• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Rational high-throughput screening for formulations that physically stabilize recombinant proteins

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Andrew L Kim
Abstract:
Protein-protein interactions and protein self-association occur in a wide range of bioprocessing environments and have taken on increased importance with the rapid development of biotechnology. In the biotechnology industry, aggregation is encountered during refolding, purification, sterilization, shipping, and storage processes. Protein precipitates are also a cause, or an associated symptom, of amyloidal and prion diseases, such as Alzheimer's and Creutzfeldt-Jakob disease (Sipe 1992; Cordell 1994). Aggregation is also the most common form of physical degradation of therapeutic proteins and often results in altered immunogenicity, reduced bioactivity, and decreased bioavailability (Manning et al. 1989). Therefore in this work we have developed a method for the rational, high throughput screening of formulations that physically stabilize proteins. A method was developed for relating the colloidal stability and conformational stability of a protein to its aggregation behavior. The conformational stability of human growth hormone was characterized by measuring the Gibbs energy of unfolding (ΔGunf ) via thermal unfolding studies using circular dichroism spectroscopy for various solution conditions. These Gunf values were then coupled to B22 measurements made via self-interaction chromatography to explain real-time and accelerated stability behavior. A weak positive correlation between the measured aggregation rate and composite rate determined from measured stability parameters suggests that this approach can be a powerful tool in forecasting the tendency of a protein to aggregate. Additional data from further studies are required to provide a stronger correlation. The sensitivity of self-interaction chromatography to differences in protein association behavior due to modification of specific interaction sites was validated for α-chymotrypsin. The well characterized dimerization behavior of α-chymotrypsin was inhibited via chemical modification at a single site by a reversible inhibitor, tosyl phenylalanyl chloromethyl ketone (TPCK). Self-interaction chromatography was able to resolve the altered self-association behavior due to the single-site modification. Cross interaction coefficients were also measured that demonstrated the inhibition of dimerization of α-chymotrypsin. Apparent dimerization equilibrium constants determined by thermodynamic analysis from the measured B22 values were compared to those measured by other techniques and found to be in good agreement. This provides assurance that self-interaction chromatography is accurately probing self-association behavior. A rational approach to the identification of key surface residues that control protein association behavior was developed using self-interaction chromatography. Human growth hormone variants selected from a phage display library were investigated with the wild-type. The altered association behavior of the human growth hormone mutants due to known mutations was probed using self-interaction chromatography and compared to that of the wild-type. SG2 variant displayed significantly reduced association behavior suggesting it may be a suitable replacement for the wild-type in therapeutic usage. Qualitative analysis of the altered association behavior and sequence differences strongly suggest several residues in the wild-type sequence as potential key interaction sites to be targeted in future studies. These results demonstrate a rational approach for the use of self-interaction chromatography to probe for the key surface residues that control self-association behavior of a protein.

Table of Contents LIST OF TABLES IV LIST OF FIGURES V ACKNOWLEDGMENTS X ABSTRACT XII 1. INTRODUCTION 1 1.1 BACKGROUND 1 1.2 PROTEIN STABILITY 3 1.2.1 Physical Stability 7 1.2.2 Relationship between Chemical and Physical Stability 9 1.3 PROTEIN-PROTEIN INTERACTIONS 10 1.4 FORMULATIONS TO PHYSICALLY STABILIZE PROTEINS 17 1.5 SEARCHING FOR EFFECTIVE FORMULATIONS 25 1.6 ANALYTICAL TECHNIQUES FOR CHARACTERIZING PROTEIN-PROTEIN INTERACTIONS 28 1.7 THESIS OBJECTIVES 31 1.8 THESIS ORGANIZATION 33 2. CONNECTING AGGREGATION BEHAVIOR TO GIBBS ENERGIES OF UNFOLDING 36 2.1 SUMMARY 36 2.2 INTRODUCTION 37 2.2.1 Conformational and Colloidal Stability 37 2.2.2 Self-Interaction Chromatography 42 2.2.3 Self-Interaction Chromatography Theory 44 2.2.4 Characterization of Conformational Stability 49 2.2.4.1 Characterization of conformational stability 50 2.2.4.2 Validity of the two-state assumption 54 2.3 MATERIALS AND METHODS 57 2.3.1 Materials 57 2.3.2 Buffer Solutions 58 2.3.3 Self-Interaction Chromatography 59 2.3.4 CD spectroscopy 61 2.3.5 Fluorescence spectroscopy 67 2.3.6 Data analysis 69 2.4 RESULTS AND DISCUSSION 72 2.4.1 Overview of excipient screening with SIC 72 2.4.2 Overview of real-time and accelerated stability studies 77 2.4.3 Comparison of SIC studies with stability studies 82 2.4.4 Conformational stability from thermal denaturation experiments ...84 2.4.5 Connection ofB22 andAGu„fvia lumped model 90

3. SENSITIVITY OF SELF-INTERACTION CHROMATOGRAPHY TO MODIFICATION OF A KNOWN INTERACTION SITE 100 3.1 SUMMARY 100 3.2 INTRODUCTION 101 3.2.J SIC as a Tool for Identifying Interaction Sites 101 3.2.2 a-Chymotrypsin 103 3.2.3 Chymotrypsinogen A 107 3.2.4 Known Dimerization Behavior of Chymotrypsin and its Inhibition 110 3.2.5 Chemical Modification of Chymotrypsin with TPCK 118 3.2.6 Reversible Inhibition by N-Acetyl-L-Tryptophan 121 3.3 MATERIALS & METHODS 122 3.3.1 Chemical Modification of a-Chymotrypsin 124 3.3.2 Kinetic Measurements of Chymotrypsin Activity 126 3.3.3 Self-Interaction Chromatography 127 3.3.4 Calculation ofVirial Coefficients 130 3.4 RESULTS AND DISCUSSION 133 3.4.1 Self-Interaction Chromatography with Chymotrypsinogen A 133 3.4.2 Effects of a Specific Modification on Self-Association 135 3.4.3 Measurement of Cross-Association Behavior 138 3.4.4 Thermodynamic Analysis of Self-Association Behavior 142 3.4.5 Effects of a Reversible Inhibitor on Self-Association 146 4. RATIONAL SCREENING TO IDENTIFY THE SURFACE RESIDUES THAT CONTROL SELF-ASSOCIATION BEHAVIOR OF A PROTEIN 155 4.1 SUMMARY 155 4.2 INTRODUCTION 156 4.2.1 Mutational Effects on Protein Stability 158 4.2.2 Human Growth Hormone Background 164 4.2.3 Aggregation Behavior of Human Growth Hormone 168 4.2.4 Methods to Introduce Mutations 169 4.3 MATERIALS & METHODS 174 4.3.1 Materials 174 4.3.2 Methods 777 4.3.3 hGH Expression 178 4.3.4 hGH Purification 188 4.3.5 Self-Interaction Chromatography 201 AA RESULTS AND DISCUSSION 202 4.4.1 Shotgun Ala Scanning: Library Design and Binding Selections.... 202 4.4.2 Measurement of Cross-Interaction Behavior for Selected SG Sequences 207 5. CONCLUSIONS AND FUTURE WORK 212 5.1 CONCLUSIONS 212 5.2 FUTURE WORK 215 5.2.1 Phage display and various screening modes 216 5.2.2 Cross coefficient Measurements (B23) 217 11

APPENDIX I: STABILITY STUDIES OF RHGH 245 1.1 LIST OF EXCIPIENT CONDITIONS EXAMINED IN REAL-TIME AND ACCELERATED STABILITY STUDIES 245 1.2 REAL-TIME AND ACCELERATED SEC SOLUTION STABILITY RESULTS (% MONOMER VERSUS EXCIPIENT CONDITION) 246 1.3 REAL-TIME AND ACCELERATED FLUORESCENCE SOLUTION STABILITY RESULTS (WAVELENGTH OF MAXIMUM EMISSION, ^MAX, VERSUS EXCIPIENT CONDITION) 247 APPENDIX II: STEP-BY-STEP PROCEDURE FOR PREPARING CD DATA FOR RUNNING CDPRO 248 APPENDIX III: MATLAB FITTING OF MELTING CURVE DATA TO DETERMINE AGUNF 250 APPENDIX IV: THERMAL UNFOLDING STUDIES OF RHGH 252 APPENDIX V: SIC STUDIES OF A-CHYMOTRYPSIN AND CHYMOTRYPSINOGEN A 253 V.l SIC STUDIES WITH IRREVERSIBLE INHIBITOR 253 V.2 SIC STUDIES WITH REVERSIBLE INHIBITOR 253 APPENDIX VI: DNA SEQUENCES OF SHOTGUN VARIANTS 254 APPENDIX VII: GENENTECH MATERIAL REQUEST WEB FORM INFORMATION 259 APPENDIX VIII: RHGH EXPRESSION AND PURIFICATION PROTOCOL . 263 iii

List of Tables TABLE 1.1 COMMON DEGRADATIVE REACTIONS IN PROTEINS 7 TABLE 1.2 ESTIMATED RESOURCES NEEDED FOR DEVELOPMENT OF FORMULATIONS THAT STABILIZE PROTEIN THERAPEUTICS 26 TABLE 1.3 ANALYTICAL TECHNIQUES FOR CHARACTERIZING PROTEIN-PROTEIN INTERACTIONS 35 TABLE 2.1 PHYSICOCHEMICAL PROPERTIES OF HUMAN GROWTH HORMONE (HGH) 57 TABLE 2.2 LIST OF INVESTIGATED EXCIPIENT CONDITIONS 58 TABLE 2.3 ABSOLUTE AND RELATIVE B22 VALUES FOR RHGH 75 TABLE 2.4 CONFORMATIONAL STABILITY PARAMETERS DETERMINED BY THERMAL DENATURATION 99 TABLE 3.1 PHYSICAL PROPERTIES OF A-CHYMOTRYPSIN AND CHYMOTRYPSINOGEN A 110 TABLE 3.2 DIMERIZATION CONSTANTS FOR CHYMOTRYPSIN AND CHYMOTRYPSINOGEN FROM THE LITERATURE 113 TABLE 3.3 KEY PARAMETERS FOR CALCULATING A VIRIAL COEFFICIENT FOR CHYMOTRYPSINOGEN AND TPCK-CHYMOTRYPSIN 132 TABLE 3.4 FREE ENERGIES OF CHYMOTRYPSINOGEN AND TPCK-CT DIMERIZATION 144 TABLE 3.5 SUMMARY OF DETERMINED B22 AND B23 VALUES IN COMPARISON WITH LITERATURE VALUES 154 TABLE 4.1 PHYSICOCHEMICAL PROPERTIES OF HUMAN GROWTH HORMONE (HGH) 175 TABLE 4.2 AP5 CULTURE MEDIUM RECIPE (1 L) 184 TABLE 4.3 AMINO ACID SEQUENCE ALIGNMENT OF RESIDUES TARGETED IN THE SITE 1 BINDING INTERFACE 206 IV

List of Figures FIGURE 2.1 REPRESENTATIVE MELTING CURVE FOR 0.025 MG/ML HUMAN GROWTH HORMONE IN 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6 WITH 0.25 MG/ML TREHALOSE. THE MONITORED QUANTITY WAS THE ELLIPTICITY AT 222 NM. THE BLUE DIAMONDS REPRESENT THE RAW DATA. THE SOLID LINE IS THE FITTED CURVE FOR A TWO-STATE UNFOLDING EQUATION AND THE DOTTED LINES REPRESENT THE PREDICTION BOUNDS FOR A 9 5 % CONFIDENCE INTERVAL FOR THE FITTED CURVE .52 FIGURE 2.2 REPRESENTATIVE SPECTRA COLLECTED AT DIFFERENT TEMPERATURES FOR 0.025 MG/ML RHGH IN 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6 WITH 0.25 MG/ML TREHALOSE .63 FIGURE 2.3 REPRESENTATIVE ESTIMATED SECONDARY STRUCTURE FRACTIONS FOR CD SPECTRA TAKEN BEFORE AND AFTER MELTING FOR 0.025 MG/ML RHGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6 WITH 0.25 MG/ML TREHALOSE. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS DETERMINED FROM THE STANDARD DEVIATION OF THE THREE ALGORITHMS USED IN CDPRO. X- AXIS LABELS DESIGNATE THE SECONDARY STRUCTURE ASSIGNMENTS .64 FIGURE 2.4 COMPARISON OF TWO NORMALIZED MELTING CURVES COLLECTED OVER DIFFERING TEMPERATURE RANGES FOR 0.75 MG/ML RHGH IN 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6. THE MONITORED QUANTITY WAS THE ELLIPTICITY AT 222 NM. THE BLUE DIAMONDS REPRESENT THE RAW DATA COLLECTED OVER THE RANGE OF 25-105° C AND THE RED SQUARES REPRESENT THE RAW DATA COLLECTED OVER THE RANGE OF 40-105° C. THERMODYNAMIC PARAMETERS WERE OBTAINED PER THE PROCEDURE DESCRIBED IN SECTION 2.3.6.1 .65 FIGURE 2.5 DIFFERENCE BETWEEN CD SPECTRA TAKEN AT 105 AND 25° C FOR A 0.05 MG/ML RHGH IN 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6 .66 FIGURE 2.6 REPRESENTATIVE THERMAL TRANSITION MONITORED BY FLUORESCENCE SPECTROSCOPY.THE SAMPLE CONTAINS 0.75 MG/ML RHGH IN 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6. EXCITATION WAS AT 295 NM AND THE DATA WAS COLLECTED AT AN EMISSION WAVELENGTH OF 340 NM .68 FIGURE 2.7 SPECTRA MEASURED FOR SAMPLE CONTAINING 0.75 MG/ML RHGH PH 7.6 BEFORE AND AFTER MELTING. THE INITIAL SCAN WAS RECORDED AT 25° C AND THE FINAL SCAN WAS RECORDED AT 99° C. THE EXCITATION WAVELENGTH WAS 295 NM .68 FIGURE 2.8 SECONDARY STRUCTURE ESTIMATES FOR THE CD SPECTRUM OF 0.75 MG/ML HGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6 WITH 7.5 MG/ML LYSINE. DATA WAS ANALYZED USING CDPRO AND THREE ALGORITHMS FROM REFERENCE SET SP43: CONTIN/LL (BLACK), SELCON3 (GRAY), AND CDSSTR (CROSSHATCH). THE AVERAGE OF EACH ESTIMATE IS ALSO SHOWN (WHITE) WITH 95% CONFIDENCE INTERVALS BOUNDED WITHIN THE ERROR BARS 71 FIGURE 2.9 RHGH B22 VALUES ACCORDING TO EXCIPIENT TYPE AND CONCENTRATION. A 5 MM SODIUM PHOSPHATE BUFFER AT PH 7.6 WAS USED FOR FORMULATION. THE PURPLE DASHED LINE AT VALUE 0.44 CORRESPONDS TO -1 X 10"4 MOL- V

ML G"2 AND THE RED DASHED LINE AT VALUE - 2.10 CORRESPONDS TO -8 X 10' 4 MOL-ML G'2. SOURCE: WILCOX2002 74 FIGURE 2.10 ACCELERATED (A) AND REAL-TIME (B) SOLUTION STABILITY RESULTS FOR RHGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6 USING SIZE-EXCLUSION CHROMATOGRAPHY. EXCIPIENT CONCENTRATIONS ARE INCLUDED IN LABELS ALONG THE X-AXIS. SOURCE: WILCOX2002 78 FIGURE 2.11 ACCELERATED (A) AND REAL-TIME (B) SOLUTION STABILITY RESULTS FOR RHGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6 USING FLUORESCENCE SPECTROSCOPY. EXCIPIENT CONCENTRATIONS ARE SHOWN IN LABELS ALONG THE X-AXIS. SOURCE: WILCOX 2002 81 FIGURE 2.12 MELTING TEMPERATURES DETERMINED FOR RHGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6. EXCIPIENT CONCENTRATIONS ARE 7.5 MG/ML UNLESS INDICATED OTHERWISE IN THE LEGEND. REFER TO TABLE 2.2 FOR A LIST OF SOLUTION CONDITIONS. ERROR BARS REPRESENT STANDARD DEVIATION VALUES FOR MEASUREMENTS IN TRIPLICATE EXCEPT FOR THE CASE OF 2.3 MG/ML LACTOSE WHICH ONLY REPRESENTS ONE DATA POINT 85 FIGURE 2.13 FREE ENERGY OF UNFOLDING VALUES DETERMINED FOR RHGH IN SODIUM PHOSPHATE BUFFER AT PH 7.6. EXCIPIENT CONCENTRATIONS ARE BE 7.5 MG/ML UNLESS INDICATED OTHERWISE IN THE LEGEND. REFER TO TABLE 2.2 FOR A LIST OF SOLUTION CONDITIONS. ERROR BARS REPRESENT STANDARD DEVIATION VALUES 87 FIGURE 2.14 SIMPLIFIED MODEL OF HGH AGGREGATION WITH KINETICS 93 FIGURE 2.15 ANALYSIS OF THE CONNECTION BETWEEN AGGREGATION BEHAVIOR TO CONFORMATIONAL STABILITY AND COLLOIDAL STABILITY FOR ALL SOLUTION CONDITIONS. PLOT SHOWS THE CORRELATION BETWEEN THE EXPERIMENTALLY DETERMINED AGGREGATION RATE OF RHGH FOR VARIOUS SOLUTION CONDITIONS AND THE COMBINED RATE FROM SIC AND THERMAL UNFOLDING DATA 95 FIGURE 2.16 ANALYSIS OF THE CONNECTION BETWEEN AGGREGATION BEHAVIOR TO CONFORMATIONAL STABILITY AND COLLOIDAL STABILITY FOR ALL SOLUTION CONDITIONS. PLOTS SHOW THE CORRELATION BETWEEN THE EXPERIMENTALLY DETERMINED AGGREGATION RATE OF RHGH AND THE COMBINED RATE FROM SIC AND THERMAL UNFOLDING DATA FOR (A) ALL SOLUTION CONDITIONS EXCEPT THOSE OF THE REDUCING SUGARS AND (B) ONLY THE REDUCING SUGARS (MANNOSE, FRUCTOSE, AND LACTOSE). DATA POINTS NOT INPUTTED INTO THE CORRELATION ARE SHOWN IN GRAY FOR EACH PLOT 97 FIGURE 3.1 RIBBON DIAGRAM OF CC-CHYMOTRYPSIN SHOWING SECONDARY STRUCTURE ELEMENTS. A-HELIX STRUCTURE IS COLORED IN MAGENTA AND P-SHEET IS COLORED IN ORANGE. THE THREE RESIDUES THAT FORM THE CATALYTIC TRIAD ARE ALSO SHOWN IN WIREFRAME FORMAT WITH ACCOMPANYING LABELS 105 FIGURE 3.2 CLOSE-UP VIEW OF THE ACTIVE SITE OF CC-CHYMOTRYPSIN. A-HELIX STRUCTURE IS COLORED IN MAGENTA AND P-SHEET IS COLORED IN ORANGE. THE THREE RESIDUES THAT FORM THE CATALYTIC TRIAD ARE ALSO SHOWN IN WIREFRAME FORMAT. SER 195 IS COLORED YELLOW, HIS 57 IS COLORED BLUE, AND ASP 102 IS COLORED RED 106 VI

FIGURE 3.3 RIBBON DIAGRAM OF CX-CHYMOTRYPSINOGEN SHOWING SECONDARY STRUCTURE ELEMENTS. A-HELIX STRUCTURE IS COLORED IN MAGENTA AND P-SHEET IS COLORED IN ORANGE. THE THREE RESIDUES THAT NORMALLY FORM THE CATALYTIC TRIAD ARE ALSO SHOWN IN WIREFRAME FORMAT WITH ACCOMPANYING LABELS .108 FIGURE 3.4 DEPENDENCE OF THE NATURAL LOGARITHM OF THE DIMERIZATION EQUILIBRIUM CONSTANT ON PH. CONDITIONS: 0.1 M NAC L, 0.01 M ACETATE BUFFER, 25°C. CIRCLES, EXPERIMENTAL POINTS; SOLID AND DASHED LINES, THEORETICAL CURVES. SOURCE: AUNE AND TlMASHEFF 1971 116 FIGURE 3.5 REACTION OF TPCK WITH CHYMOTRYPSIN TO FORM TPCK-CHYMOTRYPSIN, A MODIFIED FORM OF THE ENZYME THAT IS INACTIVE AND UNABLE TO DIMERIZE. SOURCE: WHITAKER 1993 120 FIGURE 3.6 THE PH DEPENDENCE OF THE INACTIVATION OF CHYMOTRYPSIN BY TPCK. SCHOELLMANN AND SHAW 1963 125 FIGURE 3.7 VIRIAL COEFFICIENT VALUES DETERMINED BY EXPERIMENT IN COMPARISON WITH LITERATURE VALUES FOR CHYMOTRYPSINOGEN AT PH 3. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS 134 FIGURE 3.8 VIRIAL COEFFICIENT VALUES DETERMINED BY EXPERIMENT IN COMPARISON WITH LITERATURE VALUES FOR CHYMOTRYPSINOGEN AT PH 4.5. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 134 FIGURE 3.9 VIRIAL COEFFICIENT VALUES DETERMINED BY EXPERIMENT IN COMPARISON WITH LITERATURE VALUES FOR CHYMOTRYPSINOGEN AT PH 6. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS 135 FIGURE 3.10 VIRIAL COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AND TPCK- CT (WITH MOBILE PHASE TPCK- CT) AT PH 3.0. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .136 FIGURE 3.11 VIRIAL COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AND TPCK-CT (WITH MOBILE PHASE TPCK-CT) AT PH 6.0. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .137 FIGURE 3.12 VIRIAL COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AND TPCK-CT (WITH MOBILE PHASE TPCK-CT) AT PH 4.5. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .138 FIGURE 3.13 VIRIAL CROSS COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE TPCK- CT) AND TPCK- CT (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AT PH 3.0. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .140 FIGURE 3.14 VIRIAL CROSS COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE TPCK- CT) AND TPCK- CT (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AT PH 6.0. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .141 vii

FIGURE 3.15 VIRIAL CROSS COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR CHYMOTRYPSINOGEN (WITH MOBILE PHASE TPCK- CT) AND TPCK- CT (WITH MOBILE PHASE CHYMOTRYPSINOGEN) AT PH 4.5. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS 141 FIGURE 3.16 VIRIAL COEFFICIENT VALUES DETERMINED BY SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 3.0 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 147 FIGURE 3.17 VIRIAL COEFFICIENT VALUES DETERMINED BY SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 4.5 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 148 FIGURE 3.18 VIRIAL COEFFICIENT VALUES DETERMINED SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 6.0 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 148 FIGURE 3.19 VIRIAL COEFFICIENT VALUES DETERMINED SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 3.0 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. THESE RESULTS ARE COMPARED TO PREVIOUSLY DETERMINED VALUES FOR A STUDY OF THE SAME SYSTEM WITH IRREVERSIBLE INHIBITOR, TPCK. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 150 FIGURE 3.20 VIRIAL COEFFICIENT VALUES DETERMINED SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 4.5 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. THESE RESULTS ARE COMPARED TO PREVIOUSLY DETERMINED VALUES FOR A STUDY OF THE SAME SYSTEM WITH IRREVERSIBLE INHIBITOR, TPCK. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS 150 FIGURE 3.21 VIRIAL COEFFICIENT VALUES DETERMINED SIC FOR A MOBILE PHASE CHYMOTRYPSINOGEN/STATIONARY PHASE CHYMOTRYPSINOGEN SYSTEM AT PH 6.0 IN THE ABSENCE OR PRESENCE OF REVERSIBLE INHIBITOR, N-ACETYL- L-TRYPTOPHAN, AS INDICATED ON THE GRAPH. THESE RESULTS ARE COMPARED TO PREVIOUSLY DETERMINED VALUES FOR A STUDY OF THE SAME SYSTEM WITH IRREVERSIBLE INHIBITOR, TPCK. ERROR BARS REPRESENT 9 5 % CONFIDENCE INTERVALS 150 FIGURE 4.1 A RIBBON DIAGRAM OF THE FOUR-HELIX BUNDLE STRUCTURE OF HUMAN GROWTH HORMONE 166 FIGURE 4.2 A SPACEFILL DIAGRAM THE STRUCTURE OF HUMAN GROWTH HORMONE WHEREIN TARGET RESIDUES LOCATED IN HELIX-1 ARE COLORED RED (14, 18, 21, 22, 25, 26, AND 29), MINI-HELIX TARGET RESIDUES ARE COLORED BLUE (41-48), LOOP-60 TARGET RESIDUES ARE COLORED YELLOW (60-68), AND Vlll

HELIX-4 TARGET RESIDUES ARE COLORED GREEN (164, 167, 168, 171, 172, 174,175,176,178,179,183) 172 FIGURE 4.3 TYPICAL GROWTH CURVE FOR E. couBLll EXPRESSING WT- HGH AT 29°C IN AP 5 MEDIA AND SHAKEN 2 3 0 RPM FOR A 1 L CULTURE BATCH IN A 4 L SHAKE FLASK 166 FIGURE 4.4 FLOWCHART OF THE PURIFICATION PROTOCOL USED FOR WT- HGH AND MUTANTS 166 FIGURE 4.5 DESALTING OF CLARIFIED EXTRACT BY THE USE OF TWO 5 ML DESALTING COLUMNS CONNECTED IN SERIES IN A FPLC SYSTEM. THE NOMINAL FLOW RATE WAS SET TO 1.5 ML/MIN BUT WAS LOWERED TO 0.6 ML/MIN WHEN LOADING SAMPLE ONTO THE COLUMN. THE BUFFER USED WAS 10 MM TRIS- HCL, PH 8.0 WITH 10% GLYCEROL. 3.0 ML OF CLARIFIED EXTRACT WAS LOADED ONTO THE COLUMN FOR DESALTING .166 FIGURE 4.6 ION EXCHANGE CHROMATOGRAPHY TREATMENT OF DESALTED EXTRACT BY A 5 ML HITRAP Q FF COLUMN CONNECTED TO A FPLC SYSTEM. THE NOMINAL FLOW RATE WAS SET TO 1.5 ML/MIN BUT WAS REDUCED TO 0.6 ML/MIN WHEN LOADING THE SAMPLE VOLUME ONTO THE COLUMN. SAMPLE LOAD VOLUMES VARIED AND ARE ONLY LIMITED BY THE DYNAMIC BINDING CAPACITY OF THE COLUMN USED. THE EQUILIBRATION BUFFER WAS 10 MM TRIS-HCL, PH 8.0 WITH 10% GLYCEROL WHEREAS THE GRADIENT BUFFER WAS 10 MM TRIS-HCL, PH 8.0 WITH 10% GLYCEROL AND 1 M NACL .199 FIGURE 4.7 SIZE-EXCLUSION CHROMATOGRAPHY TREATMENT OF CONCENTRATED FRACTIONS FROM THE ION EXCHANGE STEP THAT CONTAIN HGH. THE COLUMN USED IS A SUPERDEX 75 GL ( 10 MM X 300 MM) CONNECTED TO A FPLC SYSTEM. THE NOMINAL FLOWRATE WAS 0.5 ML/MIN FOR THE ENTIRE RUN WITH A SAMPLE INJECTION VOLUME OF 0 TO 500 MICROLITERS FROM A 500 MICROLITER SAMPLE LOOP. THE EQUILIBRATION BUFFER USED WAS 5 MM SODIUM PHOSPHATE BUFFER, PH 7.6 WITH 150MMNACL .201 FIGURE 4.8 SECONDARY STRUCTURE ESTIMATES FOR WILD-TYPE HGH OBTAINED FROM GENENTECH, INC., AND VARIANTS EXPRESSED IN OUR LAB. ESTIMATES WERE DETERMINED USING CDPRO AND THE ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .203 FIGURE 4.9 SPATIAL DISTRIBUTION OF THE NON-WILD TYPE RESIDUES IN THE SITE BINDING INTERFACE FOR SGI AND SG2 VARIANTS. SEQUENCE CHANGES RELATIVE TO VHGH ARE COLOR-CODED AS FOLLOWS: ALANINE SUBSTITUTIONS ARE IN BLUE, NON-WILD TYPE SUBSTITUTIONS ARE IN RED, AND UNCHANGED WILD TYPE RESIDUES ARE IN BOLD. FIGURE ADOPTED FROM KOUADIO ET AL. 2005 207 FIGURE 4.10 VIRIAL CROSS COEFFICIENT VALUES DETERMINED BY SEPARATE EXPERIMENTS FOR WT-HGH, HGHV, SGI, AND SG2 IN THE MOBILE PHASE AND WT-HGH IN THE STATIONARY PHASE AT PH 7.6. ERROR BARS REPRESENT 95% CONFIDENCE INTERVALS .209 IX

Acknowledgments On this special occasion of receiving my doctoral degree I would like to express my heartfelt gratitude to those who so greatly enriched my experience and freely gave of themselves so that I may succeed. No words can tell of the unflagging support and encouragement my parents and younger brother have provided to me. I will always remember the precious moments they lifted me up when my spirits were most downcast and the inspiration they provided through the example of their own lives. I am deeply grateful to have been able to learn from and collaborate with my advisor, Professor Todd Przybycien. His guidance and encouragement over the years have been invaluable and many times his generous outlook buttressed my hopes. His sense of integrity, eagerness to help others, and humble nature, despite his many talents, never cease to amaze me and he provides a model after whom I seek to fashion myself as I continue on in my career. I would also like to thank the members of my doctoral committee, Professors Michael Domach, Phil LeDuc, and Dennis Prieve for their constant enthusiasm, help, and encouragement. I am indebted for the genuine interest they have shown in my work and the many valuable insights they have freely shared. I cannot express enough times how grateful I am for their understanding and patience in the last hectic weeks leading up to my defense. Thank you once again. I am filled with gratitude towards my past and present colleagues at CMU. They have all contributed in innumerable ways to my well-being and academic development from friendly greetings and valuable suggestions, to commiseration and deep-rooted friendship. I thank the members of my own lab group which includes Dr. Michael Bartkovsky, Dr. Jessica Tucker, Dr. Mayra Cisneros-Ruiz, Dr. Murni Ahmad, Saana Gaspard, Beautia Dew, Sheetal Pai, and William Hum. Other members of the lab I am fortunate to have interacted with include Dr. Jeff Savard, Dr. Shane Grosser, Dr. Andy Kusumo, Dr. Oxana Selinova, Trishna Saigal, Alexandre Ribeiro, Zhizia Zhong, Agnieszka Kalinowski, and Libby Booth. I also enjoyed a wonderful collaboration with Amanda Diienno during the last year and a half of her undergraduate studies at CMU and await with eager expectation the fruits of her own budding journey into graduate research. Also within the department, I wish to thank the CheGSA Sudsuckers softball team which provided my chief, most enjoyable diversion over the years. I had the fortune to play on many talented teams over the years, but will remember with fondness the team we fielded in the summer of 2009. We mercy-ruled the Random Walkers (statistics) 12-0 and 14-1 in the double elimination championship game for a finale that cannot be beat. Thank you for sending me and Drew off as champions after so many years of frustrating second- place finishes. Many others gave of themselves despite their hectic schedules so that I may succeed. I thank Dr. Patricia Opresko, Gregory Sowd, Fujun Liu, and Rama Rao Damerla at the University of Pittsburgh not just for their friendship but for their expertise and generosity as I struggled with protein purification. Charlotte Bartosh of the interdisciplinary lab is much appreciated for her unfailing willingness to lend a helping hand. I also wish to x

thank Hayriye Ozhalici and Dr. Bruce Armitage for their generous use of the fluorospectrometer in my early work with denaturation of hGH. I can only wish all my collaborations run so smoothly in the future. I cannot forget Dr. Sachdev Sidhu (Univ. Toronto), Dr. Anthony Kossiakoff (Univ. Chicago), Dr. Jean-Louis Kouadio (Chicago), and Dr. Henry Lowman (Genentech) for their aid in obtaining plasmids and sample material for the work in the last chapter of my thesis. Dr. Philippe Lam played a most important part in providing protein samples from Genentech and was another source of generous encouragement I cannot thank enough. As I leave CMU, I take with me not just a degree or the pride of my accomplishments but many valuable friendships. I learned just as much outside of the lab as I did within and for that I have to thank the good friends I was blessed to meet in a place so far away from California. Stacey and Drew Cunningham are two wonderful friends who have taught me much not just about "all things Pittsburgh" but also of the meaning of friendship. Your generosity and thoughtfulness will be treasured always. Adam Bowles taught me everything I know about hockey and Ohio. I hope we can still be good friends even after Cal runs all over Ohio State on September 15, 2012 at the Horseshoe. David Choi, Sarah Kim, Sarah Oh, Joseph Kim, Aaron Churchill, and Brian Kim: it is a beautiful thing when friends can come together and partake in a bond that will last the ages. You are not just my friends, but true brothers and sisters that will always be a part of my life. Last but not least I thank my God, Lord, and Savior. He not only gave me life but gives me a reason for living and a hope for tomorrow. I can look forward to tomorrow because He lives and He lives in me.

Abstract Protein-protein interactions and protein self-association occur in a wide range of bioprocessing environments and have taken on increased importance with the rapid development of biotechnology. In the biotechnology industry, aggregation is encountered during refolding, purification, sterilization, shipping, and storage processes. Protein precipitates are also a cause, or an associated symptom, of amyloidal and prion diseases, such as Alzheimer's and Creutzfeldt- Jakob disease (Sipe 1992; Cordell 1994). Aggregation is also the most common form of physical degradation of therapeutic proteins and often results in altered immunogenicity, reduced bioactivity, and decreased bioavailability (Manning et al. 1989). Therefore in this work we have developed a method for the rational, high throughput screening of formulations that physically stabilize proteins. A method was developed for relating the colloidal stability and conformational stability of a protein to its aggregation behavior. The conformational stability of human growth hormone was characterized by measuring the Gibbs energy of unfolding {AGmj) via thermal unfolding studies using circular dichroism spectroscopy for various solution conditions. These Gunf values were then coupled to B22 measurements made via self-interaction chromatography to explain real-time and accelerated stability behavior. A weak positive correlation between the measured aggregation rate and composite rate determined from measured stability parameters suggests that this approach can be a powerful tool in forecasting the tendency of a protein to aggregate. Additional data from further studies are required to provide a stronger correlation. xii

The sensitivity of self-interaction chromatography to differences in protein association behavior due to modification of specific interaction sites was validated for a-chymotrypsin. The well characterized dimerization behavior of a- chymotrypsin was inhibited via chemical modification at a single site by a reversible inhibitor, tosyl phenylalanyl chloromethyl ketone (TPCK). Self- interaction chromatography was able to resolve the altered self-association behavior due to the single-site modification. Cross interaction coefficients were also measured that demonstrated the inhibition of dimerization of a-chymotrypsin. Apparent dimerization equilibrium constants determined by thermodynamic analysis from the measured B22 values were compared to those measured by other techniques and found to be in good agreement. This provides assurance that self- interaction chromatography is accurately probing self-association behavior. A rational approach to the identification of key surface residues that control protein association behavior was developed using self-interaction chromatography. Human growth hormone variants selected from a phage display library were investigated with the wild-type. The altered association behavior of the human growth hormone mutants due to known mutations was probed using self-interaction chromatography and compared to that of the wild-type. SG2 variant displayed significantly reduced association behavior suggesting it may be a suitable replacement for the wild-type in therapeutic usage. Qualitative analysis of the altered association behavior and sequence differences strongly suggest several residues in the wild-type sequence as potential key interaction sites to be targeted in future studies. These results demonstrate a rational approach for the xiii

use of self-interaction chromatography to probe for the key surface residues that control self-association behavior of a protein.

Stability itself is nothing else than a more sluggish motion. -Michel de Montaigne L Introduction 1.1 Background Desirable and undesirable protein-protein interactions and protein aggregation occur in a wide range of protein processing environments. Though poorly understood, protein aggregation and precipitation phenomena have taken on increased importance with the rapid growth of the biotechnology industry. Ongoing development in recombinant protein engineering is forecasted to spur a rise in the sale of protein and peptide therapeutics from ~$67 billion in 2006 to $118 billion in 2011 (FierceBiotech 2007). Since current protein therapeutics target only 324 (Overington et al. 2006) of an estimated 3,000 potential drug targets (Drews 2000; Burgess and Golden 2002), there exists a large driving force for growth . One factor behind the development of new therapeutic proteins is the completion of the Human Genome Project which is now beginning to yield fruit in improved understanding of biology and disease, as well as new diagnostic tests for common diseases. Another impetus for growth lies in the molecular biology advances that permitted identification of those proteins involved in disease processes such as inflammation or angiogenesis in tumor growth (Zheng et al. 2006). Thus, the importance of aggregation in this exponentially growing industry cannot be understated as it is encountered during a multitude of processing steps, including, refolding, purification, sterilization, shipping, and storage (Manning et al. 1989). 1

Protein self-association has wide-ranging implications for other processes outside the manufacturing process. One example is its role in protein crystallization (George and Wilson 1994). Presently, the growth of protein crystals is the chief bottleneck to protein structure determination; increasing demands from structural genomics presents a need for more efficient methods of protein crystallization (Saridakis and Chayen 2003). In vivo, aggregation can be a hindrance in the development of therapeutic proteins. For example, the expression of eukaryotic genes in prokaryotes such as Escherichia coli often results in the formation of precipitates, called inclusion bodies, in the host cell (Marston 1986). In this case, although recombinant technology may help overcome the low natural availability of a desired polypeptide, the degree of success may be curbed by the presence of these inclusion bodies. Conversely, inclusion bodies may turn out to be a benefit if refolding relatively simple and efficient. Protein precipitates are also a cause, or an associated symptom, of amyloidal and prion diseases, such as Down's syndrome, Alzheimer's, and Creutzfeldt-Jakob disease; in these cases, aggregation is accompanied by loss of biological function or gain in toxicity (Gasset et al. 1992; Sipe 1992; Cordell 1994). In vitro, protein aggregation limits studies of protein folding and stability. Furthermore, aggregation of therapeutic proteins often results in altered immunogenicity, reduced bioactivity (Manning et al. 1989), and decreased bioavailability (Lougheed et al. 1980). Proteins are complex structures of marginal physical stability whose biological function is highly dependent upon maintaining structural and chemical integrity. This makes them susceptible to degradation, both chemical and physical, 2

and a loss of function can occur upon exposure to various conditions during processing and storage, such as aqueous/organic interfaces, elevated temperatures, vigorous agitation, hydrophobic surfaces, and detergents. Since every protein exhibits unique stability behavior, overcoming the inherent degradative pathways begins with identifying the causes and molecular mechanisms of inactivation of the protein of interest. Yet despite the many advances in molecular biology to develop new therapeutic proteins, similar progress has not been realized for the development of formulations for therapeutic proteins. Present formulation strategies include the use of excipients (additives) or sequence modification to stabilize proteins against aggregation behavior but remain Edisonian in nature. Consequently, there is a need for a greater understanding of the mechanism by which proteins self-associate to drive the development of targeted strategies to inhibit protein aggregation. This work was undertaken to develop a more rational and rapid approach to finding formulations that physically stabilize proteins through the use of self- interaction chromatography. 1.2 Protein Stability The stability of a protein is determined by a balance of various intermolecular and intramolecular forces that may be modulated by environmental conditions. Thus protein stability plays a major role during fermentation and purification operations of the manufacturing process. In fermentation, over- expression of a protein inside a foreign host organism may lead to cell stress. This in turn causes the release of proteolytic enzymes by apoptosis which ultimately 3

Full document contains 282 pages
Abstract: Protein-protein interactions and protein self-association occur in a wide range of bioprocessing environments and have taken on increased importance with the rapid development of biotechnology. In the biotechnology industry, aggregation is encountered during refolding, purification, sterilization, shipping, and storage processes. Protein precipitates are also a cause, or an associated symptom, of amyloidal and prion diseases, such as Alzheimer's and Creutzfeldt-Jakob disease (Sipe 1992; Cordell 1994). Aggregation is also the most common form of physical degradation of therapeutic proteins and often results in altered immunogenicity, reduced bioactivity, and decreased bioavailability (Manning et al. 1989). Therefore in this work we have developed a method for the rational, high throughput screening of formulations that physically stabilize proteins. A method was developed for relating the colloidal stability and conformational stability of a protein to its aggregation behavior. The conformational stability of human growth hormone was characterized by measuring the Gibbs energy of unfolding (ΔGunf ) via thermal unfolding studies using circular dichroism spectroscopy for various solution conditions. These Gunf values were then coupled to B22 measurements made via self-interaction chromatography to explain real-time and accelerated stability behavior. A weak positive correlation between the measured aggregation rate and composite rate determined from measured stability parameters suggests that this approach can be a powerful tool in forecasting the tendency of a protein to aggregate. Additional data from further studies are required to provide a stronger correlation. The sensitivity of self-interaction chromatography to differences in protein association behavior due to modification of specific interaction sites was validated for α-chymotrypsin. The well characterized dimerization behavior of α-chymotrypsin was inhibited via chemical modification at a single site by a reversible inhibitor, tosyl phenylalanyl chloromethyl ketone (TPCK). Self-interaction chromatography was able to resolve the altered self-association behavior due to the single-site modification. Cross interaction coefficients were also measured that demonstrated the inhibition of dimerization of α-chymotrypsin. Apparent dimerization equilibrium constants determined by thermodynamic analysis from the measured B22 values were compared to those measured by other techniques and found to be in good agreement. This provides assurance that self-interaction chromatography is accurately probing self-association behavior. A rational approach to the identification of key surface residues that control protein association behavior was developed using self-interaction chromatography. Human growth hormone variants selected from a phage display library were investigated with the wild-type. The altered association behavior of the human growth hormone mutants due to known mutations was probed using self-interaction chromatography and compared to that of the wild-type. SG2 variant displayed significantly reduced association behavior suggesting it may be a suitable replacement for the wild-type in therapeutic usage. Qualitative analysis of the altered association behavior and sequence differences strongly suggest several residues in the wild-type sequence as potential key interaction sites to be targeted in future studies. These results demonstrate a rational approach for the use of self-interaction chromatography to probe for the key surface residues that control self-association behavior of a protein.