Microarray Technology for Kinetic Analysis of Vesicle Bound Receptor-Ligand Interactions

(1)

The Department of Physics, Chemistry and Biology

Master’s Thesis

Microarray Technology for Kinetic Analysis of

Vesicle Bound Receptor-Ligand Interactions

Björn Brian

LiTH-IFM-EX-07/1695-SE

The Department of Physics, Chemistry and Biology Linköpings universitet

(2)

(3)

Master’s Thesis LiTH-IFM-EX-07/1695-SE

Microarray Technology for Kinetic Analysis of

Vesicle Bound Receptor-Ligand Interactions

Björn Brian

Supervisor: Goran Klenkar

ifm, Linköpings universitet

Thomas Ederth

Examiner: Olle Inganäs

(4)

(5)

Avdelning, Institution

Division, Department

Division of Sensor Science and Molecular Physics The Department of Physics, Chemistry and Biology Linköpings universitet

SE-581 83 Linköping, Sweden

Datum Date 2007-02-09 Språk Language Svenska/Swedish Engelska/English Rapporttyp Report category Licentiatavhandling Examensarbete C-uppsats D-uppsats Övrig rapport

URL för elektronisk version

http://www.ep.liu.se

ISBN

—

ISRN

LiTH-IFM-EX-07/1695-SE

Serietitel och serienummer

Title of series, numbering

ISSN

—

Titel

Title

Mikroarrayer för vesikelbaserad receptor-ligand växelverkan

Microarray Technology for Kinetic Analysis of Vesicle Bound Receptor-Ligand In-teractions Författare Author Björn Brian Sammanfattning Abstract

A proof-of-concept for a novel microarray used to study protein-ligand interac-tion in real-time using label-free detecinterac-tion is presented. Many of todays com-mercially available instruments lack the ability to immobilize membrane proteins. At the same time, the pharmaceutical industry develops drugs directed towards membrane-bound receptors. The need to study drug-target kinetics and to be able to screen for new medical substances is high.

To study the biomolecular interactions in real-time, imaging surface plasmon resonance (iSPR) is used. A patterned sensor surface with hydrophobic barriers assisting in the piezodispensing of NeutrAvidin with complex-bound biotin-ssDNA is created. Histidine-tagged proteins are immobilized at the vesicle surface using divalent nitrilotriacetic acid.

The concept of the vesicle immobilization, the protein-binding to vesicles and the protein-ligand interaction is initially studied using a Biacore instrument. The dissociation of the ligand IFNα2 from its receptor ifnar-2 (wt) are in accordance with the literature.

In the imaging SPR experiments, it is found that the dissociation of IFNα2 from the ifnar-2 (wt) receptor is slower than expected, probably due to rebinding of the ligand. It is also found that imidazole is needed to avoid vesicle-vesicle interaction.

The immobilization of proteins had to be done on-line i.e. when the vesicles were bound to the surface. Depending on the mixture of receptors at the vesicle surface the affinity for the ligand was changed. The results achieved were repro-ducible.

Nyckelord

(6)

(7)

Abstract

A proof-of-concept for a novel microarray used to study protein-ligand interac-tion in real-time using label-free detecinterac-tion is presented. Many of todays com-mercially available instruments lack the ability to immobilize membrane proteins. At the same time, the pharmaceutical industry develops drugs directed towards membrane-bound receptors. The need to study drug-target kinetics and to be able to screen for new medical substances is high.

To study the biomolecular interactions in real-time, imaging surface plasmon resonance (iSPR) is used. A patterned sensor surface with hydrophobic barriers assisting in the piezodispensing of NeutrAvidin with complex-bound biotin-ssDNA is created. Histidine-tagged proteins are immobilized at the vesicle surface using divalent nitrilotriacetic acid.

The concept of the vesicle immobilization, the protein-binding to vesicles and the protein-ligand interaction is initially studied using a Biacore instrument. The dissociation of the ligand IFNα2 from its receptor ifnar-2 (wt) are in accordance with the literature.

In the imaging SPR experiments, it is found that the dissociation of IFNα2 from the ifnar-2 (wt) receptor is slower than expected, probably due to rebinding of the ligand. It is also found that imidazole is needed to avoid vesicle-vesicle interaction.

The immobilization of proteins had to be done on-line i.e. when the vesicles were bound to the surface. Depending on the mixture of receptors at the vesi-cle surface the affinity for the ligand was changed. The results achieved were reproducible.

Sammanfattning

En ny typ av mikroarray för att studera interaktionen mellan proteiner och deras ligander presenteras. Mätningarna görs i realtid utan inmärkning av analyterna. Många av dagens kommersiella instrument saknar möjligheten att immobilise-ra membimmobilise-ranproteiner. Samtidigt utvecklar läkemedelsindustrin läkemedel riktade mot membran-bundna receptorer. Behovet av att studera kinetiken hos läkemedel-målmolekyl samt att snabbt kunna identifiera potentiella läkemedelsmolekyler är stort.

För att studera de biomolekylära interaktionerna i realtid används avbildande ytplasmonresonans (iSPR). En mönstrad sensoryta skapas där hydrofoba barriärer

(8)

vi

hjälper till vid piezodispenseringen av NeutrAvidin med komplexbundet biotin-ssDNA. Histidin-terminerade proteiner fästs till ytan hos vesiklar med hjälp av divalent nitrilotriacetic-syra.

Vesikelimmobilisering, proteinbindningen till vesikelytan och växelverkan mel-lan protein och ligand studeras initialt med hjälp av ett Biacore-instrument. Li-ganden IFNα2 diassociation från receptorn ifnar-2 (vildtyp) överensstämmer med litteraturen.

I de avbildande SPR-försöken visas att diassociationen mellan IFNα2 och ifnar-2 (vildtyp) är långsammare än förväntat, troligtvis p.g.a. återinbindning av ligan-den. Det konstateras även att imidazole måste användas för att undvika vesikel-vesikel-interaktion.

Bindningen av proteiner måste göras on-line, d.v.s. när vesiklarna redan be-finner sig på sensorytan. Denna nackdel används för att studera hur blandningar av olika receptorer och receptormutanter på vesikelytan påverkade växelverkan mellan receptor och ligand. De uppmätta resultaten var reproducerbara.

(9)

Acknowledgments

During this work I have had the opportunity of guidance from my two super-visors Goran Klenkar and Dr. Thomas Ederth. Each of them have contributed to this work. Their great knowledge of the production of biochips and vesicles to-gether with a lot of practical laboratory hints has helped solving many problems. I would also like to thank Prof. Fredrik Höök at Lunds Tekniska Högskola and Prof. Bo Liedberg at Linköpings Tekniska Högskola for giving me the interesting problems addressed by this work. Prof. Jacob Piehler at Johann Wolfgang Goethe-University has kindly provided crucial chemicals and proteins used in the study. Dr. Gudrun Stengel and Anders Gunnarson have been most helpful with valuable knowledge on DNA and vesicles.

Many master thesis students and employees of IFM have contributed with interesting discussions. Dr. Jochen Feichtinger and employees of CR/ART 1 at Bosch GmBH, have given me a lot of inspiration during my internship in Germany. I would also like to thank all other people who have helped me complete this work.

(10)

(11)

Abbreviations

µCP Microcontact printing.

bis-NTA bis-Nitrilotriacetic acid, a divalent chelator.

AOI Angle of incidence. BSA Bovine serum albumine.

cDNA Complementary DNA.

Diglyme bis-(2-methoxyethyl) ether.

DLS Dynamic light scattering.

EDC 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride. EDTA Ethylenediaminetetraacetic acid.

EPC L-α-Phosphatidylcholine, a phospholipid also known as egg-phosphatidylcholine.

FC Flow channel.

HBS HEPES balanced saline, a running buffer. His-tag A protein tag of repeating Histidine amino acids.

HSA Human serum albumin.

IFC Integrated microfluidic cartridge. iSPR Imaging surface plasmon resonance. LMW Low molecular weight.

NA NeutrAvidin

NHS N-hydroxysuccinimide

NSB Non-specific binding. PDMS Poly-(dimethylsiloxane)

POPC 1-Palmitoyl-2-Oleoyl-sn-Glycero-3-Phosphocholine, a phospholipid. ix

(12)

x

RB Running buffer.

ROI Region of interest.

RU Resonance unit.

SAM Self-assembled monolayer. ssDNA Single-stranded DNA.

(13)

Introduction

1.1 Background

The promising development of the DNA microarray-chip in the 1990:s has enabled high throughput analysis of e.g. disease diagnostics and increased biological un-derstanding [1]. This was made possible by the engineering science at the interface of chemistry, biology and medicine.

There are many advantages associated with the DNA chip. First and foremost is the fact that a single-stranded DNA (ssDNA) molecule shows a remarkable speci-ficity for its complementary DNA (cDNA) strand. The second advantage is that the sequence of nucleotides easily can be varied and the number of combinations is theoretically infinite and practically sufficient for most biological recognition applications. Third, the DNA molecule is also the molecule-of-choice for conserva-tion of biological informaconserva-tion and is evidently a robust biomolecule. In addiconserva-tion, the production of ssDNA-based microarrays has successfully been optimized, with high production rates and impressing density [2].

The existing DNA microarray-chip has inspired research in the area of the protein chip. The protein chip is superior to the DNA-chip, usually analyzing mRNA. This is because the influence of an organism in many aspects is directly controlled by proteins, the translated and post-translationally modified product of mRNA. Unfortunately, molecular recognition does not primarily depend on the amino acid-sequence but on the fragile tertiary structure of the protein.

The development of many new drugs depend on the study of target-drug interaction[3]. Most protein analysis systems available today, e.g. Biacore, are made for water soluble proteins. However, many contemporary drugs are directed toward targets incorporated in the cell membrane. To remain in a native ter-tiary structure, it is vital that these targets are incorporated in a lipid membrane. Figures estimate that approximately 60 % of all drug targets are cell membrane receptors[4]. In the growing pharmaceutical industry, finding and developing a potential drug candidate into a commercial product may require some ten years of development and research with a cost range US$ 500 – 800 million [5]. To avoid premature termination of a research project due to an inadequate drug molecule,

(16)

2 Introduction

it is important to properly characterize the drug candidate in an early stage of the development.

In a general perspective, the biological understanding of macro-molecular mech-anisms relies on the work done on the DNA and protein level. Adding the ability to study membrane proteins will expand this knowledge. The methods used in the manufacturing of the protein chip are already available in the semi-conductor and life science industries and would be easily applicable.

1.2 Aims of the Master’s Thesis

This study is a proof-of-concept of a biochip to study extracellular receptor do-mains of cell membrane proteins. Parallel (time-resolved) measurements allow a high-throughput analysis. The studied receptors are carried by artificial cell mem-branes (vesicles). Measurements are to be done to show that the binding to the vesicle leaves the native structure of the receptor intact. The non-specific binding (NSB) of vesicles are to be minimized.

Since this work is a proof-of-concept, only functional studies are conducted in the given time frame. No extensive quantitative surface characterizations are to be made. The receptors used in this study are not complete membrane proteins interacting with the lipids of the model cell membrane and no attempt was done to solve the problem of proper membrane protein incorporation into vesicles.

1.3 Key Ideas for the Solution of the Problem

To solve the problem described above a sensitive method of detection is needed. The utilization of the imaging surface plasmon resonance (iSPR) phenomenon has been described before [6]. Also, a micro-patterned surface with biomolecular recognition has to be created. Soft lithography is used to transfer a hydropho-bic frame. The unprinted circles are filled with hydrophilic thiols (Figure 1.1b). The patterned monolayer of molecules serve as a linker-layer for a dextran hy-drogel. The dextran hydrogel acts as an anchorage for biomolecules [7]. The biomolecules responsible for recognition are surface immobilized ssDNA. The cir-cles of the micro-pattern are made to expose different ssDNA and therefore the spots of the chip have biomolecular recognition (Figure 1.1a).

Histidine-tagged (His-tagged) receptors are bound to vesicles with the use of the divalent chelator bis-nitrilotriacetic acid (bis-NTA). In addition to membrane-incorporated bis-NTA, cholesterol-terminated double-stranded DNA is fused with the lipid membrane. Since the two DNA-strands are of different length, one region of the longer ssDNA remains un-hybridized. This region is referred to as the DNA-tag of the vesicle (Figure 1.1c) enabling the addressing of vesicles to different regions of the biochip. The use of two cholesterol anchor points (and thus two DNA-strands) enables a high and persistent surface concentration of DNA [8]. The incorporation of DNA into the vesicles is done by adding two hybridized cholesterol-terminated DNA-molecules to a vesicle solution.

(17)

1.3 Key Ideas for the Solution of the Problem 3

A´

B´

(a) Cross-section of the biochip: A frame of hydrophobic thiols (cross-marked boxes) enclose hydrophilic regions of the chip. Hydrophilic thiols act as linker-layer for dextran. NeutrA-vidin with immobilized biotin-ssDNA are anchored to the dextran. Single-stranded DNA with different sequences (A’ and B’) are exposed on different regions of the chip.

(b) Top view of the biochip. Hydrophobic frame enclosing hydrophobic circles, each with the ability to expose a unique DNA-sequence.

A

Ni 2+ Ni 2+

(c) Vesicle with incorporated chelator bis-NTA. Immobilized Ni-ions secure a His-tagged recep-tor with a complex-bound ligand. Vesicle is tagged with DNA sequence A, the cDNA of A’. Figure 1.1. Principle of the used biochip and a vesicle carrying a His-tagged receptor.

(18)

(19)

Chapter 2

Theory

2.1 Protein Kinetics – A Way to Characterize

Protein-Ligand Interaction

The biochip produced in this study can be used to study interactions between proteins and ligands. Since it also provides time-resolved data, protein kinetics can be estimated through the rate constants described below.

2.1.1 General Protein Kinetics

Many of the reactions in biological systems are of the type protein-ligand interac-tion. One way of quantitatively describing a protein-ligand interaction is by rate constants.1 _{From these constants, it is possible to calculate how an interaction}

between a protein and a ligand proceeds over time. Assuming that the reaction between a receptor R and a ligand L forms a complex C according to:

R + LGGGGGGB_FGGGGGGka kb

C (2.1)

with the forward and backward rate constants ka and kd with the relationship

kd

ka

=[R][L]

[C] = KD (2.2)

at equilibrium. The bracket symbol refer to the substance concentration. Assum-ing all the active sites of the immobilized receptors are equally active and usAssum-ing Equations 2.1 and 2.2

d[C]

dt = ka[R][L] − kd[C] (2.3)

The number of free receptors in the solution are given by

1_{The term rate constant commonly used is misleading since it varies with e.g. temperature.}

(20)

6 Theory

[R] = [R]tot− [C] (2.4)

Equations 2.3 and 2.4 give

[C]0+ [C](ka[L] + kd) = [R]totka[L] (2.5)

Assuming that the ligand concentration is constant

[C] = [C]∞(1 − e−(ka[L]+kd)t) (2.6)

where [C]∞is derived from setting the formation rate of the complex, i.e. Equation

2.3 = 0

[C]∞=

ka[L][R]tot

[L]ka+ kd

(2.7) When studying the dissociation of a ligand from a receptor the concentration of the ligand drops to zero and only the right term of Equation 2.3 is left resulting in

[C] = [C]0e−kd(t−t0) (2.8)

where t0 refers to the time when the dissociation starts.

2.1.2 Studied Proteins

The ligand used to interact with the receptors is an interferon (IFN). Interferons are usually expressed as an early antiviral response, generating antiproliferative and immunomodulatory responses. The ligand used in this work is IFNα2.

The used proteins are extracellular receptor domains ifnar-1 and ifnar-2, re-sponsible for a signal transduction triggering an anti-viral response. In addition, the C-terminal part of the receptors carry a decahistidine-tag (H10). The recep-tors can individually complex-bind IFNα2. In addition, the two moieties can form a ternary complex with the ligand when the receptors are immobilized in close proximity (Figure 2.1).[9]

2.2 Vesicles

For membrane proteins to be in their native form, it is vital that they are incor-porated in a cell membrane. A cell membrane is a complicated mixture of lipids and proteins. A simple model of a cell membrane is a lipid bilayer. Lipids are biomolecules which are only soluble in non-polar solvents like chloroform. The phospholipid is a lipid consisting of five parts: fatty chains, glycerol, phosphate and alcohol (Figure 2.2). These building blocks make the molecule amphiphilic. [1] In an aqueous solution lipids spontaneously form aggregates in order to min-imize exposure of the fatty chains to the solution. The shape of the aggregate depends of the geometrical properties of the lipid. The phospholipids of interest

(21)

2.2 Vesicles 7

i

IFN

ii

IFN

(a) The complex binding of IFNα2 to ifnar-2 (wt) is transient.

i

IFN

ii

IFN

(b) Ifnar-2 (wt) and ifnar-1 (wt) in close proximity form a binding-site with good fit for IFNα2 and a stable ternary complex is formed.

Figure 2.1. The affinity for IFNα2 of the the receptor ifnar-2 (wt) is low. A stable

complex is formed by receptors ifnar-1 (wt), ifnar-2 (wt) and the ligand IFNα2.

in this work, 1-Palmitoyl-2-Oleoyl-sn-Glycero-3-Phosphocholine (POPC) and L-α-Phosphatidylcholine, commonly known as Egg-PC (EPC), spontaneously form bilayers and are shaped into spheres with a diameter ≈ 100 nm. These are used as a model for a cell membrane with the ability to carry proteins. With the use of dynamic light scattering (DLS) , the movement of the vesicles through diffusion in and out of an illuminated volume can be studied. The result of the DLS can be used to estimate the distribution of hydrodynamic radii in a vesicle population.

Fatty Chain G l y c e r o l Fatty Chain Phosphate Alcohol

Figure 2.2. Chemical structure of a phospholipid

2.2.1 Tagging the Vesicles with ssDNA

As has been described before, the use of two cholesterol anchor points for the DNA instead of one has been shown to successfully inhibit DNA-bleeding of the vesicle. This enables the control of the number of DNA-tags per vesicle and minimizes cross-talk in a mix of differently tagged vesicles. The cholesterol-terminated DNA-strands are of different length, 15 and 30 nucleotides, respectively. The 15-mer overhang serves as the part responsible identifying the vesicle. When using the term ssDNA-tag for a vesicle, actually consisting of (partly) hybridized double-stranded DNA-molecule, the 15-mer overhang is the DNA referred to (Figure 1.1c). The cholesterol is separated from the the DNA-strand by a non-hybridized 3-mer linker. [8] A thorough account of the DNA-sequences used are to be found

(22)

8 Theory

in Appendix A.

2.2.2 Nitrilotriacetic Acid (NTA)

The use of a new kind of chromatography was reported in the mid 1970s, now known as immobilized metal (ion) affinity chromatography (IMAC). This tech-nique uses immobilized metal-chelators in a column matrix. A chelator is a molecule with the ability to form a complex with a metal ion. IMAC takes advan-tage of the fact that different proteins have varying affinity for metal ions bound to the chelators.

Although the affinity varies for different amino acids, the use of Histidine-tags has been extensively used for purification purposes. Typically, a set of Histidine amino acid residues (Figure 2.3b) are placed in the C- or N-terminus of the re-combinant polypeptides, causing a retention in a chromatography column with immobilized NTA (Figure 2.3a).

The affinity of the amino acid residues can be explained by the principles of hard and soft acids and bases. According to this principle, a bond formed by two atoms requires one to be a Lewis acid and the other to be a Lewis base. The binding energy of the bond is decided by the "hardness" or "softness" rate of the two atoms, the strongest bond being formed between a "hard" acid and a "hard" base. The transition metals ions Co2+_{, Zn}2+ _{and Ni}2+ _{are intermediate acids.}

These Lewis acids form bonds with (among others) the Lewis bases composed of aromatic nitrogen atoms or the thiol group as present in e.g. histidine and cysteine, respectively. The bonds formed between these two intermediate bases and Ni2+ seem to be the strongest formed by the amino acids. [10]

To be able to attract a His-tagged receptor, an electron-accepting metal at first has to be secured by the exposed NTA at the vesicle. This is done by "loading" the surface-immobilized chelators by an injection of Ni2+. Subsequently, His-tagged receptors are flowed over the sensor surface and bind to the chelate. The proteins can then be released by injecting excess amounts of imidazole (Figure 2.3c) competing for binding sites at the chelates. (Figure 2.4)

2.3 Creating the Biochip

2.3.1 Self Assembled Monolayers (SAMs)

In the field of biosensors, the use of SAMs on a gold substrate has so far been of great importance. Although SAMs also form on other noble metals such as silver, platinum and copper, gold is the preferred metal as it does not readily form a stable oxide with the surrounding atmosphere, with resulting imperfections due to impurities. The simplest form of a SAM consists of a monolayer of molecules with a thiol head group covalently bound to an alkane tail. The length of the tail and its termination can be varied (as shown by letters n and X, respectively in Figure 2.5a) enabling different surface properties of the chip.

Hexadecanolthiol (used by Biacore in the CM5-chip production) is chosen as a candidate for the linker-layer of the dextran (Figure 2.5c). Another molecule,

(23)

2.3 Creating the Biochip 9 O H N H + OH O N H O O NH OH O O NH O N H + O H O H OH O O O N O (a) bis-NTA N H N N H₂ COOH (b) Histidine N H N (c) Imidazole

Figure 2.3. bis-NTA loaded with Ni2+has a high affinity for Histidine and imidazole

Tri-(ethylene glycol) (EG3) has been shown to resist the adhesion of proteins [11]

[12] and is therefore the second candidate for the linker-layer (Figure 2.5d). Hex-adecanthiol is chosen to be the hydrophobic frame.

A gold surface immersed in a solution of thiol molecules dissolved in ethanol readily adsorbs a monolayer during two phases. In the first, relatively quick phase (with a duration of a couple of minutes) the thiol head group chemisorbs to the gold surface in an exothermal reaction, releasing approximately 40 - 45 kcal of energy per mole [13].

Making the following assumptions:

• Dissolved molecules are constantly colliding with the surface. Only the molecules colliding with an unoccupied site can chemisorb to the surface,

i.e. adsorption is restricted to a monolayer.

• At concentration C and a constant temperature, a dynamic equilibrium ex-ists between the surface and the adsorbed monolayer.

• Adsorption sites are uniformly distributed. Sites are equivalent in the ability to bind one adsorbate only.

• Once chemisorbed to the surface, the activation barrier restricts lateral molecule movement. Enthalpy of chemisorption to a site is unaffected of surface coverage.

(24)

10 Theory Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ Ni2+ His-tagged protein Imidazole Imidazole Imidazole

Figure 2.4. Principle of His-tagged protein binding for monovalent NTA at a vesicle

surface. Figure not to scale.

S H C H₂ X n (a)

Chemical structure of alkane thiol

S H CH₃ (b) Hexadecanthiol S H OH (c) Hexadecanolthiol O O O OH O S H (d) EG3

Figure 2.5. Thiols used in the study

the adsorption of thiol molecules to a gold surface can be modeled using the Langmuir adsorption isotherm.[14]

In the slower, second phase (where the Langmuir isotherm is not valid), the un-organized thiol tails start to interact through Van der Waals’ forces. This phase

(25)

2.3 Creating the Biochip 11

is responsible for the uniformity and organization of the monolayer. A longer alkyl chain makes the intermolecular forces stronger and force a faster reaction. Also, a longer tail group gives a closer packing of molecules on the surface, minimizing pinholes and other surface defects.[13] This obviously also depends on the nature of the adsorbed molecules. If the molecule is more complex than a simple alkane thiol, pinholes due to bulky head or tail groups are not avoided by using long tail groups.

2.3.2 Microcontact Printing

A widely used technique for creating SAM micro-patterns is soft lithography or microcontact printing (µCP) . The technique uses an elastic stamp, usually made from poly-(dimethylsiloxane) (PDMS) . PDMS is made by mixing Sylgard 184 and a curing agent followed by degassing in vacuum. The mixture is then poured on top of the master relief and cured in a hot oven. Curing time and shrinkage varies with baking temperature. PDMS is an inert material with a low surface free energy (≈ 21.6 · 10−3 Jm−2 [15]) not readily adsorbing dust.

Using µCP enables the reproducible printing of structures as small as 0.2 µm with PDMS stamps.[16] As the pattern of µCP is simultaneously transferred, it is a fast method compared to e.g. electron-beam lithography or other scanning methods. Ordinary photolithography is used to create the master template. This master can then be reused numerous times to create many PDMS stamps, each with the possibility to be used some hundred times under normal atmospheric conditions, dramatically reducing fabrication costs. The drawback of µCP is that it requires a drying period of the ”ink” (i.e. the biomolecular solution to be transferred) which induces loss in biological activity, especially among proteins.[17]

2.3.3 Piezodispensing

There are different ways of transferring a defined pattern of biomolecules to a surface, often involving spotting techniques similar to an ordinary ink-jet printer. Piezodispensing enables a small volume (≈ 100 pL) to be deposited on a surface [18]. The main drawback of this molecular transfer is the problem of controlling the position of individual droplets. Dispensed droplets spontaneously aggregating would result in the mixture of protein complexes and ultimately the cross-talking of vesicles when using the biochip.

Combining the two approaches, i.e. first use µCP to print a hydrophobic barrier of a stable organic molecule assisting in the piezodispensing of sensitive biomolecules, effectively utilizes the advantages of both transfer techniques. At the same time the drawbacks are minimized. Depending on the equipment used, the lateral feature size can be as small as ≈ 100 µm.[18]

Wettability

A micropatterned surface with different surface energies has different wettability properties. In this study, the difference stems from the fact that the frame of the pattern is hydrophobic due to the methyl-termination of the thiol. The back-filled

(26)

12 Theory

circles with hydroxy-terminated thiols are hydrophilic. The adhered dextran of these areas also contributes to this property. To be able to perceive a contrast on the apparently uniform surface, water vapor is condensed on the surface. This is done by cooling the surface with a Peltier element at a high humidity. This makes the pattern appear, enabling a way to find the circular shapes where the different NeutrAvidin-Biotin-ssDNA complexes are to be spotted.

2.3.4 Dextran

The immobilization of a protein is paramount for the performance of the biosen-sor. A poor immobilization at best results in a low response. In worse cases, the result may be an absent response or even artifacts. Different approaches have been attempted, e.g. adsorption to a solid support. This method has been favored be-cause of its simplicity and mild chemistry. The problems associated with the solid support adhesion have been conformational changes and random adhesion site of the protein, both possibly affecting the affinity for the ligand. Also, the adsorption of the protein may be reversible, giving rise to dissociation and rebinding of the protein. [19] [7]

It has been shown that the use of a polymer matrix where a large part of the volume is occupied by solution resembles the natural environment for biomolecules. [20] This contributes to preserving the biological activity in the immobilized pro-tein. When comparing a flat two dimensional surface to a three dimensional coun-terpart the available sites for protein adsorption favors the matrix substrate. Also, studies conducted on the dextran matrix used in the instruments produced by Bia-core, show low levels of non-specific binding. Dextran is a high molecular polymer of glucose produced by bacteria. It is biocompatible, water soluble, and has a long shelf life. The molecular weights available range from 1 – 2000 kDa.

Dextran can be immobilized on a hydroxy-terminated linker-layer adsorbed to a gold surface as can be seen in Figure 2.6a according to the following steps:

i Epichlorhydrine dissolved in NaOH, H2O and bis-(2-methoxyethyl) ether (diglyme)

is added to the surface.

ii Epichlorhydrine covalently binds to the SAM, replacing a hydrogen atom. iii Dextran dissolved in NaOH and H2O replaces epichlorhydrine and bind to the

SAM.

iv Bromoacetic acid in NaOH and H2O is added to the surface, introducing

car-boxyl groups to be used for amine coupling to the dextran.

The immobilized dextran is then activated as described by Figure 2.6b: i 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC)

N-hydroxysuccinimide (NHS) is mixed 1:1 in H2O and added to the surface.

ii The hydrogen atom of the carboxyl group is replaced by NHS. iii The protein of interest (-R) is chemisorbed to the surface.

(27)

2.3 Creating the Biochip 13 S OH S OH S OH S OH S OH O Cl S O S OH S OH S OH S OH O S O S OH S OH S OH S OH S O S OH S OH S OH S OH COOH COOH COOH COOH

i

ii

iii

iv

(a) Dextran coupling to a hydroxy-terminated linker-layer on a gold substrate: Diglyme binds to the surface (i–ii). Diglyme is replaced by dextran (iii) which is carboxylated (iv) to be used for amine coupling.

C OOH C OOH C OOH O O C OON C OOH O O C OON C ONH-R C OOH O O C OON C ONH-R C OOH C ONH-(CH 2 )2 -OH

i

ii

iii

iv

(b) Dextran activation, coupling and deactivation: NHS replace the hydrogen atoms of carboxyl groups at the dextran (ii). This allows for coupling of primary amines in proteins (iii). The remaining reactive groups on the dextran are deactivated (iv).

Figure 2.6. Dextran is used as anchorage for primary amines found in proteins. Figure

not to scale.

2.3.5 The Neutravidin-Biotin-ssDNA Complex

The affinity for vitamin H, D-biotin found in the streptavidin protein is one of the strongest biomolecular non-covalent bindings known with KD≈ 10−13M [21].

The protein has pI ≈ 6.7 and a molecular weight of ≈ 60 kDa. The approximately 4.5 × 4.5 × 5.8 nm protein is a tetramer with four identical polymer chains, each consisting of 159 amino acid residues. Each streptavidin molecule is capable of binding up to four biotin molecules. The strength of the binding is made from numerous hydrogen bonds and van der Waals interactions. Once associated with the protein, biotin is practically irreversibly bound deep inside a β-barrel of the protein.[21] NeutrAvidin (NA) is a modified protein similar to streptavidin known

(28)

14 Theory

for its inertness towards other proteins due to its deglycosylation. As reported earlier, this modification does not affect the binding of biotin. [22]

Biotin (Figure 2.7) is a relatively small water soluble molecule that can be appended to either terminal of a ssDNA molecule. This molecule, consisting of an arbitrary number of nucleotides with an arbitrary sequence enables a tool with high specificity for a theoretically infinite number of biomolecules using the setup of this study. NH N H S COOH O

Figure 2.7. Chemical structure of biotin.

2.4 Null Ellipsometry

Ellipsometry has been widely used by the scientific community for the last hundred years to determine surface properties like layer thickness of different materials. In electromagnetic (EM) radiation, the electric field vector ~E dominates the interac-tion with non-magnetic matter and is the one of the two fields to be considered in ellipsometry.

2.4.1 The Polarization of Light

The vector ~E of monochromatic light can be described using harmonic oscillations. Defining the z-axis as the EM wave path, ~E can be described in a two-dimensional coordinate system. Polarization is the movement of the vector field at a fixed point (z) over time. Let

Ex(t) Ey(t) = X cos(−ωt + δx) Y cos(−ωt + δy) (2.9) describe ~E where X and Y are the amplitude of the oscillations, ω the angular frequency and δ the cosine argument of the two components at t = 0. As the x and y component have the same frequency but different amplitude and start phase, the resulting trajectory is an ellipse in the x, y-plane. A special case arise when the phase of the two oscillations is shifted 0◦ degenerating the ellipse into a straight line.

Let the light ray propagate in a plane perpendicular to the surface. ~E is commonly described using a coordinate system where p is in the plane of propagation and s is perpendicular to this plane. The x and y components of ~E are expressed as p and s, respectively, see Figure 2.9.

(29)

2.4 Null Ellipsometry 15

p

s (X,0) (0,Y)

(a) General case

p

s (X,0) (0,Y)

(b) Phase shift is 0◦

Figure 2.8. The trajectory of the x and y components of ~E is described by two harmonic

oscillations with amplitude X and Y

n

p

E s

Figure 2.9. ~E is resolved using axes s and p, aligned to the plane of incidence.

2.4.2 Optical Component Setup

Controlling the incident light and measuring the reflected light is done using three components. A linear polarizer (P) is used to create the situation depicted in Figure 2.8b. The direction of ~E becomes fixed and only varies in amplitude. The plane polarized light is now directed to a compensator (C) that is used to phase-shift the polarized light in a desired way. This component again gives the incident beam an elliptical shape, Figure 2.8a. After these modifications the incident beam is reflected at the surface, once again affecting the shape of the ellipse.

The analyzer (A) is also a polarizer. When lining up a pair of ”crossed” polar-izers, theoretically no light can be transmitted. This is the condition upon which the null ellipsometry rests: use P and C to generate a polarization of the light which at the surface become plane polarized and set A to extinguish the reflected light. One way to do this is by registration of the light passing A and iteratively

(30)

16 Theory

modify the relative position of P and A until the null condition is achieved. The P, C and A angles are then transformed into the ellipsometric angles Ψ and ∆. The angles are then used in mathematical models to obtain surface properties like thickness and index of refraction.[23]

2.5 Surface Plasmon Resonance (SPR)

A surface plasmon can be described as an oscillation in a confined electron gas at the interface of two media, one being a dielectric and the other a metal. The metal used in this study is gold, chosen for its chemical stability, chemical reaction with thiols (Section 2.3.1) and its physical properties in accordance with the free electron model.

In a Kretschmann SPR configuration [24], two media of different optical den-sity, e.g. glass and sensing medium, are separated by a thin gold film. Monochro-matic p-polarized light (Section 2.4.1) passes through a prism before striking the interface of the two media. The angle between the normal of the surface and the incident light is defined as θ. At the interface, incident light of angle θc < θ < θsp

is subject to total internal reflection and thus all of the incident light is directed towards the detector. However, one component of the incident light, the evanes-cent wave, penetrates approximately one wavelength into the metal film. The wave vector of the evanescent field is given by

Kev=

ω

cnpsin θ (2.10)

where np is the refractive index of the light guiding prism, ω the frequency of the

incident light and c the speed of light in vacuum. For plasmons at an interface, the wave vector is approximately

Ksp≈ ω c s mn2diel m+ n2diel (2.11) where m is the relative permittivity of the metal (at a given ω) and ndiel the

refractive index of the dielectric (sensing medium). At θspthe photon momentum

parallel to the light path is transferred to the surface plasmon and Kev = Ksp.

Solving for θsp gives

θsp= arcsin_n1 p r mn2_diel m+n2diel |m| |ndiel|2    ⇒ θsp∝ arcsin ndiel np (2.12)

The excitation of the surface plasmon at resonance condition cause a drop in intensity of the reflected light, detectable by a light sensitive sensor. The detectable resonance condition is dependent of ndiel. [24] Using SPR, the measured ndiel

primarily depends on the molecules adsorbed to the sensor surface. In addition, the Kretschmann setup allows for fast sampling of the refractive index and therefore time-resolved adhesion studies. The response of a Biacore instrument is given in resonance units (RU) where 1000 RU is equal to an adsorption of 1 ng/mm2. As

(31)

2.5 Surface Plasmon Resonance (SPR) 17

can be seen in the curvature of a typical SPR-curve, the angles around θsp in

Figure 2.10 allows for a very sensitive sensor signal.

Intensity (AU)

Incident angle (°) (90°)

(θ_c) (θsp)

Figure 2.10. Approximate response from detector. Reflected intensity dependence of

incident angle. Note the low reflected intensity at θsp.

2.5.1 Imaging SPR

Imaging SPR can be used for measuring the refractive index of a surface. The difference is that the angle of incidence is fixed and that a charge coupled device (CCD) is used as a sensor device. The advantage of this method is of course that a number of different regions can be measured in parallel, which is necessary for i micro-array. To do this, different regions of interest (ROIs) are defined at the surface (white squares in Figure 2.11). The software (EP3_{-view, Nanofilm) gives}

(32)

18 Theory

300 µm

Figure 2.11. A sensor surface docked in the flow cell is depicted using imaging SPR in a

Kretschmann configuration. White rectangles are the different ROIs. A higher refractive index corresponds to brighter areas.

(33)

Chapter 3

Experimental Details

The running buffer HEPES balanced saline (HBS) consists of 10 mM HEPES and 150 mM NaCl dissolved in purified water and adjusted to pH 7.4.

3.1 Surfaces

The surfaces used in the study were manufactured by evaporation deposition us-ing a Balzers UMS 500 P system. Base pressure was below 10−9 mbar and the evaporation pressure was below 10−7 mbar.

SPR surfaces were prepared by deposition of 300 Å of gold after evaporating a 10 Å titanium layer on a SF10 glass substrate (Schott, Germany). Titanium and gold deposition rates were 1 and 2 Å/s, respectively. Ellipsometry surfaces were made by deposition of 25 Å titanium on silicon (100) wafers followed by a 2000 Å gold deposition. Titanium and gold deposition rates were 1 and 10 Å/s, respectively.

3.2 Preparation of PDMS Master and µCP Stamp

A silicon wafer was cleaned for 10 minutes using the TL-1 protocol, described in Section 3.3. The negative photoresist SU8 was used to spin coat the wafer prior to UV light exposure through a patterned mask. The photoresist below the transparent quartz windows of the mask is exposed to the light inducing cross-linking in the photoresist. Unexposed areas of the photoresist are removed using a developer and the desired structures are produced.

The PDMS-stamp is created by mixing Sylgard 184 and a curing agent. The mixture is degassed in vacuum and poured on top of the master relief. The PDMS is then cured in a hot oven at ≈ 140℃.

(34)

20 Experimental Details

3.3 Preparation of SAMs

All surfaces, tweezers and glassware were cleaned using the TL-1 protocol to re-move organic contaminants. Objects were immersed in 5 parts of purified water, 1 part 33 % H2O2 and 1 part 27 % NH3 stirring the solution before heating to

85℃ for 5 minutes. After heating, the objects are rinsed thoroughly in purified water and dried in a stream of nitrogen gas.

3.3.1 µCP and Back-filling

A slab with a 12 × 12 matrix of the desired 300 µm diameter features was cut out from the PDMS and sonicated in 99.5 % ethanol for 5 minutes to remove low molecular weight (LMW) molecules from the polymer. Hexadecanthiol was dissolved in 99.5 % ethanol to a concentration of 10 mM. The solution was added to the PDMS stamp during 1 minute and the stamp was dried for ≈ 60 seconds in a nitrogen gas stream. The stamp was placed on the gold surface for ≈ 30 seconds. A pair of tweezers was used to gently push on top of the stamp during this last step to assure proper contact between the surface and the stamp. After patterning the surface with the hydrophobic hexadecanthiol, the chip was put in the cleaned incubation beaker containing 0.5 mM EG3 over night.

3.4 Dextran

3.4.1 Dextran Coupling to the SAM

The following protocol is used to covalently couple dextran to the SAM:

• Solutions A and B were cooled to 15 ℃. Solution A was added to a deep vessel containing the patterned surfaces. The vessel was kept in a water-bath, tempered to 15 ℃. Solution B was added two minutes after Solution A. Vessel was stirred and and left for 16 hours.

Solution A Solution B

4 M NaOH 3 ml Diglyme 6 ml

Purified water 27 ml Epichlorhydrin 3 ml

Diglyme 24 ml

• Surfaces were washed thoroughly in purified water and dried in a stream of nitrogen gas.

• 1.32 g 40 kDa dextran (GE Healthcare) was dissolved in a 27 ℃ heated vessel containing ≈ 83.3 ml 0.4 M NaOH. The solution was added to a vessel containing the dried surfaces and incubated at 27 ℃ and gently shaken for 26 hours.

• Surfaces were cleaned using copious amounts of purified water.

• 6 g bromoacetic acid was added to ≈ 21.2 ml purified water and tempered to 11 ℃. Approximately 24.5 ml 4M NaOH was added to the bromoacetic

(35)

3.5 NA-Biotin-ssDNA 21

acid during stirring for a few minutes. Surfaces were incubated in solution at 25℃ for 16 hours.

• Surfaces were cleaned using copious amounts of purified water.

• A stabilizing solution was made and surfaces were incubated and gently shaken therein for 45 minutes before drying and subsequent storage in a sealed nitrogen atmosphere at 8℃.

Stabilizing Solution D(+) sucrose 250 mg

NaHCO3 171 mg

Na2CO3 49 mg

Purified water is added to get 250 ml stabilizing solution.

3.4.2 Biotin Coupling to Dextran

The surfaces, stored as described in Section 3.4.1 were rinsed in purified water and dried in a stream of nitrogen gas. In order to activate the dextran and be able to covalently couple the aminebiotin, the following protocol, described by [7] was used:

• To activate the carboxyl groups of the dextran, frozen NHS and EDC, each dissolved in purified water, were thawn and mixed to a final concentration of 50 mM and 200 mM, respectively. The solution was immediately added to the chip surface and allowed to react for 20 minutes before rinsing with purified water and drying in a stream of nitrogen.

• 5 mM aminebiotin in purified water was allowed to chemisorb to the surface. After 20 minutes, the surface was rinsed with purified water and dried in a stream of nitrogen.

• The surface was deactivated with 0.05 mg/ml ethanolamine. After 20 min-utes, the surface was rinsed with purified water and dried in a stream of nitrogen.

3.5 NA-Biotin-ssDNA

Four complexes with different ssDNA (Medprobe) were prepared. 5 µl 1 mg/ml NeutrAvidin was added to 91.7 µl HBS and vortexed. Subsequently, 3.3 µl 50 µM biotin-ssDNA A’, B’, C’ or D’ (see Appendix A) were added and the solution was vortexed.

3.6 Piezodispensing

The humidity of the lab was controlled in the region of 80–90% by boiling water. A manual x/y-stage and a microscope was used to align the 60 µm diameter

(36)

22 Experimental Details

capillary of the piezodispenser (Microdrop, Germany). A Peltier element on the

x/y-stage was used to cool the surface just below the dew temperature and a

contrast between the hydrophobic frame and the hydrophilic circles was observed. To maintain a stable humidity, a plastic cover was kept over the x/y-stage.

The prepared NA-biotin-ssDNA complexes were allowed to react for ≈ 1 hour before aspiration into the capillary. Each complex was dispensed at four different spots, each of the four replicates with a total volume of ≈ 0.5 nL [6]. Figure 3.1 depicts the dispensing scheme used.

B‘

C‘

D‘

A‘

Figure 3.1. Each of the four different NA-biotin-ssDNA solutions were dispensed at

four different sites of the biochip.

To avoid vesicle cross-talk due to contamination of the previously dispensed system, a thorough washing involving 12 kHz sonication of the capillary was con-ducted between the dispensing of the protein complexes.

After dispensing the NA-biotin-ssDNA complexes, the surface was incubated in a 1 mg/ml bovine serum albumine (BSA) in HBS solution during shaking for 15 minutes before mounting the chip in the flow cell. This was done to saturate the surface with proteins. When proteins cover the biochip it becomes hydrophilic (the largest part of the chip area consists of the hydrophobic frame) and air is not as easily trapped at the sensor surface when docked in the flow cell. In addition, non-specific binding of proteins to the surface is minimized as BSA proteins adhere to the exposed linker-layer.

3.7 Vesicle Preparation

A flask with flat bottom and a stopper was cleaned using the TL-1 protocol. To measure the volume of solutions dissolved in chloroform, glass syringes with stainless steel needles were used. Syringes were cleaned thoroughly in chloroform and ethanol and were subsequently dried in a stream of nitrogen gas. A small arbitrary volume of chloroform, either 125 µl 40 mg/ml POPC or 200 µl 25 mg/ml

(37)

3.8 Surface Characterization 23

EPC and 280 µl 1.25 mM bis-NTA were aspired in a glass syringe and added to the flask. Both phospholipids and bis-NTA were dissolved in chloroform. The flask was rotated in a fume hood and a stream of nitrogen gas was led into the flask evaporating the chloroform for 15 minutes.

Subsequently, the flask was put in a chamber which was evacuated. The flask was left in the vacuum chamber for 1 hour at with a pressure below 10−6 mbar.

1 ml of HBS was added to the flask which was subsequently vortexed to assure a proper dissolving of the lipids.

A clean syringe was used to aspire the solution and was then mounted on an extruder with double 100 nm pore-size filters. The solution was passed over the membranes 31 times. The extruded vesicle solution was mixed 1:4 with HBS.

3.7.1 DNA-tagging of the Vesicles

Each of the ssDNA (Medprobe) A, B, C and D was mixed 1:1 with their common cDNA X and allowed to hybridize for one hour. The volume of the 5 µM solution was 250 µl. The hybridized cholesterol-terminated DNA are referred to as A&X, B&X, C&X and D&X. Each of the hybridized DNA-solutions were mixed with 250 µl of the prepared vesicle solution with subsequent vortexing, producing 500 µl ssDNA-tagged vesicles.

3.8 Surface Characterization

Imaging null ellipsometry was used to measure the topography of the surface after µCP. The wavelength was 532 nm and angle of incidence (AOI) was 60◦. The complex refractive index of the metal was set to the average of three measurements on different parts of the surface. The real and imaginary part of the refractive index of the organic film was set to 1.5 and 0, respectively [13].

3.9 SPR

For non-imaging SPR measurements, a Biacore 3000 instrument was used. For imaging SPR, an imaging null ellipsometer, EP3 _{(Nanofilm Surface}

Anal-ysis, Germany) with a xenon light source was used. The biochip was mounted in a flow cell and a 60◦ SF10 prism in a Kretschmann setup was used [24]. An in-terference filter was used to select the desired wave-length. Optimum wavelength and AOI was selected for each chip depending on the varying thickness of the gold film deposited on the different glass substrates. An intensity-image of the surface was recorded and an alignment was done to produce a vertically and horizontally flat plane before each measurement. Square regions of interest were drawn in the image of the surface, all ROIs sampled in parallel with a sample rate of 1/30 Hz. Flow was set to 10 µl/min. The reason for using these, rather moderate settings, was to avoid mechanical stress in the imaging ellipsometer and the pump. Before injecting the ligand, flow and sample rate were doubled to reduce re-binding of the ligand and to measure the critical time dependent data.

(38)

(39)

Chapter 4

Results & Discussion

4.1 Non-imaging Measurements using Biacore

In order to test the injection times to reach saturation, the binding performance of the ssDNA-tagged vesicles, adsorption responses, etc., measurements were con-ducted using a Biacore 3000 instrument and a CM5-chip. The Biacore 3000 instru-ment has the ability to measure the response of 4 flow channels (FC)s in parallel. Injections can be made to individual FCs or to all FCs connected serially. This allows for a surface modification in 4 different ways, i.e. 4 ssDNA-exposing protein complexes can be immobilized and the reactions can be measured in parallel.

Low flow rates (1 µl/min) were used to mimic the fact that the piezodispensed protein complexes were not to flow across the surface but rather be added in a lim-ited amount. The same is true for the protein complexes added by piezodispensing. Thus, replacing chemisorbed NA-biotin-ssDNA was only possible by diffusion of molecules from the limited volume above. When injecting vesicles, flow rate was set to 10 µl/min. The ssDNA used to tag the vesicles were A&X or B&X.

4.1.1 Adsorbing ssDNA-tagged Vesicles to the Sensor

Sur-face

Aims of the Experiment

In an early stage of the study, it was crucial to measure the binding performance of the NA-biotin-ssDNA complex and the vesicles tagged with the complementary ssDNA. Once aminebiotin was immobilized, the biotin-binding protein complex (NA-biotin-ssDNA) could be secured at the surface. The goal of this experiment was to show the addressability of the vesicles only depended on the ssDNA-tag. In addition, the ability to regenerate the chip by releasing the vesicles and keep the functionality of the biosensor was studied.

(40)

26 Results & Discussion Results & Discussion of the Experiment

Dextran activation, aminebiotin coupling and deactivation, each step lasting 20 minutes, was used to covalently bind aminebiotin to the surface. An approximately linear adsorption due to mass transport limitation of the two protein complexes was observed. No saturations were achieved and the injections of the NA-biotin-ssDNA were aborted. The reason for ending the injection after 65 minutes is simply a trade-off between time consumption and the ability to adsorb enough vesicles. If the number of adsorbed vesicles is too low, it is later impossible to observe the adhesion and dissociation of the receptor ligands. An adsorption of ≈ 3000 RU NA-biotin-ssDNA was observed for both systems after a 65 minute injection (Figure 4.1a). The protein complex is immobilized to the dextran in a stable manner.

Flow was increased to 10 µl/min and 100 µl untagged vesicles did not adsorb as they were injected over the sensor surface. This can be seen as the difference before and after the injection is virtually zero (at beginning and ending of Label 1 in Figure 4.1b). Missing data at the beginning of Label 1 are omitted samples due to air at the sensor surface.

A 200 µl injection of 0.5 mg/mL EPC B&X-tagged vesicles (Label 2) and a subsequent equal injection of vesicles tagged with A&X (Label 3) showed that NSB was present to some extent, especially in FC 2 (Label 4). To relate different NSB, a comparison of the non-specific and the specific response is made. In this case the response of FC 2 to an injection of A&X-tagged vesicles is divided by the response of FC 1 to the same injection. This gave FC 2 a NSB of 5.2 %.

The chip was regenerated with pulses of purified water and 50 mM NaOH. When running buffer (RB) was reintroduced at the sensor surface, roughly the same response levels were achieved, indicating that the regeneration de-hybridized the double-stranded DNA and therefore released the vesicles. As can be seen, the activity of the NA-biotin-ssDNA complex remains after the regeneration as the approximate previous response levels were reached when the injection order was permutated (compare response of FC1 Label 3 and Label 4). Again, non-specific binding was observed (8.5 % for FC 2). NSB of FC 1 was only 1.5 % and 2.4 % for first and second injection of B&X-tagged vesicles. A reference subtraction of one reference channel with no immobilized NA-biotin-ssDNA (not shown) does not reduce the NSB.

Conclusion

Vesicles with no incorporated ssDNA does not bind to the surface. Vesicles tagged with ssDNA on the other hand are addressable to the right FC. The NSB is thought to (at least partly) depend on an insufficient cleaning of the integrated microfluidic cartridge (IFC) and the injection needle. After the first immobilization of NA&A’, the protein complex is thought not to be completely washed out of the injection needle, capillaries and IFC, therefore being injected together with NA&B’ into FC 2. As can be seen in Figure 4.1b, vesicles tend to bind more unspecific to FC 2. This could reflect a poor washout of NA&A’ and so vesicles tagged with A&X are adsorbed to FC 2 to a greater extent than those tagged with B&X to FC 1.

(41)

4.1 Non-imaging Measurements using Biacore 27

Vesicles tagged with A&X seem to bind unspecific only to a small extent.

A more extensive cleaning protocol involving the on-board cleaning procedure of the IFC and injection needle together with additional washouts was used in the remaining experiments.

It was also shown that the immobilized NA-biotin-ssDNA complex is tightly bound to the dextran. The regeneration of the biochip, releasing the adsorbed vesicles is non-destructive to the surface and vesicles are addressable after a re-generation. Time (s) 0 2000 4000 6000 Response (RU) 0 500 1000 1500 2000 2500 3000 3500

(a) Linear Adsorption of NA-biotin-ssDNA due to mass transport

Time (s) 0 3500 7000 10500 14000 ) U R( es no ps e R -500 0 500 1000 1500 2000 2500 3000 FC 1 FC 2

12 3

4 5

(b) Vesicles addressed to different flow channels based on ssDNA-tag

Figure 4.1. NA-biotin-ssDNA adsorption and the addressing of ssDNA-tagged vesicles

to different flow channels based on ssDNA-sequence.

4.1.2 Detection of Receptor-Ligand Interaction

The response of the interaction between receptor ifnar-2 (wt) and the ligand IFNα2 (wt) was investigated. Two kinds of vesicles were used and vesicles were tagged with DNA A&X and B&X. B&X-tagged vesicles also had bis-NTA incorporated in the lipid bilayer allowing the binding of His-tagged receptors. NA-biotin-ssDNA protein complexes were immobilized at the sensor surface as described in Section 4.1.1 (not shown in plot).

Results & Discussion of the Experiment

Vesicles were adsorbed to the different flow channels and 10 mM NiCl2was injected

during 3 minutes over all flow channels (not shown in plot) . 65 µl ifnar-2 (wt) was injected and bound primarily to B&X-tagged vesicles in FC 2. The response to the injection was 175 RU compared to 30 RU from the channel with adhered

(42)

28 Results & Discussion

vesicles without bis-NTA. The response of FC 2 and FC 1 to the injection of 60 µl 100 nM IFNα2 (wt) was approx. 270 RU and 50 RU, respectively.

Since the vesicles immobilized in FC 1 are not supposed to be able to attract the protein this non-specific response was quite large. To be able to distinguish the amount of protein bound to the biochip and the immobilized vesicles, a reference subtraction was made, subtracting the average response of the blank FC 3 and 4 from FC 1 and 2 as can be seen (thick lines) in Figure 4.2. Dividing the response of FC 1 with FC 2 gives the NSB response of FC 1. After reference subtraction NSB is at the most 6 %.

The resulting "flat" line from the reference subtraction of FC 1 shows that the response observed in FC 1 is probably due to NSB of the biochip rather than from the immobilized vesicles (not carrying bis-NTA) of FC 1.

The dissociation constant was measured to be 0.0096 s−1. For a single mea-surement this can be seen as being in accordance with 0.011s−1_{measured by [25].}

FC 1 FC 2 FC 3 FC 4 FC 1 Ref. Subtracted FC 2 Ref. Subtracted Time (s) 0 200 400 600 800 1000 1200 1400 1600 Response (RU) 0 50 100 150 200 250 300

Figure 4.2. Adsorption of ifnar-2 (wt) and IFNα2 (wt) to vesicles with and without

bis-NTA, FC 2 and FC 1, respectively. Reference flow channels are FC 3 and 4. Reference-subtracted responses of FC 1 to 2 is indicated by thick lines. Dividing the response of FC 1 with FC 2 gives the NSB response of FC 1, NSB is at the most ≈ 6 %.

Conclusion

The adsorption of receptors to vesicles immobilized to the sensor surface is de-tectable. The receptors immobilized by bis-NTA carrying vesicles show a good specificity for the injected ligand. The measured dissociation constant kd is in

good agreement with the literature and therefore receptors are functionally

immo-bilized at the vesicle surface. In addition, it was shown that the NSB of receptors

to vesicles not carrying bis-NTA is low. Vesicles with no immobilized receptors showed a very low binding of the ligand. The maximum NSB of FC 1 was ≈ 6 %.

(43)

4.2 Qualitative Surface Characterization 29

4.1.3 Non-Specific Binding of Hexadecanolthiol and EG

3

To investigate if the SAM underneath the dextran matrix could affect NSB of the synthesized chips, two surfaces were coated with different SAMs before a 40 kDa dextran immobilization was made. The SAMs compared were hexadecanolthiol and EG3. Since the vesicles are relatively large compared to normal proteins, they

are unable to interact with the dextran-covered linker-layer. Proteins are smaller and are able to interact with the linker-layer. To make a fast, uncomplicated and easily realizable test, human serum albumin (HSA) was adsorbed to the dextran matrix of one FC. The other flow channels were left unmodified and the responses of the "blank" flow channels were used as a reference. Again, NSB is calculated as the reference response divided by the specific response.

Results & Discussion of the Experiment

0.1 mg/mL HSA in pH 4.5 was immobilized in one of the four flow channels on the two different chips. A solution containing 0.05 mg/mL anti-bodies for HSA (α-HSA) was injected over all channels. The adhesion of α-HSA to the reference flow channels and the flow channel containing the immobilized HSA was compared for the two thiols. As can be seen in Figure 4.3 the response of the reference flow channels on the EG3-chip is much smaller than that of the hexadecanolthiol-chip.

A subtraction of the bulk-contribution (9 RU, not shown in figure) of the α-HSA solution was made. Comparing the average reference response to the αHSA-HSA response gives a non-specific binding of 28.6% for the hexadecanolthiol compared to 5.36% for the EG3.

Conclusion

Replacing hexadecanolthiol with EG3 as the linker-layer for dextran seems to

minimize the NSB of proteins to the synthesized sensor surface. The results suggest that approximate 5 times less NSB can be achieved.

4.2 Qualitative Surface Characterization

Since this work is a proof-of-concept for a novel microarray rather than a study on the characterization of the synthesized surface, the focus lies mainly on functional studies. Two qualitative kinds of investigations of the surface were found to be of very important to the final results. These were imaging ellipsometry investigations of the soft lithography and microscopic wetting after the finalized matrix synthesis.

4.2.1 µCP Performance

The patterned surface created using µCP is vital for the piezodispensing. To measure the performance of the control the µCP and to assure that the technique

(44)

30 Results & Discussion Time (s) 0 200 400 600 800 Response (RU) 0 1000 2000 3000 4000 HSA on C16 Blank 1, 2, 3 (a) Adsorption of α-HSA to hexadecanolthiol-chip

HSA on EG3 Blank 1, 2, 3 Time (s) 0 200 400 600 800 1000 Response (RU) -500 0 500 1000 1500 2000 2500 3000 3500

(b) Adsorption of α-HSA to EG3- chip

Figure 4.3. Non-specific binding for the different back-filling molecules. EG3 has

ap-proximately 5 times less NSB of α-HSA compared to hexadecanolthiol.

is properly applied, imaging ellipsometry measurements were performed. Results & Discussion of the Experiment

Prior to µCP, the complex refractive index of a gold surface cleaned with the TL-1 protocol was measured using the imaging ellipsometer. After a performed µCP, the surface was again measured using the imaging ellipsometer. To generate a topological map of the surface, an optical model was used. For the organic film, the real and imaginary parts of the refractive index was set to 1.5 and 0, respectively. To minimize noise, a laser was used instead of the xenon lamp used for SPR measurements. The noise was minimized because of the superior intensity delivered by the laser. Due to limiting optical factors, only an area of ≈ 390 × 600 µm2 could be measured. The coherent light of the laser source and optical imperfections are responsible for the rippled surface. As can be seen in Figure 4.4a, the transferred pattern is clearly shown, with the hydrophobic SAM mean height 23 Å above the gold substrate. The average thickness of the "unprinted" areas are 5 Å due to contamination from the surrounding atmosphere and from the µCP.

65 % of the pixels are analyzed and accounted for in due to the uncertainty of measuring along the borders of the image and of the SAM. Note the contamination of the SAM in the lower right part of Figure 4.4a resulting in the maximum height value given in the table above. The measured thickness of the SAM is in good agreement with literature [26].

(45)

4.3 Measurements Using Imaging SPR 31

Table 4.1. Heights of the patterned surface measured using imaging ellipsometry.

Feature Pixels Mean (nm) Min (nm) Max (nm) StdDev (nm)

SAM 17931 2.3 1.3 4.2 0.202

Wells 16267 0.5 -0.6 1.6 0.204

Chip 52725

Height profiles of Figure 4.4 resemble the expected 300 µm diameter wells separated by 100 µm barriers.

Conclusion

The performance of the soft lithography was satisfactory and the results are in good agreement with the expected. Although a larger part of the sensor surface could not be analyzed by the instrument, it is likely that the measured distances are valid for the whole surface, given that the stamp used for µCP had a homogeneous pattern. Considering Figure 2.11 (the different light source enable a larger area to be depicted) the transferred pattern seems to be consistent.

4.2.2 Wetting Microscopy

To resolve the hydrophobic frame from the hydrophilic areas, a simple method called wetting microscopic is used. Figure 4.5 gives a good representation of the contrast achieved with the technique. Water vapor from the atmosphere is made to condense at the surface. Larger drops in the hydrophilic circles are formed by aggregated droplets. The smaller droplets are unable to unite due to the low surface energy of the frame.

4.3 Measurements Using Imaging SPR

In the imaging SPR experiments, the experiences from the test using the Biacore 3000 instrument were applied. Prior to using the synthesized chip, a condition-ing of the surface was made to ensure optimal function. This included several injections: 7 min purified water, 2 × 1.5 min 50 mM NaOH, 3.5 min 200 mM ethylenediaminetetraacetic acid (EDTA) and 10 minutes of ≈ 1 mg/mL BSA. All dissolved in HBS. Injections of purified water and NaOH were made to desorb the physisorbed molecules. EDTA, acting as a chelator was injected to remove possible metal traces from the surface. BSA was found to minimize NSB of vesicles.

4.3.1 Bovine Serum Albumin Minimizes NSB of Vesicles

The biochip was produced as described in Chapter 3. Four different ssDNA-sequences were exposed: A’, B’, C’ and D’, each immobilized at four replicate

(46)

32 Results & Discussion

(a) Topography of µCP surface with height-profiles indicated

Length (µm) 0 100 200 300 400 500 600 Height (nm ) 0.0 0.5 1.0 1.5 2.0 2.5

(b) Height-profile showing the approximate di-ameter (≈ 300 µm) of a "well" and the width of the frame separating two non-printed features (≈ 100 µm) Length (µm) 0 100 200 300 400 500 Height (nm ) 0.0 0.5 1.0 1.5 2.0 2.5

(c) Height-profile showing the maximum width separating two non-printed features

Figure 4.4. Topography of surface after µCP with hexadecanthiol measured using

imaging ellipsometry.

spots, occupying a total of 16 spots on the biochip (Figure 3.1). The surface was conditioned with BSA before three separate injections were made with different ssDNA-tagged vesicles (D&X, C&X and B&X). A regeneration of the biochip was made twice. The effect of not immobilizing BSA on the sensor surface was studied.

Microarray Technology for Kinetic Analysis of Vesicle Bound Receptor-Ligand Interactions

The Department of Physics, Chemistry and Biology

Master’s Thesis

Microarray Technology for Kinetic Analysis of

Vesicle Bound Receptor-Ligand Interactions

Björn Brian

Microarray Technology for Kinetic Analysis of

Vesicle Bound Receptor-Ligand Interactions

Björn Brian

Abstract

Sammanfattning

Acknowledgments

Abbreviations

Contents

Chapter 1

Introduction

1.1

Background

1.2

Aims of the Master’s Thesis

1.3

Key Ideas for the Solution of the Problem

A´

A´

B´

B´

A

Chapter 2

Theory

2.1

Protein Kinetics – A Way to Characterize

Protein-Ligand Interaction

2.1.1

General Protein Kinetics

2.1.2

Studied Proteins

2.2

Vesicles

i

ii

i

ii

2.2.1

Tagging the Vesicles with ssDNA

2.2.2

Nitrilotriacetic Acid (NTA)

2.3

Creating the Biochip

2.3.1

Self Assembled Monolayers (SAMs)

2.3.2

Microcontact Printing

2.3.3

Piezodispensing

2.3.4

Dextran

i

ii

iii

iv

i

ii

iii

iv

2.3.5

The Neutravidin-Biotin-ssDNA Complex

2.4

Null Ellipsometry

2.4.1

The Polarization of Light

2.4.2

Optical Component Setup

2.5

Surface Plasmon Resonance (SPR)

2.5.1

Imaging SPR

300 µm

Chapter 3

Experimental Details

3.1