Electrophoretic focusing in microchannels combined with mass spectrometry

Applications on amyloid beta peptides

SAARA MIKKONEN

Doctoral Thesis KTH Royal Institute of Technology School of Chemical Science and Engineering Department of Chemistry Division of Applied Physical Chemistry – Analytical Chemistry SE-100 44 Stockholm, Sweden

ii

TRITA-CHE Report 2016:36 ISSN 1654-1081 ISBN 978-91-7729-142-8

Copyright ©

Saara Mikkonen, 2016

Printed by Universitetsservice US-AB Akademisk avhandling som med tillstånd av Kungliga Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen fredagen den 4 november 2016,

kl. 10:00 i sal F3, Lindstedtsvägen 26, Stockholm. Avhandlingen presenteras på engelska.

iii

Electrophoretic focusing in microchannels combined with mass spectrometry – Applications on amyloid beta peptides, Saara Mikkonen, 2016, KTH Royal Institute of Technology

Abstract

Analysis of low-abundance components in small samples remains a challenge within bioanalytical chemistry, and new techniques for sample pretreatments followed by sensitive and informative detection are required. In this thesis, procedures for preconcentration and separation of proteins and peptides in open microchannels fabricated on silicon microchips are presented. Analyte electromigration was induced by applying a voltage along the channel length, and detection was performed either by matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) within the open channel, or by sampling a nL fraction containing the preconcentrated analytes from the channel for subsequent nano-electrospray ionization- (nESI-) or MALDI-MS. Utilizing solvent evaporation from the open system during sample supply, sample volumes exceeding the 25-75 nL channel volume could be analyzed. For preconcentration/separation of components in the discrete channel volume a lid of inert fluorocarbon liquid was used for evaporation control.

In Papers I and II, aqueous, carrier-free solutions of proteins and peptides were analyzed, and the method was successfully applied for fast and simple preconcentration of amyloid beta (Aβ) peptides, related to Alzheimer’s disease.

The impact of possible impurities in the analysis of carrier-free solutions was investigated in Paper III with the 1D simulation software GENTRANS, and a method for open-channel isoelectric focusing in a tailor-made pH gradient was developed. The latter approach was used in Paper IV for preconcentration and purification of Aβ peptides after immunoprecipitation from cerebrospinal fluid and blood plasma, followed by MALDI-MS from a micropillar chip.

Paper V includes simulations of an isotachophoretic strategy for selective enrichment of Aβ peptides. GENTRANS simulations were used to select the electrolyte composition, and 2D simulations in a geometry suitable for on-chip implementation were performed using COMSOL Multiphysics.

Keywords: Amyloid beta, computer simulation, electrophoresis, electrospray ionization, isoelectric focusing, isotachophoresis, matrix-assisted laser desorption ionization, mass spectrometry, microchannel, microchip, nano-electrospray ionization, preconcentration, separation

iv

Sammanfattning

Analys av ämnen i låga koncentrationer och små volymer är ett utmanande problem inom den bioanalytiska kemin, och det finns ett behov av nya tekniker för provupparbetning följt av känslig och informativ detektion. I denna avhandling presenteras metoder för uppkoncentrering och separation av proteiner och peptider i öppna mikrokanaler tillverkade på kiselchip. För att inducera elektromigration av analyterna applicerades spänning över kanallängden, och detektion skedde antingen med matris-assisterad laser desorptions-joniserings-masspektrometri (MALDI-MS) i den öppna kanalen, eller genom att ta upp en del av kanalvolymen innehållande de uppkoncentrerade analyterna för efterföljande nano-elektrosprayjoniserings- (nESI-) eller MALDI-MS. Genom att utnyttja avdunstningen av lösningsmedlet från det öppna systemet under provtillförseln kunde provvolymer större än kanalvolymen på 25-75 nL analyseras. För att koncentrera upp eller separera ämnen i kanalvolymen förhindrades avdunstningen med ett lock av inert fluorokolvätska.

I Artikel I och II analyserades vattenlösningar, eller buffertfria lösningar, av proteiner och peptider, och metoden tillämpades för snabb och enkel uppkoncentrering av amyloid beta-(Aβ-) peptider som är relaterade till Alzheimers sjukdom.

Påverkan av möjliga föroreningar vid analys av buffertfria lösningar undersöktes i Artikel III med hjälp av 1D-simuleringar med mjukvaran GENTRANS, och en metod för skräddarsydd isoelektrisk fokusering (IEF) i de öppna mikrokanalerna presenterades. I Artikel IV användes IEF-metoden för uppkoncentrering och upprening av Aβ-peptider efter immunoprecipitering från ryggmärgsvätska och plasma följt av MALDI-MS på ett mikropelarchip.

Artikel V innehåller simuleringar av en isotakoforetisk strategi för selektiv uppkoncentrering av Aβ-peptider. GENTRANS användes för att välja ut ett lämpligt elektrolytsystem, och en modell skapad i COMSOL Multiphysics möjliggjorde 2D-simuleringar i en geometri som kan implementeras på chip.

Sökord: Amyloid beta, datorsimulering, elektrofores, elektrosprayjonisering, isoelektrisk fokusering, isotakofores, matris-assisterad laser desorptions-jonisering, masspektrometri, mikrochip, mikrokanal, nano-elektrosprayjonisering, separation, uppkoncentrering

v

List of publications

This thesis is based on the following publications:

I Sample preconcentration in open microchannels combined with MALDI-MS Mikkonen, S., Khihon Rokhas, M., Jacksén, J., Emmer, Å., Electrophoresis, 2012, 33, 3343-3350

II Mass spectrometric analysis of nanoscale sample volumes extracted from open microchannels after sample preconcentration applied on amyloid beta peptides Mikkonen, S., Jacksén, J., Emmer, Å., Analytical and Bioanalytical Chemistry, 2014, 406, 3521-3524

III Computer simulations of sample preconcentration in carrier-free systems and isoelectric focusing in microchannels using simple ampholytes Mikkonen, S., Thormann, W., Emmer, Å., Electrophoresis, 2015, 36, 2386-2395

IV Microfluidic Isoelectric Focusing of Amyloid Beta Peptides Followed by Micropillar-Matrix-Assisted Laser Desorption Ionization-Mass Spectrometry Mikkonen, S., Jacksén, J., Thormann, W., Roeraade, J., Emmer, Å., Analytical Chemistry, 2016, DOI: 10.1021/acs.analchem.6b02324

V Selective enrichment of amyloid beta peptides using isotachophoresis Mikkonen, S., Ekström, H., Jacksén, J., Emmer, Å., Preliminary manuscript

Reprints are published with kind permission of the journals.

The contributions of the author to these papers are:

I All experiments and a major part of the writing

II All experiments and a major part of the writing

III All experiments and a major part of the computer simulations and writing

IV All computer simulations and a major part of the experiments and writing

V Part of the computer simulations and a major part of the writing

vi

Parts of this work have been presented orally by the author at the following conferences:

Sample preconcentration in microchannels Analysdagarna 2012, Uppsala, Sweden

MS analysis of nanoscale sample volumes extracted from open microchannels used for sample preconcentration 29th International Symposium on Microscale Bioseparations, MSB 2013, Charlottesville, Virginia, USA

Microfluidic isoelectric focusing of amyloid beta peptides followed by micropillar-MALDI-MS 32nd International Symposium on Microscale Separations and Bioanalysis, MSB 2016, Niagara-on-the-Lake, Ontario, Canada

Parts of this work have been presented as poster by the author at the following conferences:

Sample preconcentration in open microchannels combined with MALDI- and nano-ESI-MS Analysdagarna 2014, Djurönäset, Stockholm, Sweden and 62nd American Society for Mass Spectrometry (ASMS) Conference on Mass Spectrometry and Allied Topics, 2014, Baltimore, Maryland, USA

Microfluidic isoelectric focusing combined with MALDI- and nano-ESI-MS 63rd American Society for Mass Spectrometry (ASMS) Conference on Mass Spectrometry and Allied Topics, 2015, St Louis, Missouri, USA

In addition, the author of this thesis has contributed to:

CE analysis of single wood cells performing hydrolysis and preconcentration in open microchannels Khihon Rokhas, M., Mikkonen, S., Beyer, J., Jacksén, J., Emmer, Å., Electrophoresis, 2014, 35, 450–457

vii

Contents

1 Abbreviations and symbols ................................................................................................... 1

2 Introduction ........................................................................................................................... 3

3 Background ........................................................................................................................... 5

3.1 Electrophoresis .............................................................................................................. 5

3.1.1 Isoelectric focusing .................................................................................................. 7

3.1.2 Isotachophoresis ..................................................................................................... 9

3.1.3 Simulation of electrophoresis ................................................................................ 11

3.2 Mass spectrometry ....................................................................................................... 13

3.2.1 MALDI-MS ............................................................................................................. 13

3.2.2 ESI-MS .................................................................................................................. 15

3.2.3 Hyphenations with microchips ............................................................................... 16

3.3 Amyloid beta peptides .................................................................................................. 17

4 Instrumental setups and procedures .................................................................................. 19

4.1 Open microchannel chip and sample supply ............................................................... 19

4.2 Combinations with MS ................................................................................................. 22

4.2.1 Open channel as MALDI target ............................................................................. 22

4.2.2 Sampling from the open channel ........................................................................... 24

5 Preconcentration in carrier-free systems ............................................................................ 27

5.1 Influence of impurities .................................................................................................. 30

6 Isoelectric focusing in a mixture of simple ampholytes ....................................................... 33

6.1 IEF of amyloid beta peptides ........................................................................................ 35

7 Selective isotachophoresis ................................................................................................. 39

7.1 Electrolyte system and GENTRANS simulations ......................................................... 39

7.2 Geometry and COMSOL simulations ........................................................................... 41

viii

8 Conclusions and future perspectives .................................................................................. 43

9 Acknowledgements ............................................................................................................. 45

10 References ........................................................................................................................ 47

1

1 Abbreviations and symbols

6E10 anti-β-amyloid 1-16 antibody

Aβ amyloid beta

AD Alzheimer’s disease

APP amyloid precursor protein

c concentration

CA carrier ampholyte

CE capillary electrophoresis

CIEF capillary isoelectric focusing

CITP capillary isotachophoresis

cSer cycloserine

CytC cytochrome c

CZE capillary zone electrophoresis

DHB 2,5-dihydroxybenzoic acid

E electric field strength

ESI electrospray ionization

GE gel electrophoresis

GFb (Glu1)-Fibrinopeptide B

η viscosity

HF hydrofluoric acid

id inner diameter

IEF isoelectric focusing

IP immunoprecipitation

IPG immobilized pH gradient

ITP isotachophoresis

2

K ionization constant

KRF Kohlrausch regulating function

LE leading electrolyte

µ electrophoretic mobility

MALDI matrix-assisted laser desorption ionization

MBE moving boundary electrophoresis

ME microchip electrophoresis

MS mass spectrometry

m/z mass-to-charge ratio

nESI nano-electrospray ionization

NT neurotensin

pI isoelectric point

q charge

r radius

sAPPα soluble aminoterminal fragment alpha

S/N signal-to-noise ratio

TE terminating electrolyte

TFA trifluoroacetic acid

TOF time-of-flight

v velocity

V voltage

ZE zone electrophoresis

3

2 Introduction

Over the past few decades miniaturization has gained popularity within the fields of analytical and bioanalytical chemistry, and so called micro total analysis, or lab-on-a-chip, systems have found numerous applications in the laboratory, as well as in clinical and point-of-care settings [1]. Downscaling is accompanied by indisputable advantages such as shorter analysis time and reduced consumption of sample and reagents. In addition, miniaturized methods are suitable, or even necessary, when there are only limited amounts of sample available, e.g. for analysis of rare body fluids or single cells. Miniaturization, however, places high demands on the detection sensitivity, due to the smaller amount of sample utilized and, consequently, lower number of analytes reaching the detector. The complexity of most biological samples further requires that sample pretreatment procedures achieving both separation and preconcentration are employed.

Since electrophoretic procedures are able to handle small volumes, provide multiplexing possibilities and are highly flexible, they are well-suited for miniaturized settings. Consequently, electrophoresis is the most used separation technique in microfluidic devices [2] and numerous examples of miniaturized electrophoretic procedures within fields such as proteomics [3] and clinical analysis [4] have been presented. The electrophoretic modes isoelectric focusing (IEF) and isotachophoresis (ITP) are able to achieve both a high degree of preconcentration and separation, and have been used for sample pretreatments in microfluidic devices [5].

The phenomena involved in electrophoretic procedures influencing the achievable resolution, analysis speed and detection limits can, however, be complex, which complicates prediction of the experimental outcome. In microdevices in particular, the small scale further hampers measurement of properties such as local conductivity or pH. Dynamic computer simulations are able to provide insights into the evolution of electrophoretic separations, and are as such highly valuable in designing optimal electrophoretic systems [6]. In addition to the three general, 1D electrophoresis simulators currently in use in a number of labs [7], several multidimensional models have been developed. The latter are particularly useful to predict and optimize sample dynamics e.g. in 2D planar chip-based devices.

In microdevices electrochemical (amperometric or conductivity) [8, 9], optical (fluorescence or chemiluminescence) [10] or mass spectrometric detection is often used [11]. In comparison to the former two methods, mass spectrometry (MS) is able to achieve both sensitive detection and provide an unequivocal molecular identity for a wide variety of analytes. In bioanalysis the soft ion sources matrix-assisted laser desorption ionization (MALDI) and electrospray ionization (ESI) are the by far most used MS methods. Interfacing microdevices to MS [12-14], however, remains a challenge and approaches to transfer the often nanoliter-sized samples from the microfluidic platform to the MS ion source are required.

In this thesis a novel instrumental concept for electrophoretic preconcentration and separation in open microchannels fabricated on silicon microchips and its combination with MALDI- and nano-ESI-(nESI) MS detection is presented. Computer simulations with the electrophoresis simulator GENTRANS were performed to obtain information about the

4

focusing dynamics. The presented methods were successfully applied for analysis of the Alzheimer’s disease (AD) related amyloid beta (Aβ) peptides.

5

3 Background

In the chapter to follow a brief background to concepts included in this thesis will be presented. First, electrophoresis in general, and IEF and ITP in particular, will be described, followed by a section on simulation of electrophoresis. Thereafter, MALDI- and ESI-MS and hyphenations between these and microchips will be treated. In the last section Aβ peptides and the importance and challenges of their analysis is discussed.

3.1 Electrophoresis

Electrophoresis is the motion of charged particles in solution under the influence of an electric field [15] and is thus an essential concept in all procedures involving electromigration of solutes. In 1948 Tiselius was awarded the Nobel Prize in Chemistry for his pioneering research on electrophoresis, which included the introduction of a moving boundary electrophoresis (MBE) apparatus [16]. In MBE as well as in the electrophoretic modes zone electrophoresis (ZE) and ITP, sample components are separated based on electrophoretic mobility, μ. Upon application of an electric field the ion velocity, v, is influenced by μ and the electric field strength, E, according to:

(Eq. 1)

For spherical particles the relation between electrophoretic mobility and particle charge and size is illustrated by equation 2 [15]:

(Eq. 2)

where q: particle charge, r: particle radius and η: viscosity of the surrounding medium.

In ZE, the most common mode of electrophoresis, the separation is performed in a conductive buffer, into which a plug of sample is injected. The use of narrow-bore (typical inner diameters (id) of 25-150 µm) fused-silica capillaries for the separation, i.e. capillary zone electrophoresis (CZE), was first described by Mikkers, Everaerts and Verheggen in 1979 [17] and Jorgenson and Lukacs in 1981 [18]. Solutes are transported through the capillary by their electrophoretic migration and a bulk flow of the surrounding solution, i.e. the electroosmotic flow (EOF) [15].

The use of narrow capillaries results in efficient heat dissipation, and thereby fast analysis through the use of high electric fields. This particular feature is even more pronounced in further reduced dimensions: a fact that contributed to the development of microchip electrophoresis (ME) in the early 1990s [19, 20]. Since its introduction, ME has gained popularity, and today several modes of electrophoresis including ZE, micellar electrokinetic chromatography, electrochromatography, gel electrophoresis (GE), IEF and ITP are regularly performed both in capillaries and on-chip [2]. While speed and efficiency are often brought up as advantages of capillary electrophoresis (CE) and ME, lack of sensitivity, caused by the narrow diameter separation channels and small sample volumes, is generally considered to be their main weakness [21]. Means of overcoming this limitation include the use of preconcentration techniques and increasing the detection sensitivity.

6

Preconcentration methods containing electrophoretic components are sometimes referred to as stacking [22], or dynamic preconcentration methods [5, 23]. In these techniques electrokinetic equilibria are utilized to concentrate species at system boundaries, e.g. sample/buffer, by appropriate selection of the electrolyte composition. Thus, in comparison to non-electrophoretic preconcentration methods such as solid phase extraction and surface binding techniques, no physical barriers are required. A further division of dynamic preconcentration procedures into methods utilizing velocity changes and focusing based approaches can be made, with for example field-amplified sample stacking, ITP, dynamic pH junction and sweeping belonging to the former group of methods [5], while IEF, electric field gradient focusing [24] and temperature gradient focusing [25] are members of the electrofocusing family [26]. In the latter techniques species are focused to a self-sharpening stationary steady-state, which is independent of the initial sample distribution [27]. ITP, further described in Section 3.1.2, is characterized by a self-sharpening, although not stationary, steady-state and can thus also be considered as focusing.

7

3.1.1 Isoelectric focusing

In IEF an imposed pH gradient is used to separate amphoteric species based on differences in isoelectric point (pI), defined as the pH at which the compound has a net zero charge [28]. Proteins and large peptides generally have well-defined pIs and are thus particularly suitable to analyze using IEF. While the history of IEF dates back to the publication by Kolin in 1954 [29], the technique got its real breakthrough a few years later through the work of Svensson and Vesterberg [30-32] who generated stable pH gradients using carrier ampholytes (CA). Since then, gel-based IEF, e.g. utilizing immobilized pH gradients (IPG) [33], and capillary IEF (CIEF) [34] have been developed and extensively used. After the first demonstrations in 1999 [35, 36], several examples of IEF in microfluidic devices have also been reported [37].

In a typical, free-solution, IEF procedure the sample and the CAs are mixed in the separation column, and upon voltage application the CAs migrate to form a pH gradient increasing from anode to cathode (Fig. 1). A charged analyte migrates toward the oppositely charged electrode until it reaches a position where it has a net zero charge. Diffusion away from this point is accompanied by a change of charge. Consequently, analytes become focused at their pIs.

Figure 1. Schematic principle of IEF; A1-A6 represent CAs and the analyte molecules are denoted by circles, squares and stars.

For a compound to be suitable as CA it should have a high conductivity and buffering capacity, i.e. be isoelectric between two closely spaced pKa values (have a low ΔpKa) [31]. The amount of naturally occurring compounds exhibiting these properties is limited. Thus, Vesterberg developed synthetic procedures to generate mixtures of numerous CAs [32], allowing for the wide-spread use of IEF. CAs have been commercialized and are available under trademarks including Ampholine, Bio-Lyte, Pharmalyte and Servalyt [38]. These are typically composed of around 10-100 components and 30-500 isoforms per pH unit [38], resulting in essentially linear pH gradients enabling high-resolution separations.

The use of synthetic CAs is, however, accompanied by some disadvantages. For example, the available commercial CA systems are often chemically ill-defined [38], and if MS is used as detector, CAs may interfere [39], e.g. by suppressing analyte ionization and/or generating a high background signal. In proteomic analysis in particular, the CA MS signals may overlap with those of tryptic peptides, since they are within the same mass range [40]. Thus, other methods for forming pH gradients for IEF have been suggested. For example, it has been reported that the sample itself, consisting of e.g. protein or peptide mixtures, could form the

8

pH gradient for focusing. This approach was given the name autofocusing, as first reported by Sova in 1985 [41], and has later been used for example to separate a peptide mixture obtained after tryptic digestion of proteins prior to ESI-MS [42]. Several procedures utilizing amphoteric buffers or fewer and simpler carrier components have been described as well. For example, Yager’s group published a series of papers, in which pH gradients in amphoteric buffers were used for free-flow electrophoresis of proteins in microchannels through so called transverse IEF [43-45], and Huang et al. reported a capillary IEF (CIEF) separation coupled with whole-column imaging of two proteins in pure water [46]. IEF with amino acids, or similar small and chemically well-defined components, as carriers has been extensively studied in early work on electrophoretic simulations (see Section 3.1.3). A more recent example is the work by Zhu et al. in which amino acid carriers were used when coupling CIEF to ESI-MS [47, 48]. Regarding the material included in this thesis, preconcentration and separation of proteins and peptides in aqueous systems were investigated experimentally in Paper I and Paper II and theoretically in Paper III, and the results are discussed in Chapter 5. IEF with a few simple carriers was performed in Papers III and IV and is further described in Chapter 6.

IEF is a powerful technique with the ability of achieving both preconcentration and high-resolution separation. However, in addition to the possible disadvantages of commercial CA mixtures, a general and well-known weakness of IEF is precipitation at the pI, which may limit the loading capacity and/or cause analyte losses. A common strategy to increase solubility is using additives, e.g. zwitterionic surfactants or chaotropic agents such as urea [40]. It has also been suggested that acid wash is able to reverse precipitation [49]. Another electrophoretic method able to achieve a high degree of preconcentration – without risking pI precipitation – is ITP, which will be described next.

9

3.1.2 Isotachophoresis

In ITP components migrate with the same velocity in consecutive zones ordered by electrophoretic mobility in a discontinuous buffer system. The name ITP was established in the 1970s, although electrophoretic methods utilizing ITP principles had been reported earlier [50]. The progress in the field of analytical ITP since the 1990s has been continuously updated in a review series, most recently published in 2015 [51], and microfluidic ITP in particular has been reviewed by Smejkal et al. [52].

In a typical ITP experiment the sample is sandwiched between a leading electrolyte (LE) with a higher mobility than any of the analytes of interest and a low mobility terminating electrolyte (TE) (Fig. 2). In a single ITP experiment either anions or cations can be analyzed. Upon application of an electric field the complete system starts moving towards the oppositely charged electrode, with the highest mobility component migrating first, followed by the remaining species in the order of their net electrophoretic mobility [53]. If an ion diffuses into a neighboring zone, it will experience a different electric field, lowering or increasing its net velocity and thereby forcing it back into its own zone. Eventually, a self-sharpening steady-state configuration migrating at a constant velocity is attained.

Figure 2. Schematic principle of cationic ITP; LE: leading electrolyte, TE: terminating electrolyte, and the analyte molecules are denoted by circles, squares and stars.

Due to the self-sharpening effect, ITP is frequently used for preconcentration, sometimes combined with CZE in a single capillary setup using so called transient ITP or in a column-coupling arrangement in which separate capillaries are used for the ITP and CZE steps [54]. Strategies to further increase the concentration enhancement provided by ITP include applying a hydrodynamic counter flow and incorporating cross-sectional reductions of the separation channel [52]. The latter approach has been implemented in microdevices for example by Ivory’s group to perform cationic ITP of a cardiac biomarker [55, 56] and by Santiago’s group for nucleic acid analysis [57].

The selection of leading and terminating compounds and their mobilities determines which analytes that become stacked in the system. However, a further selectivity increase can be obtained by using electrolyte systems outside the typical ITP arrangement. For example, Schafer-Nielsen and Svendsen proposed systems in which the terminating ion was present in the LE or the leading ion was present in the TE [58], and Beckers used ITP superimposed on CZE to optimize the mobility window for isotachophoretic stacking [59, 60]. If the ITP separation is to be coupled with MS detection, the lack of available MS compatible electrolyte

10

systems, however, becomes a constraint. This was recognized by Malá et al. who recently reported electrolyte system strategies for anionic [61-63] and cationic [64] ITP in combination with ESI-MS. In a similar manner as earlier suggested [58], a small amount of the leading component was added to the TE, allowing mobility tuning and consequently increased flexibility and selectivity [62, 64]. A theoretical treatment of this approach is given in [64], using the ESI-MS compatible buffer system ammonium acetate/acetic acid. For a given concentration of acid and base in the leading zone, i.e. cA,L and cB,L, and the amount of base added to the TE, cB,T, the adjusted concentration of acid in the terminating zone, cA,T, can be calculated according to the Kohlrausch regulating function (KRF) for acidic systems:

, , , , . (Eq. 3)

For the ammonium acetate/acetic acid system A is acetic acid and B is ammonium. In this system ammonium is the leading component whereas hydrogen acts as terminating compound [65, 66]. Acetate is the common counter constituent.

This approach, which will be referred to as extended ITP, was used in Paper V, and the results are presented in Chapter 7.

11

3.1.3 Simulation of electrophoresis

The basis for the current theoretical understanding of phenomena involved in electromigration procedures can be traced back to 1897 when Kohlrausch derived the conservation law KRF together with a set of continuity equations governing electrophoretic migration [67, 68]. Their non-linear nature, however, hinders solving them analytically; instead, they are solved numerically using computer simulation. The development of dynamic computational models for electrophoresis began in the 1970s [69], after which the increased use of CE contributed to the development of 1D models such as the free-ware ZE simulator PeakMaster [70] and the general dynamic simulators GENTRANS [71], SIMUL5 [72] and SPRESSO [73]. In the latter three models, which are based on the principles of electroneutrality and conservation of mass and charge, component fluxes are computed by including contributions of electromigration, diffusion and convection (imposed flow and/or EOF) [7]. Since acid/base dissociation reactions are very fast in comparison to migration processes, component migration can typically be modeled using the average of all ionic and neutral forms of the component. By altering the initial component distribution and boundary conditions, several electrophoretic modes can be simulated [71].

Papers III, IV and V include simulations performed with GENTRANS. Since the underlying mathematical model was first introduced [71, 74, 75], the software has been further developed to include for example a specific module for the simulation of proteins (and/or large peptides) [76-78] and the possibility of including EOF [79], and interactions with an additive [80]. For low molecular mass compounds pKa and mobility values are required as input parameters. Electrophoretic mobilities of dissociating compounds vary as a function of the pH but are considered to be independent of the ionic strength. For proteins a pH vs net charge table (or titration curve) together with a diffusion coefficient are required as inputs, and the relation between mobility and ionic strength is accounted for via the Linderström-Lang approximation [77]. Due to the lack of availability of experimental titration curves, the major part of the simulations presented in this thesis were performed using theoretical titration curves, calculated based on the amino acid composition (as described in [78]) according to:

∑ ∑ (Eq. 4)

where ni and Ki are the number and ionization constant, respectively, of weakly basic groups (Arg, Lys, His and the terminal amino group) and, correspondingly nj and Kj denote acidic groups (Cys, Glu, Asp, Tyr and the terminal carboxy group). The simulation outputs include concentration, pH, conductivity and ionic strength distributions at specified time points, as well as the temporal behavior of current density and net buffer flow.

Computer simulation is a powerful tool in designing and optimizing electrophoretic procedures. Utilizing computer simulation, an increased understanding of the dynamics of electrophoresis has been obtained, including the evolution of the separation and focusing processes in IEF and ITP. IEF, which leads to a stationary steady-state, was simulated with steady-state models [81, 82] prior to the development of mathematical [83] and computational [71, 84-86] models describing transient processes. The dynamics of IEF are

12

composed of a fast separation phase and a much slower stabilizing phase [87]. Changes are initiated from the electrode boundaries and propagate inward. Analytes and carriers focus at essentially the same speed and, generally, reach their foci via the double-peak approach to equilibrium. In early work on IEF simulations simple and chemically well-defined carrier components, e.g. amino acids, were used [81, 84-86]. It has been found that the use of such components facilitates prediction of the focusing by simulation [88], and these types of systems have been used for example to predict protein fractionation in a suspended drop [89], as well as in the work presented in Paper III and Paper IV. Paper V includes GENTRANS simulations of ITP. In ITP, the sample focuses (non-isoelectrically) in the migrating LE-TE boundary. Provided a sufficient amount is sampled, a steady-state zone with a plateau concentration is formed. The length of the plateau is proportional to the amount of sample, while the concentration required to form a plateau is system-dependent, and influenced by the composition of the leading electrolyte [87]. Species present in amounts that are insufficient to form a plateau shaped zone are stacked in the boundary between the LE and TE zones and migrate in so called peak-mode. To separate them spacer components, with mobilities in between those of the analytes, can be introduced. Another approach to increase the selectivity of ITP is using electrolyte systems outside the typical ITP arrangement, as described in Section 3.1.2. A detailed description of the complex theoretical framework for ITP can be found for example in Chapter 4 in [50] or Chapter 6 in [87].

1D models are able to provide excellent insight into the dynamics of electrophoresis. However, they cannot be utilized for simulations in different geometries, e.g. in microfluidic channels with variable cross-section or across T-junctions, nor for simulation of sample injection strategies. For this purpose, multidimensional software can be used. COMSOL Multiphysics is a software package for modelling of phenomena that can be described with partial differential equations, i.e. including electrophoresis, and has been used for example by the Ivory group to simulate ITP in 1D [90, 91], 2D [92, 93] and 3D geometries [94]. COMSOL simulations of ITP are also included in Paper V.

13

3.2 Mass spectrometry

As mentioned in Section 3.1, another way to increase the detectability in miniaturized devices is using sensitive detection methods. MS, and in particular the soft ionization techniques MALDI and ESI, have since their introduction completely revolutionized the field of bioanalysis.

3.2.1 MALDI-MS

MALDI was introduced by Karas and Hillenkamp [95, 96] and Tanaka [97] in the 1980s. In a typical MALDI analysis a liquid sample is allowed to co-crystallize with a matrix compound on a stainless steel target, followed by irradiation by a high-energy laser under high vacuum (Fig. 3) [98]. The laser energy is transferred to the analytes via the matrix, resulting in formation of gas phase analyte ions. MALDI is often combined with a time-of-flight (TOF) mass analyzer. In the work presented in Papers I-IV MALDI-TOF-MS was used as detection method.

Figure 3. Schematic principle of the MALDI ion source and photograph of the analyte/matrix crystallization in a sample spot.

The achievable sensitivity in MALDI-MS is influenced by the sample preparation. The traditional dried-droplet sample preparation method typically produces so called sweet spots with significantly higher signal than the average signal in the rest of the spot [99]. This heterogeneity can result in longer analysis times and limited reproducibility. Consequently, quantitative analysis by MALDI-MS is challenging [100]. Means of increasing the spot homogeneity include pre-deposition of matrix (two-layer preparations) and using liquid matrices [99]. Electrospray sample preparation has also been suggested as an approach to improve the spot homogeneity and increase the signal intensity in MALDI-MS [101, 102]. Electrospray matrix application was used in Paper I and Paper IV (as described in Section 4.2.1).

An increased sensitivity and reduced analysis time can furthermore be obtained by confinement of the sample and matrix onto a small area [103]. A common approach to

14

achieve this is by creating small hydrophilic targets surrounded by a hydrophobic barrier with the aim of reducing lateral spreading of the sample and matrix [104]. These types of targets have been commercialized [105, 106]. In previous work at the division, micropillar MALDI targets fabricated on silicon microchips have been developed [107, 108]. The sample and matrix are confined into a very small area on top of the pillar and cannot spread to surrounding areas. Micropillar MALDI targets containing four blocks of circular 80 µm pillars with diameters of 50, 100, 200 and 500 µm [108] (Fig. 4) were used in the work presented in Paper IV. The micropillar target chip can easily be attached to a stainless steel MALDI target (using custom-made Plexiglas holders) and inserted into a conventional MALDI ion source for MS analysis.

Figure 4. Schematic illustration and photograph of micropillar MALDI target chip. Part of the figure is from [108] (unpublished work, used with permission from the authors).

In comparison to ESI, MALDI is more tolerant to possible contaminants and/or salts and surfactants, and due to the indirect ionization mechanism mainly singly protonated ions of the intact molecule are formed, which facilitates interpretation of mass spectra. In addition, the off-line format of MALDI-MS allows for parallel deposition of sample from multiple microchannels as well as repeated analysis of sample spots.

15

3.2.2 ESI-MS

ESI was first described by Dole et al. in the 1960s [109] and was thereafter further developed [110] and applied to ionize peptides and proteins [111] by Fenn and co-workers. In ESI sample is supplied in liquid phase to the atmospheric pressure ionization chamber through an emitter to which a high potential is applied (Fig. 5). When solution exits the emitter tip, an aerosol of multiply charged droplets is formed. The droplets are drawn towards a counter electrode positioned at an inlet leading to the mass analyzer. Evaporation of the solvent results in increased electrostatic repulsions within the droplets, eventually exceeding the surface tension [98]. As a result, the droplets collapse, and ultimately free gas phase ions are formed (Fig. 5).

Figure 5. Schematic principle of the ESI ion source operated in positive mode.

The quality of the spray is a critical parameter in ESI-MS, and is typically improved using volatile solvents with a low surface tension. With conventional emitters flow rates are generally in the µL/min range. nESI was first developed by Wilm and Mann in 1996 [112]. In nESI emitters with id:s in the low micrometer range are used, resulting in nL/min flow rates. Due to the concentration sensitive (as opposed to mass sensitive) behavior of ESI [113], the smaller absolute amount of ions reaching the detector does not cause loss of sensitivity. Instead, the formation of smaller and more charged droplets containing fewer analyte ions facilitates complete droplet desolvation and reduces ion suppression effects [114]. Also, miniaturized emitters can be positioned closer to the MS inlet resulting in a lower loss of ions not reaching the MS inlet. Disadvantages of the nESI-source are for example an increased risk of clogging, and that the existence of potential dead volumes becomes more critical. In the work presented in this thesis, nESI-MS with an ion trap mass analyzer was used in Papers II and III.

ESI-MS is more suitable than MALDI-MS for detection of low molecular mass components (below ca 500 Da), since they risk to be masked by the high background from the MALDI matrix. In addition, on-line coupling to continuous-flow separation techniques, such as liquid chromatography, CE or ME, is more straightforward.

16

3.2.3 Hyphenations with microchips

Interfacing microfluidic separations with MS is challenging and has been the subject of numerous reviews including those recently published by Wang et al. [12] and Feng et al. [13]. The coupling from microdevices to MALDI-MS is typically performed using off-line approaches, since the MALDI target is under vacuum in the mass spectrometer. In comparison, interfacing to ESI-MS is typically realized in an on-line fashion, which is facilitated by the compatible flow rates in microfluidic channels and nESI [12].

To transfer sample from a microdevice onto a MALDI target spotting, e.g. with a robotic spotter, can be used [115]. Another strategy is to use the microdevice or separation channel itself as MALDI target. In rapid open-access channel electrophoresis the separation was performed in open microchannels and by including the MALDI matrix in the buffer solution, MALDI-MS could be performed within the channel after solvent evaporation [116]. This approach, however, utilized a custom-built ion source. When IEF is interfaced with MS, a mobilization step of the focused, and stationary, sample typically needs to be implemented, and can be achieved for example by applying pressure or voltage [39]. Using the microdevice as MALDI target eliminates the need of mobilization and is thus a strategy particularly suitable for IEF-MS interfacing. Fujita et al. performed chip-based IEF in microchannels sealed with a removable resin tape during separation. After IEF, the sample was frozen and the tape was removed, followed by freeze-drying of the sample and application of MALDI matrix [117]. A similar approach was reported by the Sze group in which pseudo-closed separation channels manufactured on polymeric microchips with a removable cover chip were utilized for CIEF-MALDI-MS of proteins [118, 119]. Another elegant method to achieve fractionation and MALDI-MS detection after IEF is by using SlipChip-based technology. The SlipChip [120] is a microfluidic device consisting of two plates with a layer of lubricating oil between them, and by a slipping operation static IEF bands could be divided into nanoliter droplets in microwells, followed by opening of the device for MALDI-MS detection, as reported by Zhao et al. [121]. In the work presented in this thesis, the separation channel was used as MALDI target in Paper I and Paper IV.

The first reports on microfluidic ESI-interfaces were published in 1997 [122-124] and since then a variety of devices – with single, dual or multiple emitters – fabricated from different materials including glass, silicon and polymers have been developed [114, 115]. Interfacing strategies between microdevices and ESI-MS can be divided into two categories: techniques using on-chip ionization, i.e. from the microchannel outlet, and off-chip ionization using external emitters [125]. An advantage of on-chip ionization is the possibility of reaching very low, or zero, dead volumes. For a microchannel extended to the end of the device, the ESI voltage can be applied through the use of a sheath channel, also utilized to add a sheath-flow facilitating the formation of a stable spray. Sheath-flow based interfaces are also often used in CE-ESI-MS analysis [126, 127]. On-line approaches are typically fast and able to provide high throughput, but place high demands on the separation conditions, which need to be compatible with MS detection. In comparison, off-line approaches are more flexible. In Paper II and Paper III an off-line approach was used to couple the microfluidic preconcentration/separation to nESI-MS (as described in Chapter 4), whereas the initial results in Paper V (Chapter 7) are intended to be combined to nESI-MS in an on-line fashion.

17

3.3 Amyloid beta peptides

The concepts presented in Chapters 5-7 were used for preconcentration and purification of Aβ peptides (Papers II, IV and V), which are related to the pathophysiology of the neurodegenerative disorder AD. One of the pathogenic hallmarks of AD is the existence of so called senile plaques, formed by Aβ aggregation and fibrillation [128]. Aβ peptides are generated by β- and γ-secretase cleavage of the transmembrane protein amyloid precursor protein (APP) via the amyloidogenic pathway (Fig. 6) [129]. The heterogeneity of γ-secretase cleavage results in that several Aβ isoforms are formed [128]. In body fluids, such as CSF and blood plasma, Aβ 1-40 is the most common isoform [130, 131], whereas Aβ 1-42 is the most common form in amyloid plaques [128]. A selective reduction of Aβ 1-42 in CSF, sometimes measured as the ratio between Aβ 1-40 and 1-42, is an established AD biomarker [132].

Figure 6. APP processing. In the non-amyloidogenic pathway α-secretase cleavage generates the soluble aminoterminal fragment alpha (sAPPα), leaving behind the fragment C83, followed by

γ-secretase cleavage to release the p3 peptide and leave the APP-intracellular domain (AICD) in the membrane. The amyloidogenic pathway releases the soluble aminoterminal fragment beta (sAPPβ) by β-secretase cleavage, leaving behind C99, followed by γ-secretase cleavage generating Aβ peptides.

Analysis of Aβ peptides is complicated by their aggregation propensity and low abundance. The solubility of Aβ peptides is increased in highly acidic or alkaline conditions, and it has been shown that the monomeric form dominates at pH values below 4 and above 8 [133]. The concentration of Aβ 1-42 in CSF and blood plasma is in the pg/mL or pM range [129]. Thus, high-sensitive detection schemes must be employed. Immunoassays, e.g. enzyme-linked immunosorbent assay, are often used for detection of Aβ 1-40 and 1-42 [134]. It is, however, possible that cross-reactions with the employed antibodies occur, causing faulty results. In comparison, MS is able to provide an unambiguous molecular identity, enables detection of post translational modifications and facilitates simultaneous detection of several species. The latter aspect is important since Aβ peptides other than those for which immunoassays are

18

commercially available (i.e. Aβ 1-42, 1-40 and 1-38) have attracted clinical research interest [134]. Also, the pattern of multiple Aβ isoforms in plasma has been suggested to be of use as a measure of response to treatment [135] and there are ongoing investigations on finding novel Aβ species with potential to serve as plasma-based AD biomarkers [136, 137]. Prior to MS detection it is, however, necessary to include a purification and/or preconcentration step. Immunoprecipitation (IP), for example using magnetic beads coated with a capture antibody, is commonly used for this purpose [138, 139].

Examples of Aβ peptide analysis utilizing electrophoretic procedures include CE in combination with UV [140, 141] and ESI-MS [142] detection, and ZE [143] and GE [144] in microchips followed by fluorescence detection. IEF procedures used for Aβ analysis include gel-based IPG of Aβ peptides in human plasma [145] and CIEF-immunoassay of amino-terminal Aβ variants [146]. Recently capillary ITP (CITP)-UV of the Aβ isoform 1-40 was reported, resulting in that concentrations lower than previously reached (using UV detection) could be detected [147, 148].

19

4 Instrumental setups and procedures

In this chapter the instrumental strategies developed and utilized in Papers I-IV will be presented (the used chemicals and materials are described in detail in the respective papers). A major part of the experimental work has been performed in an open microchannel setup. In comparison to closed, or sealed, separation channels, fabricated e.g. by bonding of a lid [149], used in many chip-based microfluidic applications, open channels provide an increased flexibility regarding sample supply and combination with MS detection.

4.1 Open microchannel chip and sample supply

The open microchannels were fabricated on silicon microchips, using a single photolithography, deep-reactive ion etching and oxidation [150]. In other work at the division, these open channel chips have been used in combination with CE and/or MALDI [150, 151]. The channels have a rectangular shape and the following dimensions: length 1 or 3 cm, width 50-150 µm and depth 50 µm. Herein, only 1 cm long channels, with a volume of 25-75 nL, have been used. Some channels have wells at the ends, for positioning of devices such as electrodes and capillaries. Photographs of a microchip and the various well designs are shown in Fig. 7.

Figure 7. Photographs of open microchannel chip and the various well designs. (Modified from Paper I)

A schematic overview of the developed instrumental setups is given in Fig. 8. For preconcentration and separation in the open microchannels different approaches were used. In Paper I and Paper II preconcentration was performed during sample supply. As depicted schematically in Fig. 8A, sample was continuously supplied from both ends through fused-silica capillaries connected to a syringe pump, while a voltage was applied along the channel length. For voltage application, platinum wire electrodes attached to x,y,z-micropositioning devices were used. Utilizing this methodology, the sample volume possible to apply is not restricted to the channel volume. Instead, the evaporation of the solvent (water) during sample supply is exploited as means of achieving a further concentration enhancement, in addition to preconcentration based on analyte electomigration. As a result, an increased detectability can be obtained.

It can, however, be advantageous to be able to work with smaller sample volumes. For analysis of sample volumes in the range of the channel volume, i.e. 25-75 nL, the rapid evaporation from the channel must instead be limited. To achieve this, the sample, supplied to the open channel using a pipette or a fused-silica capillary, can be covered by a transparent

20

and chemically inert fluorocarbon liquid, creating a semi-open system. This method for evaporation control of small volumes of liquid samples has been used in previous work at the division [152-154]. Preconcentration in a discrete channel volume, as illustrated in Fig. 8B, was used in Paper III.

In Paper IV a combination of the two previously described strategies was used. First, a µL range sample was loaded into the open channel through a fused-silica capillary by allowing the water in the sample to evaporate during the sample supply. Thereafter, separation buffer was added, followed by addition of the fluorocarbon lid and application of voltage to further preconcentrate the analytes present in the 75 nL channel volume.

In addition to analyte migration, voltage application will result in ionic decomposition of water at the electrodes, leading to formation of oxygen gas and protons at the anode and hydrogen gas and hydroxide ions at the cathode. An advantage of using an open system is that the generated gas bubbles can go into the atmosphere instead of into the channel, where they could disrupt the current. Protons and hydroxide ions can, however, migrate into the channel, in which case pH changes can occur in poorly buffered systems.

21

Figure 8. Schematic overview of instrumental setups of preconcentration approaches and coupling to MS. (Modified from Paper III)

22

4.2 Combinations with MS

The use of a closed microchannel format can be advantageous, or even necessary, for continuous-flow procedures or for pumping through densely packed stationary phases. However, for focusing based approaches leading to a stationary end configuration, the use of an open, or semi-open, system is beneficial since it allows several combinations with MS detection without the need of mobilization of the complete sample. The strategies used in this work are described below.

4.2.1 Open channel as MALDI target

The open silicon microchannel was used directly as MALDI target in Papers I and IV. In Paper IV the setup depicted in Fig. 8B was used. Thus, as a first step, the liquid fluorocarbon lid was removed using suction with a pipette, after which the procedure followed in the same way as in Paper I. In Paper I preconcentration during continuous sample supply (Fig. 8A) was performed. After having supplied the desired sample volume, the preconcentrated/separated sample was allowed to dry in the channel while maintaining the voltage across the channel. Thereafter MALDI matrix was added to the channel.

2,5-dihydroxybenzoic acid (DHB) is commonly used as matrix for protein and peptide analysis [99] and was used as matrix compound throughout all MALDI-MS results presented in this thesis. In initial experiments, matrix was applied with an airbrush. This approach, however, required using relatively high matrix concentrations and large solution volumes to reach a sufficient degree of crystallization in the narrow channels located 50 µm below the chip surface. Thus, in further work an electrospray matrix application procedure was developed, resulting in that the volume of matrix solution could be decreased from 30 to 0.2 µL per channel and the DHB concentration could be lowered. The instrumental setup for electrospray matrix application is depicted schematically in Fig. 9. The chip was placed on an electrically grounded metal plate, and the cone-shaped tip of a fused-silica capillary, to which a positive potential was applied, was moved over the channel generating positively charged droplets drawn towards the channel. The spraying was monitored using a USB camera.

Figure 9. Schematic instrumental setup for, and photograph of, electrospray matrix application and photograph of the open microchannel chip attached to the metal plate to be inserted in the MALDI ion

source.

23

Thereafter, double-sided adhesive tape or custom made Plexiglas holders were used to attach the chip to a stainless steel target, which was inserted into the MALDI ion source. In some experiments (Paper IV) the chip was instead attached to a TLC-MALDI adapter (Bruker Daltonik). In the MALDI ion source laser pulses were shot along the channel length and the signal-to-noise ratio (S/N) of the analyte relative to an internal standard component included in the matrix solution was measured. In Fig. 10 photographs of the open microchannel prior to and after electrospray matrix application are shown, together with a photograph of the channel after exposure to laser pulses.

Figure 10. Photographs of microchannel chip before (A) and after (B) electrospray matrix application and of the open microchannel after exposure to laser pulses (C).

By performing MALDI-MS analysis within the open channel the analyte distribution in the channel can be detected in a convenient and straight-forward way. In the next section, an alternative approach – direct sampling of the preconcentrated analytes from the open system – is described.

24

4.2.2 Sampling from the open channel

The access to the sample from the open, or semi-open, instrumental setup enables direct sampling of the analytes from the microchannel while the preconcentration/separation voltage is kept on. As described below, this provides a simple approach of combining the open channel preconcentration with nESI-MS. Furthermore, being able to retain the concentrated target analytes in a nanoliter range volume can bring an increased sensitivity in MALDI-MS if this volume is deposited and confined onto a miniaturized MALDI target.

The setup for sampling is illustrated schematically in Fig. 11. A nESI emitter or a short fused-silica capillary was inserted into a rubber septum of a small glass air vessel attached to an x,y,z-positioning stage. In the experiments included in Paper II the chip was placed on a manually controlled x,y-stage, whereas in Paper III and Paper IV the complete instrumental setup was situated on a programmable x,y,z-translational stage equipped with two CCD-cameras [107].

Figure 11. Schematic instrumental setup for sampling from the open microchannel.

nESI-MS detection of the channel contents was performed in Paper II and Paper III by sampling with a nESI emitter (Fig. 12). The emitter tip was immersed into the solution in the channel at the position of the analytes, and vacuum suction was applied to sample a nanoliter fraction from the channel, after which the emitter was transferred to an off-line nanospray interface of an ESI-ion trap-MS system. In the MS instrument, the sampled species generated a short pulse of spray. Filling the nESI-emitter by suction from the outlet, instead of manual filling from the back of the emitter using a micropipette, enables nESI-MS analysis of smaller volumes, as described in earlier work at the division [153].

25

Figure 12. Photograph of nESI emitter positioned in open channel together with electrodes and fused-silica capillaries for supply of sample solution.

To instead perform MALDI-MS detection of the analytes, sampling was performed with a 2-3 cm long fused-silica capillary (id/outer diameter 10-20/150). The capillary tip was chemically etched to a cone-shape using hydrofluoric acid (HF). By immersing the tip into the solution in the microchannel, the capillary was filled by capillary forces. The capillary dimensions were selected based on the desired sampling volume.

Deposition of the sampled volume was performed by applying a pulse of compressed air (Fig. 11). To minimize spreading, and thereby dilution, of the concentrated analytes, miniaturized MALDI targets were used, and the MALDI matrix was applied through a fused-silica capillary. In Papers II and III sampling was performed onto commercial Bruker AnchorChipTM MALDI targets (Fig. 13A), whereas micropillar MALDI targets (see Chapter 3, Section 3.2.1) were used in Paper IV (Fig. 13B). As shown in Fig. 13C, sampling and deposition with several capillaries simultaneously is also possible. Using this approach a complete fractionation of the channel contents could be achieved in one step.

Figure 13. Photographs of sample deposition with fused-silica capillaries onto a Bruker AnchorChipTM target (A), a 100 µm micropillar (B) and simultaneous deposition onto three 200 µm

micropillars (C).

Sampling from the open microchannel can be performed both from the setup open to the surroundings (Fig. 8A, Paper II) and through the fluorocarbon liquid lid (Fig. 8B, Papers III

26

and IV). Moreover, placing also the MALDI target under the fluorocarbon lid further enables chemical reactions to be performed on-plate in nanodroplets, as described in Paper II.

27

5 Preconcentration in carrier-free systems

In Paper I the main focus was development of the instrumental setup described in the previous chapter. Thus, experiments were performed using simple, carrier-free systems consisting of the analytes dissolved in either pure water or acidic solution. In these systems, the preconcentration is solely based on the analyte charge, i.e. a positively charged compound will migrate towards the cathode and a negatively charged to the anode. Cytochrome C (CytC), with a pI around 10 (positively charged in an aqueous solution), was used as model protein, since its coloration enables visual observation of a concentration increase. Photographs of CytC preconcentration in an open glass microchannel (fabricated by chemical etching using HF) under a lid of fluorocarbon liquid are shown in Fig. 14.

Figure 14. Photographs of preconcentration (12 V/cm applied for 5, 60 and 120 min) of aqueous CytC solution. The anode is on the left.

In the open silicon microchannel setup, initial experiments utilizing optical detection were performed with fluorescent analytes [155], whereas the distribution of CytC after preconcentration during continuous supply of sample (Fig. 8A) was detected by performing MALDI in the channel (Fig. 8C). A zone of concentrated protein could be detected at a channel position close to the cathode for initial CytC concentrations down to 1 nM (Fig. 15).

Figure 15. MALDI spectrum obtained 1 mm from the cathode in the open silicon channel after preconcentration (15 µL/h with 36.4 V applied for 5 min) of 1 nM CytC (1.25 fmol supplied to the

channel). 1 µM ubiquitin (Uq) was added to the matrix solution as internal standard protein. (Modified from Paper I)

A procedure was developed also to detect the preconcentrated species using nESI-MS (as described in Section 4.2.2). A 500 nM aqueous solution of the model peptide neurotensin (NT, pI 9.7, positively charged in an aqueous solution) was continuously supplied to the open channel for 1 min during application of 36.4 V. Sampling was performed with a nESI emitter (Fig. 8D) from a channel position next to the cathode, and nESI-MS analysis revealed that

28

the preconcentration procedure resulted in a S/N enhancement by a factor of 15 compared to analysis of the same solution without preconcentration (Fig. 16).

Figure 16. a) nESI mass spectrum obtained after preconcentration of a 500 nM neurotensin (NT) (20 µL/h, 36.4 V, 1 min) and sampling from a position next to the cathode in the channel. b) nESI

mass spectrum of the same solution without preconcentration (data collected for the same amount of spray time). (Modified from Paper II)

As mentioned in Chapter 4, the semi-open instrumental setup allows for combining the preconcentration with on-plate chemical reactions followed by MALDI-MS detection, such as trypsin digestion of a preconcentrated protein, as described in Paper II. CytC was preconcentrated in the open channel, and using a short fused-silica capillary, 7.5 nL containing the preconcentrated protein were sampled and deposited onto a MALDI target covered by fluorocarbon liquid. The trypsin digestion was performed in a manner similar to what was reported in [153]. To the sampled droplet digestion buffer was added and after 2.5 h the digestion was terminated by addition of MALDI matrix and removal of the fluorocarbon liquid. In Fig. 17 the resulting MALDI-MS spectrum for an initial CytC concentration of 5 µM is shown.

Figure 17. MALDI-MS spectrum obtained after preconcentration of 5 µM CytC (15 µL/h, 36.4 V, 5 min) in an open microchannel, sampling of 7.5 nL containing the concentrated protein and on-plate trypsin digestion (2.5 h) in a nanodroplet. The assigned m/z values include CytC fragments with zero

and one missed cleavage.

In addition to model peptides, preconcentration of Aβ peptides was performed in carrier-free systems. As shown in Paper II, preconcentration followed by sampling onto an AnchorChipTM

29

MALDI target (Fig. 8D and Fig. 13A) enabled MALDI-MS detection of 1 nM Aβ 1-38 and 1-40 and their oxidized forms. The method was used also on Aβ immunoprecipitated from a cell culture (containing 41.8 ng/mL Aβ 1-40 before IP) and eluted in 0.1% formic acid (resulting in a net positive charge for the Aβ peptides). The preconcentration resulted in significantly increased MALDI-MS signals (Fig. 18a) in comparison to analysis of the same the volume without preconcentration (Fig. 18b) – clearly demonstrating the method’s applicability as a fast and simple method for controlled preconcentration of small volumes.

Figure 18. a) MALDI mass spectrum obtained after preconcentration (36.4 V applied for 3 min) of Aβ after IP from a cell culture and sampling of 8.8 nL from a position next to the cathode in the channel.

b) MALDI mass spectrum of the same sample solution and volume (8.8 nL) without preconcentration. (Modified from Paper II)

30

5.1 Influence of impurities

It is important to consider that in carrier-free systems consisting of aqueous solutions of the analytes possible contaminants, e.g. from dissolved atmospheric carbon dioxide or remains from salts used in the preparation of proteins or peptides, may have a considerable influence on the system behavior. This was investigated with computer simulations in Paper III, by comparing the experimental results reported in Paper I to GENTRANS simulations performed with carbonic acid and sodium chloride (model impurities) as additional input parameters. The potential presence of carbonic acid is attributed dissolved atmospheric carbon dioxide; the theoretical pH of pure water exposed to the atmosphere is 5.6 [156], corresponding to approx. 10 µM carbonic acid. The potential remains from substances used in the protein preparation and purification procedures were simulated using various micromolar range concentrations of NaCl.

In Fig. 19A distribution plots, detected by MALDI-MS in the open channel (Fig. 8C), from three experiments of a 5 min preconcentration of an aqueous solution of 1 µM CytC are shown. The corresponding simulation results for this system are displayed in Fig. 19B after 5, 10 and 30 min of voltage application, respectively. Fig. 19C-F show the simulated CytC distribution profiles in the presence of 10 µM carbonic acid and/or 0.5 or 5 µM NaCl.

Figure 19. Preconcentration of CytC. A) Experimental protein distribution for three different 50 µm wide channels detected with MALDI-MS analysis after preconcentration of 1 µM CytC with 36.4 V

applied for 5 min. B)-F) Simulation data of 1 µM CytC concentrated at a constant 36.4 V for 5, 10 and 30 min in B) pure water, C) 10 µM H2CO3, D) 0.5 µM NaCl, E) 0.5 µM NaCl and 10 µM H2CO3, and

F) 5 µM NaCl and 10 µM H2CO3. The anode is on the left. (Paper III)

31

A qualitative agreement between the simulated distribution (Fig. 19B) and one of the experimental profiles can be observed (Fig. 19A); however, the computer predicted time is significantly longer than the experimentally observed time. Since the presence of carbonic acid was predicted to accelerate the protein migration (Fig. 19C and 19E), it is possible that carbonic acid was present in the experimental system. Fig. 19D and 19F show that while a low amount of NaCl does not significantly alter the protein distribution (Fig. 19D), at higher concentration, 5 µM (Fig. 19F), the accumulation of sodium at the cathode affects the protein distribution. This could be a reason for the bump observed in one of the experimental profiles (Fig. 19A).

In Paper I the open channel preconcentration procedure was also used to simultaneously preconcentrate differently charged compounds using a trypsin digest of CytC. Computer simulations of this system were performed in Paper III for five of the most abundant peptides observed in MALDI-MS analysis (i.e. the peptides with masses 633, 778, 964, 1168 and 1495) in the presence of 10 µM carbonic acid and 0.5 µM NaCl. In agreement with the experimental results obtained by MALDI-MS analysis in the open channel (Fig. 20A) simulation predicted that the most basic peptides (633, 778 and 1168), would accumulate mainly in the cathodic part of the channel whereas the acidic species (964 and 1495), would migrate to the anode (Fig. 20B).

Figure 20. Preconcentration of CytC digest with 36.4 V/cm applied. A) MALDI spectra obtained at 1, 5 and 9 mm from the cathode in the open channel after 5 min preconcentration of 5 µM CytC trypsin digest. (Paper I) B) Simulated concentration profiles for five peptides (initial concentration 1 µM) in the presence of 10 µM H2CO3 and 0.5 µM NaCl. The anode is on the left. (Modified from Paper III)

32

In conclusion, the results presented in Chapter 5 show that preconcentration in carrier-free systems can be successfully applied both for analysis of real biological samples (e.g. Aβ peptides, Fig. 18) and to achieve a division of acidic and basic species of a protein digest (Fig. 20). Nevertheless, in these types of systems, peptides and proteins will always be distributed across part of the microchannel and not be tightly focused, and the potential presence of carbonic acid or other impurities may affect the sample dynamics. Thus, to improve the robustness and efficiency of the procedure, and thereby increase the number of its potential applications, simple ampholytes can be used as carriers for IEF. This will be discussed in the next chapter.

33

6 Isoelectric focusing in a mixture of simple ampholytes

In a mixture of amphoteric carriers tightly focused analyte zones can be formed by IEF. The use of a small number of carriers results in a steep pH-step, and a high concentration efficiency compared to linear gradients generated using commercial CA mixtures. As mentioned in Section 3.1.3, the use of small and chemically well-defined amphoteric carriers further facilitates accurate prediction of the location of the focused species by computer simulation. Thus, prior to carrying out IEF in the open microchannel setup, GENTRANS simulations were performed.

An initial uniform component distribution of three amino acid carriers, Glu, His and Arg, and two model peptides, NT and (Glu1)-Fibrinopeptide B (GFb), was subjected to IEF. In Fig. 21 the computer predicted focusing dynamics for this system are shown in terms of concentration, pH and conductivity profiles after 1.0, 1.7 and 4.0 min voltage application (35 V/cm), respectively. The carriers migrate to take up the anodic (Glu), cathodic (Arg) and center (His) part of the channel (Fig. 21A) generating a step-wise pH gradient (Fig. 21C). The zone boundaries are demarcated by changes in conductivity (Fig. 21D). As seen in Fig. 21B, the peptides, with pIs of around 4 (GFb) and 10 (NT), position themselves in the Glu-His and His-Arg boundaries, respectively.

Figure 21. Simulated distributions after 1.0 (top graphs), 1.7 (center graphs) and 4.0 min (bottom graphs) application of 35 V to a configuration with three amino acids and two peptides. A) Amino acid concentration profiles (initial concentrations 10 mM each), B) concentration profiles of NT and GFb (initial concentration 1 µM each), C) pH profiles (initial pH 7.4), and D) conductivity distributions

(initial conductivity 52.6 mS/m). The anode is on the left. (Paper III)

The corresponding laboratory experiments were performed in the open microchannel setup (depicted in Fig. 8B and 8D). Due to their low mass, nESI-MS was used to detect the amino acids. After focusing, sampling of approx. 15 nL was performed with nESI emitters from different parts of the channel. To dilute the carriers (present in mM concentration) and improve the spray conditions, 2 µL of 7/3 acetonitrile/0.1% formic acid was added to the emitter using a micropipette. As shown in Fig. 22A, a clear separation of the carriers was

34

obtained. The peptides were instead detected using MALDI-MS. Simulation predicted that GFb would focus at a position of approx. 3 mm from the anode, and NT at ca 7 mm from the anode. Thus, a short fused-silica capillary was used to sample 9.4 nL from these positions followed by deposition onto AnchorChipTM MALDI targets (Fig. 13A). As shown in Fig. 22B, an agreement with the simulation results, and a complete separation of the model peptides, was achieved. A somewhat larger volume than predicted by simulation was sampled to allow for a small potential variation of the zone location.

Figure 22. MS spectra after 15 min IEF at 35 V in 150 µm wide microchannels with a liquid lid. A) nESI-MS spectra after IEF of Glu, His and Arg (10 mM each). Sampling was performed 1 mm from

the anode (top graph), from the channel center (center graph) and 1 mm from the cathode (bottom graph). B) MALDI-MS spectra after IEF of 1 µM NT (m/z 1673) and 1 µM GFb (m/z 1571) in a mixture

of Glu, His and Arg (10 mM each). A 9.4 nL sample was collected 3 mm from the anode (top graph) and 7 mm from the anode (bottom graph). (Paper III)

The presented results suggest that by proper choice of carrier components, pH gradients tailored for specific analytes can be designed. This is discussed in the next section in which a system tailored for the preconcentration and purification of Aβ peptides was created.

35

6.1 IEF of amyloid beta peptides

In previous work at the division it was found that the eluate obtained after IP of Aβ from CSF or plasma contained unknown contaminants [157]. Consequently, no more than approx. 5 nL of the eluate could be deposited onto a micropillar MALDI target (see Section 3.2.1) without risking loss of MS signal due to the formation of a viscous consistency (Fig. 23) hindering co-crystallization with the MALDI matrix.

Figure 23. Photographs illustrating the consistency obtained after deposition of approx. 10 nL of the eluate obtained after IP of Aβ onto micropillars.

It is likely that the contaminants were remains of either the capture antibody used for IP, anti-β-amyloid 1-16 antibody (6E10), or the soluble aminoterminal fragment alpha (sAPPα) (produced in the non-amyloidogenic pathway of APP processing, see Fig. 6, Section 3.3), which also contains the epitope recognized by 6E10. Thus, the carrier components for IEF – Glu, aspartyl-histidine (Asp-His), cycloserine (cSer) and Arg – were chosen to achieve an efficient removal of such components while focusing the Aβ peptides in a pH-step around their pI. Most C-terminal Aβ isoforms starting with Asp(1), including Aβ 1-38, 1-40 and 1-42, have a pI of ca 5.3 (Paper IV), and are thereby predicted to focus in the boundary between Asp-His (pH 4.9) and cSer (pH 5.9). To adjust the pH (see Section 3.3), trifluoroacetic acid (TFA) was also included in the IEF buffer. The focusing dynamics were again studied using GENTRANS simulation. Fig. 24 shows the computer predicted concentration profiles for the carriers and TFA (Fig. 24A) and Aβ peptides (Aβ 1-38 and 1-40 were used as model species in the simulations), sAPPα and 6E10 (Fig. 24B) together with the resulting pH gradient (Fig. 24C) after 10 min of voltage application.

Figure 24. IEF of 1 µM Aβ 1-38 and 1-40 and 0.1 µM sAPPα and 6E10 in a mixture composed of Glu, Asp-His, cSer, and Arg (5 mM each) and 0.1% TFA. The plots show the computer predicted

distributions of ampholytes and TFA (A), sample components (B) and pH (C) after 10 min application of 74.0 V. The initial pH was 3.21. The anode is on the left. (Modified from Paper IV)

36

Next, IEF was performed in open microchannels. Prior to sampling onto micropillar MALDI targets the distribution of Aβ 1-38 and 1-40 after IEF was detected by MALDI-MS within the open channel (Fig. 25). Compared to the simulation results, a slight position shift towards the cathode was observed, probably due to the presence of salts that had been added in the standard preparation procedures.

Figure 25. MALDI-MS spectra obtained along the channel length after IEF of 1 µM Aβ 1-38 and 1-40 in Glu, Asp-His, cSer, and Arg (5 mM each). The spectrum furthest away is from the anode and the

closest spectrum from the cathode.

Thereafter CSF (from patients diagnosed with AD) and plasma samples were analyzed. After IP, the elution was performed in pure water, and for these samples a better agreement with the simulation results in Fig. 24 was obtained, with the Aβ peptides focused at a position 7 mm from the cathode. As mentioned in Chapter 4, using open channels enables analyzing sample volumes larger than the channel volume as an approach to increase the detectability. Thus, during controlled evaporation of water, a µL range volume IP eluate was supplied to a microchannel, followed by addition of IEF buffer and the fluorocarbon lid and application of voltage. After focusing, a nL fraction containing the Aβ peptides was sampled and deposited onto a micropillar. The additional confinement of the sample obtained using micropillar MALDI targets, in comparison to conventional miniaturized targets, was illustrated in Fig. 13. To limit the potential loss of sample due to pI precipitation, application of MALDI matrix was followed by recrystallization using 1% TFA (see Section 3.1.1). In Fig. 26 a comparison between the MALDI-MS signal for Aβ 1-40 after direct deposition of 2 nL IP eluate (top) and after loading of 1 µL (center) and 2 µL (bottom) followed by IEF and sampling of 2 nL is shown.

37

Figure 26. Micropillar-MALDI-MS signal of Aβ 1-40 after IP from CSF. Top: 2 nL IP eluate deposited (no IEF), center: 1 µL supplied to the channel, sampling of 2 nL after IEF, bottom: 2 µL supplied to the

channel, sampling of 2 nL after IEF.

Complete MALDI-MS spectra after IP from CSF (left) and plasma (right) utilizing this approach are shown in Fig. 27. The upper spectra (A and C) display the best achievable signals for these samples without the IEF purification step, whereas the lower spectra (B and D) were obtained after IEF.

Figure 27. Micropillar-MALDI-MS spectra of 5 nL deposited on 200 µm pillars of Aβ after IP from CSF (A and B) and plasma (C and D). A) IP eluate (in water), no IEF. B) 1.5 µL supplied to

microchannel, IEF in 5 mM Glu-(Asp-His)-cSer-Arg with 0.1% TFA (10 min, 74 V) and sampling 7 mm from anode. C) IP eluate (in water), no IEF. D) 1 µL supplied to channel, IEF in 5 mM Glu-(Asp-His)-

cSer-Arg with 0.1% TFA (10 min, 74 V) and sampling 7 mm from anode. (Modification of figures in Paper IV)

38

The presented approach of IP-IEF-micropillar-MALDI-MS is able to detect Aβ peptides in human CSF and plasma, in which they are present in pg/mL concentrations. Thus, this technique could have the potential to contribute to AD related research, for example aimed at developing an increased understanding of the disease mechanisms, evaluating possible treatments or enabling early AD diagnosis.

39

7 Selective isotachophoresis

As mentioned in Section 3.1.2, extended ITP allows for mobility tuning and thereby tailoring of the system to target specific components. It could thus be an alternative strategy to IEF, able to achieve both preconcentration and purification. Paper V includes simulations of extended ITP in the ESI-MS compatible buffer system ammonium acetate/acetic acid. The presented methodology is intended to be applied for ITP-ESI-MS analysis of Aβ peptides in a miniaturized, chip-based format.

7.1 Electrolyte system and GENTRANS simulations

To select a suitable electrolyte composition, Eq. 3 (Section 3.1.2) was used, together with GENTRANS computer simulations. For two extended ITP electrolyte systems, the theoretically predicted (Fig. 28A and 28C) and experimentally observed (initial results using CITP-ESI-MS with a 100 cm long capillary [158], Fig. 28B and 28D) migration order of five tryptic peptides of CytC was compared. As shown, a satisfactory agreement was obtained.

Figure 28. Simulated (A and C) and experimental (B and D) profiles after extended ITP of five peptides from a CytC digest. A and B: LE: 10 mM NH4+, 20 mM HAc, TE: 2.5 mM NH4+, 22.95 mM

HAc (electrolyte system from [64]), C and D: LE: 5 mM NH4+, 30 mM HAc, TE: 2 mM NH4+, 31.181 mM HAc. The figure is a modification from [158].

The electrolyte system in Fig 28A and 28B was taken from [64], whereas the electrolyte composition in Fig. 28C and 28D was selected with the aim of creating a mobility window for targeting Aβ peptides. GENTRANS computer simulations were performed with Aβ 1-40 as model peptide. For a LE composition of 30 mM HAc and 5 mM NH4

+ the amount of NH4+ in

the TE was varied from 0 to 2 mM. Without NH4+ in the TE, all components expected to be

present after IP of Aβ from CSF or plasma (see Section 6.1) migrate with the ITP boundary, as

40

seen in Fig. 29A. The change in conductivity demarcates the ITP boundary. By increasing the amount of NH4

+ in the TE to 1 mM, 6E10 no longer migrates with the ITP boundary (Fig. 29B), whereas a further increase of NH4

+ to 2 mM also eliminates sAPPα (Fig. 29C).

Figure 29. Component migration in an electrolyte system with LE: 30 mM HAc, 5 mM NH4+ and TE: A) 31.968 mM HAc, B) 31.574 mM HAc, 1 mM NH4+ and C) 31.181 mM HAc, 2 mM NH4+ after 20 s

application of 100 V/cm. The lower level plots show a magnification of the boundary region. The anode is on the left. (Paper V)

Simulations were performed with other C-terminal Aβ isoforms as well, which were found to also migrate with the ITP boundary. Thus, the results suggest that the presented method could potentially be used for purification and preconcentration of Aβ peptides. Since the final aim is to use this method in a miniaturized, chip-based setting, a 2D model was also developed, as described in the next section.

41

7.2 Geometry and COMSOL simulations

COMSOL was used for simulations in a geometry that could be incorporated onto a microchip. A potential design is shown schematically in Fig. 30A. It includes two reservoirs for application of injection voltage, Vinj, a sample loading reservoir and a separation channel with a buffer reservoir. The separation channel consists of a wide (500 µm) channel followed by a narrower channel (250 µm) to facilitate loading of a large (µL range) sample volume. To enable direct infusion into ESI-MS from the chip corner a sheath-flow channel is included. A slightly simplified version of this geometry was used in the COMSOL simulations (Fig. 30B).

Figure 30. A) Scheme showing possible geometry for incorporation onto a microchip. B) Scheme showing the geometry used in COMSOL simulations. (Paper V)

In COMSOL a 2D model was created using the Tertiary Current Distribution Nernst-Planck interface. The proteins were treated in a simplified manner, i.e. by assigning a single charge to each species (the theoretically predicted value from the calculated titration curves, see Section 3.1.3). The separate sample reservoir included in Fig. 30B, allows for injection of a large portion of the sample compared to the amount of sample injected by applying the injection voltage to the sample reservoir (Paper V).

In Fig. 31, the concentration of Aβ 1-40 in the separation channel is shown every 5 s after application of the separation voltage. It can be seen that a peak is formed and focuses as it migrates along the channel. As the peptide enters the narrower channel, i.e. at an arc length of approx. 0.06 m, an additional enrichment is obtained. In future work, this system will be developed experimentally.

42

Figure 31. Simulated Aβ 1-40 peak position along the separation axis at every 5 s after application of 5000 V. The anode is on the left. (Paper V)

43

8 Conclusions and future perspectives

In this thesis, new approaches for electrophoretic preconcentration and separation of peptides and proteins in microchannels have been introduced, together with their successful combination with MS detection.

Initial experiments were performed in carrier-free systems. Although it was found that complete absence of buffer affects the system performance in terms of robustness and achievable degree of preconcentration, it was possible to preconcentrate simple samples consisting of model proteins or peptides. For a protein digest a division of acidic and basic peptides could be achieved. When the preconcentration was performed in acidic solution, Aβ peptides could be detected in concentrations down to 1 nM, and the method was applicable as a fast and simple preconcentration of Aβ peptides immunoprecipitated from a cell culture.

The combination of computer simulation and experiments in the open microchannel setup was used to create a pH gradient designed for IEF of Aβ peptides. When small and chemically well-defined amphoteric carriers were used, an agreement between the computer-predicted and experimentally observed location of the focused zones was obtained. IEF of Aβ after IP from CSF and plasma, in which they are present in pg/mL or pM concentrations, followed by MALDI-MS detection could be performed. This method thereby has potential to contribute to AD related research. By selecting other carrier components, the pH-step for focusing can easily be altered, and if a higher number of carriers are included, a more selective fractionation could be achieved. Thus, the presented approach with pH gradients tailor-made for specific analytes could find numerous applications within bioanalysis.

As an alternative strategy, extended ITP was used to create a mobility window aimed for preconcentration and purification of Aβ peptides in an ESI-MS compatible electrolyte system. Computer simulation was again utilized to obtain information about the system behavior. Simulation predicted that an efficient removal of possible contaminants after IP could be achieved. An agreement between computer-predicted and experimental results (with CITP-ESI-MS) was obtained for a CytC digest. Future work of interest would be performing corresponding experiments with Aβ peptides, and miniaturizing the setup from a capillary-based to a chip-based format, which was the final aim of this project.

The flexibility of the presented open microchannel setup is demonstrated e.g. by its ability to handle various sample volumes; while analysis of nanoliter-sized volumes can be achieved, the open setup also allows for supplying sample volumes exceeding the 25-75 nL microchannel volume by utilizing evaporation of the solvent during the sample supply. During voltage application and/or sampling the inert fluorocarbon liquid instead hinders evaporation from the channel while maintaining access to the channel contents. Consequently, the open, or semi-open, nature of the setup also allows for several interfacing strategies to MALDI- and nESI-MS; the open silicon channel can be used directly as MALDI target, or a fraction of the channel content containing the preconcentrated analytes can be sampled. Using a nano-ESI emitter for sampling enables performing direct nESI-MS analysis of the sampled species. Alternatively, the sampled volume can be deposited onto a MALDI target. As demonstrated, the latter case further allows for performing on-plate reactions in the sampled nanodroplets under the fluorocarbon lid.

44

In summary, a simple yet versatile preconcentration concept has been introduced. Its usefulness has been successfully demonstrated by analysis of Aβ peptides. The potential applications of the presented concepts can likely be extended to a wide variety of samples and analytes, and could thereby contribute to important findings within fields such as bioanalysis, proteomics and clinical analysis.

45

9 Acknowledgements

First of all I would like to express my sincere gratitude to my supervisors Åsa Emmer and Johan Jacksén for of your support and guidance in all parts of this work.

I would also like to thank my coauthors, Maria Khihon Rokhas, Wolfgang Thormann, Johan Roeraade and Henrik Ekström, for your important contributions to the presented papers.

Wolfgang Thormann is particularly acknowledged for providing and demonstrating the software GENTRANS, and for many interesting discussions about electrophoresis.

KTH Royal Institute of Technology, the Swedish Research Council, the EU-project NaDiNe and Stiftelsen Claes Adelskölds medalj och minnesfond (the Royal Swedish Academy of Sciences, Kungl. Vetenskapsakademien) are acknowledged for financial support. Furthermore, California Separation Science Society (CASSS), Gålöstiftelsen, Linders stiftelse and Klasons stiftelse (KTH Royal Institute of Technology) and the ÅForsk Foundation are acknowledged for conference scholarships and travel grants.

Also, I would like to thank Thomas Frisk for fabricating the open channel microchips, Erik Tingborn for performing initial experiments on the extended ITP systems and Hans Klafki for helpful discussions about Aβ peptides.

Thank you to present and former colleagues in the group of Analytical Chemistry, Leila Josefsson, Linus Svenberg, Yuye Zhou, Catharina Silfverbrand Lindh and Patrik Ek for all the good times that we have had.

Lastly, many thanks to my friends and family, especially David, for your love, patience and for always believing in me!

46

47

10 References

[1] Kovarik, M. L., Gach, P. C., Ornoff, D. M., Wang, Y., Balowski, J., Farrag, L., Allbritton, N. L., Analytical Chemistry 2012, 84, 516-540.

[2] Castro, E. R., Manz, A., Journal of Chromatography A 2015, 1382, 66-85.

[3] Štěpánová, S., Kašička, V., Journal of Separation Science 2016, 39, 198-211.

[4] Guihen, E., Electrophoresis 2014, 35, 138-146.

[5] Lin, C. C., Hsu, J. L., Lee, G. B., Microfluidics and Nanofluidics 2011, 10, 481-511.

[6] Thormann, W., Breadmore, M. C., Caslavska, J., Mosher, R. A., Electrophoresis 2010, 31, 726-754.

[7] Mosher, R. A., Breadmore, M. C., Thormann, W., Electrophoresis 2011, 32, 532-541.

[8] Wang, J., Talanta 2002, 56, 223-231.

[9] Mark, J. J. P., Scholz, R., Matysik, F. M., Journal of Chromatography A 2012, 1267, 45-64.

[10] Götz, S., Karst, U., Analytical and Bioanalytical Chemistry 2007, 387, 183-192.

[11] Uchiyama, K., Nakajima, H., Hobo, T., Analytical and Bioanalytical Chemistry 2004, 379, 375-382.

[12] Wang, X., Yi, L., Mukhitov, N., Schrell, A. M., Dhumpa, R., Roper, M. G., Journal of Chromatography A 2015, 1382, 98-116.

[13] Feng, X., Liu, B. F., Li, J., Liu, X., Mass Spectrometry Reviews 2015, 34, 535-557.

[14] He, X., Chen, Q., Zhang, Y., Lin, J. M., TrAC - Trends in Analytical Chemistry 2014, 53, 84-97.

[15] Kok, W., Capillary Electrophoresis: Instrumentation and Operation, H. Weinheimer 2000.

[16] Tiselius, A., Transactions of the Faraday Society 1937, 33, 524-531.

[17] Mikkers, F. E. P., Everaerts, F. M., Verheggen, Th. P. E. M., Journal of Chromatography A 1979, 169, 11-20.

[18] Jorgenson, J. W., Lukacs, K. D., Analytical Chemistry 1981, 53, 1298-1302.

[19] Harrison, D. J., Manz, A., Fan, Z., Lüdi, H., Widmer, H. M., Analytical Chemistry 1992, 64, 1926-1932.

48

[20] Harrison, D. J., Fluri, K., Seiler, K., Fan, Z., Effenhauser, C. S., Manz, A., Science 1993, 261, 895-897.

[21] Breadmore, M. C., Journal of Chromatography A 2012, 1221, 42-55.

[22] Breadmore, M. C., Tubaon, R. M., Shallan, A. I., Phung, S. C., Abdul Keyon, A. S., Gstoettenmayr, D., Prapatpong, P., Alhusban, A. A., Ranjbar, L., See, H. H., Dawod, M., Quirino, J. P., Electrophoresis, 36, 36-61.

[23] Sueyoshi, K., Kitagawa, F., Otsuka, K., Journal of Separation Science 2008, 31, 2650-2666.

[24] Koegler, W. S., Ivory, C. F., Journal of Chromatography A 1996, 726, 229-236.

[25] Ross, D., Locascio, L. E., Analytical Chemistry 2002, 74, 2556-2564.

[26] Ivory, C. F., Separation Science and Technology 2000, 35, 1777-1793.

[27] Ivory, C. F., Electrophoresis 2007, 28, 15-25.

[28] Righetti, P. G., Fasoli, E., Righetti, S. C., Protein Purification: Principles, High Resolution Methods, and Applications: Third Edition 2011, pp. 379-409.

[29] Kolin, A., The Journal of Chemical Physics 1954, 22, 1628-1629.

[30] Svensson, H., Acta Chemica Scandinavica 1961, 15, 325-341.

[31] Svensson, H., Acta Chemica Scandinavica 1962, 16, 456-466.

[32] Vesterberg, O., Acta Chemica Scandinavica 1969, 23, 2653-2666.

[33] Bjellqvist, B., Ek, K., Righetti, P. G., Gianazza, E., Görg, A., Westermeier, R., Postel, W., Journal of Biochemical and Biophysical Methods 1982, 6, 317-339.

[34] Hjertén, S., Zhu, M. d., Journal of Chromatography A 1985, 346, 265-270.

[35] Hofmann, O., Che, D., Cruickshank, K. A., Muller, U. R., Analytical Chemistry 1999, 71, 678-686.

[36] Mao, Q., Pawliszyn, J., Analyst 1999, 124, 637-641.

[37] Sommer, G. J., Hatch, A. V., Electrophoresis 2009, 30, 742-757.

[38] Righetti, P. G., Simó, C., Sebastiano, R., Citterio, A., Electrophoresis 2007, 28, 3799-3810.

[39] Hühner, J., Lämmerhofer, M., Neusüß, C., Electrophoresis 2015, 36, 2670-2686.

[40] Righetti, P. G., Sebastiano, R., Citterio, A., Proteomics 2013, 13, 325-340.

49

[41] Sova, O., Journal of Chromatography A 1985, 320, 15-22.

[42] Tomáš, R., Yan, L., Křenková, J., Foret, F., Electrophoresis 2007, 28, 2283-2290.

[43] Macounová, K., Cabrera, C. R., Holl, M. R., Yager, P., Analytical Chemistry 2000, 72, 3745-3751.

[44] Cabrera, C. R., Finlayson, B., Yager, P., Analytical Chemistry 2001, 73, 658-666.

[45] Macounová, K., Cabrera, C. R., Yager, P., Analytical Chemistry 2001, 73, 1627-1633.

[46] Huang, T., Wu, X. Z., Pawliszyn, J., Analytical Chemistry 2000, 72, 4758-4761.

[47] Zhu, G., Sun, L., Keithley, R. B., Dovichi, N. J., Analytical Chemistry 2013, 85, 7221-7229.

[48] Zhu, G., Sun, L., Yang, P., Dovichi, N. J., Analytica Chimica Acta 2012, 750, 207-211.

[49] Tran, J. C., Doucette, A. A., Journal of Proteome Research 2008, 7, 1761-1766.

[50] Everaerts, F. M., Beckers, J. L., Verheggen, Th. P. E. M., Isotachophoresis Theory, Instrumentation and Applications, Elsevier, Amsterdam 1976.

[51] Malá, Z., Gebauer, P., Boček, P., Electrophoresis 2015, 36, 2-14.

[52] Smejkal, P., Bottenus, D., Breadmore, M. C., Guijt, R. M., Ivory, C. F., Foret, F., Macka, M., Electrophoresis 2013, 34, 1493-1509.

[53] Petr, J., Maier, V., Horáková, J., Ševcík, J., Stránský, Z., Journal of Separation Science 2006, 29, 2705-2715.

[54] Foret, F., Szoko, E., Karger, B. L., Journal of Chromatography A 1992, 608, 3-12.

[55] Bottenus, D., Hossan, M. R., Ouyang, Y., Dong, W. J., Dutta, P., Ivory, C. F., Lab on a Chip - Miniaturisation for Chemistry and Biology 2011, 11, 3793-3801.

[56] Bottenus, D., Jubery, T. Z., Ouyang, Y., Dong, W. J., Dutta, P., Ivory, C. F., Lab on a Chip - Miniaturisation for Chemistry and Biology 2011, 11, 890-898.

[57] Bahga, S. S., Kaigala, G. V., Bercovici, M., Santiago, J. G., Electrophoresis 2011, 32, 563-572.

[58] Schafer-Nielsen, C., Svendsen, P. J., Analytical Biochemistry 1981, 114, 244-262.

[59] Beckers, J. L., Journal of Chromatography A 1993, 641, 363-373.

[60] Beckers, J. L., Electrophoresis 2000, 21, 2788-2796.

[61] Malá, Z., Gebauer, P., Boček, P., Electrophoresis 2013, 34, 3072-3078.

50

[62] Gebauer, P., Malá, Z., Boček, P., Electrophoresis 2013, 34, 3245-3251.

[63] Gebauer, P., Malá, Z., Boček, P., Electrophoresis 2014, 35, 746-754.

[64] Malá, Z., Pantůčková, P., Gebauer, P., Boček, P., Electrophoresis 2013, 34, 777-784.

[65] Boček, P., Gebauer, P., Deml, M., Journal of Chromatography A 1981, 217, 209-224.

[66] Boček, P., Gebauer, P., Deml, M., Journal of Chromatography A 1981, 219, 21-28.

[67] Kohlrausch, F., Annalen der Physik 1897, 298, 209-239.

[68] Hruška, V., Gaš, B., Electrophoresis 2007, 28, 3-14.

[69] Thormann, W., Caslavska, J., Breadmore, M. C., Mosher, R. A., Electrophoresis 2009, 30, S16-S26.

[70] Jaroš, M., Hruška, V., Štědrý, M., Zusková, I., Gaš, B., Electrophoresis 2004, 25, 3080-3085.

[71] Bier, M., Palusinski, O. A., Mosher, R. A., Saville, D. A., Science 1983, 219, 1281-1287.

[72] Hruška, V., Jaroš, M., Gaš, B., Electrophoresis 2006, 27, 984-991.

[73] Bercovici, M., Lele, S. K., Santiago, J. G., Journal of Chromatography A 2009, 1216, 1008-1018.

[74] Saville, D. A., Palusinski, O. A., AIChE Journal 1986, 32, 207-214.

[75] Palusinski, O. A., Graham, A., Mosher, R. A., Bier, M., Saville, D. A., AIChE Journal 1986, 32, 215-223.

[76] Mosher, R. A., Dewey, D., Thormann, W., Saville, D. A., Bier, M., Analytical Chemistry 1989, 61, 362-366.

[77] Mosher, R. A., Gebauer, P., Caslavska, J., Thormann, W., Analytical Chemistry 1992, 64, 2991-2997.

[78] Mosher, R. A., Gebauer, P., Thormann, W., Journal of Chromatography 1993, 638, 155-164.

[79] Thormann, W., Zhang, C. X., Caslavska, J., Gebauer, P., Mosher, R. A., Analytical Chemistry 1998, 70, 549-562.

[80] Breadmore, M. C., Quirino, J. P., Thormann, W., Electrophoresis 2009, 30, 570-578.

[81] Bier, M., Mosher, R. A., Palusinski, O. A., Journal of Chromatography A 1981, 211, 313-335.

51

[82] Palusinski, O. A., Allgyer, T. T., Mosher, R. A., Bier, M., Saville, D. A., Biophysical Chemistry 1981, 13, 193-202.

[83] Palusinski, O. A., Bier, M., Saville, D. A., Biophysical Chemistry 1981, 14, 389-397.

[84] Thormann, W., Mosher, R. A., Bier, M., Journal of Chromatography A 1986, 351, 17-29.

[85] Mosher, R. A., Thormann, W., Bier, M., Journal of Chromatography A 1988, 436, 191-204.

[86] Mosher, R. A., Thormann, W., Kuhn, R., Wagner, H., Journal of Chromatography 1989, 478, 39-49.

[87] Mosher, R. A., Saville, D. A., Thormann, W., The Dynamics of Electrophoresis, VCH, Weinheim 1992.

[88] Mosher, R. A., Thormann, W., Electrophoresis 2008, 29, 1036-1047.

[89] Egatz-Gomez, A., Thormann, W., Electrophoresis 2011, 32, 1433-1437.

[90] Dixon, D. R., Clark, S. B., Ivory, C. F., Electrophoresis 2012, 33, 880-888.

[91] Jacroux, T., Bottenus, D., Rieck, B., Ivory, C. F., Dong, W. J. I., Electrophoresis 2014, 35, 2029-2038.

[92] Cui, H., Dutta, P., Ivory, C. F., Electrophoresis 2007, 28, 1138-1145.

[93] Cong, Y., Bottenus, D., Liu, B., Clark, S. B., Ivory, C. F., Electrophoresis 2014, 35, 646-653.

[94] Bottenus, D., Jubery, T. Z., Dutta, P., Ivory, C. F., Electrophoresis 2011, 32, 550-562.

[95] Karas, M., Bachmann, D., Hillenkamp, F., Analytical Chemistry 1985, 57, 2935-2939.

[96] Karas, M., Hillenkamp, F., Analytical Chemistry 1988, 60, 2299-2301.

[97] Tanaka, K., Waki, H., Ido, Y., Akita, S., Yoshida, Y., Yoshida, T., Matsuo, T., Rapid Communications in Mass Spectrometry 1988, 2, 151-153.

[98] El-Aneed, A., Cohen, A., Banoub, J., Applied Spectroscopy Reviews 2009, 44, 210-230.

[99] Hillenkamp, F., Jaskolla, T. W., Karas, M., The MALDI Process and Method pp. 1-28 in MALDI MS: A Practical Guide to Instrumentation, Methods, and Applications, Wiley-VCH Verlag GmbH & Co. KGaA, 2014.

[100] Duncan, M. W., Roder, H., Hunsucker, S. W., Briefings in Functional Genomics and Proteomics 2008, 7, 355-370.

52

[101] Axelsson, J., Hoberg, A. M., Waterson, C., Myatt, P., Shield, G. L., Varney, J., Haddleton, D. M., Derrick, P. J., Rapid Communications in Mass Spectrometry 1997, 11, 209-213.

[102] Wei, H., Nolkrants, K., Powell, D. H., Woods, J. H., Ko, M. C., Kennedy, R. T., Rapid Communications in Mass Spectrometry 2004, 18, 1193-1200.

[103] Tu, T., Gross, M. L., TrAC - Trends in Analytical Chemistry 2009, 28, 833-841.

[104] DeVoe, D. L., Lee, C. S., Electrophoresis 2006, 27, 3559-3568.

[105] Schuerenberg, M., Luebbert, C., Eickhoff, H., Kalkum, M., Lehrach, H., Nordhoff, E., Analytical Chemistry 2000, 72, 3436-3442.

[106] Nordhoff, E., Schürenberg, M., Thiele, G., Lübbert, C., Kloeppel, K. D., Theiss, D., Lehrach, H., Gobom, J., International Journal of Mass Spectrometry 2003, 226, 163-180.

[107] Sjödahl, J., Kempka, M., Hermansson, K., Thorsén, A., Roeraade, J., Analytical Chemistry 2005, 77, 827-832.

[108] Jacksén, J., Ek, P., Schönberg, T., Roeraade, J., Manuscript in preparation.

[109] Dole, M., Mack, L. L., Hines, R. L., Chemistry, D. O., Mobley, R. C., Ferguson, L. D., Alice, M. B., The Journal of Chemical Physics 1968, 49, 2240-2249.

[110] Yamashita, M., Fenn, J. B., Journal of Physical Chemistry 1984, 88, 4451-4459.

[111] Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., Whitehouse, C. M., Science 1989, 246, 64-71.

[112] Wilm, M., Mann, M., Analytical Chemistry 1996, 68, 1-8.

[113] Bruins, A. P., Journal of Chromatography A 1998, 794, 345-357.

[114] Koster, S., Verpoorte, E., Lab on a Chip - Miniaturisation for Chemistry and Biology 2007, 7, 1394-1412.

[115] Gao, D., Liu, H., Jiang, Y., Lin, J.-M., Lab on a Chip 2013, 13, 3309-3322.

[116] Liu, J., Tseng, K., Garcia, B., Lebrilla, C. B., Mukerjee, E., Collins, S., Smith, R., Analytical Chemistry 2001, 73, 2147-2151.

[117] Fujita, M., Hattori, W., Sano, T., Baba, M., Someya, H., Miyazaki, K., Kamijo, K., Takahashi, K., Kawaura, H., Journal of Chromatography A 2006, 1111, 200-205.

[118] Mok, M. L. S., Hua, L., Phua, J. B. C., Wee, M. K. T., Sze, N. S. K., Analyst 2004, 129, 109-110.

[119] Guo, X., Chan-Park, M. B., Yoon, S. F., Chun, J. H., Hua, L., Sze, N. S. K., Analytical Chemistry 2006, 78, 3249-3256.

53

[120] Du, W., Li, L., Nichols, K. P., Ismagilov, R. F., Lab on a Chip - Miniaturisation for Chemistry and Biology 2009, 9, 2286-2292.

[121] Wang, S., Chen, S., Wang, J., Xu, P., Luo, Y., Nie, Z., Du, W., Electrophoresis 2014, 35, 2528-2533.

[122] Xue, Q., Foret, F., Dunayevskiy, Y. M., Zavracky, P. M., McGruer, N. E., Karger, B. L., Analytical Chemistry 1997, 69, 426-430.

[123] Figeys, D., Ning, Y., Aebersold, R., Analytical Chemistry 1997, 69, 3153-3160.

[124] Ramsay, R. S., Ramsey, J. M., Analytical Chemistry 1997, 69, 1174-1178.

[125] Sikanen, T., Franssila, S., Kauppila, T. J., Kostiainen, R., Kotiaho, T., Ketola, R. A., Mass Spectrometry Reviews 2010, 29, 351-391.

[126] Bonvin, G., Schappler, J., Rudaz, S., Journal of Chromatography A 2012, 1267, 17-31.

[127] Zhong, X., Zhang, Z., Jiang, S., Li, L., Electrophoresis 2014, 35, 1214-1225.

[128] Güntert, A., Döbeli, H., Bohrmann, B., Neuroscience 2006, 143, 461-475.

[129] Galozzi, S., Marcus, K., Barkovits, K., Expert Review of Proteomics 2015, 12, 343-354.

[130] Wiltfang, J., Esselmann, H., Bibl, M., Smirnov, A., Otto, M., Paul, S., Schmidt, B., Klafki, H. W., Maler, M., Dyrks, T., Bienert, M., Beyermann, M., Rüther, E., Kornhuber, J., Journal of Neurochemistry 2002, 81, 481-496.

[131] Lewczuk, P., Esselmann, H., Bibl, M., Paul, S., Svitek, J., Miertschischk, J., Meyrer, R., Smirnov, A., Maler, J. M., Klein, C., Otto, M., Bleich, S., Sperling, W., Kornhuber, J., Rüther, E., Wiltfang, J., Electrophoresis 2004, 25, 3336-3343.

[132] Olsson, B., Lautner, R., Andreasson, U., Öhrfelt, A., Portelius, E., Bjerke, M., Hölttä, M., Rosén, C., Olsson, C., Strobel, G., Wu, E., Dakin, K., Petzold, M., Blennow, K., Zetterberg, H., The Lancet Neurology 2016, 15, 673-684.

[133] Barrow, C. J., Yasuda, A., Kenny, P. T. M., Zagorski, M. G., Journal of Molecular Biology 1992, 225, 1075-1093.

[134] Bros, P., Delatour, V., Vialaret, J., Lalere, B., Barthelemy, N., Gabelle, A., Lehmann, S., Hirtz, C., Clinical Chemistry and Laboratory Medicine 2015, 53, 1483-1493.

[135] Pannee, J., Törnqvist, U., Westerlund, A., Ingelsson, M., Lannfelt, L., Brinkmalm, G., Persson, R., Gobom, J., Svensson, J., Johansson, P., Zetterberg, H., Blennow, K., Portelius, E., Neuroscience Letters 2014, 573, 7-12.

[136] Kaneko, N., Yamamoto, R., Sato, T. A., Tanaka, K., Proceedings of the Japan Academy Series B: Physical and Biological Sciences 2014, 90, 104-117.

54

[137] Kaneko, N., Nakamura, A., Washimi, Y., Kato, T., Sakurai, T., Arahata, Y., Bundo, M., Takeda, A., Niida, S., Ito, K., Toba, K., Tanaka, K., Yanagisawa, K., Proceedings of the Japan Academy Series B: Physical and Biological Sciences 2014, 90, 353-364.

[138] Portelius, E., Westman-Brinkmalm, A., Zetterberg, H., Blennow, K., Journal of Proteome Research 2006, 5, 1010-1016.

[139] Portelius, E., Zetterberg, H., Andreasson, U., Brinkmalm, G., Andreasen, N., Wallin, A., Westman-Brinkmalm, A., Blennow, K., Neuroscience Letters 2006, 409, 215-219.

[140] Picou, R. A., Kheterpal, I., Wellman, A. D., Minnamreddy, M., Ku, G., Gilman, S. D., Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences 2011, 879, 627-632.

[141] Verpillot, R., Otto, M., Klafki, H., Taverna, M., Journal of Chromatography A 2008, 1214, 157-164.

[142] Varesio, E., Rudaz, S., Krause, K. H., Veuthey, J. L., Journal of Chromatography A 2002, 974, 135-142.

[143] Mesbah, K., Verpillot, R., Chiari, M., Pallandre, A., Taverna, M., Analyst 2014, 139, 6547-6555.

[144] Mohamadi, M. R., Svobodova, Z., Verpillot, R., Esselmann, H., Wiltfang, J., Otto, M., Taverna, M., Bilkova, Z., Viovy, J. L., Analytical Chemistry 2010, 82, 7611-7617.

[145] Maler, J. M., Klafki, H. W., Paul, S., Spitzer, P., Groemer, T. W., Henkel, A. W., Esselmann, H., Lewczuk, P., Kornhuber, J., Wiltfang, J., Proteomics 2007, 7, 3815-3820.

[146] Haußmann, U., Jahn, O., Linning, P., Janßen, C., Liepold, T., Portelius, E., Zetterberg, H., Bauer, C., Schuchhardt, J., Knölker, H. J., Klafki, H., Wiltfang, J., Analytical Chemistry 2013, 85, 8142-8149.

[147] Oukacine, F., Taverna, M., Analytical Chemistry 2014, 86, 3317-3322.

[148] Mai, T. D., Oukacine, F., Taverna, M., Journal of Chromatography A 2016, 1453, 116-123.

[149] Székely, L., Guttman, A., Electrophoresis 2005, 26, 4590-4604.

[150] Jacksén, J., Frisk, T., Redeby, T., Parmar, V., van der Wijngaart, W., Stemme, G., Emmer, Å., Electrophoresis 2007, 28, 2458-2465.

[151] Khihon Rokhas, M., Mikkonen, S., Beyer, J., Jacksén, J., Emmer, Å., Electrophoresis 2014, 35, 450-457.

[152] Litborn, E., Roeraade, J., Journal of Chromatography B: Biomedical Sciences and Applications 2000, 745, 137-147.

55

[153] Ek, P., Stjernström, M., Emmer, Å., Roeraade, J., Rapid Communications in Mass Spectrometry 2010, 24, 2561-2568.

[154] Hartmann, M., Sjödahl, J., Stjernström, M., Redeby, J., Joos, T., Roeraade, J., Analytical and Bioanalytical Chemistry 2009, 393, 591-598.

[155] Mikkonen, S., Diploma thesis, Dept. of Chemistry, Div. of Applied Physical Chemistry - Analytical Chemistry, KTH Royal Institute of Technology, Stockholm, Sweden 2011.

[156] Persat, A., Chambers, R. D., Santiago, J. G., Lab on a Chip - Miniaturisation for Chemistry and Biology 2009, 9, 2437-2453.

[157] Jacksén, J., Roeraade, J., Unpublished work, part of EU Project NaDiNe, Dept. of Chemistry, Div. of Applied Physical Chemistry - Analytical Chemistry, KTH Royal Institute of Technology, Stockholm, Sweden 2014.

[158] Tingborn, E., Master Thesis, Dept of Chemistry, Div. of Applied Physical Chemistry - Analytical Chemistry, KTH Royal Institute of Technology, Stockholm, Sweden 2016.