in the regeneration process are neurotrophins. These growth factors are involved in neuronal survival, development and functionality [18,19,20]. We decided to focus our attention studying brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF) and glial-derived neurotrophic factor (GDNF). Also another group of protein molecules involved in the regeneration process are filament proteins such as nestin and vimentin.

Figure 4.2: a) MTS assay of dASCs grown on rGO coverslips (purple line), GO coverslips (red line), and bare glass coverslips (blue line). b) Live/Dead assay results of dASCs grown on the different coverslips.

* p <0.05, ** p<0.01 , *** p<0.001, **** p<0.0001

112

Nestin is a protein, which is involved in the axonal growth and is normally up regulated after nerve injury [21,22]. Moreover a recent study highlighted that nestin-positive hair- follicle pluripotent stem cells were able to promote peripheral nerve regeneration [23]. Vimentin is reported to be up regulated during peripheral nerve regeneration and higher expression of this protein has been reported to augment the peripheral nerve regeneration [24]. Lastly we decided to investigate also the gene expression of neurotrophins receptor such as TrkB, TrkC and Ret.

GDNF expression is increased on rGO coverslips (0.64 ± 0.06 vs glass, p< 0.01, n=3) and no statistical difference is observed between GO coverslips and the glass controls. The expression of BDNF is increased on rGO and GO coverslips (0.53 ± 0.08, p  0.01 vs glass, n=3 on rGO substrates) and (0.68 ± 0.01, p<0.001 vs glass, n=3 on GO substrates). NGF expression increased on rGO coverslips (0.67 ± 0.14, p-value  0.01) while GO substrates behaved as glass controls. (Fig. 4.3 a-c)

Nestin expression is increased on both rGO and GO substrates (0.44 ± 0.05, p<0.01 vs glass, n=3 on rGO substrates) and (0.31 ± 0.11, p<0.05 vs glass, n=3 on GO substrates). Vimentin expression is increased on both rGO and GO substrates (0.66 ± 0.04, p 0.001 vs glass, n=3 on rGO substrates) and (0.67 ± 0.01, p<0.001 vs glass, n=3 on GO substrates. (Fig. 4.3 d-e)

TrkC expression is increased on rGO substrates (1.86 ± 0.29, p 0.01 vs glass, n=3) while no statistical difference is observed between GO substrates and the glass controls. The same trend is followed by the gene expression of TrkB receptor. The expression of this gene is increased on rGO substrates (0.85 ± 0.02, p<0.01 vs glass, n=3). The expression of Ret receptor is marginally increased on rGO substrates although not statistically significant due to the high variability of the rGO substrates. The expression of this gene on GO shows the same behaviour as the glass controls.(Fig. 4.3f-h).

Faroni et al. [25] studied the gene expression changes when ASCs are differentiated towards Schwann-like dASCs. The gene expression of GDNF, BDNF, TrkC, Ret, nestin and vimentin markers was reported to be upregulated after the differentiation protocol compared to non-differentiated ASC. The expression of NGF and TrkB was reported to be downregulated after the differentiation protocol compared to non-differentiated ASC.

113

Figure 4.3: Gene expression of dASCs on the different substrates: a) GDNF; b) BDNF; c) NGF; d) Nestin; e) Vimentin; f) TrkC ; g) Ret; h) TrkB

The increase in gene expression can be due either to a positive effect of graphene and related nanomaterials in increasing neuroglial genes expression or to a cellular response to a stress induced by the GO or rGO coatings. Other results in the literature seems to 114 highlight that reduced GO nanoscaffolds do not increase cytotoxivity in the spinal cord in vivo [26] and graphene functionalized scaffolds do reduce inflammatory response in the adult brain [27]. Therefore we can postulate that rGO induces higher gene expression of crucial proteins and growth factor involved in the peripheral nerve regeneration.

4 Conclusion

The development of a protocol that permanently differentiate ASC cells into dASC is crucial to implement a stem-cell based therapy strategy for peripheral nerve regeneration. dASC cells were found able to express glial markers and to promote nerve regeneration, myelination and to increase the speed of the conduction in the nerve [6-9]. The main obstacle in the clinical translation of dASC is the maintenance of the differentiated phenotype. Faroni et al [25] proved that after the withdrawal of the growth factors, dASC started to decrease the expression of glial markers and to reverse into ASC characteristics. There is consequently the need to develop better differentiation protocols and to test new materials that help maintaining the dASC phenotype even after the withdrawal of the growth factors. Graphene and related nanomaterials have been widely reported as suitable material to support stem cell growth and differentiation [31-33]. In this study, the proliferation rate and the biocompatibility of these substrates were studied by two different assays and we can conclude although there is a slower proliferation rate of dASC on rGO substrates, the amount of live cells is comparable between all the different substrates studies indicating that GO and rGO substrates do not cause cytotoxicity after 48 hours. Importantly, the analysis of gene expression of important glial markers increased after 48 hours of cellular growth on GO and rGO substrates. The expression of neurotrophins and their receptors is statistically increased on rGO substrates and to a lesser extent on GO substrates. Moreover, the expression of filament proteins such as nestin and vimentin is statistically increased on both and rGO substrates. As the initial results on 48 hours are encouraging, further studies need to be conducted to establish if the gene expression of these markers will be increased at longer time points as the entire differentiation protocol required 2 weeks of treatment.

115

References:

1. Son, Y.-J.; Thompson, W. J., Schwann cell processes guide regeneration of peripheral axons. Neuron 1995, 14, 125-132.

2. Faroni, A.; Rothwell, S. W.; Grolla, A. A.; Terenghi, G.; Magnaghi, V.; Verkhratsky, A., Differentiation of adipose-derived stem cells into Schwann cell phenotype induces expression of P2X receptors that control cell death. Cell Death Dis 2013, 4, e743.

3. Kingham, P. J.; Kalbermatten, D. F.; Mahay, D.; Armstrong, S. J.; Wiberg, M.; Terenghi, G., Adipose-derived stem cells differentiate into a Schwann cell phenotype and promote neurite outgrowth in vitro. Experimental Neurology 2007, 207, 267-274.

4. Tomita, K.; Madura, T.; Mantovani, C.; Terenghi, G., Differentiated adipose- derived stem cells promote myelination and enhance functional recovery in a rat model of chronic denervation. Journal of Neuroscience Research 2012, 90, 1392-1402.

5. Xu, Y.; Liu, L.; Li, Y.; Zhou, C.; Xiong, F.; Liu, Z.; Gu, R.; Hou, X.; Zhang, C., Myelin-forming ability of Schwann cell-like cells induced from rat adipose-derived stem cells in vitro. Brain Research 2008, 1239, 49-55.

6. Georgiou, M.; Golding, J. P.; Loughlin, A. J.; Kingham, P. J.; Phillips, J. B., Engineered neural tissue with aligned, differentiated adipose-derived stem cells promotes peripheral nerve regeneration across a critical sized defect in rat sciatic nerve. Biomaterials 2015, 37, 242-251.

7. di Summa, P. G.; Kalbermatten, D. F.; Pralong, E.; Raffoul, W.; Kingham, P. J.; Terenghi, G., Long-term in vivo regeneration of peripheral nerves through bioengineered nerve grafts. Neuroscience 2011, 181, 278-291.

8. di Summa, P. G.; Kingham, P. J.; Raffoul, W.; Wiberg, M.; Terenghi, G.; Kalbermatten, D. F., Adipose-derived stem cells enhance peripheral nerve regeneration. Journal of Plastic, Reconstructive & Aesthetic Surgery 2010, 63, 1544-1552.

9. di Summa, P. G.; Kingham, P. J.; Campisi, C. C.; Raffoul, W.; Kalbermatten, D. F., Collagen (NeuraGen®) nerve conduits and stem cells for peripheral nerve gap repair. Neuroscience Letters 2014, 572, 26-31.

10. Li, N.; Zhang, X.; Song, Q.; Su, R.; Zhang, Q.; Kong, T.; Liu, L.; Jin, G.; Tang, M.; Cheng, G., The promotion of neurite sprouting and outgrowth of mouse hippocampal cells in culture by graphene substrates. Biomaterials 2011, 32, 9374-9382. 116

11. Tang, M.; Song, Q.; Li, N.; Jiang, Z.; Huang, R.; Cheng, G., Enhancement of electrical signaling in neural networks on graphene films. Biomaterials 2013, 34, 6402- 6411.

12. Tu, Q.; Pang, L.; Wang, L.; Zhang, Y.; Zhang, R.; Wang, J., Biomimetic Choline- Like Graphene Oxide Composites for Neurite Sprouting and Outgrowth. ACS Applied Materials & Interfaces 2013, 5, 13188-13197.

13. Tu, Q.; Pang, L.; Chen, Y.; Zhang, Y.; Zhang, R.; Lu, B.; Wang, J., Effects of surface charges of graphene oxide on neuronal outgrowth and branching. Analyst 2014, 139, 105-115.

14. Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M., Improved Synthesis of Graphene Oxide. ACS Nano 2010, 4, 4806-4814.

15. Hummers, W. S.; Offeman, R. E., Preparation of Graphitic Oxide. Journal of the American Chemical Society 1958, 80, 1339-1339.

16. Michio, K.; Hikaru, T.; Kazuto, H.; Shinsuke, M.; Chikako, O.; Asami, F.; Takaaki, T.; Yasumichi, M., Analysis of Reduced Graphene Oxides by X-ray Photoelectron Spectroscopy and Electrochemical Capacitance. Chemistry Letters 2013, 42, 924-926.

17. Ganguly, A.; Sharma, S.; Papakonstantinou, P.; Hamilton, J., Probing the Thermal Deoxygenation of Graphene Oxide Using High-Resolution In Situ X-ray-Based Spectroscopies. The Journal of Physical Chemistry C 2011, 115, 17009-17019.

18. Hempstead, BL.; Dissecting the Diverse Actions of Pro- and Mature Neurotrophins. Current Alzheimer Research 2006, 3, 19-24.

19. Reichardt, L. F., Neurotrophin-regulated signalling pathways. Philosophical Transactions of the Royal Society B: Biological Sciences 2006, 361, 1545.

20. Lentz, S. I.; Knudson, C. M.; Korsmeyer, S. J.; Snider, W. D., Neurotrophins Support the Development of Diverse Sensory Axon Morphologies. The Journal of Neuroscience 1999, 19, 1038-1048.

21. Sahin Kaya, S.; Mahmood, A.; Li, Y.; Yavuz, E.; Chopp, M., Expression of nestin after traumatic brain injury in rat brain. Brain Research 1999, 840, 153-157.

117

22. Frisén, J.; Johansson, C. B.; Török, C.; Risling, M.; Lendahl, U., Rapid, widespread, and longlasting induction of nestin contributes to the generation of glial scar tissue after CNS injury. The Journal of Cell Biology 1995, 131, 453-464.

23. Amoh, Y.; Aki, R.; Hamada, Y.; Niiyama, S.; Eshima, K.; Kawahara, K.; Sato, Y.; Tani, Y.; Hoffman, R. M.; Katsuoka, K., Nestin-positive hair follicle pluripotent stem cells can promote regeneration of impinged peripheral nerve injury. The Journal of Dermatology 2012, 39, 33-38.

24. Perlson, E.; Hanz, S.; Ben-Yaakov, K.; Segal-Ruder, Y.; Seger, R.; Fainzilber, M., Vimentin-Dependent Spatial Translocation of an Activated MAP Kinase in Injured Nerve. Neuron 2005, 45, 715-726.

25. Faroni, A.; Smith, R. J. P.; Lu, L.; Reid, A. J., Human Schwann-like cells derived from adipose-derived mesenchymal stem cells rapidly de-differentiate in the absence of stimulating medium. European Journal of Neuroscience 2016, 43, 417-430.

26. Palejwala, A. H.; Fridley, J. S.; Mata, J. A.; Samuel, E. L. G.; Luerssen, T. G.; Perlaky, L.; Kent, T. A.; Tour, J. M.; Jea, A., Biocompatibility of reduced graphene oxide nanoscaffolds following acute spinal cord injury in rats. Surgical Neurology International 2016, 7, 75.

27. Zhou, K.; Motamed, S.; Thouas, G. A.; Bernard, C. C.; Li, D.; Parkington, H. C.; Coleman, H. A.; Finkelstein, D. I.; Forsythe, J. S., Graphene Functionalized Scaffolds Reduce the Inflammatory Response and Supports Endogenous Neuroblast Migration when Implanted in the Adult Brain. PLoS ONE 2016, 11, e0151589.

28. Bussy C.; Ali-Boucetta, H.; Kostarelos, K., Safety Considerations for Graphene: Lessons Learnt from Carbon Nanotubes. Acc. Chem. Res., 2013, 46 (3), pp 692–701

29. Kim, J.; Choi, K. S.; Kim, Y.; Lim, K.-T.; Seonwoo, H.; Park, Y.; Kim, D.-H.; Choung, P.-H.; Cho, C.-S.; Kim, S. Y.; Choung, Y.-H.; Chung, J. H., Bioactive effects of graphene oxide cell culture substratum on structure and function of human adipose- derived stem cells. Journal of Biomedical Materials Research Part A 2013, 101, 3520- 3530.

30. Lee, W. C.; Lim, C. H. Y. X.; Shi, H.; Tang, L. A. L.; Wang, Y.; Lim, C. T.; Loh, K. P., Origin of Enhanced Stem Cell Growth and Differentiation on Graphene and Graphene Oxide. ACS Nano 2011, 5, 7334-7341.

118

31. Yoo, J.; Kim, J.; Baek, S.; Park, Y.; Im, H.; Kim, J., Cell reprogramming into the pluripotent state using graphene based substrates. Biomaterials 2014, 35, 8321-8329.

32. Garcia-Alegria, E.; Iliut, M.; Stefanska, M.; Silva, C.; Heeg, S.; Kimber, S. J.; Kouskoff, V.; Lacaud, G.; Vijayaraghavan, A.; Batta, K., Graphene Oxide promotes embryonic stem cell differentiation to haematopoietic lineage. Scientific Reports 2016, 6, 25917.

33. Nayak, T. R.; Andersen, H.; Makam, V. S.; Khaw, C.; Bae, S.; Xu, X.; Ee, P.-L. R.; Ahn, J.-H.; Hong, B. H.; Pastorin, G.; Özyilmaz, B., Graphene for Controlled and Accelerated Osteogenic Differentiation of Human Mesenchymal Stem Cells. ACS Nano 2011, 5, 4670-4678.

119

Introduction to Chapter 5

In this chapter, the attachment of His-tagged Orco nanodiscs on graphene oxide surfaces is performed. Firstly the carboxylation of graphene oxide is carried out by reacting chloroacetic acid and sodium hydroxide. The alkali environment opens the epoxy ring and produces diols. Diols and hydroxyl groups from graphene oxide are then converted into carboxylic groups by the action of chloroacetic acid.

Carboxylic groups are then reacted with carbodiimide to create ester bonds. Esters bonds are reacted with Nα,Nα-Bis(Carboxymethyl)-L-lysine hydrate(NTA-NH2) by the formation of amide bonds. The grafting of nitrilotriacetic acid (NTA) molecules onto graphene oxide allows the immobilization of nickel ions through the interaction of nickel ions and NTA. Nickel ions are then able to interact with the His-tag sequence of the nanodiscs allowing their attachment.

The evaluation of the carboxylation of GO is characterized by X-Ray Photoelectron spectroscopy while the immobilization of the nanodiscs and their detachment from the surface by treatment with imidazole is characterized by atomic force microscopy.

The possible applications of such graphene functionalization range from protein purification matrices to the production of graphene based biosensors. Many biotechnological companies and research laboratories worldwide use His-tagged recombinant proteins. The immobilization of odorant receptors such as Orco can be exploited in the production of graphene-based electrical nose sensors.

Authors Contribution

Andrea Francesco Verre, Stefan Goodwin and Colm Carraher carried out the functionalization of graphene oxide with the His-tagged Orco nanodiscs. Colm Carraher performed the expression of Orco receptors and the production of His-tagged Orco nanodiscs. Andrea Francesco Verre performed the sample preparation for XPS analysis, which was performed by Christopher Muryn. Andrea Francesco Verre performed the analysis of the XPS data. Andrea Francesco Verre performed the atomic force microscopy measurement. Aravind Vijayaraghavan, Ernie Hill, Bruce Grieve and Andrew Kralicek, supervised the project, Andrea Francesco Verre and Aravind Vijayaraghavan jointly interpreted the data and write the manuscripts.

120

Selective immobilization of membrane proteins carried in nanodiscs on functionalized graphene oxide

Andrea F Verre,1 Colm Carraher, 2 Christopher Muryn, 3 Stefan Goodwin, 4 Bruce Grieve, 5 Ernie Hill, 4 Andrew Kralicek, 2 Aravind Vijayaraghavan1,*

1 School of Materials and National Graphene Institute, University of Manchester, Manchester M13 9PL, UK 2 Faculty of Science, University of Auckland, Auckland 1142, NZ 3 School of Chemistry, University of Manchester, Manchester M13 9PL, UK 4 School of Computer Science, University of Manchester, Manchester M13 9PL, UK 5 School of Electrical and Electronic Engineering, University of Manchester, Manchester M13 9PL, UK * Corresponding author: [email protected]

Abstract: Nanodiscs are synthetic carriers of biomimetic membranes and membrane proteins; they are comprized of membrane scaffolding proteins encapsulating a phospholipid bilayer, which offers a biomimetic environment for membrane proteins. The immobilization of membrane proteins on a graphene surface could enable a new route to understanding their functioning mechanisms and enable novel bio-chemical sensors. In this paper, we demonstrate a chemical functionalization route to bind histidine tagged odorant receptor co-receptor (Orco) which is housed in a nanodisc on to the surface of graphene oxide. The immobilization was confirmed by X-ray photoelectron spectroscopy and atomic force microscopy. The ability of attaching and detaching His-tagged proteins can be exploited also in affinity chromatography, a widely used technique in protein purification.

121

1. Introduction

The attachment of proteins from the surface is at the heart of many biotechnological processes such affinity chromatography and bio-sensing. The presence of molecular tags which can be expressed in recombinant proteins can be used by various surfaces to selectively bind the recombinant protein, for instance from the cellular lysate in affinity chromatography [1-3]. Subsequently, the binding needs to be dissociated so the recombinant protein can be eluted and the purified protein product can be yielded. In biosensing, it is crucial to attach biomolecules such molecular receptors to the transduction surface, which can then specifically recognize biomarkers, gas-phase or biomolecules of interest. [4,5]

Carbon nanotubes (CNT), graphene and related nanomaterials have high specific surface area and their polyaromatic structure enables various routes of organic chemistry to be used to bind the molecular tags. CNTs were oxidized and used to bind histidine (His)- tagged proteins selectively on its surface [6]. Biotinilated graphene oxide (GO) has been shown to be a useful matrix for the purification of streptavidin-tagged proteins [7]. GO has been functionalized by hydrothermal reduction in a solution of nickel cations by hydrazine hydrate in the presence of poly(sodium-p-styrenesulfonate). Reduced GO flakes decorated with nickel nanoparticles have been used as a matrix to selective bind His-tagged recombinant proteins [8]. Graphene has been functionalized with carboxylated diazonium salts, to create carboxylic groups on the surface of the molecule. The carboxylic groups were then used to link His-tagged green fluorescent proteins to realize a graphene-based filed effect transistors (FET) characterized by wavelength- dependent photoresponses. [9,10]

In this work, we functionalized GO for the controlled binding of odorant receptor co- receptor (Orco), which is housed in the biomimetic environment of a nanodisc. A schematic of the Orco/Nanodisc system is shown in Figure 1(a).

Orco is a receptor highly conserved in all the insect orders and is important in sensorial neurons as it allows the start of the signalling evoked by odors. Membrane integral proteins such as molecular receptors are assembled within the plasmatic membranes of the living cells. When they are attached to surfaces, it is ideal to create a lipid biomimetic interface, which can preserve the natural folding of the protein, and to not decrease the protein functionality. Orco could play a key role in developing a graphene-based bioelectronic nose, and conversely graphene could be used as a transducer to understand the functioning of Orco in a biomimetic environment. [11,12] 122

Nanodiscs are circular shaped phospholipid bilayers, which are self-assembled with integral membrane receptors. This self-assembling technique is very useful as it allows chemically binding receptor proteins, which are self-assembled in a biomimetic membrane [13,14]. Examples of proteins successfully assembled in nanodiscs were cytochrome P450 [15,16], blood coagulation and human tissue factor [17], bacteriorhodopsin [18,19], G-protein coupled receptors [20,21,22] and in this case, Orco.

2. Methodology

2.1. Preparation and functionalization of GO flakes and imidazole elution of His-tagged proteins GO was prepared by a modified Hummer’s method. Briefly, graphite oxide was synthesized by a modified Hummer`s method [14,15] and exfoliated down to monolayer to yield GO as explained more in details in Chapter 2.

Carboxylation of GO with NaOH and chloracetic acid was then performed in solution using three different concentration of chloroacetic acid (0.5M, 1M and 3M respectively) dissolved in 4M solution of NaOH as explained in Chapter 2.

0.7 mg/mL carboxylated GO is spun down on the surface of 290 nm SiO2/Si chips by spin coating at the speed of 2,500 rpm, at acceleration of 250 rpm/sec for 2 minutes and then the functionalization is carried out as explained in Chapter 2 in the Section 2.2.

The reaction scheme is explained in Fig 5.1

After the Orco nanodiscs binding, the chips were immersed in 300 mM imidazole solution and left to incubate in a 50 mL glass Petri dish for one hour. Afterward, the chips were rinsed three times with deionized water and blown dry using nitrogen gun.

2.2 XPS and AFM Characterization

Sample for XPS were prepared by drop-casting the different dispersions on top of silicon dioxide wafers and dried in vacuum. The sample was analysed with a SPECS NAP-XPS system employing a monochromatic Al Kα source (1486.6 eV). XPS C1s peak and wide scan spectra were analysed and fitted by using CasaXPS software. AFM measurements were performed by FastScan microsope in tapping mode on the spin coated substrates.

3. Results and Discussion

3.1. XPS characterisation of the carboxylation of graphene oxide The carboxylation of GO in this reaction follows a two step mechanism: firstly the opening of the epoxy rings by the action of strong alkali and lastly the grafting of acetic 123 groups due to the reaction between chloroacetic acid and hydroxyl groups in graphene oxide. The alkali solution generates the formation of hydroxyl ions. Hydroxyl ions are nucleophile species which tends to bind carbon atoms of the epoxy rings forming diols group. Hydroxyl groups from opened epoxy rings and the ones already present originally in GO react with the carbon atom directly binding the chlorine atoms. Chlorine atoms behave as leaving group allowing the grafting of acetic group in the structure of GO.

Figure 5.1: (a,) Schematic of the functionalization of GO performed in this study. b) Illustration of the protein immobilisation strategy performed in this study.

124

XPS C1s spectrum of GO (Fig.2a) showed C=C peak at 284.6 eV, C-C at 285.1 eV, C- OH at 286 eV, C-O-C at 287.4 eV, C=O at 288 eV and carboxyl peak at 289 eV. Epoxy groups show an atomic percentage of 48%, carboxyl peak shows an atomic percentage of 4.43 % and hydroxyl peaks shows an atomic percentage of 4.6%.

Analyzing the C1s spectra after the reaction with 0.5 M chloracetic group and 4M NaOH (Fig 5.2b), it is possible to notice a clear decrease in the C1s peak with an atomic percentage of 18.50% and an increase in the atomic percentage of hydroxyl peak to 5.2 %. This demonstrates clearly the successful opening of the epoxy rings by treatment with NaOH. The atomic percentage of the carboxyl peak is narrowly increased compared to GO showing a value of 5.75%.

Analyzing the C1s spectra after the reaction with 1 M chloracetic group and 4M NaOH (Fig 5.2c), the atomic percentage of the epoxy peak decreases reaching a value of 18.46% confirming the effect of NaOH in opening the epoxy rings. The atomic percentage of hydroxyl and carboxyl peaks increase up to 12.28% and to 10% respectively

Figure 5.2: (a) XPS characterization of GO before carboxylation, (b) XPS characterization of GO reacted with 0.5 M of chloroacetic acid and 4M of NaOH. (c) XPS characterization of GO reacted with 1M of chloroacetic acid and 4M of NaOH. (d) XPS characterization of GO reacted with 3M of chloroacetic acid and 4M of NaOH

125

This confirms chloroacetic acid at the concentration of 1M is able to chemically convert hydroxyl groups into carboxylic groups.

Analysing the C1s spectra after the reaction with 3 M chloracetic group and 4M NaOH (Fig 5.2d), it is noticed a very small decrease in the atomic percentage of the epoxy group that shows a value of 45.12%. This is reasonably due to the fact that increasing concentration of chloroacetic acid brings the pH of the solution towards the neutrality. This decreases the efficacy of the opening of the epoxy rings.

The atomic percentage of the hydroxyl and carboxyl peaks is 5.23 and 6.44 % respectively. Because the opening of the epoxy rings is not happening, the grafting of acetyl groups is not happening due to the small number of hydroxyl ions that can boost up the second step of the reaction. [26,27,28]

Ultimately, the XPS characterisation of this reaction shows a successful carboxylation of graphene oxide and that the reaction with 1M of chloracetic acid and 4M NaOH shows the best outcome.

3.2. AFM studies of the protein binding on functionalized GO After established the successful carboxylation of GO, we have used the presence of carboxylic acids as binding sites to attach Orco-DMPG nanodiscs to GO surface. Carboxylic groups were first activated by carbodiimide coupling reaction. Carbodiimide coupling chemistry is a common functionalization route to bind biomolecules to carboxylic groups. Carboxylic groups are not efficient leaving group and they required to be activated by carbodiimide to form esters group. After this reaction, NTA-NH2 reacts with esterified GO and the formation of amide bonds occurs. NTA molecule is reported to bind nickel ions through the formation of electrostatic interactions between positively charged nickel ions and the negatively charged oxygen atoms from NTA. [29]

The deposition of nickel ions on the surface of GO allows immobilising the Orco-DMPG on the surface due to the interaction between histidine molecules from the His-tag sequence of the protein and nickel ions. To characterize this binding experiment we employed AFM and analysed how the height profile changes from reference GO topography to the topography profile of the functionalized GO. To be certain of the selectivity of our functionalization, we reversed the binding using imidazole. Imidazole is a molecule, which has a stronger binding affinity towards nickel ions compared to histidine, therefore allowing the detachment of His-tagged Orco-DMPG-nanodiscs. AFM topography image of non-functionalized GO (Fig 5.3a) shows height profile of 1 -

126

2 nanometers, which is indicative of GO mono and bilayers. The distribution of the height peaks confirming that the majority of the flakes show a height distribution between 1 and 2 nanometers with no measurement higher than 3 nanometers. AFM topography images of functionalized GO (Fig 5.3b) shows a distribution profile with two main peaks at 0 nm, which is attributable to the bare substrates, and another peak at between 1 and two nanometers attributable to the GO flakes thickness. One important difference between the distribution profiles is the presence of height profiles of 5-6 nanometers, which is indicative of successful attachment of Orco-DMPG nanodiscs attachment. The average thickness of the bright particles is 5 ± 1 nanometers (n=50).

As final investigation on the successful attachment of Orco-DMPG nanodiscs, we have analyzed the distribution of height profiles after reacting the functionalized GO flakes with imidazole (Fig 3c). The distribution profile shows as for the previous samples two

Figure 5.3: (a) AFM characterization of GO as control reference, (b) AFM and height profile measurement of GO flakes after His-tagged Orco immobilisation (c) AFM and height profile measurement of GO flakes after imidazole elution of the His-tagged Orco from the surface of the GO flakes. (A and B in the AFM height graphs stands for Height and Count percentage respectively.

127 main peaks: one at 0 nm attributable to the bare substrate and another one at 1 - 2 nanometers attributable to GO flakes thickness.

We did not notice anymore the presence of height profiles above 3 nanometres confirming the action of imidazole in detaching His-tagged Orco-DMPG nanodiscs from GO surfaces.

The average thickness of the bright particles observed in the Fig.5.3d is 2.5 ± 1 nanometers (n=50).

Overall, it has been demonstrated the possibility of exploiting the presence of carboxylic group to attach His-tagged proteins to GO surfaces. One major issue is the small amount of carboxylic groups in the structure which are detected by XPS. Using a previously reported protocol [25], it has been shown that epoxy rings can be opened and form diols which can react with chloroacetic group increasing the amount of carboxylic groups. XPS data showed that the best condition for this reaction is using 1M of chloroacetic acid and 4M of NaOH. In fact, when the concentration of acid was 3M the value of pH came close to the neutrality, limiting the opening of the epoxy ring. After having obtained an increased number of carboxylic groups, these functional binding sites can be used to graft NTA which can entrap nickel ions through ionic interactions. The presence of nickel ions can be used to bind His-tagged proteins as demonstrated by AFM topography measurements. In fact, the height profile of the protein particles on GO functionalized flakes showed an increased thickness compared to non-functionalized flakes. When functionalized flakes were treated with 300 mM imidazole solution, the thickness of the particles on GO flakes decrease due to the interaction of imidazole and nickel ions which detached His-tagged proteins from GO surfaces.

4. Conclusions In this paper, we have shown the XPS characterization of the carboxylation process of GO flakes. Increasing the concentration of carboxylic functional groups allows increasing the possibility of attaching biomolecules not only on the edges of GO flakes but also on the basal plane of GO. Through this functionalization strategy it is shown how to functionalize GO in order to bind selectively His-tagged proteins to the surface of GO flakes. This process is reversible and chemicals such imidazole can detach selectively His-tagged proteins from GO surfaces. In this instance, we have demonstrated the binding of Orco housed in nanodiscs, which opens a route to application of the Orco/graphene system for biosensing and biomedical research. This functionalization of GO flakes can open up the possibility of using GO-based matrices in the affinity chromatography 128 process for His-tagged protein purification. The system also has significant potential in agriculture, specifically pest control, since Orco is a key protein in the olfaction process of insects.

129

References:

1. Zhao, X.; Li, G.; Liang, S., Several Affinity Tags Commonly Used in Chromatographic Purification. Journal of Analytical Methods in Chemistry 2013, 2013, 8.

2. Janknecht, R.; de Martynoff, G.; Lou, J.; Hipskind, R. A.; Nordheim, A.; Stunnenberg, H. G., Rapid and efficient purification of native histidine-tagged protein expressed by recombinant vaccinia virus. Proceedings of the National Academy of Sciences of the United States of America 1991, 88, 8972-8976.

3. Junttila, M. R.; Saarinen, S.; Schmidt, T.; Kast, J.; Westermarck, J., Single-step Strep-tag® purification for the isolation and identification of protein complexes from mammalian cells. PROTEOMICS 2005, 5, 1199-1203.

4. Park, S. J.; Kwon, O. S.; Lee, S. H.; Song, H. S.; Park, T. H.; Jang, J., Ultrasensitive Flexible Graphene Based Field-Effect Transistor (FET)-Type Bioelectronic Nose. Nano Letters 2012, 12, 5082-5090.

5. Vockenroth, I. K.; Atanasova, P. P.; Knoll, W.; Koper, I.; Toby, A.; Jenkins, A. In Functional tethered bilayer membranes as a biosensor platform, IEEE Sensors, 2005., Oct. 30 2005-Nov. 3 2005; 2005; p 3 pp.

6. Graff, R. A.; Swanson, T. M.; Strano, M. S., Synthesis of Nickel−Nitrilotriacetic Acid Coupled Single-Walled Carbon Nanotubes for Directed Self-Assembly with Polyhistidine-Tagged Proteins. Chemistry of Materials 2008, 20, 1824-1829.

7. Liu, Z.; Jiang, L.; Galli, F.; Nederlof, I.; Olsthoorn, R. C. L.; Lamers, G. E. M.; Oosterkamp, T. H.; Abrahams, J. P., A Graphene Oxide˙Streptavidin Complex for Biorecognition – Towards Affinity Purification. Advanced Functional Materials 2010, 20, 2857-2865.

8. Jia-Wei, L.; Ting, Y.; Lin-Yu, M.; Xu-Wei, C.; Jian-Hua, W., Nickel nanoparticle decorated graphene for highly selective isolation of polyhistidine-tagged proteins. Nanotechnology 2013, 24, 505704.

9. Lu, Y.; Lerner, M. B.; John Qi, Z.; Mitala, J. J.; Hsien Lim, J.; Discher, B. M.; Charlie Johnson, A. T., Graphene-protein bioelectronic devices with wavelength- dependent photoresponse. Applied Physics Letters 2012, 100, 033110.

130

10. Zhang, H.; Li, Z.-f.; Snyder, A.; Xie, J.; Stanciu, L. A., Functionalized graphene oxide for the fabrication of paraoxon biosensors. Analytica Chimica Acta 2014, 827, 86- 94.

11. Carraher, C.; Nazmi, A. R.; Newcomb, R. D.; Kralicek, A., Recombinant expression, detergent solubilisation and purification of insect odorant receptor subunits. Protein Expression and Purification 2013, 90, 160-169.

12. Lundin, C.; Käll, L.; Kreher, S. A.; Kapp, K.; Sonnhammer, E. L.; Carlson, J. R.; von Heijne, G.; Nilsson, I., Membrane topology of the Drosophila OR83b odorant receptor. FEBS Letters 2007, 581, 5601-5604.

13. Bayburt, T. H.; Grinkova, Y. V.; Sligar, S. G., Self-Assembly of Discoidal Phospholipid Bilayer Nanoparticles with Membrane Scaffold Proteins. Nano Letters 2002, 2, 853-856.

14. Bayburt, T. H.; Sligar, S. G., Membrane protein assembly into Nanodiscs. FEBS Letters 2010, 584, 1721-1727.

15. Nath, A.; Koo, P. K.; Rhoades, E.; Atkins, W. M., Allosteric Effects on Substrate Dissociation from Cytochrome P450 3A4 in Nanodiscs Observed by Ensemble and Single-Molecule Fluorescence Spectroscopy. Journal of the American Chemical Society 2008, 130, 15746-15747.

16. Plucinski, L.; Ranjan Gartia, M.; Arnold, W. R.; Ameen, A.; Chang, T.-W.; Hsiao, A.; Logan Liu, G.; Das, A., Substrate binding to cytochrome P450-2J2 in Nanodiscs detected by nanoplasmonic Lycurgus cup arrays. Biosensors and Bioelectronics 2016, 75, 337-346.

17. Morrissey, J. H.; Pureza, V.; Davis-Harrison, R. L.; Sligar, S. G.; Ohkubo, Y. Z.; Tajkhorshid, E., Blood clotting reactions on nanoscale phospholipid bilayers. Thrombosis research 2008, 122, S23-S26.

18. Bayburt, T. H.; Sligar, S. G., Self-assembly of single integral membrane proteins into soluble nanoscale phospholipid bilayers. Protein Science 2003, 12, 2476-2481.

19. Bayburt, T. H.; Grinkova, Y. V.; Sligar, S. G., Assembly of single bacteriorhodopsin trimers in bilayer nanodiscs. Archives of Biochemistry and Biophysics 2006, 450, 215-222.

20. Whorton, M. R.; Bokoch, M. P.; Rasmussen, S. G. F.; Huang, B.; Zare, R. N.; Kobilka, B.; Sunahara, R. K., A monomeric G protein-coupled receptor isolated in a high- 131 density lipoprotein particle efficiently activates its G protein. Proceedings of the National Academy of Sciences 2007, 104, 7682-7687.

21. Borch, J.; Torta, F.; Sligar, S. G.; Roepstorff, P., Nanodiscs for Immobilization of Lipid Bilayers and Membrane Receptors: Kinetic Analysis of Cholera Toxin Binding to a Glycolipid Receptor. Analytical Chemistry 2008, 80, 6245-6252.

22. Raschle, T.; Hiller, S.; Yu, T.-Y.; Rice, A. J.; Walz, T.; Wagner, G., Structural and Functional Characterization of the Integral Membrane Protein VDAC-1 in Lipid Bilayer Nanodiscs. Journal of the American Chemical Society 2009, 131, 17777-17779.

23. Hummers, W. S.; Offeman, R. E., Preparation of Graphitic Oxide. Journal of the American Chemical Society 1958, 80, 1339-1339.

24. Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M., Improved Synthesis of Graphene Oxide. ACS Nano 2010, 4, 4806-4814.

25. Imani, R. E., Shahriar Hojjati Faghihi, Shahab, Nano-graphene oxide carboxylation for efficient bioconjugation applications: a quantitative optimization approach. Journal of Nanoparticle Research 2015, 17, 1-15.

26. Michio, K.; Hikaru, T.; Kazuto, H.; Shinsuke, M.; Chikako, O.; Asami, F.; Takaaki, T.; Yasumichi, M., Analysis of Reduced Graphene Oxides by X-ray Photoelectron Spectroscopy and Electrochemical Capacitance. Chemistry Letters 2013, 42, 924-926.

27. Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S., Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 2007, 45, 1558-1565.

28. Yang, D.; Velamakanni, A.; Bozoklu, G.; Park, S.; Stoller, M.; Piner, R. D.; Stankovich, S.; Jung, I.; Field, D. A.; Ventrice Jr, C. A.; Ruoff, R. S., Chemical analysis of graphene oxide films after heat and chemical treatments by X-ray photoelectron and Micro-Raman spectroscopy. Carbon 2009, 47, 145-152

132

Introduction to Chapter 6

In this chapter, the fabrication of graphene-based microarray as substrates to perform Lipid Dip-Pen nanolithography to produce phospholipids biomimetic membranes as biological interfaces for graphene-based biosensors is investigated. In the literature, the formation of phospholipid biomimetic membranes on graphene surfaces has already been demonstrated. The retain of the lipid functionalisation was demonstrated by binding experiments between biotin-functionalized phospholipids and streptavidin. The aim of this study is to analyse the fabrication of graphene-based microarray as substrate to the production of biomimetic lipid membranes. This biological platform is tested to check wether biological insertion of tail-anchoring proteins is occurring or not.

In this chapter, we use a cytochrome B5-synaptobrevin 2 recombinant proteins, which has the property of inserting into phospholipids bilayers in vivo.

The possible applications of graphene based membrane proteins arrays range from drug- discovery platforms, electrical nose sensors, and as sensing platform to detect antigen- antibody recognition event. The fabrication of the graphene array is performed firstly by transferring CVD graphene to a Si/SiO2 substrates. The patterning of CVD graphene films is then performed by optical lithography. The direct deposition of phospholipids on the graphene array is performed by Lipid-Dip Pen Nanolithography while the insertion of the tail-anchoring proteins is characterising after immunohistochemical assay by fluorescence microscopy.

Authors Contribution

Andrea Francesco Verre and Monica Alberto carried out the transfer of CVD graphene on top of the Si/SiO2 . Andrea Francesco Verre performed the optical lithography on the CVD graphene and the array characterization by Raman spectroscopy and Atomic Force microscopy. Andrea Francesco Verre and Sylwia Sekula-Neuer performed the Lipid- DPN writing on the graphene arrays while Abobakr Abdel Rehim expressed and purified the recombinant tail-anchoring proteins. Andrea Francesco Verre performed the protein insertion and the immunohistochemical staining. Michael Hirtz, Stephen High and Aravind Vijayaraghavan supervised the project, interpreted the data and helped in the writing of manuscript jointly with Andrea Francesco Verre

133

Tail-anchoring proteins insertion into phospholipid biomimetic membranes on graphene

Andrea Francesco Verre,1 Michael Hirtz,2 Monica Alberto,3 Sylwia Sekula-Neuer,2 Abobakr Abdel Rehim,4 Stephen High,4 Aravind Vijayaraghavan1,*

1 School of Materials and National Graphene Institute, University of Manchester, Manchester M13 9PL, UK 2 Institute of Nanotechnology, Karlsruhe Institute of Technology, 76131 Karlsruhe, Germany 3 School of Chemical Engineering and Analytical Science, University of Manchester, Manchester M13 9PL, UK 4 Faculty of Life Science, University of Manchester, Manchester M13 9PL, UK * Corresponding author: [email protected]

Abstract:

Graphene has been shown to be an excellent substrate to support the self-assembly of phospholipid membranes that mimic the cell membrane. The interaction of a cell with its environment and with other cells is key to the existence of complex life, and this is governed by proteins and other biomolecules that reside and function only in the cell membrane. In this study, we report on the insertion of tail-anchored protein (cytochrome- B5-synaptobrevin-2) into the biomimetic membranes supported on graphene. Furthermore, we demonstrate the binding of anti-synaptobrevin-2 antibody by fluorescence microscopy on the graphene supported bio-membranes. This offers a novel route for the immobilization of large biomolecules on graphene using a non-covalent approach which are subsequently functional to specific binding, and this system serves both as a route to fabricate biomimetic cell membrane models to study cellular processes that govern life as well as developing a platform for chemical- and biosensors.

134

1. Introduction

Graphene is a promising material for applications in chemical- and bio-sensors, owing to its combination of maximum specific surface area and versatile chemical functionalizability. The electronic system of graphene is readily influenced by interactions with its environment, such as charge transfer doping; graphene’s ambipolar nature allows for both electron and hole doping to influence its conductivity. Graphene- based bio-sensors for DNA sequencing [1], electronic nose [2] and electrochemical glucose sensors [3] have been previously demonstrated. Graphene typically needs to be chemically functionalized to yield specificity to a target analyte, but two considerations arise in the functionalisation of graphene for bio-sensors. Firstly, covalent functionalisation of graphene disturbs the pi-conjugation and reduced the electronic conductivity and mobility and non-covalent methods to immobilize biomolecules on graphene are sought. Other interactions such as pi-pi stacking, van der Waals interactions or hydrophobic interactions can also be used to adsorb biomolecules on to graphene [4]. Secondly, a number of biomolecules exist and function only in specific biological environments, for example, membrane proteins and ion channels exist in the cell membrane and only perform their function in that environment. In order to study their function or to harness their specific binding properties, they must be immobilized on graphene while retaining their native bio-membrane environment. [5]

Dip Pen Nanolithography (DPN) is an emerging and scalable technology for the targeted delivery of chemical and biomolecules on to various surfaces with nanoscale control over location and pattern of delivery [6-10]. Recently, the use of DPN to achieve molecular deposition on graphene surfaces has been demonstrated, specifically, the delivery of phospholipids in a process known as lipid DPN (L-DPN) [11-12]. L-DPN allows the fabrication of phospholipid biomimetic membranes on graphene and other surfaces, which can create tailored and spatially controlled lipid patches that can serve as a biological overlay to provide a natural environment to biomoelcules in the sensor fabrication process.

The patterning of nano-scale features comprized of self-assembled lipid membranes on graphene surfaces and the demonstration of subsequent binding functionality sets the scene for the application of such a system in biosensing. Atomic force microscopy (AFM) characterization showed the formation of inverted lipid bilayers in air on graphene surfaces in air, where the hydrophobic tail groups are exposed to the graphene surface and surrounding air medium while the hydrophilic tail groups occur on the inside. When 135 the lipid patches are immersed in an aqueous buffer solution, the membrane rearranges into a biologically oriented monolayer with the phosphate groups of the phospholipid heads are facing the hydrophilic solution. This configuration allows biological interactions between the lipid patches and bioactive molecules. The binding was proven by studying the interaction between biotin-functionalized phospholipid head-groups and streptavidin. [11,12]

A further step towards the fabrication of successful biosensors is the insertion of biological molecules into this biomimetic membrane, such as a molecular receptor or an antibody which can allow the detection of HIV or cancer markers though the formation of biological interaction [13,14]. Tail-anchor proteins are a class of integral membrane proteins, which are able to be fully inserted in a lipid membrane by the presence of hydrophobic amino acids in the carboxyl-terminus of the protein while the amino- terminus is facing towards the cytosol [15-20]. This study aims to demonstrate the insertion of tail-anchored proteins into biomimetic lipid patches created by L-DPN on large-scale arrays of graphene patches which would represent the core of a lab-on-a-chip type bio-sensing device [21]. This system is shown schematically in figure 6.1. The successful formation of such biomimetic structures can be exploited in a variety of biosensing devices such drug screening, biomarkers and analytes detection.

2. Materials and methods

2.1 Graphene-Array Fabrication

Chemical vapour deposition (CVD) grown monolayer graphene was transferred from its native copper foil substrate on top of a freshly oxygen plasma cleaned Si/SiO2 wafer using a sacrificial polymer as described in detail in Chapter 2 section 2.6.1.

Figure 6.1: Illustration of the graphene-based array used for the biomimetic phospholipid membrane by DPN and the tail-anchor protein insertion performed in this study. Antibody directed towards the Cytochrome B5 –Synaptobrevin-2 protein is needed to study the protein binding by fluorescence microscopy. 136

After CVD graphene is transferred on top of the silicon dioxide wafer, the graphene arrays were fabricated employing photolithography as explained in detail in Chapter 2 section 2.6.2.

2.2 Lipid Deposition by L-DPN

L-DPN was used to deliver the lipid on to the graphene to create biomimetic lipid membranes on the graphene surface. Multiple 12-pen arrays were used to perform simultaneous lipid DPN on NLP 2000 instrument (Nanoink Inc., USA). The lipids spread on the graphene patch and are confined within the each square by the difference in hydrophilicity between the graphene and the surrounding SiO2 surface as described [12]. The 12-pen array combined with self-confined spreading of lipids on graphene patches allows for rapid wafer-scale formation of biomimetic membranes on graphene arrays. The lipid patterning by L-DPN was performed as explained in Chapter 2 section 2.7.

2.3 Protein Insertion and Immunostaining

The chips were washed by pipetting 50 μL of phosphate-buffered saline (PBS) and then imaged by Zeiss Axio Imager set-up equipped with SARFUS Software (Nanolane) fluorescent microscope. Note that after wetting, the lipid biomimetic membranes became too thin to be image solely by optical microscopy contrast. The chips were then prepared for the protein-insertion assay. Briefly, the areas of the chips not covered by lipids were blocked to avoid non-specific protein binding. The surface was blocked by immersing the chips in a 10% (v/v) bovine serum albumin (BSA) solution in PBS for one hour. After the blocking, the chips were washed by immersion in PBS for 10 minutes. The chips were washed three times. After the wash, the protein insertion assay was performed. 40 μL (at the concentration of 1.7 mg/mL) of recombinant protein Cytochrome b5- synaptobrevin 2 was pipetted in each chip and left to react for 3 hours. After the insertion period, the chips were washed three times in PBS. Each wash lasted 10 minutes.

Following this, for each chip 5 μg of monoclonal mouse anti-synaptobrevin-2 (Synaptic Systems, Germany) were diluted in 50 μL of PBS. Each chip was then incubated in the 50 μL and then left to incubate overnight. The next day, the chips were washed three times in PBS for 10 minutes each.

3.3 Protein Insertion and Immunostaining

The chips were washed by pipetting 50 μL of phosphate-buffered saline (PBS) and then imaged by Zeiss Axio Imager set-up equipped with SARFUS Software (Nanolane) fluorescent microscope. Note that after wetting, the lipid biomimetic membranes became 137 too thin to be image solely by optical microscopy contrast. The chips were then prepared for the protein-insertion assay. Briefly, the areas of the chips not covered by lipids were blocked to avoid non-specific protein binding. The surface was blocked by immersing the chips in a 10% (v/v) bovine serum albumin (BSA) solution in PBS for one hour and the chips were washed three times by covering the whole surface of the chips with 40 µL of PBS as explained in more details in Chapter 2 section 2.7.

Subsequently, an Alexa-488 labelled anti-mouse secondary antibody at 1:500 dilution in PBS was used for the immunostaining of the chips. Each chip was immersed in 50 μL of the secondary antibody dilution and left one hour to react.

Figure 2 (a) Optical characterization of the graphene-based microarray. (b) Average Raman spectrum acquired inside the graphene squares. (c) AFM topography image of one graphene square after the fabrication process. (d) Height profile measurement of graphene-based array after the fabrication process.

Figure 6.2: (a) Optical characterization of the graphene-based microarray. (b) Average Raman spectrum acquired inside the graphene squares. (c) AFM topography image of one graphene square after the fabrication process. (d) Height profile measurement of graphene-based array after the fabrication process.

138

After one hour the samples were washed as previously. The chips were then mounted on a microscope glass coverslips and imaged on a Zeiss Axio Imager set-up equipped with SARFUS Software (Nanolane).

3. Results and Discussion

The formation of the array was characterized by optical microscopy, AFM and by Raman spectroscopy. The result of photolithographic patterning of the graphene was validated by optical microscopy (Fig.6.2a), specifically, the faithful reproduction of the square dimensions and pitch. The pitch is critical, as this needs to match the distance between each DPN pens in the cantilever array in order to realize multiplexed DPN writing.

Raman spectroscopy was used to validate that the lithography process does not result in defects on the graphene by monitoring the intensities of the defect related D-peak to G- peak ratio. An increased ID/IG ratio would correlate to the formation of structural defects in the lattice. Our results show that no D-peak is observed in the Raman mapping in the interior of each square after the photolithography process (Fig.6.2b). This indicates that this fabrication protocol does not create structural defects within the graphene layer. [16]

AFM height profile of the graphene square was employed to evaluate the level of polymer contamination after the photolithography. PMMA tends to adhere to graphene through strong hydrophobic interactions. AFM measurement shows that the lift off procedure has been able to remove the layer of PMMA almost completely as the height profile of each square is between 1 to 3 nm (Fig.6.2c,d).

Optical and fluorescence microscopy was employed to evaluate the lipid deposition on the array both in air and in liquid. As seen in Fig.6.3a, the formation of lipid biomimetic membranes is visible in optical reflectance and fluorescence microscopy. In reflectance, the lipids are visible as bright patches on the graphene squares, compared to graphene squares with no lipids that occurs on the 6th and 12th column corresponding to the uninked pens. The brightness at this stage is no-uniform as excess quantities of lipids have been delivered to each square, and the quantity of lipids flowing off each pen is not identical. Fluorescence microscopy reveals red fluorescence from the membrane containing rhodamine-functionalized phospholipids in column 1 and 7 corresponding to the 1st and 7th pens (Fig.6.3b). Both the techniques confirm that the lipid membranes do not spread outside the graphene surface when the substrate is kept on air. This has been attributed to the contrast in the hydrophobicity/hydrophobicity of the graphene and the surrounding

SiO2 surface. When the substrate is immersed in the liquid, lipid membranes rearranges

139

Figure 6.3: (a) Optical characterization of the graphene-based microarray after lipid patterning (b) Fluorescence microscopy of the lipid patterning in air. (c) Fluorescence microscopy image of lipid patterning in liquid and lipid excess is washed off, resulting ideally in a single monolayer (or half bilayer) of lipids on the graphene surface. The membranes now become too thin for sufficient contrast in optical reflectance. A weak red fluorescence is still visible on the fluorescence microscopy images (Fig.6.3c), which confirms the presence of the lipid membrane within the graphene square.

Fluorescence microscopy is also employed to investigate the insertion of the tail-anchored protein into the biomimetic membrane and subsequent antibody binding. The anti- synaptobrevin-2 is tagged with Alexa 488, a green fluorescent dye. Unfortunately, as the binding protocol requires several washing steps, we were unable to keep the lipid patches strictly confined within the bounds of the graphene square, and this issue will require a

Figure 6.4: (a, b) Fluorescence microscopy of the lipid-protein interaction (lipid-rhodamine fluorescence in red- alexa 488 fluorescence from anti-synaptobrevin 2 after immunohistochemical assays.

140 solution. This spreading out of the lipids is likely caused by the stress on the membranes due to the several immersions of the substrates on liquid buffer (Fig.6.4a). The lipid membranes are stained in green and they are placed outside the surface of the graphene squares. Their green stained is due to the binding of the anti-synaptobrevin-2 antibody to the proteins immobilized in the lipid membrane by means of the tail anchor. The binding is also confirmed when imaging the protein binding on rhodamine-functionalized membranes. In this situation, we observe a fluorescence overlay between the red signal from the lipids and the green signal from the antibody (Fig.6.4b). This clearly indicates the presence of binding event between the protein and the biomimetic membranes.

The fabrication of biomimetic lipid membranes by lipid DPN has been established on a variety of substrates such as silicon dioxide and glass for instance. Graphene has emerged recently as a new substrate for lipid DPN patterning for a number of reason. Graphene in contrast with insulating substrates allows the fabrication of fluorescence quenching biological sensors due to the electronic charge transfers between the lipid membrane and graphene. Moreover, the spatial movement of the phospholipids is higher on graphene substrates compared to glass or silicon dioxide due to the increased hydrophobicity of graphene. This phenomenon allows the uniform spreading of the phospholipids and the formation of a biomimetic membrane stable to be immersed in liquid environment to bind biomolecules. In this study, it was attempted to test the insertion of tail-anchoring proteins into the biomimetic membrane as a model system for the fabrication of a graphene-based biological interface. Although it has been demonstrated the successful insertion of tail- anchoring proteins into the biomimetic membrane, obstacles need to be overcome in the future. It is in fact crucial to control the spreading of the biomimetic membranes in liquid environment because with an increased number of liquid immersion, they tend to be spread outside the graphene square surface.

4. Conclusions

These results confirm the successful fabrication of large area graphene array with specific binding functionality, where the specificity is provided by a membrane protein that is immobilized within its native environment in a non-covalent manner. Fluorescence microscopy analysis of the protein binding seems to suggest that the lipid membranes are well confined within the squares in air and after initial liquid immersion. After the protein binding steps, the binding between the protein and the lipid membranes seems to happen as indicated the co-localized fluorescence between the rhodamine-functionalized lipid membranes and the anti-synaptobrevin-2 antibodies.

141

On the other hand, two main problems need to be overcome in future studies. First, the lipid membranes need to be confined better in the graphene squares for the fabrication of successful biosensors. To achieve this goal, new protocols to study the binding of the protein with the membranes needs to be established. Possible solutions are increasing the hydrophilicity of the substrate to stop the mobility of the lipid within graphene surfaces or decreasing the number of substrate immersion of the chips to cause less stress on the membrane. Secondly, the surface blocking by BSA needs to be improved as minor unspecific binding events are taking place as proved by higher background fluorescence signals due to unspecific binding of labelled antibody to the surface.

When these issues are resolved, the scheme outlined here could serve as a novel system to produce large area, highly multiplexed, lab on a chip devices that rely on membrane proteins serving as specific recognition elements atop a graphene sensing element that provides the signal transduction related to the binding events. This scheme would be particularly applicable for an ‘electronic nose’, since a plethora of odorant binding membrane proteins specific to odour molecules have been identified.

142

References:

1. Chaitanya, S.; Anuj, G.; Jean-Pierre, L.; Klaus, S., Electronic detection of dsDNA transition from helical to zipper conformation using graphene nanopores. Nanotechnology 2014, 25, 445105.

2. Park, S. J.; Kwon, O. S.; Lee, S. H.; Song, H. S.; Park, T. H.; Jang, J., Ultrasensitive Flexible Graphene Based Field-Effect Transistor (FET)-Type Bioelectronic Nose. Nano Letters 2012, 12, 5082-5090.

3. Kang, X.; Wang, J.; Wu, H.; Aksay, I. A.; Liu, J.; Lin, Y., Glucose Oxidase– graphene–chitosan modified electrode for direct electrochemistry and glucose sensing. Biosensors and Bioelectronics 2009, 25, 901-905.

4. Georgakilas, V.; Otyepka, M.; Bourlinos, A. B.; Chandra, V.; Kim, N.; Kemp, K. C.; Hobza, P.; Zboril, R.; Kim, K. S., Functionalization of Graphene: Covalent and Non- Covalent Approaches, Derivatives and Applications. Chemical Reviews 2012, 112, 6156- 6214.

5. White, S. H.; Wimley, W. C., MEMBRANE PROTEIN FOLDING AND STABILITY: Physical Principles. Annual Review of Biophysics and Biomolecular Structure 1999, 28, 319-365.

6. Piner, R. D.; Zhu, J.; Xu, F.; Hong, S.; Mirkin, C. A., "Dip-Pen" Nanolithography. Science 1999, 283, 661-663.

7. Sekula, S.; Fuchs, J.; Weg-Remers, S.; Nagel, P.; Schuppler, S.; Fragala, J.; Theilacker, N.; Franzreb, M.; Wingren, C.; Ellmark, P.; Borrebaeck, C. A. K.; Mirkin, C. A.; Fuchs, H.; Lenhert, S., Multiplexed Lipid Dip-Pen Nanolithography on Subcellular Scales for the Templating of Functional Proteins and Cell Culture. Small 2008, 4, 1785- 1793.

8. Sekula-Neuner, S.; Maier, J.; Oppong, E.; Cato, A. C. B.; Hirtz, M.; Fuchs, H., Allergen Arrays for Antibody Screening and Immune Cell Activation Profiling Generated by Parallel Lipid Dip-Pen Nanolithography. Small 2012, 8, 585-591.

9. Ielasi, F. S.; Hirtz, M.; Sekula-Neuner, S.; Laue, T.; Fuchs, H.; Willaert, R. G., Dip-Pen Nanolithography-Assisted Protein Crystallization. Journal of the American Chemical Society 2015, 137, 154-157.

10. Hirtz, M.; Brglez, J.; Fuchs, H.; Niemeyer, C. M., Selective Binding of DNA Origami on Biomimetic Lipid Patches. Small 2015, 11, 5752-5758. 143

11. Hirtz, M.; Oikonomou, A.; Georgiou, T.; Fuchs, H.; Vijayaraghavan, A., Multiplexed biomimetic lipid membranes on graphene by dip-pen nanolithography. Nature Communications 2013, 4, 2591.

12. Hirtz, M.; Oikonomou, A.; Clark, N.; Kim, Y.-J.; Fuchs, H.; Vijayaraghavan, A., Self-limiting multiplexed assembly of lipid membranes on large-area graphene sensor arrays. Nanoscale 2016, 8, 15147-15151.

13. Lee, K.-B.; Kim, E.-Y.; Mirkin, C. A.; Wolinsky, S. M., The Use of Nanoarrays for Highly Sensitive and Selective Detection of Human Immunodeficiency Virus Type 1 in Plasma. Nano Letters 2004, 4, 1869-1872.

14. Irvine, E. J.; Hernandez-Santana, A.; Faulds, K.; Graham, D., Fabricating protein immunoassay arrays on nitrocellulose using Dip-pen lithography techniques. Analyst 2011, 136, 2925-2930.

15. Borgese, N.; Colombo, S.; Pedrazzini, E., The tale of tail-anchored proteins. Journal of Cell Biology 2003, 161, 1013-1019.

16. Suk, J. W.; Kitt, A.; Magnuson, C. W.; Hao, Y.; Ahmed, S.; An, J.; Swan, A. K.; Goldberg, B. B.; Ruoff, R. S., Transfer of CVD-Grown Monolayer Graphene onto Arbitrary Substrates. ACS Nano 2011, 5, 6916-6924.

17. Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K., Raman Spectrum of Graphene and Graphene Layers. Physical Review Letters 2006, 97, 187401.

18. Borgese, N.; Fasana, E., Targeting pathways of C-tail-anchored proteins. Biochimica et Biophysica Acta (BBA) - Biomembranes 2011, 1808, 937-946.

19. Whitley, P.; Grahn, E.; Kutay, U.; Rapoport, T. A.; von Heijne, G., A 12-Residue- long Polyleucine Tail Is Sufficient to Anchor Synaptobrevin to the Endoplasmic Reticulum Membrane. Journal of Biological Chemistry 1996, 271, 7583-7586.

20. Beilharz, T.; Egan, B.; Silver, P. A.; Hofmann, K.; Lithgow, T., Bipartite Signals Mediate Subcellular Targeting of Tail-anchored Membrane Proteins in Saccharomyces cerevisiae. Journal of Biological Chemistry 2003, 278, 8219-8223.

21. Bog, U.; Laue, T.; Grossmann, T.; Beck, T.; Wienhold, T.; Richter, B.; Hirtz, M.; Fuchs, H.; Kalt, H.; Mappes, T., On-chip microlasers for biomolecular detection via highly localized deposition of a multifunctional phospholipid ink. Lab on a Chip 2013, 13, 2701-2707. 144

22. Della Pia, E. A.; Holm, J. V.; Lloret, N.; Le Bon, C.; Popot, J.-L.; Zoonens, M.; Nygård, J.; Martinez, K. L., A Step Closer to Membrane Protein Multiplexed Nanoarrays Using Biotin-Doped Polypyrrole. ACS Nano 2014, 8, 1844-1853.

23. Fang, Y.; Frutos, A. G.; Lahiri, J., Membrane Protein Microarrays. Journal of the American Chemical Society 2002, 124, 2394-2395.

145

Chapter 7: Conclusions and Future Works

Graphene and its parental material graphene oxide are very attractive materials in biotechnology due to their mechanical properties, ultra-high surface area, ease of functionalisation and biocompatibility. Graphene is also an electrical conductor and it has been proposed as electro-active biological interface for biosensing applications. Future applications for these materials range from biosensing, as carrier in drug and gene delivery, in cancer photo-thermal therapy to graphene-based substrates in stem cells differentiation.

During these years, many researchers have been tested graphene and related nanomaterials to grow different type of stem cells. In this thesis, the aim was to evaluate the effect of chemical functionalization on the differentiation of adipose-derived mesenchymal stem cells as a strategy to direct stem cell differentiation to a neuroglial phenotype. In Chapter 3, it was demonstrated how non-functionalized graphene oxide and reduced graphene oxide substrates tend to direct adipose-derived mesenchymal stem cells differentiation toward osteoblasts and chondrocytes. The interesting effect of IKVAV functionalisation clearly regards the feasibility of reversing these differentiation pathways and directing the adipose-derived mesenchymal stem cells towards neuroglial phenotype by chemical functionalisation of graphene oxide substrates. The ease of functionalisation and the demonstrated biocompatibility that graphene oxide offers makes this material very attractive in stem cell culture and differentiation technology.

In Chapter 4, reduced graphene oxide and graphene oxide based substrates were used as substrates for the growth of Schwann-like differentiated adipose stem cells. Differentiated adipose stem are able to produce proteins and growth factors useful in the peripheral nerve regeneration. There is a need to find new materials to sustain the growth these cells to increase the expression of neurotrophins and intermediate filament proteins. Results produced in this thesis highlighted that although the proliferation rate of these cells is slowed down by the graphene-based coatings compared to the glass controls, the vast majority of the cells were found alive by biocompatibility assays. Results showed how reduced graphene oxide increased the gene expression of neurotrophins such as BDNF and GDNF and intermediate filament proteins like nestin and vimentin. The increased expression of these genes can be used in graphene-based scaffolds to improve the nerve regeneration in the treatment of peripheral neuropathies.

146

Graphene and related nanomaterials are promising materials to be used as transducer in biosensing application due to electronic and optoelectronic properties. The main obstacle that these materials need to overcome is the binding of molecular receptors which can increase both the selectivity and specificity. In this thesis two different strategies were of chemical functionalization were tested: in Chapter 5 membrane proteins carried in lipid nanodiscs were covalently bound to graphene oxide and in Chapter 6 membrane protein were inserted on biomimetic membrane deposited on graphene surface by expoiting hydrophobic interaction.

In Chapter 5, the successful increase of carboxylic groups in the first step of the reaction is characterized by X-Ray photoelectron spectroscopy while the immobilization of the nanodiscs is evaluated by atomic force microscopy measuring the height profile of the GO flakes before the functionalization, after the functionalization and after the imidazole elution. The successful attachment of His-tagged proteins can be extremely useful in biotechnological applications such as protein purification and biosensing.

Lastly in Chapter 6, it was demonstrated the fabrication of graphene-based microarray as substrates to perform lipid Dip-Pen nanolithography to produce phospholipids biomimetic membranes as biological interfaces for graphene-based biosensors. In this thesis, demonstration of successful insertion of tail-anchored recombinant proteins was provided as demonstrated by fluorescence microscopy experiments. Still there are problems to overcome as the biomimetic membranes are not confined within the graphene squares during the protein insertion assay. Once this issue will be solved, this fabrication strategy will be extremely efficient in the realization of graphene-based membrane proteins arrays for sensing biomolecular interactions as for instance the presence of analytes in blood samples.

A major limitation of the results in Chapter 3 and Chapter 4 is the time-point of the gene expression analysis. For what concerns Chapter 3, future works needs to be carried out in fabricating graphene-oxide substrates to perform the whole differentiation of ASC towards the Schwann-like phenotypes on these substrates to evaluate the impact of these materials in the whole process. New functionalisation of graphene oxide will be established to direct adipose derived mesenchymal-stem cells differentiation into different phenotypes other than neuroglial differentiation. For what concerns Chapter 4, analyzing the gene expression on prolonged time-points can allow establishing if rGO and GO are promising material to sustain the Schwann-like differentiated adipose stem cells phenotype. 147

Concerning the results obtained in Chapter 6, future works needs to be performed on limiting the spreading of lipid biomimetic membranes within the graphene surface during the insertion of tail-anchoring protein. This issue can be solved either by increasing the hydrophilicity of the substrates to confine lipid mobility within the array surface or by causing less stress on the biomimetic membranes changing the insertion protocol. One possible change would be reducing the number of binding steps by direct functionalisation of the tail-anchoring protein with a fluorophore reporter. This functionalisation will allow not to use the primary and secondary antibody therefore reducing the number of binding steps.

148

Introduction to the Appendix

Appendix 1 is a peer-reviewed journal article where the author of this thesis contributed in the preparation and characterization of GO dispersions. In this study, GO dispersion were tested to check the effect of GO in differentiate cancer stem cells. Cancer stem cells are difficult to eradicate with conventional chemotherapeutic therapeutic approaches. Differentiation-based therapy consists in differentiating cancer stem cells into a less aggressive cellular phenotype which can be treated by radiotherapy or chemiotherapy. Results in Appendix 1 showed the positive effect of GO in decreasing the formation of tumorspheres under low-attachment conditions in an assay which measures the proliferation of cancer stem cells. Moreover, the expression of cancer stem cell markers was decreased after GO treatment.

Appendix 2 is a manuscript submitted to a peer reviewed journal where the author of this thesis contributed to the fabrication of graphene and graphene oxide samples. The aim of this study was to analyze theoretically and experimentally the structure and the thickness of biomimetic phospholipid membranes produced by lipid DPN on graphene and GO substrates. The molecular structure of the biomimetic membranes on graphene and GO in air and in liquid is crucial in the development of such graphene-based biological sensors.

149

www.impactjournals.com/oncotarget/ Oncotarget, Advance Publications 2015

Graphene oxide selectively targets cancer stem cells, across multiple tumor types: Implications for non-toxic cancer treatment, via “differentiation-based nano-therapy”

Marco Fiorillo1,2,3, Andrea F. Verre4, Maria Iliut4, Maria Peiris-Pagés1,2, Bela Ozsvari1,2, Ricardo Gandara1,2, Anna Rita Cappello3, Federica Sotgia1,2, Aravind Vijayaraghavan4 and Michael P. Lisanti1,2 1 The Manchester Centre for Cellular Metabolism (MCCM), Institute of Cancer Sciences, University of Manchester, UK 2 The Breakthrough Breast Cancer Research Unit, Institute of Cancer Sciences, University of Manchester, UK 3 The Department of Pharmacy, Health and Nutritional Sciences, The University of Calabria, Italy 4 School of Materials and National Graphene Institute, University of Manchester, UK Correspondence to: Michael P. Lisanti, email: [email protected] Correspondence to: Aravind Vijayaraghavan, email: [email protected] Keywords: nanomaterials, graphene oxide, cancer stem cells, multiple cancer types: breast, ovarian, prostate, lung, pancreatic, and glioblastoma (brain), differentiation therapy Received: January 01, 2015 Accepted: February 12, 2015 Published: February 24, 2015

This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT Tumor-initiating cells (TICs), a.k.a. cancer stem cells (CSCs), are difficult to eradicate with conventional approaches to cancer treatment, such as chemo-therapy and radiation. As a consequence, the survival of residual CSCs is thought to drive the onset of tumor recurrence, distant metastasis, and drug-resistance, which is a significant clinical problem for the effective treatment of cancer. Thus, novel approaches to cancer therapy are needed urgently, to address this clinical need. Towards this end, here we have investigated the therapeutic potential of graphene oxide to target cancer stem cells. Graphene and its derivatives are well-known, relatively inert and potentially non-toxic nano-materials that form stable dispersions in a variety of solvents. Here, we show that graphene oxide (of both big and small flake sizes) can be used to selectively inhibit the proliferative expansion of cancer stem cells, across multiple tumor types. For this purpose, we employed the tumor- sphere assay, which functionally measures the clonal expansion of single cancer stem cells under anchorage-independent conditions. More specifically, we show that graphene oxide effectively inhibits tumor-sphere formation in multiple cell lines, across 6 different cancer types, including breast, ovarian, prostate, lung and pancreatic cancers, as well as glioblastoma (brain). In striking contrast, graphene oxide is non- toxic for “bulk” cancer cells (non-stem) and normal fibroblasts. Mechanistically, we present evidence that GO exerts its striking effects on CSCs by inhibiting several key signal transduction pathways (WNT, Notch and STAT-signaling) and thereby inducing CSC differentiation. Thus, graphene oxide may be an effective non-toxic therapeutic strategy for the eradication of cancer stem cells, via differentiation-based nano- therapy.

INTRODUCTION consequence, they have been directly implicated in the disease pathogenesis of tumor recurrence and distant Cancer stem cells (CSCs) are resistant to metastasis [4, 5]. In addition, drug-resistant CSCs conventional therapeutic approaches [1-3]. As a have been linked to unfavorable clinical outcomes, www.impactjournals.com/oncotarget 1 Oncotarget across different tumor types [6-8]. As only a very RESULTS small percentage of cancer cells have “stem-like” and “tumor-initiating” properties, they are difficult to study and their key distinguishing features remain relatively GO flakes target breast cancer stem cells uncharacterized, although they appears to resist both chemo-therapy and radiation. Interestingly, CSCs share many properties with Here, we tested the efficacy of graphene oxide as a normal stem cells, including immortality and resistance potential new anti-cancer agent, for the selective targeting to stress, as well as asymmetric cell division [9, 10]. A of CSCs. Graphene, described as two-dimensional sheets particular distinguishing characteristic of CSCs is their of carbon atoms without any additional functional groups, ability to initiate tumors and to undergo anchorage- does not forms stable dispersions in water or other independent growth, when cultured in suspension biologically relevant solvents [17]. On the other hand, [11]. Under these particular cell culture conditions, graphene oxide [18] is the water-soluble derivative of CSCs proliferate and form 3D-spheroid-like structures, graphene which can be produced with various sizes and containing CSCs and progenitor cells, which are known bearing varied functional groups and which can be more as “tumor-spheres” or “onco-spheres” [12, 13]. In easily manipulated experimentally, especially in biological striking contrast, the vast majority of non-CSCs undergo systems. Sterile GO dispersions were prepared in a 5% a specialized form of apoptosis in suspension cultures, mixture of DMSO in double-distilled water for these called anoikis. Importantly, each 3D-spheroid originates studies. from the clonal proliferation of a single CSC, is not to due Initially, we tested the ability of GO to affect CSC to the self-aggregation of cancer cells. Thus, tumor-sphere proliferation, using MCF7 cells, a well-established ER(+) formation is an efficient means to selectively enrich for breast cancer cell line. Two grades of GO were used, CSCs. The CSC population is resistant to DNA-damage, small GO (s-GO) with flake sizes of 0.2 – 2 µm and big and shows lower levels of ROS production, as well [14, GO (b-GO) with flake sizes of 5 – 20 µm, to represent 15]. 3D-tumor-spheres derived from breast cancer cells are two size classes, where the flakes are either smaller or also known as mammo-spheres [12, 13]. larger than the target cells (Figure 1). For this purpose, we Clinically, there is an urgent need to identify new assessed the effects of graphene oxide on the anchorage- therapeutic strategies for selectively targeting CSCs. independent clonal expansion of MCF7 CSCs, using Here, we show that graphene oxide (GO) demonstrates the tumor-sphere assay. This functional assay directly this selectivity. As such, our current study provides a new measures CSCs proliferative expansion [12]. Figure 2 rationale for exploiting graphene oxide itself as an anti- shows the results of this analysis. Using s-GO flakes, we cancer therapeutic, rather than simply as a drug-delivery observed a dose-dependent inhibition of tumor-sphere agent [16]. formation, in the range of 1.25 to 25 µg/ml, with an IC-

Figure 1: Graphene Oxide (GO) grades. Left and Right panels show atomic force microscopy images of graphene oxide on a silicon dioxide substrate indicating the flake size distribution and monolayer thickness of the flakes. Inset shows a vial of b-GO stock dispersion in DMSO and water, at concentration of 2.3 mg/ml. www.impactjournals.com/oncotarget 2 Oncotarget Figure 2: Graphene Oxide (GO) selectively targets cancer stem cells (CSCs) in breast cancer cells. Upper Panels. Note that GO (big and small flakes) inhibits the anchorage-independent proliferation of MCF7 CSCs, as evidenced by inhibition of mammosphere formation. Lower Panels. In contrast, GO (big and small flakes) does not affect cell viability of the total MCF7 cell population. An * indicates p < 0.05 (Student’s t-test).

Figure 3: Graphene Oxide (GO) selectively targets cancer stem cells (CSCs) of multiple cancer cell types. GO (big flakes) inhibits the anchorage-independent proliferation of SKOV3 ovarian cancer cells (A), U87 glioblastoma cells (B), PC3 prostate cancer cells (C), A549 lung cancer cells (D), as well as pancreatic cancer cells (E), in a concentration-independent manner. These results indicate that GO inhibits sphere formations of multiple cancer types. An * indicates p < 0.05 (Student’s t-test). www.impactjournals.com/oncotarget 3 Oncotarget Table 1: Six Cancer Cell Models with Broad GO-based mechanistic studies: Effects on well- Applicability. established CSC signaling pathways

To gain mechanistic insights into how GO flakes target cancer stem cells, we next analyzed their effects on a series of well-established signal transduction pathways, which have been shown to contribute towards “stemness” [21-23]. For this purpose, we used a panel of MCF7 cell lines that were stably-transfected with different luciferase reporters, that allows one to quantitatively measure the activation-state of a given signal transduction pathway. Interestingly, Figure 7 shows that a number of signaling pathways were significantly inhibited by GO treatment. More specifically, GO treatment inhibited WNT- and Notch-driven signaling, as well as STAT1/3 signaling and 50 of ~12.5 µg/ml. Similarly, using b-GO flakes, we also the NRF2-dependent anti-oxidant response. However, observed a dose-dependent inhibition of tumor-sphere little or no effect was observed on TGF-beta/SMAD- formation, in the range of 6.25 to 100 µg/ml, again with an IC-50 of ~12.5 µg/ml. Importantly, both small and big GO flakes did not affect the viability of the bulk non-CSC population of MCF7 cells, indicating selectivity towards CSCs (Figure 2).

GO flakes target CSCs, across multiple cancer types

Since both the small and big GO flakes showed similar potency, we focused more on evaluating the efficacy of b-GO flakes. We next evaluated whether GO also showed efficacy against CSCs from multiple cancer types, such as ovarian, prostate, pancreatic and lung cancers, as well as glioblastoma (brain) (the six cell lines tested are summarized in Table 1). For simplicity, we tested b-GO flakes at two doses, namely 25 and 50 µg/ml. Figure 3 shows that b-GO flakes also effectively inhibited tumor-sphere formation in these 5 other cell lines. Thus, our results indicate that GO must be targeting a relatively specific and highly-conserved phenotypic property of CSCs, across multiple cancer types. Representative images of tumor-sphere inhibition by GO treatment are shown in Figure 4. Interestingly, the viability of bulk non-CSCs from these five cancer cell lines (SKOV3, PC3, A549, Mia PaCa2, and U-87) was not affected by GO (Figure 5), further highlighting its specificity and selectivity for CSCs. Importantly, b-GO flakes also did not affect the viability of a normal skin fibroblast cell line (hTERT- BJ1), indicating that GO is relatively non-toxic for normal body cells (Figure 6). This is consistent with the findings Figure 4: Graphene Oxide (GO) selectively targets of many other laboratories, i.e., that GO is non-toxic for cancer stem cells (CSCs) of multiple cancer cell types. multiple normal cell types [19, 20]. Representative images showing that GO (big flakes, 25μg/ml) inhibits the anchorage-independent proliferation of MCF7 breast cancer cells, SKOV3 ovarian cancer cells, PC3 prostate cancer cells, U87 glioblastoma cells, A549 lung cancer cells as well as MIA-PaCa-2 pancreatic cancer cells. www.impactjournals.com/oncotarget 4 Oncotarget signaling (Figure 7). and CD24), and quantitatively analyzed their expression Thus, it appears that GO treatment somehow targets by FACS analysis. The results of these studies are several different signal transduction pathways in cancer presented in Figure 8. cells, to reduce overall “stemness”. Briefly, MCF7 cells were treated as monolayer cultures with GO (50μg/ml) for 48 hours or left untreated GO promotes the differentiation of breast cancer (vehicle alone control). Then, cells were trypsinized and stem cells plated on low-attachment plates for 10 hours to induce anoikis and enrich for CSCs. Single cells were then analysed by FACS to quantitate the CD44(+)CD24-/ To further validate the idea GO was reducing low population, which represents the breast CSCs. As stemness in MCF7-derived CSCs, we used a series of predicted, the CD44(+)CD24-/low population is greatly well-established breast cancer stem cell markers (CD44 enriched after 10 hours in low-attachment conditions

Figure 5: Graphene Oxide (GO) does not affect cell viability of the total population of cancer cells. Cell viability was assessed using an SRB assay. Note that GO (big flakes) does not affect cell viability of the total cell population of SKOV3 ovarian cancer cells (A), U87 glioblastoma cells (B), PC3 prostate cancer cells (C), A540 lung cancer cells (D) as well as MIA-PaCa-2 pancreatic cancer cells.

Figure 6: Graphene Oxide (GO) does not affect the cell viability normal fibroblasts.Cell viability of hTERT-BJ1 fibroblasts was assessed using an SRB assay. Note that GO (big flakes) does not affect cell viability of the total cell population of normal fibroblasts. www.impactjournals.com/oncotarget 5 Oncotarget Figure 7: Graphene Oxide (GO) inhibits signaling pathways related to cancer stem cells, antioxidant responses and interferon. MCF7 breast cancer cells carrying luciferase-reporters (Cignal, QIAGEN) were generated to monitor the activation of a variety of signaling networks, including Wnt, STAT3, Notch, NRF2-dependent antioxidant responses, Interferonγ-STAT1 and SMAD- TGFβ pathways. MCF7-Luc reporter cells were treated with GO (big flakes) for 48 hours and luminescence was determined as a measure of pathway activation status. Note that GO inhibits cancer stem cell signaling (WNT, STAT3 and Notch), NRF2-antioxidant responses, as well as INFΥ-STAT1 signaling. No effects were observed for the SMAD-TGFβ-pathway. An * indicates p < 0.05; ** indicates p < 0.01 (Student’s t-test).

Figure 8: Graphene Oxide (GO) promotes the differentiation of breast cancer stem cells. MCF7 cells were treated as monolayer cultures with small or big GO (50μg/ml) for 48 hours or left untreated (vehicle alone control). Then, cells were trypsinized and plated on low-attachment plates for 10 hours to induce anoikis and enrich for cancer stem cells. Single cells were then analyzed by FACS to quantitate the CD44(+)CD24-/low population, which represents the cancer stem cells. A. Note that, as expected, the CD44(+)CD24-/ low population is greatly enriched after 10 hours in low-attachment conditions (vehicle alone control). B. Interestingly, GO does not reduce the total number of anoikis-resistant cells (data not shown), but rather induces the expression of CD24, thereby significantly reducing the CD44(+)CD24-/low population. This suggests that GO inhibits mammosphere formation by promoting the differentiation of breast cancer stem cells. An * indicates p < 0.05 (Student’s t-test). www.impactjournals.com/oncotarget 6 Oncotarget (vehicle alone control) (Figure 8A). Thus, GO may reduce the number of bonafide CSCs that Interestingly, GO does not reduce the total number are capable of forming tumor-spheres, by inducing their of anoikis-resistant cells (data not shown), but rather differentiation and inhibiting their proliferation. However, induces the expression of CD24, thereby significantly additional mechanistic studies are clearly warranted. reducing the CD44(+)CD24-/low population (Figure 8B). Importantly, our preliminary results indicate that This suggests that GO inhibits mammosphere formation GO treatment does not significantly affect oxidative by promoting the differentiation of breast cancer stem mitochondrial metabolism (OXPHOS) in this context cells, supporting our results from the analysis of multiple (data not shown), suggesting that GO does not target signal transduction pathways mitochondria. This is in contrast to our previous studies where a number of mitochondrially-targeted FDA- DISCUSSION approved antibiotics effectively eradiated CSCs [24]. Thus, GO and mitochondrially-targeted antibiotics appear Here, we show that treatment with GO is sufficient to work differently, via separate and distinct molecular to inhibit tumor-sphere formation in six independent mechanism(s). cancer cell lines, across multiple tumor types (breast, Also, since b-GO flakes are 5-to-20µm in size, ovarian, prostate, lung, and pancreatic cancer, as well they must be exerting their effects at the cell surface, as glioblastoma (brain cancer)). These results suggest that they are too large to be internalized within cells and are GO specifically targets a global phenotypic property of actually larger than a single cell. This is consistent with CSCs that is highly conserved in multiple tumor types. our findings that GO-treatment dampens the activation of Moreover, using MCF7 cells expressing a panel of several stem cell associated signal transduction pathways, luciferase reporters, we observed that GO treatment was which are initiated at the cell surface. This could then indeed sufficient to inhibit a number of different signal mechanistically induce CSC differentiation, which we transduction pathways, including WNT, Notch, STAT1/3 observed experimentally (summarized in Figure 9). and NRF-2, but did not effect TGB-beta/SMAD signaling. Previous studies have shown that GO (or its related Finally, using a panel of specific well-established breast derivatives) can inhibit “bulk” cancer cell migration CSC markers (namely CD44 and CD24), we show that or prevent tumor growth in xenograft models [25-28]. GO appears to reduce the number of CSCs by inducing However, none of these studies connected GO treatment their differentiation, as they now begin to express CD24. to the CSC phenotype or indicated that it could be used for “differentiation” therapy.

Figure 9: Graphene oxide (GO): Targeting cancer stem cells with differentiation-based nano-therapy. Our current mechanistic studies suggest that GO could directly be used as a therapeutic for targeting CSCs, because of its ability to induce differentiation. In this context, we might envision that GO could used to clear residual CSCs, with the aim of preventing tumor recurrence and distant metastasis, thereby providing a practical means for achieving “differentiation-based nano-therapy”. www.impactjournals.com/oncotarget 7 Oncotarget Interestingly, several studies have shown that Tumor-sphere culture GO is non-toxic for normal stem cells, and indeed GO promotes their differentiation. More specifically, it was A single cell suspension was prepared using demonstrated that culturing normal pluripotent stem cells enzymatic (1x Trypsin-EDTA, Sigma Aldrich, #T3924), on GO as a substrate induces their terminal differentiation and manual disaggregation (25 gauge needle) to create a towards multiple cell lineages, including neurons, single cell suspension. Cells were plated at a density of chondrocytes and adipocytes [29-33]. These properties are 500 cells/cm2 in mammosphere medium (DMEM-F12/ currently being actively exploited for tissue engineering B27/20ng/ml EGF/PenStrep) in non-adherent conditions, and regenerative medicine, by using GO as a scaffold for in culture dishes coated with (2-hydroxyethylmethacrylate) bone reconstruction and neural regeneration. (poly-HEMA, Sigma, #P3932) [12]. Cells were grown for Our new mechanistic studies suggest that GO could 5 days and maintained in a humidified incubator at 37°C directly be used as a therapeutic for targeting CSCs, at an atmospheric pressure in 5% (v/v) carbon dioxide/ possibly as a differentiation agent. In this context, we envision that GO could be delivered i.v. or p.o., as a new air. After 5 days for culture, spheres >50 μ were counted anti-cancer therapeutic, depending on the location of using an eye graticule, and the percentage of tumor- the tumor. Alternatively, GO flakes could also be used sphere formation was normalized to 100% for vehicle as a lavage solution during surgery, to clear the tumor alone control (1 = 100 % TSF) [12]. All experiments were excision site or the peritoneal cavity (as in ovarian or performed in triplicate, three times independently, such other peritoneal cancers) of residual CSCs, with the aim that each data point represents the average of 9 replicates. of preventing tumor recurrence and distant metastasis, via differentiation-based nano-therapy (Figure 9). Evaluation of CSC signalling pathways

MATERIALS & METHODS The Cignal Lenti reporter assay (luc) system (Qiagen) was chosen for monitoring the activity of several signal transduction pathways in MCF7 cells. Materials The responsive luciferase constructs encode the firefly luciferase reporter gene under the control of a minimal (m) CMV promoter and tandem repeats of response elements Cancer cell lines were purchased from the ATCC or for each pathway. The following constructs were used: other commercially available sources. Gibco-brand cell TCF/LEF(luc) for Wnt signal transduction (CLS-018L); culture media (DMEM/F12) was purchased from Life STAT3(luc) for transcriptional activity of STAT3 (CLS- Technologies. 6028L); RBP-Jk(luc) for Notch-induced signaling (CLS- 014L); ARE(luc) for Nrf2- and Nrf1-mediated antioxidant Graphene oxide response (CLS-2020L); GAS(luc) for Interferon gamma-induced Stat1-signal transduction (CLS-009L); Graphene oxide was prepared by using the Hummers SMAD(luc) for TGFβ-induced signal transduction (CLS- method with modifications [34, 35]. The individual 017L). Briefly, 1 x 105 MCF7-GFP cells were seeded in graphite oxide flakes contain carboxyl groups mainly 12-well plates. Once cells were attached, the viral particles at the edges, and epoxide, hydroxide and ketone groups were diluted 1:10 in complete culture media containing mainly on the basal plane. The C to O ratio is usually polybrene (sc-134220, Santa Cruz), and added to the cells. slightly lower or slightly higher than 1 as determined by Puromycin treatment (P9620, Sigma) was started 48 hours X-ray photoemission spectroscopy. The graphene oxide later in order to select stably infected cells. flakes of different sizes were separated by centrifuging graphene oxide suspensions at various rpm and Luciferase assays collecting different phases of the suspension. The AFM characterization of graphene oxide flakes was performed The Luciferase Assay System (E1501, Promega Kit) on a Bruker Dimension FastScan AFM system by using was used on all luciferase reporter MCF7 cells treated taping mode. The substrates were prepared by spin-casting with GO. Briefly, 6 × 103 MCF7-GFP cells were seeded the suspension on a Si/SiO2 substrate to yield monolayer in black-walled 96-well plates and then were treated with film, followed by AFM imaging. Concentrations were GO (50 μg/ml). As controls, vehicle-treated cells were run obtained from UV-Vis spectra, which were recorded in 10 in parallel. Four replicates were used for each condition. mm path length quartz cells using a PerkinElmer Lambda After 48 hours of treatment, luciferase assays were – 1050 UV-Vis-NIR spectrometer. The dispersions were performed according to the manufacturer’s instructions. diluted to give the absorption intensity lower than 1. Light signal was acquired for 2 minutes in photons/second in the Xenogen VivoVision IVIS Lumina (Caliper Life www.impactjournals.com/oncotarget 8 Oncotarget Sciences), and the results were analysed using Living resistance. Molecular cell. 2014; 54(5):716-727. Image 3.2 sofware (Caliper Life Sciences). Luminescence 4. Dawood S, Austin L and Cristofanilli M. Cancer Stem was normalized using SRB (to determine total cellular Cells: Implications for Cancer Therapy. Oncology. 2014; protein), as a measure of MCF7 cell viability. 28(12). 5. Colak S and Medema JP. Cancer stem cells--important Anoikis and CSC differentiation analysis players in tumor therapy resistance. The FEBS journal. 2014; 281(21):4779-4791. Following GO treatment, the CSC population was 6. Filipova A, Seifrtova M, Mokry J, Dvorak J, Rezacova M, enriched by seeding on low-attachment plates. Under Filip S and Diaz-Garcia D. Breast cancer and cancer stem these conditions, the non-CSCs undergo anoikis (a form of cells: a mini-review. Tumori. 2014; 100(4):363-369. apoptosis induced by lack of proper attachment) and CSCs 7. Scopelliti A, Cammareri P, Catalano V, Saladino V, are believed to survive. The expression of differentiation Todaro M and Stassi G. Therapeutic implications of Cancer markers by the surviving “CSC fraction” was analyzed by Initiating Cells. Expert opinion on biological therapy. 2009; FACS analysis. Briefly, 1 x 104 MCF7 cells were treated 9(8):1005-1016. with GO (50μg/ml) for 48h in 6-well plates, grown as a 8. Dean M. ABC transporters, drug resistance, and cancer stem monolayer. Then, the monolayer cells were trypsinized and cells. Journal of mammary gland biology and neoplasia. seeded in low-attachment plates in mammosphere media. 2009; 14(1):3-9. After 10h under low-attachment conditions, MCF7 cells 9. Reya T, Morrison SJ, Clarke MF and Weissman IL. were spun down and incubated with CD24 (IOTest CD24- Stem cells, cancer, and cancer stem cells. Nature. 2001; PE, Beckman Coulter) and CD44 (APC mouse Anti- 414(6859):105-111. Human CD44, BD Pharmingen cat.559942) antibodies for 10. Karsten U and Goletz S. What makes cancer stem cell 15 minutes on ice. Cells were rinsed twice and incubated markers different? SpringerPlus. 2013; 2(1):301. with LIVE/DEAD dye (Fixable Dead Violet reactive dye; Invitrogen) for 10 minutes. Samples were then analyzed 11. Magee JA, Piskounova E and Morrison SJ. Cancer stem by FACS (Fortessa, BD Bioscence). Only the live cells: impact, heterogeneity, and uncertainty. Cancer cell. population, as identified by the LIVE/DEAD dye staining, 2012; 21(3):283-296. was analyzed for CD24/CD44 expression. Data were 12. Shaw FL, Harrison H, Spence K, Ablett MP, Simoes BM, analysed using FlowJo software. Virtually identical results Farnie G and Clarke RB. A detailed mammosphere assay were also obtained using 7-AAD (7-Aminoactinomycin protocol for the quantification of breast stem cell activity. D; Life Technologies) to distinguish between the live and Journal of mammary gland biology and neoplasia. 2012; dead populations of cells (cell viability), during anoikis. 17(2):111-117. 13. Fillmore CM and Kuperwasser C. Human breast cancer cell ACKNOWLEDGEMENTS lines contain stem-like cells that self-renew, give rise to phenotypically diverse progeny and survive chemotherapy. We thank the University of Manchester for Breast cancer research : BCR. 2008; 10(2):R25. providing start-up funds that contributed to the success 14. Cojoc M, Mabert K, Muders MH and Dubrovska A. A role of this study (to Federica Sotgia and Michael P. Lisanti). for cancer stem cells in therapy resistance: Cellular and Aravind Vijayaraghavan, Maria Iliut and Andrea F. Verre molecular mechanisms. Seminars in cancer biology. 2014. were supported by Engineering and Physical Sciences 15. Skvortsov S, Debbage P, Lukas P and Skvortsova I. Research Council (EPSRC) grants EP/G03737X/1 and Crosstalk between DNA repair and cancer stem cell (CSC) EP/G035954/1. associated intracellular pathways. Seminars in cancer biology. 2014. REFERENCES 16. Liu Z, Robinson JT, Sun X and Dai H. PEGylated nanographene oxide for delivery of water-insoluble cancer 1. Sinha N, Mukhopadhyay S, Das DN, Panda PK and drugs. Journal of the American Chemical Society. 2008; Bhutia SK. Relevance of cancer initiating/stem cells in 130(33):10876-10877. carcinogenesis and therapy resistance in oral cancer. Oral 17. Hernandez Y, Lotya M, Rickard D, Bergin SD and Coleman oncology. 2013; 49(9):854-862. JN. Measurement of Multicomponent Solubility Parameters 2. Xin H, Kong Y, Jiang X, Wang K, Qin X, Miao ZH, Zhu for Graphene Facilitates Solvent Discovery. Langmuir. Y and Tan W. Multi-drug-resistant cells enriched from 2010; 26(5):3208-3213. chronic myeloid leukemia cells by Doxorubicin possess 18. Dreyer DR, Park S, Bielawski CW and Ruoff RS. The tumor-initiating-cell properties. Journal of pharmacological chemistry of graphene oxide. Chem Soc Rev. 2010; sciences. 2013; 122(4):299-304. 39(1):228-240. 3. Easwaran H, Tsai HC and Baylin SB. Cancer epigenetics: 19. Liao KH, Lin YS, Macosko CW and Haynes CL. tumor heterogeneity, plasticity of stem-like states, and drug Cytotoxicity of graphene oxide and graphene in human www.impactjournals.com/oncotarget 9 Oncotarget ACS Nano

This document is confidential and is proprietary to the American Chemical Society and its authors. Do not copy or disclose without written permission. If you have received this item in error, notify the sender and delete all copies.

Biomimetic Phospholipid Membrane Organization on Graphene and Graphene Oxide Surfaces: a Molecular Dynamics Simulation Study

Journal: ACS Nano

Manuscript ID nn-2016-07352p.R1

Manuscript Type: Article

Date Submitted by the Author: n/a

Complete List of Authors: Willems, Nathalie; Univ of Oxford, Biochem Urtizberea, Ainhoa; Instituto de Ciencia de Materiales de Aragón, CSIC and Universidad de Zaragoza, Verre, Andrea; University of Manchester, School of Materials Iliut, Maria; University of Manchester, School of Materials Lelimousin, Mickael; Centre de Recherches sur les Macromolecules Vegetales, Hirtz, Michael; Karlsruher Institut für Technologie (KIT), Institut für Nanotechnologie (INT) Vijayaraghavan, Aravind; University of Manchester, School of Materials Sansom, Mark; Univ of Oxford, Biochem; Univ of Oxford, Biochem

ACS Paragon Plus Environment Page 1 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 Biomimetic Phospholipid Membrane Organization on Graphene and 3 4 Graphene Oxide Surfaces: a Molecular Dynamics Simulation Study 5 6 † ‡ § § || 7 Nathalie Willems , Ainhoa Urtizberea, Andrea F. Verre, Maria Iliut, , Mickael Lelimousin , 8 Michael Hirtz, ‡,* Aravind Vijayaraghavan, §* and Mark S.P Sansom †* 9 10 11 † Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, United 12 13 Kingdom. 14 15 || CERMAV UPR5301, CNRS and Université Grenoble Alpes, BP 53, 38041 Grenoble cedex 9, 16 17 France 18 19 ‡ Institute of Nanotechnology (INT) and Karlsruhe Nano Micro Facility (KNMF), Karlsruhe 20 21 Institute of Technology (KIT), 76344 EggensteinLeopoldshafen, Germany. 22 23 24 § School of Materials and National Graphene Institute, University of Manchester, Manchester M13 25 9PL, UK 26 27 28 ◊ Physical Institute and Center for Nanotechnology (CeNTech), Westfälische WilhelmsUniversität, 29 48149 Münster, Germany 30 31 32 33 34 35 *corresponding authors: email: [email protected] & [email protected] 36 37 38 39 40 ACS Nano. ms. nn201607352p 41 42

43 44 45 KEYWORDS: Molecular Dynamics, Phospholipid Bilayer, Supported Lipid Membranes, DipPen 46 47 Nanolithography, Polymer Pen Lithography 48 49 50 51 52 53 54 55 56 57 58 59 1 60 ACS Paragon Plus Environment ACS Nano Page 2 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 ABSTRACT: 3 4 Supported phospholipid membrane patches stabilized on graphene surfaces have shown potential in 5 6 sensor device functionalization, including biosensors and biocatalysis. Lipid dippen 7 nanolithography (LDPN) is a method useful in generating supported membrane structures that 8 9 maintain lipid functionality, such as exhibiting specific interactions with protein molecules. Here, 10 11 we have integrated LDPN, atomic force microscopy, and coarsegrained molecular dynamics 12 simulation methods to characterize the molecular properties of supported lipid membranes (SLMs) 13 14 on graphene and graphene oxide supports. We observed substantial differences in the topologies of 15 16 the stabilized lipid structures depending on the nature of the surface (polar graphene oxide vs non 17 polar graphene). Furthermore, the addition of water to SLM systems resulted in largescale 18 19 reorganization of the lipid structures, with measurable effects on lipid lateral mobility within the 20 21 supported membranes. We also observed reduced lipid ordering within the supported structures 22 relative to freestanding lipid bilayers, attributed to the strong hydrophobic interactions between the 23 24 lipids and support. Together, our results provide insight into the molecular effects of graphene and 25 26 graphene oxide surfaces on lipid bilayer membranes. This will be important in design of these 27 surfaces for applications as biosensor devices. 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 2 60 ACS Paragon Plus Environment Page 3 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 1 2 2 Dippen nanolithography (DPN) with phospholipids (LDPN) and the use of polymer pen 3 lithography (PPL) 3 and other stamping techniques for the generation of lipid membranes on 4 5 supports 4–6 have gained increasing interest in recent years 7. Supported lipid bilayers have diverse 6 8–11 12–17 7 applications in biomedical research , sensor and device functionalization , protein 8 crystallization 18 and in generating lipid sensor structures.18–20 In LDPN, the tip of an atomic force 9 10 microscope (AFM) is covered with phospholipid and brought into close contact with a solid 11 12 support, allowing the lipid ink to transfer to the substrate and selfassemble into stacks of 13 membranes 21 . This method has the benefit of direct and precise spatial control during lipid 14 15 deposition onto the support surface, with the ability to tailor the lipid mixture in the ‘ink’ to a 16 17 desired composition. Other methods of lipid deposition, including imprinting, have proved useful in 18 elucidating the spreading behavior of lipid membranes on support surfaces of varying degrees of 19 20 hydrophilicity and roughness, and have indicated that different spreading mechanisms and 21 22,23 22 velocities may occur depending on the nature of the surface. An interesting difference when 23 generating membrane patches by LDPN compared with other methods (e.g. vesicle fusion 24 , 24 25 micropipettes 25,26 , or spreading membranes in microfluidic systems 27,28 ) is that the fabrication of the 26 27 lipid patches generally takes place in air ( i.e. under ambient conditions). Therefore, the structures 28 will usually be transferred into liquid only after the lithographic step is completed, as most 29 30 applications take place in liquid phase. This suggests that the lipid structures may rearrange, and 31 32 different structural models have been proposed for the molecular organization of the lipid 33 membrane in air and water and on surfaces with varying hydrophilicity.13,28–30 34 35 36 Computational methods such as molecular dynamics (MD) simulations have been used to study 37 lipid membrane organization and dynamics on different support materials 31–35 . Coarsegrained (CG) 38 39 simulations of lipid bilayers on hydrophilic model surfaces indicate that preformed lipid patches 40 41 are more ordered on these surfaces compared to freestanding bilayers, affecting lateral lipid 42 diffusion in the lipid leaflet that directly contacts the support, and resulting in reduced lipid 43 44 mobility. 35 Increasing surface “roughness” reduces lipid ordering, and results in higher lipid 45 46 mobility relative to smooth surfaces, confirming that lipid membrane organization is directly 47 influenced by the underlying surface topology of the support.35 Additionally, CG simulations 48 49 modeling lipid selfassembly on supports of differing hydrophilicity suggest that lipid diffusion is 50 51 more strongly affected, and reduced, on hydrophilic supports compared to on hydrophobic supports, 52 due to stronger attractive forces.36 The nature of the supporting surface is thus important. Recent 53 54 studies have focused on parameterizing models of various support materials, for example graphite, 55 56 to more accurately represent interactions of organic molecules, such as longchain alkanes, with the 57 58 59 3 60 ACS Paragon Plus Environment ACS Nano Page 4 of 36 2303214_File000005_39618591.docx 1/13/2017

1 37 2 support. These models were shown to reproduce phase transitions and molecular organization of 3 organic molecules on the surface, in line with experimental data. 31,37 4 5 6 MD simulations are therefore useful in studying the molecular properties of lipidsurface 7 interactions, allowing microscopic details to be characterized. In this study, we examine the 8 9 organization of supported lipid bilayers as generated by LDPN or PPL using CG molecular 10 11 dynamics (CGMD) simulations, and correlate the results with experimental evidence. Our 12 simulation results are in close agreement with AFM measurements of lipid layers on graphene and 13 14 graphene oxide supports. The AFM experiments suggested that inverted bilayer configurations are 15 16 formed on pristine graphene, whilst altered lipid structures can be formed on a graphene oxide 17 surface. Our simulations reproduce this behavior, capturing the dynamic process underlying the 18 19 altered lipid configurations. Furthermore, the simulations revealed effects on lipid ordering and 20 21 diffusion within the bilayer structures relative to freestanding bilayers in water, depending on the 22 support surface, Together, the combined experimental and simulation results provide an integrated 23 24 study of the dynamical properties of lipid membranes on graphenebased supports. 25 26 27 28 Results: 29 30 Previous studies have shown that the formation of stable lipid structures, such as multilayered lipid 31 32 membranes, on graphene, silicon dioxide, and other substrates can be generated by LDPN, and are 33 13,38,39,40 34 influenced by surface morphology and experimental conditions. Here, we used CGMD 35 simulations to investigate the molecular details of lipid interactions with pristine graphene and with 36 37 graphene oxide surfaces. The simulated systems were constructed to match the experimental 38 39 conditions used during LDPN and AFM studies as well as possible. These experiments were either 40 performed for systems in air or systems in solution. 13 Experiments in air were typically performed 41 42 at 2030% relative humidity (R.H). This corresponds to < 1 CG water particle within the volume of 43 3 44 the simulation box at 2030% R.H, even for the largest systems simulated (30 x 20 x 20 nm ). The 45 experiments in air were therefore reasonably well approximated by simulations in vacuum. 46 47 48 Lipid interactions with graphene 49 50 LDPN generated lipid structures on pristine graphene surfaces in air are thought to form “inverted” 51 bilayer structures.13,39 We investigated the stability of these inverted bilayer structures on a 52 53 graphene surface in vacuum and in water using CGMD. Two different graphene surface areas were 54 2 2 55 modeled: a small (10 x 10 nm ) and large (30 x 30 nm ) surface. Preformed inverted bilayers 56 consisting of 512 1,2dioleoylsnglycero3phosphocholine (DOPC) molecules (256 lipids per 57 58 leaflet) were individually placed above the small and larger graphene surfaces and simulated for 0.5 59 4 60 ACS Paragon Plus Environment Page 5 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 – 1 µs. Simulations of the systems in vacuum (4 replicate simulations each of 4 µs duration) 3 resulted in rapid adsorption of the lipid layers on the surface, mediated by strong hydrophobic 4 5 interactions between the graphene surface and the lipid tails (Fig. 1A). The lipid structures 6 7 maintained their inverted configurations (Fig. 1C), as shown by monitoring the center of mass 8 (COM) distance between the lipid phosphate headgroup and the graphene surface (Fig. 1E). Visual 9 10 inspection also indicated some degree of enhanced ordering of the lipid tails in the graphene 11 12 proximal layer. Polar interactions between the headgroups resulted in clustering of the lipid head 13 groups, stabilizing the inverted bilayer topology. Furthermore, CGMD simulations of lipid 14 15 monolayers with the small graphene surface (again, 4 replicate simulations each of 4 µs 16 17 duration)resulted in spontaneous lipid reorganization to form an inverted bilayer structure in 18 vacuum (Fig. 1B, D and F). 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 5 60 ACS Paragon Plus Environment ACS Nano Page 6 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Figure 1. CGMD simulations of the interaction of (A) a preformed inverted bilayer and (B) a lipid 47 2 48 monolayer with a small (10 x 10 nm ) graphene surface in vacuum. The lipid choline, phosphate, 49 glycerol and carbon groups are shown as licorice representations colored in green, red, yellow and 50 grey respectively. The graphene surface is shown as black van der Waals spheres. The Hyperballs 51 41 52 program was used for image generation. Partial density profiles were calculated for the last 25% 53 of the simulation time for the inverted bilayer ( C) and monolayer ( D) simulations. (E) and (F) show 54 the time evolution of the average COM distance between the lipid phosphate headgroups and the 55 graphene surface (dashed line) for the inverted bilayer and monolayer simulations in ( A) and ( B) 56 57 respectively. 58 59 6 60 ACS Paragon Plus Environment Page 7 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 These results corroborate AFM measurements of LDPN deposited lipid layers on pristine graphene 3 surfaces, 13,39 revealing that phospholipids form a flat inverted bilayer of uniform thickness (Fig. 2A) 4 5 on the hydrophobic support in vacuum (simulation) or air (experiment). The measured thickness of 6 7 the inverted bilayer structure from the simulations compares well to the height measurements from 8 the AFM experiments, corresponding to ~ 4 nm. This can be seen from the density peaks of lipid 9 10 tail groups as shown in the partial density profiles in Fig. 1B and D. This matches the average 11 12 height of ~44.5 nm of the inverted bilayer as measured by AFM (Fig. 2A). 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Figure 2 . AFM images of LDPN generated lipid membranes (DOPC) on pristine graphene (A) and 46 graphene oxide (B) surfaces in air. (C, D ) AFM height measurements of the same patches measured 47 48 between the red dots shown on the left images. The top section profile shows a smooth single step 13 49 membrane compatible with the thickness of a single inverted bilayer on graphene (in air). The 50 bottom profile section reveals a bilayer membrane on top of a lipid monolayer (wetting layer). The 51 wetting layer is similar to the layers observed on silicon dioxide, though thinner, probably due to 52 28 53 reduce layer density. 54 However, the inverted bilayer topology was less stable (in terms of maintenance of the inverted 55 56 bilayer structure after interactions with the support surface) in CGMD simulations with the large 57 2 2 58 graphene surface (30 x 30 nm ) compared to the small graphene surface (10 x 10 nm ). In 59 7 60 ACS Paragon Plus Environment ACS Nano Page 8 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 simulations with the large graphene surface, the strong hydrophobic interactions between the lipid 3 tails and graphene beads dominated lipid organization, resulting in disassembly of the layer and 4 5 spreading of lipids across the available surface area, aligning with the surface in a lateral fashion 6 7 (Fig. S2A). The same behavior was also observed in simulations of preformed regular bilayers with 8 the large graphene surface (Fig. S2C). This contrasts with our initial observations of stable inverted 9 10 bilayer configurations during simulations with the small graphene surface area, which did not 11 12 disassemble. The difference between these simulations is the overall surface area of the graphene 13 model. The preformed inverted lipid structures are roughly 10.5 x 10.5 nm 2 in x,y dimensions 14 15 (prior to simulation), and thus occupy a slightly larger surface area than the small graphene surface 16 2 17 (10 x 10 nm ). Therefore, the larger density of lipids on this small graphene surface may affect the 18 stability of the observed lipid configuration compared to the large graphene surface, resulting in 19 20 maintenance of the inverted structure. Experimentally, inverted bilayer patches generated by L 21 2 22 DPN are stable on more extended graphene surfaces (20 x 20 µm ), for which lipid spreading is 23 thought to be a selflimiting process. 39 Consequently, the difference in the lipid density on the large 24 25 graphene surface areas within the AFM experiment compared the simulations may be a deciding 26 27 factor in the observed disassembly of the inverted bilayer structure (512 lipids) on the large 28 graphene surface in the simulation. Interestingly, simulation of larger inverted bilayers (2110 lipids) 29 30 exhibited higher stability and maintained the inverted configuration to a larger extent compared to 31 32 the smaller bilayer systems (512 lipids) (Fig. S2E). However, lipid tail entanglement and lateral 33 interactions with the surface, as well as headgroup clustering, resulted in deviation from the more 34 35 "ideal" inverted bilayer configuration observed in simulations of the smaller bilayers on the small 36 2 37 graphene surface in vacuum (10 x 10 nm ;Fig. 1). Importantly, the increased stability of larger lipid 38 layers suggests that lipid density on the graphene surface is another determinant of the topology of a 39 40 supported membrane. 41 42 Lipid interactions with graphene oxide 43 44 Surface polarity is known to affect the molecular properties of supported lipid bilayers, with effects 45 46 on topology 29,30 , diffusion 42 and lipid order 36,43 . To investigate these effects, LDPN was also 47 48 performed on graphene oxide substrates in air. AFM height measurements of the deposited lipids on 49 the hydrophilic substrate indicated that the lipids organized into a “1.5” bilayer configuration (Fig. 50 51 2B), as has also been suggested for lipid structures on silicon dioxide surfaces 13 . Two distinct lipid 52 53 layers could be distinguished: a “wetting” layer composed of phospholipids with their headgroups 54 oriented towards the hydrophilic support surface, and a second inverted bilayer that formed on top 55 56 of this wetting layer. The same lipid organization is seen in CGMD simulations (of duration 5 µs) 57 58 of (noninverted) bilayers with graphene oxide in vacuum. The lipid bilayer interacted with the 59 8 60 ACS Paragon Plus Environment Page 9 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 surface, undergoing substantial reorganization within ~ 0.2 µs to form a 1.5 bilayer on top of the 3 oxidised surface (Fig. 3A). Simulation (also of duration 5 µs) of a preformed inverted bilayer 4 5 configuration positioned above the oxide surface converged to a similar 1.5 bilayer structure (Fig. 6 7 3B), indicating that the outcome was robust to the initial bilayer (inverted vs. noninverted) model 8 used. The significant rearrangements facilitating formation of the 1.5 bilayer structure are thought 9 10 to be driven by polar headgroup interactions with the hydrophilic graphene oxide surface. Partial 11 12 density profiles calculated for the last 25% of CGMD simulation time show very similar peaks in 13 lipid headgroup and tail densities for both the inverted and noninverted bilayer systems (Fig. 3C 14 15 and D), confirming that the differing starting structures converged to similar end structures. These 16 17 results support the interpretation of a 1.5 bilayer configuration by AFM height measurements for 18 supported lipid membranes on graphene oxide, suggesting that this arrangement is the preferred 19 20 molecular state of lipid layers on polar oxide surfaces in air. 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 9 60 ACS Paragon Plus Environment ACS Nano Page 10 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 Figure 3. CGMD simulation of a preformed (A) regular ( i.e. noninverted) bilayer or (B) inverted 42 lipid bilayer on a small (10 x 10 nm 2) graphene oxide surface in vacuum. Color scheme is the same 43 as in Fig. 1. The graphene oxide surface is shown in black (carbons) and red (oxygens). (C) and (D) 44 45 show partial densities of the lipid tail and phosphate headgroups in the simulation box, calculated 46 for the last 25% of simulation time from the simulations shown in (A) and (B) . The dashed line 47 represents the position of the graphene oxide surface in the simulation box. 48 49 50 51 In contrast, the 1.5 lipid bilayer configuration (which exposes the hydrophobic tails of the lipid 52 molecules at the ‘uppermost’ layer) was not observed in CGMD simulations of the lipidgraphene 53 54 oxide systems in water. Instead, preformed lipid bilayers reorganized to form bicellelike 55 56 configurations on both small and larger graphene oxide surfaces in water (Fig. 4EH). The 57 rearrangement of inverted bilayers was driven by lipid headgroup interactions with the underlying 58 59 10 60 ACS Paragon Plus Environment Page 11 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 support, either directly or through bridging water particles, as well as by interactions with the 3 surrounding solvent, resulting in a stable bicelle structure which does not subsequently move as a 4 5 whole on the graphene oxide surface. Conversely, the addition of water to lipidcovered graphene 6 7 systems in the CGMD simulations resulted in destabilization of the previously formed inverted 8 bilayers (formed in vacuum), and triggered disassembly of the structures (Fig. 4AD). 9 10 Subsequently, the lipids spread over the available surface area with headgroups oriented toward the 11 12 surrounding solvent molecules. Importantly, similar behavior is observed for lipidcovered 13 graphene transferred to aqueous solution by AFM measurements. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 11 60 ACS Paragon Plus Environment ACS Nano Page 12 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 Figure 4. Images of the first and final frames of CGMD simulations of preformed inverted (A, C) 50 and regular (noninverted) lipid bilayers (B, D) with 30 x 30 nm 2 graphene ( A, B ) and graphene 51 oxide ( C, D) surfaces in water (shown as a light blue background for clarity). Partial density 52 profiles were calculated for the last 25% of the simulation time shown for the graphene ( E, G ) and 53 graphene oxide ( F, H ) systems. The same color scheme is used as in Fig. 1. The graphene oxide 54 surface is shown in black (carbons) and red (oxygens). The graphene surface is shown in black. 55 56 57 58 59 12 60 ACS Paragon Plus Environment Page 13 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 Two observations can be made from the combined simulation and experimental data. First, the 3 addition of aqueous solvent can result in different lipid configurations on the same surface e.g. 4 5 bicellelike formations on graphene oxide in an aqueous environment compared to a 1.5 bilayer 6 13,21,28 7 configurations in air (or vacuum). Second, the polarity difference resulting from the oxygen 8 containing functional groups in graphene oxide compared to the hydrophobic surface presented by 9 10 pristine graphene resulted in stabilization of different lipid configurations e.g . 1.5 layers vs inverted 11 12 bilayers. These observations imply that the configuration of the supported membrane can be tuned 13 as a function of system solvation as well as underlying surface polarity, resulting in relatively 14 15 different end structures. 16 17 Direct observation of lipid reorganization on graphene and silicon dioxide in liquid by AFM has 18 19 been reported by a previous study. 13 These observations necessitated the use of bovine serum 20 21 albumin (BSA), which binds to the surface, to block excessive lipid spreading and stabilize the 22 membrane patches in aqueous solution. In this study, direct AFM measurements of the lipid layer 23 24 structures on graphene oxide in liquid has been hindered by the blocking layers of BSA needed to 25 26 stabilize the small membrane patches during scanning. Given that specific binding interactions in L 27 DPN applications are usually conveyed by functionalization at the lipids headgroup, it was inferred 28 29 that the lipid patches undergo reorganization on graphene oxide, exposing the (otherwise buried 30 31 functional groups) to the liquid phase, similar to the lipid configurations identified from the CG 32 simulations of the graphene oxide systems in water.13,44 33 34 Lipid layer properties on graphene and graphene oxide: lipid order parameters 35 36 37 So far, we have provided an overview of the topological differences generated in lipid organization 38 on surfaces of varying polarity and system solvation. How does this affect the dynamic properties of 39 40 the lipids within the observed structures? Calculation of lipid order parameters (S) as a function of 41 42 simulation time provides a metric for lipid rearrangements on a support surface (see Methods). 43 Briefly, the lipid order parameter is calculated by estimating the angle between the bonds 44 45 connecting the lipid particles and the zaxis of the simulation box ( i.e. the normal to the bilayer). 46 47 Thus alignment of the lipid particles with the zaxis yields an order parameter of S = 1, whereas a 48 value of S = 0 corresponds to randomly orientated lipids. Order parameters were calculated for the 49 50 PC lipids of selected systems and compared to those for a freestanding bilayer (i.e . no graphene 51 52 surface) simulated in water (Table S1). 53 54 Systems that underwent substantial lipid reorganization showed major changes in the lipid order 55 parameters over time, particularly for the simulations starting from an inverted bilayer on the 56 57 graphene oxide surface in vacuum (Fig. 5). In this system, the average order parameter values 58 59 13 60 ACS Paragon Plus Environment ACS Nano Page 14 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 change significantly during the first 2 µs of simulation time, particularly for the phosphateglycerol 3 bond, and consecutive bonds for both sn1 and sn2 lipid tails up to the third CG particle. The 4 5 changes represent altered configurations of the lipids as the molecules reorganized to form the 1.5 6 7 bilayer structure. Specifically, the average order parameters for particular lipid bonds decrease from 8 around 0.3 to ~0.1 during the first microsecond of the simulation time, indicating a random 9 10 arrangement for many lipids during formation of an inverted bilayer on top of the monolayer 11 12 (wetting layer). Towards the end of the rearrangement (~2 µs), the average order parameters for e.g. 13 the phosphateglycerol bond returns to 0.3, suggesting lipid alignment with the zaxis following the 14 15 transition (Fig. 5). Other order parameter values, such as the C2C3 bond however, did not 16 17 completely recover their initial value, which may be attributed to a degree of bilayer distortion 18 brought about by micellelike lipids within the bilayer formed on top of the wetting layer (Fig. 3A). 19 20 This value (~0.1) is reflected in those for subsequent bonds ( e.g. C3C4) in both sn1 and sn2 21 22 chains, characterizing the random arrangement of lipids tails within the 1.5 bilayer structure on 23 graphene oxide. 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 14 60 ACS Paragon Plus Environment Page 15 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Figure 5. Lipid order parameters calculated from a CGMD simulation of an inverted bilayer 43 graphene oxide system in vacuum. The upper panel (A) shows the position of all the lipid phosphate 44 groups in the system (shown as transparent red van der Waals spheres), as well as the position a 45 single typical lipid in the layer, at different time frames during the simulation. The middle panel (B) 46 47 shows the alignment of this single lipid with the zaxis of the simulation box, as the lipid layer re 48 organized over the course of the simulation. The same color scheme as Fig. 1 is used for the lipids. 49 The graphene oxide surface is shown in black (carbons) and red (oxygens). (C) Average lipid order 50 51 parameters as a function of time. These were calculated for every bond between two consecutive 52 beads in the DOPC lipid (B nBn+1 , where B = bead) compared to the zaxis of the simulation box, 53 i.e. the normal to the graphene oxide plane. The initial inverted configuration of the bilayer is 54 labelled I. 55 56 57 58 59 15 60 ACS Paragon Plus Environment ACS Nano Page 16 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 In general, the average lipid tail order parameters for all systems suggest that the tails were less 3 ordered on both graphene and graphene oxide surfaces compared to a freestanding bilayer in water 4 5 (Fig. 6; Table S1). For example, an average order parameter of 0.18 is obtained for lipid tails in an 6 7 inverted bilayergraphene simulation in vacuum (small surface area), compared to 0.36 for lipids 8 within the freestanding bilayer. However, inspection of the time evolution of the order parameters 9 10 from the inverted bilayergraphene simulation in vacuum reveal a slight gain in order over the last 11 12 0.5 µs of simulation time, particularly for the sn2 chain of the lipids ( e.g C1BC5B) (Fig. S3A). 13 This transition indicates the increased alignment of the lipid tails, as the inverted bilayer 14 15 configuration becomes more stable on the pristine graphene surface. Furthermore, simulations of 16 17 different starting lipid structures with graphene oxide resulted in similar order parameter values, 18 particularly towards the end of the simulations, further indicating convergence to the 1.5 bilayer 19 20 organization on this surface in vacuum (Fig. S3C and D). The overall values compare well with 21 36 22 lipid tail order parameters calculated from CG simulations of related systems. 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 Figure 6. Average lipid order parameters compared with a freestanding bilayer in water (free bil; 41 horizontal broken line). S = small, L = large, GO = graphene oxide, G = graphene, inv = inverted 42 43 bilayer, reg = regular (i.e. noninverted) bilayer, vac = vacuum, wat = water. 44 45 46 The overall reduced lipid order on the graphene (oxide) supports could be a result of lateral 47 48 interactions between the lipid tails and the surface (partial alignment of the lipid tails with the 49 50 surface), as is observed in the pristine graphene systems (due to strong hydrophobic forces), and/or 51 headgroup interactions with the graphene oxide surface, disrupting lipid order in the layer. 52 53 54 Lipid layer properties on graphene and graphene oxide: Lipid diffusion 55 56 Characterization of the lateral diffusion of lipids within supported membrane systems is 57 fundamental to understanding the dynamic properties of phospholipids within these membranes. 45– 58 59 16 60 ACS Paragon Plus Environment Page 17 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 48 2 Experimental studies report both linear and anomalous diffusion regimes of lipids depending on 3 the systems, and increasingly highlight the importance of the time and length scale over which the 4 5 diffusion data are collected when making a distinction between anomalous and linear 6 42,46,47,49 7 diffusion. To characterize lateral diffusion of lipids within our simulated supported bilayer 8 systems, both linear and anomalous diffusion models were considered. Mean square displacements 9 10 (MSDs) were collected by tracking the displacement of lipids (see Methods). The resultant MSDs 11 50 51 12 were then fitted with either a linear or an anomalous diffusion equation. 13 14 This analysis revealed differences in the lateral displacement of lipids on surfaces relative to those 15 16 freestanding bilayers. The diffusion of lipids within supported membranes, on either graphene or 17 graphene oxide, was slower in the presence of the support compared to freestanding bilayers (Fig. 18 19 7A).52 The presence of water appeared to result in increased lipid diffusion on the graphene oxide 20 21 surface (Fig. 7B), suggesting the water layers could act as a ‘lubricant’ for lipid diffusion within the 22 layer. Indeed, bridging water particles between the lipids and the support surface were often 23 24 observed within the simulations with this surface. Similar effects have also been observed in 25 46 26 experimental systems of supported lipid membranes on hydrophilic surfaces in aqueous solution. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 17 60 ACS Paragon Plus Environment ACS Nano Page 18 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Figure 7. MSDs (mean squared displacements) of lipids within a graphene oxide supported 34 membrane compared with a free lipid bilayer. The supported bilayer data correspond to a CGMD 35 36 simulation of a regular (noninverted) bilayer with a small graphene oxide surface in vacuum (A) 37 and in water (B) , as well as a freestanding bilayer in water (Bil). MSD data was sampled at time 38 sampling windows ranging from 0.2200 ns (xaxis). The data was then fitted with a linear or 39 40 anomalous diffusion equation, and the resultant diffusion coefficients (D) and anomalous exponent 7 2 7 2 α 41 (α) values are shown in the plots (units of D are: x 10 cm /s for linear diffusion, x 10 cm /s for 42 anomalous diffusion). Errors bars indicate standard deviations. 43 44 45 46 Furthermore, fitting MSD data using the anomalous diffusion equation produced α values of 0.7 to 47 48 0.9 for all the supported membrane systems, suggesting that the lipids within the supported 49 membranes may exhibit different diffusion regimes as a consequence of their interactions with the 50 51 surface (Fig. 8). In particular, lipids exhibited subdiffusion for all supported membrane 52 53 simulations. 54 55 56 57 58 59 18 60 ACS Paragon Plus Environment Page 19 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Figure 8. (A) Diffusion coefficients calculated from either linear or anomalous fits of MSD data for 38 39 each of the simulated systems in this study . (B) The α values calculated for the anomalous diffusion 40 coefficients shown in (A). The dotted line represents the α value calculated from the anomalous 41 diffusion fit for a CG simulation of the freestanding bilayer in water. 42 43 44 45 Importantly, the displacement of lipids within the freestanding bilayer were broadly consistent with 46 7 2 47 a linear diffusion model, producing a value of D = 3.9 x 10 cm /s (± 0.5). This value should be 48 53 49 scaled by a factor of 4 to account for the reduced degrees of freedom within CG simulations in 7 2 50 order to compare with experimental estimates, resulting in Dlin = 0.98 x 10 cm /s, which compares 51 7 2 54,55 52 well to atomistic simulations of DOPC bilayers in water (1.5 x 10 cm /s). However, it does 53 54 remain challenging to compare diffusion data for membranes on surfaces with experimental values, 55 or even other simulation studies, given the dependence on both the time scale over which diffusion 56 42,45,47,49 57 is measured, and the length scale of the system. Nonetheless, we can reasonably conclude 58 59 19 60 ACS Paragon Plus Environment ACS Nano Page 20 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 from the MSD data and fits across all of the supported membrane systems (see SI Fig. S56) that in 3 general D is lower for lipids in the supported bilayers relative to a freefloating bilayer, with the 4 5 largest reduction (more than twofold, with α = 0.7) for a membrane on pristine graphene or on 6 7 graphene oxide in vacuum. 8 9 10 11 Discussion: 12 13 The dynamic organization of lipid membranes on graphene supports has been investigated using 14 15 both experimental and computational techniques. As has been observed in a number of other studies 16 17 of supported lipid membranes, the dynamic properties of the lipid configurations presented here 18 were affected by the polar/apolar nature of the surface, as well as system hydration.32,34–36,56 AFM 19 20 measurements suggested the formation of inverted bilayer topology on pristine graphene, whilst a 21 13 22 1.5 lipid structure was observed on graphene oxide in air (Fig. 2). CG simulations of these 23 systems supported initial observations of the differing lipid topologies on the different surfaces. 24 25 Encouragingly, different preformed starting lipid structures spontaneously rearranged to converge 26 27 to the same end structure as observed by the AFM studies. For example, regular bilayers formed 1.5 28 layers on graphene oxide, whilst single lipid monolayers formed inverted bilayer structures on 29 30 pristine graphene, converging to the observed end structures despite different starting configurations 31 32 (Fig. 1 and 3). Furthermore, the CG simulations suggested that stability of the lipid organization 33 was related to lipid density on the surface. Whilst small inverted bilayers were observed to 34 35 disassemble on large graphene surfaces due to strong hydrophobic interactions with the surface, 36 37 larger bilayers were more stable (Fig. S2). Importantly, initial AFM characterization of 38 phospholipid patches on pristine graphene in air indicated that lipids are more mobile on this 39 40 surface than on hydrophilic surfaces such as silicon dioxide, also attributed to the strong 41 13,39 42 hydrophobic interactions between lipid hydrocarbon tails and graphene. Spreading of a very 43 small membrane patch on a large graphene surface is thus perhaps not unexpected. 44 45 System hydration was an important factor affecting lipid stability and organization on support 46 47 surfaces. Specifically, lipid organization on graphene oxide is affected by the addition of water, 48 49 resulting in formation of stable bicellelike structures on graphene oxide surfaces, contrasting the 50 preferred 1.5 bilayer lipid organization observed on the same surface in vacuum (simulation) or in 51 52 air (AFM). Given the apparent high stability of bicelle structures on graphene oxide in water, 53 54 patterning of hydrophilic surfaces in a discrete manner should be possible in aqueous support 55 systems. A recent study investigating lipid localization across lipid monolayerbilayer junctions 56 57 exploited precisely this topological difference in lipid layering, which results from the different 58 59 20 60 ACS Paragon Plus Environment Page 21 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 57 2 energies of interaction of lipids with surfaces of varying hydrophobicity. Thus, lipid monolayer 3 bilayer junctions were established by patterning glass surfaces with hydrophobic molecules, 4 5 resulting in formation of lipid monolayers on the hydrophobic sections and lipid bilayers on the 6 57 7 surrounding hydrophilic (glass) sections. Lipid bilayer formation on graphene oxide in solution 8 has also been observed by Okamoto et al , who used AFM measurements to propose that both single 9 10 and double lipid membranes are thermally stable on graphene oxide supports after vesicle fusion 11 30 12 with the surface. Furthermore, a recent study investigating lipid assembly on oxidized graphene 13 surfaces confirmed that planar lipid bilayers were stable on this support.58 The same study also 14 15 observed that, in contrast with lipid bilayer formation on oxidized graphene, lipid monolayers were 16 17 formed on a pristine graphene surface in solution, in which lipid tails interacted with the underlying 18 graphene support. This is similar to what is suggested by our CGMD simulations of preformed 19 20 lipid bilayers on larger graphene surfaces, which disassembled into monolayers in the presence of 21 22 water, again suggesting that the hydrophobic interactions between the lipid carbon tails and 23 graphene surface dominate lipid organization on this support. A recent functional study of LDPN 24 25 generated membranes on pristine graphene observed the same lipid behavior upon the addition of 26 27 buffer solution, in which fluorescence was used to verify the solventexposed orientation of the 28 lipid headgroups with respect to surface. 39 The study also indicated that lipid spreading on pristine 29 30 graphene is selflimiting, and is more favorable than lipid interactions with silicon dioxide surfaces, 31 39 32 attributed to the contrast in hydrophilicity between the two supports. A similar trend in lipid 33 spreading behavior has also been observed on other support surfaces, including glass (hydrophilic) 34 35 and noctadecyltrichlorosilane (hydrophobic). 59 Using ellipsometry to determine lipid film heights, 36 37 it was suggested that lipids preferentially spread as monolayers on hydrophobic surfaces, whilst 38 lipid bilayers form on more hydrophilic surfaces. 59,60 39 40 41 We also observed that lipids may undergo dynamic changes in their configurations on either support 42 surface, as is evidenced by the time evolution of lipid tail order parameters. This is particularly true 43 44 for simulations of regular (i.e. noninverted) lipid bilayers on the small graphene oxide surface, in 45 46 which complete restructuring of the lipid layers occurred to form the apparently more favorable 1.5 47 bilayer configuration (Fig. 5). Furthermore, characterization of lipid mobility within the supported 48 49 membrane structures suggested that lipid diffusion was anomalous, exhibiting subdiffusion for 50 51 most systems. This is expected given that interactions between the lipid and the support alter mean 52 square displacement values over the longer time sampling windows. Tero et al also identified sub 53 54 diffusive behavior of DOPC molecules within supported lipid membranes on hydrophilic titanium 55 42 56 oxide, and related this to the atomic topology of the support surface. Furthermore, other 57 experimental studies of lipid diffusion in supported membranes have reported both linear and 58 59 21 60 ACS Paragon Plus Environment ACS Nano Page 22 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 anomalous diffusion regimes, and related this to lipid interactions with the support surface, surface 3 topology of the support, membrane composition and topology ( e.g . planar bilayer vs vesicles), as 4 5 well as system hydration. 42,46,47,49 The distinction between diffusion regimes is however also related 6 45 7 to the temporal and spatial scales at which diffusion is measured. It would therefore be of interest 8 in the future to extend our simulations to substantially larger length scales and longer time scales, 9 10 enabling more direct comparison with experimental data for similar systems.61 11 12 The use of CG resolution simulations of the lipid membranegraphene systems has allowed us to 13 14 characterize of supported membrane properties on longer (i.e . several microsecond) simulation 15 16 times than would be readily available to e.g. all atom simulations. A number of previous studies 17 have indicated the utility of CG simulation methods for studying lipid behavior on surfaces. 31,34–36 It 18 19 is important to note that the mapping scheme used to model the CG graphene and graphene oxide 20 21 surfaces (2 atoms:1 CG particle) differs from the 4:1 mapping scheme used for the lipid molecules. 22 However, the Martini forcefield was parameterized to remain internally consistent and compatible 23 24 between parameterization efforts (e.g. comparing Martini v2.1 and Martini v2.2). 37,53 Consequently, 25 26 the parameters used to model the interactions between the lipid molecules and graphene/graphene 27 oxide surfaces are largely independent of the mapping schemes employed to generate the graphene 28 29 and lipid models. This is because the interactions between the different components are modeled 30 53,54 31 using an interaction matrix, with different values assigned to interactions of each bead type. The 32 original parameterization of the graphite surface on which the graphene (oxide) model in this study 33 34 is based, involved reproducing thermodynamic values and adsorption behavior of longchain 35 37 36 alkanes on the graphite support. The nonbonding interaction parameters between the different 37 beads within the Martini models were adjusted to reproduce this behavior, so that the different 38 39 mapping schemes did not affect the overall behavior of the systems. 40 41 Full atomistic simulations would be of interest to capture in more detail the interactions between 42 43 lipid molecules and the support surface in more detail. Importantly, the electronic structure of 44 graphene and the polarizability of interacting molecules, which may play key roles in biomolecular 45 46 adsorption and interactions,62–64 are not captured by the CG model. With the development of 47 62,65,66 48 polarizable forcefields and further characterization of the unique properties of graphene 49 supports,63 future simulations encompassing these aspects whilst starting from the CG models 50 51 described in the current study would provide enhanced insight into the interaction forces underlying 52 53 different supported membrane configurations. It would be of interest to explore the free energy 54 landscape underlying the different lipid configurations observed on these support surfaces, as this 55 56 would be likely to provide further insights into the origin of stability of the lipid structures observed 57 58 for each of the surfaces. Characterization of the energetic differences between the different lipid 59 22 60 ACS Paragon Plus Environment Page 23 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 structures formed on pristine graphene and graphene oxide depending on system solvation might 3 also be expected to provide insights into the link between surface wettability and the eventual lipid 4 5 structure formed. However, accurate estimation of free energy andscapes for lipid/graphene 6 7 inteactions would be challenging in terms of achieving adequate sampling of the system to reach 8 convergence. In a previous study, we have shown that both the choice of a suitable collective 9 10 variable and adequate sampling are key parameters in obtaining a converged free energy landscape 11 67 12 even for relatively simple systems, such as single proteinlipid interactions. 13 14 In conclusion, our combined simulation and experimental results highlight the effects of both 15 16 surface polarity and the solvent environment on phospholipid membrane organization when 17 interacting with a support surface. We demonstrate that lipids can form multilayered structures such 18 19 as 1.5 bilayers on hydrophilic surfaces such as graphene oxide, and can spontaneously rearrange to 20 21 form a preferred topology even when starting from different structures ( e.g. regular bilayers). 22 Hydrophobic surfaces such as pristine graphene interact highly favorably with lipids, affecting lipid 23 24 bilayer membrane stability on this surface. Characterizing these interactions will provide important 25 26 insight into the applications of graphene/graphene oxide within biotechnology, including sensor 27 devices and drug delivery systems.68,69 28 29 30 31 32 Materials and Methods: 33 34 Coarsegrained graphene (oxide) models 35 53,70 37 36 CG simulations were performed using the Martini forcefield. The Martini model for graphite 37 38 was used to construct the graphene and graphene oxide model surfaces (single layer; Fig. S1). 39 Briefly, the model involved a 2:1 mapping scheme of atomistic carbon atoms into hexagonally 40 41 packed graphene beads (SG4) making up the sheet. The published parameterization of this model 42 43 was based on reproducing the adsorption and topological behavior of longchain alkane molecules 44 on graphite, suggesting the model is suitable to study the behavior of lipids on graphene surfaces. 37 45 46 The CG graphene model was used to build the graphene oxide model. To do this, the carbon beads 47 48 (SG4) comprising the pristine graphene surface were randomly substituted by oxygen beads (SP1), 49 which represent either a COC (epoxy) group or a COH group,71 eventually reproducing the 50 51 oxygen content of the graphene oxide substrates used in the experiments (2.82% COC: 49.3% C 52 53 OH based on Xray photoelectron spectroscopy data) (Fig. S1). 54 55 Simulation details 56 57 58 59 23 60 ACS Paragon Plus Environment ACS Nano Page 24 of 36 2303214_File000005_39618591.docx 1/13/2017

1 72 53,54,71 2 The GROMACS 4.6 (www.gromacs.org ) simulation suite and the Martini 2.2 forcefield was 3 used to perform all CG simulations. Lipid bilayers consisting of DOPC molecules were constructed 4 5 by selfassembly simulations; this lipid composition reflects the main lipids employed in the 6 13,39 7 experiments and related studies. Lipid molecules were randomly inserted into cubic box of 8 dimensions 10 x 10 x 5 nm 3 (small bilayers) or 20 x 20 x 5 nm 3 (large bilayer); the systems were 9 10 then energy minimized using the steepest descent algorithm for 10,000 steps. The zdimension of 11 12 the box was then extended to 15 nm or 30 nm. The system was subsequently solvated and simulated 13 for 100 ns to allow bilayer selfassembly to occur. The inverted bilayers that mimic lipid structures 14 15 seen in previous AFM studies 13,39 were constructed by placing two lipid monolayers on top of each 16 17 other, with the lipid headgroups pointing towards each other., using the editconf tool from 18 GROMACS. These lipid configurations were then placed above either graphene or graphene oxide 19 20 surfaces. 21 22 The polarizable Martini water model 73 was employed in simulations containing water. A round of 23 24 equilibration simulations were performed for the solvated systems. This involved performing 25 26 10,000 steps of equilibration using incremental simulation time steps of: 1 fs, 2 fs, 5 fs, 10 fs, and 27 20 fs. The equilibrated systems were used to initiate production simulations performed at constant 28 29 temperature and pressure (NPT ensemble). Temperature was regulated used the Berendsen 30 74 31 thermostat and a coupling constant of 0.3 ps at 298 K. Pressure was regulated using the Berendsen 32 barostat 74 with a coupling constant of 3.0 ps, applying anisotropic pressure coupling using a 33 34 compressibility of 0.5 x 10 5 bar 1 in x and y, and 3.0 x 10 5 bar 1 in the zdirection. Simulations of 35 36 systems in vacuum were performed at constant temperature and volume (NVT ensemble). The 37 lipids, water and graphene were coupled to separate external baths. Nonbonding interactions were 38 39 modeled using shift functions; both LJ and Coulombic interactions were evaluated within a 1.2 nm 40 41 cutoff, and shifted within a 0.9 nm cutoff distance. These parameters reflect those applied in 42 parameterization studies of the CG graphene model. 37 43 44 Lipid order parameter analysis 45 46 nd 47 2 rank order parameters were calculated for all of the lipids in the select CG systems (Fig. S3 and 48 cc nd S4). These were calculated according to the S n 2 rank order equation: 49 50 S = 0.5 (3 < cos 2(θ) > 1) 51 52 53 where θ is the angle between the bond connecting lipid bead particles (B nBn1) connecting beads 54 within the lipid and the zaxis of the simulation box, normal to the bilayer. The angle brackets 55 nd 56 denote the mean angle calculated for all of the lipids in the system (ensemble average). The 2 rank 57 58 59 24 60 ACS Paragon Plus Environment Page 25 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 75 2 term refers to the second order Legendre polynomial used to describe the order parameter. This 3 has been used in many experimental and computational studies of similar bilayer systems.55,76–78 4 5 6 Diffusion analysis 7 8 The diffusion of lipids within the simulated systems was analyzed using documented opensource 9 code (http://dx.doi.org/10.5281/zenodo.11827). This code employs an algorithm which calculates 10 11 the mean square displacement (MSD) values of individual lipid centroids over a range of time 12 13 sampling windows, including: 1, 2, 5, 10, 20, 50, 70, 100, and 200 ns. The diffusion coefficients are 14 then calculated by fitting the MSD vs time data using a linear diffusion equation 50 or an anomalous 15 16 diffusion equation. 51 The linear approximation uses a least squares first degree fit of the data, 17 18 whereas the anomalous approximation uses a nonlinear least squares fitting to a twoparameter 19 equation of the form: 20 21 α 22 MSD = 4 Dαt 0 < α < 2 23 2 α 24 where Dα is the fractional diffusion coefficient, measured in units of length time . 25 26 For the linear diffusion fit, standard deviations of the diffusion coefficients were estimated as the 27 28 difference of the slopes from the first and second halves of the MSD vs time data. This is the 29 approach used by the GROMACS tools function g_msd . For the anomalous diffusion fit, the scaling 30 31 exponent (α) was estimated as the slope of the log MSD vs log time data. The standard deviations of 32 33 both parameters ( Dα and α) were calculated from the square root of the diagonal of the covariance 34 matrix from the anomalous fit. This code has been used by previous simulation studies of lipid 35 79,80 36 diffusion in virus particles. Diffusion constants are reported without correction for the reduced 37 38 degrees of freedom resulting from applying the Martini CG model. A simple scaling factor of 4 has 39 previously been applied to compare diffusion of lipid and water particles in CG systems to 40 54 41 experimental measurements. 42 43 Lipid dippen nanolithography (LDPN) 44 45 LDPN was performed on graphene and graphene oxide samples. Lithography took place in a 46 47 DPN5000 system (Nanoink, USA) with a single cantilever probe (AType, ACST, USA). The 48 cantilever was coated with 1,2dioleoylsnglycero3phosphocholine (DOPC) using a microfluidic 49 50 inkwell (ACST, USA) at high humidity (~70%). The tip was then conditioned by writing some 51 52 sacrificial patterns to strip off excessive ink. Membranes used for AFM imaging were written at 53 54 25% humidity. 55 56 Atomic force microscopy (AFM) 57 58 59 25 60 ACS Paragon Plus Environment ACS Nano Page 26 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 AFM imaging was performed on a Dimension Icon setup (Bruker) in tapping mode. NSC15 3 cantilevers (MikroMasch) with a nominal force constant of 46 N/m and a resonance frequency of 4 5 325 kHz were used. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 26 60 ACS Paragon Plus Environment Page 27 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 ASSOCIATED CONTENT 3 4 Supporting Information . Additional details of the simulated graphene and graphene oxide models, 5 6 as well as lipid diffusion and order parameter analyses, are reported in the supporting information. 7 Movies corresponding to the simulations in Figures 1 and 3 are provided as Web Enhanced Objects. 8 9 This material is available free of charge via the Internet at http://pubs.acs.org. 10 11 Author Contributions 12 13 All authors conceived the study and devised the experiments and models. NW performed all 14 15 molecular dynamics simulations and analysis. Graphene samples were fabricated by AFV and MI. 16 17 DPN and AFM was performed by AU, ML and MH. The manuscript was written through 18 contributions of all authors. All authors have given approval to the final version of the manuscript. 19 20 Funding Sources 21 22 23 This research was supported by the Biotechnology and Biological Sciences Research Council 24 (BBSRC) [grant number BB/J014427/1], Engineering and Physical Sciences Research Council 25 26 (EPSRC) [grant number EP/L01548X/1 and the Karlsruhe Nano Micro Facility (KNMF, 27 28 www.knmf.kit.edu), a Helmholtz Research Infrastructure at Karlsruhe Institute of Technology 29 (KIT, www.kit.edu). A.U. acknowledges the People Programme (Marie Curie Actions) of the 30 31 European Union’s Seventh Framework Programme FP7/2007–2013 under REA grant agreement 32 33 no. 328163. 34 35 Acknowledgement 36 37 Our thanks to our colleagues for their interest in this work, and especially to Matthieu Chavent for 38 39 his help with preparation of figures and movies using Hyperballs. 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 27 60 ACS Paragon Plus Environment ACS Nano Page 28 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 REFERENCES 3 4 (1) Piner, R. D.; Zhu, J.; Xu, F.; Hong, S.; Mirkin, C. A. “DipPen” Nanolithography. Sci. 1999 , 5 6 283 , 661–663. 7 8 9 (2) Lenhert, S.; Sun, P.; Wang, Y.; Fuchs, H.; Mirkin, C. A. Massively Parallel DipPen 10 Nanolithography of Heterogeneous Supported Phospholipid Multilayer Patterns. Small 2007 , 11 12 3, 71–75. 13 14 15 (3) Huo, F.; Zheng, Z.; Zheng, G.; Giam, L. R.; Zhang, H.; Mirkin, C. A. Polymer Pen 16 Lithography. Sci. 2008 , 321 , 1658–1660. 17 18 19 (4) Brinkmann, F.; Hirtz, M.; Greiner, A. M.; Weschenfelder, M.; Waterkotte, B.; Bastmeyer, 20 21 M.; Fuchs, H. Interdigitated Multicolored Bioink Micropatterns by Multiplexed Polymer Pen 22 Lithography. Small 2013 , 9, 3266–3275. 23 24 25 (5) Nafday, O. A.; Lowry, T. W.; Lenhert, S. Multifunctional Lipid Multilayer Stamping. Small 26 27 2012 , 8, 1021–1028. 28 29 (6) Lowry, T. W.; KusiAppiah, A.; Guan, J.; Van Winkle, D. H.; Davidson, M. W.; Lenhert, S. 30 31 Materials Integration by Nanointaglio. Adv. Mater. Interfaces 2014 , 1, n/an/a. 32 33 34 (7) Hirtz, M.; SekulaNeuner, S.; Urtizberea, A.; Fuchs, H. Functional Lipid Assemblies by Dip 35 Pen Nanolithography and Polymer Pen Lithography. In Soft Matter Nanotechnology ; Wiley 36 37 VCH Verlag GmbH & Co. KGaA, 2015; pp. 161–186. 38 39 40 (8) SekulaNeuner, S.; Maier, J.; Oppong, E.; Cato, A. C. B.; Hirtz, M.; Fuchs, H. Allergen 41 Arrays for Antibody Screening and Immune Cell Activation Profiling Generated by Parallel 42 43 Lipid DipPen Nanolithography. Small 2012 , 8, 585–591. 44 45 (9) Oppong, E.; Hedde, P. N.; SekulaNeuner, S.; Yang, L.; Brinkmann, F.; Dörlich, R. M.; 46 47 Hirtz, M.; Fuchs, H.; Nienhaus, G. U.; Cato, A. C. B. Localization and Dynamics of 48 49 Glucocorticoid Receptor at the Plasma Membrane of Activated Mast Cells. Small 2014 , 10 , 50 1991–1998. 51 52 53 (10) KusiAppiah, A. E.; Lowry, T. W.; Darrow, E. M.; Wilson, K. A.; Chadwick, B. P.; 54 55 Davidson, M. W.; Lenhert, S. Quantitative DoseResponse Curves from Subcellular Lipid 56 Multilayer Microarrays. Lab Chip 2015 , 15 , 3397–3404. 57 58 59 28 60 ACS Paragon Plus Environment Page 29 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 (11) Vafai, N.; Lowry, T. W.; Wilson, K. A.; Davidson, M. W.; Lenhert, S. Evaporative Edge 3 Lithography of a Liposomal Drug Microarray for Cell Migration Assays. Nanofabrication 4 5 2015 , 2, 32–42. 6 7 8 (12) Mitsakakis, K.; SekulaNeuner, S.; Lenhert, S.; Fuchs, H.; Gizeli, E. Convergence of Dip 9 Pen Nanolithography and Acoustic Biosensors towards a RapidAnalysis MultiSample 10 11 Microsystem. Analyst 2012 , 137 , 3076–3082. 12 13 14 (13) Hirtz, M.; Oikonomou, A.; Georgiou, T.; Fuchs, H.; Vijayaraghavan, A. Multiplexed 15 Biomimetic Lipid Membranes on Graphene by DipPen Nanolithography. Nat. Commun. 16 17 2013 , 4, 1–8. 18 19 20 (14) Bog, U.; Laue, T.; Grossmann, T.; Beck, T.; Wienhold, T.; Richter, B.; Hirtz, M.; Fuchs, H.; 21 Kalt, H.; Mappes, T. OnChip Microlasers for Biomolecular Detection via Highly Localized 22 23 Deposition of a Multifunctional Phospholipid Ink. Lab Chip 2013 , 13 , 2701–2707. 24 25 26 (15) Bog, U.; Brinkmann, F.; Kalt, H.; Koos, C.; Mappes, T.; Hirtz, M.; Fuchs, H.; Köber, S. 27 LargeScale Parallel Surface Functionalization of GobletType Whispering Gallery Mode 28 29 Microcavity Arrays for Biosensing Applications. Small 2014 , 10 , 3863–3868. 30 31 32 (16) Rath, P.; Hirtz, M.; LewesMalandrakis, G.; Brink, D.; Nebel, C.; Pernice, W. H. P. Diamond 33 Nanophotonic Circuits Functionalized by DipPen Nanolithography. Adv. Opt. Mater. 2015 , 34 35 3, 328–335. 36 37 38 (17) Bog, U.; Brinkmann, F.; Wondimu, S. F.; Wienhold, T.; Kraemmer, S.; Koos, C.; Kalt, H.; 39 Hirtz, M.; Fuchs, H.; Koeber, S.; et al. Densely Packed Microgoblet Laser Pairs for Cross 40 41 Referenced Biomolecular Detection. Adv Sci 2015 , 1500066. 42 43 44 (18) Ielasi, F. S.; Hirtz, M.; SekulaNeuner, S.; Laue, T.; Fuchs, H.; Willaert, R. G. DipPen 45 NanolithographyAssisted Protein Crystallization. J. Am. Chem. Soc. 2015 , 137 , 154–157. 46 47 48 (19) Lenhert, S.; Brinkmann, F.; Laue, T.; Walheim, S.; Vannahme, C.; Klinkhammer, S.; Xu, M.; 49 50 Sekula, S.; Mappes, T.; Schimmel, T.; et al. Lipid Multilayer Gratings. Nat. Nanotechnol. 51 2010 , 5, 275–279. 52 53 54 (20) Lowry, T. W.; Prommapan, P.; Rainer, Q.; Van Winkle, D.; Lenhert, S. Lipid Multilayer 55 Grating Arrays Integrated by Nanointaglio for Vapor Sensing by an Optical Nose. Sensors 56 57 (Basel). 2015 , 15 , 20863–20872. 58 59 29 60 ACS Paragon Plus Environment ACS Nano Page 30 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 (21) Urtizberea, A.; Hirtz, M. A Diffusive Ink Transport Model for Lipid DipPen 3 Nanolithography. Nanoscale 2015 , 7, 15618–15634. 4 5 6 (22) Raedler, J.; Strey, H.; Sackmann, E. Phenomenology and Kinetics of Lipid Bilayer Spreading 7 8 on Hydrophilic Surfaces. Langmuir 1995 , 11 , 4539–4548. 9 10 (23) Nissen, J.; Gritsch, S.; Wiegand, G.; Raedler, J. O. Wetting of Phospholipid Membranes on 11 12 Hydrophilic Surfaces Concepts towards SelfHealing Membranes. Eur. Phys. J. B 1999 , 10 , 13 14 335–344. 15 16 (24) Reviakine, I.; Brisson, A. Formation of Supported Phospholipid Bilayers from Unilamellar 17 18 Vesicles Investigated by Atomic Force Microscopy. Langmuir 2000 , 16 , 1806–1815. 19 20 21 (25) Ainla, A.; Gözen, I.; Hakonen, B.; Jesorka, A. Lab on a Biomembrane: Rapid Prototyping 22 and Manipulation of 2D Fluidic Lipid Bilayer Circuits. Sci. Rep. 2013 , 3, 2743. 23 24 25 (26) Smith, K. A.; Gale, B. K.; Conboy, J. C. Micropatterned Fluid Lipid Bilayer Arrays Created 26 27 Using a Continuous Flow Microspotter. Anal. Chem. 2008 , 80 , 7980–7987. 28 29 (27) Jönsson, P.; Beech, J. P.; Tegenfeldt, J. O.; Höök, F. ShearDriven Motion of Supported 30 31 Lipid Bilayers in Microfluidic Channels. J. Am. Chem. Soc. 2009 , 131 , 5294–5297. 32 33 34 (28) Hirtz, M.; Corso, R.; SekulaNeuner, S.; Fuchs, H. Comparative Height Measurements of 35 DipPen NanolithographyProduced Lipid Membrane Stacks with Atomic Force, 36 37 Fluorescence, and SurfaceEnhanced Ellipsometric Contrast Microscopy. Langmuir 2011 , 38 39 27 , 11605–11608. 40 41 (29) Tsuzuki, K.; Okamoto, Y.; Iwasa, S.; Ishikawa, R.; Sandhu, a; Tero, R. Reduced Graphene 42 43 Oxide as the Support for Lipid Bilayer Membrane. J. Phys. Conf. Ser. 2012 , 352 , 12016. 44 45 (30) Okamoto, Y.; Tsuzuki, K.; Iwasa, S.; Ishikawa, R.; Sandhu, a; Tero, R. Fabrication of 46 47 Supported Lipid Bilayer on Graphene Oxide. J. Phys. Conf. Ser. 2012 , 352 , 12017. 48 49 50 (31) Wu, D.; Yang, X. CoarseGrained Molecular Simulation of SelfAssembly for Nonionic 51 Surfactants on Graphene Nanostructures. J. Phys. Chem. B 2012 , 116 , 12048–12056. 52 53 54 (32) Liu, C.; Faller, R. Conformational, Dynamical. and Tensional Study of Tethered Bilayer 55 56 Lipid Membranes in CoarseGrained Molecular Simulations. Langmuir 2012 , 28 , 15907– 57 15915. 58 59 30 60 ACS Paragon Plus Environment Page 31 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 (33) Wang, J.; Wei, Y.; Shi, X.; Gao, H. Cellular Entry of Graphene Nanosheets: The Role of 3 Thickness, Oxidation and Surface Adsorption. RSC Adv. 2013 , 3, 15776. 4 5 6 (34) Xing, C.; Faller, R. CoarseGrained Simulations of Supported and Unsupported Lipid 7 8 Monolayers. Soft Matter 2009 , 5, 4526. 9 10 (35) Xing, C.; Faller, R. Interactions of Lipid Bilayers with Supports: A CoarseGrained 11 12 Molecular Simulation Study. J. Phys. Chem. B 2008 , 112 , 7086–7094. 13 14 15 (36) Mhashal, A. R.; Roy, S. SelfAssembly of Phospholipids on Flat Supports. Phys. Chem. 16 Chem. Phys. 2015 , 17 , 31152–31160. 17 18 19 (37) Gobbo, C.; Beurroies, I.; de Ridder, D.; Eelkema, R.; Marrink, S. J.; De Feyter, S.; van Esch, 20 21 J. H.; de Vries, A. H. MARTINI Model for Physisorption of Organic Molecules on Graphite. 22 J. Phys. Chem. C 2013 , 117 , 15623–15631. 23 24 25 (38) Lenhert, S.; Mirkin, C. a; Fuchs, H. In Situ Lipid DipPen Nanolithography under Water. 26 27 Scanning 2010 , 32 , 15–23. 28 29 (39) Hirtz, M.; Oikonomou, A.; Clark, N.; Kim, Y.J.; Fuchs, H.; Vijayaraghavan, A. Self 30 31 Limiting Multiplexed Assembly of Lipid Membranes on LargeArea Graphene Sensor 32 33 Arrays. Nanoscale 2016 , 0, 1–3. 34 35 (40) Hirtz, M.; Oikonomou, A.; Varey, S.; Fuchs, H.; Vijayaraghavan, A. Supplementary 36 37 Multiplexed Biomimetic Lipid Membranes on Graphene by DipPen Nanolithography. 38 39 Microsc. Microanal. 2014 , 20 , 2058–2059. 40 41 (41) Chavent, M.; Vanel, A.; Tek, A.; Levy, B.; Robert, S.; Raffin, B.; Baaden, M. GPU 42 43 Accelerated Atom and Dynamic Bond Visualization Using Hyperballs: A Unified Algorithm 44 45 for Balls, Sticks, and Hyperboloids. J. Comput. Chem. 2011 , 32 , 2924–2935. 46 47 (42) Tero, R.; Sazaki, G.; Ujihara, T.; Urisu, T. Anomalous Diffusion in Supported Lipid Bilayers 48 49 Induced by Oxide Surface Nanostructures. Langmuir 2011 , 27 , 9662–9665. 50 51 (43) Tero, R. Substrate Effects on the Formation Process, Structure and Physicochemical 52 53 Properties of Supported Lipid Bilayers. Materials (Basel). 2012 , 5, 2658–2680. 54 55 56 (44) Sekula, S.; Fuchs, J.; WegRemers, S.; Nagel, P.; Schuppler, S.; Fragala, J.; Theilacker, N.; 57 Franzreb, M.; Wingren, C.; Ellmark, P.; et al. Multiplexed Lipid DipPen Nanolithography 58 59 31 60 ACS Paragon Plus Environment ACS Nano Page 32 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 on Subcellular Scales for the Templating of Functional Proteins and Cell Culture. Small 3 2008 , 4, 1785–1793. 4 5 6 (45) Machan, R.; Hof, M. Lipid Diffusion in Planar Membranes Investigated by Fluorescence 7 8 Correlation Spectroscopy. Biochim. Biophys. Acta Biomembr. 2010 , 1798 , 1377–1391. 9 10 (46) Renner, L.; Osaki, T.; Chiantia, S.; Schwille, P.; Pompe, T.; Werner, C. Supported Lipid 11 12 Bilayers on Spacious and pHResponsive Polymer Cushions with Varied Hydrophilicity. J. 13 14 Phys. Chem. B 2008 , 112 , 6373–6378. 15 16 (47) Ratto, T. V.; Longo, M. L. Anomalous Subdiffusion in Heterogeneous Lipid Bilayers. 17 18 Langmuir 2003 , 19 , 1788–1793. 19 20 21 (48) Benda, A.; Benes, M.; Marec, V.; Lhotsky, A.; Hof, M. How To Determine Diffusion 22 Coefficients in Planar Phospholipid Systems by Confocal Fluorescence Correlation 23 24 Spectroscopy. 2003 , 9, 4120–4126. 25 26 27 (49) Schütz, G. J.; Schindler, H.; Schmidt, T. SingleMolecule Microscopy on Model Membranes 28 Reveals Anomalous Diffusion. Biophys. J. 1997 , 73 , 1073–1080. 29 30 31 (50) Einstein, A. Über Die von Der Molekularkinetischen Theorie Der Wärme Geforderte 32 33 Bewegung von in Ruhenden Flüssigkeiten Suspendierten Teilchen. Ann. Phys. 1905 , 322 , 34 549–560. 35 36 37 (51) Kneller, G. R.; Baczynski, K.; PasenkiewiczGierula, M. Communication: Consistent Picture 38 39 of Lateral Subdiffusion in Lipid Bilayers: Molecular Dynamics Simulation and Exact 40 Results. J. Chem. Phys. 2011 , 135 , 2–5. 41 42 43 (52) Shi, X.; Kohram, M.; Zhuang, X.; Smith, A. W. Interactions and Translational Dynamics of 44 45 Phosphatidylinositol Bisphosphate (PIP2) Lipids in Asymmetric Lipid Bilayers. Langmuir 46 2016 , 32 , 1732–1741. 47 48 49 (53) Marrink, S. J.; Risselada, H. J.; Yefimov, S.; Tieleman, D. P.; Vries, A. H. De. The 50 MARTINI Force Field : Coarse Grained Model for Biomolecular Simulations. J. Phys. 51 52 Chem. B 2007 , 111 , 7812–7824. 53 54 55 (54) Marrink, S. J.; de Vries, A. H.; Mark, A. E. Coarse Grained Model for Semiquantitative 56 Lipid Simulations. J. Phys. Chem. B 2004 , 108 , 750–760. 57 58 59 32 60 ACS Paragon Plus Environment Page 33 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 (55) Reddy, A. S.; Warshaviak, D. T.; Chachisvilis, M. Effect of Membrane Tension on the 3 Physical Properties of DOPC Lipid Bilayer Membrane. Biochim. Biophys. Acta Biomembr. 4 5 2012 , 1818 , 2271–2281. 6 7 8 (56) Lamberg, A.; Taniguchi, T. CoarseGrained Computational Studies of Supported Bilayers: 9 Current Problems and Their Root Causes. J Phys Chem B 2014 , 118 , 10643–10652. 10 11 12 (57) Ryu, Y.S.; Wittenberg, N. J.; Suh, J.H.; Lee, S.W.; Sohn, Y.; Oh, S.H.; Parikh, A. N.; 13 14 Lee, S.D. Continuity of MonolayerBilayer Junctions for Localization of Lipid Raft 15 Microdomains in Model Membranes. Sci. Rep. 2016 , 6, 26823. 16 17 18 (58) Tabaei, S. R.; Ng, W. B.; Cho, S.J.; Cho, N.J. Controlling the Formation of Phospholipid 19 20 Monolayer, Bilayer, and Intact Vesicle Layer on Graphene. ACS Appl. Mater. Interfaces 21 2016 , acsami.6b02837. 22 23 24 (59) Sanii, B.; Parikh, A. N. SurfaceEnergy Dependent Spreading of Lipid Monolayers and 25 26 Bilayers. Soft Matter 2007 , 3, 974–977. 27 28 (60) Parikh, A. N. MembraneSubstrate Interface: Phospholipid Bilayers at Chemically and 29 30 Topographically Structured Surfaces. Biointerphases 2008 , 3, FA22. 31 32 33 (61) Chavent, M.; Duncan, A. L.; Sansom, M. S. Molecular Dynamics Simulations of Membrane 34 Proteins and Their Interactions: From Nanoscale to Mesoscale. Curr. Opin. Struct. Biol. 35 36 2016 , 40 , 8–16. 37 38 39 (62) Tomasio, S. M.; Walsh, T. R.; Toma, S. M.; Walsh, T. R. Modeling the Binding Affinity of 40 Peptides for Graphitic Surfaces. Influences of Aromatic Content and Interfacial Shape. J. 41 42 Phys. Chem. C 2009 , 113 , 8778–8785. 43 44 45 (63) De Leo, F.; Magistrato, A.; Bonifazi, D. Interfacing Proteins with Graphitic Nanomaterials: 46 From Spontaneous Attraction to Tailored Assemblies. Chem. Soc. Rev. 2015 . 47 48 49 (64) Kim, S. N.; Kuang, Z.; Slocik, J. M.; Jones, S. E.; Cui, Y.; Farmer, B. L.; McAlpine, M. C.; 50 Naik, R. R. Preferential Binding of Peptides to Graphene Edges and Planes. J. Am. Chem. 51 52 Soc. 2011 , 133 , 14480–14483. 53 54 55 (65) Ho, T. a; Striolo, A. Polarizability Effects in Molecular Dynamics Simulations of the 56 GrapheneWater Interface. J. Chem. Phys. 2013 , 138 , 54117. 57 58 59 33 60 ACS Paragon Plus Environment ACS Nano Page 34 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 (66) Hughes, Z. E.; Tomásio, S. M.; Walsh, T. R. Efficient Simulations of the Aqueous Bio 3 Interface of Graphitic Nanostructures with a Polarisable Model. Nanoscale 2014 , 6, 5438. 4 5 6 (67) Domański, J.; Hedger, G.; Best, R.; Stansfeld, P. J.; Sansom, M. S. P. Convergence and 7 8 Sampling in Determining Free Energy Landscapes for Membrane Protein Association. J Phys 9 Chem B 2016 . 10 11 12 (68) Wang, F.; Liu, B.; Ip, A. C.F.; Liu, J. Orthogonal Adsorption onto NanoGraphene Oxide 13 14 Using Different Intermolecular Forces for Multiplexed Delivery. Adv. Mater. 2013 , 25 , 15 4087–4092. 16 17 18 (69) Lelimousin, M.; Sansom, M. S. P. Membrane Perturbation by Carbon Nanotube Insertion: 19 20 Pathways to Internalization. Small 2013 , 9, 3639–3646. 21 22 (70) Marrink, S. J.; Tieleman, D. P. Perspective on the Martini Model. Chem. Soc. Rev. 2013 , 42 , 23 24 6801–6822. 25 26 27 (71) Monticelli, L.; Kandasamy, S. K.; Periole, X.; Larson, R. G.; Tieleman, D. P.; Marrink, S.J. 28 The MARTINI GoarseGrained Force Field: Extension to Proteins. J. Chem. Theory Comput. 29 30 2008 , 4, 819–834. 31 32 33 (72) Hess, B.; Kutzer, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly 34 Efficient, LoadBalanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 35 36 2008 , 4, 435–447. 37 38 39 (73) Yesylevskyy, S. O.; Schäfer, L. V; Sengupta, D.; Marrink, S. J. Polarizable Water Model for 40 the CoarseGrained MARTINI Force Field. PLoS Comput. Biol. 2010 , 6, 1–17. 41 42 43 (74) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. 44 45 Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984 , 81 , 3684– 46 3690. 47 48 49 (75) Abramovitz, M.; Stegun, I. . Handbook of Mathematical Functions ; Abramovitz, M.; Stegun, 50 I. ., Eds.; 10th ed.; Dover: Washington, 1972. 51 52 53 (76) Vermeer, L. S.; de Groot, B. L.; Réat, V.; Milon, A.; Czaplicki, J. Acyl Chain Order 54 55 Parameter Profiles in Phospholipid Bilayers: Computation from Molecular Dynamics 56 Simulations and Comparison with 2H NMR Experiments. Eur. Biophys. J. 2007 , 36 , 919– 57 58 59 34 60 ACS Paragon Plus Environment Page 35 of 36 ACS Nano 2303214_File000005_39618591.docx 1/13/2017

1 2 931. 3 4 (77) Douliez, J.P. P.; Léonard, A.; Dufourc, E. J.; Leonard, A. Restatement of Order Parameters 5 6 in Biomembranes: Calculation of CC Bond Order Parameters from CD Quadrupolar 7 8 Splittings. Biophys J 1995 , 68 , 1727–1739. 9 10 (78) Douliez, J. P.; Léonard, a.; Dufourc, E. J.; JP, D.; a, L.; J, D. E. Conformational Order of 11 12 DMPC Sn1 versus Sn2 Chains and Membrane Thickness: An Approach to Molecular 13 14 Protrusion by Solid State 2HNMR and Neutron Diffraction. J. Phys. Chem. 1996 , 100 , 15 18450–18457. 16 17 18 (79) Reddy, T.; Shorthouse, D.; Parton, D. L.; Jefferys, E.; Fowler, P. W.; Chavent, M.; Baaden, 19 20 M.; Sansom, M. S. P. Nothing to Sneeze At: A Dynamic and Integrative Computational 21 Model of an Influenza A Virion. Structure 2015 , 23 , 584–597. 22 23 24 (80) Reddy, T.; Sansom, M. S. P. The Role of the Membrane in the Structure and Biophysical 25 26 Robustness of the Dengue Virion Envelope. Structure 2016 , 24 , 375–382. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 35 60 ACS Paragon Plus Environment ACS Nano Page 36 of 36 2303214_File000005_39618591.docx 1/13/2017

1 2 Table of Contents Graphic 3 4 5 6 7 8 graphene 9 10 11 12 13 14 15 graphene 16 oxide 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 36 60 ACS Paragon Plus Environment