1 Novel Nanomaterials and Chromatographic System for Enhanced Separation and Characterization of Biomacromolecules and Nanoparti

Novel Nanomaterials and Chromatographic System for Enhanced Separation and

Characterization of Biomacromolecules and Nanoparticles

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Yanhui Wang, M.S.

Graduate Program in Chemistry

The Ohio State University

2018

Dissertation Committee

Dr. Susan V. Olesik, Advisor

Dr. Philip Grandinetti

Dr. Abraham Badu-Tawiah

Copyrighted by

Yanhui Wang

2018

Abstract

With recent advances in technologies and methodologies, proteomics, which is the large-scale analysis of proteins, has been continuously developed in the field of bioinformatics, biotherapeutics and biomarker discovery. Top-down proteomics, which focuses on the analysis of intact proteins, has emerged within the last decade with significant advantages over the traditional bottom-up approach, such as the characterization of labile protein structures and the universal detection of all existing modifications. The front-end separation technologies for intact proteins are of the primary importance for the successful implementation of top-down proteomics. The work reported herein focuses the development of miniaturized liquid chromatography (LC) system and an effective and eco-friendly solvent system to address the challenges faced in intact protein separation and characterization.

Electrospun nanofibers featuring effective chromatographic performance as the stationary phase of the ultrathin layer chromatography (UTLC) was developed in this work for the separation of amino acids and intact proteins. Nafion, a synthetic perfluorinated cationic polymer, was incorporated into a carrier polymer, polyacrylonitrile (PAN), to fabricate the nanofibrous stationary phase via electrospinning method. The separation of charged amino acids and proteins on the Nafion-PAN UTLC was based on the ion exchange mechanism (IEX). Design of experiments (DOEs) ii methods were applied to optimize the Nafion-PAN stationary phase and separation conditions. The nanofibers exhibited excellent mechanical stability and solvent compatibility. The separation of amino acids confirmed the feasibility of Nafion-PAN nanofibers as the ion exchange UTLC stationary phase. The separation of intact proteins has illustrated that Nafion-PAN stationary phase is also suitable for separation of large biomolecules. The retention of proteins on the Nafion-PAN UTLC largely depends on the properties of proteins including the net charge, hydropathicity, molecular size and structure. The Nafion-PAN UTLC demonstrated high separation efficiency for both amino acids and intact proteins.

In addition to intact protein separation, a micellar liquid chromatography (MLC) system was developed using the polyacrylonitrile UTLC device for the size characterization of polyethylene glycol (PEG)ylated gold nanoparticle (AuNP), an emerging agent in cancer therapeutics, of which the cellular uptakes and cytotoxicity are highly dependent on its size. PEGylated AuNPs with different sizes in the range of 10-80 nm were well separated from each other. The developed method also permitted the separation of AuNPs capped with different molecular weight of PEG in the range of 2-20 kDa. Micellar mobile phases were adopted to provide a highly biodegradable chromatographic system. This method exhibited excellent separation performance with smallest plate heights < 2 µm and resolution of each pair of AuNPs > 1.5. Decent separations for all PEGylated AuNPs could be achieved within 5 min. This method was applied to monitor the transformation of AuNPs in serum protein, serving as a rapid and

iii convenient tool for characterization of size distribution and modification of PEGylated

AuNPs.

Efforts towards improving the mobile phase system of intact protein separation were also made by adopting enhanced fluidity liquid chromatography (EFLC), which involves the addition of liquefied gas, such as carbon dioxide, to the conventional liquid mobile phases. The addition of liquefied CO2 provides increased diffusivity and decreased viscosity of the mobile phase, which inherently leads to more efficient separation. Herein, EFLC was first time applied to hydrophobic interaction chromatography (HIC) to study the impact of liquefied CO2 on the chromatographic behaviors of proteins. Since conventional HIC is known to preserve the native structure of proteins, the effects of liquefied CO2 on protein structures, charge state distributions

(CSD) and ionization efficiencies would be more pronounced by adapting EFLC to HIC online coupled to electrospray ionization mass spectrometry (ESI-MS). In this work,

EFLC offered improved chromatographic behaviors including a shorter analysis time, better peak shapes and a higher plate number. Liquefied CO2 proved to be an ESI friendly and “supercharging” reagent without sacrificing chromatographic performance, which can be used to improve peptide and protein identification in the large-scale application.

The EFLC system was also applied to reversed phase chromatography (RPC) for intact protein separation. The EFLC solvents utilizing liquefied CO2/methanol/water mixture could be considered as a “greener” alternative to traditional water/acetonitrile mobile phase in LC with better separation efficiency and peak symmetry. Various mobile

iv phase additives were compared under RPC mode to provide the optimum condition for integrated EFLC-MS system for intact protein separation and characterization.

Dedication

To my beloved grandma,

I miss you everyday

Acknowledgments

The completion of this dissertation would have not been possible without the support, guidance and efforts of many special individuals. I would like to extend my sincere gratitude to all of them.

First, I would like to express my special thanks to my advisor, Dr. Susan Olesik for her generous support, invaluable guidance and mentorship over the past five years.

Thank you for encouraging my research and allowing me to grow as a scientist. You have led me way farther than I thought I could go.

I would also like to thank every member from the Olesik research group for their dedicated efforts and cooperative attitude. I would especially like to recognize Michael

Beilke, Martin Beres, Raffeal Bennett, Jiayi Liu, Juan Bian, Brian Fitch and Rebekah

Gibson for helping me both professionally and personally. I’m grateful to have all of you who could share laughter, dreams and frustrations with me all along the way.

Finally, I would like to express my deepest gratitude to my family. Thank you to my parents for always believing in me and teaching me the most important job in life is to learn how to understand myself and others, to give out love and to be happy. Thank you to my big brother, Yanfeng, for taking good care of Mom and Dad. Grandma, I hope

I have made you proud. Words cannot express how much I miss all of you. To my husband-to-be, Wey Jian, this journey could never be possible without you. Thank you vii for always being there for me through thick and thin. Thank you for never letting me give up on myself and pushing me to be a better person day by day.

viii

Vita

2010 to 2013 ...... B.S. Medicinal Chemistry, University at Buffalo, SUNY,

Buffalo, NY

2013 to 2015 ...... M.S. Chemistry, The Ohio State University, Columbus, OH

2014 to 2018 ...... Graduate Teaching Assistant and Graduate Research

Assistant, Department of Chemistry and

Biochemistry, The Ohio State University, Columbus, OH

Publications

Wang, Y.; Olesik, V. S. Separation of PEGylated Gold Nanoparticles by Micellar

Enhanced Electrospun Fiber Based Ultrathin Layer Chromatography. Analytical

Chemistry 2018, 90 (4), pp 2662–2670.

Wang, Y.; Olesik, V. S. Electrospun Nafion/PAN as Ion Exchange Ultrathin Layer

Chromatography Stationary Phase. Analytica Chimica Acta 2017, 970, 82-90.

Deuro, R. E.; Lieker, K. M.; Wang, Y.; Caras, C. A.; Milillo, T. M.; Bright, F. V. Denim

Fiber Characterization Using Multispectral Luminescence Imaging. Applied Spectroscopy

2015, 69,103-114.

Fields of Study

Major Field: Chemistry

Table of Contents

Abstract ...... ii Dedication ...... vi Acknowledgments...... vii Vita ...... ix List of Tables ...... xv List of Figures ...... xvii Chapter 1. Introduction ...... 1 1.1 Overview ...... 1 1.2 Basic chromatography fundamentals ...... 2 1.2.1 Retention Factor and Retardation Factor ...... 2 1.2.2 Selectivity ...... 3 1.2.3 Chromatographic Efficiency, Peak Asymmetry and Resolution ...... 5 1.2.4 Peak Capacity...... 8 1.3 Top Down Proteomics ...... 9 1.4 Separation Mechanisms for Intact Proteins in Liquid Chromatography ...... 12 1.4.1 Reversed-phase chromatography ...... 12 1.4.2 Hydrophobic Interaction Chromatography ...... 14 1.4.3 Ion Exchange Chromatography ...... 18 1.4.4 Size Exclusion Chromatography...... 19 1.4.5 Affinity Chromatography...... 19 1.5 Gold Nanoparticles ...... 20 1.5.1 PEGylation ...... 21 1.5.2 Size Separation and Characterization Methods for AuNPs ...... 22 1.6 Ultrathin Layer Chromatography and Electrospinning...... 24 1.7 Enhanced Fluidity Liquid Chromatography ...... 27 xi

1.8 Research Focus ...... 28 1.9 References ...... 29 Chapter 2. Electrospun Nafion-Polyacrylonitrile Nanofibers as an Ion Exchange Ultrathin Layer Chromatographic Stationary Phase ...... 34 2.1 Introduction ...... 34 2.2 Materials and Methods ...... 38 2.2.1 Materials ...... 38 2.2.2 Instrumentation ...... 39 2.2.3 Preparation of Nafion-PAN Solution ...... 39 2.2.4 Electrospinning ...... 39 2.2.5 Ion Exchange Capacity ...... 40 2.2.6 Ultrathin Layer Chromatography ...... 40 2.2.7 Visualization and Data Collection ...... 41 2.2.8 Experimental Designs ...... 42 2.3 Results and Discussion ...... 43 2.3.1 Optimization of the UTLC Conditions for Amino Acids ...... 43 2.3.2 Nanofiber Stability in Mobile Phases ...... 53 2.3.3 Mobile Phase Velocity ...... 55 ...... 58 2.3.4 Band Broadening ...... 59 2.3.5 Ion Exchange Capacity ...... 62 2.3.6 Separation of Amino Acids ...... 64 2.3.7 Separation of Proteins ...... 67 2.4 Conclusions ...... 76 2.5 References ...... 77 Chapter 3. Separation of PEGylated Gold Nanoparticles by Micellar Ultrathin Layer Chromatography using Electrospun Polyacrylonitrile Nanofibers as the Stationary Phase ...... 80 3.1 Introduction ...... 80 3.2 Materials and methods ...... 83 3.2.1 Materials ...... 83 3.2.2 Instrumentation ...... 84 3.2.3 Electrospining ...... 85

xii

3.2.4 Ultrathin Layer Chromatography ...... 85 3.2.5 PEGylation of AuNPs ...... 86 3.3 Results and Discussion ...... 86 3.3.1 Electrospun Nanofibers ...... 86 3.3.2 Mobile Phase Conditions ...... 88 3.3.2.1 Effect of Surfactants ...... 88 3.3.2.2 Effect of Alcohol in the Mobile Phase ...... 91 3.3.2.3 Effect of pH...... 94 3.3.3 Separation of AuNPs with Different Core Sizes and PEG Lengths ...... 95 3.3.4 PEGylated AuNPs in SDS Micellar UTLC ...... 104 3.3.5 Retention Order and PEG Density ...... 108 3.3.6 Chromatographic Performance of Electrospun UTLC ...... 114 3.3.7 Monitoring the Transformation of AuNPs in Serum Protein...... 119 3.4 Conclusions ...... 122 3.5 References ...... 123 Chapter 4. Enhanced-Fluidity Liquid Chromatography-Mass Spectrometry for Intact Protein Separation and Characterization ...... 127 4.1 Introduction ...... 127 4.2 Materials and Methods ...... 129 4.2.1 Chemicals and Reagents ...... 129 4.2.2 Sample Preparation ...... 130 4.2.3 EFLC-MS ...... 130 4.2.4 Data Acquisition and Analysis...... 131 4.3 Results and Discussion ...... 133 4.3.1 Characterization of the HIC Column for Protein Separation ...... 133

4.3.2 Addition of Liquefied CO2 to Isocratic Elution ...... 136 4.3.3 EFLC Dual Gradient Elution ...... 140

4.3.3.1 Effect of CO2 on Chromatographic Performance ...... 140 4.3.3.2 Sensitivity of EFLC Method to Minor Modifications ...... 145

4.3.3.3 Effect of CO2 on CSD and Ionization Efficiency ...... 147 4.3.4 Potential Factors that Affected CSD and Ionization Efficiency ...... 154 4.3.4.1 Solution pH ...... 154 4.3.4.2 Apparent Gas Phase Basicity of Protein ...... 155 xiii

4.3.4.3 Solvent Gas Phase Basicity ...... 156 4.3.4.4 Desolvation Process ...... 157

4.3.5 Supercharging effect of Liquefied CO2 ...... 161 4.4 Conclusions ...... 162 4.5 References ...... 163 Chapter 5. Enhanced-Fluidity Liquid as the Alternative Mobile Phase in Reversed Phase Chromatography for Intact Protein Separation ...... 167 5.1 Introduction ...... 167 5.2 Materials and Methods ...... 170 5.2.1 Chemicals and Reagents ...... 170 5.2.2 Sample Preparation ...... 170 5.2.3 Instrumentation ...... 171 5.2.4 Data Acquisition and Analysis...... 171 5.3 Results and Discussion ...... 172 5.3.1 Selection of Organic Cosolvent for EFLC ...... 172 5.3.2 Comparison between EFLC and HPLC with Optimized Mobile Phases ...... 178 5.3.3 Chromatographic Performance of EFLC and HPLC ...... 180 5.3.4 Additive choice for EFLC-MS...... 182 5.4 Conclusions ...... 190 5.5 References ...... 191 Chapter 6. Summary and Future Work ...... 193 6.1 Summary of Research ...... 193 6.2 Future Work ...... 196 6.3 References ...... 198 Bibliography ...... 200

xiv

List of Tables

Table 2.1 Summary of factors and levels used in Box-Behnken design...... 49

Table 2.2 ANOVA for response surface quadratic model for optimization of UTLC conditions for amino acids...... 50

Table 2.3 Ion exchange capacity of Nafion-PAN nanofibers...... 63

Table 2.4 Isoelectric point (pI), charge, Rf and plate height of amino acids on Nafion-

PAN UTLC...... 65

Table 2.5 Separation of amino acids on pure PAN UTLC in three repetitions (n=3)...... 66

Table 2.6 LOD and LOQ of Amino Acid...... 66

Table 2.7 Summary of factors and levels used in Box-Behnken design for the mobile phase in protein separation...... 68

Table 2.8 ANOVA for response surface quadratic model for the mobile phase in protein separation...... 69

Table 2.9 Molecular weights, pI, GRAVY scores and the separation results of the proteins used in this work...... 74

Table 2.10 Separation of intact proteins on pure PAN UTLC in three repetitions (n=3).75

Table 2.11 LOD and LOQ of Proteins...... 75

Table 3.1 Core diameter, concentration, surface area and hydrodynamic diameters of differently sized AuNP...... 96 xv

Table 3.2 Hydrodynamic diameter and zeta potential of Au-PEG in aqueous solution without/ with SDS above its CMC...... 106

Table 3.3 Summary of Flory radius (RF), layer thickness (L), PEG-PEG distance (D),

Area per molecule (A), density (d) and effective number of PEG (Neff) for different sized

PEGylated AuNPs...... 112

Table 3.4 Summary of Flory radius (RF), layer thickness (L), PEG-PEG distance (D),

Area per molecule (A), density (d), effective number (Neff) and molecular weight of PEG

(MWeff) for different sized PEGylated AuNPs...... 112

Table 3.5 Retardation factor, plate height and resolution of PEGylated AuNPs with different molecular weight of PEG...... 117

Table 4.1 Average protein charge state without/with liquefied CO2 and signal enhancement with addition of liquefied CO2...... 151

Table 5.1 Selected physical and chemical properties of water, acetonitrile, methanol and isopropanol...... 175

xvi

List of Figures

Figure 1.1 The effect of peak width on separation...... 4

Figure 1.2 Illustration of (A) peak asymmetry; (B) resolution...... 7

Figure 1.3 Workflows of (A) bottom up proteomics; (B) top down proteomics...... 11

Figure 1.4 Retention and elution steps in a HIC separation...... 16

Figure 1.5 Hofmeister series...... 17

Figure 1.6 Electrospinning apparatus...... 26

Figure 2.1 Molecular structure of Nafion. For 1100 equivalent weight Nafion used in this study, x= 6.5 (average value) and y= 1...... 37

Figure 2.2 (A) Standardized Pareto chart of the fractional factorial design for resolution,

(B) Main effects plot for resolution...... 46

Figure 2.3 (A) Contour plots using the Box-Behnken design showing the effect of (A)

Nafion concentration and migration distance on resolution (B) Nafion concentration and organic solvent on resolution (C) organic solvent and migration distance on resolution. 51

Figure 2.4 Response optimization plot...... 52

Figure 2.5 SEM images of Nafion-PAN nanofibers (A) before soaking (B) after being soaked in formate buffer: methanol: n-butanol (90: 5: 5 v/v/v) after 60 min, (C) in H2O:

2-propanol: NH4OH (54:43:3, v/v/v) + 0.15M NaCl after 60 min...... 54

xvii

Figure 2.6 (A) Comparison of mobile phase migration rate on UTLC plates containing

20% Nafion (●) with R2=0.998, 30% Nafion (▲) with R2=0.9936, 45% Nafion (■) with R2=0.999 and 60% Nafion (◆) with R2=0.998 (B) Fiber diameter of different

Nafion concentration...... 57

Figure 2.7 Comparison of contact angles of formate buffer: methanol: n-butanol (90: 5: 5 v/v/v) on (A) pure PAN UTLC; (B) 45% Nafion-PAN UTLC...... 58

Figure 2.8 Change in plate height H of phenylalanine with increasing solvent migration distance on Nafion-PAN UTLC. Error bars represent the standard deviation...... 61

Figure 2.9 Chromatogram for separation of (1) Arg, (2) His, (3) Lys, (4) Phe, (5) Val and

(6) Ala on electrospun Nafion-PAN UTLC using 30mM formate buffer (pH 3.5): methanol: n-butanol (90: 5: 5 v/v/v)...... 65

Figure 2.10 (A) Contour plots using the Box-Behnken design showing the effect of (A) organic solvent and salt concentration on resolution; (B) organic solvent and additive on resolution; (C) salt concentration and additive on resolution...... 70

Figure 2.11 Response optimization plot for protein separation...... 71

Figure 2.12 Chromatogram for separation of (1) LYZ, (2) BSA, (3) MGB, 4) CHY on electrospun Nafion-PAN UTLC using 0.15 M NaCl in H2O: 2-propanol: NH4OH (54: 43:

3, v/v/v)...... 74

Figure 3.1 (A) SEM image and (B) the fiber diameter distribution of electro-spun PAN nanofibers...... 87

Figure 3.2 Effect of (A) SDS, (B) CTAB, (C) Brij-35 on retardation factors of 10 nm, 30 nm, 50 nm and 80 nm Au-PEG...... 90

xviii

Figure 3.3 Effective hydrodynamic diameter of PEGylated AuNPs measured by DLS at different concentrations of 2-propanol in Tris buffer (10 mM, pH 8.5) with 3% (w/v)

SDS...... 92

Figure 3.4 Effect of (A) in micellar mobile phase, (B) 2-Propanol in aqueous phase, (C) pH on retardation factors of 10 nm, 30 nm, 50 nm and 80 nm Au-PEG...... 93

Figure 3.5 Separation of differently sized PEGylated AuNPs using 3% SDS in Tris buffer

(pH 8.5, 10 mM) with 7.5% 2-Propanol as the mobile phase: lane 1, 10 nm; lane 2, 30 nm; lane 3, 50 nm; lane 4, 80 nm; lane 5, mixture of 10 nm and 50 nm...... 97

Figure 3.6 Au ions on UTLC: a. before mobile phase development, b. after mobile phase development; 5 kDa PEG on UTLC: c. before mobile phase development, d. after mobile phase development...... 98

Figure 3.7 UV Spectra of Au-citrate and Au-PEG, Spectra of 30 nm AuNPs before and after PEGylation, the inset shows the clear red shift of about 2 nm after PEGylation. .. 100

Figure 3.8 Effect of surfactants on retardation factors of Au-PEG with different MW of

PEG. A: SDS, B: CTAB, C: Brij-35...... 101

Figure 3.9 Effect of A: 2-Propanol in micellar mobile phase; B: pH on retardation factors of Au-PEG with different MW of PEG...... 102

Figure 3.10 Separation of AuNPs with different MW of PEG using 2% SDS in CAPS buffer (pH 10, 10 Mm) with 7.5 % 2-Propanol as the mobile phase: lane 1, 2 kDa; lane 2,

5 kDa; lane 3, 10 kDa; lane 4, 20 kDa; lane 5, mixture of 2 kDa and 20 kDa...... 103

xix

Figure 3.11 Separation of differently sized PEGylated AuNPs on commercial Silica gel

60 F254 HPTLC using 3% SDS in Tris buffer (pH 8.5, 10 mM) with 7.5% 2-Propanol as the mobile phase: lane 1, 10 nm; lane 2, 30 nm; lane 3, 50 nm; lane 4, 80 nm...... 107

Figure 3.12 Scheme of effective interaction between grafted PEG on AuNP and the ligands on the nanofiber stationary phase...... 111

Figure 3.13 Plot of (A) Retardation factor vs. Log (Neff) for different sized Au-PEG with

2 R =0.996; (B) Retardation factor vs. Log (MWeff) for AuNPs capped with different molecular weight of PEG, R2 = 0.998...... 113

Figure 3.14 Separation efficiency: (A) Plate height of 10 nm, 30 nm, 50 nm and 80 nm

Au-PEG as a function of migration distance; (B) resolution of 10 nm/30 nm, 30 nm/50 nm and 50 nm/80 nm Au-PEG as a function of migration distance; (C) Mobile phase velocity on electrospun PAN UTLC with R2 = 0.997...... 116

Figure 3.15 A. UTLC of a) nonincubated Au-PEG, b) incubated Au-PEG before washing, c) incubated Au-PEG after washing, d) non-incubated Au-citrate, e) incubated Au- citrated before washing, d) non-incubated Au-citrate after washing. B. Rf values of incubated Au-PEG before and after washing at different incubation period...... 121

Figure 4.1 Ternary pump setup used in this work to run EFLC dual gradient...... 132

Figure 4.2 HPLC separations of protein mixture: 1. RNase A, 2. TI, 3. Chy, 4. Lys, 5.

ChA on PolyBUTYL A column (A, B, C, D) and PolyPENTYL A column (E, F, G, H).

MPA: 1 M ammonium acetate for all conditions. MPB: organic solvent in 20 mM ammonium acetate. Organic solvent in MPB: (A, E)-50% ACN; (B, F)-50% MeOH; (C,

G)-60% MeOH; (D, H)-70% MeOH. Gradient: 10 min MPA to MPB at 25 °C. Flow rate:

1 mL/min...... 135

Figure 4.3 (A): Addition of various amount of liquified CO2 to isocratic elution for Lys on PolyPENTYL A column. Volume percentage of CO2 in 300 mM ammonium acetate with 70% MeOH: (-) 0%; (-) 2.5%; (-) 5%; (-) 7.5%; (-) 10%; (-) 12.5%; (-) 15%. (B):

Mass spectrum of Lys with 0% CO2 in the mobile phase. Flow rate: 1 mL/min...... 138

Figure 4.4 Effect of the addition of liquified CO2 on the system back pressure. Mobile phase: various amount of CO2 in 300 mM ammonium acetate with 70 v% MeOH. Flow rate: 1 mL/min...... 139

Figure 4.5 DAD chromatogram at 280 nm of protein mixture separation on PolyPENTYL

A column at 25 °C. Mobile phase A: 1 M AmAc. Mobile phase B: MeOH. (A) HPLC gradient: 10 min 100 v% mobile phase A to 100 v% mobile phase B; (B) EFLC dual gradient: (1) 10 min 100 v% mobile phase A to 100 v% mobile phase B, (2) 0-6 min 1-5 v% CO2, 6-10 min 5-30 v% CO2. Flow rate: 1 mL/min. Analyte key: 1. RNase A, 2. Trp,

3. Chy, 4. ChA, 5. Lys...... 143

Figure 4.6 A comparison of HPLC gradient (A, C) and EFLC dual gradient (B, D) for chromatographic performances of ChA (A, B) and Lys (C, D): (o) data, (-) gaussian fits.

...... 144

Figure 4.7 EFLC-MS of Chy (A) and ChA (B) showing chromatographically separated proteoforms with minor modifications under dual gradient condition as described in

Figure 4.5C. Inset shows the mass spectra of two peaks (a1, a2 and b1, b2) of Chy and

ChA, respectively. The deconvoluted spectrum showing a 16 Da difference...... 146

xxi

Figure 4.8 Mass spectrum of Lys with (A) 0 v% CO2 in the mobile phase. (B) 2.5-15 v%

CO2 in the mobile phase. Flow rate: 1 mL/min...... 149

Figure 4.9 Mass spectrum of RNase A (A1), TI (B1), Chy (C1), ChA (D1) and Lys (E1) in LC gradient condition as described in Figure 4.5A; and mass spectrum of RNase A

(A2), TI (B2), Chy (C2), ChA (D2) and Lys (E2) in EFLC dual gradient condition as described in Figure 4.5B...... 150

Figure 4.10 Effect of (A, B) 0% v/v TEAA, (C, D) 5% v/v TEAA, (E, F) 10% v/v TEAA on the retention and peak shape of ChA (A, C, E) and Lys (B, D, F) in EFLC dual gradient elution...... 152

Figure 4.11 Effect of (A, B) 0% v/v TEAA, (C, D) 5% v/v TEAA, (E, F) 10% v/v TEAA on the CSD and ionization efficiency of ChA (A, C, E) and Lys (B, D, F) in EFLC dual gradient elution...... 153

Figure 4.12 Illustration of a charged droplet containing four molecules with different sizes and shapes at three different stages of solvent evaporation.54 ...... 158

Figure 4.13 Mass spectra of ChA using EFLC dual gradient condition as described in

Figure 1C with different gas temperatures at AJS electrospray ion source: (A) sheath gas:

350 °C, drying gas: 300 °C; (B) sheath gas: 250 °C, drying gas: 300 °C; (C) sheath gas:

150, drying gas: 200 °C...... 159

Figure 4.14 Mass spectra of ChA using EFLC dual gradient condition as described in

Figure 1C with different gas temperatures at AJS electrospray ion source: (A) sheath gas:

350 °C, drying gas: 300 °C; (B) sheath gas: 250 °C, drying gas: 300 °C; (C) sheath gas:

150, drying gas: 200 °C...... 160

xxii

Figure 5.1 DAD chromatogram at 210 nm of protein mixture using (A) IPA gradient: 0-

30 min from 20% to 40% IPA at 40 °C; (B) ACN gradient: 0-30 min from 20% to 60%

ACN at 40 °C; (C) MeOH gradient: 0-30 min from 20% to 60% MeOH at 40 °C; (D)

MeOH gradient: 0-30 min from 20% to 60% MeOH at 80 °C. Flow rate for all separations: 0.5 mL/min. Each mobile phase contained 0.2% v/v of TFA as additive.

Analyte key: 1. RNase A; 2. Cyt C.; 3. Lys; 4. BSA; 5: MGB; 6: ChA...... 174

Figure 5.2 Changes in system backpressure as the concentration of organic solvents increases in the H2O/Organic mixture. Flow rate: 0.5 mL/min...... 177

Figure 5.3 DAD chromatogram at 210 nm of protein mixture using (A) EFLC gradient:

2-5 min min from 2% to 15% CO2 at 40 °C; (B) HPLC-ACN gradient: 0-8 min from 30% to 60% ACN at 40 °C; (C) HPLC-MeOH gradient: 0-5-8 min from 60% to 80% to 100%

MeOH at 40 °C. Flow rate for all separations: 1 mL/min. Each mobile phase contained

0.2% v/v of TFA as additive. Analyte key: 1. RNase A; 2. Cyt C.; 3. Lys; 4. BSA; 5:

MGB...... 179

Figure 5.4 Comparison of peak asymmetry using EFLC mobile phase, HPLC mobile phases containing ACN and MeOH under optimized conditions (mobile phase conditions listed in Figure 5.3)...... 181

Figure 5.5 Ion pairing effect of TFA on proteins...... 183

Figure 5.6 Illustration of various important aspects of silica surface chemistry in RPC.183

Figure 5.7 Effect of different additives on separations of protein mixture. (A) 0.2% FA;

(B) 0.2% AA; (C) 10 mM pH 6.8 AmFm; (D) 10 mM pH 6.8 AmAc. EFLC gradient: 2-5

xxiii min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min. Analyte key: 1. RNase A;

2. Cyt C.; 3. Lys; 4. BSA; 5: MGB...... 187

Figure 5.8 Effect of different additives on MS sensitivity of MGB (A) no additive; (B)

0.2% TFA; (C) 0.2% FA; (D) 0.2% AA; (E) 10 mM pH 6 AmFm; (F) 10 mM pH 6

AmAc. EFLC gradient: 2-5 min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min...... 188

Figure 5.9 MS signal of (A) RNase A; (B) Cyt C; (C) Lys; (D) BSA; (E) Heme; (F) MGB using EFLC gradient: 2-5 min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min.

...... 189

Figure 6.1 Illustration of LDI, MALDI and SALDI for sample analysis on the surface of a TLC plate.5 ...... 197

xxiv

Chapter 1. Introduction

1.1 Overview

Proteins are the primary building block of life processes, and they are responsible for the basic function and structure for all living organisms. With recent advances in technologies and methodologies, proteomics, which is the large-scale analysis of proteins, has been continuously developed in the field of bioinformatics,1 biotherapeutics2 and biomarker discovery.3 Top-down proteomics, a mass spectrometry-based approach, that focuses on the analysis of intact proteins, allows for 100% sequence coverage and full characterization of proteins.4 However, to successfully implement top-down approach, many technically difficulties need to be overcome for proteome analysis at intact protein level. The front-end separation methods in top-down strategy have always represented a major challenge for intact proteins due to their large molecular weight

(MW), structural complexity and poor recovery. Herein the development of novel strategies to improve the intact protein separations are described. The novel approaches reported in this work are in two categories: (1) novel polymeric nanomaterials as the stationary phase for a miniaturized ultrathin layer chromatography (UTLC); (2) enhanced fluidity liquid (EFL) as the mobile phase to perform highly efficient intact protein separation on the current state-of-the-art instrumentations.

Owing to their unique physiochemical properties, gold nanoparticles (AuNPs) are being employed widely in biochemical applications for disease diagnosis and therapy.5,6

The properties and efficacy of AuNPs are governed by their morphology (e.g. size and shape) and surface chemistry (e.g. polyethylene glycol [PEG] functionalized).7

Separation of spherical AuNPs with varying core sizes and different PEG chain lengths using the developed micellar mobile phase on the UTLC device is also detailed herein.

1.2 Basic chromatography fundamentals

Chromatographic separation of macromolecules like proteins can be challenging.

Evaluating the chromatographic parameters is essential for the optimization of methods and conditions to achieve successful separation.

1.2.1 Retention Factor and Retardation Factor

In almost all the chromatographic mechanisms, the separation is driven by the differences in the relative distribution of analyte molecules between the stationary phase and the mobile phase. In column chromatography, retention time (tR) describes the amount of time taken for a molecule pass through a column from the time of injection to the time of detection. Retention factor, k, measures the time a substance spends in the stationary phase relative to the time it spends in the mobile phase. It is defined by

Equation 1.1:

푡 − 푡 푘 = 푅 푀 (1.1) 푡푀

2 where tM is the measure of the time it takes for the mobile phase or an unretained solute to flow through the column, which is commonly referred as the “dead time.”8 In thin layer chromatography, retardation factor Rf is commonly-used to qualitatively evaluate the retention of a compound:

푧푠 푅푓 = (1.2) 푧푓 − 푧0

Where zs is the migration distance of substance from the sample origin, zf is the migration distance of solvent from the sample origin, and z0 is the distance between immersion line and the sample origin.9 For instance, if a sample is totally unretained in a chromatographic system, the retention factor k is 0 and the retardation factor Rf is 1.

1.2.2 Selectivity

The relative position for two chromatographic peaks is measured by the separation factor, or “selectivity”, α.

푘 훼 = 퐵 (1.3) 푘퐴 where kB is the retention factor of a more retained solute and kA is the retention factor for a less retain solute. The larger the value of α, the larger the difference in retention of two solutes. However, the separation factor is not the only factor to determine how well two components are resolved. As illustrated in Figure 1.1, the retention factors for the two components are the same in both chromatograms, therefore the α values are also the same. Yet, it is obvious that the two peaks in Figure 1.1B are not fully resolved due to the widths of the peaks. The width and the shape of the peaks also need to be considered.

Figure 1.1 The effect of peak width on separation.

1.2.3 Chromatographic Efficiency, Peak Asymmetry and Resolution

Theoretical plate number, N, also referred as efficiency, is a measure of how broad a solute band can spread out for a given time it spends in a column. It is given by

Equation 1.4.

2 푡 2 푡 푁 = 16 ( 푅 ) = 5.54 ( 푅 ) (1.4) 푤푏 푤1 ⁄2 where wb is the baseline peak width and w1/2 is the peak width at half maximum height.

Both expressions assume a Gaussian shaped chromatographic band. Plate height (H) can also be used to evaluate the chromatographic efficiency when columns with different lengths (L) are compared.

퐿 퐻 = (1.5) 푁

Therefore, large N values and small H values indicate narrow peaks and high chromatographic efficiency.

The shape of the peak is another important factor in a chromatographic separation. Ideally, all peaks are expected to elute with a Gaussian profile; however, the reality can be quite different for many chromatographic peaks. The Asymmetry factor,

As, indicates the extent of peak deviation from a Gaussian shape. It is measured by comparing b the distance from the peak midpoint to the trailing edge of the peak at 10% of the peak height and a is the distance from the peak midpoint to the front edge of the peak at 10% of the peak height (Figure 1.2A):

푏 퐴 = (1.6) 푠 푎

5 when As is equal to 1, the peak is symmetrical or Gaussian; when As is greater than 1, the peak is has tailing; when As is less than 1, the peak has fronting. Peak asymmetry could arise from many factors such as extra-column dead volume, column secondary interactions, slow mass transfer processes, incomplete resolution of sample components, buildup contamination on column or column packing voids.10,11,12

The degree of separation between two peaks is defined by resolution Rs, which takes into account the separation of the two peak maxima (∆ t) and the widths of the peaks (Figure 1.2B and Equation 1.7).

2∆푡 1.18∆푡 푅푠 = = (1.7) (푤푏1 + 푤푏2) (푤ℎ1 + 푤ℎ2) where wb1 and wb2 are the peak widths at base, and wh1 and wh2 are the peak width at half height. For overlapping peaks, calculation using peak width at half height can be used instead of peak weak at base. Assuming the chromatographic bands are Gaussian, a Rs of

1.5 is required for two adjacent peaks to be fully separated at baseline. For two gaussian peaks have similar height and width, a Rs value of 1 indicates a valley separation of about

95.4% and band overlap of 2.3%, which is the generally considered as an adequate goal for optimization.8

Figure 1.2 Illustration of (A) peak asymmetry; (B) resolution.

1.2.4 Peak Capacity

The separation power of a chromatographic system can be characterized by its peak capacity, which describes the maximum number of peaks that can be separated with a given set of conditions and time interval with a certain resolution. For Rs of 1, for isocratic elution, which employs a mobile phase of fixed composition throughout the entire separation, the peak capacity (n) can be given by Equation 1.8.

√푁 푡 푛 = 1 + 푙푛 푅 (1.8) 4 푡푀 where N is the efficiency, tR is the retention time, and tM is the dead time. For gradient elution, in which the composition of mobile phase varies during the separation, the peak capacity (n) can be calculated by Equation 1.9.

푡푔 푛 = 1 + (1.9) 푤푏 where tg is the gradient time and wb is the peak width at base.

1.3 Top Down Proteomics

Mass spectrometry (MS) in combination with a variety of separation methods is the principle approach for proteomics.13 The two fundamental strategies to identify and characterize proteins using MS are the “bottom up” and “top down” proteomics. In bottom up approach (also known as “shotgun proteomics”), proteolytic digestion is directly utilized to cleave intact protein mixtures into peptides, which are then separated by single or multi-dimensional separation methods. The separated peptides are then analyzed by MS, and the proteins can be identified from the generated tandem mass spectra by matching the database (Figure 1.3A). The well-developed separation methods for peptides such as reversed-phase liquid chromatography (RPLC) and the high sensitivity of peptide signals on common mass spectrometers make bottom up proteomics the most mature and widely-used approach for protein identification and characterization in the last two decades.14 There are some limitations associated with bottom up proteomics. Only a small and variable fraction of total peptide population of a protein can be recovered in the bottom up approach leading to a limited protein sequence coverage, which means a significant amount of information about labile post translational modifications (PTMs) or sequence variants may be lost. Also, a peptide or several peptides may not be specific to an individual protein which can cause mapping ambiguity.15 These problems can be overcome by using the top down approach, which allows the proteins to be separated and MS analyzed in their intact form without enzymatic digestion (Figure 1.3B). This ensures the full sequence coverage of a protein

9 and the accurate localization and characterization of any PTMs.16 Despite many advantages top down proteomics offers, it is still a relative new strategy and faces many technical difficulties before it can be considered as a routine approach for proteomics.

One of the major obstacles in top down proteomics is the protein fractionation step because the separation of intact proteins is inherently more difficult compared to smaller peptides due to their large MW and structural complexity. Also, due to the complicated

ESI spectra of intact proteins resulting from multiple charges, mass analyzers with very high accuracy such as Fourier transform ion cyclotron resonance (FT-ICR) are often required.17 Moreover, there is a lack of suitable software for reconstructing protein identity for top down analysis.16 In the last decade, multidisciplinary efforts have been made to address the challenges of top down proteomics from the front-end separation techniques,18,19,20 advanced MS instrumentations21 to back-end computational tools.22,23

With the continuously development of techniques, top down analysis is hoped to become a more widespread strategy for proteomics.

A B

Figure 1.3 Workflows of (A) bottom up proteomics; (B) top down proteomics.

1.4 Separation Mechanisms for Intact Proteins in Liquid Chromatography

Because of the enormous complexity of proteomic samples with differences in protein concentrations (e.g. 0.03-5000 mg/dL in human plasma),24 effective separations are critical for reducing sample complexity and improving the dynamic range of detection. The separation of intact proteins has always represented a challenge owing to the low solubility and poor recovery of proteins compared to peptides. The separation of intact proteins is commonly accomplished by two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) due to its exceptional peak capacity, which can resolve up to

10,000 proteins from large sets of complex mixtures.25 However, difficulties in extracting

26,27 intact proteins from the gel matrix result in poor recovery.

Liquid chromatography (LC) is considered as an indispensable technique for intact protein separation owing to its high speed, high resolving power and the possibility of online coupling with ESI-MS.28 LC separation depends on the distribution of solute between the liquid mobile phase and the station phase. The current LC methods for intact proteins employ various retention mechanisms including reversed-phase chromatography

(RPC), hydrophobic interaction chromatography (HIC), ion-exchange chromatography

(IEX), size exclusion chromatography (SEC) and affinity chromatography (AC).

1.4.1 Reversed-phase chromatography

Separation of intact proteins by RPC is based on the reversible adsorption and desorption of proteins with various degrees of hydrophobicity to the hydrophobic

12 stationary phase. RPC is well-established and the method of choice for peptide separation.29 However, compared to low MW molecules and peptides, RPC for intact protein is always more problematic. Irreversible adsorption of proteins on the stationary phase with strong hydrophobic character such as long alkyl chain phase C18 has been reported repeatedly.30,31 Moreover, due to the secondary interactions between charged groups on the protein and the residual silanol groups on the silica sorbent, peak tailing is often noticed as the kinetics of secondary ionic interactions are slower than that of

32,33 hydrophobic interactions. In addition, high MW proteins have larger radii and lower diffusion coefficients than small molecules and peptides, and this may cause peak broadening when conventional porous packing materials with 80-120 Å pore sizes are used.33 In the last two decades, significant efforts have been made in improving column and packing designs, and the mobile phase properties to address the issues associated with RPC for intact proteins.34 For example, shorter alkyl chain length like C4 is adapted preferably for intact proteins to facilitate the recovery.35 Porous layer open tubular

(PLOT) and sub-2 µm particles have been employed to improve the kinetic performance and minimize peak broadening effects.36,37 Ion pairing reagents or elevated temperature of mobile phases have also been used to improve the peak shapes.38,39 Because the mobile phases used in RPC are generally water and organic solvents such as methanol and acetonitrile, which are highly compatible with ESI-MS. The straightforward coupling to

MS makes RPC increasingly important for intact protein analysis.

1.4.2 Hydrophobic Interaction Chromatography

HIC separates proteins according to their surface hydrophobicity. Although the retentions in both RPC and HIC rely on the hydrophobic interaction between proteins and the stationary phases, the conditions used in HIC are much milder than those in RPC, which allows the biological activity of proteins to be maintained.40 Therefore, HIC is particularly attractive for intact protein separation. Basically, the hydrophobic ligands are weaker (alkyl chains or aryl groups) with smaller density of substitution on the HIC media compared to RPC. Instead of using harsh mobile phase conditions like extreme pH or large proportions of organic solvents, the interaction between hydrophobic proteins and the HIC ligands can be controlled by the presence of salts in the buffered mobile phase.

A high salt concentration in the mobile phase promotes the stationary phase ligand-protein interactions and results in protein retention on the stationary phase.

Proteins with minimum or no surface hydrophobicity are eluted with the high salt mobile phase. The adsorbed proteins can be eluted by stepwise or gradient elution with decreasing salt concentration (Figure 1.4). The ability of a salt to promote the ligand- protein interaction depends on the type of the ionic species in the salt and the salt concentration. The ability of an ion to precipitate proteins from aqueous solution follows the Hofmeister series (Figure 1.5). Salts at the beginning of the series with a strong salting out effect are called kosmotropic (order-making) agents, and they are considered to stabilize the protein structure; whereas salts at the end of the series with salting in

14 effect are called chaotropic (order-breaking) agents, which disrupt the structure of proteins and decrease the hydrophobic interactions.41 In practice, sodium, potassium and ammonium sulfates are the commonly-used salts to produce relatively high precipitation of proteins in HIC separations.40 Unfortunately, these salts are incompatible with ESI-

MS. Therefore, the confirmation of HIC peak identities typically requires offline fraction collection followed by desalting or buffer exchange steps and subsequent MS analysis.

Figure 1.4 Retention and elution steps in a HIC separation.

Increasing salting out effect

3- 2 ------Anions: PO4 SO4 -, CH3COO , Cl , Br , NO3 . ClO4 , I , SCN

+ + + + + 2+, 2+, 2+, 2+ Cations: NH4 , Rb , K , Na , Cs , Li Mg Ca Ba

Increasing salting in effect

Figure 1.5 Hofmeister series.

1.4.3 Ion Exchange Chromatography

In IEX, the stationary phase is charged to provide retention to analytes with opposite charges, and the elution is realized by exchange of analyte ions with the counterions in the mobile phase or by altering the pH of the mobile phase. IEX is extensively used in intact protein separation because biomolecules like proteins generally involve many charged groups on their surfaces, which change with the pH of the solution.

In cation exchange chromatography (CEX), the stationary phase carries negatively charged groups which attract positively charged proteins at acidic pH; while in anion exchange chromatography (AEX), the stationary phase carries positively charged groups which bind to negatively charged protein at basic pH. Depending on the type of charged groups on the stationary phase, CEX and AEX can be further divided into strong cation exchange (SCX) and anion exchange (SAX), and weak cation exchange (WCX) and anion exchange (WAX). Strong ion exchangers such as sulfonate groups and quaternary amines maintain their charged form independent of the mobile phase pH, while weak ion exchangers like carboxylic acids and tertiary amines are very sensitive to pH change.42

The elution conditions in IEX allow proteins to be characterized in their nondenaturing conformation; on the other hand, desalting or buffer exchange is required because of the presence of nonvolatile salts in the mobile phase. The selectivity of IEX remains a challenge and multi-dimensional separation is often required for complex mixtures.43

1.4.4 Size Exclusion Chromatography

Unlike RPC, HIC or IEX, which all achieve separation by taking advantage of the ability of molecules participate in intermolecular interactions (e.g. hydrophobic interaction or electrostatic interaction), SEC is a noninteractive method to separate macromolecules and polymers according to their hydrodynamic volume. The packing material in SEC column consists of various sized pores. Very large molecules which cannot penetrate many of the pores will elute from the column first, whereas smaller molecules which can diffuse further into the pores and flow through both the interparticle space and the pore volume will exit the column last. SEC is routinely used for intact proteins separations due to its simplicity and reproducibility; however, the separation capacity of this method is limited.44 As a rule of thumb, sample molecules need have differences in MW by ~10% to be completely resolved by SEC.42

1.4.5 Affinity Chromatography

AC is based on the specific binding of an analyte molecule to a ligand immobilized on the stationary phase. The affinity ligands are covalently bonded to the packing material via a spacer arm. When the sample is passed through the column, only the molecule selectively binds to the ligand is retained, and all other sample components pass through the column unretained. The adsorbed molecule can be eluted either by competitive displacement or by changing the conformation of the molecule through a change in mobile phase conditions such as pH or ionic strength. Because of the tremendous specificity of AC, it is most suitable for target protein purification with high 19 yield.42 The main drawbacks of AC are the relatively low capacity, high costs of the immobilized ligands, and limited lifetime of the packing materials.45

1.5 Gold Nanoparticles

In the past two decades, significant progress has been made in the area of nanomaterials to solve problems in healthcare,46 energy,47 electronics48 and many other areas. One of the most promising and exciting applications of nanomaterials is the design of engineered nanoparticles to address biomedical challenges. Among the different types of nanoparticles, gold nanoparticles (AuNPs) have attracted tremendous attention due to its unique optical, electronic and thermal properties.49 AuNPs prove to be an important tool for a wide range of biomedical applications such as drug delivery,50 imaging,51 cancer therapeutics,52 chemical and biological sensing.53 The size of AuNPs is the key parameter as slight size variation can have significant impact on the cellular uptake and biological processes of AuNPs.54 Therefore, an accurate characterization of the AuNP size is essential. Moreover, one of the primary goals outlined in “Nanotechnology

Research Directions for Societal Needs in 2020” is to develop separation and fractionation methods to realize a library of nanomaterials with monodispersity in composition, size and shape.55

1.5.1 PEGylation

In 1977, Davis and Abuschowski reported for the first time that by covalently attaching polyethylene glycol (PEG) to bovine serum albumin and liver catalase proteins, the protein therapeutics showed extended systemic circulation time and decreased immunogenicity without compromising activity.56 Since then, PEGylation, the process of covalent or noncovalent attachment of PEG to molecules and macrostructures, has become a fast-growing technology in drug delivery systems. To date, there are more than ten FDA-approved PEGylated protein therapeutics on the market, and many new products are continuously being developed and in different stages of clinical trials.57

The success of PEGylated proteins has led to the investigations of PEGylation on

AuNPs. Recognized as foreign objects, AuNPs are readily shuttled out from the systematic circulation to liver, spleen or bone marrow causing nonspecific binding of

AuNPs to nontargeted areas.58 Moreover, aggregation of bare AuNPs can lead to capillary occlusion and subsequent organ damage.59 PEGylation of AuNPs can overcome these challenges. Similar to the PEGylated proteins, the PEG coating on AuNPs can reduce cellular uptakes and increase the circulation time.60 Also, the protective coating also prevents AuNPs from aggregation and association with nontargeted serum and tissue proteins.61 PEGylation can effectively increase the hydrodynamic size of AuNPs, therefore, it becomes increasingly important to develop rapid and accurate methods for the separation and characterization of PEGylated AuNPs with different core sizes and

PEG lengths.

1.5.2 Size Separation and Characterization Methods for AuNPs

Size characterization of AuNPs is traditionally performed using scanning/ transmission electron microscopy (SEM/TEM) and dynamic light scattering (DLS). Even when samples do not contain complex matrix, SEM and TEM require tedious sample preparation and long measurement time for a large number of particles to afford a representative population. DLS measures the Brownian motion, which is the random movement of particles due to bombardment by the surrounding solvent molecules, and the velocity of the Brownian motion, also known as the translational diffusion coefficient

(D), is related to the size of the particles according to the Stoke-Einstein Equation:

푘푇 푑(퐻) = (1.10) 3휋휂퐷 where d(H) is the hydrodynamic diameter, k is the Boltzmann’s constant, T is the absolute temperature, η is the viscosity.62 DLS is limited by its low resolution, which cannot precisely measure a polydisperse sample. Large particle present in even small quantities can be take into account during data analysis and lead to skewed results.63

Thus, separation step is necessary prior to measurement for polydisperse samples.

Separation techniques including SEC, hydrodynamic chromatography (HDC), field-flow fractionation (FFF), capillary electrophoresis (CE), and gel electrophoresis

(GE), have been used to evaluate the size of AuNPs. SEC is a well-established method for separation of macromolecules according to their size, as mentioned earlier.

Nonetheless, this method suffers from low resolution and irreversible adsorption of NPs, which may be minimized by addition of detergent in the mobile phase.64 In HDC, 22 particles are separated in a packed or open tubular capillary column in the order of decreasing size, as in SEC. Under laminar (parabolic) flow profile, large particles preferentially remain near the center of the capillary, where the flow rate is maximum, whereas smaller particles stay closer to capillary wall, experiencing slower flow rate.65

The main advantage of HDC is the rapid and convenient analysis, but this technique is intrinsically limited by low resolution and small peak capacity. FFF was first introduced by J. Calvin Giddings in 1966.66 The separation is carried out by applying an external field perpendicular to the direction of sample flow (also in a parabolic flow profile) through an empty and narrow ribbon-like channel. FFF is a considered as a powerful tool due to the fact that it can separate materials over a wide size range from 1 nm to 10 µm while maintaining high resolution. Currently, FFF still lacks the maturity of LC such as the number of instruments and published methods.67 CE separates charged molecules based on their electrophoretic mobility in an electroosmotic flow of an electrolyte solution under a high voltage. CE offers high speed of analysis and high separation resolution but suffers from limited sample loading capacity, poor reproducibility and low sensitivity with UV detection.68 In GE, an electric field is applied to drive the molecules through the agarose or polyacrylamide gel, and molecules are separated according to their size and charge. The distinct color of AuNPs enable direct observation of the sample in the gel; however, GE often requires a large sample volume and the separation can be time consuming.

1.6 Ultrathin Layer Chromatography and Electrospinning

Ultrathin layer chromatography (UTLC) was introduced by Hauck et al. in 2001 using silica monolith as the stationary phase.69 UTLC is a more efficient method compared to the conventional high-performance thin layer chromatography (HPTLC) owing to its layer thickness and the mobile phase migration distance. The layer thickness of HPTLC is generally 100-250 µm, whereas the layer thickness of UTLC is only about

10 µm. The mobile phase migration distance on the HPTLC is in the range of 8-10 cm; instead, only 1-3 cm of migration distance is required in UTLC to achieve a good separation. Therefore, UTLC offers much fast analysis but the silica monolith stationary phase did not significantly improve the chromatographic efficiency.69

Electrospun nanofibers have been applied as the stationary phase of UTLC by the

Olesik group since 2009.70 This nano-scaled material has shown promising separation efficiency for small molecules like laser dyes, steroidal compounds and amino acids.70,71

The fabrication of electrospun nanofibers can be carried out by a simple and fast technique called electrospinning. The apparatus of the electrospinning consists a high voltage power supply, a syringe pump and a grounded collector (Figure 1.6). During the electrospinning process, a polymer solution is injected at a controlled flow rate through a syringe needle which is charged to a high voltage (typically 10 to 30 kV). When the sufficient voltage is achieved, the surface tension of the liquid droplet at the needle tip is overcome by the electrostatic repulsion to form a Taylor cone. A charged liquid jet is

24 then stretched from the Taylor cone and travels towards the collector via a whipping process, where solvent evaporation and fiber elongation take place.

Electrospun nanofibers are particularly valuable as the stationary phase in UTLC.

Not only the electrospinning process is simple and cost effective but also various polymers can be electrospun into nanofibers according to desired functionalities on the stationary phase. With such versatility, many separation mechanisms can be realized on the UTLC devices. Moreover, electrospun nanofibers deposit on the collector with very high surface area, thus no binder is required to attach the fiber mat to the substrate. Most importantly, the fibers in nanoscale with high uniformity are expected to achieve exceptional peak efficiency. The small device utilizing electrospun nanofiber UTLC has great potential to separate macrostructures like intact proteins and PEGylated AuNPs.

Taylor Cone

Figure 1.6 Electrospinning apparatus.

1.7 Enhanced Fluidity Liquid Chromatography

Enhanced fluidity liquid chromatography (EFLC) involves the addition of liquefied gas (most commonly, carbon dioxide) to the conventional liquid mobile phase.

In supercritical liquid chromatography (SFC), CO2 is used as the primary mobile phase with only small amount of organic solvent (e.g. methanol or ethanol) as the additive, making SFC limited to relatively nonpolar compounds. In contrast, EFLC employed the liquefied CO2 as the additive to organic solvents making it applicable to a broader range of analyte polarity. Compared to conventional LC, the mobile phase used in enhanced fluidity liquids have higher diffusivity and lower viscosity due to the addition of liquefied

CO2. The increased diffusivity can improve the mass transfer inherently leading to more efficient separation. In addition, the lower solvent viscosity can lower the system back pressure, which allows faster flow rate to be performed. Therefore, EFLC is ideal for fast and high efficient separation for a wide range of compounds.

Previously, EFLC was successfully applied in several separation mechanisms including RPC,72 SEC,73 IEX,74 normal-phase,75 chiral76 and hydrophilic liquid chromatography (HILIC).77,78 With its striking features, EFLC can potentially address some of the challenges in intact protein separation.

1.8 Research Focus

This dissertation describes the development of novel nanofiber-based stationary phase and innovative solvent system for the separation of intact proteins and PEGylated

AuNPs. Chapter 2 evaluates an ion exchange stationary phase using electrospun Nafion- polyacrylonitrile nanofibers for amino acids and intact proteins. Design of experiment methods are described in detail for the optimization of electrospinning parameters and chromatographic conditions. Chapter 3 describes the study of using electrospun nanofibrous UTLC and a micellar mobile phase system for size separation and characterization of PEGylated AuNPs. Chapter 4 describes the advancements of the integrate EFLC-MS system using a HIC column for intact protein separation and characterization. The significant impact of liquefied CO2 on both the chromatographic and mass spectroscopic performances is examined. Chapter 5 continues to assess the

EFLC mobile phase as an alternative to traditional RPC mobile phase for intact protein separation. Together, this work provides new insights into intact protein separation for top down proteomics and macrostructure characterization.

1.9 References

1 Blueggel, M.; Chamrad, D; Meyer, H. E. Curr Pharm Biotechnol. 2004, 5, 79-88.

2 Lee, J.; Kohn, E. C. Ann. Oncol. 2010, 21, 205-210.

3 Rifai, N.; Gillette, M. A.; Carr, S. A. Nature Biotechnol. 2006, 24, 971-983.

4 Smith, L. M.; Kelleher, N. L. Nat Methods. 2013, 10, 186–187.

5 Dykman, L.; Khlebstov, N. Chem. Soc. Rev. 2012, 41, 2256-2282.

6 Huang, X.; Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Nanomedicine 2007, 2, 681- 693.

7 Yeh, Y.; Creran, B.; Rotello, V. M. Nanoscale 2012, 4, 1871-1880.

8 Poole, C.F. The Essence of Chromatography; Elsevier: New York, 2003.

9 Spangenberg, B.; Poole, C. F.; Weins, Ch. Quantitative Thin-Layer Chromatography, A Practical Survey; Springer: New York, 2011.

10 Gotmar, G.; Fornstedt, T.; Guiochon, G. J. Chromatogr. A 1999, 831, 17-35.

11 Conder, J. R. J. High Resolut. Chromatogr. 1982, 5, 341-348 and 397-403.

12 Castillo, A.; Roig-Navarro, A. F.; Pozo, O. J. J. Chromatogr. A 2007, 1172, 179-185.

13 Capriotti, A. L.; Cavaliere, C.; Foglia, P.; Samperi, R.; Lagana, A. J. Chromatogr. A 2011, 1218, 8760-8776.

14 Yates, J. R.; Ruse, C. I.; Nakorchevsky, A. Annu. Rev. Biomed. Eng. 2009, 11, 49-79.

15 Zhang, Y.; Fonslow, B. R.; Shan, B.; Baek, M.; Yates, J. R. Chem Rev. 2013, 113, 2343–2394.

16 Kellie, J. F.; Tran, J. C.; Lee, J. E.; Ahlf, D. R.; Thomas, H. M.; Ntai, I.; Catherman, A. D.; Durbin, K. R.; Zamdborg, L.; Vellaichamy, A.; Thomas, P. M.; Kelleher, N. L. Mol. BioSyst. 2010, 6, 1532–1539.

17 Han, X.; Jin, M.; Breuker, K.; McLafferty, F. W. Science 2006, 314, 109-112.

18 Lee, J. E.; Kellie, J. F.; Tran, J. C.; Tipton, J. D.; Catherman, A. D.; Thomas, H. M.; Ahlf, D. R.; Durbin, K. R.; Vellaichamy, A.; Ntai, I.; Marshall, A. G.; Kelleher, N. L. J. Am. Soc. Mass Spectrom. 2009, 20, 2183–2191.

19 Gargano, A. F. G.; Roca, L. S.; Feller, R. T.; Bocxe, M.; Dominguez-Vega, E.; Somsen, G. W. Anal. Chem. 2018, 90, 6601−6609.

20 Xiu, L.; Valeja, S. G.; Alpert, A. J.; Jin, S.; Ge, Y. Anal. Chem. 2014, 86, 7899−7906.

21 Parks, B. A.; Jiang, L.; Thomas, P. M.; Wenger, C. D.; Roth, M. J.; Boyne, M. T., 2nd; Burke, P. V.; Kwast, K. E.; Kelleher, N. L. Anal. Chem. 2007, 79, 7984–7991.

22 Leduc, R. D.; Kelleher, N. L. Current Protocols in Bioinformatics; Wiley: New York, 2007; Chapter 13, Unit 13.16.

23 Cai, W.; Guner, H.; Gregorich, Z. R.; Chen, A. J.; Guner, S. A.; Peng, Y.; Valeja, S. G.; Liu, X.; Ge, Y. Mol. Cell. Proteomics 2016, 15, 703-714.

24 Rossi, D.; Fangazio, M.; Paoli, L. D.; Puma, A. P.; Riccomagno, P.; Pinto, V.; Zigrossi, P.; Ramponi, A.; Monga, G.; Gaidano, G. Cancer, 2010, 116, 2188-2200.

25 Wittmann-Liebold, B.; Graack, H. R.; Pohl, T. Proteomics 2006, 6, 4688-703.

26 Rabilloud T. Proteomics 2002, 2, 3–10.

27 Gygi, S. P.; Corthals, G. L.; Zhang, Y.; Rochon, Y.; Aebersold, R. Proc. Natl. Acad. Sci. 2000, 97, 9390-9395.

28 Xie, F.; Smith, R. D.; Shen, Y. J. Chromatogr. A 2012, 1261, 78-90.

29 Peterson, A.; Hohmann, L.; Huang, L.; Kim, B.; Eng, J. K.; Martin, D. B. J. Proteome Res. 2009, 8, 4161–4168.

30 Eschelbach, J. W.; Jorgenson, J. W. Anal. Chem. 2006, 78, 1697-1706.

31 Bobaly, B.; Mikola, V.; Sipko, E.; Marta, Z.; Fekete, J. J. Chromatogr. Sci. 2015, 53, 1078-1083.

32 Stella, C.; Rudaz, S.; Veuthey, J. L.; Tchapla, A. Chromatographia 2001, 53, 113–131.

33 Staub, A.; Guillarme, D.; Schappler, J.; Veuthey, J.; Rudaz, S. J. Pharm. Biomed. Anal. 2011, 55, 810-822.

34 Svec, F.; Tennikova, T. B.; Deyl, Z. Monolithic Materials: Preparation, Properties and Applications; Elsevier: Amsterdam, 2003. 30

35 Hamada, T.; Tanaka, H.; Izumine, H.; Ohira, M. J. Chromatogr. A 2004, 1043, 27–32.

36 Rogeberg, M.; Wilson, S. R.; Greibrokk, T.; Lundanes, E. J. Chromatogr. A 2010, 1217, 2782–2786.

37 Everley, R. A.; Croley, T. R. J. Chromatogr. A 2008, 1192, 239–247.

38 Chakraborty, A. B.; Berger, S. J. J. Biomol. Tech. 2005, 16, 327-335.

39 Blackler, A. R.; Speer,s A. E.; Wu, C. C. Proteomics 2008, 8, 3956-3964.

40 Queiroz, J. A.; Tomaz, C. T.; Cabral, J. M. S. J. Biotechnol. 2001, 87, 143-159.

41 Russo, D. Chem. Phys. 2008, 245, 200-211.

42 Vitha, M. F. Chromatography Principles and Instrumentation; John Wiley & Sons: NJ, USA, 2017.

43 Mitulovic, G.; Mechtler, K. Brief. Funct. Genomics. 2006, 5, 249-260.

44 Hong, P.; Koza, S.; Bouvier, E. S. P. J. Liq. Chromatogr. Relat. Technol. 2012, 35, 2923-2950.

45 Goding, J. W. Monoclonal Antibodies: Principle and Practice, 3rd Ed.; Elsevier: New York, 1996.

46 Boisseau, P.; Loubaton, B. C. R. Physique 2011, 12, 620–636.

47 Zhang, Q.; Uchaker, E.; Candelaria, S. L.; Cao, G. Chem. Soc. Rev. 2013, 42, 3127- 3171.

48 Jiao, Y.; Du, A.; Hankel, M.; Zhu, Z.; Rudolph, V.; Smith, S. C. Chem. Commun. 2011, 47, 11843-11845.

49 Elahi, N.; Kamali, M.; Baghersad, M. H. Talanta 2018, 184, 537–556.

50 Ghosh, P.; Han, G.; De, M.; Kim, C. K.; Rotello, V. M. Adv. Drug Deliv. Rev. 2008, 60, 1307-1315.

51 Popovtzer, R.; Agrawal, A.; Kotov, N. A.; Popovtzer, A.; Balter, J.; Carey, T. E.; Kopelman, R. Nano Lett. 2008, 8, 4593–4596.

52 Zhao J.; Lee, P.; Wallace, M. J.; Melancon, M. P. Curr. Pharm. Des. 2015, 21, 4240- 51.

53 Saha, K.; Agasti, S. S.; Kim, C.; Li, X.; Rotello, V. M. Chem. Rev. 2012, 112, 2739– 2779.

54 He, C.; Hu, Y.; Yin, L.; Tang, C.; Yin, C. Biomaterials 2010, 31, 3657–3666.

55 Roco, M. C.; Mirkin, C. A.; Hersam, M. C. Nanotechnology Research Directions for Societal Needs; Springer: Dordrecht, The Netherlands, 2011.

56 Abuchowski, A.; McCoy, J. R.; Palczuk, N. C.; van Es, T.; Davis, F. F. J. Biol. Chem. 1977, 252, 3582-3586.

57 Suk, J. S.; Xu, Q.; Kim, N.; Hanes, J.; Ensign, L. M. Adv Drug Deliv Rev. 2016, 99, 28-51.

58 Zhang, X.; Wu, H.; Wu, D.; Wang, Y.; Chang, J.; Zhai, Z.; Meng, A.; Liu, P.; Zhang, L.; Fan, F. Int. J. Nanomedicine 2010, 5, 771–781.

59 Thakor, A. S.; Jokerst, J.; Zaveleta, C.; Massoud, T. F.; Gambhir, S. S. Nano Lett. 2011, 11, 4029-4036.

60 van Vlerken, L. E.; Vyas, T. K.; Amiji, M. M. Pharm. Res. 2007, 24, 1405–1414.

61 Jokerst, J. V.; Lobovkina, T.; Zare, R. N.; Gambhir, S. S. Nanomedicine 2011, 6, 715– 728.

62 Einstein, A. Investigation on the Theory of the Brownian Movement; Dover: New York, 1926.

63 Filipe, V.; Hawe, A.; Jiskoot, W. Pharm. Res. 2010. 27, 796-810.

64 Wei, G.; Liu, F.; Wang, C. R. C. Anal. Chem. 1999, 71, 2085-2091.

65 Striegel, A. M.; Brewer, A. K. Annu. Rev. Anal. Chem. 2012, 5, 15–34.

66 Giddings, J. C. J. Sep. Sci. 1966, 1, 123–125.

67 Messaud, F. A.; Sanderson, R. D.; Runyon, J. R.; Otte, T.; Pasch, H.; Williams, S. K. R. Prog. Polym. Sci. 2009, 34, 351-368.

68 Huang, Y. F.; Huang, C. C.; Hu, C. C.; Chang, H. T. Electrophoresis 2006, 27, 3503- 3522.

69 Hauck, H. E.; Bund, O.; Fischer, W.; Schulz, M. J. Planar Chromatogr. 2001, 14, 234.

70 Clark, J. E.; Olesik, S. V. Anal. Chem. 2009, 81, 4121–4129. 32

71 Lu, T.; Olesik, S. V. J. Chromatogr. A 2013, 912, 98-104.

72 Cui, Y.; Olesik, S. V. J. Chromatogr. A 1995, 691, 151-162.

73 Yuan, H.; Olesik, S. V. J. Chromatogr. A 1997, 785, 35-48.

74 Beres, M. J.; Olesik, S. V. J. Sep. Sci. 2015, 38, 3119-3129.

75 Lee, S. T.; Olesik, S. V. J. Chromatogr. A 1995, 707, 217-224.

76 Sun, Q.; Olesik, S. V. Anal. Chem. 1999, 97, 2139-2145.

77 Treadway, J. W.; Philibert, G. S.; Olesik, S. V. J. Chromatogr. A 2011, 1218, 5897- 5902.

78 Beilke, M. C.; Beres, M. J.; Olesik, S. V. J. Chromatogr. A 2016, 1436, 84-90.

Chapter 2. Electrospun Nafion-Polyacrylonitrile Nanofibers as an Ion Exchange

Ultrathin Layer Chromatographic Stationary Phase

2.1 Introduction

Thin layer chromatography (TLC) is a simple chromatographic method that involves the separation of multiple samples and standards on an open layer developed by a mobile phase.1 It is used in many areas including synthetic chemistry,2 food science,3 pharmaceutical industry4 as well as clinical research5 due to its simplicity, rapidness and low cost. High performance thin layer chromatography (HPTLC), an extension of TLC, was developed in 1970s using smaller diameter particles (5-20 µm).6 HPTLC offers better chromatographic efficiency, faster separation, lower analyte and mobile phase consumption and lower detection limits.7 In order to further improve the chromatographic performance and reduce analysis time and the amount of consumables, ultrathin layer chromatography (UTLC) was introduced in 2001 using monolithic silica as the stationary phase. Compared to classic TLC with thicknesses of 100-400 µm, UTLC utilizes sorbent layers as thin as 5-25 µm, approximately.8 There are many materials that have been developed as stationary phases for TLC, including the most commonly-used

34 silica gel and modified silica gel, and less frequently used aluminum oxide, cellulose, polyamides and ion exchange resins.9 Recently, our group has reported electrospun nanofibers as the stationary phases for UTLC. Various polymers can be electrospun to produce nanofibers. Polyacrylonitrile (PAN),10 glassy carbon,11 polyvinyl alcohol

(PVA),12 silica,13 carbon nanorod-filled polyacrylonitrile14 were successfully fabricated as stationary phases for UTLC using the electrospinning method. All of these electrospun

UTLC plates showed enhanced separation efficiency, decreased use of solvents and increased speed of analysis compared to commercial HPTLC plates.

Nafion is a synthetic perfluorinated cationic polymer developed by Dupont in the

1970s.15 As shown in Figure 2.1, a hydrophobic tetrafluoroethylene (Teflon) backbone makes Nafion highly chemical resistant, while sulfonate groups on the side chains provides proton conductivity. Scaling down Nafion into the nanometer scale via electrospinning can further improve its proton conductivity.16 During the electrospinning process, the shear force elongates fibers and orients ionic domains along the fiber axis direction, and the aligned ionic structures result in higher conductivity.16, 17

Unfortunately, the pure Nafion polymers or polymer solutions have a low shear viscosity that does not allow electrospinning of the polymer.18 Therefore, another polymer is needed to blend with Nafion for electrospinning to take place. Several researchers have successfully blended Nafion with carrier polymers, such as poly (acrylic acid) (PAA),17,19 poly (vinyl) alcohol (PVA),20,21 poly (ethylene oxide) (PEO)15,22,23,24 and polyacrylonitrile

(PAN).25,26 In our group, PAN and multi-walled carbon nanotubes (MWNT) filled PAN have been previously electrospun as UTLC stationary phases and showed substantially

35 improved chromatographic performance.10,14 Thus, in this work, PAN was chosen as the carrier polymer to electrospin with Nafion to create ion exchange UTLC station phases.

Separations of amino acids and proteins are essential in food science,27 agricultural science28 and pharmaceutical industry.29 Ion exchange chromatography (IEX) is one of the most popular chromatographic techniques for separating amino acids and proteins owing to its high capacity, high resolving power and easy controllability of the separation process.30 In practice, IEX for amino acids and proteins is more often conducted in column chromatography, which exhibits disadvantages in costly equipment, time consuming experiments and difficulties in detection.31 Employing the ion exchange technique on UTLC eliminates these requirements for time and expensive instruments.

Also, the detection in UTLC can be realized by a simple spray reaction, by absorption of ultraviolet light, by using fluorescent labeling, or even direct observation in the case of colored analytes.32 Electrospun Nafion-PAN nanofibers have particularly attractive properties as the ion exchange UTLC stationary phase, including excellent chemical stability and easy accessibility of ion exchange sites which allow sufficient interaction between analytes and stationary phase.

In this dissertation, we report the first cation exchange electrospun UTLC method and an evaluation of its chromatographic performance by separating amino acids and intact proteins. Fractional factorial design and response surface methodology were used to optimize the Nafion-PAN stationary phase and separation conditions. The chemical stability of the electrospoun nanofiber mat was also investigated.

Figure 2.1 Molecular structure of Nafion. For 1100 equivalent weight Nafion used in this study, x= 6.5 (average value) and y= 1.

2.2 Materials and Methods

2.2.1 Materials

Nafion containing solution LIQUION 1115 (15% by weight NAFION®, 1100 equivalent weight) was purchased from Ion Power Inc. (New Castle, DE).

Polyacrylonitrile (PAN), average Mw 150,000 g mol-1, was purchased from Sigma-

Aldrich (St. Louis, MO). N, N-Dimethylformamide (DMF) (99.8%), HPLC grade methanol (99.9%), acetonitrile (99.9%), 2-propanol (99.9%) and methylene chloride

(99.9%) were acquired from Fisher Scientific (Fair Lawn, NJ). Ethanol (91%) was purchased from Decon Labs Inc. (King of Prussia, PA). Buffer reagents MES hydrate

(minimum 99.5%) and ammonium formate (≥ 99%) were purchased from Sigma-Aldrich; sodium chloride (100%) and ammonium hydroxide (certified ACS plus) were purchased from Fisher Scientific. Water was purified to 18.1 MΩ by a Barnstead Nanopure Infinity

System from Thermal Scientific Inc. (Odessa, TX).

The amino acids including arginine (Arg), lysine (Lys), histidine (His), valine

(Val), phenylalanine (Phe) and alanine (Ala); Arg, Lys and Ala were purchased from

Sigma-Aldrich, Val was purchased from Amresco (Solon, OH), His was purchased from

Matheson Coleman & Bell (Gardena, CA), and Phe was purchased from Eastman

Chemical Company (Kingsport, TN). Proteins included lysozyme (Lys) from chicken egg, bovine serum albumin (BSA), α-chymotrypsin (Chy) from bovine pancreas and myoglobin (MGB) from equine skeletal muscle. Lys, BSA and MGB were obtained from

Sigma-Aldrich, and Chy was purchased from Worthington Biochemical Corporation

(Lakewood, NJ). Visualization spray reagents ninhydrin (≥ 98%), fluorescamine (≥ 98%) and triethylamine (≥ 99.5%) were purchased from Sigma-Aldrich.

2.2.2 Instrumentation

The morphology of electrospun nanofiber mats was characterized using a FEI

Nova NanoSEM 400 (FEI Corporate, North America NanoPort, Hillsboro, OR) scanning electron microscope (SEM). Each sample was sputter coated with AuPd for 90s at 17 mA using a Cressington sputter coater (Cressington Scientific Instruments, Watford, UK) to create a conductive surface before SEM analysis. Fiber diameters were measured on

SEM images using ImageJ software (Available from the National Institute of Health at http://www.rsbweb.nih.gov/ij/index.html).

2.2.3 Preparation of Nafion-PAN Solution

Pure Nafion was obtained following the drying/curing procedure provided by Ion

Power Inc. Dried Nafion and PAN were dissolved in DMF and stirred for at least 24 h using a magnetic stirrer under gentle heating.

2.2.4 Electrospinning

39 control and monitor the relative humidity in the closed chamber while electrospinning.

Nanofibers were electrospun at room temperature with relative humidity below 20%. The distance between the collector and the tip of the needle was kept at 15 cm. The voltage was kept at 16 kV, and the flow rate was kept at 0.3 mL/h. Each nanofiber mat was electrospun for 10 min.

2.2.5 Ion Exchange Capacity

Ion exchange capacity (IEC) of electrospun Nafion-PAN nanofibers was determined by titration. Nafion-PAN nanofibers were immersed in 1.0 M NaCl solution for 48 hours to exchange Na+ ions with H+ ions in the Nafion. The solution was then titrated with 0.01M NaOH using phenolphthalein as indicator. The titrated IEC of the

Nafion-PAN nanofibers was determined by:

0.01×푉 퐼퐸퐶 = 푁푎푂퐻 (2.1) 푊푛푎푛표푓𝑖푏푒푟푠 where VNaOH is the titrated volume of NaOH and Wnanofibers is the weight of the electrospun Nafion-PAN nanofibers.

2.2.6 Ultrathin Layer Chromatography

After electrospun nanofiber mats were prepared, they were cut into 3 cm × 6 cm

UTLC plates. Fused silica capillary tubes with an inner diameter of 250 µm were used for spotting analytes onto the origin line drawn at 1 cm from the bottom of the plate. The volume of analytes spotted, which was approximately 500 nL, was calculated as the volume difference in the capillary tube before and after spotting. Mixtures of organic 40 solvents and aqueous buffers were required to prepare analyte solutions instead of using

100% aqueous buffers or water, otherwise analytes could not be spotted onto the plates due the hydrophobic nature of Nafion-PAN nanofibers. Amino acids (15 mM) were prepared in formate buffer (pH=3.0, 10mM)/ methanol (60:40, v/v). Proteins were prepared in MES buffer (pH=6.5, 10 mM)/ acetonitrile (80:20, v/v) with the concentration of 2 mg/mL. UTLC plates were then placed in a 250 mL developing chamber containing 5 mL of mobile phase. Prior to each development, the mobile phase was allowed to equilibrate for 10 min.

2.2.7 Visualization and Data Collection

After development, UTLC plates were dried at room temperature. For separations of amino acids, plates were sprayed with freshly prepared ninhydrin solution (0.3 g of ninhydrin dissolved in 100 mL of n-butanol with 3 mL of HOAc) using a TLC reagent nebulizer (Kimble-Chase Vineland, NJ), and then heated in the oven at 110 ºC for 10 min. For separations of proteins, plates were treated according to the fluorescamine spray procedure: spray with a solution of 10% triethylamine in methylene chloride and air dry for several seconds; spray with a solution of 0.05% fluorescamine in acetone and air dry for several seconds; re-spray with a solution of 10% triethylamine in methylene chloride.33 Proteins were visualized under UV radiation at λ= 365 nm. A Canon A650IS

12.1 MP digital camera was utilized to take photographs after separations, and the digital images were then analyzed using ImageJ 1.48v. Chromatograms were obtained from digital images using TLC analyzer (available at http://www.sciencebuddies.org/science-

41 research-papers/tlc_analyzer.shtml), and PeakFitTM (version 4, SPSS Inc.) was used to fit the curves into Gaussians. For quantification, UTLC plates were visualized using

Typhoon FLA 9500 (GE Healthcare), and the peak area was then measured using ImageJ

1.48v. The limit of detection (LOD) and limit of quantification (LOQ) values were determined on the basis of specific calibration curve using respective concentrations: 0.5,

1.0, 1.5, 2.0 and 3.0 µg/spot for amino acids; 1.0, 1.5, 2.0, 3.0 and 5.0 µg/spot for proteins.

2.2.8 Experimental Designs

To identify the most important factors that affect the separation performance, screening designs were carried out using fractional factorial designs for both amino acids and proteins. Factorial designs allow the levels of all factors to be varied at the same time instead of one at a time so that not only the effect of each factor on the response is investigated but also the effects of interactions between factors on the response can be examined. A full factorial design measures the responses at all possible combinations of the factor levels. A fractional factorial design only conducts a subset of the runs in a full factorial design to avoid high cost and time-consuming experiments. For example, the number of runs for a 2-level full factorial design containing k factors is 2k, whereas the number of runs for a 2-level factorial design containing k factors, where only a fraction

(p) of the full design is used, is 2k-p.34 Herein, five factors including Nafion concentration, mobile phase ionic strength, mobile phase pH, the organic solvent content, and the migration distance of mobile phase on the UTLC plate were examined at their low and

5-2 high levels by conducting a quarter fraction of the full factorial design of 2 in duplicate.

Center points were also added in triplicate to the design to detect curvature of the fitted data. After significant factors were determined by the screening design, response surface methodology using Box-Behnken design was then used to optimize the response. All the statistical experimental designs and the optimization calculations were performed using

Minitab 17 (Minitab, Inc., State College, PA).

2.3 Results and Discussion

2.3.1 Optimization of the UTLC Conditions for Amino Acids

Electrospinnability of Nafion-PAN as a function of blend composition and total solid concentration in DMF was previously studied by Tran, et al.25 The optimum composition of Nafion-PAN in DMF for a bead-free and uniform UTLC sorbent with sufficient ionic capacity was further explored in this work. As an initial step, a fractional factorial design was performed to identify the significant variables involved in the separations using Nafion-PAN UTLC.

A two level (25-2) fractional factorial design was applied in duplicate with a total of 16 experiments. Meanwhile, three replicates are performed in the center points to examine the curvature of the fitted data. The factors and their levels evaluated by the screening experimental design included the Nafion concentration in the electrospun nanofibers (20% and 60%), the ionic strength (10 mM and 50 mM), pH (3.0 and 4.0), the percentage of organic solvents (10% and 40%), which consist of 1:1 (v/v) methanol and n-butanol, and the migration distance of mobile phase on the UTLC plate (2 cm and 4 43 cm). The resolution Rs was chosen as the response to evaluate the influence of the variables (Equation 2.2)

푍푠2−푍푠1 푅푠 = 2 (2.2) 푤1+푤2 where Zs is the distance migrated by the analyte, and w is the developed spot width.

Therefore, the greater difference between migrated distances of analytes 1 and 2 (Zs1 and

1 Zs2) and the smaller their spot widths, the greater resolution of the separation. Among the six amino acids, the pair of Lys and His was always difficult to separate and had the smallest Rs value in each separation. Thus, the Rs between Lys and His was used as the response to optimize. Once the significant factors are determined, response surface methodology using Box-Behnken design35 was applied to find the factor settings that optimize the response. Box-Behnken design is a type of response surface design, which is more efficient and less expensive to do than a central composite design with the same number of factors. In a Box-Behnken design, the treatment combinations are at the midpoints of the edges and at the center of the experimental space. Box-Behnken design allows efficient estimation of the first and second order coefficients.35

The results obtained from the evaluation of significant factors by fractional factorial design are summarized in the Pareto chart of effects, which is depicted in Figure

2.2A. Migration distance, organic solvent and Nafion concentration have significant effects on the resolution at 95% confidence level. Figure 2.2B shows the main effect of each factor on the resolution. The steeper the slope of the line, the greater the magnitude of the main effect. The higher concentration of Nafion (60%) resulted in better separation

44 compared to 20% since the larger amount of Nafion provides more ion exchange sites thus larger ionic capacity. The lower percentage of organic solvents in the mobile phase resulted in a higher resolution, but when the organic solvents were decreased to be lower than 10%, the migration rate of the mobile phase became extremely slow due to the hydrophobic nature of Nafion. The migration distance of 4 cm gave a relatively poor resolution, and this can be a result of longitudinal diffusion when the migration rate decreases as the solvents moves up on the UTLC plate. Ionic strength and pH of the buffer also have effects on the resolution, however, these effects were not significant.

Therefore, these two variables were held constant at the average level: ionic strength=

30mM, pH=3.5, while the Nafion concentration, organic solvent and migration distance were further evaluated for optimization purpose.

Figure 2.2 (A) Standardized Pareto chart of the fractional factorial design for resolution,

(B) Main effects plot for resolution.

Box-Behnken design was applied to optimize the three significant factors determined from fractional factorial design. The levels of the three factors and theirs corresponding values are listed in Table 2.1. The number of experiments (N) required for the development of Box-Behnken design is defined as:

푁 = 2푓 (푓 − 1) + 퐶0 (2.3)

35 where f is number of factors and C0 is the number of replicates on the central point.

Thus, a total of 15 experiments were performed. The experimental data were then fitted to the following second order polynomial regression equation:

2 2 2 푌 = 푏0 + 푏1푥1 + 푏2푥2 + 푏3푥3 + 푏11푥1 + 푏22푥2 + 푏33푥3

+푏12푥1푥2 + 푏13푥1푥3 + 푏23푥2푥3 (2.4) where Y = predicted response, b0 = a constant, b1, b2, b3 = linear coefficient, b11, b22, b33 = squared coefficient, b12, b13, b23 = interaction coefficient, and x1, x2 and x3 are the independent variables. The statistical significance of each term in this quadratic model was evaluated by analysis of variance (ANOVA) (Table 2.2). The predicted model using coded units from Box-Behnken design can be described as:

2 2 푅푠 = 0.6373 + 0.0376 푥1 − 0.1886 푥2 − 0.0722푥3 − 0.3589 푥1 − 0.0802푥2 −

2 0.2491푥3 − 0.1635 푥1푥2 − 0.0010푥1푥3 + 0.1502푥2푥3 (2.5) where x1 is the Nafion concentration, x2 is the percentage of organic solvent and x3 is the migration distance. The ANOVA shows an insignificant lack of fit, which reveals an adequate representation of the experimental data using this model. When the insignificant factors (p > 0.05) were omitted from the model, the fitting of the proposed model became worse (R2 =70.87% instead of R2= 91.88%). Therefore, the full response model 47 containing all factors was used for further optimization. The contour plots generated by this design, which can be used as a guide to obtain the optimal settings, are illustrated in

Figure 2.2. These contour plots show how the resolution varies with the Nafion concentration, the percentage of organic solvent and the migration distance. The function optimization leads to 45.45% Nafion concentration, 10.0% organic solvent and 2.545 cm migration distance, which were rounded to the nearest integer. As shown in the response optimization plot (Figure 2.3.), the optimum conditions were determined as: Nafion concentration of 45%, organic solvents of 10%, migration distance of 2.5 cm.

Table 2.1 Summary of factors and levels used in Box-Behnken design.

Level Factor Low (-1) Central (0) High (+1)

Nafion Concentration 20 40 60 (%), x1 Organic Solvents (%), 10 25 40 x2 Migration Distance 2 3 4 (cm), x3

Table 2.2 ANOVA for response surface quadratic model for optimization of UTLC conditions for amino acids.

Term DF SS MS F-Value p-Value

Model 9 1.198 0.133 6.28 0.028

Nafion Concentration (x1) 1 0.011 0.011 0.53 0.498

Organic Solvent (x2) 1 0.285 0.285 13.44 0.014

Migration Distance (x3) 1 0.042 0.042 1.97 0.220

2 2 Nafion Concentration (x1 ) 1 0.476 0.476 22.46 0.005

2 2 Organic Solvent (x2 ) 1 0.024 0.024 1.12 0.338

2 2 Migration Distance (x3 ) 1 0.229 0.229 10.82 0.022 Nafion Concentration X 1 0.107 0.107 5.05 0.075 Organic Solvent (x1x2) Nafion Concentration X 1 0.000 0.000 0.01 0.989 Migration Distance (x1x3) Organic Solvent X Migration 1 0.090 0.090 4.26 0.094 Distance (x2x3) Residual 5 0.106 0.021 - -

Lack of Fit 3 0.098 0.033 8.55 0.106

Pure Error 2 0.008 0.0038 - -

DF: Degree of freedom; SS: Sum of square; MS: Mean square; R2 = 91.88%; R2 (adj)=

77.26%; Standard error of the regression = 0.146.

Figure 2.3 (A) Contour plots using the Box-Behnken design showing the effect of (A)

Nafion concentration and migration distance on resolution (B) Nafion concentration and organic solvent on resolution (C) organic solvent and migration distance on resolution.

Figure 2.4 Response optimization plot.

2.3.2 Nanofiber Stability in Mobile Phases

The electrospun Nafion-PAN nanofibers have smooth and uniform fiber morphology as shown in Figure 2.4. In order to confirm the stability of the stationary phase for the use in UTLC, Nafion-PAN fiber mats were immersed in the optimized mobile phase mixtures for 60 min. After soaking, the plates were allowed to dry, and

SEM images were taken for the measurements of fiber diameters. Figure 2.4B and C show that there was no swelling or fiber dissolution after soaking Nafion-PAN fiber mats in the optimized mobile phase mixtures for 60 min. This illustrates the superb chemical resistance of the Teflon backbone of Nafion. In addition, compared to previously reported PVA UTLC for amino acids, which required crosslinking treatment to avoid mat fusion and mass loss when it was exposed to mobile phases, PAN is insoluble in water and common organic solvents except DMF, dimethyl acetamide (DMAA), dimethyl sulfoxide (DMSO).12,36

Figure 2.5 SEM images of Nafion-PAN nanofibers (A) before soaking (B) after being soaked in formate buffer: methanol: n-butanol (90: 5: 5 v/v/v) after 60 min, (C) in H2O:

2-propanol: NH4OH (54:43:3, v/v/v) + 0.15M NaCl after 60 min.

2.3.3 Mobile Phase Velocity

Electrospinning produces nanofibers formed by randomly oriented fibers lying loosely on top of each other37 which creates a porous network of stationary phase. The mobile phase transport through the porous layer by capillary action can be described by the following relationship:

푍푓 = √к푡 (2.6) where Zf is the total distance moved by the solvent front from the origin, к is the velocity constant, and t is the development time.38 The mobile phase velocity can be determined by plotting migration distance versus the square root of time. The velocity constant к

(cm2/s) is also related to the properties of the stationary and mobile phases:

к = 2퐾0푑푝(훾⁄ƞ) 푐표푠 휃 (2.7) where K0 is the permeability constant, dp is the average particle diameter of the stationary phase, Ƴ is the surface tension of the mobile phase, ƞ is the viscosity of the mobile phase and Ɵ is the contact angle between mobile and stationary phases. In this study, the optimized mobile phase for amino acids was used on Nafion-PAN plates containing different Nafion concentrations to allow the comparison under the same condition. Figure

2.5A shows the variation of the migration distance as a function of the square root of time for the Nafion-PAN plates, and the slope is к0.5 based on Equation 2.6. By comparing the slopes, the more the Nafion concentration on the UTLC plate, the slower the mobile phase transport. Equation 2.7 indicates that the mobile phase velocity is directly proportional to the particle diameter. The fiber diameters of Nafion-PAN plates 55 containing different Nafion concentrations are illustrated in Figure 2.5B; hence the K0 term was smaller for Nafion-PAN plates containing more Nafion. The contact angle between mobile phase and stationary phase became larger with the addition of Nafion into the nanofibers (Figure 2.6); therefore, cos Ɵ term was smaller for the UTLC plate with a higher content of Nafion. The decreased mobile phase velocity was observed as a result of the decrease in fiber diameter and permeability, and an increase in contact angle between the mobile phase and the Nafion-PAN stationary phase. As observed in the experimental designs, when the Nafion concentration was above 45%, the resolution started to decrease, and this can be ascribed to the slow migration velocity on those

2 0.5 plates. Figure 2.5A shows a good linearity (R > 0.99) between Zf and t , which verifies the applicability of Equation 2.6, and also demonstrated the homogeneity of the Nafion-

PAN stationary phases.

Figure 2.6 (A) Comparison of mobile phase migration rate on UTLC plates containing

20% Nafion (●) with R2=0.998, 30% Nafion (▲) with R2=0.9936, 45% Nafion (■) with R2=0.999 and 60% Nafion (◆) with R2=0.998 (B) Fiber diameter of different

Nafion concentration.

A B

Figure 2.7 Comparison of contact angles of formate buffer: methanol: n-butanol (90: 5: 5 v/v/v) on (A) pure PAN UTLC; (B) 45% Nafion-PAN UTLC.

2.3.4 Band Broadening

The efficiency of the separation can be described by the plate number (N) or the plate height (H). Their relationships with the migration distance (Zs) and the spot width

(w) of the analyte are described in the following equations:

푍 푁 = 16 ( 푠)2 (2.8) 푤

푍 퐻 = 푆 (2.9) 푁

As shown in Equation 2.8, the plate number N is inversely proportional to the square of developed spot width. Band broadening in TLC can be expressed by plate number N or plate height H. The goal of efforts to improve the performance of TLC is to obtain small H values and maximum N values.1 Phenylalanine was selected to examine the changes in plate height as a function of solvent migration distance on the Nafion-PAN

UTLC plate. As shown in Figure 2.7, the plate height H remained approximately the same with increasing distance from 5 to 25 mm and increased for distances of 30 mm or greater.

The van Deemter equation (Equation 2.10) describes the dependence of plate height H versus distance traveled and evaluates components that are causing band broadening.39,40

퐵 퐻 = 퐴 + + 퐶푢 (2.10) 푢

where u is the mobile phase velocity. The A term describes the contribution from multipath flow taken by sample solutes when they travel through the sorbent layer. For

59 fine particles with diameters smaller than 10 µm, the A term is small and typically negligible. The B term describes the contribution to the broadening from diffusion along the flow direction, i.e. the longitudinal diffusion in mobile phase. The C term describes the broadening resulting from the resistant to mass transfer in the stationary phase and mobile phase. Under capillary flow condition in TLC, the C term is also negligible.41

Therefore, in this case, only longitudinal diffusion (B term) contributes to band broadening. In TLC, due to capillary action, the mobile phase velocity gradually slows down as the solvent moves up, thus the H value would increase significantly at longer migration distance, which is in consistent with the trend observed in Figure 2.7.

Figure 2.8 Change in plate height H of phenylalanine with increasing solvent migration distance on Nafion-PAN UTLC. Error bars represent the standard deviation.

2.3.5 Ion Exchange Capacity

The IEC of 45% Nafion-PAN nanofibers was determined to be 0.408 ± 0.09 mmol/g. Based on the structure of Nafion (Figure 2.1), each ionomer contains 39 F and 1

42 SO3H. The fluorine content in 45% Nafion-PAN nanofibers was then calculated as 16.0 mmol/g. As shown in Table 2.3, the IEC increases as the Nafion concentration increases in nanofiber composite. A high IEC is desirable for an ion exchange stationary phase; however, it cannot be achieved by simply increasing the Nafion concentration in the nanofibers. The hydrophobic backbone (% F) and the sulfonate group (%SO3H) of

Nafion is interdependent, thus, an increase in the concentration of Nafion introduces not only a high IEC but also a large hydrophobicity of the nanofibers. As mentioned in section 2.3.3 and 2.3.4, a higher Nafion concentration (higher hydrophobicity) exhibited a slower mobile phase velocity, which resulted in band broadening. The equivalent weight

(EW) of the Nafion used in this study is 1100 g polymer/ SO3H group. The smaller the

EW value, the higher the density of the ion exchange sites and the lower the content of the hydrophobic backbone.43 Using different manufacturing methods, the EW of Nafion can be varied. To further enhance the IEC of the Nafion-PAN nanofibers, Nafion with a lower EW can be considered. Nevertheless, with the decrease in EW, the nanofibers are expected to swell more easily in the aqueous mobile phases. Therefore, there would be an optimum EW to achieve the highest IEC while maintaining the chemical stability of the

Nafion-PAN nanofibers. Future work is necessary to compare Nafion with different EW for the application of ion exchange UTLC.

Table 2.3 Ion exchange capacity of Nafion-PAN nanofibers.

Nafion Concentration (%) IEC (mmol/g) 20 0.185 ± 0.010 30 0.268 ± 0.009 45 0.408 ± 0.009 60 0.563 ±0.010

2.3.6 Separation of Amino Acids

As amino acids can be separated on an ion exchange plate based on the difference in their charges, using the optimized mobile phase with a pH of 3.5, amino acids with different pI values were expected to migrate different distances. Therefore, several basic and neutral amino acids were selected to test the performance of the Nafion-PAN stationary phase. The pI values and their charges in the formate buffer mobile phase are listed in Table 2.4. Three basic amino acids Arg, Lys and His with doubly positive charges, were expected to be mostly retained by sulfonate groups on the stationary phase.

Neutral amino acids Phe, Val and Ala which carried one positive charge in the mobile phase, would be relatively less retained on the Nafion-PAN plate.

The retardation factors (Rf) and plate heights (H) of these amino acids separated on the Nafion-PAN plate are listed in Table 2.4. From the chromatogram shown in

Figure 2.8, as expected, Ala, Val and Phe are less retained than the basic amino acids

Lys, His, and Arg. It is worth noting that good selectivity among the amino acids with the same amount of charges was also realized on the Nafion-PAN plate. Separation on pure

PAN plate (Table 2.5) using the same mobile phase condition was compared to Nafion-

PAN. Without addition of Nafion, pure PAN plate provides no selectivity and minimum retention for all six amino acids. This result further confirm that Nafion is an effective ion exchange component in the stationary phase. The LOD and the LOQ of amino acids on

Nafion-PAN UTLC were determined in the range of 0.26-0.49 µg/spot and 0.79-1.50

µg/spot. (Table 2.6).

Table 2.4 Isoelectric point (pI), charge, Rf and plate height of amino acids on Nafion-

PAN UTLC.

Intraday (n=3) Interday (n=3)

Amino Charge in R H (µm) R H (µm) pI f f Acid pH 3.5 (%RSD) (%RSD) (%RSD) (%RSD) Arg 10.76 2+ 0.08 (2.2) 48.7 (17.4) 0.08 (5.7) 49.2 (20.7) His 7.59 2+ 0.24 (1.5) 31.2 (7.40) 0.23 (2.4) 31.4 (12.9) Lys 9.74 2+ 0.30 (1.2) 24.6 (10.1) 0.30 (3.5) 26.3 (10.8) Phe 5.48 1+ 0.68 (5.1) 9.2 (6.1) 0.69 (6.3) 9.2 (8.5) Val 5.96 1+ 0.87 (3.4) 5.2 (13) 0.87 (3.5) 5.1 (12) Ala 6.00 1+ 0.93 (2.1) 2.1 (25) 0.92 (2.5) 2.5 (24)

Figure 2.9 Chromatogram for separation of (1) Arg, (2) His, (3) Lys, (4) Phe, (5) Val and

(6) Ala on electrospun Nafion-PAN UTLC using 30mM formate buffer (pH 3.5): methanol: n-butanol (90: 5: 5 v/v/v).

Table 2.5 Separation of amino acids on pure PAN UTLC in three repetitions (n=3).

Analyte Rf (% RSD) H (% RSD) Arg 0.92 (6.3) 18.2 (6.95) His 0.90 (5.6) 14.9 (6.13) Lys 0.89 (4.6) 17.6 (7.17) Amino Acids Phe 0.94 (1.4) 15.8 (3.40) Val 0.94 (1.0) 16.4 (8.14) Ala 0.96 (3.1) 18.4 (11.0)

Mobile Phase: 30mM formate buffer (pH 3.5): methanol: n-butanol (90: 5: 5 v/v/v).

Table 2.6 LOD and LOQ of Amino Acid.

Amino Acids Linear Equationα R2 LOD (µg/spot) LOQ (µg/spot) Arg y = 2155.1x + 228.48 0.9979 0.26 0.79 Lys y = 937.94x + 337.14 0.9988 0.36 1.10 His y = 1307.9x + 1071.1 0.9982 0.46 1.38 Phe y = 1295x + 11.854 0.9985 0.41 1.38 Val y = 789.26x + 719.1 0.9978 0.49 1.50 Ala y = 740.52x + 541.36 0.9989 0.35 1.05

α y= peak area and x= concentration of the analyte (µg/spot)

2.3.7 Separation of Proteins

Lysozyme, bovine serum albumin, α-chymotrypsin and myoglobin were used as model proteins to test the chromatographic performance of the Nafion-PAN UTLC for intact proteins. Similar to the optimization procedure for amino acids, the organic solvent content, ionic strength of the mobile phase and the amount of additive (NH4OH) were optimized using a Box-Behnken design for the protein separation based on improved resolution. The levels of the three factors and theirs corresponding values are listed in

Table 2.7. The statistical significance of each factor in the quadratic model was evaluated by ANOVA (Table 2.8). The predicted model using coded units from Box-Behnken design can be described as:

2 2 Rs = 1.4665 − 0.1898 x1 − 0.0525 x2 + 0.1791 x3 − 0.4723 x1 − 0.6470 x2

2 − 0.5757x3 − 0.2046 x1x2 − 0.1567x1x3 − 0.1653x2x3 (2.11)

The response surface plots are shown in Figure 2.9, and the optimized mobile phase for intact protein samples on the Nafion-PAN UTLC containing H2O: 2-propanol: NH4OH

(54:43:3, v/v/v) and 0.15 M NaCl (Figure 2.10).

Table 2.7 Summary of factors and levels used in Box-Behnken design for the mobile phase in protein separation.

Level Factor Low (-1) Central (0) High (+1) Organic Solvent (%), 20 50 80 x1

NaCl (M), x2 0 0.15 0.3

Additive (%), x3 0 2.5 5

Table 2.8 ANOVA for response surface quadratic model for the mobile phase in protein separation.

Term DF SS MS F-Value p-Value

Model 9 4.068 0.452 18.38 0.003

Organic Solvent (x1) 1 0.288 0.288 11.72 0.019

NaCl (x2) 1 0.022 0.022 0.90 0.387

Additive (x3) 1 0.257 0.257 10.43 0.023

2 2 Organic Solvent (x1 ) 1 0.824 0.824 33.49 0.002

2 NaCl (x2 ) 1 1.546 1.546 62.85 0.001

2 2 Additive (x3 ) 1 1.224 1.224 49.75 0.001 Organic Solvent X 1 0.167 0.167 6.81 0.048 NaCl (x1x2) Organic Solvent X 1 0.098 0.098 4.00 0.102 Additive (x1x3) NaCl X Additive 1 0.109 0.109 4.45 0.089 (x2x3) Residual 5 0.123 0.024 - -

Lack of Fit 3 0.119 0.040 18.33 0.052

Pure Error 2 0.004 0.002 - -

DF: Degree of freedom; SS: Sum of square; MS: Mean square; R2 = 97.07%; R2 (adj)=

91.78%; Standard error of the regression = 0.157.

Figure 2.11 Response optimization plot for protein separation.

The different molecular weights, pI values and hydropathicities of these proteins

(Table 2.9) would make them have very distinct retention behaviors on this ion exchange stationary phase. The hydropathicities of proteins were quantified by grand average hydropathy (GRAVY) score, which was calculated by adding the hydropathy value for each residue and dividing by the length of the sequence.44 Due to the extremely strong electrostatic interaction between proteins and the stationary phase, elongated spots were observed when acidic and neutral mobile phase conditions were applied; therefore, a basic mobile phase was used to reduce the net positive charges on the proteins, which in turn reduced the retention. The separation of these four proteins is shown in Figure 2.11.

The Rf and plate height values of each protein are listed in Table 2.9, and an excellent separation efficiency was achieved with plate height as low as 8.7 µm observed. In this basic mobile phase system, only Lys is positively charged, and the other three proteins are negatively charged. As a relatively polar protein with positive charges, the retention of Lys was most likely due to the ionic interaction and ion-dipole interaction with the sulfonate group of Nafion. Even though Chy and MGB have more basic pI values which were expected to be more retained than BSA on the cation exchange stationary phase,

BSA was the second mostly retained protein. The retention of BSA may be the result of its high molecular weight because a large molecule like BSA has more chance to form extremely strong binding through multipoint attachments (hydrogen bonds) with the sorbent layer than the smaller molecules.45 As can be seen in Figure 2.11, MGB was more retained than Chy on the Nafion-PAN plate. With the similar pI values and the close molecular weights, the selectivity between Chy and MGB could be mainly ascribed

72 to their different hydropathicites. The more hydrophilic MGB (GRAVY= -0.381) comprises more polar groups on the side chains, which would bind to the sulfonate groups more strongly via ion-dipole interaction compared to the relatively hydrophobic

Chy (GRAVY= 0.039). Separation of proteins on pure PAN plate was also compared to the results on Nafion-PAN plate (Table 2.10). In the absence of Nafion, pure PAN provides no selectivity for Lys, BSA and Chy, and the retention of all four proteins are relatively weak on pure PAN. The LOD and the LOQ of proteins on Nafion-PAN UTLC were determined in the range of 0.59-0.88 µg/spot and 1.82-2.68 µg/spot. (Table 2.11).

Table 2.9 Molecular weights, pI, GRAVY scores and the separation results of the proteins used in this work.

Intraday (n=3) Interday (n=3)

MW GRAV R H (µm) R H (µm) Protein pI f f (kDa) Y (%RSD) (%RSD) (%RSD) (%RSD)

BSA 66 4.7 -0.479 0.55 (4.0) 13.9 (12.6) 0.57 (5.7) 14.5 (13.3)

MGB 17 7.2 -0.381 0.64 (2.2) 16.8 (16.5) 0.62 (4.9) 16.5 (14.1)

Chy 25 8.8 0.039 0.78 (3.2) 8.7 (11) 0.78 (4.8) 8.3 (13)

Lys 14.3 11.4 -0.472 0.48 (4.2) 29.9 (14.6) 0.44 (8.2) 20.3 (14.3)

Figure 2.12 Chromatogram for separation of (1) LYZ, (2) BSA, (3) MGB, 4) CHY on electrospun Nafion-PAN UTLC using 0.15 M NaCl in H2O: 2-propanol: NH4OH (54: 43:

3, v/v/v).

Table 2.10 Separation of intact proteins on pure PAN UTLC in three repetitions (n=3).

Analyte Rf (% RSD) H (% RSD)

BSA 0.83 (1.9) 7.8 (13)

Proteinsb MGB 0.81 (4.1) 11.0 (12.0)

Chy 0.92 (1.7) 20.2 (10.3)

Lys 0.84 (0.81) 9.0 (14)

Mobile Phase: 0.15 M NaCl in H2O: 2-propanol: NH4OH (54: 43: 3, v/v/v).

Table 2.11 LOD and LOQ of Proteins.

Proteins Linear Equationα R2 LOD (µg/spot) LOQ (µg/spot)

BSA y = 306.81x + 7.4783 0.9959 0.88 2.68

MGB y = 370x + 247.78 0.9993 0.76 2.31

Chy y = 95.098x + 243.98 0.9973 0.59 1.82

Lys y = 84.355x + 751.45 0.9987 0.60 1.82

α y= peak area and x= concentration of the analyte (µg/spot).

2.4 Conclusions

Nafion-PAN nanofibers were successfully fabricated via the electrospinning method. To the best of our knowledge, it is the first work that Nafion-PAN nanofibers have been used as the stationary phase of UTLC which provides ion exchange mechanism. Fractional factorial design and response surface methodology were applied to optimize the Nafion-PAN stationary phase and separation conditions. The smooth and uniform nanofibers obtained exhibited excellent mechanical stability and solvent compatibility. The separation of amino acids confirmed the feasibility of Nafion-PAN nanofibers as the UTLC stationary phase. The Nafion-PAN UTLC demonstrated high separation efficiency for amino acids. The decreased mobile phase velocity was resulted from the reduced fiber diameter and the increased contact angle between the mobile phase and the Nafion-PAN stationary phase. The separation of proteins illustrated that

Nafion-PAN nanofibers are also suitable for separation of large biomolecules. The retention of proteins on the Nafion-PAN UTLC largely depends on the properties of proteins including the net charge, hydropathicity, molecular size and structure.

2.5 References

1 Sherma, J.; Fried, B. Handbook of Thin-Layer Chromatography, 3rd ed.; Mercel Dekker: New York, 2013.

2 Zhao, C.; Xiong, Y.; Yu, L.; Li, S.; Wu, Z. J. Chromatogr. Sci. 2010, 48, 685-689.

3 de Andrade, F. I.; Florindo Guedes, M. I.; Pinto Vieira, I. G.; Pereira Mendes, F. N.; Salmito Rodrigues, P. A.; Costa Maia, C. S.; Marques Avila, M. M.; de Matos Ribeiro, L. Food. Chem. 2014, 157, 193-198.

4 Shewiyo, D. H.; Kaale, E.; Risha, P. G.; Dejaegher, B.; Smeyers-Verbeke, J.; Vander Heyden, Y. J. Pharm. Biomed. Anal. 2012, 66, 11-23.

5 Le Roux, A. M.; Wium, C. A.; Jouber, J. R.; Van Jaarsveld, P. P. J. Chromatogr. 1992, 581, 306-309.

6 Komsta, L.; Waksmundzka-Hajnos, M. Thin Layer Chromatography in Drug Analysis; CRC Press: Florida, 2014.

7 Sherma, J. Anal. Chem. 1988, 60, 74 R-86 R.

8 Bezuidenhout, L. W.; Brett, M. J. J. Chromatogr. A 2008, 1183, 179-185.

9 Fried, B.; Sherma, J. Thin-Layer Chromatography, 4th ed.; Marcel Dekker: New York, 1999.

10 Clark, J. E.; Olesik, S. V. Anal. Chem. 2009, 81, 4121-4129.

11 Clark, J. E.; Olesik, S. V. J. Chromatogr. A 2010, 1217, 4655-4662.

12 Lu, T.; Olesik, S. V. J. Chromatogr. B 2013, 913, 98-104.

13 Newsome, T. E.; Olesik, S. V. Anal. Chem. 2014, 86, 10961-10969.

14 Fang, X.; Olesik, S. V. Anal. Chim. Acta 2014, 830, 1-10.

15 Grot, W. G. U.S. Patent 3, 770, 567, 1973.

16 Dong, B.; Gwee, L.; Cruz, D.; Winey, K.; Elabd, Y. Nano Lett. 2010, 10, 3785-3790.

17 Mauritz, K. A.; Moore, R. B. Chem. Rev. 2004, 104, 4535–4585.

18 Zhou, C.; Liu, Z.; Dai, J.; Xiao, D. Analyst 2010, 135, 1004-1009.

19 Chen, H.; Snyder, J.; Elabd, Y. Macromolecules 2008, 41, 128–135.

20 Zhao, J.; Yuan, W. Z.; Xu, A.; Ai, F.; Lu, Y.; Zhang, Y. React. Funct. Polym. 2011, 71, 1102-1109.

21 Sharma, D. K.; Li, F.; Wu, Y. Colloids Surf. A 2014, 457, 236-243.

22 Ballengee, J.; Pintauro, P. J. Electrochem. Soc. 2011, 158, B568-B572.

23 Laforgue, A.; Robitaille, L.; Mokrini, A.; Ajji, A. Macromol. Mater. Eng. 2007, 292, 1229-1236.

24 Nah, C.; Lee, Y. S.; Cho, B. H.; Yu, H. C.; Akle, B.; Leo, D. J. Compos. Sci. Technol. 2008, 68, 2960-2964.

25 Tran, C.; Kalra, V. Soft Matter 2013, 9, 846-852.

26 Sharma, D. K.; Shen, J.; Li, F. RSC Adv. 2014, 4, 39110-39117.

27 Ueno, H.; Wang, J.; Kaji, N.; Tokeshi, M.; Baba, Y. J. Sep. Sci. 2008, 31, 898 – 903.

28 Martens, D. A.; Loeffelmann, K. L. J. Agric. Food Chem. 2003, 51, 6521–6529.

29 Zhou, L.; Mao, B.; Ge, Z. J. Pharm. Biomed. Anal. 2008, 46, 898–906.

30 Bonnerjea, J.; Oh, S.; Hoare, M.; Dunnill, P. Biotechnol. J. 1986, 4, 954-958.

31 Touchstone, J. C.; Dobbins, M. F. Practice of thin layer chromatography, 2nd ed.; Wiley: New York, 1978.

32 Devenyi, T. Hung. Sci. Instrum. 1974, 30, 13-22.

33 Felix, A. M.; Jimenez, M. H. J. Chromatogr. 1974, 89, 361-364.

34 Box, G. E.; Hunter, J. S.; Hunter, W. G. Statistics for Experimenters: Design, Innovation, and Discovery, 2nd Ed.; Wiley: Hoboken, NJ, 2005.

35 Ferreira, S. L. C.; Bruns, R. E.; Ferreira, H. S.; Matos, G. D.; David, J. M.; Brandao, G. C.; da Silva, E. G. P.; Portugal, L. A.; dos Reis, P. S.; Souza, A. S.; dos Santos, W. N. L. Anal. Chim. Acta 2007, 597, 179–186.

36 Iovleva, M. M.; Smirnova, V. N.; Budnitskii, G. A. Fibre Chem. 2001, 33, 262-264.

37 Li, W.; Laurencin, C. T.; Caterson, E. J.; Tuan, R. S.; Ko, F. K. J. Biomed. Mater. Res. 2002, 60, 613-621.

38 Acikara, O. B. Ion-Exchange Chromatography and Its Applications, Column Chromatography, Dr. Dean Martin ed.; InTech, DOI: 10.5772/55744, 2013.

39 Spagenberg, B. Quantitative Thin-Layer Chromatography; Springer, New York, 2011.

40 Carr, P. W.; Sun, L. J. Microcolumn Sep. 1998, 10, 149-152.

41 Geiss, F. Fundamentals of Thin Layer Chromatography; Huethig: Heidelberg, Germany, 1987.

42 Chen, T.; Leddy, J. Langmuir 2000, 16, 2866–2871.

43 Mauritz, K. A.; Moore, R. B. Chem. Rev. 2004, 104, 4535-4585.

44 Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157, 105-132.

45 Cole, R. D. J. Bio. Chem. 1961, 236, 1369-1371.

Chapter 3. Separation of PEGylated Gold Nanoparticles by Micellar Ultrathin

Layer Chromatography using Electrospun Polyacrylonitrile Nanofibers as the

Stationary Phase

Chemical Society.

3.1 Introduction

Due to their extraordinary physicochemical properties, gold nanoparticles

(AuNPs) are increasingly used in many fields including electronics,1 food industry,2 cosmetics,3 and especially biomedical applications, such as bioimaging,4 targeted drug and gene delivery,5 cancer diagnostics and immunotherapy.6 For biomedical applications, one of most crucial factors is the surface functionality of AuNPs.7 Unfunctionalized

AuNPs are unstable and prone to agglomerate quickly in biological environments, which in turn will exhibit unpredictable effects that can enormously complicate their interactions with biological molecules and cell organelles.8 Thiol-containing polyethylene glycol (PEG) is one of the most widely used biocompatible capping agents to stabilize

AuNPs.9,10,11 PEGylated AuNPs have increased stability in vivo, reduced non-specific binding to proteins, decreased cellular uptake and enhanced systemic circulation times 80 compared to unfunctionalized AuNPs or AuNPs with other capping agents, like citrate and hexadecyltrimethylammonium bromide (CTAB).12,13 The size of the AuNPs is the other most critical factor that can largely affect their performance in biomedical applications. Both cellular uptake and cytotoxicity of AuNPs are greatly dependent on nanoparticle size.14,15 On the other hand, most synthetic methods of producing AuNPs result in a distribution of sizes. Furthermore, the length of the end capping compound can effectively change the hydrodynamic radius of AuNPs. Therefore, attention is turning towards developing efficient methods for size characterization.

Traditionally, the size of AuNPs is evaluated by microscopic or spectroscopic techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), atomic force microscopy (AFM) and dynamic light scattering (DLS).

However, electron microscopic methods may be time consuming. Moreover, artifacts can be generated during sample preparation.16 Most importantly, these techniques do not involve a separation process.17 Although DLS offers an obvious advantage over microscopic methods as it measures the hydrodynamic radius, which includes the size of soft coatings on AuNPs, it is still susceptible to skewed results when a sample is polydisperse.18

In recent years, separation techniques such as size exclusion chromatography

(SEC),17,19 hydrodynamic chromatography (HDC),20 field-flow fractionation (FFF),21,22 capillary electrophoresis (CE)23,24 and gel electrophoresis (GE)25,26 was applied to fractionate AuNPs by size. In SEC, sample degradation or irreversible adsorption of NPs onto the stationary phase may occur, although this effect can be minimized by adding

81 surfactants into the mobile phase.17 HDC separates NPs based on their hydrodynamic radius and the separation is independent on the coating or surface charge of NPs.16 FFF provides continuous and high-resolution separation for NPs in the size range of 1 nm to 1

µm.16 Hansen et al. showed that asymmetric-flow FFF (AF4) is a useful tool for the separation of 30 nm AuNPs coated with different molecular weights of polyethylene glycol.22 However, FFF usually suffers overloading, incomplete sample recovery, and requires expensive instrumentation.16 CE is commonly used for the separation of AuNPs based on their size and charge, but it provides low sensitivity to UV detection as only very small volumes of sample can be loaded into CE capillaries.16 PEGylated Au NPs were extensively studied by gel electrophoresis due to the ease of detection via their surface plasmon resonance (SPR).25,26 Nevertheless, this technique is time consuming (30 min to several hours) to achieve good separations and requires large sample volume.25

Recently, Yan et al. reported the separation of AuNPs using commercial silica gel high performance thin layer chromatography (HPTLC).27,28 Differently sized citrate- functionalized AuNPs were separated with sufficient resolving power, however, the development time was 20 min. Consequently, there is still demand for other techniques for the separation of AuNPs.

Ultrathin layer chromatography (UTLC) was first introduced in 2001 using monolithic silica as the stationary phase. Our group reported electrospun nanofibers as the stationary phases for UTLC. Electrospun UTLC offers several distinct advantages: 1)

This method has minimum instrumentation requirements. The fabrication of electrospun

UTLC plates is simple, fast and low cost. Moreover, a wide range of polymers or

82 materials can be electrospun into nanofibers;29 2) With almost 10 times thinner sorbent layer compared to commercial HPTLC plates and in the absence of a binder, the reported electrospun UTLC plates have showed enhanced separation efficiency, reduced sample and solvent consumption and increased speed of analysis;30 3) Crude samples can be directly applied onto the UTLC plates with minimum sample preparation.27 Previously, electrospun polyacrylonitrile (PAN) and composite nanofibers were successfully applied for separations of laser dyes,30 polycyclic aromatic hydrocarbons (PAHs),31 amino acids and proteins.32

In this dissertation, for the first time, UTLC was employed to separate PEGylated

AuNPs with different sizes and AuNPs coated with various molecular weights of PEG.

3.2 Materials and methods

3.2.1 Materials

10, 30, 50 and 80 nm AuNPs (PEGylated, 5 kDa) and 10 and 30 nm citrated stabilized AuNPs were purchased from NanoComposix Inc. (San Diego, CA). Methoxy- poly (ethylene glycol)-thiol (mPEG-SH) was purchased from Laysan Bio Inc. with different lengths: 2, 5, 10 and 20 kDa. PAN, average Mw 150,000 g mol-1, 3-

(cyclohexylamino)-1-propanesulfonic acid (CAPS, ≥ 98%), sodium dodecyl sulfate

(SDS, BioReagent, ≥98.5), hexadecyltrimethylammonium bromide (CTAB, ≥ 98%), polyoxyethylene (23) lauryl ether (Brij-35) and bovine serum albumin (BSA, lyophilized pow-der, ≥ 96%) were purchased from Sigma-Aldrich (St. Louis, MO). N, N-

Dimethylformamide (DMF) (99.8%), HPLC grade 2-propanol (99.9%) and tris(hydroxymethyl)aminomethane (Tris, molecular biology grade) were acquired from

Fisher Scientific (Fair Lawn, NJ). Milli-Q water was purified to 18.1 MΩ by a Barnstead

Nano-pure Infinity System from Thermal Scientific Inc. (Odessa, TX).

3.2.2 Instrumentation

The morphology of electrospun nanofibers was characterized using a FEI Nova

NanoSEM 400 (FEI Corporate, North America NanoPort, Hillsboro, OR) scanning electron microscope (SEM). Each sample was sputter coated with AuPd for 90 s at 17 mA using a sputter coater (Cressington Scientific Inst., Watford, UK) to create a conductive surface before SEM analysis. Fiber diameters were measured on SEM images using ImageJ software (Available at http://www.rsbweb.nih.gov/ij/index.html). A Canon

A650IS 12.1 MP digital camera was used for acquisition and documentation of UTLC images. Hydrodynamic diameters of AuNPs were measured using dynamic light scattering instrument (Brookhaven Instruments Corporation, BI-200SM). Zeta potentials of AuNPs were measured using NanoBrook ZetaPALS Potential Analyzer (Brookhaven

Instruments Corporation). Absorption spectra of AuNPs were recorded using an Agilent

Cary 5000 UV-Vis-NIR spectrometer; water was used as the reference solvent.

3.2.3 Electrospining

A syringe pump (Harvard Apparatus, Holliston, MA), a high voltage power supply (Spellman, Hauppauge, NY), a stainless-steel collector covered with aluminum foil (Reynold Super Strength) and a Plexiglas enclosure were used as the electrospinning apparatus as previously described.30 A nitrogen purge and a VWR Traceable® humidity sensor was used to control and monitor the relative humidity in the closed chamber while electrospinning. Nanofibers were electrospun at room temperature with relative humidity below 20%. The distance between the collector and the tip of the needle was kept at 15 cm. The voltage was kept at 16 kV, and the flow rate was kept at 0.3 mL/h. Each nanofiber mat was electrospun for 10 min.

3.2.4 Ultrathin Layer Chromatography

After the electrospun nanofiber mats were prepared, they were cut into 3 cm × 6 cm UTLC plates. Hamilton® micro syringe (10 µL) were used for spotting AuNPs onto the origin line drawn at 1 cm from the bottom of the plate. All the AuNPs samples have mass concentration of 0.05 mg/mL. The volume of each AuNP solution spotted on UTLC was controlled at 0.25 µL. UTLC plates were then placed in a 250 mL developing chamber containing 5 mL of mobile phase. Prior to each development, the mobile phase was allowed to equilibrate for 10 min.

3.2.5 PEGylation of AuNPs

AuNPs with different molecular weight of PEG were prepared according to the method reported by Smith et al.33 Briefly, PEG powder was dissolved in Milli-Q water with the concentration of 6, 15, 30 and 60 mg/mL for the 2, 5, 10 and 20 kDa PEG, respectively. The 1.2 mL of each PEG solution was added to 10 mL of 30 nm citrate stabilized AuNPs. The mixtures were then magnetically stirred at room temperature for

24 h to allow complete exchange of PEG with citrate. Each PEGylated AuNP solution was then centrifuged at 14,000 rpm for 60 min. After centrifugation, the supernatant containing excess PEG was removed and the AuNPs were re-dispersed in Milli-Q water by vortex mixing.

3.3 Results and Discussion

3.3.1 Electrospun Nanofibers

Electrospun PAN nanofibers were characterized by SEM (Figure 3.1). PAN nanofibers exhibited uniform morphology, which is highly desirable in the application of

UTLC. The average fiber diameter was 296 ± 26 nm. The electrospun PAN UTLC possesses microscale interstitial spaces between the nanofibers, which are large enough for AuNPs to enter the spaces and be retained on the surface of nanofibers.

Figure 3.1 (A) SEM image and (B) the fiber diameter distribution of electro-spun PAN nanofibers.

3.3.2 Mobile Phase Conditions

Mobile phase components that may affect the retardation factor (Rf) and efficiency of the sample spots were evaluated.

3.3.2.1 Effect of Surfactants

SDS, CTAB, and Brij-35 represents surfactants that differ according to the charge of their polar head: anionic, cationic and nonionic, respectively. These surfactants were used to investigate the effect of surfactant type on the retardation factors of PEGylated

AuNPs with size of 10, 30, 50 and 80 nm. Micelles form in the mobile phase when the concentration of the surfactant is above its critical micelle concentration (CMC). A concentration range including below the CMC to concentrations well above CMC was studied for each surfactant.

As shown in Figure 3.2, the selectivity between all AuNPs improved with an increase of SDS concentration above its CMC; on the contrary, all the AuNPs remained strongly retained regardless the concentration change of CTAB in the mobile phase, which indicated that there was minimal association between the PEGylated AuNPs and

CTAB. Previous studies noted that surfactants may bind cooperatively to nonionic water- soluble polymers such as PEG to form micelle-polymer complexes,34,35 but these interactions are largely restricted to anionic surfactants (e.g. SDS).36,37 Manna et al. also found that there was no direct or indirect associative interaction between CTAB micelles and PEG.38 Interactions between PEGylated AuNPs and SDS could be mainly due to the

88 electrostatic interaction between PEG and the headgroup of the SDS and the hydrophobic interaction between the hydrophobic sections of these two species. The Rf values of all

AuNPs increased with increasing Brij-35 concentration; however, the selectivity among the AuNPs did not increase for Brij-35 as much as that observed for SDS. This may be a result of the identical structures of PEG and the polar head of Brij-35, therefore, the strong eluent strength of the Brij-35 mobile phase hindered its selectivity. Consequently, the 3% SDS concentration was adopted for the subsequent studies unless otherwise noted.

Figure 3.2 Effect of (A) SDS, (B) CTAB, (C) Brij-35 on retardation factors of 10 nm, 30 nm, 50 nm and 80 nm Au-PEG.

3.3.2.2 Effect of Alcohol in the Mobile Phase

The effect of the addition of 2-propanol as the organic modifier into the micellar mobile phase was studied. Since hydrophobic interaction is the driving force for the micelle formation, large proportions of modifiers can totally disrupt the micelle structure.39 Therefore, the volume percentage of 2-propanol added to the micellar mobile phase was maintained below 25%. Tsianou et al. illustrated that the addition of 2- propanol caused increased clustering of PEG when both SDS and PEG were in an aqueous solution and the SDS micelle shape changed from ellipsoidal to a spherical.

Furthermore, the surface area of an SDS micelle decreases from 45 to 16 nm2. However, this decrease in micelle size would also increase the total amount of SDS micelles that would be attached to the PEG.40 Increasing amount of SDS micelles on the Au-PEG can essentially increase the hydrodynamic diameter of AuNPs. As shown in Figure 3.3, the hydrodynamic diameter of AuNPs increases with the addition of 2-propanol in the aqueous mobile phase with 3% of SDS. This increase in the hydrodynamic diameter of the AuNPs would cause the observed increase in retention with addition of 2-propanol.

Figure 3.4A shows clearly that less than 10% of 2-propanol, four differently sized AuNPs could be well separated from each other. On the other hand, when aqueous-2-propanol mobile phase system was used, 40-60% of 2-propanol was needed to separate these

AuNPs with relatively poor selectivity (Figure 3.4B). In this regard, micellar enhanced

UTLC offers not only satisfactory separation for PEGylated AuNPs, but also an eco-

91 friendly chromatographic system. For the subsequent studies, 7.5% (v/v) of 2- propanol/water was used.

Figure 3.3 Effective hydrodynamic diameter of PEGylated AuNPs measured by DLS at different concentrations of 2-propanol in Tris buffer (10 mM, pH 8.5) with 3% (w/v)

SDS.

Figure 3.4 Effect of (A) in micellar mobile phase, (B) 2-Propanol in aqueous phase, (C) pH on retardation factors of 10 nm, 30 nm, 50 nm and 80 nm Au-PEG.

3.3.2.3 Effect of pH

For PEGylated AuNPs, changes in pH alters their surface. Specifically, the lone electron pair on the ether oxygen atom of PEG binds to hydrogen ions in aqueous solution making the PEG layer on AuNPs positively charged, which favors the adsorption of the anionic SDS micelles.37,41 As a result, the electrostatic interaction between SDS and PEG depends on the concentration of hydrogen ions in the solution.37 Therefore, it is necessary to optimize the pH of the mobile phase. The effect of pH on the retardation factors of AuNPs was investigated over the pH range of 3-11. As shown in Figure 3C, all the AuNPs were more retained and better separated for pH values ≥ 5. Since the positive charge on PEG is reduced as the pH increases, the increased retention of the AuNPs likely results from decreased electrostatic attraction between the PEGylated AuNPs and the negatively charged SDS micelles in the mobile phase. Basic condition near pH 8-9 provided the highest selectivity. Therefore, a pH value of 8.5 was used to separate differently sized PEGylated AuNPs.

3.3.3 Separation of AuNPs with Different Core Sizes and PEG Lengths

Under the optimized condition, PEGylated AuNPs with size of 10, 30, 50 and 80 nm were well separated from each other (Figure 3.5). The results of this work clearly demonstrated that the smaller the size of AuNPs, the faster the migration on the UTLC.

This may be attributed to the difference in the surface area of these AuNPs. As listed in

Table 3.1, smaller AuNPs have a smaller surface area, thus, their chance to interact with the PAN stationary phase is smaller than larger AuNPs. Moreover, for the same length of

PEG, the number of PEG capped on a smaller AuNP is less than the one on a larger particle,42 therefore, smaller AuNPs can be eluted more easily in the micellar mobile phase.

It is worth noting that there were stains remaining at the origin during the separation of AuNPs; they may be from the impurities in the solution of AuNPs, such as excess Au ions or PEG molecules. To clarify this assumption, standard ionic Au and pure

5 kDa PEG were spotted separately on the PAN UTLC. As shown in Figure 3.6, after the micellar mobile phase development, Au ions remained at the origin, while 5 kDa PEG migrated to Rf = 0.4. Yan et al. also observed the strong retention of Au ions at the starting point on silica HPTLC under micellar mobile phase condition.27 Therefore, the stains at the origin are ascribed to the excess ionic gold species in the AuNP solution.

Table 3.1 Core diameter, concentration, surface area and hydrodynamic diameters of differently sized AuNP.

Au-PEG (5 kDa) Au Nanoparticles 10 nm 30 nm 50 nm 80 nm

Core Diameter (nm)a 11 ± 1 33 ± 4 51 ± 7 82 ± 11

Mass Concentration 0.051 0.053 0.052 0.054 (mg/mL)

Surface Area (nm2) 314 2830 7850 20100

Hydrodynamic Diameter 43.2 57.4 78.5 111.0 (nm)

a Values obtained from nanoComposix.

Figure 3.5 Separation of differently sized PEGylated AuNPs using 3% SDS in Tris buffer

(pH 8.5, 10 mM) with 7.5% 2-Propanol as the mobile phase: lane 1, 10 nm; lane 2, 30 nm; lane 3, 50 nm; lane 4, 80 nm; lane 5, mixture of 10 nm and 50 nm.

Figure 3.6 Au ions on UTLC: a. before mobile phase development, b. after mobile phase development; 5 kDa PEG on UTLC: c. before mobile phase development, d. after mobile phase development.

The chain length of PEG can also influence the biocompatibility of AuNPs.33

Hence, it is important to separate AuNPs capped with different molecular weight of PEG.

In this work, we coated 30 nm AuNPs with 2, 5, 10 and 20 kDa PEG and separated them on electrospun PAN UTLC. After coating and prior to separation, the PEGylated AuNPs and the citrated stabilized AuNPs (starting material) were both characterized by UV-vis spectroscopy. After PEGylation, a slight red shift (2 nm) in the plasmon absorption was observed in Figure 3.7, which is in accordance with the results of other studies.42 This shift is due to the increase in the dielectric constant of AuNPs and indicates that the surface of AuNPs was PEGylated.43 To separate these AuNPs with different length of

PEG, the composition of a suitable micellar mobile phase was examined in Figure 3.8 and 3.9. Mobile phase composed of 2% SDS in CAPS buffer (pH 10, 10 mM) with 7.5 %

2-propanol was then selected. As shown in Figure 3.10, four AuNPs coated with different molecular weight of PEG were sufficiently separated from each other with very compact spots. Figure 3.10 demonstrates that increasing the PEG molecular weight leads to a slower migration of AuNPs on UTLC. Longer PEG chains can essentially increase the hydrodynamic diameter of the AuNPs. Although all these AuNPs have the same core size of 30 nm, the hydrodynamic diameters of 2, 5, 10 and 20 kDa PEGylated AuNPs measured by DLS are 38, 48, 64 and 96 nm, respectively. Therefore, the trend observed in the separation of differently sized AuNPs holds true for the AuNPs with different length of PEG: the smaller sized AuNPs migrated faster. In addition, the selectivity between these four PEGylated AuNPs can be ascribed to the difference in their

99 hydrophobicity. Increasing the polymer length of PEG substantially enhances its hydrophobicity because the ethylene repeat unit is extended.44

Figure 3.7 UV Spectra of Au-citrate and Au-PEG, Spectra of 30 nm AuNPs before and after PEGylation, the inset shows the clear red shift of about 2 nm after PEGylation.

100

Figure 3.8 Effect of surfactants on retardation factors of Au-PEG with different MW of

PEG. A: SDS, B: CTAB, C: Brij-35.

101

Figure 3.9 Effect of A: 2-Propanol in micellar mobile phase; B: pH on retardation factors of Au-PEG with different MW of PEG.

102

Figure 3.10 Separation of AuNPs with different MW of PEG using 2% SDS in CAPS buffer (pH 10, 10 Mm) with 7.5 % 2-Propanol as the mobile phase: lane 1, 2 kDa; lane 2,

5 kDa; lane 3, 10 kDa; lane 4, 20 kDa; lane 5, mixture of 2 kDa and 20 kDa.

103

3.3.4 PEGylated AuNPs in SDS Micellar UTLC

Upon addition of SDS into the mobile phase, the PAN stationary phase was modified by the adsorption of SDS monomers with their charged heads oriented inward and the hydrophobic tails outward.45 The modified PAN stationary phase acted like a pseudo-alkyl bonded phase, which bound to PEG through hydrophobic interaction. When the concentration of SDS exceeded the CMC, micelles were formed in the mobile phase and they would bind to PEG polymer chains around AuNPs. As the SDS concentration further increased, the number of micelles in the mobile phase increased, whereas the number of free monomers in the mobile phase and the bounded monomers on the PAN remained constant to give a stable modified stationary phase. The presence of SDS micelles can essentially change the conformation of PEG chains. Studies using small- angle neutron scattering illustrate that PEG exhibits a random coil like conformation in diluted aqueous solution without SDS micelles.46 When SDS micelles are present in the solution, they bind along the PEG polymer chain to give an extended and “beaded necklace” structure.40 For instance, in the presence of 3% (w/v) SDS in the mobile phase, the hydrodynamic diameter of a 10 nm Au-PEG measured by DLS changed from 43.2 nm to 63.7 nm (Table 3.2). The zeta potential of Au-PEG became more negative after the addition of SDS above the CMC, which also confirms the association of negative SDS micelles with the PEG polymers (Table 3.2). The extended polymer conformations increased the likelihood of their interactions with the stationary phase.

While Au-PEG were well separated using the optimized condition on electrospun

UTLC (polyacrylonitrile nanofibers), they were all strongly retained on commercial

104

Silica Gel HPTLC plates. This is most likely due to the different surface functionality on each stationary phase (cyano groups on nanofibers, silica on commercial HPTLC) (Figure

3.11). SDS in the mobile phase can modify cyano groups to generate pseudo-alkyl bonded phase, whereas hydrophilic surface is generated when using SDS on the negatively charged silanol group.45 The completely opposite chromatography mode gave distinct retention of Au-PEG.

105

Table 3.2 Hydrodynamic diameter and zeta potential of Au-PEG in aqueous solution without/ with SDS above its CMC.

Au-PEG 10 nm 30 nm 50 nm 80 nm

no SDS 43.2 57.4 78.5 111 Hydrodynamic Diameter (nm) 3% w/v 63.7 78.6 101.5 142 SDS

no SDS N/A -24.14 -32.16 -44.43 Zeta Potential (mV) 3% w/v -48.63 -52.81 -41.59 -52.07 SDS

106

Figure 3.11 Separation of differently sized PEGylated AuNPs on commercial Silica gel

60 F254 HPTLC using 3% SDS in Tris buffer (pH 8.5, 10 mM) with 7.5% 2-Propanol as the mobile phase: lane 1, 10 nm; lane 2, 30 nm; lane 3, 50 nm; lane 4, 80 nm.

107

3.3.5 Retention Order and PEG Density

To understand the retention order, the density of PEG molecules attached to the

AuNP must first be confirmed. There are two essential parameters that used to understanding the grafting density of PEG: Flory radius (RF) and the distance between

47,48 the individual grafted PEG chains (D). RF can be calculated by knowing the number of monomers (N) and the length of one monomer (a):

3/5 푅퐹 = 푎푁 (3.1)

For 5 kDa of PEG, N=113.5 and a= 0.35 nm. As shown in equation 3.2, the D value is correlated to the thickness of PEG layer (L), which can be determined by DLS in a pure

47 solvent. The calculated RF, L and D values for PEG on different sized AuNP are listed in Table 3.3. When D is greater than RF, the PEG adopts “mushroom” regime with the chains folding back over the AuNP surface resulting a thin PEG layer (small L). When D is smaller than RF, the PEG exists in “brush” regime with the chains stretching out from

48 the AuNP resulting a thick PEG layer (large L). By comparing the calculated D and RF for each sized Au-PEG in Table 3.3, the results suggested that PEG chains on all AuNPs used in this study existed in “brush” regime coating.

푁푎5⁄3 퐷 = ( )3⁄2 (3.2) 퐿

Knowing the distance between grafted PEG molecules, the area (A) occupied by each

PEG chain (nm2) on a spherical AuNP can be calculated using the following equation:

퐷 2 퐷 2 퐷 퐴 = 휋 (( ) + ℎ2) = 휋 ( ) , |ℎ| ≪ | | (3.3) 2 2 2

108 where, h is the height of a spherical cap, and its value is negligible in the case of a very small area taken by a molecule. Hence, the density(d) of PEG (molecule/nm2) on the

AuNPs can be determined by taking 1/A. The total number of PEG (N) on the AuNPs can be also calculated by knowing the surface area of each AuNP. In this work, we used the effective number (Neff) of PEG on each AuNP that interacted with the fibrous stationary phase to study their relationship with the retention order. For a spherical nanoparticle, only the portion facing the nanofiber would interact effectively with the ligands on stationary phase as illustrated in Figure 3.12. This portion of the nanoparticle can be considered as a spherical cap; hence, the effective surface area can be calculated using

Equation 3.4. We assume that grafted PEG on the bottom portion of each AuNP can have effective interactions, and this portion depends on the angle (θ) we take. In other words, when a PEG ligand exceeds certain angle on the AuNP, its orientation hinders its interaction with the stationary phase. In this calculation, we used θ =120° to determine the effective surface area for each AuNP. This angle might not be the exact angle during the interacting process; nonetheless, when we take a fixed θ value for all AuNP, based on

Equation 3.4, the effective surface area is always proportional to the square of the particle size (r). After the effective surface area is determined, the effective number and effective molecular weight of PEG can be calculated knowing the capping density.

2 휃 2 푆퐴푒푓푓 = 2휋푟 (1 − 푐표푠 ⁄2), SAeff ∝ SA ∝ r (3.4)

Nonetheless, the SAeff is always proportional to the total SA of a nanoparticle. For different sized AuNPs, the Neff depends on both the capping density and the surface area of the nanoparticle:

109

푁푒푓푓 = 푑푆퐴푒푓푓 (3.5)

By plotting the retention order of four different sized AuNPs as a function of log (Neff), it’s found that the AuNPs have stronger retention with an increase in the effective number of PEG (Figure 3.13A). The obtained linear relationship (R2=0.9961) demonstrates that the developed method can be used to determine either the size of an AuNP with a known capping density or the capping density of PEG with a known size of the nanoparticle.

Hence, this method can potentially be used to isolate AuNPs when synthetic methods produce a sample with a distribution of different sizes or capping densities.

Similarly, the confirmation of PEG with different molecular weight on the 30 nm

AuNPs was also determined to be “brush” regime. The effective numbers of PEG with different molecular weight on the 30 nm AuNP were also determined using the calculations above (Table 3.4). Because of the difference in molecular weight, only considering the effective number of PEG is not conclusive in determining their relationship with the retention order. Therefore, we adopted effective molecular weight of

PEG using the following equation:

푀푊푒푓푓 = 푁푒푓푓푀푊 (3.6)

As shown in Figure 3.13B, with the same core size of AuNPs, the retention of AuNPs increased with an increase in the effective molecular weight of the grated PEG. The plot

2 of retardation factor versus the log (Meff) shows a good linearity (R =0.9983), which indicates that this method can be applicable to the determination of the molecular weight of the grafted PEG on a certain sized nanoparticle with a known capping density.

110

Figure 3.12 Scheme of effective interaction between grafted PEG on AuNP and the ligands on the nanofiber stationary phase.

111

Table 3.3 Summary of Flory radius (RF), layer thickness (L), PEG-PEG distance (D),

Area per molecule (A), density (d) and effective number of PEG (Neff) for different sized

PEGylated AuNPs.

2 -2 Au-PEG (nm) RF (nm) L (nm) D (nm) A (nm ) d (nm ) Neff 10 6.0 16.6 1.3 1.3 0.76 59.3 30 6.0 13.7 1.7 2.3 0.43 301.1 50 6.0 14.3 1.6 2.1 0.48 941.7 80 6.0 15.5 1.4 1.6 0.62 3103.2

Table 3.4 Summary of Flory radius (RF), layer thickness (L), PEG-PEG distance (D),

Area per molecule (A), density (d), effective number (Neff) and molecular weight of PEG

(MWeff) for different sized PEGylated AuNPs.

Au-PEG 2 -2 MWeff RF (nm) L (nm) D (nm) A (nm ) d (nm ) Neff (kDa) (kDa) 2 3.4 3.9 2.9 6.4 0.16 110.0 219.5

5 6.0 8.9 3.3 8.6 0.12 82.4 411.8

10 9.1 17.0 3.6 9.9 0.10 71.1 710.9

20 13.8 33.0 3.7 10.8 0.09 65.5 1309.1

112

Figure 3.13 Plot of (A) Retardation factor vs. Log (Neff) for different sized Au-PEG with

2 R =0.996; (B) Retardation factor vs. Log (MWeff) for AuNPs capped with different molecular weight of PEG, R2 = 0.998.

113

3.3.6 Chromatographic Performance of Electrospun UTLC

The chromatographic performance of electropsun ULTC was evaluated in terms of plate height (H), resolution (Rs) and development time (t). The plate number (N) describes the separation efficiency of TLC as it is related to the migration distance (Zs) and the spot width (w) of the sample:

푍 2 푁 = 16 ( 푠) (3.7) 푤

H can be obtained by using the equation H = Zs/N. Therefore, H is inversely proportional to N and directly proportional to w. The goal of efforts to minimize band broadening in

TLC is to obtain small H values.49 In this work, the changes in H of each PEGylated

AuNPs were examined as a function of solvent migration distance on electrospun PAN

UTLC. As shown in Figure 6A, the H values remained approximately the same with increasing distance from 5 to 25 mm for all AuNPs except 80 nm AuNP. The H values of all AuNPs increased significantly at longer migration distance, which were mainly attributed to the longitudinal diffusion, because the mobile phase velocity gradually slowed down as it migrated up on the plate due to the capillary driven force in TLC.49

Based on Figure 3.14A, the smallest H value can be achieved was ~1.2 µm for 10 nm

AuNP, and H for all AuNPs were below 10 µm when the migration distance was within

20 mm. Migration distance exceeding 25 mm could result in band broadening to some extent. However, H itself cannot adequately tell the appropriate migration distance for a good separation to occur. The resolution, Rs, which defines how well two neighboring substances could be differentiated, should also be taken into consideration:

114

푍푠2 − 푍푠1 푅푠 = 2 (3.8) 푤1 + 푤2

A large Rs value, which indicates a high chromatographic selectivity, results from the large difference between migrated distances of analytes 1 and 2 (Zs1 and Zs2) and their

20 small spot widths (w1 and w2). Thus, the Rs for each neighboring pair of AuNPs as the function of mobile phase migration distance was also investigated. As shown in Figure

3.14B, when mobile phase traveled further than 25 mm, the Rs for each pair of AuNPs started to decrease. Corresponding to the H study, band broadening observed after 25 mm migration distance could contribute to overlapping of sample spots. Notably, with only 15 mm migration distance of the mobile phase, each pair of AuNPs could be separated with

Rs value > 1.5 and up to 3.1. As previously stated, band broadening was observed at 25 mm migration distance or longer for all AuNPs. Consequently, 15 mm solvent migration distance was used for all the separations in this work. The H and Rs for AuNPs capped with different molecular weight of PEG were also measured under optimized condition and listed in Table 3.5. Electrospun PAN UTLC exhibited comparable chromatographic performance for PEGylated AuNPs from both categories (size and molecular weight).

115

Figure 3.14 Separation efficiency: (A) Plate height of 10 nm, 30 nm, 50 nm and 80 nm

116

Table 3.5 Retardation factor, plate height and resolution of PEGylated AuNPs with different molecular weight of PEG.

Au-PEG (Different MW)

MW of PEG Rf H (µm)

(kDa) (%RSD, n=3) (%RSD, n=3)

2 0.81 (1.2) 3.4 (19) 5 0.70 (1.4) 4.2 (17) 10 0.63 (1.1) 4.6 (4.5) 20 0.53 (1.7) 6.5 (3.2)

Rs (%RSD, n=3)

2/5 1.7 (12) 5/10 1.5 (9.0) 10/20 1.7 (16)

117

To evaluate the speed of the separation, mobile phase velocity was studied using the following equation:

2 푍푓 = 휅푡 (3.9) where, Zf is the total distance traveled by the mobile phase from the origin, к is the

50 2 velocity constant, and t is the development time. By plotting Zf versus t (Figure 3.12C), a straight line was obtained, and the mobile phase velocity can be determined as the slope. Deviations from the straight line suggest abnormal flow the mobile phase due to experimental errors or the heterogeneity of the stationary phase.51 Figure 3.14C shows

2 2 linearity (R > 0.99) between Zf and t, which demonstrated the homogeneity of the electrospun PAN stationary phase. On the electrospun PAN UTLC, solvent migration of

15 mm required less than 5 min (285 s) to achieve. In current literature, SEC and CE required at least 8-10 min and HPTLC took 20 min to separate AuNPs with two or three different sizes;19,23,27 FFF required 40 min and GE took 4 h to separate AuNPs coated with different length of PEG.22,25 In comparison, this developed method using electrospun UTLC offers much faster separations with excellent chromatographic performance.

118

3.3.7 Monitoring the Transformation of AuNPs in Serum Protein

When AuNPs are introduced into bloodstream, proteins rapidly bind to the surface of AuNPs to form a biological coating, known as protein corona (PC). The formation of

PC can affect the biological identity of the AuNPs and change the structure of adsorbed proteins, which may further eliminate their physiological functions.52,53 In recent years, a large number of studies have devoted to characterizing AuNP-PC complexes in order to further investigate their impact on physiological systems.15,26,54 In this work, we applied the developed method using electrospun UTLC to track the transformation of AuNPs in serum protein. Bovine serum albumin (BSA) is the most abundant protein in bovine plasma and highly homologous to its human counterpart (HSA). To monitor the protein adsorption on nanoparticles, 10 nm PEGlyated AuNPs (Au-PEG) and 10 nm citrated stabilized AuNPs (Au-citrate) were incubated with 5 mg/mL bovine serum albumin

(BSA) at 37 °C for 1 h. The 5 mg/mL of BSA was prepared in phosphate buffer saline

(PBS) to mimic the physiological condition. After incubation, without washing, Au-PEG and Au-citrate were spotted on the UTLC plates, and non-incubated AuNPs were also spotted as standard references. As shown in Figure 3.15A (spot a, b, d, e) significant decay in migration distance was observed for both Au-PEG and Au-citrate after incubation, which indicates the increase in their hydrodynamic sizes and BSA adsorption.

To confirm the stability of the protein corona, incubated Au-PEG and Au-citrated were washed three times with Milli-Q water. After washing, according to Figure 3.15A (spot a, c, d, f), Au-PEG migrated the same distance for both incubated and non-incubated NPs, whereas Au-citrated still migrated to a shorter distance than its non-incubated 119 counterpart. This observation can be illustrated using the two different types of protein corona: hard corona and soft corona. Proteins that tightly bounded on the surface of NPs are hard corona, while proteins loosely bounded on the surface are soft corona.55 Soft corona is secondary outer layer and has weak interaction with hard corona, therefore, it is dynamic and can be rinsed off or replaced by proteins with higher affinity to AuNPs over time.56 For Au-citrate, it is believed that there were both hard and soft corona on the surface, and the stable hard corona still presented after washing to give an increased hydrodynamic size of AuNPs. For Au-PEG, the particle size remained stable even when the incubation period was extended to one week (Figure 3.15B). It is well accepted that there is only soft corona covering the surface of PEGylated AuNPs, and no hard corona is observed because thiol terminated PEG creates a protective barrier on AuNPs to resist protein adsorption.25,57 Hence, PEGylation is now the most widely applied strategy to block nonspecific protein adsorption on AuNPs and increase their biocompatibility.58

Using the developed method, electrospun PAN UTLC can be applied to monitor the size transformation of AuNPs in a biological environment.

120

121

3.4 Conclusions

To the best of our knowledge, this is the first use of UTLC for the separation of nanoparticles. PEGylated AuNPs with different sizes in the range of 10-80 nm were well separated from each other. It demonstrated that the Rf value increases with decreasing size of AuNPs. This method also permits the separation of AuNPs capped with different molecular weight of PEG in the range of 2-20 kDa. The results indicated that the Rf value increases with decreasing length of the PEG coating. Micellar mobile phases were adopted to provide a highly biodegradable chromatographic system. This novel method exhibits excellent separation performance with the smallest plate heights < 2 µm and resolution of each pair of AuNPs > 1.5. Electrospun PAN UTLC shows uniform morphology and allows fast mobile phase migration velocity. Separations for all

PEGylated AuNPs could be achieved within 5 min. This method was applied to monitor the transformation of AuNPs in serum protein and provided clear evidence of size changes for Au-citrate and the stability of Au-PEG.

122

3.5 References

1 Huang, D.; Liao, F.; Molesa, S.; Redinger, D.; Subramanian, V. J. Electrochem. Soc. 2003, 150, G412-G417.

2 Zhou, Q.; Liu, N.; Qie, Z.; Wang, Y.; Ning, B.; Gao, Z. J. Agric. Food Chem. 2011, 59, 12006-12011.

3 Raj, S.; Jose, S.; Sumod, U. S.; Sabitha, M. J. Pharm. Bioall. Sci. 2012, 4, 186–193.

4 Hutter, E.; Maysinger, D. Microsc. Res. Tech. 2011, 74, 592-604. 5 Ghosh, P.; Han, G.; De, M.; Kim, C. K.; Rotello, V. M. Adv. Drug Deliv. Rev. 2008, 60, 1307-1315.

6 Almeida, J. P. M.; Figueroa, E. R.; Drezek, R. A. Nanomedicine 2014, 10, 503–514.

7 Jiang, Y.; Huo, S.; Mizuhara, T.; Das, R.; Lee, Y.; Hou, S.; Moyano, D. F.; Duncan, B.; Liang, X. J.; Rotello, V. M. ACS Nano 2015, 9, 9986–9993.

8 Albanese, A.; Chan, W. C. ACS Nano 2011, 5, 5478-89.

9 Zhang, X. D.; Wu, D.; Shen, X.; Liu, P. X.; Yang, N.; Zhao, B.; Zhang, H.; Sun, Y. M.; Zhang, L. A.; Fan, F. Y. Int. J. Nanomedicine 2011, 6, 2071-2081.

10 Eck, W.; Craig, G.; Sigdel, A.; Ritter, G.; Old, L. J.; Tang, L.; Brennan, M. F.; Allen, P. J.; Mason, M. D. ACS Nano 2008, 11, 2263–2272.

11 Lipka, J.; Semmler-Behnke, M.; Sperling, R. A.; Wenk, A.; Takenaka, S.; Schleh, C.; Kissel, T.; Parak, W. J.; Kreyling, W. G. Biomaterials 2010, 31, 6574-6581.

12 Albanese, A.; Tang, P. S.; Chan, W. C. W. Annu. Rev. Biomed. Eng. 2012, 14, 1-16.

13 Alkilany, A. M.; Murphy, C. J. J. Nanopart. Res. 2010, 12, 2313–2333.

14 Pan, Y.; Leifert, A.; Ruau, D.; Neuss, S.; Bornemann, J.; Schmid, G.; Brandau, W.; Simon, U.; Jahnen-Dechent, W. Small 2009, 5, 2067–2076.

15 Pan, Y.; Neuss, S.; Leifert, A.; Fischler, M.; Wen, F. Small 2007, 3, 1941 – 1949.

16 Valcárcel, M.; López-Lorente, Á. Gold Nanoparticles in Separation Science, 1st ed.; Elsevier: Amsterdam, Netherlands, 2014.

17 Wei, G. T.; Liu, F. K. J. Chromatogr. A 1999, 836, 253–260. 123

18 Hassellov, M.; Readman, J.; Ranville, J.; Tiede, K. Ecotoxicology 2008, 17, 344-361.

19 Zhou, X.; Liu, J.; Jiang, G. Environ. Sci. Technol. 2017, 51, 3892–3901.

20 Pergantis, S. A.; Jones-Lepp, T. L.; Heithmar, E. M. Anal. Chem. 2012, 84, 6454– 6462.

21 Schmidt, B.; Loeschner, K.; Hadrup, N.; Mortensen, A.; Sloth, J. J.; Koch, C. B.; Larsen, E. H. Anal. Chem. 2011, 83, 2461–2468.

22 Hansen, M.; Smith, M. C.; Crist, R. M.; Clogston, J. D.; McNeil, S. E. Anal. Bioanal. Chem. 2015, 407, 8661-8672.

23 Franze, B.; Engelhard, C. Anal. Chem. 2014, 86, 5713-5720.

24 López-Lorente, Á. I.; Soriano, M. L.; Valcárcel, M. Microchimica Acta 2014, 181, 1789–1796.

25 Doane, T. L.; Cheng, Y.; Babar, A.; Hill, R. J.; Burda, C. J. Am. Chem. Soc. 2010, 132, 15624-15631.

26 Harrison, E.; Nicol, J. R.; Macias–Montero, M.; Burke, G. A.; Coulter, J. A.; Meenan, B. J.; Dixona, D. Mater. Sci. Eng. C 2016, 62, 710–718.

27 Yan, N.; Zhu, Z.; Jin, L.; Wei, G.; Gan, Y.; Hu, S. Anal. Chem. 2015, 87, 6079-6087.

28 Yan, N.; Zhu, Z.; He, D.; Jin, L.; Zheng, H.; Hu, S. Sci. Rep. 2016, 6, 24577, doi: 10.1038/srep24577.

29 Huang, Z. M.; Zhang, Y. Z.; Kotaki, M.; Ramakrishna, S. Compos. Sci. Technol. 2003, 63, 2223–2253.

30 Clark, J. E.; Olesik, S. V. Anal. Chem. 2009, 81, 4121-4129.

31 Fang, X.; Olesik, S. V. Anal. Chim. Acta 2014, 830, 1-10.

32 Wang, Y.; Olesik, S. V. Anal. Chim. Acta 2017, 970, 82-90.

33 Smith, M. C.; Crist, R. M.; Clogston, J. D.; McNeil, S. E. Anal. Bioanal. Chem. 2015, 407, 3705-3716.

34 Goddard, E. D. Colloids Surf. 1986, 19, 301-329.

35 Dai, S.; Tama, K. C. Physicochem. Eng. Aspects 2006, 289, 200–206.

36 Cao, M.; Hai, M. J. Chem. Eng. Data 2006, 51, 1576-1581. 124

37 Schwuger, M. J. J. Colloid Interface Sci. 1973, 43, 491-498.

38 Manna, K.; Panda, A. K. J. Surfact. Deterg. 2011, 14, 563-576.

39 Berthod, A.; Garcia-Alvarez-Coque, C. Micellar Liquid Chromatography, 1st ed.; Taylor & Francis: Oxfordshire, UK, 2000.

40 Fajalia, A. I.; Tsianou, M. J. Phys. Chem. B 2014, 118, 10725−10739.

41 Guo, W.; Sun, Y. W.; Luo, G. S.; Wang, Y. J. Colloids Surf., A 2005, 252, 71–77.

42 Rahme, K.; Chen, L.; Hobbs, R. G.; Morris, M. A.; O'Driscoll, C.; Holmes, J. D. RSC Adv. 2013, 3, 6085-6094.

43 Rayford, C. E.; Schatz, G.; Shuford, K. Nanoscape 2005, 2, 27-33.

44 Bekale, L.; Agudelo, D.; Tajmir-Riahi, H. A. Colloids Surf., B 2015, 130, 141–148.

45 Ruiz-Angel, M. J.; Carda-Broch, S.; Torres-Lapasio, J. R.; Garcia-Alvarez-Coque, M. C. J. Chromatogr. A 2009, 1216, 1798-1814.

46 Hammouda, B.; Ho, D. L. J. Polym.Sci., Part B: Polym. Phys. 2007, 45, 2196-2200.

47 deGennes, P. G. Macromolecules 1980, 13, 1069-1075.

48 Perry, J. L.; Reuter, K. G.; Kai, M. P.; Herlihy, K. P.; Jones, S. W.; Luft, J. C.; Napier, M.; Bear, J. E.; DeSimone, J. M. Nano Lett. 2012, 12, 5304-5310.

49 Sherma, J.; Fried, B. Handbook of Thin-Layer Chromatography, 3rd ed.; Mercel Dekker: New York, 2013.

50 Acikara, O. B. Ion-Exchange Chromatography and Its Applications, Column Chromatography; Dr. Dean Martin ed.; InTech, 2013, DOI: 10.5772/55744.

51 Geiss, F. Fundamentals of Thin Layer Chromatography (Planar Chromatography); Huethig: Heidelberg, 1987.

52 Tenzer, S.; Docter, D.; Kuharev, J.; Musyanovych, A.; Fetz, V.; Hecht, R.; Schlenk, F.; Fischer, D.; Kiouptsi, K.; Reinhardt, C.; Landfester, K.; Schild, H.; Maskos, M.; Knauer, S. K.; Stauber, R. H. Nat. Nanotechnol 2013, 8, 772-781.

53 Wang, P.; Wang, X.; Wang, L.; Hou, X.; Liu, W.; Chen, C. Sci. Technol. Adv. Mater. 2015, 16, 034610.

54 Dobrovolskaia, M. A.; Patri, A. K.; Zheng, J.; Clogston, J. D.; Ayub, N.; Aggarwal, P.; Neun, B. W.; Hall, J. B.; McNeil, S. E. Nanomedicine 2009, 5, 106-117. 125

55 Rahman, M.; Laurent, S.; Tawil, N.; Yahia, L.; Mahmoudi, M. Protein-Nanoparticle interactions: The bio-nano interface; Springer: New York, 2013.

56 Kari, O. K.; Rojalin, T.; Salmaso, S.; Barattin, M.; Jarva, H.; Meri, S.; Yliperttula, M.; Viitala, T.; Urtti, A. Drug Deliv. and Transl. Res. 2017, 7, 228-240.

57 Walkey, C. D.; Olsen, J. B.; Guo, H.; Emili, A.; Chan, W. C. W. J. Am. Chem. Soc. 2012, 134, 2139-2147.

58 Jokerst, J. V.; Lobovkina, T.; Zare, R. N.; Gambhir. S. S. Nanomedicine (Lond). 2011, 6, 715-728.

126

Chapter 4. Enhanced-Fluidity Liquid Chromatography-Mass Spectrometry for

Intact Protein Separation and Characterization

4.1 Introduction

The emergence of top down proteomics allows the characterization of intact proteins and the examination of protein post-translational modifications (PTMs).1,2

However, unlike the well-established bottom-up strategy for protein analysis, top down proteomics faces many technical challenges to be considered as a routine approach for the study of proteomes. Due to the complexity of proteome, one of the challenges is the separation of intact proteins.3,4

Liquid chromatography (LC) is widely used for analysis of biomolecules.5,6,7

Proteins are separated using LC based on several properties: charge, size, bio-specific interaction and hydrophobicity. LC with volatile mobile phases which are compatible with electrospray mass spectrometry (ESI-MS) is preferred for the separation of complex biological mixtures. Ion exchange chromatography (IEX) has been directly coupled to

MS for purification of proteins.8,9 However, IEX is limited by poor selectivity, therefore multi-dimensional separation is often required for complex protein mixtures.10 In size exclusion chromatography (SEC), biological activities of a protein can be highly preserved because there’s no interaction between solutes and the stationary phase, but

127

SEC gives the least resolution among LC techniques.11 Affinity chromatography separates proteins based on their reversible interactions with ligands on the column matrix hence it offers extremely high specificity and high purification power.12 Despite such promises, expensive biospecific ligands, complex ligand coupling techniques and the leakage of ligands hinder this separation method from routine applications.13 Both reversed phase chromatography (RPC) and hydrophobic interaction chromatography

(HIC) separate proteins based on their hydrophobicity. Even though RPC is still one of the most popular technique for protein separation, the large proportion of organic solvent in the mobile phase generally leads to denaturization of proteins which causes poor protein recovery and severe peak broadening.14,15 Alternatively, proteins can retain their conformational structures and biological activities using hydrophobic interaction chromatography, HIC, which involves mild binding interactions with the stationary phase and elution by using a salt gradient.16 Recent work by Chen et. al highlighted the possibility of online coupling of HIC with MS using ammonium acetate under LC condition.17

Recent advances in enhanced fluidity liquid chromatography (EFLC) demonstrated the improvements in separation efficiency and speed of analysis for biological compounds such as amino acids,18 nucleosides and nucleotides,19 and oligosaccharides.20 In addition, Bennett et. al illustrated the feasibility of EFLC for separation of peptides and proteins by hydrophilic interaction chromatography (HILIC), in which, a large portion of organic solvent was required for adequate retention.21

Compared to conventional LC, EFLC uses carbon dioxide (CO2) as a modifier in the

128 mobile phase to provide enhanced diffusivity, faster solute mass transfer, and reduced system backpressure.

In this work, EFLC mobile phases are applied to separate proteins by online HIC-

MS for the first time. The effect of liquid CO2 on the retention factor, separation and ionization efficiency of several proteins and protein complexes was also evaluated.

4.2 Materials and Methods

4.2.1 Chemicals and Reagents

Ammonium acetate (for molecular biology, ≥98%) and protein standards including ribonuclease A from bovine pancreas (RNaseA), trypsin inhibitor from glycine max (soybean) (TI), α-chymotrypsin from bovine pancreas (Chy), α-chymotrypsinogen A from bovine pancreas (ChA), lysozyme from chicken egg white (Lys) and concanavalin

A from canavalia ensiformis (jack bean) (ConA) were purchased from Sigma-Aldrich (St.

Louis, MO). Supercritical fluid extraction grade CO2 (99.999%) was obtained from

Praxair, Inc. (Danbury, CT). LC-MS grade methanol (MeOH), acetonitrile (ACN) were purchased from Fisher Scientific (Fair Lawn, NJ). Water was purified to 18.3 MΩ by a

Barnstead Nanopure Infinity System from Thermal Scientific Inc. (Odessa, TX).

129

4.2.2 Sample Preparation

Standard protein samples RNase A, TI, Chy, ChA and Lys were prepared without additional purification. Nanopure water was used to prepare 1 M ammonium acetate

(AmAc) solution, and proteins were subsequently dissolved in 1 M AmAc at 1mg/mL for

RNase A, ChA and Lys, and 2 mg/mL for TI and Chy, followed by syringe filtration using 0.2 µm Captiva syringe filter (Agilent Technologies, Santa Clara, CA). Protein complex ConA was prepared following the same procedure at 4 mg/mL in 1 M AmAc.

4.2.3 EFLC-MS

An Agilent 1260 Infinity HPLC system (Agilent Technologies, Santa Clara, CA) was used to perform HIC salt gradient analysis. A hybrid of an Agilent 1260 Infinity

Analytical HPLC and an Agilent 1260 Infinity Analytical SFC was adapted to perform dual gradient elution (CO2 gradient on top of the HIC salt gradient) as shown in Figure

4.1. The hybrid system included an Aurora SFC Fusion ™ A5 module with maximum

400 bar pressure limit, two 1260 HiP degasser, a 1260 SFC binary pump and a 1260

HPLC binary pump, a 1260 autosampler, 1290 thermostatted column compartment, a

1200 Diode Array Detector (DAD), and an Agilent 6530 ESI-Q-TOF mass spectrometer.

The back-pressure regulator (BPR) was set to 100 bar and the temperature to 50 °C to maintain the backpressure over the system. A splitter assembly was connected to the outlet of the DAD to split flow between the BPR and the MS. Both PolyBUTYL A and

PolyPENTYL A HIC columns (100 mm x 4.6 mm i.d., 3 µm, 1500 Å, Poly-LC) were

130 evaluated for the separation of protein mixture. The HIC salt gradient was delivered through the HPLC binary pump with 1 M AmAc as the mobile phase A and various percentage of organic solvent in 20 mM AmAc as the mobile phase B. The eluting proteins were allowed to pass through DAD and subsequently electrosprayed into a Q-

TOF mass spectrometer.

4.2.4 Data Acquisition and Analysis

The DAD was set to a wavelength of 280 nm with a bandwidth of 4 nm, a sampling rate of 2.5 Hz, and a background wavelength of 360 nm with a 100 nm bandwidth. All mass spectra were collected in positive ion mode. The capillary voltage was optimized at 3500V. The nebulizer was set to 35 psig. The drying gas was set to 8

L/min at 300 °C. The sheath temperature was at 300 °C degree, unless otherwise noted.

All data were collected using LC/MS Data Acquisition Version B 06.01 in Agilent

MassHunter Workstation Software. All data were analyzed using Qualitative Analysis

Version B 07.00 in Agilent MassHunter Workstation Software. PeakFit Version 4 software (SPSS Inc. Chicago, IL) was used to perform gaussian fits to chromatographic peaks. mMass Version 5.5.0 software was used to deconvolute protein mass spectra.

131

Figure 4.1 Ternary pump setup used in this work to run EFLC dual gradient.

132

4.3 Results and Discussion

4.3.1 Characterization of the HIC Column for Protein Separation

In conventional HIC method, nonvolatile salts such as potassium chloride, ammonium sulfate and ammonium tartrate are generally used to induce “salt out” effect of proteins. However, these nonvolatile salts can strongly interfere with the ionization process and accumulate in the ion source and ion transfer system to affect the MS performance.22 Hence, conventional HIC is often performed offline, and a buffer exchange or a desalting procedure is necessary prior to MS analysis. Previous studies were reported using volatile salts such as ammonium acetate on HIC columns to separate proteins.23,24,25 Even though the use of volatile salts could potentially simplify the desalting procedures, in these studies, 2 to 4 M of ammonium acetate was required to achieve adequate retention of proteins, which was not in the MS compatible range.

Recently, Chen et al. have demonstrated the feasibility of online HIC-MS using a series of HIC materials with more hydrophobic ligands such as butyl, pentyl, hexyl and heptyl.17 In their study, 1M or less of ammonium acetate, was used which is compatible with ESI-MS to retain and elute proteins.17 The retention mechanism for these columns under these conditions was a HIC-RPC mixed mode interaction as some organic solvent had to be added into the salt gradient to promote protein elution in a reasonable time frame while maintaining the low charge states of eluted proteins. This mix-mode feature is desirable for our investigation of EFLC-MS system for proteins because in order to add

133 liquefied CO2 to HIC salt gradient, an organic cosolvent (such as methanol, ethanol, acetonitrile) is required to avoid solvent demixing.18

The two HIC columns were evaluated first under conventional HPLC condition to study optimum retention and selectivity of protein mixture. Commonly-used organic solvents in RPC such as acetonitrile (ACN) and methanol (MeOH) were added to low salt solution in mobile phase B at various percentage to examine their elution strength and selectivity. A salt gradient between (1 M to 20 mM) was acceptable for both

PolyPENTYL A and Poly BUTYL A. As expected, PolyPENTYL A provided stronger retention than PolyBUTYL A for all proteins except for RNase A which eluted at dead time under all evaluated conditions (Figure 4.2). On both columns, 60 v% MeOH showed similar solvent strength to 50 v% ACN. However, ACN provided better selectivity for the more retained proteins Lys and ChA on PolyBUTYL A column whereas MeOH gave better selectivity than ACN on PolyPENTYL A column. The retention order changed for

Lys and ChA from PolyBUTYL A to PolyPENTYL A as shown in Figure 4.2. This could be due to the higher accessibility of sequestered hydrophobic domain on Lys than ChA as the ligand length increased on the HIC stationary phase. PolyPENTYL A column was then selected for EFLC-MS study using MeOH as the cosolvent for the following reasons. MeOH is a greener and cheaper substitute for ACN; MeOH/H2O mixtures are

21 miscible with greater amounts of liquefied CO2 than ACN/H2O mixtures; Gradients containing MeOH provided the best separation for protein mixture on PolyPENTYL A.

134

Figure 4.2 HPLC separations of protein mixture: 1. RNase A, 2. TI, 3. Chy, 4. Lys, 5.

ChA on PolyBUTYL A column (A, B, C, D) and PolyPENTYL A column (E, F, G, H).

MPA: 1 M ammonium acetate for all conditions. MPB: organic solvent in 20 mM ammonium acetate. Organic solvent in MPB: (A, E)-50% ACN; (B, F)-50% MeOH; (C,

G)-60% MeOH; (D, H)-70% MeOH. Gradient: 10 min MPA to MPB at 25 °C. Flow rate:

1 mL/min.

135

4.3.2 Addition of Liquefied CO2 to Isocratic Elution

To examine the effect of liquefied CO2 on the protein retention, charge state distribution and ionization efficiency of proteins, a series of isocratic elution experiments was carried out using Lys as the model protein because it was more strongly retained than other proteins on PolyPENTYL A. Under isocratic condition, with other mobile phase composition maintained constant, the effect of liquefied CO2 can be easily observed.

During this study, a fixed amount of CO2 from 0-15 v % in 2.5 v% increment was added to 300 mM AmAc with 70 v% MeOH. As shown in Figure 4.3A, Lys became less retained with a narrower and more symmetric peak shape as the volume percent of CO2 in the mobile phase increased. This indicated that CO2 acted as a strong eluent in HIC-

RPC retention mechanism due to its hydrophobic nature. Nonetheless, isocratic elution was not preferred for protein separation as peak tailing was observed regardless the amount of CO2 added into the MeOH/AmAc mixture. This was most likely due to the large amount of MeOH in the mobile phase from the starting condition which could lead to some protein denaturization and prolonged interaction between denatured Lys and the hydrophobic ligands.

In HPLC, the system backpressure (∆P) is directly proportional to solvent viscosity (η), which changes with the mobile phase composition. MeOH-aqueous mixtures have higher viscosity than ACN-aqueous mixtures, therefore under the same column condition including column length (L), column radius (r), particle size (dp) and

136 permeability (K0), the application of MeOH-aqueous mixture as the mobile phase is restricted at a relatively low flow rate (F) compared to ACN-aqueous mixture.

ηFL ΔP = 0 2 2 (4.1) K πr dp

In EFLC, the addition of liquefied CO2 to MeOH-aqueous mixture lowers the backpressure drop by decreasing the overall viscosity of the mobile phase.26 As shown in

Figure 4.4, while keeping other mobile phase composition (300 mM AmAc with 70 v%

MeOH) and the flow rate constant (1mL/min), the system backpressure decreased linearly with increasing amount of liquefied CO2. This ensured that higher flow rates could be performed on the same instrumentation with stronger elution strength and better peak symmetry, promoting a short analysis time.

As mentioned earlier, isocratic elution of lysozyme only served as a model to directly examine the effect of liquefied CO2. It is not preferred for separation of protein mixtures with different polarities due to broad peak shapes and reduced sensitivity for retained analytes, and poor selectivity for less retained analytes.

137

Mass spectrum of Lys with 0% CO2 in the mobile phase. Flow rate: 1 mL/min.

138

Figure 4.4 Effect of the addition of liquified CO2 on the system back pressure. Mobile phase: various amount of CO2 in 300 mM ammonium acetate with 70 v% MeOH. Flow rate: 1 mL/min.

139

4.3.3 EFLC Dual Gradient Elution

4.3.3.1 Effect of CO2 on Chromatographic Performance

Many problems associated with isocratic analysis for proteins can be overcome by using gradient elution. As discussed in section 4.3.1, gradient elution using 1 M AmAc to various volume percentage of MeOH exhibited good separations for protein mixtures on

PolyPENTYL A column. Herein, EFLC was compared to optimized LC gradient condition. Figures 4.5A shows the optimized separation of the protein mixture using a LC solvent gradient from 100 v% 1 M AmAc to 100 v% MeOH in 10 min. The addition of liquefied CO2 was performed based on the optimized LC method as in Figure 4.5A. Due to the limited miscibility of CO2 in aqueous solution, liquefied CO2 was added in low proportions (1 to 5 v%) in the first 6 min of the solvent gradient (<60 v% MeOH). As the

MeOH content increased in the mobile phase, more CO2 was introduced to the solvent mixture. As shown in Figure 4.5B, the dual gradient was carried out with conventional

LC solvent changing from 1 M AmAc to 100 v% MeOH in 10 min; liquefied CO2 from

1-5 v% was added from 0-6 min, and after 6 min, the amount of liquefied CO2 was increased linearly to 30 v% till 10 min. Comparing Figure 4.5A and B, the retention time of Chy, ChA and Lys decreased with the addition of liquefied CO2, which was in consistency with the observation for Lys in the isocratic condition. Therefore, in the dual gradient elution, liquefied CO2 also enhanced the solvent strength of the mobile phase mixture. It’s worth noting that in addition to the reduced retention, the peak shapes of

Chy, ChA and Lys were drastically improved. Especially for the ChA and Lys, the 140 extracted ion chromatographic peaks (Figure 4.6) under dual gradient elution showed

2.55 ± 0.02 times decrease in peak width compared to the conventional LC gradient. This could be ascribed to the high diffusivity of the EFLC solvent, which leads to fast mass transfer. As shown in Equation 4.2, where u is the linear velocity, the plate height (H) depends on the multi-path diffusion (A), longitudinal diffusion (B) and the resistance to

27 the mass transfer between mobile phase (Cm) and the stationary phase (Cs). According to Equation 4.3, where ω is a constant, when the particle diameter (dp) is fixed, the Cm term is inversely proportional to the diffusion coefficient of the mobile phase. Higher diffusivity leads to lower resistance to mass transfer which diminishes band broadening.

In other words, a faster mass transfer using EFLC can generate lower plate height and higher efficiency.

B H = A + + (C + C )u̅ (4.2) u̅ m s

2 ωdp Cm = (4.3) Dm

ChA and Lys peaks fitted well to a Gaussian distribution for both conventional

LC and EFLC dual gradients. For gradient separtions, peak capacity (nc) can be

28 calculated based on the gradient time (tg) and the peak width at base (wb= 4σ):

tg nc = 1 + (4.4) wb

For a Gaussian function, the full width at half maximum (FWHM=2.35σ) can be used to give equal estimation of peak capacity:

tg n = 1 + (4.5) c 1.7 FWHM 141

Using Eq. 4.5, for both ChA and Lys in a 10 min gradient separation, the peak capacity of EFLC dual gradient was 31, which was ~2.5 folds higher than the peak capacity of conventional LC gradient (nc=13). Therefore, the addition of liquefied CO2 in the EFLC dual gradient provided significant advantages in separation efficiency and peak capacity. The improvements of analysis speed and plate number were most notable for the more retained analytes.

142

Figure 4.5 DAD chromatogram at 280 nm of protein mixture separation on PolyPENTYL

3. Chy, 4. ChA, 5. Lys.

143

Figure 4.6 A comparison of HPLC gradient (A, C) and EFLC dual gradient (B, D) for chromatographic performances of ChA (A, B) and Lys (C, D): (o) data, (-) gaussian fits.

144

4.3.3.2 Sensitivity of EFLC Method to Minor Modifications

Chy was eluted in two peaks on PolyPENTYL A. Especially in the dual gradient elution, the sensitivity of these two peaks of Chy was improved with the addition of liquefied CO2 (Figure 4.5). As shown in Figure 4.7A, the deconvoluted mass of the first eluted peak had a 16 Da mass difference from the second peak. Therefore, the early eluted peak was the oxidized form of Chy. In Figure 4.7B, a similar elution pattern was observed for ChA after two weeks of storage at 4° C or one week of storage at room temperature, and the oxidized ChA peak slowly overtook the original ChA peak during storage. The targets of protein oxidation could be the protein backbones or the amino acid side chains.29 Oxidative attack on the protein backbones would lead to protein fragmentation, and the oxidation of amino acid side chains would result in the formation of various oxidation products.30 The oxidized Chy and ChA observed in this study were both in their intact molecular weights, thus they were the products of the side-chain oxidation. Among all the amino acid residues, cysteine and methionine are particularly sensitive to oxidation by a wide range of oxidants, and the oxidation of these residues are often protein inactivating modification leading to impaired protein functions.31,32

Therefore, it is important to monitor the oxidative modification of proteins. This EFLC-

MS method can be employed as a simple and sensitive strategy to identify oxidative damage of intact proteins.

145

Figure 4.7 EFLC-MS of Chy (A) and ChA (B) showing chromatographically separated proteoforms with minor modifications under dual gradient condition as described in

Figure 4.5C. Inset shows the mass spectra of two peaks (a1, a2 and b1, b2) of Chy and

ChA, respectively. The deconvoluted spectrum showing a 16 Da difference.

146

4.3.3.3 Effect of CO2 on CSD and Ionization Efficiency

The impact of CO2 on the proteins’ ionization during the ESI process was examined, and the ESI source parameters were kept constant for direct comparison. As mentioned earlier, in the isocratic elution, the large amount of MeOH in the mobile phase could lead to protein denaturization, surprisingly, the mass spectrum of the eluted Lys peak using 300 mM AmAc with 70 v% MeOH still exhibited relative narrow CSD in low charge states. which corresponded to the characteristic of the mass spectrum of a nondenatured protein (Figure 4.8A). Upon the addition of 2.5 v% CO2, the charge states of Lys immediately shifted to lower m/z (8+ to 12+) with only a slightly wider charge distribution. Also, the further addition of CO2 concentration did not result in a shift to higher charge states (Figure 4.8B). These results implied that only small amounts of CO2 could effectively increase the charge state of Lys.

In the dual gradient elution, the charge states of all the proteins shifted to higher charge states (lower m/z). Even for RNase A and TI, which were eluted with the amount of liquefied CO2 < 2 v% and <5 v%, respectively, a shift in CSD was observed. This indicated that the effect of CO2 on CSD was perceptible at very low amount of CO2 addition, and it was not restricted to a specific protein. From Figure 4.9, the most abundance charge state of RNase A and Lys shifted from 7+, 8+ to 10+, 12+ in the dual gradient elution, while those of TI, Chy and ChA shifted from 9+, 10+, 10+ to 20+, 22+,

19+, respectively. The smaller charge state shift of RNase A and Lys could be attributed to the disulfide bonds within the structure maintaining a folded conformation.33 Other than the CSD shift, the observed ion intensity for each protein exhibited nearly 20 to 330- 147 fold enhancement in the dual gradient elution (Table 4.1). This was the result of the improved nebulization efficiency in the presence of CO2 in the ESI solvent, and this mechanism will be further discussed in this study.

Under the circumstances of highly complicated protein mixtures, where the separation step is insufficient to isolate one protein from the other, increased charge states and wider CSDs could increase the chance of spectral overlapping between coeluted proteins resulting in a higher spectral complexity.34 Charge reduction is often performed to address this problem.35,36 In this study, triethylammonium acetate (TEAA) was added to the mobile phase for charge reduction. Two adjacently eluted proteins, ChA and Lys were used to examine the charge reducing effect of TEAA. In Figure 4.10, the addition of

5 v% and 10 v% of 1M TEAA in the dual gradient elution didn’t necessarily alter the retention time and the peak shapes of ChA and Lys on PolyPENTYL A column. The addition of TEAA on the ESI-MS signals was significant. With increasing amount of

TEAA, the charge states of ChA and Lys were shifted to higher m/z with narrower distribution (Figure 4.11). However, severe ion suppression effect was noted with the addition of TEAA since ion pairing with the positive charges on the protein can occur during ESI.37 Hence, a charge reducing reagent such as TEAA could be added to the mobile phase to manipulate the CSD when proteins are not resolved in the separation, but

TEAA should be avoided in high concentrations.

148

Figure 4.8 Mass spectrum of Lys with (A) 0 v% CO2 in the mobile phase. (B) 2.5-15 v%

CO2 in the mobile phase. Flow rate: 1 mL/min.

149

Figure 4.9 Mass spectrum of RNase A (A1), TI (B1), Chy (C1), ChA (D1) and Lys (E1) in LC gradient condition as described in Figure 4.5A; and mass spectrum of RNase A

(A2), TI (B2), Chy (C2), ChA (D2) and Lys (E2) in EFLC dual gradient condition as described in Figure 4.5B.

150

Table 4.1 Average protein charge state without/with liquefied CO2 and signal enhancement with addition of liquefied CO2.

Signal Enhancement Protein No CO2 CO2 (Fold increase) RNase A 7+ 10+ 59 TI 9+ 20+ 158 Chy 10+ 24+ 335 ChA 10+ 18+ 56 Lys 8+ 12+ 24

151

Figure 4.10 Effect of (A, B) 0% v/v TEAA, (C, D) 5% v/v TEAA, (E, F) 10% v/v TEAA on the retention and peak shape of ChA (A, C, E) and Lys (B, D, F) in EFLC dual gradient elution.

152

Figure 4.11 Effect of (A, B) 0% v/v TEAA, (C, D) 5% v/v TEAA, (E, F) 10% v/v TEAA on the CSD and ionization efficiency of ChA (A, C, E) and Lys (B, D, F) in EFLC dual gradient elution.

153

4.3.4 Potential Factors that Affected CSD and Ionization Efficiency

There are many factors that affect the CSD and ionization efficiency of proteins during ESI process. Besides the physical dimension of protein molecules in solution,38 factors such as solution pH,39 the apparent gas phase basicity of protein,40 the gas phase basicity of the mobile phase solvents,41 and the desolvation process also influence the protein CSD.42 Understanding the effect of these parameters is important for optimal design of solvent condition and ionization process of MS. These critical factors were studied when the EFLC dual gradient eluent with liquefied CO2 was used as the ESI solvent.

4.3.4.1 Solution pH

Addition of CO2 to aqueous solution may alter the pH of the mobile phase by forming carbonic acid with H2O. Toews et al. reported that pH values of supercritical

43 CO2 in pure H2O were 2.8-2.83 under pressure of 70-200 atm at room temperature.

However, AmAc at neutral pH was adapted as the aqueous medium. Although AmAc has poor buffer capacity at pH 7, it can still alleviate any pH alternation in the elution and electrospray process. In the case of acidification of the mobile phase by introducing liquefied CO2, AmAc can lessen this effect to ensure that the solution pH is not lower than 4.75 ± 1, which is the pKa of acetate/acetic acid pair. In the positive mode ESI process, there are other factors that can lead to acidification even without CO2, such as oxidation and droplet shrinking.44,45 For example, Van Berkel et al. illustrated that the pH of unbuffered solution at the ESI emitter changed from neutral range to pH of 3, and 154 even down to 1.4 under some conditions.46 In this study, using AmAc solution in the mobile phase mixture mitigated these acidification effects by buffering the solution in the pH range of 4.75 ± 1 when analyte was released to positive mode ESI.

4.3.4.2 Apparent Gas Phase Basicity of Protein

The basicity of macromolecules in gas phase is useful in predicting the sites available for protonation.47 Using the charge residue model (CRM) of ESI process, for a folded protein (globular), the maximum charge state corresponds to the charges on the droplet at Rayleigh limit:

ZRe = 8πr√ϵ0γr (4.6)

where ZR is the charge number on the droplet at the Rayleigh instability limit; e is the elementary charge; r is the droplet radius; ε0 is the electric permittivity and γ is the surface tension.48 Schnier et al. proposed a model to estimate the maximum charge state of proteins employing the apparent gas phase basicity (GBapp) of proteins and modeling protein in their extended form. The GBapp value could be determined based on the intrinsic proton transfer reactivity (GBintrinsic) and the Coulomb energy from the interacting charges:

n 2 app q GB = GBintrinsic − ∑ (4.7) 4πε0εrri,t 1

where q is the charge, ε0 is the electric permittivity, εr is the effective shielding

41 and ri,t is the distance between charges. Using this model, they reported that the

155 maximum charge state of Lys was 22+ and 14+ for disulfide-reduced and disulfide-intact

41 structure, respectively. The maximum charge state of Lys observed with CO2 in the system was 14+, and it didn’t shift to higher charges when more CO2 was added. Even though Lys may not be in its native conformation when maximum charge state 14+ was observed, it should be at least partially folded with disulfide bond intact and with increased number of surface-exposed basic groups available when liquefied CO2 was added. Similarly, the structure of RNase A was also stabilized by its four disulfide bonds, therefore, relatively small shift in CSD was observed for RNase A and Lys. On the contrary, the CSD of TI, Chy and ChA experienced a significant shift and a wider distribution, which can be ascribed to the increased basic sites available to accept protonation when EFLC condition was applied.

4.3.4.3 Solvent Gas Phase Basicity

Apart from the GBapp of the basic sites in the protein, solvent gas phase basicity

(GB) is another important factor to significantly affect the CSD of the protein. Iavarone et al. investigated the effects of solvent composition on the maximum charge state and

CSD of proteins in the ESI process.49 In their study, the CSD of cytochrome C and myoglobin formed in 47:50:3 (v/v/v) H2O/solvent/acetic acid, shifted to lower charge states (higher m/z) with increasing solvent GB when the 50% solvent fraction changed from water (GB= 158 kcal/mol) to MeOH (GB= 173 kcal/mol), to ACN (GB= 179 kcal/mol), to isopropanol (GB= 182 kcal/mol). The GB value of CO2 is 123 kcal/mol, which is significantly lower than other constituents in the mobile phase and the

156 electrospray solvent where the GB values for ammonia, acetic acid, and methanol are 196 kcal/mol, 180 kcal/mol and 173 kcal/mol, respectively.50 The low GB of the solvent in presence of CO2 could inherently reduce the proton transfer from the charged protein to organic base resulting in higher charge state of protein.

4.3.4.4 Desolvation Process

The desolvation process greatly affects the ionization efficiency and CSD of

42 proteins. Liquefied CO2 turned into gaseous phase shortly after emitting from the ESI nebulizer. As the gas quickly releases from the droplets, the droplet size is expected to reduce. According to Fenn’s model of ion formation (Figure 4.12), which is an extension of ion evaporation model, the charges move closer together to give an increased surface charge density as the droplet evaporates.51 When the smaller droplets are formed in the early stage of the ESI process, a protein will spend a longer time in the regime of the high surface charge density leading to higher charge states.52 In addition to the increased surface charge density, the lower surface tension requires fewer Coulombic fission events to create gas-phase ions, resulting in higher ionization efficiency.42

To determine whether the faster desolvation caused by introducing the CO2 was the dominant factor for the charge state shift, other factors that affect the desolvation process were adjusted to counter this effect. Agilent Jet Stream (AJS) ESI nebulization assembly was used in this study. It utilizes a super-heated nitrogen sheath gas to confine the spray and increase the ion density in front of the capillary. Decreasing the sheath temperature on the ESI nebulization assembly could reduce the desolvation efficiency. As

157 shown in Figure 4.13, while keeping the CO2 content and other ESI parameters consistent, the CSD of ChA shifted back to lower charge states with a narrower distribution when the sheath gas temperature decreased from 350 °C to 150 °C.

Decreasing the sheath temperature down to room temperature didn’t cause further charge state shift compared to the CSD at 150 °C, however, decreasing drying gas temperature from 300 °C to 200 °C promoted an even narrower distribution but not necessarily lower charge state (Figure 4.13). Reducing the nebulizer flow could also diminish the desolvation. As shown in Figure 4.14, a change in the nebulizer pressure from 35 psig to

15 psig, caused the CSD of TI to shift to lower charge states with a narrower distribution, and this effect applied to all other studied proteins used. Accordingly, the enhanced desolvation using the EFLC mobile phase was one of the predominate factors for the

CSD shift and the improved ionization efficiency of proteins. Optimization of these conditions allows improved ionization efficiency.

Figure 4.12 Illustration of a charged droplet containing four molecules with different sizes and shapes at three different stages of solvent evaporation.51

158

Figure 4.13 Mass spectra of ChA using EFLC dual gradient condition as described in

Figure 1C with different gas temperatures at AJS electrospray ion source: (A) sheath gas:

350 °C, drying gas: 300 °C; (B) sheath gas: 250 °C, drying gas: 300 °C; (C) sheath gas:

150, drying gas: 200 °C.

159

Figure 4.14 Mass spectra of ChA using EFLC dual gradient condition as described in

Figure 1C with different gas temperatures at AJS electrospray ion source: (A) sheath gas:

350 °C, drying gas: 300 °C; (B) sheath gas: 250 °C, drying gas: 300 °C; (C) sheath gas:

150, drying gas: 200 °C.

160

4.3.5 Supercharging effect of Liquefied CO2

Other than improving the separation efficiency of intact proteins, the addition of liquefied CO2 in the EFLC methods also dramatically improved the ionization efficiency and increased the charge states of proteins. This effect is called “supercharging”, which was termed by William et al. in 2003 to describe the increased charging observed in ESI solvents with a variety of additives53. Supercharging the intact protein ions offers several merits for ESI-MS characterization. Enhancing the charge states of large biomolecules can benefit the high mass accuracy measurements and it’s valuable for studying very large macromolecules.54 Supercharging is highly desirable because the performance of most mass analyzers improves with higher charge states.55 Moreover, supercharging of intact proteins can significantly enhance the tandem MS ionization efficiency with higher sequence coverage.56 Therefore, supercharging effect is especially valuable for top-down proteomics. The most commonly used supercharging reagents such as m-nitro benzyl alcohol (m-NBA), dimethyl sulfoxide (DMSO) and tetramethylene sulfone (sulfolane) can inversely affect the LC performance like peak broadening and lower resolution.57,58

Thus, these supercharging reagents worked more effectively by post-column addition.59

Liquefied CO2 used in this work ensured a fully integrated LC-MS strategy with promising efficiency in both separation and characterization for intact proteins.

161

4.4 Conclusions

A dual gradient system consisted of HIC salt-organic gradient and EFLC CO2 gradient was successfully applied to intact protein separation by using the custom instrument configuration for the first time. Decreased viscosity of the mobile phase by applying EFLC lowered the system back pressure which could facilitate higher flow rates on the current state-of-the-art instruments. EFLC offered improved chromatographic behaviors including a shorter analysis time, better peak shapes and high plate numbers.

Charge state shift and enhanced ionization efficiency were obtained for all protein samples including a labile protein complex under EFLC condition. These effects were mainly attributed to the improved desolvation with the addition of CO2 in the ESI process. Therefore, liquefied CO2 proved to be an ESI friendly and supercharging reagent without sacrificing the chromatographic performance. This system can be used to improve peptide and protein identification in large-scale application.

162

4.5 References

1 Siuti, N.; Kelleher, N. L. Nat Methods 2007, 4, 817–821.

2 Smith, L. M.; Kelleher, N. L. Nat Methods 2013, 10, 186–187.

3 Gregorich, Z. R.; Ge, Y. Proteomics 2014, 14, 1195–1210.

4 Kelleher, N.L. Anal. Chem. 2004, 76, 196A-203A.

5 Zheng, S.; Yoo, C.; Delmotte, N.; Miller, F. R.; Huber, C. G.; Lubman, D. M. Anal. Chem. 2006, 78, 5198-5204.

6 Kortz, L.; Dorow, J.; Ceglarek, U. J. Chromatogr. B 2014, 964, 1–11.

7 Rea, J. C.; Moreno, G. T.; Lou, Y.; Farnan, D. J. Pharm. Biomed. Ana. 2011, 54, 317- 323.

8 Muneeruddin, K.; Nazzaro, M.; Kaltashov, I. A. Anal. Chem. 2015, 87, 10138–10145.

9 Leblanc, Y.; Ramon, C.; Bihoreau, N.; Chevreux, G. J. Chromatogr. B 2017, 1048, 130–139.

10 Zhang, X.; Fang, A.; Riley, C. P.; Wang, M.; Regnier, F. E.; Buck, C. Anal Chim Acta. 2010, 664, 101–113.

11 Hong, P.; Koza, S.; Bouvier, S. P. J. Liq. Chromatogr. Relat. Technol. 2012, 35, 2923- 2950.

12 Grodzki, A. C.; Berenstein, E. Methods Mol Biol. 2010, 588, 33-41.

13 Urh, M.; Simpson, D.; Zhao, K. Methods Enzymol. 2009, 463, 417-438.

14 Eschelbach, J. W.; Jorgenson, J. W. Anal. Chem. 2006, 78, 1697-1706.

15 Fekete, S.; Veuthey, J.; Guillarme, D. J. Pharm. Biomed. Anal. 2012, 69, 9-27.

16 Queiroz, J. A.; Tomaz, C. T.; Cabral, J. M. S. J. Biotechnol. 2001, 87, 143-159.

17 Chen, B.; Peng, Y.; Valeja, S. G.; Xiu, L.; Alpert, A. J.; Ge, Y. Anal. Chem. 2016, 88, 1885–1891.

18 Beres, M. J.; Olesik, S. V. J. Sep. Sci. 2015, 00, 1–11.

163

19 Beilke, M. C.; Beres, M. J. Olesik, S. V. J. Chromatogr. A 2016, 1436, 84-90.

20 Bennett, R.; Olesik, S. V. Anal. Chim. Acta 2017, 960, 151-159.

21 Bennett, R.; Olesik, S. V. J. Chromatogr. A 2017, 1523, 257-264.

22 Cappiello, A.; Famiglini, G. Rossi, L.; Magnani, M. Anal. Chem. 1997, 69, 5136-5141.

23 Gooding, D. L.; Schmuck, M. N.; Nowlan, M. P.; Gooding, K. M. J. Chromatogr. 1986, 359, 331−337.

24 Nakamura, H.; Konishi, T.; Kamada, M. Anal. Sci. 1990, 6, 137-138.

25 Konishi, T.; Kamada, M. J. Chromatogr. 1990, 515, 279−283.

26 Cui, Y.; Olesik, S. V. Anal. Chem. 1991, 63, 1812-1819.

27 Poole, C. F. The essence of Chromatography, 1st ed.; Elsevier: New York, 2003.

28 Dolan, J. W.; Snyder, L. R.; Djordjevic, N. M.; Hill, D. W.; Waeghe, T. J. J. Chromatogr. A 1999, 857, 1–20.

29 Davies, M. J. Biochim. Biophys. Acta 2005, 1703, 93-109.

30 Berlett, B. S.; Stadtman, E. R. J. Biol. Chem. 1997, 272, 29313-29316.

31 Kim, G.; Weiss, S. J.; Levine, R. L. Biochim. Biophys. Acta 2014, 1840, 2-12.

32 Sanchez, R.; Riddle, M.; Woo, J.; Momand, J. Protein Sci. 2008, 17, 473-481.

33 Zhang, T.; Bertelsen, E.; Alber, T. Nat. Struct. Mol. Biol. 1994, 1, 434-438.

34 Krusemark, C. J.; Frey, B. L.; Belshaw, P. J.; Smith, L. M. J. Am. Soc. Mass Spectrom. 2009, 20, 1617–1625.

35 Pagel, K.; Hyung, S. J.; Ruotolo, B. T.; Robinson, C. V. Anal. Chem. 2010, 82, 5363– 5372.

36 Zhou, M.; Dagan, S.; Wysocki, V. H. Angew. Chem. Int. Ed. 2012, 51,4336–4339.

37 Sharma, V. K.; Glick, J.; Vouros, P. J. Chromatogr. A 2012, 1245, 65-74.

38 Kaltashov, I. A.; Abzalimov, R. R. J. Am. Soc. Mass Spectrom. 2008, 19, 1239-1246.

39 Konermann, L.; Douglas, D. J. Biochemistry 1997, 36, 12296-12302.

40 Prakash, H.; Mazumdar, S. J. Am. Soc. Mass164 Spectrom. 2005, 16, 1409-1421.

41 Schnier., P. D.; Gross, D. S.; William, E. R. J. Am. Soc. Mass Spectrom. 1995, 6, 1086- 1097.

42 Page, J. S.; Kelly, R. T.; Tang, K.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2007, 18, 1582-1590.

43 Toews, K. L.; Shroll, R. M.; Wai, C. M. Anal. Chem. 1995, 67, 4040–4043.

44 Van Berkel, G. J.; Kertesz, V. Anal. Chem. 2007, 79, 5511–5520.

45 Iyengar, S. S.; Day, T. J. F.; Voth, G. A. J. Am. Soc. Mass Spectrom. 2005, 241, 197- 204.

46 Van Berkel, G. J.; Asano, K. G.; Schnier, P. D. J. Am. Soc. Mass Spectrom. 2001, 12, 853-862.

47 Touboul, D.; Jecklin, M. C.; Zenobi, R. J. Phys. Chem. B 2007, 111, 11629-11631.

48 Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem. Phys. 1968, 49, 2240–2249.

49 Iavarone, A. T.; Jurchen, J. C.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2000, 11, 976–985.

50 Hunter, E. P.; Lias, S. G. J. Phys. Chem. Ref. Data 1998, 27, 3, 413-656.

51 Fenn, J. B. J. Am. Soc. Mass Spectrom. 1993, 4, 524-535.

52 Karki, S.; Flanigan IV, P. M.; Perez, J. J.; Archer, J. J.; Levis, R. J. J. Am. Soc. Mass Spectrom. 2015, 26, 706-715.

53 Iavarone, A. T.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2003, 125, 2319–2327.

54 Han, X.; Jin, M.; Breuker, K.; McLafferty, F. W. Science 2006, 314, 109-112.

55 Schaub, T. M.; Hendrickson, C. L.; Horning, S.; Quinn, J. P.; Senko, M. W.; Marshall, A. G. Anal. Chem. 2008, 80, 3985–3990.

56 Tsybin, Y. O.; Fornelli, L.; Stoermer, C.; Luebeck, M.; Parra, J.; Nallet, S.; Wurm, F. M.; Hartmer, R. Anal. Chem. 2011, 83, 8919−8927.

57 Kjeldsen, F.; Giessing, A. M. B.; Ingrell, C. R.; Jensen, O. N. Anal. Chem. 2007, 79, 9243-9252.

58 Meyer, J. G.; Komives, E. A. J. Am. Soc. Mass Spectrom. 2012, 23, 1390-1399. 165

59 Miladinovic, S. M.; Fornelli, L.; Lu, Y.; Piech, K. M; Girault, H. H.; Tsybin, Y. O. Anal. Chem. 2012, 84, 4647−4651.

166

Chapter 5. Enhanced-Fluidity Liquid as the Alternative Mobile Phase in Reversed

Phase Chromatography for Intact Protein Separation

5.1 Introduction

With its superior resolving power, versatility, convenience and stability, reversed phase chromatography (RPC) is the most popular separation strategy for a wide range of analytes. Due to the increasing emphasis on protein biopharmaceuticals, there is a growing demand for the development of effective RPC separations for macromolecules like intact proteins. Because the mobile phase used in RPC methods is highly compatible with mass spectrometry (MS), RPC is an indispensable choice in proteomic research.

RPC separates compounds according to their hydrophobicity. Unlike hydrophobic interaction chromatography (HIC) discussed previously, the stationary phase in RPC mode is usually more hydrophobic (e.g. linear alkyl chains). RPC provides excellent resolution for small molecules and peptides and has been applied extensively in bottom up proteomics for peptide fractionation in both initial analysis and final large-scale

1,2 purification. However, RPC separations for intact proteins are not as powerful as for small peptides due to the structural complexity and large molecular weight of proteins.

To successfully perform an RPC separation for intact protein, many parameters need to be carefully considered and optimized. These parameters can be sorted into two categories: characteristics of stationary phase and solvent choice for mobile phase.

167

The most important factors associated with RPC stationary phase include the surface chemistry, alkyl chain length, particle size, pore size and the mechanical stability.3 The stationary phase of RPC is commonly composed of linear alkyl chains (e.g.

C4, C8 and C18) linked to porous silica particles. The residual silanol groups on the RPC stationary support can have secondary interaction (ionic interaction) with the charged sites on proteins leading to peak tailing, low recovery and sometimes irreversible adsorption.4 To minimize secondary interaction, endcapping or ion pairing reagent can be used to restrict the access to residual silanol groups.5 Shorter alkyl chains such as C4 or

C8 with less retentive character are more suitable for intact proteins because large proteins usually have increased hydrophobicity which can experience extremely strong interaction with longer alkyl ligands like C18 and result in incomplete elution. C18 can be used for very small proteins (< 10 kDa) with low hydrophobicity. The diffusion rates of macromolecules are significantly slow in the conventional particle with 80-120 Å pore size; therefore, packing materials with larger pore sizes such as 300 or 1000 Å are preferred for intact proteins to avoid peak broadening caused by slow diffusion.6

Moreover, because the multipath diffusion and resistance to mass transfer are directly related to the particle size, columns packed with reducing particle size (sub 2-µm or nanoscale) can substantially improve the separation efficiency.7

The use of large amount of organic solvent in RPC can often result in protein denaturization and loss of its biological activity. This can further induce protein precipitation and peak broadening during the elution process.8 Careful selection of organic solvents in RPC is crucial in order to achieve successful separation. Among

168 commonly used organic solvents for RPC, acetonitrile (ACN) has been the solvent of choice for intact proteins because of the high resolution and good peak shape attainable

9 mainly due to the low viscosity of H2O/ACN mixture and sufficient solvent strength for denature proteins. However, since 2008, a combination of economic and environmental reasons has caused a sudden shortage of ACN with a sharp price increase.10 This ACN crisis gave a strong “wake-up call” for both academia and industry to seek sustainable substitutes for RPC separations. In the past decade, researchers have made great efforts to reduce the usage of ACN by developing miniatured LC devices with shorter columns and smaller particles11 or replacing ACN completely with more readily available solvents like methanol (MeOH) or isopropanol (IPA).12 However, due to several chemical properties of alcohol-based solvents, using MeOH or IPA alone rarely provides comparable chromatographic performance to ACN especially on the common HPLC columns.

The mobile phase used in enhanced fluidity liquid chromatography (EFLC), which composed of liquefied CO2 as the additive in the conventional liquid mobile phase

(commonly water/alcohol), has been demonstrated as the “green” alternative for ACN in hydrophilic interaction chromatography (HILIC) separation with comparable or even better efficiency for biomolecules including nucleotides, nucleosides13 and proteins.14

In this work, EFLC is first time applied for intact protein separation in RPC mode.

A commercial C4 column with 300 Å pore size was used to evaluate the performance of

EFLC mobile phases compared to conventional liquid chromatography mixtures

(H2O/ACN and H2O/MeOH) in term of analysis time, separation efficiency and peak shape. Comparison of several MS compatible RPC additives will also be illustrated in

169 this study under optimized EFLC condition to provide insights for better solvent choices in EFLC-MS analysis.

5.2 Materials and Methods

5.2.1 Chemicals and Reagents

Protein standards including ribonuclease A from bovine pancreas (RNaseA), cytochrome c from equine heart (Cyt C), lysozyme from chicken egg white (Lys), lysozyme from chicken egg white (Lys), bovine serum albumin (BSA), myoglobin from equine heart (MGB) and α-chymotrypsinogen A from bovine pancreas (ChA) were purchased from Sigma-Aldrich (St. Louis, MO). LC-MS grade methanol (MeOH), acetonitrile (ACN), isopropanol (IPA), trifluoroacetic acid (TFA, >99%), formic acid

(FA, >99%) and glacial acetic acid (ACS reagent, 99.7%) were purchased from Fisher

Scientific (Fair Lawn, NJ). Ammonium formate (AmFm, BioUltra ≥99%) and ammonium acetate (AmAc, for molecular biology, ≥98%) were purchased from Sigma-

Aldrich (St. Louis, MO). Water was purified to 18.3 MΩ by a Barnstead Nanopure

Infinity System from Thermal Scientific Inc. (Odessa, TX). Supercritical fluid extraction grade CO2 (99.999%) was obtained from Praxair, Inc. (Danbury, CT).

5.2.2 Sample Preparation

Standard protein samples RNase A, Cyt C, Lys, BSA and MGB were prepared without additional purification. Proteins were dissolved in nanopure H2O at 2mg/mL,

170 followed by syringe filtration using 0.2 µm Captiva syringe filter (Agilent Technologies,

Santa Clara, CA).

5.2.3 Instrumentation

An Agilent 1260 Infinity HPLC system (Agilent Technologies, Santa Clara, CA) was used to perform conventional H2O/organic mobile phase analysis. The system included a 1260 HiP degasser, a 1260 binary pump, 1260 autosampler, 1260 thermostatted column compartment (80°C maximum) and a 1200 Diode Array Detector

(DAD). The EFLC analysis was conducted on an Angilent 1260 Infinity SFC system

(Agilent Technologies, Santa Clara, CA) which included an Aurora SFC Fusion ™ A5 module (400 bar pressure limit), 1260 HiP degasser, a 1260 SFC binary pump, a 1260

SFC ALS autosampler, 1290 thermostatted column compartment and a 1200 DAD. The back-pressure regulator (BPR) was set to 100 bar and the temperature to 50 °C to maintain the backpressure over the system. An Agilent 6530 ESI-Q-TOF mass spectrometer (MS) was used for both HPLC and EFLC studies. A splitter assembly was connected to the outlet of the DAD to split flow between the BPR and the MS. An

XBridge Protein BEH C4 column (Waters, 4.6 × 150 mm, 3.5 µm, and 300 Å) was used for both HPLC and EFLC separations of protein mixture.

5.2.4 Data Acquisition and Analysis

The DAD was set to a wavelength of 220 nm with a bandwidth of 4 nm, a sampling rate of 2.5 Hz, and a background wavelength of 360 nm with a 100 nm bandwidth. All mass spectra were collected in positive ion mode. The capillary voltage 171 was optimized at 3500V. The nebulizer was set to 35 psig. The drying gas was set to 8

L/min at 300 °C. All data were collected using LC/MS Data Acquisition Version B 06.01 in Agilent MassHunter Workstation Software. All data were analyzed using Qualitative

Analysis Version B 07.00 in Agilent MassHunter Workstation Software. PeakFit Version

4 software (SPSS Inc. Chicago, IL) was used to perform gaussian fits.

5.3 Results and Discussion

5.3.1 Selection of Organic Cosolvent for EFLC

To find optimum EFLC solvent mixture, organic solvents including IPA, MeOH and ACN were evaluated initially under HPLC conditions for model proteins. The protein standards used in the evaluation represented a range of molecular weight from ~12 to

66.5 kDa, isoelectric points from 4.7 to 11, and with various hydrophilicities. As shown in Figure 5.1, IPA exhibited the strongest solvent strength. With even shallower gradient profile (20-40% in 30 min) compared to ACN and MeOH (20-60% in 30 min), all proteins were eluted within the shortest time scale with H2O/IPA solvent system.

Unfortunately, the less retention of proteins was also accompanied with reduced resolution. Lys and BSA was coeluted when IPA was used as the organic solvent.

Comparing Figure 5.1B and C, with the same gradient profile, proteins experienced much stronger retention in H2O/MeOH solvent system. In RPC, increasing the temperature of mobile phase can reduce the retention of analytes because elevated temperatures lead to reduced mobile phase viscosity and enhanced analyte diffusivity.15 However, even at 80

°C of column temperature, the elution strength of MeOH was still not comparable to that 172 of ACN at 40 °C. Knowing that the packing materials in RPC column usually have an upper temperature limit to ensure the stability and separation reproducibility. The shorter the alkyl ligands, the lower the temperature limit. The maximum temperature for C4 column is 90 °C. Moreover, the temperature limit on the column compartment is 80 °C; therefore, further increase the mobile temperature for H2O/MeOH was not an option. In order to achieve similar solvent strength as ACN, higher concentration of MeOH is required without promise in similar separation efficiency and peak shape.

Baseline drift was observed in separations using H2O/IPA and H2O/MeOH, and this was mainly due to the changes in the refractive index of the solvent during gradient elution. The UV cutoff values of the solvents used in this comparison are summarized in

Table 5.1. The remedy to severe baseline drift is to use an alternative wavelength or a UV transparent solvent like ACN. Additionally, in HPLC, peptide bonds on proteins contribute absorbance in the 210-220 nm range, therefore the ideal mobile phase solvent should have minimum UV absorbance inn this range to avoid noise in UV detection.16

Unfortunately, neither IPA nor MeOH has lower UV absorption than ACN at short UV wavelength.

173

Figure 5.1 DAD chromatogram at 210 nm of protein mixture using (A) IPA gradient: 0-

30 min from 20% to 40% IPA at 40 °C; (B) ACN gradient: 0-30 min from 20% to 60%

ACN at 40 °C; (C) MeOH gradient: 0-30 min from 20% to 60% MeOH at 40 °C; (D)

MeOH gradient: 0-30 min from 20% to 60% MeOH at 80 °C. Flow rate for all separations: 0.5 mL/min. Each mobile phase contained 0.2% v/v of TFA as additive.

Analyte key: 1. RNase A; 2. Cyt C.; 3. Lys; 4. BSA; 5: MGB; 6: ChA.

174

Table 5.1 Selected physical and chemical properties of water, acetonitrile, methanol and isopropanol.

Property H2O ACN MeOH IPA

Density [g/mL] 1.00 0.786 0.792 0.786

Viscosity [cP] at 25 °C 0.89 0.38 0.55 1.96

UV Cutoff [nm] 200 190 205 205

175

Another major challenge in replacing ACN with an alcohol-based solvent is high viscosity associated with H2O/alcohol mixture. Not only water, MeOH and IPA has much higher viscosity individually compared to ACN (Table 5.1), the viscosity of water/alcohol mixtures also changes more drastically than water/ACN mixture when the concentration of organic solvent increases. As shown in Figure 5.2, the pressure experienced by the chromatographic system varies depending on the mixing ratio of organic solvents. H2O/IPA had its maximum viscosity at 50-60% (v/v) of IPA corresponding to the highest system pressure in a H2O/IPA gradient elution. Given the same flow rate, the system pressure using H2O/MeOH showed the similar trend with

H2O/IPA but with lower overall viscosity. In contrast, the system pressure decreased when more ACN was added into aqueous solution. Therefore, when running a gradient, an appropriate flow rate need to be carefully determined for alcohol-based mobile phase because the initial system pressure does not represent the maximum pressure in the entire run. The significantly high viscosity of H2O/IPA hindered fast flow rates to be performed.

Addition of liquefied CO2 was expected to correct several disadvantages with

H2O/alcohol mobile phase. Instead of running a H2O/alcohol gradient, a CO2 gradient can be performed in a fixed H2O/alcohol ratio, and liquefied CO2 has extremely low viscosity which is close to that of gases, which eliminates the concern of exceeding the pressure limit for a gradient run. Moreover, miscibility of CO2 in H2O/alcohol is higher than that

17 in H2O/ACN, allowing more CO2 to be blended in the mobile phase. Due to its nonpolar nature, CO2 was expected to act as a strong solvent in RPC elution. Because of 176 the strong elution and poor resolution obtained using IPA even at low concentration, adding CO2 can further increase the solvent strength and reduce the resolution; moreover, low concentration of IPA in water has limited solubility for liquefied CO2. Therefore,

H2O/MeOH was chosen over H2O/IPA for EFLC analysis.

Figure 5.2 Changes in system backpressure as the concentration of organic solvents increases in the H2O/Organic mixture. Flow rate: 0.5 mL/min.

177

5.3.2 Comparison between EFLC and HPLC with Optimized Mobile Phases

The goal of optimizing the mobile phase condition is to achieve the highest resolution for all the analytes in the shortest amount of time. The CO2 gradient was optimized in 40:60 (v/v) H2O/MeOH mixture. A higher MeOH content (e.g. 65%) in the

EFLC mobile phase could cause coelution of RNase A and Cyt C at the dead time. A CO2 gradient of 2-15% v/v from 2 to 5 min was implemented and the separation is shown in

Figure 5.3A. It should be noted that there is a limit to the amount of liquefied CO2 that can added to H2O/organic mixtures before demixing issues arise. The volume portion beyond 15% (v/v) of CO2 in 60% (v/v) MeOH can cause solvent demixing and significant baseline noise. As shown in Figure 5.3, EFLC separation showed similar analysis time and better peak shapes compared to the optimized separation using

H2O/ACN gradient in HPLC condition. Noted that the optimized H2O/MeOH gradient in

HPLC still gave slow elution and broad peaks. In addition, the amount of MeOH in

EFLC was fixed at 60% (v/v), whereas the HPLC condition required up to 100% (v/v)

MeOH to achieve sufficient elution in a reasonable time scale. Therefore, the EFLC mobile phase can serve as an effective alternative for ACN in RPC separations of proteins in terms of analysis time and organic solvent consumption. The chromatographic performance of both EFLC and HPLC solvent systems including efficiency (N) and peak shape will be examined in greater detail in the next section.

178

Figure 5.3 DAD chromatogram at 210 nm of protein mixture using (A) EFLC gradient:

2-5 min min from 2% to 15% CO2 at 40 °C; (B) HPLC-ACN gradient: 0-8 min from 30% to 60% ACN at 40 °C; (C) HPLC-MeOH gradient: 0-5-8 min from 60% to 80% to 100%

MeOH at 40 °C. Flow rate for all separations: 1 mL/min. Each mobile phase contained

0.2% v/v of TFA as additive. Analyte key: 1. RNase A; 2. Cyt C.; 3. Lys; 4. BSA; 5:

MGB.

179

5.3.3 Chromatographic Performance of EFLC and HPLC

As discussed in Chapter 1, peak shape is an important factor to access the chromatographic performance. Either peak tailing or fronting is undesired in a successful separation. As shown in Figure 5.4, asymmetry factor (As) in the range of 1-1.5 was obtained for all the protein samples using EFLC mobile phase. In practice, optimum

18 separation should strive for As values no more than 1.5 for all sample peaks. Several peaks in the HPLC separations using ACN or MeOH as the organic solvent in the mobile phase exhibited As values >1.5 or even larger than 2. As mentioned earlier, peak asymmetry can be caused by many factors such as column impurity, poorly packed column, extra-column effect, sample overloading and weak mobile phase.19 Under the same column condition, the better peak shapes observed in EFLC separation can be mainly ascribed to the improved solvent strength and fast mass transfer.

180

Figure 5.4 Comparison of peak asymmetry using EFLC mobile phase, HPLC mobile phases containing ACN and MeOH under optimized conditions (mobile phase conditions listed in Figure 5.3).

181

5.3.4 Additive choice for EFLC-MS

In RPC, additives are often used to improve the separation.20 Trifluoroacetic acid

(TFA) is the most commonly used mobile phase additive for analysis of intact proteins in

RPC owing to the sharp, symmetrical peak shapes and excellent resolution it can offer.21

The use of TFA can serve two functions in the RPC separation: 1) due to its low pKa, trifluoroacetate ions (Figure 5.6) bind to the positive charged sites on protein structure and increase the hydrophobic character of proteins. This ion pairing effect can provide adequate retention for proteins on the RPC column which usually leads to improved resolution. 2) as mentioned previously, the silica support used in RPC comprises very complicated surface chemistry (Figure 5.7). In particular, the highly active residual silanol groups (component 2 and 5 in Figure 5.7) can have secondary interaction with the polar or ionic groups on proteins resulting in peak tailing. However, this effect can be controlled by the pH of the mobile phase. The pKa of silanol is 3.8-4.2, therefore when mobile phase pH is higher than 6, the silanol groups will be fully ionized which can interact strongly with the basic sites on proteins. The addition of TFA (pKa~0.23) significantly reduces the solution pH, in which the silanol groups will be protonated and the secondary interaction are then lessened.22 Combining the ion pairing and pH controlling effects, TFA is a powerful reagent for intact proteins in RPC separations.

182

Figure 5.5 Ion pairing effect of TFA on proteins.

Figure 5.6 Illustration of various important aspects of silica surface chemistry in RPC.22

183

Unfortunately, the strong ion pairing character of TFA can have adverse effect when MS detection is used. The formed ion-pairs are so strongly attached that they are rarely broken apart during the electrospray ionization process, which results in charge suppression of proteins.23 In addition, addition of TFA leads to high conductivity and surface tension of the mobile phase giving rise to spray instability and sensitivity reduction.24 Many efforts have been made to overcome the ion suppression issue caused by TFA. For example, Apffel et al. reported the use of post-column addition of highly concentrated weak acid to compete with TFA and promote protonation of analytes.24 In order to balance the chromatographic efficiency and MS sensitivity in an integrated system, several MS compatible additives including formic acid (FA), acetic acid (AA), ammonium formate (AmFm) and ammonium acetate (AmAc) were evaluated in this study. All the additives were examined under the optimized EFLC mobile phase condition. As shown in Figure 5.8 A and B, when TFA was replaced by FA and AA, the less retained proteins such as RNase A, Cyt C and Lys were coeluted as a single peak at the dead time. This was due to the diminished ion pairing effect accompanied by reduced hydrophobicity of proteins. Meanwhile, the relatively large protein BSA (MW 66.5 kDA) and the more hydrophobic protein MGB (GRAVY -0.381) showed significant peak broadening when FA and AA were used as the mobile phase additive. Because the pKa values of FA (pKa 3.75) and AA (pKa 4.76) are similar to the one of residual silanol groups on RPC column and much higher than that of TFA, the silanol groups were more susceptible to forming secondary interaction with the polar and ionizable sites on proteins, noted that more polar sites would be available on a large protein when its not

184 ion-paired. Additionally, due to the lower anion concentration using FA and AA, mutual repulsion between the adsorbed protein ions with the same charge could limit sample loading capacity resulting in peak broadening.25 Therefore, the severe band broadening for the more retained proteins was the result of increased secondary interaction and mutual repulsion effect when FA and AA were used.

A pair of MS compatible buffers including ammonium formate (AmFm) and ammonium acetate (AmAc) was also evaluated with unadjusted pH of 6.8 and ionic strength of 10 mM. As shown in Figure 5.8 C and D, the EFLC containing AmFm and

AmAc as additives provided much better resolution for RNase A, Cyt C and Lys compared to their acid counterparts. Notably, the peak shapes of BSA and MGB were significantly improved using AmFm and AmAc. As mentioned above, overloading of peptides and proteins can occur at very low sample amounts in RPC when FA and AA were adapted as the additives. The strong ionic strength of ammonium salts alleviated the mutual repulsion effect between ions and improved sample load tolerance.26 Furthermore, ammonium ions were likely to mask the residual silanol group on the RPC column and lessen secondary interactions leading to improved peak shapes.

MS spectra of the most retained model protein, MGB, were examined to compare the impact of different additives on ionization efficiency (Figure 5.9). Compared to the condition without any additive, the ion intensities using TFA were reduced by 3 times due to the ion suppression effect discussed previously. Additives including FA, AA,

AmFm and AmAm showed 4 to 6 times higher MS signal response than TFA.

Meanwhile, the charge states were shifted to higher charges under the condition of FA,

185

AA, AmFm and AmAc. Even though broad chromatographic peaks of MGB were eluted into MS using FA and AA as the additive, the MGB MS signals using these additives were in the similar intensity range compared to those obtained using AmFm and AmAc, which indicated that FA and AA provided the highest ionization efficiency among all the tested additives. It’s worth noting that while other protein samples still maintained intact

MW (Figure 5.10), the deconvoluted MW of MGB using all the conditions was ~16.9 kDa, corresponding to apo-MGB, which lacked a heme unit (616 Da) from the structure.

This was expected for the denaturization of proteins when harsh eluting conditions like

RPC was applied.

186

Figure 5.7 Effect of different additives on separations of protein mixture. (A) 0.2% FA;

(B) 0.2% AA; (C) 10 mM pH 6.8 AmFm; (D) 10 mM pH 6.8 AmAc. EFLC gradient: 2-5 min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min. Analyte key: 1. RNase A;

2. Cyt C.; 3. Lys; 4. BSA; 5: MGB.

187

Figure 5.8 Effect of different additives on MS sensitivity of MGB (A) no additive; (B)

0.2% TFA; (C) 0.2% FA; (D) 0.2% AA; (E) 10 mM pH 6 AmFm; (F) 10 mM pH 6

AmAc. EFLC gradient: 2-5 min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min.

188

Figure 5.9 MS signal of (A) RNase A; (B) Cyt C; (C) Lys; (D) BSA; (E) Heme; (F) MGB using EFLC gradient: 2-5 min min from 2% to 15% CO2 at 40 °C. Flow rate: 1 mL/min.

189

5.4 Conclusions

In this work, EFLC was applied for intact protein separation in RPC mode for the first time. The enhanced fluidity liquid served as an efficient alternative to the traditional

H2O/ACN mobile phase in RPC. Small amount of liquefied CO2 in the mixture of

H2O/MeOH provided comparable solvent strength to achieve similar analysis time to

H2O/ACN. Better peak symmetry was obtained with EFLC compared to H2O/ACN and

H2O/MeOH mobile phases in HPLC. Special consideration of mobile phase additive must be made when optimizing RPC-MS method. TFA offered the best chromatographic performance but with ion suppression effect. With improved MS signal response, AmFm and AmAc outperformed FA and AA in terms of chromatographic resolution and peak shapes, and they can be the additives of choice to balance separation performance while maintaining sensitive MS detection.

190

5.5 References

1 Xu, P.; Duong, D. M.; Peng, J. J. Proteome Res. 2009, 8, 3944-3950.

2 Delmott, N.; Lasaosa, M.; Tholey, A.; Heinzle, E.; Huber, C. G. J. Proteome Res. 2007, 6, 4363–4373

3 Giacometti, J.; Josic, D. Liquid Chromatography: Applications; Elsevier: New York, 2013.

4 Mant, C. T.; Hodges, R. S. High Performance Liquid Chromatography of Peptides and Proteins: Separation, Analysis, and Conformation; CRC Press: Florida, 1991.

5 Aguilar, M. HPLC of Peptides and Proteins: Methods and Protocols; Humana Press: Totowa, New Jersey, 2004.

6 Wilson, K. J.; Wieringen, E. V.; Klauser, S.; Berchtold, M. W. J. Chromatogr. 1982, 237, 407-416.

7 Everley, R. A.; Croley, T. R. J. Chromatogr. A 2008, 1192, 239–247.

8 Gekko, K.; Ohmae, E.; Kameyama, K.; Takagi, T. Biochim. Biophys. Acta 1998, 1387, 195-205.

9 Rubinstein, M. Anal. Biochem. 1979, 98, 1-7,

10 Venkatasami, G.; Sowa Jr, J. R. Anal. Chim. Acta 2010, 665, 227-230.

11 Cielecka-Piontek, J.; Zalewski, P.; Jelinska, A.; Garbacki, P. Chromatographia 2013, 76, 1429–1437.

12 Desai, A. M.; Andreae, M.; Mullen, D. G.; Banaszak Holl, M. M.; Baker Jr, J. R. Anal Methods. 2011, 3, 56-58.

13 Beilke, M. C.; Beres, M. J.; Olesik, S. V. J. Chromatogr. A 2016, 1436, 84-90.

14 Bennett, R.; Olesik, S. V. J. Chromatogr. A 2017, 1523, 257-264.

15 Antia, F. D.; Horvath, Cs. J. Chromatogr. 1988, 453, 1–15.

16 Reversed Phase Chromatography: Principles and Methods; Amersham Pharmacia Biotech: Vastra, Sweden, 1997.

191

17 Lee, S. T.; Reighard, T. S.; Olesik, S. V. Fluid Phase Equilib. 1996, 122, 223-241.

18 Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. Practical HPLC Method Development, 2nd ed.; John Wiley & Sons: New York, USA, 1997.

19 Papai, Zs.; Pap, T. L. J. Chromatogr. A 2002, 953, 31-38.

20 Li, S.; Tian, M.; Row, K. H. Int. J. Mol. Sci. 2010, 11, 2229-2240.

21 Bobaly, B.; Beck, A.; Fekete, J.; Guillarme, D.; Fekete, S. Talanta, 2015, 136, 60-67.

22 Crawford Scientific. Silica for HPLC Stationary Phases. https://www.crawfordscientific.com(accessed April 20, 2018).

23 Garcia, M. C. J. Chromatogr. B 2005, 825, 111-123.

24 Apffel, A.; Fischer, S.; Goldberg, G.; Goodey, P. C.; Kuhlmann, F. E. J. Chromatogr. A 1995, 712, 177-190.

25 McCalley, D. V. J. Chromatogr. A 2004, 1038, 77-84.

26 Johnson, D.; Boyes, B.; Orlando, R. J. Biomol. Tech. 2013, 24, 187-197.

192

Chapter 6. Summary and Future Work

6.1 Summary of Research

The research detailed in this document describes the development of novel nanomaterials and chromatographic solvent systems for separations of intact proteins and biocompatible macrostructures (PEGylated AuNPs). Nanofibers fabricated via electrospinning technique has been demonstrated as a convenient and highly effective stationary phase for rapid analysis of biomolecules and nanoparticles in Chapter 2 and 3.

Chapter 4 and 5 has clearly illustrated the benefits of using enhanced fluidity liquids

(EFL) as the mobile phase for intact protein separations in both hydrophobic interaction chromatography (HIC) and reversed phase chromatography (RPC).

Chapter 2 described the capability of electrospun Nafion-PAN nanofibers as a strong cation exchange (SCX) stationary phase for ultrathin layer chromatography

(UTLC). Fractional factorial design was successfully implemented to recognize the significant factors that affect the chromatographic performance for amino acids and intact proteins. Response surface methodology (Box-Behnken design) was further applied to optimize parameters related to stationary phase properties and mobile phase constituents to achieve optimum separation. The developed Nafion-PAN UTLC devices exhibited high chemical stability under various mobile phase conditions. The separation of amino acids showed high selectivity and separation efficiency. Despite the rather complicated

193 retention mechanism of intact proteins on this developed separation system, which relied on a combination of protein properties including the net charge, hydropathicity, molecular size and structure, the successfully separation of four model proteins with molecular weights ranging from 14.3 to 66.5 kDa demonstrated the feasibility of Nafion-

PAN UTLC for the separation of large biomolecules.

The application of nanofibrous UTLC was extended to size fractionation and characterization of synthetic nanoparticles in Chapter 3. AuNPs with thiol terminated polyethylene glycol (PEG) coating are of great interest in biomedical applications.

Combined with the optimized micellar mobile phase system, the electrospun polyacrylonitrile (PAN) UTLC represented a rapid and inexpensive tool for separation and characterization of PEGylated AuNPs with sizes ranging from 10 to 80 nm and 30 nm AuNPs capped with PEG lengths in the range of 2-20 kDa. The electrospun PAN

UTLC device required minimal mobile phase consumption (~5 mL) while offering faster separation and higher resolution compared to other separation methods for AuNPs.1,2,3,4

The highly biodegradable micellar mobile phase contained only 7.5% v/v of organic solvent. Using the developed micellar enhanced UTLC system, AuNPs with different core sizes and different PEG lengths were well separated within 5 min with lowest plate height < 2 µm and resolution value > 1.5.

In Chapter 4, intact proteins were separated by a dual gradient system consisted of

HIC salt-organic gradient and EFLC CO2 gradient using a custom instrument configuration for the first time. EFLC mobile phase offered increased diffusivity and lower viscosity than convention LC mobile phases. A shorter analysis time, better peak

194 shapes and higher efficiency were obtained using EFLC dual gradient for the protein mixture compared to traditional HPLC methods. All model proteins including a labile protein complex, ConA, experienced charge state shifts to higher charges and markedly improved ionization efficiency under EFLC condition. These effects were primarily due to the enhanced desolvation with the addition of CO2 in the ESI process. EFLC eluents containing liquefied CO2 presented supercharging effect for macromolecules in ESI-MS while still offering excellent chromatographic performance. This developed EFLC-MS method can be a valuable approach to separate and characterize intact proteins for top- down proteomics.

EFLC solvents consisted of H2O/MeOH/ CO2 has been demonstrated in Chapter 5 as an effective substitute for H2O/ACN in RPC for proteins. Adding small amount of liquefied CO2 (2-15% v/v) in the H2O/MeOH mixture offered comparable solvent strength to H2O/ACN. Similar speed of analysis was achieved using EFLC gradient compared to traditional H2O/ACN gradient. Better separation efficiency and peak symmetry were obtained with EFLC than using H2O/ACN and H2O/MeOH mobile phases in HPLC. To achieve better MS sensitivity in addition to successful EFLC separations, several mobile phase additives were evaluated. Among all the examined additives, TFA offered the best chromatographic performance but with ion suppression effect, while AmFm and AmAc provided better chromatographic resolution and peak shapes than their acid counterparts along with sensitive MS detection.

Overall, this dissertation provides new insights for material and method development for intact protein separations and nanoparticle characterizations.

195

6.2 Future Work

UTLC using electrospun nanofibers has shown great promise for separation of both small and large biomolecule. For complex samples, even though high quality separations are achievable, unambiguous identification of sample spots remains a challenge for using UTLC alone. Traditionally, indirect MS analysis can be performed by scraping the sample spots off the plate followed by extraction, purification and concentration of the analytes for MS detection.5 To avoid this extremely tedious sample pretreatment process, desorption ionization techniques can be adapted to perform direct analysis (Figure 6.1). Various compounds including dyes,6 alkaloids,7 gangliosides,8 lipids9 and peptides10 have been separated and characterized by conventional TLC coupled to matrixed assisted laser desorption ionization (MALDI) MS. However, the thickness of the commercial silica gel TLC has significant impact on the MS sensitivity.

Schiller et al. showed that the MS signals of lipids were markedly increased when a thinner layer (60 µm) was used instead of the common 200 µm TLC plates.11 The thickness of electrospun UTLC (10 µm) is expected to improve the MS sensitivity drastically. With the advances of elelctrospun nanofibers in UTLC applications, Olesik group also developed surface-assisted laser desorption ionization (SALDI) using polymer and carbon based electrospun nanofibers as the substrates.12 This method eliminates the matrix interference by substituting the chemical matrix with the nanofiber surface and provides enhanced sensitivity. Using electrospun UTLC as the SALDI platform after mobile phase development, drying and derivatization for direct analysis of resolved

196 analytes can be extremely valuable for identification and characterization of both small and large molecules from complex sample matrices.

Figure 6.1 Illustration of LDI, MALDI and SALDI for sample analysis on the surface of a TLC plate.5

Therapeutic monoclonal antibodies (mAbs) has become one the fastest growing drugs to treat cancer, autoimmune diseases and infections.13,14 Monitoring the heterogeneity and post translational modifications (PTMs) of mAbs is required by health authorities.15 Therefore, characterization of mAbs in their intact state is necessary. HIC is particularly useful for protein variant characterization.16 The developed EFLC-MS technique on HIC column has shown capability in separating proteins from their oxidized species. Supercharging effect offered by EFLC mobile phase can improve the

197 fragmentation efficiency in tandem MS leading to better characterization of minor PTMs.

In addition, EFLC gradients, which generated narrow peaks on HIC column, holds potential for high resolution separation of antibody drug conjugates (ADCs) varied by the number of small drugs per antibody to precisely monitor the drug distribution in ADCs.

6.3 References

1 Zhou, X.; Liu, J.; Jiang, G. Environ. Sci. Technol. 2017, 51, 3892 –3901.

2 Hansen, M.; Smith, M. C.; Crist, R. M.; Clogston, J. D.; McNeil, S. E. Anal. Bioanal. Chem. 2015, 407, 8661-8672.

3 Franze, B.; Engelhard, C. Anal. Chem. 2014, 86, 5713-5720.

4 Doane, T. L.; Cheng, Y.; Babar, A.; Hill, R. J.; Burda, C. J. Am. Chem. Soc. 2010, 132, 15624-15631.

5 Cheng, S.; Huang, M.; Shiea, J. J. Chromatogr. A 2011, 1218, 2700-2711.

6 Mehl, J. T.; Gusev, A. I.; Hercules, D. M. Chromatogr. 1997, 46, 358-364.

7 Malinowska, I.; Studzinski, M.; Niezabitowska, K.; Gadzikowska, M. Chromatographia 2013, 76, 1327-1332.

8 Dreisewerd, K.; Muthing, J.; Rohlfing, A.; Meisen, I.; Vukelic, Z.; Peter-Katalinic, J.; Hillenkamp, F.; Berkenkamp, S. Anal. Chem. 2005, 77, 4098-4107.

9 Stubiger, G.; Belgacem, O.; Rehulka, P.; Bicker, W.; Binder, B. R.; Bochkov, V. Anal. Chem. 2010, 82, 5502–5510

10 Mehl, J.; Hercules, D. M. Anal. Chem. 2000, 72, 68–73

11 Griesinger, H.; Fuchs, B.; Sub, R.; Matheis, K.; Schulz, M.; Schiller, J. Anal. Biochem. 2014, 451, 45-47.

12 Lu, T.; Olesik, S. V. Anal. Chem. 2013, 85, 4384-4391.

13 Weiner, G. J. Nat. Rev. Cancer 2015, 15, 361−370.

198

14 Leavy, O. Nat. Rev. Immunol. 2010, 10, 297−297.

15 Haverick, M.; Mengisen, S.; Shameem, M.; Ambrogelly, A. mAbs 2014, 6, 852-858.

16 Fekete, S.; Veuthey, J.; Beck, A.; Guillarme, D. J. Pharm. Biomed. Anal. 2016, 130, 3- 18.

199

Bibliography

Chapter 1

1 Blueggel, M.; Chamrad, D; Meyer, H. E. Curr Pharm Biotechnol. 2004, 5, 79-88.

2 Lee, J.; Kohn, E. C. Ann. Oncol. 2010, 21, 205-210.

3 Rifai, N.; Gillette, M. A.; Carr, S. A. Nature Biotechnol. 2006, 24, 971-983.

4 Smith, L. M.; Kelleher, N. L. Nat Methods. 2013, 10, 186–187.

5 Dykman, L.; Khlebstov, N. Chem. Soc. Rev. 2012, 41, 2256-2282.

6 Huang, X.; Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Nanomedicine 2007, 2, 681- 693.

7 Yeh, Y.; Creran, B.; Rotello, V. M. Nanoscale 2012, 4, 1871-1880.

8 Poole, C.F. The essence of chromatography; Elsevier: New York, 2003.