Using Computational Tools to Understand Interactions Between Osmolytes and Optimize the Preservation of Heterogeneous Populations of Primary Cells

Home , Cryoprotectant

Using Computational Tools to Understand Interactions between Osmolytes and Optimize the Preservation of Heterogeneous Populations of Primary Cells

A Dissertation SUBMITTED TO THE FACULTY OF UNIVERSITY OF MINNESOTA BY

Chia-Hsing Pi

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Allison Hubel, Ph.D., Advisor

November, 2019

Acknowledgment

There are a number of people who sincerely deserve my deep thanks for their effort in the support of my study and the doctoral work. I firstly would like to express sincere gratitude to my advisor Professor Allison Hubel for leading me to cryopreservation, and continuous guidance and teaching of my Ph.D. study. Professor Hubel is so intelligent, innovative and optimism on research and training me to be a qualified Ph.D. I would also like to thank Professor David Odde, Professor Ognjen Ilic, and Professor Suhasa Kodandaramaiah for being on my committee and their insightful comments and suggestions that polish my research from various perspectives. I would like to thank Prof. John Bischof and Prof. Diana Negoescu for serving on the committee for written and oral preliminary exams. I would like to thank Prof. Peter Dosa for giving many valuable comments and suggestions on my research and scientific writing. I would thank Prof. Ashley Petersen for helping me to build up the statistical models.

Dozens of people have helped and taught me in the Hubel lab. These intelligent lab mates have not only helped me with experiments but also made my life in the Ph.D. program more enjoyable. Thank you to Dr. Guanglin Yu for experiments of Raman Spectroscopy, to Rui Li for teaching me cell culture and everything biomedical, to Kathlyn Hornberger for experiments with NK cells and PBMCs, to Rachel Johnson for giving me writing and speaking corrections. Several Rockstar undergraduates helped me for experiments, and they are Elizabeth Moy, Paul Esslinger, and Jacob Herbers.

I would like to thank Vanessa L. Reynolds for giving useful information about PBMCs. I also would like to sincerely thank Dr. Rose M. Wangen and Natalie Eichten for their support on the project of cryopreserving PBMC at Translational Therapy Laboratory. Dr. Wangen gave me insightful comments and suggestions on experimental design and Natalie performed tremendous work on flow cytometry. I would also like to thank Prof. Keli Hippen and Sophia Shani for sharing PBMC in the preliminary work and giving me helpful suggestions.

A sincere thank you to Prof. Kuo-Shen Chen and Prof. Kevin T. Turner, my B.S. and M.S. advisor, respectively, for keeping interested in my research, life and career after leaving the National Cheng Kung University and the University of Pennsylvania. Performing research and development with you was a really good experience in my life. I can’t be here without the foundation you established.

A quite delight thank you to my aunt, Yi Bi, and her husband, Kang Huang, for their unaccountable help such as grocery shopping and car purchasing since I have moved to Minnesota. Sheng Huang and Lisa Huang are my lovely cousins. May God’s mercy is always with you all.

I would also like to thank all my friends in Life Spring Campus Fellowship in Twin Cities Chinese Christian Church for unaccountable support and encouragement. Specially thank Bo Zhou, Mengen Zhang, Sheng Chen, Jiadi Fan, Yiao Wang and Sheng Sang. Eating fried chicken and sharing the Christian faith with these “permanent head damage” students is a unique experience in my life. All brothers and sisters make me feel happier in my Ph.D. program. I would also like to thank my friends in NCKU campus evangelical fellowship for over 10 years of friendship in Jesus Christ. They are Sheng-Wei Chang, Cheng-Cheng Wang, Bang-Shiuh Chen, Yi-Ping Lin, Chu-En Hsu, Ron-Can Hong. I can’t believe our undergraduate lives, those old and good days, were already far far far away to us.

A deep thanks to all my friends in ME 370 from various research groups. I met many smart and friendly Ph.D. students there. They are Jia Hu, Leila Ghanbari, Matthew Rynes, Zhejie Zhu, Raito Su, Rui Luo, Mian Wang. We have worked in the same building for several years and I wish all of you will be great in your career.

I like to thank Dr. Chun Liu, Han-Pin Lin and Chieh Huang for our long-term friendship at 4304 Ludlow Street in Philadelphia since 2012. You guys are the best housemates ever! I still remember everything there and miss our lives in Philadelphia. I can’t survive in the USA without all of you! We shall reunite soon.

Thank my special senior in NCKU, Dr. Chi-Chung Wang, for his countless encouragement, suggestions, and reminders during my past years in graduate school even I haven’t met you in person for many years. Thank you for sharing your stories to support me when I failed or lose faith. Thank another special mentor, Prof. Tian-Shiang Yang, for his encouragement although I haven’t taken any courses or done any research with him. He gives me the impression of a righteous professor.

Sincerely thank Pastor Chang-Le Chu and Hsiao-Feng Tsai for their teaching and praying. They are my spiritual parents in Jesus Christ. I really miss both of you. Specially thank my Muses, Yu-An Chen, for her encouragement and support in addition to her fantastic piano performance. Thank you for accompanying me in the last year of my Ph.D. program.

Last but certainly not least, I owe a huge amount of thanks to my tremendously supportive parents, Kuo-Hong Pi and Cheng-Chen Wu, and my elder sister, Chia-Jun Pi, for their endless love since I was born. I know I am so lucky to grow up in this family. Thank you for always supporting me and giving love. Without the gigantic love from my beloved family, I wouldn’t be where I am today without you.

This work is funded by the National Institute of Health under contract number R01EB023880. Parts of this work were carried out in the Characterization Facility, University of Minnesota, which received partial support from NSF through the MRSEC program. Part of this research was supported by the National Institutes of Health’s National Center for Advancing Translational Sciences, grant UL1TR002494. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health’s National Center for Advancing Translational Sciences.

Soli Deo Gloria

iii

I dedicate this dissertation to Jesus Christ for giving me faith and guidance throughout my life

I also dedicate this dissertation to my grandmothers, my parents, and my older sister for whom without their love and support I would have never found the strength to finish my Doctor of Philosophy

Abstract

Immunotherapies such as chimeric antigen receptor (CAR) T-cell therapy are emerging therapies for the treatment of cancers and persistent viral infections. It is common for immunotherapy products to be collected in one site and processed in another site.

Cryopreservation is the technology to stabilize cells at a low temperature for a variety of applications including diagnosis and treatment of disease. However, cryopreservation with the current gold standard, DMSO, can result in poor post-thaw recoveries and adverse reactions to patients upon transfusion.

In this work, we propose to understand and optimize DSMO-free cryoprotectants with combinations of non-toxic and natural osmolytes including sugars, sugar alcohols and amino acids. The post-thaw recoveries of Jurkat cells display comparable performance to

DMSO and non-linear interactions between osmolytes. Raman spectroscopy observes different protective properties of osmolytes, and statistical modeling characterizes the importance of osmolytes and their interactions. The differential evolution algorithm was applied to optimize the formulations of cryoprotectants previously, but the suboptimal control parameters reduce the performance. The influence of control parameters and four types of differential evolution algorithms are examined and optimized for DMSO-free cryoprotectants specifically. Additionally, we demonstrate that these DMSO-free cryoprotectants can cryopreserve human peripheral blood mononuclear cells as good as conventional DMSO. The advantages of DMSO-free cryoprotectants can improve the accessibility of cell therapy. It will also be critical to providing a methodology to develop multiple component DMSO-free cryoprotectants for other cell types.

Table of Content

ACKNOWLEDGMENT ...... I

ABSTRACT ...... V

TABLE OF CONTENT ...... VI

FIGURE LIST ...... XI

TABLE LIST ...... XX

ABBREVIATIONS ...... XXI

UNITS ...... XXII

CHAPTER 1: INTRODUCTION ...... 1

1.1 Motivation ...... 1

1.2 Objective ...... 3

CHAPTER 2: BACKGROUND ...... 7

2.1 Jurkat cells ...... 7

2.2 Peripheral blood mononuclear cells (PBMCs) ...... 7

2.3 Mechanism of cryopreservation ...... 8

2.3.1 Mechanism of cell damage during freezing ...... 9

2.3.2 Mechanism of cell damage during thawing ...... 10

2.4 Cryoprotectants ...... 12

2.4.1 Cryopreservation with dimethyl sulfoxide (DMSO) ...... 13

2.5 Cryopreservation Optimization ...... 14

2.6 Differential evolution algorithm ...... 16

CHAPTER 3: MATERIAL AND METHOD ...... 19

3.1 Equipment and Materials ...... 19

3.2 Jurkat cell culture ...... 22

3.3 Peripheral blood mononuclear cells (PBMC) ...... 22

3.4 Freezing and thawing ...... 23

3.4.1 Plate freezing for high-throughput screening ...... 23

3.4.2 Vial freezing of Jurkat and PBMC ...... 24

3.4.3 Thawing ...... 24

3.5 Measurement of Post-thaw viability for high-throughput screening ...... 25

3.6 Raman spectroscopy and thermally controlled stage ...... 26

3.7 Raman images/spectra analysis ...... 27

3.8 Statistical regression ...... 29

3.9 Statistical analysis ...... 31

PART 1: DEVELOPING AND UNDERSTANDING THE MULTICOMPONENT

CRYOPROTECTANT USING OSMOLYTES ...... 32

CHAPTER 4: CHARACTERIZING THE “SWEET SPOT” FOR THE

PRESERVATION OF A T-CELL LINE USING OSMOLYTES ...... 33

4.1 Introduction ...... 33

4.2 Methods...... 35

4.2.1 General methods ...... 35

4.2.2 Osmolarity...... 35

4.2.3 Toxicity ...... 35

4.3 Results ...... 36

4.3.1 Toxicity ...... 36

4.3.2 Single component studies ...... 38

vii

4.3.3 Variation in responses with cooling rate ...... 42

4.3.4 Post-thaw recovery of multiple components ...... 43

4.3.5 Raman images ...... 47

4.3.6 Statistical modeling of multicomponent solutions ...... 50

4.4 Discussion and Conclusion ...... 55

CHAPTER 5: CHARACTERIZING MODES OF ACTION AND INTERACTION

FOR MULTICOMPONENT OSMOLYTE SOLUTIONS ON JURKAT CELLS .. 59

5.1 Introduction ...... 59

5.2 Methods...... 60

5.2.1 General Methods ...... 60

5.2.2 DMSO-free cryoprotectants ...... 60

5.3 Results ...... 61

5.3.1 Variation in response with cooling rate ...... 61

5.3.2 Post-thaw recoveries of SGI, SGC, TGC, SMC ...... 63

5.3.3 Statistical modeling ...... 67

5.3.4 Raman images ...... 72

5.4 Discussion and Conclusion ...... 77

PART 2: LOCALIZING THE OPTIMAL FORMULATION IN A

MULTICOMPONENT CRYOPROTECTANT USING DIFFERENTIAL

EVOLUTION ALGORITHM ...... 82

CHAPTER 6: DIFFERENTIAL EVOLUTION FOR THE OPTIMIZATION OF

DMSO-FREE CRYOPROTECTANTS: INFLUENCE OF CONTROL

PARAMETERS...... 83

viii

6.1 Introduction ...... 83

6.2 Methods...... 85

6.2.1 General Methods ...... 85

6.2.2 Differential Evolution Algorithm ...... 85

6.2.3 Python GUI ...... 86

6.3 Results ...... 87

6.3.1 Post-thaw recovery for multi-component solutions ...... 87

6.3.2 Mutation and Crossover ...... 91

6.3.3 Population size ...... 94

6.3.4 Validation ...... 96

6.4 Discussion ...... 100

6.5 Conclusion ...... 103

PART 3: APPLY MULTICOMPONENT DMSO-FREE CRYOPROTECTANTS

TO CLINICALLY-RELEVANT CELL THERAPY PRODUCTS ...... 104

CHAPTER 7: UNDERSTANDING AND OPTIMIZING FREEZING RESPONSES

TO PBMC CRYOPRESERVED WITH DMSO-FREE CRYOPROTECTANTS . 105

7.1 Introduction ...... 105

7.2 Methods...... 107

7.2.1 General methods ...... 107

7.2.2 Differential scanning calorimetry (DSC) ...... 107

7.2.3 Characterize phenotype and viability using flow cytometry ...... 107

7.3 Results ...... 110

7.3.1 Comparison of post-thaw recovery of Jurkat, PBMC and Specific

subpopulations to osmolyte solutions ...... 110

7.3.2 Comparisons between Jurkat cells and T-cell subpopulations ...... 113

7.3.3 Optimization of DMSO-free cryoprotectant to cryopreserve dual subsets of

T-cells 115

7.3.4 Statistical modeling of freezing response between PBMC subsets ...... 117

7.3.5 Differential Scanning Calorimetry (DSC) ...... 120

7.4 Discussion and Conclusion ...... 121

CHAPTER 8: CONCLUSIONS AND FUTURE WORK ...... 124

8.1 Conclusions ...... 124

8.2 Future Work ...... 132

8.2.1 Understanding the role of amino acids using novel microscopy ...... 132

8.2.2 Simulating the interactions between cell, ice and osmolytes using molecular

dynamic simulation ...... 132

8.2.3 Optimizing cryoprotectant formulations using other computational

algorithms ...... 133

8.2.4 Improving the formulations to cryopreserve cytotoxic T-cell and other

subsets 133

BIBLIOGRAPHY ...... 135

APPENDIX ...... 147

Figure list

Chapter 3 Figure 3.1 (a) Raman spectra and images of ice, amide I, sucrose, glycerol and mannitol. Raman images were rendered based on the specific Raman signals indicated on the spectra. (b) IIF and cell boundary for AIC calculation, intracellular ice formation (IIF) was determined by the presence of OH stretch peak at 3125 cm-1. (c) Raman spectra of cell section without IIF. (d) Raman spectra of cell section with IIF. The arrow indicates the Raman signal of ice. (e) Raman spectra of SMC353 and sucrose. Shadowed area was from mannitol and used to rendered Raman images. (f) Method of calculating ellipticity, where “a” is the major axis and “b” is the minor axis. Boundary and area of ice crystal are generated through MATLAB...... 28

Chapter 4 Figure 4.1 Viabilities of Jurkat cells incubated at different time points post-exposure in different concentrations of (a) sucrose, (b) glycerol, (c) isoleucine and (d) Normosol-R and SGI155...... 37

Figure 4.2 Post-thaw recoveries of Jurkat cells cryopreserved at −1ºC/min as a function of (a) sucrose concentration; (b) glycerol concentration; and (c) isoleucine concentration...... 40

Figure 4.3 (a) Raman images of ice, amide I, and sucrose of cells cryopreserved in 730mM sucrose solution. (b) Raman images of ice, amide I, and sucrose of cells cryopreserved in 1460mM sucrose solution. (c) AIC of cells cryopreserved in 730mM and 1460mM sucrose solution (n=8, p=0.033). (d) Normalized concentration of sucrose along the white arrow in (a). (e) Normalized concentration of sucrose along the white arrow in (b). (f) Cross-sectional area of cells cryopreserved in 730 mM and 1460 mM (n=8, p<0.001). (g) Raman images of ice, amide I, and glycerol of cells cryopreserved in 4% glycerol solution...... 41

Figure 4.4 Post-thaw recoveries of 8 formulations in the corners of the parameter space (level 0 or level 5 of a given component), the optimal formulation (SGI155) and 10% DMSO control as a function of three cooling rates (−1C/min, −3C /min and −10C /min) ...... 42

Figure 4.5 Post-thaw recovery of Jurkat cells cryopreserved at 1ºC/min as a function of cryoprotectant osmolarities ...... 44

Figure 4.6 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min and plotted to show (a) the effect of sucrose with coloring by level of glycerol, (b) the effect of sucrose with coloring by level of isoleucine, (c) the effect of glycerol with coloring by level of sucrose, (d) the effect of glycerol with coloring by level of isoleucine, (e) the effect of isoleucine with coloring by level of sucrose, and (f) the effect of isoleucine with coloring by level of glycerol. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. The dashed lines indicate the post-thaw recoveries for the single-component solutions. 46

Figure 4.7 (a) Raman images of ice, amide I, and sucrose of cells cryopreserved in 146mM sucrose solution. (b) Raman images of ice, amide I, and glycerol of cells cryopreserved in 10% glycerol solution. (c) Raman images of ice, amide I, and glycerol of cells cryopreserved in the SGI155 solution. (d) Normalized concentration of sucrose along the white arrow in (a). (e) Normalized concentration of glycerol along the white arrow in (b). (f) Normalized concentration of glycerol along the white arrow in (c). (g) AIC of cells cryopreserved in 146mM sucrose solution, 10% glycerol solution and SGI155 solution (n=8, p=0.1253 between Sucrose (Suc) and Glycerol (Gly), p=0.0002 between Suc and SGI155, p=0.0009 between Gly and SGI155). (h) The cross-sectional area of cells cryopreserved in 146mM sucrose solution, 10% glycerol solution and SGI155 solution (n=8, p=0.0004 between Suc and Gly, p=0.4504 between Suc and SGI155, p=0.0007 between Gly and SGI155). (i) Cell boundary of cells cryopreserved in 146mM sucrose solution, 10% glycerol solution and SGI155 solution...... 49

xii

Figure 4.8 Estimated log odds of post-thaw recovery from the quasi-binomial model without interactions and with coloring by (a) level of glycerol and (b) level of isoleucine; and estimated log odds of post-thaw recovery from the quasi-binomial model with interactions and coloring by isoleucine level and for a glycerol level of (c) 0, (d) 1, (e) 2, (f) 3, (g) 4, and (h) 5 ...... 52

Figure 4.9 The estimated log odds of post-thaw recovery from the quasi-binomial model with interactions and coloring by glycerol level and for an isoleucine level of (a) 0, (b) 1, (c) 2, (d) 3, (e) 4, and (f) 5 ...... 53

Figure 4.10 (a) Raman images of ice, amide I, and glycerol of cells cryopreserved in the SGI353 solution. (b) AIC between cells cryopreserved in SGI155 and SGI353 solution (n=8, p=0.0001). (c) Normalized concentration of sucrose along the white arrow in (a). (d) The cross-sectional area of cells cryopreserved in SGI155 and SGI 353 (n=10, p<0.001) ...... 54

Figure 4.11 Raman spectra of an extracellular spot (labeled 1), an intracellular spot (labeled 2) and the third spot at interface between cell and extracellular ice (labeled 3) for a cell cryopreserved in 730mM sucrose solution...... 58

Chapter 5 Figure 5.1 Post-thaw recoveries of Jurkat cells cryopreserved at 1C/min, 3C/min and 10C/min with eight formulations at the corners of the parameter space (level 0 and level 5 of a given component) and 10% DMSO as a control of (a) SGC, (b) TGC, and (c) SMC...... 62

xiii sucrose for SMC. (f) The effect of glycerol with coloring level of creatine for SMC. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. Each color represents the level of sugar with all six levels of creatine. The dashed lines indicate the post-thaw recoveries for the single- component solutions...... 65

Figure 5.3 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min for varying solution compositions: (a) The effect of sucrose with coloring level of glycerol for SGC. (b) The effect of sucrose with coloring level of creatine for SGC. (c) The effect of trehalose with coloring level of glycerol for TGC. (d) The effect of trehalose with coloring level of creatine for TGC. (e) The effect of sucrose with coloring level of mannitol for SMC. (f) The effect of sucrose with coloring level of creatine for SMC. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. Each color represents the level of sugar alcohol with all six levels of creatine. The dashed lines indicate the post-thaw recoveries for the single-component solutions...... 66

Figure 5.4 Estimated log odds of post-thaw recovery from the quasi-binomial model without interactions for (a) sucrose level coloring by glycerol level for SGC, (b) sucrose level coloring by creatine level for SGC, (c) trehalose level coloring by glycerol level for TGC, (d) trehalose level coloring by creatine level for TGC, (e) sucrose level coloring by mannitol level for SMC, (f) sucrose level coloring by creatine level for SMC...... 69

Figure 5.5 Estimated log odds of post-thaw recovery from the quasi-binomial model with interactions for (a) sucrose level coloring by glycerol level for SGC, (b) sucrose level coloring by creatine level for SGC, (c) trehalose level coloring by glycerol level for TGC, (d) trehalose level coloring by creatine level for TGC, (e) sucrose level coloring by mannitol level for SMC, (f) sucrose level coloring by creatine level for SMC...... 70

xiv and mannitol of cells cryopreserved in the SMC353 solution. (c) AIC of cells cryopreserved in SGC353 and SMC353 solution (n=8, p<0.001). (d) Normalized concentration of glycerol along the white arrow in (a). (e) Normalized concentration of mannitol along the white arrow in (b) ...... 74

Figure 5.7 (a) Raman images of ice and non-frozen solution of SGI353 at −50C. (b) Raman images of ice and non-frozen solution of SGC353 at −50C. (c) Raman images of ice and non-frozen solution of TGC353 at −50C. (d) Raman images of ice and non- frozen solution of SMCC353 at −50C. (e) Ellipticity of ice crystals formed after freezing of solution compositions of SGI353, SGC353, TGC353 and SMC353 from (a), (b), (c) and (d), respectively (n=10). (e) Area of ice crystal of SGI353, SGC353, TGC353 and SMC353 from (a), (b), (c) and (d), respectively (n=10)...... 75

Figure 5.8 (a) Raman images of ice and non-frozen solution of SGC453 at −50C. (b) Ellipticity of ice crystals formed after freezing of solution compositions of SGC353 and SGC453 from Figure 5.7(b) and (a), respectively (n=10). (c) Area of ice crystal of SGC353 and SGC453 from Figure 5.7(b) and (a), respectively (n=10). (d) Raman images rendered of the signal of ice, amide I, and mannitol of Jurkat cells cryopreserved in SGC453 solution. (e) AIC of cells cryopreserved in SGC353 and SGC453 solution (n=8, p<0.01)...... 76

Chapter 6 Figure 6.1 Post-thaw recoveries of Jurkat cells cryopreserved in varying concentrations of glycerol and creatine for a given concentration of sucrose (SGC) at a cooling rate of 1ºC/min for sucrose concentration (a) 0 mM, (b) 146 mM, (c) 292 mM, (d) 438 mM, (e) 584 mM and (f) 730 mM...... 89

Figure 6.2 Post-thaw recoveries of Jurkat cells cryopreserved in varying concentrations of glycerol and isoleucine for a given concentration of sucrose (SGI) at a cooling rate of

1ºC/min for sucrose concentration (a) 0 mM, (b) 146 mM, (c) 292 mM, (d) 438 mM, (e) 584 mM and (f) 730 mM...... 90

Figure 6.3 Accuracy of post-thaw recovery to (a) DE/rand/1/bin, (b) DE/best/1/bin, (c) DE/local-to-best/1/bin and (d) DE/1/local-to-best/1/bin with self-adaption for the SGC data set with NP=9 and 0.1 increment of mutation and crossover. Grey area represents the region with accuracy higher than 95%...... 92

Figure 6.4 Accuracy of post-thaw recovery to (a) DE/rand/1/bin, (b) DE/best/1/bin, (c) DE/local-to-best/1/bin and (d) DE/1/local-to-best/1/bin with self-adaption for the SGI data set with NP=9 and 0.1 increment of mutation and crossover. Grey area represents the region with accuracy higher than 95%...... 93

Figure 6.5 The accuracy of post-thaw recovery to DE/best/1/bin with NP=9, 13, 17, 21 and 25 for the SGC data set of (a) F=0.5, Cr=0.9, (b) F=0.9, Cr=0.5, and for the SGI data set of (c) F=0.5, Cr=0.9 and (d) F=0.9, Cr=0.5...... 95

Figure 6.6 TGC optimized DE/local-to-best/1/bin with self-adaptive using initial F=0.9, Cr=0.5 for Jurkat cells. For NP=9, (a) Post-thaw recoveries of all formulations in every generation. The best formulations are presented in black. (b) Post-thaw recovery of the best member per generation. (c) The number of improved formulations per generation. For NP=18, (d) Post-thaw recoveries of all formulations in every generation. The formulations are presented in black. (e) Post-thaw recovery of the best member per generation. (f) The number of improved formulations per generation...... 98

Figure 6.7 TGI optimized DE/local-to-best/1/bin with self-adaptive using initial F=0.9, Cr=0.5 for Jurkat cells. For NP=9, (a) Post-thaw recoveries of all formulations in every generation. The best formulations are presented in black. (b) Post-thaw recovery of the best member per generation. (c) The number of improved formulations per generation. For NP=18, (d) Post-thaw recoveries of all formulations in every generation. The formulations are presented in black. (e) Post-thaw recovery of the best member per generation. (f) The number of improved formulations per generation...... 99

xvi

Chapter 7 Figure 7.1 Gating strategy to identify proportions of PBMC subsets through flow cytometry. A total of 30000 singlet events was collected (top left plot). First, white blood cell (CD45+) was identified from singlets. Granulocytes (CD15+), monocytes (CD14+) and lymphocytes (CD14-CD15-) were identified from white blood cells (CD45+). B cells (CD19+), natural killer cells (CD3-CD56+), natural killer T-cells (CD3+CD56+) and T- cells (CD3+CD56-) were identified from lymphocytes. Helper T-cells (CD3+CD4+) and cytotoxic T-cells (CD3+CD8+) were identified from T-cells...... 109

Figure 7.2 Comparison of normalized post-thaw recoveries between Jurkat and PBMC (CD45+) based on the DE algorithm for cryoprotectants (a) SGI and (b) TGI. The post- thaw recoveries were normalized to a 10% DMSO control. SGI155 and TGI155 were the optimal formulations (boxed)...... 111

Figure 7.3 (a) The average proportions major populations present in PBMC samples (left) and the proportions of lymphocyte subpopulations (n=10), (b) The post-thaw recoveries of Jurkat cells and lymphocyte subpopulations with three DMSO-free and one DMSO- containing cryoprotectant under cooling rate 1ºC/min (n=10). Sampling and measurement uncertainties for small populations might result in post-thaw recoveries over 100%. ... 112

Figure 7.4 Post-thaw recoveries of Jurkat cells, T-cells, Helper T-cells, and Cytotoxic T- cells cryopreserved in TGI, SGI and MGI cryoprotectants as well as 10% DMSO (n=10)...... 114

Figure 7.5 TGI formulations were optimized using a differential evolution of the multiobjective (DEMO) algorithm to maximize helper T-cell and cytotoxic t-cell recovery. (a) Post-thaw recoveries of helper T-cell (CD3+CD4+) using TGI formulations from Generation 0 to Generation 3. (b) Post-thaw recoveries of cytotoxic (CD3+CD8+) T-cells using TGI formulations from Generation 0 to Generation 3. (c) Post-thaw recoveries of CD4+ and CD8+ to TGI formulations generated via DEMO from Generation 0 to Generation 3. (d) Post-thaw recoveries of NKT (CD3+CD56+) cells in TGI formulations from Generation 0 to Generation 3...... 116

xvii

Figure 7.6 The predicted vs actual of post-thaw recoveries using statistical models for (a) helper T-cell (CD3+CD4+), (b) cytotoxic (CD3+CD8+) and (c) NK T-cell (CD3+CD56+). The diagonal line indicates where predicted values equal actual values. (N=32) ...... 119

Appendix Figure A.1 Post-thaw recovery of Jurkat cryopreserved with (a) SGC and (b) TGC at 1ºC/min as a function of cryoprotectant osmolarities ...... 148

Figure A.2 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min and plotted to show (a) the effect of trehalose with coloring by level of glycerol, (b) the effect of trehalose with coloring by level of isoleucine, (c) the effect of glycerol with coloring by level of trehalose, (d) the effect of glycerol with coloring by level of isoleucine, (e) the effect of isoleucine with coloring by level of trehalose, and (f) the effect of isoleucine with coloring by level of glycerol. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. The dashed lines indicate the post-thaw recoveries for the single-component solutions...... 149

Figure A.3 Estimated log odds of post-thaw recovery from the quasi-binomial model with interactions for (a) trehalose level coloring by glycerol level for TGI, (b) trehalose level coloring by isoleucine level for TGI, and (c) The predicted vs actual of post-thaw recoveries using statistical models for TGI. The diagonal line indicates where predicted values equal actual values. (N=216)...... 150

Figure A.4 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min and plotted to show (a) the effect of trehalose with coloring by level of glycerol, (b) the effect of trehalose with coloring by level of creatine, (c) the effect of glycerol with coloring by level of trehalose, (d) the effect of glycerol with coloring by level of creatine, (e) the effect of creatine with coloring by level of trehalose, and (f) the effect of creatine with coloring by level of glycerol. Each solid line demonstrates the effect of the x-axis xviii osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. The dashed lines indicate the post-thaw recoveries for the single-component solutions...... 151

Figure A.5 Estimated log odds of post-thaw recovery from the quasi-binomial model with interactions for (a) trehalose level coloring by glycerol level for TGC, (b) trehalose level coloring by creatine for TGC, and (c) The predicted vs actual of post-thaw recoveries using statistical models for TGC. The diagonal line indicates where predicted values equal actual values. (N=216)...... 152

Figure A.6 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with SGI for (a) G=0 and 1, (b) G=2 and 3, and (c) G=3 and 5 under cooling rate 1ºC/min (n=10)...... 153

Figure A.7 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with SGC for (a) G=0 and 1, (b) G=2 and 3, and (c) G=4 and 5 under cooling rate 1ºC/min...... 154

Figure A.8 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with TGC for (a) G=0 and 1, (b) G=2 and 3, and (c) G=4 and 5 under cooling rate 1ºC/min...... 155

Figure A.9 The post-thaw recoveries of Jurkat cells as a function of the molar ratio between sugar alcohol and sugar tof (a) SGI, (b) SGC, (c) TGI, and (d) TGC ...... 156

xix

Table list

Chapter 3 Table 3.1 Equipment ...... 19 Table 3.2 Materials ...... 20 Table 3.3 The properties of cryoprotective agents used in this work ...... 21 Table 3.4 Wavenumber Assignments for Raman Spectra ...... 28

Chapter 4 Table 4.1 Definition of concentration level and corresponding absolute concentration for the components tested ...... 43

Chapter 5 Table 5.1 Definition of concentration level and corresponding absolute concentration for the tested components ...... 61 Table 5.2 Coefficients and p-values of each term in interaction models ...... 71

Chapter 7 Table 7.1 Information on antibodies and fluorophores of phenotype characterization .. 108 Table 7.2 The coefficients of statistical models to helper T-cell, cytotoxic and NK T-cell ...... 118 Table 7.3 The thermophysical properties of both DMSO and DMSO-free cryoprotectants ...... 120

Abbreviations

AIC The ratio of the cross-sectional area of IIF to the cross-sectional area of the cell AO Acridine Orange CAR Chimeric antigen receptor Calcein-AM Calcein Acetoxymethyl CRF Controlled rate freezer DE Differential evolution DEMO Differential evolution for multi-objective DMSO Dimethyl sulfoxide DPBS Dulbecco’s Phosphate Buffered Saline DSC Differential scanning calorimetry FBS Fetal bovine serum FDA Food and Drug Administration IIF Intracellular ice formation MGI Maltose-Glycerol-Isoleucine NK Natural killer PI Propidium iodide PBMC Peripheral blood mononuclear cell SGI Sucrose-Glycerol-Isoleucine SGC Sucrose-Glycerol-Creatine SMC Sucrose-Mannitol-Creatine TGI Trehalose-Glycerol-Isoleucine TGC Trehalose-Glycerol-Creatine

xxi

Units

℃ Degree Celsius ℃/min Degree Celsius per minute (cooling rate) cm-1 Wavenumber µL Microliter mL Milliliter mM Millimolar mOsm milliOsmolar min Minutes

xxii

Chapter 1: Introduction

1.1 Motivation

Over the past several years, immunotherapy has emerged as the “fourth pillar” of cancer treatment. Chimeric antigen receptor (CAR) T-cell therapy is a rapidly growing therapy for the treatment of cancer [1–3]. The U.S. Food and Drug Administration (FDA) approved two CAR T-cell therapies in 2017: Kymriah, developed by Novartis for the treatment of children with acute lymphoblastic leukemia, and Yescarta, developed by

Kite for adults with advanced lymphomas. Further progress with the use of immunotherapies for the treatment of cancer as well as other diseases is also anticipated.

In these treatments, immune cells are harvested from patients, genetically modified to better enable them to target cancer cells, and then reinfused into the patients [2]. Effective preservation of these cells at each stage of the process is critical to clinical use as it enables transportation from the site of collection to the site of manufacture and to the site of use. It also allows for coordination of the therapy with patient availability as well as proper safety and quality control testing.

In many biomedical applications, cells are collected in one location for use in a different location later. Without appropriate storage conditions during transport, these cells degrade or die, greatly limiting their utility. Cryopreservation is a commonly used technique to reduce this degradation. Cryopreservation of cells is a critical supporting technology for a variety of fields including cell therapy, cell banking, and biotechnology

[4–9]. Dimethyl sulfoxide (DMSO) has been the standard cryopreservation agent for

1 freezing cells since the 1960s [10]. However, DMSO is toxic upon infusion to patients and can lead to side effects from mild (such as nausea and vomiting) to severe (such as cardiovascular problems) or even cause death [11]. When exposed to DMSO, cells lose viability and function over time [12]. For hematopoietic cells, exposure to DMSO is typically limited to 30 min [13]. This practice adds to the complexity of the workflow associated with the preservation of cells using DMSO. Because of these drawbacks, there has been a long-standing interest in finding alternatives to DMSO for the preservation of

T-cells [14] including studying the use of osmolytes. These compounds are used in nature to stabilize biological systems subjected to environmental extremes [15].

DMSO-free cryoprotectants with biologically non-toxic components could improve both viability and functionality of immune cells and make immunotherapy more accessible to more patients, because DMSO-free cryoprotectants would require fewer processing steps post-thaw to be handling as a safe and functional product. A novel DMSO-free cryoprotectant may require optimizing the formulation and the corresponding protocol.

However, empirical (i.e., trial-and-error) approach is the current methodology to optimize the cryopreservation protocol. This method not only consumes a huge amount of time and money, but also limits the experimental size. An optimized methodology for optimizing cryoprotectant may be necessary for this work. This work will complement our knowledge of the development of DMSO-free cryoprotectants and improve associated cryopreservation protocols.

1.2 Objective

This research proposed to develop a DMSO-free cryoprotectant with combinations of osmolytes and understand their mechanism of interactions through low-temperature

Raman spectroscopy and statistical modeling, improve the optimizing efficiency of the differential evolution algorithm and apply DMSO-free cryoprotectants from immortalized cell lines to human peripheral blood mononuclear cells. In order to fulfill this objective, the following hypotheses and aims were proposed.

Hypothesis 1: Osmolytes act in concert to improve cell viability and interactions between different osmolytes is critical for the overall performance of the solutions.

This hypothesis can be tested using the following aims:

Aim 1.1: Measure post-thaw recovery of Jurkat cells cryopreserved with single osmolytes and combinations of osmolytes using high-throughput screening.

The post-thaw recoveries of Jurkat cells cryopreserved with sucrose, glycerol, and isoleucine individually as well as the combinations of sucrose, trehalose, glycerol, mannitol, creatine, and isoleucine will be measured through high-throughput screening.

Three cooling rates will be tested initially to find the appropriate one for the high- throughput screening.

Aim 1.2: Characterize the osmolyte distribution of lymphocytes and ice morphology during freezing using Raman spectroscopy.

Jurkat cells cryopreserved with DMSO and four combinations of osmolytes under the same cooling rate will be examined in this work. The interactions among cell membrane, extracellular/intercellular ice and cryoprotectants (both penetrating and non-penetrating) will be characterized with Raman spectroscopy in detail to reveal the mechanism of intracellular ice formation and protective properties of cryoprotectants.

The ice morphology including size and ellipticity of four DMSO-free cryoprotectants are also analyzed to understand the interactions between water and cryoprotectants and correlate to the freezing response of cells.

Aim 1.3: Analyze the main effects and interactions of osmolytes using statistical modeling.

The role of each osmolyte and its interactions with other osmolytes on post-thaw recovery is poorly understood. Statistical modeling was used to analyze the experimental data from high-throughput screening in order to sort out the influence of different osmolytes and their interactions with one another. This analysis will provide the foundation for a molecular model of protection and osmolyte interaction.

Hypothesis 2: Different controlling parameters influence the ability of the differential evolution algorithm to optimize the compositions of DMSO-free cryoprotectants.

This hypothesis can be tested using the following aim:

Aim 2: The experimental data is used to test several types of differential evolution algorithms and their control parameters including mutation, crossover and population size.

In order to accelerate the development process of DMSO-free cryoprotectants, a differential evolution algorithm has been utilized in the previous work. It has been indicated that using non-optimal control parameters results in lower efficiency or even stagnation. The efficiencies of several variants and the influences of their control parameters will be examined to tune this computational algorithm to specifically optimize the formulations of DMSO-free cryoprotectants. High-throughput screening data are used to select appropriate types of differential evolution algorithms and understand the influences of control parameters.

Four different types of differential evolution algorithms include random, best, local-to- best and self-adaption will be tested as well as the effects of their control parameters including mutation, crossover and population size. Two combinations of osmolytes will be used to validate selected types and tuned control parameters.

Hypothesis 3: The DMSO-free cryoprotectants developed based on an immortalized cell line provide equivalent performance to various types of human immune cells in comparison to a conventional DMSO-containing cryoprotectant.

This hypothesis can be tested using the following aim:

Aim 3.1: Measure the compositions and viabilities of peripheral blood mononuclear cells subsets with DMSO-free cryoprotectants as well as DMSO.

The development and characterization of DMSO-free cryoprotectants and tuning of the differential evolution algorithm are based on the immortalized T-cell line in Chapter 4-6.

In Chapter 7, we use these DMSO-free cryoprotectants to cryopreserve human peripheral blood mononuclear cells and understand the freezing responses of different subsets.

Peripheral blood mononuclear cells from various human donors are cryopreserved with three DMSO-free cryoprotectants. The proportions of cell subsets and their viabilities were characterized using flow cytometry. The flow cytometry characterization specifically focuses on the subsets of lymphocytes including B cell, natural killer cell, T- cell (helper T, cytotoxic T, and natural killer T).

Aim 3.2: Optimize the formulation of DMSO-free cryoprotectants with differential evolution algorithm.

A differential evolution with the multi-objective (DEMO) algorithm is used to optimize the formulations of DMSO-free cryoprotectants for improving the post-thaw recoveries of different subsets simultaneously. In addition to post-thaw recoveries, the post-thaw proliferation and apoptosis are tested.

Chapter 2: Background

2.1 Jurkat cells

Jurkat cells used in this work were established from the peripheral blood of a 14-year-old boy with T-cell leukemia [16]. Jurkat cells were initially used to produce IL-2 and eventually found its significance by serving as a model to explore T-cell receptor signaling. In this work, Jurkat cells were selected as a model of primary immune cells to study the responses of T-cells to various conditions of cryopreservation.

2.2 Peripheral blood mononuclear cells (PBMCs)

Peripheral blood mononuclear cells (PBMCs) are essential components of the immune system and have multiple clinical applications. Peripheral blood mononuclear cells

(PBMCs) are any peripheral blood cells having a round nucleus. PBMCs consist of lymphocytes (T cells, B cells, NK cells) and monocytes. In humans, lymphocytes make up the majority of the PBMC population, followed by monocytes, and only a small percentage of dendritic cells. Scientists conduct research in the fields of immunology, vaccine development, transplant immunology and high-throughput screening with

PBMCs.

2.3 Mechanism of cryopreservation

The cooling rate is the most critical parameter in cryopreservation. In general, cells cannot survive either too fast or too slow cooling rates and most cells exhibit an inverted

U-shaped relationship between cellular survival and cooling rate. In the freezing process, the freezing medium becomes supercooled after its temperature drops below the freezing point. As temperature decreases further, spontaneous ice crystals are formed in the extracellular solution, increasing the solution concentration and resulting in a concentration gradient across the cell membrane [17,18]. In order to maintain the chemical and biological equilibrium across the cell membrane, water is drawn out of the cell to the extracellular solution. For high cooling rates, intracellular water does not have enough time to diffuse to the extracellular space. As freezing continues, the temperature of the intracellular solution becomes lower than the freezing point and the intracellular ice formation increases. The interaction between solidifying water and biological cells has been investigated by both experimental and numerical studies [19–24]. It is widely accepted that the interaction between extracellular ice and the cell membrane is involved in the formation of intracellular ice crystals [25–27].

Toner proposed that the surface-catalyzed heterogeneous nucleation makes the cell membrane an energetically favorable location for ice nucleation [28]. Morries reviewed the processes of ice nucleation and crystal growth during cryopreservation [29]. Dowgert and Steponkus proposed that the disruption of the cell membrane by extracellular ice is the mechanism of intracellular ice formation (IIF) [30]. For slow cooling rates, cells have sufficient time to draw out water and reduce cellular size.

Slow cooling rates cause “solution injury”, and the damage mechanism has several hypotheses. Lovelock proposed the increased concentration of solutes is the reason for slow freezing damage [31]. Steponkus proposed that excessive cell dehydration and corresponding plasma membrane destabilization is the reason for slow freezing injury

[32–34].

2.3.1 Mechanism of cell damage during freezing

If cells are frozen improperly, large distortions in the cellular system, such as membrane destruction, might happen due to ice crystal formation [34,35]. During the freezing process, ice solidifies anisotropically and rejects cells and proteins from the solid phase.

Both extracellular and intracellular ice formation can result in mechanical stress on either the cell membrane or cytoskeleton. Solutes in the cryoprotectants help to inhibit ice formation or shift the freezing temperature, resulting in more control over ice nucleation and smaller ice crystals, which limits the damage to cells. Ice nucleation is a stochastic process, and controlling ice nucleation using a controlled rate freezer has been shown to improve the cell viability [36,37].

The interactions between the intrinsic properties of cells and cryoprotectants might also affect the post-thaw recovery [38]. Cryopreservation performance depends on the cooling rate [39]. In general, slow cooling rates (<10C/min) induce extracellular ice formation, concentrate cryoprotectants in the extracellular fluid, and increase the osmotic gradient across the cell membrane to draw out water. That process, named osmotic dehydration, helps to limit intracellular ice formation which can lead to cell injury before total

9 solidification happens at slow cooling rates [17,40]. The concentration of cryoprotectants is another important factor. Increasing concentrations may destabilize the cell membrane.

Several cryobiologists investigated the effects of cryoprotectants including DMSO, glycerol, and sugars [41–44].

Cells might be damaged during the following steps in cryopreservation, including sample preparation, freezing, storage, thawing, and post-thaw assessment. Suitable cryoprotectants, controlled rate cooling, and rapid thawing can mitigate damage during cooling and thawing where cells undergo the greatest change of environments. However, cell death or post-thaw apoptosis might still happen.

Membrane integrity can be used as an index to evaluate post-thaw viability immediately with fluorescent dyes. However, thawed cells may still test positive for membrane integrity in early apoptosis, resulting in poorer viability than expected post-thaw function of the sample. Damage to the cytoskeleton or nucleus might be also present in cells post- thaw [45,46]. Blebbing of the cell membrane is a characteristic of stress and is frequently observed in apoptosis [47,48].

2.3.2 Mechanism of cell damage during thawing

Cryopreserved cells need to be thawed under a suitable warming rate for further clinical or research use, such as regenerative medicine and cell therapy. The optimal warming rate with the highest post-thaw recovery depends on the cooling rate used to freeze cells.

Cells cryopreserved at a high cooling rate need to be thawed at a rapid warming rate to

10 prevent the recrystallization of small ice crystals that formed during freezing [49,50].

Studies indicated the correlation between the recrystallization of intracellular ice and cell death for cells cooled rapidly but thawed slowly [39]. The recrystallization of intracellular ice has been demonstrated by calorimetric measurements based on a population of cells. For cells cryopreserved at a low cooling rate precludes intracellular ice formation, the mechanism of slow thawing rate is more complicated [51,52]. Slow thawing rate has no significant effect on cell recovery, can be less damaging than rapid thawing rates, or can be more damaging than rapid thawing rates as shown in three different groups [51–53].

2.4 Cryoprotectants

Cryoprotectants are classified into two categories based on their capability of diffusing across the cell membrane: penetrating cryoprotectant and non-penetrating cryoprotectant.

DMSO and glycerol are typical penetrating cryoprotectants, which have been used in cryopreservation for over half a century [10,54]. The protective properties of penetrating cryoprotectants are believed to be associated with the effects of decreasing freezing point and diluting the concentration of salts in physiological solutions.

Long-chain polymers and sugars are common non-penetrating cryoprotectants. Non- penetrating cryoprotectants can dehydrate cells and reduce the chance of intracellular ice formation due to high osmotic stress. In addition, the protective properties of non- penetrating cryoprotectants may also come from specific interactions with cell membranes or proteins. The interactions between sugars including sucrose and trehalose, and macromolecules including proteins and membranes have been examined. Many hypotheses have been proposed to explain the interaction and protective property including the water-replacement hypothesis [55–58] and the vitrification hypothesis [59].

Antifreeze proteins are another category of cryoprotectants, which are able to reduce the size and nucleation of ice crystals, and inhabit ice crystal reformation during thawing [60].

In addition, it is widely believed that antifreeze proteins can protect the cell membrane during freezing. Several articles reviewed the differences and functions of cryoprotectants [5,61].

2.4.1 Cryopreservation with dimethyl sulfoxide (DMSO)

Dimethyl sulfoxide (DMSO) was first introduced into cryopreservation in 1959 by

Lovelock and Bishop [10] and has become the universal cryoprotectant until today.

Although DMSO has been the gold standard in cryopreservation, the mechanism of protection is still not fully understood. DMSO is a small molecule and is able to penetrate the cell membrane for efficient osmotic equilibration across the cell membrane during freezing and thawing [62]. It also widely accepted that DMSO thins the cell membrane and decreases the membrane rigidity as a stabilizer. Several groups applied the molecular dynamic simulation to understand the interaction between DMSO and membrane [63,64].

Even though DMSO is easy-to-use and cheap, it is systemically toxic not only to patients

[11–13] but also to cellular products [14,65]. Current treatment protocols require washing out DMSO for thawed cellular products before transfusion, which may cause cell loss during centrifugation and aspiration, but remaining DMSO may still cause side effects.

Novel washing techniques have been proposed. For example, the Hubel group developed a microfluidic device to wash out DMSO [66,67]. For treatments with multiple transfusion of cryopreserved/thawed cellular products, using DMSO-free cryoprotectants is desirable and beneficial. Several DMSO-free cryoprotectants have been developed for cryopreserving various cell types including stem cells, immune cells and immortalized cell lines [68–72].

2.5 Cryopreservation Optimization

Many cryobiologists have attempted to optimize cryopreservation protocols for various cell types. Factorial protocol optimization is the most common method to empirically determine the global or local optimum by changing a limited range of conditions. Several parameters of the cryopreservation procedure can be tuned to identify the optimums including components of cryoprotectant, cryoprotectant concentrations, addition/removal rates, pre-freeze incubation time/temperature, cooling profile (hold time, nucleation temperature, cooling rate, steps of cooling), storage temperature, and thawing time/temperature. Dijkstra-Tiekstra et al. [43] studied a total of 32 factorial combinations to measure the effects of pre-cooling, DMSO concentration, cooling rate and storage temperature on hematopoietic progenitor cell quality. Nezarpour et al. [73] tested various

DMSO concentrations, FBS concentrations, and cooling rates on human PBMCs. Dong et al. [74] examined 64 combinations to understand differences in pre-cooling, cryoprotectants, and cooling/thawing rates to optimize rhesus monkey sperm cryopreservation. Freimark et al. [75] tested 36 combinations of cryoprotectants, equilibration period, and cooling rate to optimize the cryopreservation of mesenchymal stem cells. Chaytor et al. used 8 sugars with various concentrations and combinations of sugars and DMSO to optimize embryonic liver cells and human liver hepatocellular carcinoma cells. de Paz et al. [42] tested 5 concentrations of glycerol and 3 cooling rates for freezing brown bear ejaculated spermatozoa. Rusco et al. [76] tested various cryoprotectants and their concentrations as well as the thawing rates. Ting et al. [77] studied 16 combinations of vitrification solutions incubation times, and polymer additive to improve macaque ovarian tissue vitrification. Kearney et al. [78] tested 36

14 combinations of cryoprotectants, permeation times, and cooling rates to optimize murine skin cell cryopreservation. In general, the factorial optimization schemes examined 2-6 different values for 2-6 different parameters, limiting the total conditions tested in these studies to a controllable experimental size.

2.6 Differential evolution algorithm

The DE algorithm was developed by Storn and Price [79] and utilizes stochastic direct search and independent perturbation of population vectors to identify a global maximum from within the parameter space. An initial random vector consisting of a given population (NP) across the parameter space is selected 푥푖,퐺, 푖 = 1, 2, 3 … 푁푃 where 퐺 denotes the generation. For the purpose of this investigation, the vector is composed of a given number of solution compositions and the number of candidate solution compositions tested in each generation remains constant during the optimization process.

Cells are resuspended in a candidate solution of interest, frozen and thawed and post- thaw recovery measured. The DE algorithm utilizes the post-thaw recovery associated with a given population vector/generation to predicts solutions that may result in more favorable live cell recovery. In the classical DE algorithm, there are three factors: mutation (F), crossover (Cr) and selection that affect the manner by which a next- generation vector is chosen. These factors must be properly selected for the proper function of the optimization process [80], and different applications of DE usually require different combinations of control parameters [81]. The general convention used in the DE community is DE/x/y/z, where DE means “differential evolution”, x stands for the base vector to be perturbed, y is the number of difference vectors considered for perturbation of x and z is the type of crossover being used.

Mutation

The mutation process involves a target vector from the current generation (푥푖,퐺), a donor vector (푣푖,퐺+1) which is a mutated vector and a trial vector (푢푖,퐺+1) formed from the

16 combination of the target and donor vector. Three common mutation types including

“random”, “best” and “local-to-best” are used. For each type, a donor vector 푣푖,퐺+1 is generated with a different target vector as below.

DE/random/1/bin:

푣푖,퐺+1 = 푥푟1,퐺 + 퐹(푥푟2,퐺 − 푥푟3,퐺) (2.1)

DE/best/1/bin

푣푖,퐺+1 = 푥푏푒푠푡,퐺 + 퐹(푥푟2,퐺 − 푥푟3,퐺) (2.2)

DE/local-to-best/1/bin

푣푖,퐺+1 = 푥푖,퐺 + 퐹(푥푏푒푠푡,퐺 − 푥푖,퐺) + 퐹(푥푟2,퐺 − 푥푟3,퐺) (2.3) where 푟1, 푟2, 푟3 ∈ [1, 푁푃] are randomly chosen indices, 푥푏푒푠푡,퐺 is the vector with the highest performance in Gth generation, 퐹 is a mutation factor (퐹 ∈ [0, 1]) that controls the amplification of vector difference.

Crossover (Cr)

Crossover is used to enhance the diversity of the population after generating the donor vector through mutation. The donor vector exchanges its component with the target vector under crossover to form the trial vector. The crossover is performed on each of the

D variables whenever a randomly generated number between 0 and 1 is less than or equal to the crossover value. The scheme is outlined as Equation (2.4)

푣푗푖,퐺+1 푖푓 푟푎푛푑(푗) ≤ 퐶푟 푢푗푖,퐺+1 = { (2.4) 푥푗푖,퐺 표푡ℎ푒푟푤푖푠푒

For 푗 = 1, 2, … , 퐷, 푟푎푛푑(푗) ∈ [0, 1] is the jth evaluation of a uniform random number. 퐶푅 is the crossover constant 퐶푅 ∈ [0, 1]

Selection

Selection is used to determine whether the target or the trial vector survives to the next generation. A selection scheme is used as Equation (2.5)

푢 푖,퐺 푖푓 푓(푢푖,퐺+1) > 푓(푥푖,퐺) 푓표푟 푚푎푥푖푚푖푧푎푡푖표푛 푥푖,퐺+1 = { (2.5) 푥푖,퐺 표푡ℎ푒푟푤푖푠푒

If and only if the trial vector 푢푖,퐺+1 yields a better result than 푥푖,퐺, then 푥푖,퐺+1 is set to

푢푖,퐺; otherwise, the old value is reused. According to Storn et al. [79], the efficiency of

DE is sensitive to these controlled parameters. The common choices are: 퐹 ∈ [0.5, 1],

퐶푅 ∈ [0.8, 1], and 푁푃 = 10퐷.

Self-adaptive (SA) DE

The trial-and-error method used for tuning parameters requires several runs. According to

Brest et al. [81], self-adaptive DE was proposed to adjust both crossover (Cr) and mutation factor (F) in each generation for eliminating the manual tuning of control parameters. The self-adaptive strategy is as Equation (2.6-7)

퐹푙 + 푟푎푛푑1 ∙ 퐹푢 푖푓 푟푎푛푑2 < 휏1 퐹푖,퐺+1 = { (2.6) 퐹푖,퐺 표푡ℎ푒푟푤푖푠푒

푟푎푛푑3 푖푓 푟푎푛푑4 < 휏2 퐶푅푖,퐺+1 = { (2.7) 퐶푅푖,퐺 표푡ℎ푒푟푤푖푠푒 where 푟푎푛푑푗, 푗 ∈ {1, 2, 3, 4}, are uniformly distributed random values between 0 and 1;

휏1 and 휏2 are constants, 0.1, which represent the probabilities that update both mutation and crossover; 퐹푙 and 퐹푢 are constants values, 0.1 and 0.9, respectively. The new F and new CR take a value from [0.5, 1] and [0, 1], respectively.

Chapter 3: Material and Method

3.1 Equipment and Materials

All equipment used in this work is listed in Table 3.1 along with the manufacturer information. All equipment is from the USA unless indicated. All materials used in this work are listed in Table 3.2 along with abbreviations and manufacturer information. The properties of osmolytes and DMSO are listed in Table 3.3. All materials are from the

USA unless indicated.

Table 3.1 Equipment

Equipment Details Manufacturer Controlled rate freezer Planer Series III Kryo 10 Planer Middlesex, UK Synergy HT multi-mode Biotek Biotek plate reader Winooski, VT WITec confocal Raman WITec system Ulm, Germany 100X air objective NA 0.90 Nikon Instrument Melville, NY Water bath ThermoFisher Scientific Waltham, MA Osmometer OsmetteTM Precision systems Natick, MA Flow cytometry LSRII BD Biosciences San Jose, CA

Table 3.2 Materials

Materials Details Abbreviations Manufacturer Dimethyl Sulfoxide DMSO Bioniche Pharma Belleville, ON, Canada Dulbecco’s DPBS Invitrogen Phosphate Grand Island, NY Buffered Saline Fetal Bovine Serum FBS Hyclone-Thermo Scientific Waltham, MA t-Flasks Tissue culture Corning treated Corning, NY Normosol-RTM PH 7.4 Hospira Lake Forest, IL RPMI High glucose Life Technologies Carlsbad, CA Acridine Orange AO Life Technologies Carlsbad, CA Propidium Iodide PI Life Technologies Carlsbad, CA 96-well plate Black wall Corning Clear bottom Corning, NY Cryovials Nunc Thermo Scientific Waltham, MA Neubauer Corning hemocytometer Corning, NY Sucrose S5-3 Fisher Scientific Pittsburgh, PA Trehalose T5251 Sigma St. Louis, MO Maltose Sigma St. Louis, MO Lactose Sigma St. Louis, MO Glycerol NDC 0395- Sigma 1031-16 St. Louis, MO Isoleucine I2752 Sigma St. Louis, MO Mannitol M4125 Sigma St. Louis, MO Creatine C780 Sigma St. Louis, MO

Table 3.3 The properties of cryoprotective agents used in this work

Properties Chemical Molar Solubility Solubility Structure formula mass in water in water Chemicals (g/mol) (g/L) (mM)

Sucrose C12H22O11 342.30 2100.00 5840

Trehalose C12H22O11 378.33 689 1821

Mannitol C6H14O6 182.17 182.17 1000

Glycerol C3H8O3 92.09 miscible miscible

Creatine C4H9N3O2 131.14 13.30 100

Isoleucine C6H13NO2 131.18 34.40 260

Dimethyl C2H6OS 78.13 miscible miscible sulfoxide (DMSO)

3.2 Jurkat cell culture

Jurkat cells (ATCC TIB-152), a T-cell line, were used in this study. The cell line identity was confirmed by Short Tandem Repeat (STR) by the American Type Culture Collection.

Cells were cultured in high-glucose RPMI 1640 (Life Technologies, Carlsbad, CA, USA) with 10% fetal bovine serum (FBS; Qualified, Life Technologies). Cultures were maintained at densities ranging between 1×105 and 3×106 cells/mL. Samples for Raman spectroscopy were prepared by washing and centrifuging cells twice in Dulbecco’s

Phosphate Buffered Saline at 125×g for 10 min. Cells were then resuspended in the experimental solution of interest and frozen using a thermally controlled stage or cryopreserved in 96-well plates as described below.

3.3 Peripheral blood mononuclear cells (PBMC)

Samples were collected from healthy volunteers (n=10 including 3 females and 7 males, age: 39-70, mean age: 55.7 and median: 57.5), with informed consent and IRB approval at Memorial Blood Center (Saint Paul, MN). Peripheral blood mononuclear cells

(PBMCs) were recovered from leukoreduction system chambers (LSCs) [82]. PBMCs were stored overnight at room temperature and atmospheric conditions. The average cell number is 1 × 109 in approximately 5-10 mL. Demographic information including age, gender and race were released if disclosed by the donor. There is no significant difference between male and female donors in terms of the compositions of PBMCs pre-freezing.

3.4 Freezing and thawing

3.4.1 Plate freezing for high-throughput screening

Cells were frozen in 96-well microplates (Corning, NY, USA) for high-throughput freezing experiments. Test solutions were made at 2× of their final concentration in

Normosol-R (Hospira, Lake Forest, IL). Cells were centrifuged and resuspended in

Normosol-R and then combined 1:1 with the 2× solution, using a single-step addition in clear-bottom black 96-well plates to produce a 1× concentration of cryoprotectant solution with a total volume of 50μL and a cell concentration of 300,000 cells/well (6 million cells/mL). Cells were frozen in 10% DMSO with culture media as a control. All experimental studies were performed in triplicate wells on each plate. The cells were incubated in the solutions of interest for one hour at room temperature under atmospheric conditions in the plates before being sealed with silicone round well covers (Laboratory

Supply Distributers, Millville, NJ, USA) to prevent desiccation during freezing and storage.

A controlled-rate freezer (Series III Kryo 10; Planer, Middlesex, UK) was used to cryopreserve all samples. The plates were placed in a plastic rack in a controlled-rate freezer, and frozen using the following profile: (1) start at 20C, (2) −10C/min to 0C, (3) hold at 0C for 15 min, (4) −1C/min to −8C, (5) −50C/min to −45C, (6) +15C/min to −12C, (7) −1C/min to −60C, and (8) −10C/min to −100C. The rapid cooling and rewarming (steps 5 and 6) are used to induce nucleation in the extracellular solution at

−8C. After the freezing procedure was completed, plates were stored in vapor-phase liquid nitrogen for a minimum of 24 hours up to several days until thawed.

3.4.2 Vial freezing of Jurkat and PBMC

Cells diluted in Normosol-R blank solution were transferred to cryovials and then stepwise added an equal volume of cryoprotectants at 2× designed final concentrations.

10% DMSO controls were also prepared. After 1-hour incubation, vials were transferred to the control rate freezer (CRF) (Planar Series III Kyro 10) with the same cooling as described in Section 3.4.1.

3.4.3 Thawing

Frozen plates or vials were thawed in a 37C water bath in less than 3 mins. Plates or vials were submerged halfway and smoothly agitated until only minuscule ice pellet remained. Cells were assessed for viability immediately post-thaw.

3.5 Measurement of Post-thaw viability for high-throughput screening

Calcein acetoxymethyl (Calcein-AM, Life Technologies) and propidium iodine (PI, Life

Technologies) were used to determine post-thaw recovery/viability. Calcein-AM/PI dye was added to each well at a 1:1 ratio between the dye and tested solution volume. After the addition of the dye, the plates were wrapped in aluminum foil to protect from light exposure and incubated for a half-hour at 37C and 5% CO2. The fluorescence of each plate was read at 530/590nm and 485/528nm (SynergyTM HT, BioTek Instruments,

Winooski, VT, USA). A control curve was obtained by reading plates with known numbers of live and dead cells in each well. The fluorescence readings for an experimental plate were compared to the control curve for correlating the amount of live and dead cells in each well. The post-thaw recovery was defined as the ratio of the number of live cells post-thaw to the number of seeded live cells.

3.6 Raman spectroscopy and thermally controlled stage

Raman spectroscopy measurements were conducted using a Confocal Raman Microscope

System Alpha 300R (WITec, Ulm, Germany) with a UHTS300 spectrometer and DV401

CCD detector with 600/mm grating. The WITec spectrometer was calibrated with a

Mercury-Argon lamp. A Nd: YAG laser (532 nm wavelength) was used as an excitation source. A 100× air objective (NA 0.90; Nikon Instrument, Melville, NY, USA) was used for focusing the 532nm excitation laser to the sample. The laser at the objective was

10mW, as measured by an optical power meter (Thorlabs, Newton, NJ, USA). The lateral resolution of the microscope was about 296 nm according to Abbe’s diffraction formula.

Cell samples were frozen using a four-stage Peltier (Thermonamic Electronics Corp,

Jiangxi, China) and a series 800 temperature controller (Alpha Omega Instruments Corp,

Lincoln, RI, USA). About 1-3µL of cell suspension was placed on the stage, covered with a piece of mica (TED PELLA, Redding, CA, USA) and sealed with Kapton tape

(Dupont, Wilmington, DE, USA) to prevent sample evaporation/sublimation.

The temperature of the cooling stage was maintained at 1°C and it took several seconds to cool the sample from 1°C to the seeding temperature of −6°C, at which ice was nucleated in samples using a liquid nitrogen-cooled needle, cooled at 1oC/min to a holding temperature of −50oC and held for 20 mins before imaging.

3.7 Raman images/spectra analysis

Raman images were generated by integrating spectra at each pixel based on the characteristic wavelength of common intracellular and extracellular materials (Figure

3.1(a)). Raman signals and associated wavenumbers selected for these studies are given in Table 3.4. Amide I (mainly associated with the C=O stretching vibration) and Alkyl

C=C stretches were used to generate a distribution of protein and lipid to define the area of frozen cells. Images of ice were generated with background subtraction at both sides of the peak range to separate ice and water signals. The image size was 15µm×15µm and each image had 45×45 pixels with an integration time of 0.2 seconds for each pixel. The cell boundary was determined by applying the contour function on the Raman image of amide I in WITec Project FOUR software (Figure 3.1(b)). Intracellular ice formation (IIF) was determined by the presence of an OH stretch peak at 3125cm−1. Raman spectra of cell sections with IIF showed the presence of an OH stretch peak, while Raman spectra of cell section without IIF showed the absence of an OH stretch peak (Figure 3.1(c) and (d)).

The ratio of the cross-sectional area of IIF to the cross-sectional area of the cell was calculated in ImageJ and termed an area of ice-to-cell (AIC). Due to the partial overlap of

Raman spectra of mannitol and sucrose, Raman images of mannitol were rendered based on a peak from which the contribution of sucrose was subtracted (Figure 3.1(e)). The ellipticity of ice crystals was defined as the ratio of the difference between the major axis and minor axis to the major axis (Figure 3.1(f)). The fitted ellipses of ice crystals were generated through MATLAB R2016b (MathWorks, Natick, MA, USA). The boundary and corresponding area of ice crystal were generated through MATLAB R2016b.

(a) (b) (c)

(d)

(e) (f)

Figure 3.1 (a) Raman spectra and images of ice, amide I, sucrose, glycerol and mannitol. Raman images were rendered based on the specific Raman signals indicated on the spectra. (b) IIF and cell boundary for AIC calculation, intracellular ice formation (IIF) was determined by the presence of OH stretch peak at 3125 cm-1. (c) Raman spectra of cell section without IIF. (d) Raman spectra of cell section with IIF. The arrow indicates the Raman signal of ice. (e) Raman spectra of SMC353 and sucrose. Shadowed area was from mannitol and used to rendered Raman images. (f) Method of calculating ellipticity, where “a” is the major axis and “b” is the minor axis. Boundary and area of ice crystal are generated through MATLAB.

Table 3.4 Wavenumber Assignments for Raman Spectra Substance Wavenumber (cm−1) Assignments [83–85] Ice 3125 OH stretching Protein and Lipid (Cell) 1660 Amide I and Alkyl C=C stretching Glycerol 480 CCO rock Mannitol 880 CCO stretching

3.8 Statistical regression

The first hypothesis is that molecules act in combination to improve post-thaw recovery.

It may describe the interrelationship between the different components statistically. For example, with SGI, the hypothesis is that sucrose, glycerol and isoleucine all act to preserve cells during freezing and thawing but also potential interactions between sucrose and glycerol or sucrose and isoleucine or glycerol and isoleucine may also play a role in improved outcome. Statistical inference is used to test hypotheses, to generate a measure of effect, to describe and model the relationship within the data and in other functions.

The goal in all data analysis is to extract the hidden meaning and accurate estimation from raw data. One of the common questions concerning if there is a statistical relationship between a dependent variable (or response) and the explanatory variable(s)

(or predictor(s)). There are several types of statistical models, but only the logistic regression is introduced and applied in this work.

Each cell is coded in a 96-well plate as a data point, then the post-thaw behavior is binary

(either alive or dead). The probability of post-thaw recovery can be modeled by aggregating all the cells in each well. The formula of this probability is logistic function.

It takes any real number input such as the concentrations of all components, whereas the output always takes the value between zero and unity just like the post-thaw recovery.

The variation in post-thaw recovery with composition was modeled using a quasi- binomial model. Two models were fit: (1) the main effects model to quantify the influence of each osmolyte on post-thaw recovery as Equation 3.1

푝 ln ( ) = 푎 푥 + 푎 푥 + 푎 푥 (3.1) 1−푝 1 푆 2 푆퐴 3 퐴퐴 and (2) the model with interactions to test for pairwise interactions between osmolytes as

Equation 3.2.

푝 ln ( ) = 푎 + 푎 푥 + 푎 푥 + 푎 푥 + 푎 푥 푥 + 푎 푥 푥 + 푎 푥 푥 (3.2) 1−푝 0 1 푆 2 푆퐴 3 퐴퐴 4 푆 푆퐴 5 푆 퐴퐴 6 푆퐴 퐴퐴

Where 푝 is the post-thaw recovery, 푎푖 represents the fitted coefficients, 푥푖 represents the concentration levels of osmolytes, and subscripts S, SA and AA represent sugar, sugar alcohol and amino acid, respectively. Note 푎1, 푎4 and 푎5 vary with the concentration levels of sugar.

The main effects model included predictors for the concentration levels of sugars, sugar alcohols and amino acids. The interaction models included the main effects plus the three pairwise interactions between sugars, sugar alcohols and amino acids. In both models, the concentration level of sugars was modeled as a categorical variable to allow the possibility of a non-monotonic relationship, as was previously observed that post-thaw recovery of single sugar varied non-monotonically with concentration.

3.9 Statistical analysis

Both statistical regression and analysis were performed using R programming language, version 3.4.0 (https://www.R-project.org/) for Windows OS. The statistical models were fitted using the glm function from the built-in R package with a quasibinomial family.

Mean plus/minus standard error was reported for all measurements unless otherwise noted. The sample sizes were determined with power analysis with power and alpha were

0.80 and 0.05, respectively. Two-tailed Student’s t-tests were performed for two-sample comparisons to obtain p-values.

Part 1: developing and understanding the multicomponent cryoprotectant using osmolytes

Hypothesis: Osmolytes act in concert to improve cell viability and interactions between different osmolytes is critical for the overall performance of the solutions.

Much of the text and figures in this chapter have previously appeared in the publication below

1. Pi, Chia-Hsing, Yu, Guanglin, Petersen, Ashley, and Hubel, Allison, 2018,

“Characterizing the ‘Sweet Spot’ for the Preservation of a T-Cell Line Using

Osmolytes,” Scientific Reports, 8, p. 16223

2. Pi, Chia-Hsing, Yu, Guanglin, Dosa, Peter I., Hubel, Allison, “Characterizing modes

of action and interaction for multicomponent osmolyte solutions on Jurkat cells”,

Biotechnology and Bioengineering, Volume 116, Issue 3, pp. 631-643

Guanglin Yu performed experiments with Raman spectroscopy

Chapter 4: Characterizing the “sweet spot” for the preservation of a T-cell line using osmolytes

4.1 Introduction

Diverse biological systems (plants, insects, etc.) survive high salt environments, dehydration, drought, freezing temperatures, and other stresses through the use of osmolytes [15,86–88]. In the human kidney, a mixture of five osmolytes is used to stabilize the cells [89]. Recently we developed a method of preserving cells with combinations of osmolytes [90–92]. These studies demonstrated that a combination of three different osmolytes including sugar, sugar alcohol, and amino acids/proteins could stabilize Jurkat cells and mesenchymal stromal cells (MSCs) during freezing. Each of the components plays a role in the stabilization of the cell during freezing. Sugars are associated with stabilization of the cell membrane [55] and interaction via hydrogen bonding with water [93], thereby changing solidification patterns. Glycerol also interacts strongly with water [94] via hydrogen bonds, penetrates the cell membrane [95] and is associated with the stabilization of proteins [96]. Amino acids help stabilize sugars during freezing so that they do not precipitate out of solution [97]. It is noteworthy that higher levels of osmolytes did not necessarily correspond to higher post-thaw viability

[70]. The osmolytes appeared to act in concert to improve post-thaw recovery.

The objective of this investigation is to understand in more detail the relationships amongst the osmolytes present in these solutions and Jurkat cell recovery. Raman spectroscopy has been widely used in characterizing subcellular structures such as mitochondrion, lysosome, and nucleus because it is label-free and has high spatial

33 resolution [98]. Moreover, Raman spectroscopy can identify the phase of water (liquid or solid) and the location of cryoprotective agents. For this study, low-temperature Raman spectroscopy was used to interrogate freezing responses of cells cryopreserved in different combinations of osmolytes. This tool enables us to quantify intracellular ice formation (IIF), distribution of cryoprotective agents, damage to subcellular compartments and other cell behaviors during freezing.

In a previous study, we demonstrated that osmolytes act in concert to improve cell viability [70]. A recent study demonstrated that combinations of osmolytes had a strong effect on the crystallization of water and form natural deep eutectic systems (NADES)

[99]. The next phase of the investigation will involve characterizing the role of a given osmolyte and its interactions with other osmolytes on post-thaw recovery using a statistical model. This type of analysis will provide the foundation for a molecular model of protection and osmolyte interaction. This knowledge is critical for the development of improved cryopreservation protocols for high-value cells such as cell therapies.

4.2 Methods

4.2.1 General methods

Jurkat cells were cultured according to the methods described in Section 3.2. Plate freezing and thawing were performed as described in Section 3.4. Post-thaw recovery was measure according to Section 0. Cells for Raman spectroscopy on a temperature- controlled cooling stage according to Section 3.6 and 3.7. Statistical analysis was according to section 3.8.

4.2.2 Osmolarity

The osmolarity of solutions was measured using an OSMETTETM osmometer (Precision

Systems, Natick, MA) for each solution and all measurements were repeated in triplicate.

4.2.3 Toxicity

Cryopreservation solutions are typically not physiological and exposure to the solutions can result in cell losses. In order to determine the toxicity of the candidate solutions,

Jurkat cells were exposed to candidate solutions at room temperature. The viability of the cells was determined at different time points post-exposure. The highest acceptable cell losses were set to 10% (90% viability). Cells were incubated in 96-well plates (Corning,

NY, USA) for all candidate solutions studies using the same procedure as the plate- freezing studies. The staining of live/dead cells with Calcein-AM/PI was as same as in the post-thaw assessment. All experimental studies were performed in sextuplicate wells on each plate.

4.3 Results

4.3.1 Toxicity

Initially, the variation of cell survival as a function of solution composition was determined for a single component (sugar, sugar alcohol, and amino acid). The concentration of a given osmolyte was varied from 0% to 100% of the solubility limit or alternatively the toxicity limit for the cell to screen the space with all possible formulations.

Preliminary toxicity studies were performed to determine the parameter space for the single component study. Cell losses > 10% were considered unacceptable and the upper- level of cryoprotective agents were based on that level of acceptable cell losses. For concentrations of sucrose above 730 mM, cell losses with time increased rapidly after 1- hour incubation, but cell loss was still acceptable for 2190 mM for 1-hour incubation and the upper threshold of single component studies for sucrose was set at 2190 mM (Figure

4.1(a)). Cell losses in glycerol were high for all concentrations above 10% and for times greater than one hour (Figure 4.1(b)) and as a result, the upper threshold of glycerol concentration was set at 10%. The viability of Jurkat cells in isoleucine was independent of concentration and incubation time (Figure 4.1(c)) and the upper limit of isoleucine used was based on the solubility limit. It is noteworthy that Jurkat cells incubated in

SGI155 exhibited minimal losses over the 4-hour period studied (Figure 4.1(d)).

(a) (b)

Figure 4.1 Viabilities of Jurkat cells incubated at different time points post-exposure in different concentrations of (a) sucrose, (b) glycerol, (c) isoleucine and (d) Normosol-R and SGI155.

4.3.2 Single component studies

Post-thaw recovery for Jurkat cells in sucrose varied between roughly 3% and 10% over the range of concentrations based on toxicity studies (Figure 4.2(a)). The cooling rate for single component studies was 1C/min according to previously published work[91]. The maximum post-thaw recovery occurred at roughly 730 mM. In contrast, the post-thaw recovery of cells cryopreserved in glycerol increased with increasing concentration to a threshold concentration of ~8% (Figure 4.2(b)) and achieved a maximum recovery of

40%. The post-thaw recovery of cells cryopreserved in isoleucine was low (~7%) and remained largely unchanged across the range of tested concentrations (Figure 4.2(c)).

As indicated in Figure 4.2(a), the recovery of Jurkat cells cryopreserved in sucrose solutions varied with concentration. To explore the effects of sucrose concentration on the freezing response of cells, Jurkat cells in 730mM and 1460mM sucrose solution were cryopreserved at a constant cooling rate of 1 C/min down to −50C, and Raman images rendered on the signals associated with ice, amide I and sucrose were generated (Figure

4.3(a) and Figure 4.3(b)). Cells cryopreserved in 730 mM sucrose solution showed small ice crystals (indicated by the white arrow in the image of ice) based on the presence of

OH stretching peak. In contrast, large pieces of pure ice crystals were observed in the center of cells cryopreserved in 1460 mM sucrose solution (3 out of 8 cells). Accordingly,

AIC of cells cryopreserved in 1460 mM sucrose solution was significantly greater than that of cells cryopreserved in 730mM sucrose solution (Figure 4.3(c)). For cells cryopreserved in 730 mM sucrose solution, Raman images showed that sucrose was predominantly distributed in the unfrozen solution, forming a thin layer encircling the

38 frozen cell (<1 m) (Figure 4.3(d)). For cells cryopreserved in 1460 mM sucrose solution, substantial penetration of sucrose into cells was detected in five of the eight cells studied, suggesting cell membrane of those cells was possibly damaged (Figure 4.3(e)). Raman images of amide I also showed that cells cryopreserved in 730 mM sucrose solution maintained normal but smaller cell size. The cross-sectional area of 1460 mM sucrose (57

µm2) was significantly larger (p=0.009) than 730 mM sucrose (39 µm2) (Figure 4.3(f)) once again suggesting damage to the cell membrane. On the contrary, cells cryopreserved in 1460mM sucrose solution showed irregular cell shape.

(a) (b)

(c)

Figure 4.2 Post-thaw recoveries of Jurkat cells cryopreserved at −1ºC/min as a function of (a) sucrose concentration; (b) glycerol concentration; and (c) isoleucine concentration.

(a) (b) (c)

(d) (e)

(f)

4.3.3 Variation in responses with cooling rate

The cooling rate is a key factor in post-thaw recovery and the composition can also influence the optimum cooling rate. As a result, the influence of cooling rate on post- thaw recoveries for multicomponent solutions was determined before screening the entire operation space. Eight formulations spanning the extremes of the parameter space (level

0 or level 5 of a given component) and 10% DMSO were tested with three cooling rates

(1C/min, 3C/min and 10C /min). The post-thaw recoveries were higher at 1C/min than those observed at 3C/min and 10C/min for the formulations tested (Figure 4.4). As a result, a cooling rate of 1C/min was used for subsequent experiments.

4.3.4 Post-thaw recovery of multiple components

The concentration limit of sucrose was truncated to 730 mM (the peak of post-thaw recovery) based on the single component freezing studies, glycerol was limited to 10% and isoleucine was limited to 43 mM based on the toxicity studies described above. The concentration space of each component was discretized to six levels with equal scale (216 formulations total, Table 4.1). The actual composition was described using these levels.

For example, 353 was the combination of level-three sucrose, level-five glycerol, and level-three isoleucine. The post-thaw recovery as a function of composition was determined across all 216 formulations.

Table 4.1 Definition of concentration level and corresponding absolute concentration for the components tested

Sucrose (mM) Glycerol (%) Isoleucine (mM) Level 0 0 0 0 Level 1 146 2 8.67 Level 2 292 4 17.33 Level 3 438 6 26.00 Level 4 584 8 34.67 Level 5 730 10 43.33

Post-thaw recovery was plotted as a function of osmolarity for different combinations of sucrose, glycerol, and isoleucine tested (Figure 4.5). Over a range of osmolarity from 200 to 1600mOsm/kg, there was little correlation between post-thaw recovery and osmolarity

(R2=0.2293). This result is consistent with what we have observed previously with other cell types [70].

Figure 4.5 Post-thaw recovery of Jurkat cells cryopreserved at 1ºC/min as a function of cryoprotectant osmolarities

The optimal formulation was the combination of 146 mM sucrose (level 1), 10% glycerol

(level 5) and 46 mM isoleucine (level 5) solution (SGI155) with 84% post-thaw recovery.

To visualize the interactions between osmolytes, spaghetti plots of sucrose, glycerol, isoleucine and post-thaw recovery were presented (Figure 4.6). Each subfigure showed a plot of the mean post-thaw recovery vs concentration level of one osmolyte with colors used to indicate the concentration levels of the other osmolytes. The dashed line presented the post-thaw recovery for the single component solution. For sucrose and isoleucine, the post-thaw recoveries of cells cryopreserved in multicomponent solutions were consistently higher than those for the single-component solutions (Figure 4.6(a), (b),

(e) and (f)). It is also noteworthy that the highest post-thaw recovery of sucrose alone is observed at moderate concentration but the highest post-thaw recovery for SGI was shifted to a lower concentration (146 mM). Glycerol exhibited lower post-thaw recovery for some compositions of SGI than that of the single component (Figure 4.6(c) and (d)).

Isoleucine presented a disordering effect of post-thaw recovery in both sucrose and glycerol solutions (Figure 4.6(b) and (d)). Unlike single component studies, the variation in post-thaw recovery with composition rose and fell over the parameter space.

(a) (b)

(e) (f)

4.3.5 Raman images

Post-thaw recovery of cells frozen in the SGI solution was generally higher than that of single-component solutions. In order to understand the difference, cells were cryopreserved in 146 mM sucrose solution (sucrose level 1), 10% glycerol solution

(glycerol level 5), or combination of 146 mM sucrose, 10% glycerol and 46 mM isoleucine solution (SGI155) at a constant cooling rate of 1ºC/min down to −50ºC, and typical Raman images rendered on the signals associated with ice, amide I, sucrose or glycerol were generated (Figure 4.7(a), (b) and (c)). Normalized concentrations of sucrose and glycerol determined using spectroscopy showed that sucrose was present in the extracellular space and not the intracellular (Figure 4.7(d)). Glycerol, however, was present both inside and outside the cell for cells cryopreserved in 10% glycerol (Figure

4.7(e)) and SGI155 (Figure 4.7(f)), respectively.

Cells cryopreserved in 146 mM sucrose solution displayed both small ice crystals and/or large pieces of ice. On the contrary, only small ice crystals were formed in cells cryopreserved in 10% glycerol solution. For cells cryopreserved in the SGI155 solution, little IIF was observed. The AIC of cells cryopreserved in a single-component solution was significantly greater than that of cells cryopreserved in a multicomponent solution

(Figure 4.7(g)). In contrast with sucrose, Raman images of glycerol showed considerable penetration of glycerol into all frozen cells. It was noteworthy that cells cryopreserved in single component glycerol solution appeared in a larger size (57 µm2) than those cryopreserved in solutions containing sucrose (41 µm2) as well as SGI155 (40 µm2), suggesting lower water content for cells in the multicomponent osmolyte solutions

(Figure 4.7(h)). Cells cryopreserved in single component glycerol solution also showed irregularities on the cell membrane consistent with blebbing (Figure 4.7(i)).

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

4.3.6 Statistical modeling of multicomponent solutions

The main effects model considered the individual, additive effects of each osmolyte without interactions. It showed that post-thaw recovery was dominated by increasing glycerol level (Figure 4.8(a)) while increasing the isoleucine level only led to small improvement (Figure 4.8(b)). Increasing glycerol by one level was associated with 34% higher odds of post-thaw recovery (95% CI: 29-33% higher; p<0.001). Increasing isoleucine by one level was associated with 3% higher odds of post-thaw recovery (95%

CI: 0-6% higher; p=0.09). Sucrose had a statistically significant effect on post-thaw recovery (p<0.001) with its effect peaking at level 1 and then declining (Figure 4.8(a) and

(b)).

We used the interaction model to test for pairwise interactions between osmolytes. There was evidence of interactions between sucrose and isoleucine (p=0.012) and sucrose and glycerol (p=0.014). There was no evidence of an interaction between glycerol and isoleucine (p=0.36). For the interaction model, we visualize the impact of the osmolyte levels on the estimated log odds of post-thaw recovery (Figure 4.8(c) to (h)). We see that more isoleucine is generally better unless there is a high level of sucrose, in which case isoleucine degrades the post-thaw recovery. The overall post-thaw recovery was proportional to glycerol levels, but the trends were distinct within glycerol levels (Figure

4.8(c) to 6(h)). For example, the variation of post-thaw recovery between isoleucine levels was negligible for sucrose level 2 and glycerol level 5 (Figure 4.8(h)) in comparison to variation for the same sucrose level and glycerol level 0 (Figure 4.8(c)).

Last, the best post-thaw recovery is estimated to be for sucrose level 1 and isoleucine

50 level 5 for all glycerol levels, which is consistent with experimental data. Glycerol was estimated to always have a positive association with post-thaw recovery, though the size of the effect varied based on the level of sucrose (Figure 4.9).

Raman images of ice, amide I and glycerol for cells cryopreserved in SGI155 (i.e., optimal) and SGI353 solution were generated and were consistent with the conclusions of the statistical model (Figure 4.10(a)). More IIF was observed in cells cryopreserved in solution SGI353 solution than a cell in SGI155 solution, accordingly, AIC of cells cryopreserved in SGI353 solution was greater than that of cells in SGI155 solution

(Figure 4.10(b)). Normalized glycerol concentration determined using Raman spectroscopy revealed that glycerol was also present inside the cells Figure 4.10(c)).

However, the cross-sectional area of cells cryopreserved in SGI155 (40 µm2) was significantly smaller than SGI353 (60 µm2) (Figure 4.10(d)).

(a) (b) (c)

(d) (e) (f)

(g) (h)

(a) (b)

(e) (f)

(a) (b)

4.4 Discussion and Conclusion

There has been tremendous interest in the replacement of DMSO. Trehalose, other sugars and specialty polymers have been studied as replacements for DMSO [86,100–104].

Glycerol has been used to preserve red blood cells [6,105,106]. None of these studies have found a single molecule capable of replacing DMSO. Osmolyte mixtures have been used for protein stabilization [107–109] but have not been used for cryopreserving cells.

This work used osmolyte mixtures to improve the post-thaw recovery of Jurkat cells, which was consistent with our previous study using mesenchymal stem cells [70,92]

It has long been known that water content inside the cell is an important factor in cell response during freezing [49]. In this investigation, cell size is noted as a surrogate for intracellular water content. As noted in the results, the cell size varied between the different single and multi-component solutions tested. The presence of sucrose in a solution resulted in small cell size and therefore low intracellular water content. It is noteworthy that in Figure 4.7, the area of the cells in 146 mM sucrose and SGI155 were roughly the same but the AIC for cells in the sucrose solution was very high (~0.3) with little or no ice found in the cells frozen SGI 155. Therefore, cell area/water content alone does not correlate with freezing response. Cells in the presence of glycerol alone or higher levels of sucrose exhibited larger cell sizes and therefore higher water content. In the case of the larger cell size for SGI353, the presence of intracellular ice increased the cell volume measured.

The outcome of this investigation and other studies can be used to understand molecular mechanisms of action for the osmolytes. It has long been hypothesized that disaccharides such as trehalose and sucrose could lower the transition temperature of membranes by replacing the water molecules in lipid headgroups [57,58,110,111], or by vitrification of the stabilizing solutes [112]. The spatial distribution of osmolytes was examined using a cell cryopreserved in a 730mM sucrose solution. The Raman spectra of three spots were selected from the Raman images (Figure 4.11). The Raman spectra of spot 1 showed a strong peak of sucrose but no peak for amide I, which suggested spot 1 was extracellular.

On the contrary, the Raman spectra of spot 2 showed a strong peak of amide I but no peak of sucrose, which demonstrated that sucrose did not penetrate the cell and this spot was in the cell interior. However, both signals of amide I and sucrose were detected from the Raman spectra of spot 3, indicating that the sucrose and cell had overlap on the barrier between extracellular and intracellular, the cell membrane. The observed phenomenon was consistent with the long-held theory that the protective properties of sucrose partially result from its interaction and stabilization of membranes and consistent with other Raman studies of sugars and cell membrane interactions [57,113]. A recent study has found that non-penetrating cryoprotectants can also provide protection [114] suggesting that stabilization of the cell membrane may be critical for post-thaw recovery.

Sugars such as sucrose also interact with water. Sucrose has been shown to have a destructuring effect on the water tetrahedral hydrogen bond network that has been observed in both experimental studies and molecular dynamics simulations [115,116].

For high concentration sucrose solutions, it was found that all the water molecules were

56 involved in hydrogen bonds with sucrose [93] and that the hydrogen bonds formed between sucrose and water significantly slowed down the water dynamics [117]. The interaction between sucrose and water can manifest on a macroscale. Bailey and colleagues found that the addition of sucrose to dimethyl sulfoxide changed the ice crystal patterns observed upon freezing [118].

The statistical model suggests that glycerol plays a major role in cell survival and interactions between glycerol and sucrose influence post-thaw recovery as well. The influence of glycerol on cell survival has been known for over 60 years [119]. Glycerol has long been associated with the stabilization of proteins [120]. As demonstrated in

Figure 4.7 and Figure 4.10, glycerol penetrates the cell membrane and provides a stabilizing benefit in the intracellular space.

As with sugars, the results in this study suggest that sugar alcohols act on water molecules. Previous studies have shown that the hydrogen bonding between glycerol and water plays a significant role to inhibit ice crystallization and the structure of ice crystals formed during freezing [84,94,121,122]. A recent study demonstrated changes in the structure of ice formed in the presence of different sugar alcohols [69]. The result of this investigation is consistent with those previous studies.

Interactions between sucrose and isoleucine determined with the statistical model are consistent with the observation by Wen and colleagues that the presence of specific

57 proteins actually stabilizes trehalose during freezing and prevents precipitation [97] and suggest an important role in the solution.

Figure 4.11 Raman spectra of an extracellular spot (labeled 1), an intracellular spot (labeled 2) and the third spot at the interface between cell and extracellular ice (labeled 3) for a cell cryopreserved in 730mM sucrose solution.

Chapter 5: Characterizing modes of action and interaction for multicomponent osmolyte solutions on Jurkat cells

5.1 Introduction

In previous studies, we demonstrated that various osmolytes can act in concert to improve cell viability during cryopreservation [69,70,85,91,92]. We have shown that certain combinations of sugars, sugar alcohols and amino acids not only maintained post-thaw recovery but also resulted in cells that had improved function compared to those that were cryopreserved with DMSO. Interactions between sugars and sugar alcohols have also been observed by other groups and they have discovered that these molecules interact with each other and water to form deep natural eutectic systems [99].

The objective of this investigation is to use statistical models to analyze how individual osmolytes and interactions between different combinations of osmolytes affect post-thaw recovery. Low-temperature Raman spectroscopy will be used to observe cell response for differences in composition. These studies using Jurkat cells, an immortalized line of T- cells, will help us understand the nature of action and interaction between different molecules and their influence on post-thaw recovery. Understanding these interactions is a key step in developing improved cryopreservation methods that can be used clinically.

5.2 Methods

5.2.1 General Methods

5.2.2 DMSO-free cryoprotectants

In order to characterize the role of specific osmolytes as well as their interaction with other osmolytes, the concentration space of each component (sugar, sugar alcohol and amino acid) was discretized to six levels with equal scale (Table 5.1) based on solubility of the component or levels that result in cell death (toxicity). Three combinations were tested: trehalose-glycerol-creatine (TGC), sucrose-glycerol-creatine (SGC) and sucrose- mannitol-creatine (SMC). For each three-component combination, a total of 216 formulations were tested. The specific composition was described using these levels. For example, SGC123 was the combination of level-one sucrose, level-two glycerol, and level-three creatine. The post-thaw recoveries of all combinations as a function of its composition were determined across all 216 formulations.

5.3 Results

5.3.1 Variation in response with cooling rate

Cell survival can vary strongly with the cooling rate. As a result, the cooling rate selected is a critical parameter. An initial screening study was performed to determine the appropriate cooling rate to be used in the study. The screening was performed by measuring post-thaw recovery for eight formulations at the corners of the parameter space (level 0 and level 5 of a given component) and 10% DMSO as a control. Three different cooling rates (1 C/min, 3 C/min and 10 C /min) were tested. The post-thaw recovery for the different compositions tested was typically highest at 1 C/min when compared to that observed at 3 C/min and 10 C /min. As a result, a cooling rate of 1

C/min was used for all subsequent studies (Figure 5.1).

Table 5.1 Definition of concentration level and corresponding absolute concentration for the tested components

Sucrose Trehalose Glycerol Mannitol Creatine Isoleucine

(mM) (mM) (%) (mM) (mM) (mM) Level 0 0 0 0 0 0 0 Level 1 146 60.66 2 8.66 3.33 8.67 Level 2 292 121.33 4 17.33 6.66 17.33 Level 3 438 182 6 26 10 26.00 Level 4 584 242.66 8 34.66 13.33 34.67 Level 5 730 303.33 10 43.33 16.66 43.33

(a)

(b)

(c)

Figure 5.1 Post-thaw recoveries of Jurkat cells cryopreserved at 1C/min, 3C/min and 10C/min with eight formulations at the corners of the parameter space (level 0 and level 5 of a given component) and 10% DMSO as a control of (a) SGC, (b) TGC, and (c) SMC.

5.3.2 Post-thaw recoveries of SGI, SGC, TGC, SMC

Post-thaw recovery as a function of composition for a cooling rate of 1 C/min was determined for TGC, SGC, and SMC across the parameter space. The mean post-thaw recovery vs concentration of sugar alcohol with varying levels of sugar or amino acid is shown in Figure 5.2. The mean post-thaw recovery vs concentration of sucrose with varying levels of sugar alcohol and amino acid is also shown in Figure 5.3. The black dashed line presented the post-thaw recovery for the single component solution.

For most formulations tested in TGC and SGC, post-thaw recovery for the multicomponent solutions was higher than that observed for a single sugar solution. Data trends suggest that the overall post-thaw recovery was proportional to the concentration of glycerol for both SGC and TGC (Figure 5.2(a) and (c)). Post-thaw recovery increased then decreased as sugar concentration increased for both SGC and TGC (Figure 5.3(a) and (c)). Plots showed post-thaw recovery for each level of sugar strongly depended on glycerol concentration and weakly depended upon creatine levels (Figure 5.3(b) and (d)).

Two saddle nodes, defined as either a local maximum or a local minimum, were observed with variations in sucrose level for a given concentration of glycerol (Figure 5.3(a), (b),

(c), (d)). Formulations of SMC uniformly performed poorly, with recoveries of < 30%

(Figure 5.2(e), (f) and Figure 5.3(e), (f)). There were no clear trends in post-thaw recovery with creatine concentration observed (Figure 5.2(b), (d), (f) and Figure 5.3(b),

(d), (f)).

The non-linear relationship between post-thaw recovery and composition for TGC and

SGC as a function of glycerol level suggests that there are interactions between the different osmolytes. The nature of those interactions will be described in more detail with the statistical model described below. It is noteworthy that some of the multicomponent compositions tested result in a lower post-thaw recovery than freezing in glycerol alone.

This outcome suggests that more is not necessarily better. The formulations associated with the highest post-thaw recovery for TGC, SGC and SMC were 152, 353 and 354 with

83±5%, 79±7% and 27±5%, respectively. The post-thaw recovery of 10% DMSO as control was 85±5%. There is no significant difference (p>0.05) between optimal TGC and SGC, but a significant difference (p<0.01) between TGC and SMC as well as SGC and SMC. The cell concentration and viability after thawing and 24 hours post-thawing for Jurkat cells cryopreserved in both optimal formulation and DMSO control were measured, and both cell concentration and viability were comparable between our formulation and DMSO control.

(a) (b)

(e) (f)

Figure 5.2 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min for varying solution compositions: (a) The effect of glycerol with coloring level of sucrose for SGC. (b) The effect of glycerol with coloring level of creatine for SGC. (c) The effect of glycerol with coloring level of trehalose for TGC. (d) The effect of glycerol with coloring level of creatine for TGC. (e) The effect of mannitol with coloring level of sucrose for SMC. (f) The effect of glycerol with coloring level of creatine for SMC. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. Each color represents the level of sugar with all six levels of creatine. The dashed lines indicate the post-thaw recoveries for the single- component solutions.

(a) (b)

(e) (f)

5.3.3 Statistical modeling

Statistical models have been used previously to characterize the role and interactions between different components [69]. Main effects models were used to describe the additive effects of each osmolyte without interactions, and these models were consistent with experimental data. The main effects model results suggested that both SGC and

TGC exhibited a strong dependence on glycerol concentration (Figure 5.4(a) and (c)), consistent with experimental trends described above (Figure 5.2 and Figure 5.3).

Increasing glycerol by one level was associated with 36% and 34% higher odds of post- thaw recovery to SGC and TGC, respectively. There was little dependence on post-thaw recovery on both mannitol and creatine (Figure 5.4(b), (d), (e) and (f)). Increasing either mannitol or creatine by one level was associated with less than 3% higher odds of post- thaw recovery. These observations are statistically significant with a p-value smaller than

0.001 via F-test.

Potential interactions between osmolytes are also of interest and can be tested using an interaction model. Interaction models indicated that strong positive interactions between sugar and sugar alcohol were observed for sucrose and glycerol (p<0.001) as well as trehalose and glycerol (p<0.001). The interactions varied with the level of sugar and not all interactions improved post-thaw recovery (Figure 5.5(a) and (c)). The statistical models indicated that sugars (sucrose and trehalose) interact with creatine (p<0.05). The variation in log odds of post-thaw recovery at level 3 of sucrose for all creatine levels in

SGC (the circled region in Figure 5.5(b)) reflects this interaction. A similar effect can be seen at level 2 of trehalose (Figure 5.5(d)). It is noteworthy that the interaction between

67 level 3 sucrose and creatine is associated with highest post-thaw recovery for this combination of molecules. Detailed information about both p-values and coefficients are listed in Table 5.2. In contrast to what was observed with SGC and TGC, interactions between sucrose and mannitol act to reduce post-thaw recovery in SMC (Figure 5.5(e) and (f)).

(a) (b)

(e) (f)

(a) (b)

(e) (f)

Table 5.2 Coefficients and p-values of each term in interaction models

TGC SGC SMC (coefficient/p-value) (coefficient/p-value) (coefficient/p-value) Intercept −1.47806 < 2e−16 −1.45156 < 2e−16 −1.88553 < 2e−16 Level 1 sugar 0.241766 0.1142 0.211161 0.183208 −0.01271 0.897683 Level 2 sugar −0.16667 0.2920 −0.01555 0.923132 0.061704 0.533402 Level 3 sugar −0.05169 0.7418 −0.68274 8.84e−05 0.091794 0.330244 Level 4 sugar −0.35108 0.0331 −0.52044 0.002922 0.558270 4.57e−09 Level 5 sugar 0.020637 0.8952 −0.15831 0.331864 0.607310 2.18e−10 Sugar alcohol 0.310126 < 2e−16 0.248203 2.50e−12 0.023966 0.234994 Creatine 0.007020 0.8280 −0.02239 0.509021 −0.02612 0.200104 Level 1 sugar 0.123746 0.0015 0.136974 0.000685 −0.02643 0.294256 *sugar alcohol Level 2 sugar 0.156567 9.9e−05 0.14842 0.000285 −0.10005 0.000117 *sugar alcohol Level 3 sugar 0.025384 0.5131 0.206608 1.36e−06 0.081955 0.000630 *sugar alcohol Level 4 sugar −0.02924 0.4655 0.050701 0.230458 −0.07947 0.000764 *sugar alcohol Level 5 sugar −0.02217 0.5675 0.114932 0.004713 −0.05338 0.022469 *sugar alcohol Level 1 sugar 0.017257 0.6417 0.008537 0.824594 0.012261 0.626075 *creatine Level 2 sugar −0.03477 0.3572 0.002507 0.948533 0.031722 0.213577 *creatine Level 3 sugar −0.01014 0.7874 0.112197 0.005149 0.023216 0.324646 *creatine Level 4 sugar 0.010715 0.7829 0.024607 0.547486 0.034663 0.137355 *creatine Level 5 sugar −0.02271 0.5472 0.042785 0.273012 −0.00669 0.773263 *creatine Sugar alcohol * −0.00527 0.4290 0.001840 0.790682 0.004374 0.268218 creatine

5.3.4 Raman images

The studies described above demonstrated relatively subtle changes in composition can result in distinctly different post-thaw recoveries. Raman spectroscopy was used to understand the role of specific components in stabilizing cells during freezing. For example, Raman microscopy was used to characterize freezing responses of cells cryopreserved in SGC353, which resulted in the highest post-thaw recovery (83%), and cells cryopreserved in SMC353, which exhibited a much lower post-thaw recovery (24%).

Both compositions had the same concentration of sucrose and creatine and the only difference was the sugar alcohol used (glycerol or mannitol). Typical Raman images were rendered based on the Raman signals of ice, amide I, glycerol and mannitol (Figure

5.6(a) and (b)). A distinct difference in the amount of intracellular ice formation (IIF) was observed between the two compositions (Figure 5.6(c)) and AIC for SGC 353 was much lower than that of SMC 353 (p<0.001). Normalized glycerol and mannitol concentration revealed that glycerol was present inside the cell whereas little mannitol was detected inside the cell (Figure 5.6(d) and (e)).

Sugar alcohols can also interact with water and alteration of ice structure by osmolytes has been demonstrated to be influential in cryopreservation outcome [118]. In order to examine the interactions between osmolytes and water during freezing, Raman images of osmolyte solutions without cells cooled down to −50ºC with cooling rate 1ºC/min were obtained for SGI353 [69], SGC353, TGC353, and SMC353 solutions. At this temperature, both ice and the unfrozen liquid were present and distinct differences in the ice morphology were observed with the different solution compositions. The characterization

72 of ice crystal morphology was analyzed in terms of area and ellipticity (Figure 5.7(a), (b),

While both the area and ellipticity of SGC353 and SGI353 were similar, other combinations showed greater differences. With SGC353 and TGC353 (Figure 5.7(b) and

(c)), similar ellipticity was observed (Figure 5.7(e)), but the area of TGC353 was significantly larger (Figure 5.7(f)). Most notably, the area and ellipticity of SMC353 were significantly different (p<0.05) from that of SGI353, SGC353 and TGC353. This outcome and the differing amounts of intracellular ice formation observed with the

SGC353 and SMC353 compositions strongly suggest that glycerol is affecting the freezing behavior of water both inside and outside cells.

Subtle changes in the concentration of one component can also change post-thaw recovery. For example, the post-thaw recovery was 83% and 40% respectively for cells cryopreserved in SGC353 and SGC453 solution. In order to understand the difference,

Raman images of ice and nonfrozen solution of SGC454 were obtained (Figure 5.8(a)) and compared to the images of SGC353 previously obtained (Figure 5.7(b)). The more effective SGC353 solution showed significantly (p<0.05) less ice ellipticity (Figure

5.8(b)) and area (Figure 5.8(c)) than the SGC453 solution. Raman images also showed significantly (p<0.01) higher IIF for Jurkat cells cryopreserved in SGC453 in comparison to SGC353 (Figure 5.8(d) and (e)).

(a) (b) (c)

(d) (e)

Figure 5.6 (a) Raman images rendered of the signal of ice, amide I, and glycerol of Jurkat cells cryopreserved in the SGC353 solution. Regions of light color correspond to areas of high concentration of the signal. (b) Raman images rendered the signal of ice, amide I, and mannitol of cells cryopreserved in the SMC353 solution. (c) AIC of cells cryopreserved in SGC353 and SMC353 solution (n=8, p<0.001). (d) Normalized concentration of glycerol along the white arrow in (a). (e) Normalized concentration of mannitol along the white arrow in (b)

(a) (b) (c) (d)

(e) (f)

(a) (b) (c)

(d) (e)

Figure 5.8 (a) Raman images of ice and non-frozen solution of SGC453 at −50C. (b) The ellipticity of ice crystals formed after freezing of solution compositions of SGC353 and SGC453 from Figure 5.7(b) and (a), respectively (n=10). (c) Area of ice crystal of SGC353 and SGC453 from Figure 5.7(b) and (a), respectively (n=10). (d) Raman images rendered of the signal of ice, amide I, and mannitol of Jurkat cells cryopreserved in SGC453 solution. (e) AIC of cells cryopreserved in SGC353 and SGC453 solution (n=8, p<0.01).

5.4 Discussion and Conclusion

The search for cryoprotectants that can serve as alternatives to DMSO has been ongoing for several decades. The conventional approach has been to look for a single molecule capable of replacing DMSO and conventional wisdom has been that higher levels of cryoprotective agents are better than lower. Higher levels shift the phase diagram for the solution and promote vitrification [123]. This conventional wisdom ignores the strategy developed by nature to use multiple cryoprotective agents to stabilize biological systems against environmental extremes. The outcome of this investigation demonstrates that cryoprotective molecules can influence post-thaw recovery alone and can interact with each other to stabilize cells. The study also demonstrates that this is not a unique property; more than one combination (sucrose and glycerol as well as trehalose and glycerol) of molecules can interact to preserve cells. The long-term goal should be a molecular-level understanding of cryoprotection where specific molecules interact with biological structures (such as the cell membrane) as well as each other and water to stabilize cells.

There is a growing body of literature that demonstrates the cryoprotective behavior of the osmolytes studied in this investigation. The cryoprotective benefits of sugars, especially trehalose and sucrose, have been studied extensively [102,103]. Sugar alcohols such as glycerol have been used to cryopreserve platelets [124] and red blood cells [6].

Sugars

Our models suggest that sugars improve post-thaw recovery and that sugars interact with both sugar alcohols and amino acids to influence post-thaw recovery. There appears to be a threshold level for the stabilizing effects of the sugars: the post-thaw recovery increases with increasing level of sugar and then decreases for increasing level of sugar beyond that threshold level for both SGC and TGC. The outcome is consistent with that observed in our previous work [69].

It is noteworthy that the sugars tested do not readily penetrate the cell membrane. As a result, the molecule acts principally in the extracellular space. This outcome is consistent with previous work in our lab [113]. Ice crystal formation in the extracellular space is also influenced by the presence of sugars. Trehalose and sucrose influenced both the shape and size of ice crystals in the extracellular space. This outcome is consistent with previous studies, which found that sugars suppressed the growth rate of ice crystals due to the strong hydrogen bonding of sugar with water molecules [125–127]. These results are also consistent with the findings of Bailey and colleagues, who found that the addition of trehalose to dimethyl sulfoxide changed the ice crystal patterns observed upon freezing [118].

The stabilizing effects of sugars also reflect the interaction with the cell membrane.

Crowe and colleagues [57] postulated that sugars replaced the water in the cell membrane during freezing. A recent study [113] used low-temperature Raman spectroscopy to image interactions between sugars and the cell membrane. The molecular structure of the

78 sugar influences the stabilizing effects. We have established previously that disaccharides are more effective than monosaccharides [70,128] in improving post-thaw recovery of cells. Disaccharides can reduce the free water due to hydration effect for proteins in the cell when compared to monosaccharides [125,126] suggesting that disaccharides are more effective in stabilizing cell membrane proteins than monosaccharides.

Sugar alcohols

Our models suggest that glycerol plays a major role in cell survival and that glycerol interacts with sugars as well. The influence of glycerol on cell survival has been known for over 60 years [119]. This small molecule penetrates the cell membrane. There is a significant difference in post-thaw recovery for cells frozen in SGC when compared to

SMC, and this may come from not only the absolute concentrations used but also from the intrinsic molecular difference between glycerol and mannitol. The Raman images obtained suggest that the higher post-thaw recovery for cells cryopreserved in SGC results from the ability of the molecule to penetrate the cell membrane (Figure 5.6). The importance of penetrating cryoprotectants on post-thaw recovery has long been known

[49], but a recent study has found that non-penetrating cryoprotectants can also provide protection [114]. In addition, cell types may respond differently to the same formulation.

For example, SMC showed a moderate post-thaw recovery to mesenchymal stem cells but a low post-thaw recovery in this work [70]. Glycerol has long been associated with the stabilization of proteins [120]. The ability of glycerol to penetrate the cell may result in enhanced stability of intracellular proteins, which in turn resulted in improved post- thaw recovery.

As with sugars, the results in this study suggest that sugar alcohols act on water molecules. Raman images of ice formed in the presence of the solutions of interest demonstrated differences in shape and size of ice crystal with different compositions.

This outcome is consistent with previous studies that have shown that the hydrogen bonding between glycerol and water plays a significant role to inhibit ice crystallization and the structure of ice crystals formed during freezing [84,94,121,122]. Taken as a whole, these studies demonstrate that molecular level interactions result in macroscale changes in ice growth.

Amino acids

Our models suggest that amino acids interact with sugars to alter post-thaw recovery. For example, SGC had more formulations with higher post-thaw recovery than TGC under the same combination of glycerol and creatine, and our interaction models suggest the interaction of creatine to sucrose is stronger than it to trehalose. Interactions between sugars and amino acids are consistent to the work by Wen et al. that specific proteins stabilize trehalose during freezing and inhabit ice formation [97,129], that amino acids strengthen the local interactions between water molecules.

Interactions between osmolytes

Our model suggests that sugar alcohols interact with sugars for the range of molecules and concentrations tested. Castro and colleagues observed a similar outcome when they determined that trehalose and glycerol formed a natural deep eutectic system [99] when combined with water. Specifically, the combination of molecules exhibited a stronger

80 than expected influence on the eutectic temperature of the solution. This outcome suggests that one potential explanation for the interaction of sugar and sugar alcohol could be the formation of a deep eutectic system that influences the solidification of water. Studies have observed that the stability of a protein is strongly influenced by the interactions between sugar and sugar alcohol [130,131]. It is noteworthy that combinations of osmolytes have also been used to stabilize proteins, suggesting another potential method of action for combinations of osmolytes. Finally, recent studies began to clearly demonstrate the beneficial effect of combinations of osmolytes on post-thaw recovery [69,70].

Part 2: localizing the optimal formulation in a multicomponent cryoprotectant using differential evolution algorithm

Hypothesis: Different operating parameters influence the ability of the differential evolution to optimize solution composition.

Much of the text and figures in this chapter will be published soon as the article below

Pi C-H, Dosa PI, Hubel A. Differential evolution for the optimization of DMSO-free cryoprotectants: influence of control parameters. J Biomechanical Engineering (under review)

Chapter 6: Differential evolution for the optimization of

DMSO-free cryoprotectants: influence of control parameters

6.1 Introduction

There is no universally optimal set of cryopreservation conditions, and thus conditions need to be reoptimized for each new cell type. A variety of factors influence the post- thaw recovery of a cell after cryopreservation: the composition of the freezing medium, the method of introduction and removal of the medium, the freezing and thawing rate and storage conditions [12]. Protocols are typically developed using empirical approaches: the composition, cooling rate or other parameters are varied and the resulting post-thaw recovery is measured. This approach is costly and time-consuming [132,133].

An alternative to this inefficient empirical approach is to use an optimization scheme designed to significantly reduce the number of experiments necessary to arrive at the optimized cryopreservation protocol. Among optimization methods, differential evolution

(DE) algorithm is a simple and powerful technique for multi-dimensional and global optimization [79]. Advantages of DE algorithm includes (1) simple and straightforward implementation in comparison to other optimization algorithms; (2) a small numbers of control parameters including mutation (F), crossover (Cr) and population (NP) in classical DE; and (3) low space complexity of DE is low as compared to some of the most competitive real parameter optimizers [134]. In addition, DE is also applicable to optimize discontinuous space [135,136], which is helpful in extending DE for handling large-scale and expensive optimization problems. Due to these advantages, the DE algorithm and its variants have been applied in many fields such as bioinformatics,

83 chemical engineering, molecular configuration, and urban energy management [134,137].

Recently, DE has been adopted in optimizing culture media formulations [138]. We have developed and applied DE to the development of DMSO-free cryopreservation protocols in our lab [91] and we demonstrated the ability to optimize both solution composition and cooling rate using a relatively small number of experiments [92].

The robust application of the DE algorithm requires the selection of the proper control parameters for the algorithm. Improper selection of control parameters may reduce the optimization efficiency or even result in stagnation. Reducing the number of experiments required to optimize the preservation of a given biological system is critical for applications in which the material is rare or hard to acquire (gametes from endangered species) or expensive (cell therapies are valued at hundreds of thousands of dollars). In this work, we use data sets (post-thaw recovery for a range of solution compositions spanning the parameter space) to rationally select the control parameters settings for the

DE algorithm to arrive at optimal cryopreservation conditions most efficiently. We then validate the performance of the optimized DE algorithm by optimizing other DMSO-free cryoprotectant conditions using other data sets. The knowledge obtained from this study is critical to expanding the use of this approach in future studies designed to optimize cryopreservation protocols to maximize high post-thaw recovery.

6.2 Methods

6.2.1 General Methods

Post-thaw recovery as a function of solution composition was obtained from previous studies [68,69]. Jurkat cells (ATCC TIB-152) were cryopreserved in candidate cryoprotectant solutions in 96-well plates using a cooling rate of 1C/min and later thawed in a 37 C water bath as described in Section 3.2 and 3.4. Calcein-AM/PI was used to measure the numbers of live and dead cells in each well post-thaw as described in

Section 3.5.

6.2.2 Differential Evolution Algorithm

The Differential Evolution algorithm and its variants were coded with Python programming language version 3.4.0 (Python Software Foundation, https://www.python.org/) in this work. Each cryoprotectant formulation was taken as a vector in the software. Concentration level was used instead of absolute concentration for convenience as with previous publications [68,69]. The code searched the corresponding post-thaw recovery automatically and suggested the composition of cryoprotectant to test in the following iteration.

Four experimental data sets were used to find optimal control parameters and most effective type base for the two indices to evaluate the performance of each type: (1) accuracy, which is the ratio of the post-thaw recovery of converged solution and optimal post-thaw recovery, and (2) convergence speed, which is defined as the generations to convergence without a change of accuracy. Convergence can be measured by observing

85 an increase in a cumulative best formulation, a decrease in the number of improved solutions within the emergent population after each generation, or by the average of each generation. For the purpose of this investigation, the convergence was defined when the same best member was observed in three consecutive generations and decreasing number of improved solutions is also observed.

The accuracy was defined as Equation (6.1)

푙표푐푎푙푖푧푒푑 표푝푡푖푚푎푙 푟푒푐표푣푒푟푦 A푐푐푢푟푎푐푦 = × 100% (6.1) 푟푒푎푙 표푝푡푖푚푎푙 푟푒푐표푣푒푟푦

Each experiment was repeated for 50 runs to get statically meaningful data.

6.2.3 Python GUI

The graphical user interface (GUI) includes the parameter inputs, automatic saving/loading, and data display. On the first page, users can control parameters such as the population size, the dimensions, and file names to either save or load data. On the second page, one needs to select three different scenarios, (1) start from generation zero,

(2) have a generation and need a mutant, or (3) have both a generation and a mutant and need to perform the selection. In the third page, the DE algorithm will perform either formulation generation, mutation, or selection based on your choice above. The Python

GUI can automatically save/load data to a text file. Once the convergence criteria fulfilled, the notification will be displayed.

6.3 Results

6.3.1 Post-thaw recovery for multi-component solutions

Previous studies have established that multicomponent osmolyte solutions are effective in cryopreserving cells [68–70,92]. The post-thaw recovery of Jurkat cells cryopreserved with two combinations of osmolytes, sucrose-glycerol-creatine (SGC) and sucrose- glycerol-isoleucine (SGI), were screened over the entire spaces [68,69] as shown in

Figure 6.1 and Figure 6.2, respectively. The concentration space of each component was discretized to six levels with equal scale, with the highest level being determined by either toxicity limits or solubility limits. A total of 216 formulations were tested across the parameter space for one DMSO-free cryoprotectant. As described previously [68,69], the SGC optimal formulation composition was 438 mM sucrose, 10% glycerol and 10 mM creation with 80% post-thaw recovery (Figure 6.1(d)). The SGI optimal formulation composition was 146 mM sucrose, 10% glycerol and 43 mM isoleucine with 84% post- thaw recovery (Figure 6.2(b)).

These studies determined that there is a narrow range of optimum composition for the three components (sugar, sugar alcohol, and amino acid) associated with maximum post- thaw recovery. This type of behavior provides an excellent opportunity to test the DE algorithm and tune the selection of the mutation (F), crossover (Cr) and population size

(NP) for DE optimization of cryopreservation protocols.

The 3D plots of post-thaw recovery for different combinations of osmolytes were presented as a function of composition and shown in Figure 6.1 and Figure 6.2. The

87 topology of SGI was complex and non-linear with a sharp peak around the optimal formulation. The topology of SGC was smoother than SGI with broader regions of high post-thaw recovery. Statistical analysis of the data suggests that the overall post-thaw recovery was proportional to the concentration of glycerol for both SGC and SGI [68].

Post-thaw recovery increased then decreased as sugar concentration increased for both

SGC and SGI. These experimental data were utilized to test different types of DE algorithm and to adjust the control parameters in order to achieve optimal accuracy and convergence speed.

(a) (b)

(e) (f)

Figure 6.1 Post-thaw recoveries of Jurkat cells cryopreserved in varying concentrations of glycerol and creatine for a given concentration of sucrose (SGC) at a cooling rate of 1ºC/min for sucrose concentration (a) 0 mM, (b) 146 mM, (c) 292 mM, (d) 438 mM, (e) 584 mM and (f) 730 mM.

(a) (b)

(e) (f)

Figure 6.2 Post-thaw recoveries of Jurkat cells cryopreserved in varying concentrations of glycerol and isoleucine for a given concentration of sucrose (SGI) at a cooling rate of 1ºC/min for sucrose concentration (a) 0 mM, (b) 146 mM, (c) 292 mM, (d) 438 mM, (e) 584 mM and (f) 730 mM.

6.3.2 Mutation and Crossover

The next phase of the investigation involved using the experimental data sets described above to determine the influence of mutation factor (F) and crossover (Cr) to different

DE types. Four different DE types including random, best, local-to-best, local-to-best with self-adaption (SA) were examined as well in order to investigate the influence of mutation strategies.

The accuracy maps of all combinations of mutation and crossover with 0.1 increment under the same population size for all experimental data are shown in Figure 6.3 and

Figure 6.4. The threshold of accuracy was set to 95% in order to filter combinations of mutation and crossover as grey area. These figures indicated that SA was tolerant of combinations of mutation and crossover for SGC but was sensitive for SGI.

DE/best/1/bin showed the smallest acceptable range of mutation and crossover for SGC.

The DE algorithm operated more efficiently for larger values of crossover but accepted a narrow range of mutation factors. For all DE types, the selection of mutation factor and crossover was critical for SGI. It is noteworthy that DE/rand/1/bin, the classical DE, was tolerant of mutation factor and crossover for SGI in comparison to other DE types (Figure

6.4).

(a) (b)

6.3.3 Population size

The accuracy was proportional to the population size for all experimental data sets analyzed as shown in Figure 6.5. The accuracy for the SGC data set was proportional to population size and was below 95% for NP<17 for F=0.5 and Cr=0.9 as shown in Figure

6.5(a) but the accuracies were all above 95% for all NP and no significant differences while NP was larger than 13 for F=0.9 and Cr=0.5 as shown in Figure 6.5(b).

For the SGI data set, the accuracies were proportional to population size and all below

90% independent of population size using F=0.5 and Cr=0.9 as shown in Figure 6.5(c).

The accuracy achieved over 95% while F=0.9, Cr=0.5 and NP>17 as shown in Figure

6.5(d). The searching converged after 5 generations for F=0.5 and Cr=0.9, which exhibited stagnation as shown in Figure 6.5(a) and (c). In other words, the searching process became stuck at the local optimum. Adjusting the mutation and crossover to

F=0.9 and Cr=0.5 resulted in improved accuracy for both SGI and SGC data sets for the same population size. The accuracy map for four DE types for both SGC and SGI data sets showed that the acceptable combinations of mutation factor and crossover were proportional to the population. When applied to data sets with a smooth topology, F>0.7 and the full range of Cr tested achieved >95% accuracy. For data sets with a sharp topology such as SGI, F>0.7 and Cr>0.3 to DE/rand/1/bin and F>0.7 and Cr > 0.3 to others achieved >95%.

(a) (b)

6.3.4 Validation

Based on the studies described above, the best control parameters are F>0.7 and Cr>0.3 in order to optimize the compositions of DMSO-free cryoprotectants based on multiple osmolytes. The next phase of the investigation involved evaluating the universality and consistency of these control parameters using two new DMSO-free cryoprotectant formulations. Post-thaw recovery for two candidate solutions (trehalose-glycerol-creatine

(TGC) and trehalose-glycerol-isoleucine (TGI) were optimized using the DE algorithm and the control parameters described above. The outcomes of this optimization of TGC and TGI are shown in Figure 6.6 and Figure 6.7, respectively. The solution composition associated with the highest post-thaw recovery emerged in a relatively small number of generations. The optimum solution formulation associated with the optimal post-thaw recovery and converged after generation 6 (seven freezing experiments). Figure 6.6(a) and (d) as well as Figure 6.7(a) and (d) demonstrated that the optimal cryopreservation solution composition occurred at either generation 2 or 3 and persisted until the convergence criteria were fulfilled. Figure 6.6(b) and (e) as well as Figure 6.7(b) and (e) showed recovery associated with the best member solution increases and plateaus as the algorithm converged. Figure 6.6(c) and (f) as well as Figure 6.7(c) and (f) presented the number of improved solutions in each generation, which decreased as the algorithm converged. The population size influenced the rate at which the number of improved solutions declined. For NP=18, the number of improved solutions per generation rapidly declined for generations > 1. For NP=9, the number of improved solutions declined after generations > 3. The TGC optimal formulation composition of 61 mM trehalose, 10% glycerol and 7 mM creatine with 83% post-thaw recovery. The TGI optimal formulation

96 composition of 61 mM trehalose, 10% glycerol and 43 mM isoleucine with 84% post- thaw recovery. Experiments spanning the entire parameter space confirmed that the best member identified in the algorithm represented the overall global maximum. It is noteworthy that the DE algorithm determined the optimum using 60 formulations and spanning the parameter space requires at least 216 formulations.

(a) (b)

(e) (f)

(a) (b)

(e) (f)

6.4 Discussion

Protocols for cryopreservation have typically been determined empirically by exhaustively varying composition and cooling rates. We have previously developed

DMSO-free conditions for cryopreserving cells that used combinations of osmolytes including sugars, sugar alcohols and amino acids [68–70]. However, the performance (i.e. post-thaw recovery) of these conditions highly depended on the composition of the osmolytes and had to be redetermined for each new cell type. Finding the optimal formulation with traditional trial-and-error methodologies proved to be time-consuming and costly. We previously applied a differential evolution algorithm to help optimize the composition of cryoprotectants [91], but three control parameters including mutation (F) and crossover (Cr) and population size (NP) may significantly influence the optimization performance of DE. In this study, we examine the manner by which varying the control parameters affect the performance of the DE algorithm.

One approach for testing a new DE variant usually involved measuring the accuracy and convergence speed of the algorithm with several mathematic test functions [139,140].

This approach allows testing of large population size in order to select mutation and crossover. In contrast, this study used actual data sets for tuning the crossover and mutation.

Topology of data

For experimental data sets with smooth, broad peaks, a larger range of acceptable combinations of control parameters can be used. For example, in this situation the

100

DE/local-to-best/1/bin with self-adaptive can achieve 95% accuracy with most initial combination of mutation and crossover, F>0.1 and Cr>0.2, with larger population size as shown in Figure 6.3(d), and other DE strategies presented similar acceptable combinations, F > 0.7 and Cr > 0.3 as shown in Figure 6.3(a)-(c). However, when presented with a sharp maximum as seen in the SGI data set, the number of acceptable combinations of mutation and crossover were for F>0.7 and Cr>0.3 for DE/rand/1/bin only (Figure 6.4(a)), which is significantly lower. Other strategies cannot search the global optimum with 95% accuracy as shown in Figure 6.4(b)-(d). Operating outside of that range of mutation and crossover values would result in the algorithm getting stuck in local optimum or passing over the global optimum during mutation, crossover or even roundoff truncation.

Mutation, crossover and population

In comparison to suggested control parameters for bioprocess, biomedical and bioinformatics [79,134,141], using the DE algorithm to optimize cryoprotectants requires a narrow range of mutation factor and a wide range of crossover. Specifically, the accuracy of all DE types is greater for mutation factors ranging from 0.7 to 1.0 and crossovers ranging from 0.3 to 1.0. For SGI, three DE strategies presented undesired performance of below 95% as shown in Figure 6.4(b)-(d) but showed acceptable combinations for F>0.7 and Cr > 0.3 as shown in Figure 6.4(a). Increasing the population size increased the acceptable combinations of mutation and crossover. For example,

F=0.5, Cr=0.9 and NP=9 resulted in low accuracy as shown in Figure 6.5(a) and (c), but

NP=27 improved accuracy 10% and 20% to SGC and SGI, respectively. The

101 improvement through increasing population size was distinct especially to the data with a sharp peak or discontinuity as shown in Figure 6.5(c) and (d).

Selecting population size is a trade-off between the convergence speed and cost. In addition, when biological experiments need to be performed, practical considerations also need to be considered. These include the availability of the system is preserved, duration and budget and the availability and capacity of labor and instrumentation. For example, in our experiments, NP=18 was at the limit of testable formulations when factors such as the capacity of the controlled rate freezer and the time necessary for liquid handling and sample preparations were considered.

Future work

The current approach to optimizing cryoprotectant composition assumes a single output

(viability) for a homogeneous population. Preservation of a heterogeneous cell population may be desired and future work could extend the application of DE to heterogeneous cell populations using multi-objective differential evolution (MODE) [142–145].

Alternatively, it may also be desirable to optimize several post-thaw metrics for a single cell type (e.g. viability, surface markers and functionality).

102

6.5 Conclusion

This work investigated the effects of control parameter selection for use in the differential evolution algorithm for the purpose of optimizing compositions of DMSO-free cryoprotectants. The accuracy of the classical DE algorithm depended on the combinations of mutation factor and crossover. Screening results showed typical mutation factor and crossover should be between [0.7, 1] and [0.3, 1], respectively. The self-adaption modification reduced the effects of control parameter selections but classical DE with classical mutation method (DE/rand/1/bin) presented better accuracy to the discontinuous space. The topology of experimental data was also another critical issue in the optimization process. If the global optimum was a sharp peak, the DE algorithm might pass over that during mutation, crossover or even numerical truncation and eventually get stuck in the local optimum. This work will help with more efficiently determining the optimal concentrations of multicomponent cryopreservation solutions using a differential evolution algorithm approach.

103

Part 3: Apply multicomponent DMSO-free cryoprotectants to clinically-relevant cell therapy products

Hypothesis: DMSO-free cryoprotectants can maintain immune cell viability and function during cryopreservation

Most figures and texts will be published in the article below:

Pi C-H, Hornberger K, Dosa P I, Hubel A. Understanding the freezing responses of different subsets of human peripheral blood mononuclear cells using DSMO-free cryoprotectants with combinations of osmolytes. Cytotherapy (preparation)

104

Chapter 7: understanding and optimizing freezing responses to

PBMC cryopreserved with DMSO-free cryoprotectants

7.1 Introduction

Previous studies of the cryopreservation of PBMCs [14,73,146–150] treated PBMCs as a homogeneous population. However, PBMCs are a heterogeneous population comprised of multiple cell types, each with different membrane features and cell sizes [151]. For example, monocytes are approximately 15 to 25 µm in diameter, but small lymphocytes are 7 to 10 µm in diameter and large lymphocytes are approximately 14 to 20 µm in diameter. The lymphocytes have a more regular cytoplasmic border without the cytoplasmic blebbing and pseudopods that are present in monocytes. The monocyte has a grainy, gritty texture that is absent in the lymphocyte, and monocytes can develop into either dendritic cells or macrophages. Neglecting the differential responses of the different cell types that make up PBMCs can lead to suboptimal outcomes as certain cell classes may be greatly damaged during cryopreservation even though overall PBMC survival may be strong.

In our previous work [68–70,85,91,113], we optimized DMSO-free cryoprotectants for

Jurkat cells using mixtures of sugars, sugar alcohols and amino acids. These natural osmolytes act by stabilizing biological systems subjected to environmental extremes [15].

Using the Jurkat immortalized lymphocyte cell line allowed for high-throughput screening of cryoprotectant solutions. Additionally, the preservation mechanisms and interactions between osmolytes were previously characterized using Raman spectroscopy and statistical modeling [68,69]. Raman images showed the influence of osmolytes and combinations of osmolytes on ice crystal shape, which reflected the interactions between

105 osmolytes and water. Both higher concentrations of glycerol and interactions between sugars and glycerol were found to typically increase the post‐thaw recovery. Jurkat cells were used for screening osmolyte solution compositions [68,69] and the optimal compositions for each formulation were then used in this study to cryopreserve PBMCs.

The post-thaw recoveries of Jurkat cells cryopreserved with optimal formulations of SGI,

TGI and MGI were 84%, 84% and 85%, respectively [68,69].

In this work, we conduct a comparative study to evaluate the effects of three DMSO-free cryoprotectants on human PBMC subsets. The evaluations include the post-thaw recovery of subtypes of lymphocytes including B cell (CD19+), natural killer cell (CD3-

CD56+), helper T-cell (CD3+CD4+) and cytotoxic T-cell (CD3+CD8+). This work provides critical insight into the potential of DMSO-free cryoprotectants in terms of preserving a heterogeneous population and analyzing the freezing response of each subpopulation to osmolyte-based cryoprotectants.

106

7.2 Methods

7.2.1 General methods

Cell culture of Jurkat cells and collections of PBMC were described in Section 3.2 and

3.3, respectively. Plate freezing for high-throughput screening was performed as described in Section 3.4.1. Vial freezing for understanding the freezing response of

PBMC subsets was described in Section 3.4.2. Thawing and measurement of post-thaw recovery were mentioned according to Section 3.4.3 and 3.5, respectively. Statistical analysis was according to section 3.8.

7.2.2 Differential scanning calorimetry (DSC)

Differential scanning calorimetry (DSC) has been widely applied in cryopreservation to measure thermodynamic properties [152–154]. DSC was performed on a TA Differential

Scanning Calorimeter Q1000. Cryoprotectants without cells were loaded into T zero pans and hermetically sealed. Samples were frozen to −150ºC using the following protocol: (1) start at 20ºC, (2) cool to −150ºC at 10ºC/min, (3) hold for 3 min at −150ºC, and (4) warm to 20ºC at 10ºC/min. DSC results were analyzed using TA Universal Analysis software.

7.2.3 Characterize phenotype and viability using flow cytometry

Freshly isolated and thawed PBMC were used for flow cytometry to analyze the proportion of each population and viability of each population. For each PBMC sample, phenotype staining was done before freezing (fresh) and immediately after thawing in order to calculate the post-thaw recoveries. Nine panels of antibodies were used to determine the proportions of PBMC subsets and their viabilities. The information for

107 fluorophores, antibodies and manufacturers is shown in Table 7.1. The gating strategy of all PBMC subsets was determined by first identifying all white blood cells (CD45+).

From this plot, granulocytes (CD15+), monocytes (CD14+), and lymphocytes (CD14-

CD15-) were identified. From the lymphocyte plot, B cells (CD19+), NK cells (CD3-

CD56+), NKT cells (CD3+CD56+), and T-cells (CD3+CD56-) were determined. Finally, from the T-cell subset, helper T-cells (CD3+CD4+) and cytotoxic T-cells (CD3+CD8+) were identified. The full gating strategy is shown in Figure 7.1. The data was collected with a BD LSR II (BD Bioscience, San Jose, CA) flow cytometry with FACSDiva software (version 8.0.1), and the data analysis was performed using FlowJo 9.8.5 software (Tree Star, Ashland, OR) according to established protocols.

Table 7.1 Information on antibodies and fluorophores of phenotype characterization

Fluorophore Antibody Manufacturer V500 CD45 BD Horizon PE-TR CD3 Invitrogen PerCP-Cy5.5 CD4 Biolegend Pacific blue CD8 Invitrogen BV605 CD56 Biolegend PE CD14 BD Pharmingen APC CD15 BD Pharmingen FITC CD19 BD Biosciences APC-ef870 Viability Invitrogen

The staining protocol is briefly described as following. Antibody cocktails were added to

PBMC samples and incubated for 30 mins at 4ºC under a low light environment. BD

FACS Lysing 10x solution (BD Biosciences, San Jose, CA) was added and cells were

108 incubated for 15 mins at room temperature. Cells were centrifuged for 5 min at 500×g, the supernatant was removed, and a staining buffer was added (BD Pharmingen, San Jose,

CA). Samples were recentrifuged for 5 min at 500×g, supernatant was removed and fixation buffer was added. Samples were stored at 4ºC under low light environment until acquisition.

Figure 7.1 Gating strategy to identify proportions of PBMC subsets through flow cytometry. A total of 30000 singlet events was collected (top left plot). First, white blood cell (CD45+) was identified from singlets. Granulocytes (CD15+), monocytes (CD14+) and lymphocytes (CD14-CD15-) were identified from white blood cells (CD45+). B cells (CD19+), natural killer cells (CD3-CD56+), natural killer T-cells (CD3+CD56+) and T- cells (CD3+CD56-) were identified from lymphocytes. Helper T-cells (CD3+CD4+) and cytotoxic T-cells (CD3+CD8+) were identified from T-cells.

109

7.3 Results

7.3.1 Comparison of post-thaw recovery of Jurkat, PBMC and Specific

subpopulations to osmolyte solutions

The first phase of the investigation involved comparing the freezing response of PBMCs to that of Jurkat cells, a cell line intended to model the freezing behavior of T-cells.

PBMCs or Jurkat cells were cryopreserved in 96-well plates and the post-thaw recovery under different formulations of SGI and TGI was determined in order to compare their freezing responses as shown in Figure 7.2(a) and (b). Test formulations were generated spanning the parameter space and included the optimum formulation. The post-thaw recovery for PBMCs and Jurkat cells were comparable for optimized solution formulations; however, post-thaw viability of Jurkat cells decreased more rapidly for suboptimal formulations when compared to PBMCs.

PBMCs products used in this investigation contain lymphocytes (CD14-CD15-), a major subset (84%), followed by monocytes (CD14+, 14%) and granulocytes (CD15+, 2%). Of the lymphocyte population, Helper T (CD3+CD4+, 41%) was the major subset, followed by B (CD19+, 16%) Cytotoxic T (CD3+CD8+, 14%) and NK (CD3-CD56+10%) and

NK T (CD3+CD56+, 3%) as shown in Figure 7.3(a). The standard error of each population was less than 5%. The comparison of post-thaw recoveries for subpopulations of PBMC (CD45+) and Jurkat cells for three optimal formulations of DMSO-free cryoprotectants (SGI, TGI, MGI) is shown in Figure 7.3(b). The post-thaw recovery of

CD45+ cells was equivalent to Jurkat cells (p>0.05) and there was no significant difference between DMSO-free and DMSO-containing (p>0.05). However, NK (CD3-

110

CD56+) and cytotoxic T-cells (CD8+) frozen in DMSO had higher post-thaw recovery than cells frozen in DMSO-free cryoprotectants.

(a) (b)

111

(a)

(b)

112

7.3.2 Comparisons between Jurkat cells and T-cell subpopulations

Subpopulations of T-cells are used therapeutically and as a result, a comparison of post- thaw recovery for Jurkat cells and T-cell subpopulations is of interest. The post-thaw recovery of primary T-cells (CD3+ cells) was higher than Jurkat cells for the DMSO-free cryoprotectants. DMSO-free solutions had higher post-thaw recoveries compared to 10%

DMSO for lymphocytes and their subpopulations including B cell (CD19+), T-cell

(CD3+CD56-) and helper T-cell (CD3+CD4+). DMSO-free cryoprotectants also showed higher post-thaw recoveries of human helper T-cells (CD3+CD4+) compared to Jurkat cells (p<0.05), but there was no significant difference between human T-cell and Jurkat cells frozen in the DMSO-containing cryoprotectant (p>0.05) as shown in Figure 7.4.

DMSO-free cryoprotectants showed lower post-thaw recoveries for cytotoxic T-cells

(CD3+CD8+) compared to Jurkat cells, but the DMSO-containing solution showed equivalent (p>0.05) post-thaw recoveries between human cytotoxic T-cells (CD3+CD8+) and Jurkat cells as shown in Figure 7.4. T-cells and Jurkat cells had the same post-thaw recovery when preserved in conventional DMSO-containing solution.

There was no significant difference between all three DMSO-free cryoprotectants for primary human T-cells. For human helper T-cells, there was no significant difference between the three DMSO-free cryoprotectants, but there was a significant difference between TGI and DMSO (p=0.0225) as wells as SGI and DMSO (p=0.0418). There was no significant difference between MGI and DMSO. For human cytotoxic T-cells, there was no significant difference between all four cryoprotectants. In other words, all cryoprotectants were equivalent in terms of CD8+ post-thaw recoveries. In comparison

113 between post-thaw recoveries of CD4+ and CD8+, there were significant differences for

TGI (p=0.0012), SGI (p=0.0005) and MGI (p=0.0039) and no significant difference for

DMSO. The differences in post-thaw recoveries between helper T-cell (CD3+CD4+) and cytotoxic T-cells (CD3+CD8+) in DMSO-free cryoprotectants indicated the necessity to optimize the formulations in order to improve post-thaw recoveries of both groups as high as possible and as equivalent as possible.

Figure 7.4 Post-thaw recoveries of Jurkat cells, T-cells, Helper T-cells, and Cytotoxic T- cells cryopreserved in TGI, SGI and MGI cryoprotectants as well as 10% DMSO (n=10).

114

7.3.3 Optimization of DMSO-free cryoprotectant to cryopreserve dual subsets of

T-cells

TGI was selected to be optimized with the differential evolution algorithm due to higher post-thaw recoveries in the average of T-cell (CD3+), helper T-cell (CD3+CD4+) and cytotoxic T-cell (CD3+CD8+) among all three DMSO-free cryoprotectants (Figure 7.4).

The variant of the DE algorithm, differential evolution of multi-objective (DEMO), was used to identify the optimal formulations with the highest post-thaw recoveries of helper

T-cell (CD3+CD4+) and cytotoxic T-cell (CD3+CD8+) generation by generation as shown in Figure 7.5(a) and (b), respectively. The generation 0, 1 and 2 showed diversified variations of post-thaw recoveries of both CD4+ and CD8+, but by generation

3 variability had been reduced to below 20%. The evolutionary optimization showed the improvement of two subsets simultaneously as shown in Figure 7.5(c) from divergence to convergence as well as from low to high post-thaw recovery. The post-thaw recoveries of helper T-cells were generally proportional to but higher than cytotoxic T-cell recovery.

The maximum post-thaw recoveries of helper and cytotoxic were 103% and 73%, respectively.

The optimization of two subsets of CD3+ T-cells might deoptimize another subset of

CD3+ T-cells. For example, two TGI formulations exhibited over 80% post-thaw recoveries of NK T-cell (CD3+CD56+) in generation 1, but these two formulations were excluded in generation 3 because they resulted in relatively lower post-thaw recoveries of helper T-cells and cytotoxic T-cells as shown in Figure 7.5(d). In addition, the average post-thaw recovery of generation 1 was greater in generation 1 (56%) than generation 2

(33%) but had rebounded by generation 3 (58%).

115

(a) (b)

116

7.3.4 Statistical modeling of freezing response between PBMC subsets

Statistical models can be used to characterize the role and interactions between different osmolytes in the TGI cryoprotectant across different PBMC subsets [68,69]. The statistical models indicated the difference between helper T-cell (CD3+CD4+) and cytotoxic T-cell (CD3+CD8+), even though both are T-cells (CD3+) as shown in Table 2.

For helper T-cells, the level 1 trehalose and its interaction with glycerol and isoleucine dominated the post-thaw recovery and high-level trehalose resulted in negative effects.

This trend was similar to Jurkat cells as seen in our previous work [68,69]. For cytotoxic

T-cells, however, a specific level of trehalose was preferred, and there were weak interactions between trehalose and glycerol and trehalose and isoleucine.

The corresponding statistical model for NK T-cells was also established as shown in

Table 2. The statistical model suggested low-level trehalose was dominant but the interaction of low-level trehalose and glycerol was not beneficial, which was distinct to both helper T-cell and cytotoxic T-cell. The prediction versus truth plots showed all predicted points had over 80% accuracy for helper T-cells and cytotoxic T-cells as shown in Figure 7.6(a) and (b), respectively. The statistical model underestimated several individuals in the prediction of the post-thaw recovery of NK T-cells as shown in Figure

7.6(c).

117

Table 7.2 The coefficients of statistical models to helper T-cell, cytotoxic and NK T-cell

Cell type Helper T-cell Cytotoxic T-cell NK T-cell (CD3+CD4+) (CD3+CD8+) (CD3+CD56+) Intercept 0.02619 −1.06837 −0.27805 Level 1 trehalose 4.04120 0.80401 2.74033 Level 2 trehalose −1.48409 −0.39872 0.61609 Level 3 trehalose −2.02170 −1.27156 −1.62816 Level 4 trehalose −0.06433 1.42700 −0.73732 Level 5 trehalose −1.64595 −0.57918 0.85361 Glycerol 0.44619 0.30054 −0.02072 Isoleucine 0.04204 0.02716 −0.19452 Level 1 trehalose *glycerol −2.46496 −0.24038 −0.79640 Level 2 trehalose *glycerol 0.47758 0.24386 −0.30011 Level 3 trehalose *glycerol 0.71940 0.78952 0.68677 Level 4 trehalose *glycerol 0.63765 −0.16140 0.12593 Level 5 trehalose *glycerol 0.64156 0.11695 −0.64214 Level 1 trehalose *isoleucine 1.82606 0.06212 −0.04865 Level 2 trehalose *isoleucine 0.35771 −0.08998 −0.10984 Level 3 trehalose *isoleucine −0.14175 −0.63488 −0.88787 Level 4 trehalose *isoleucine NA NA NA Level 5 trehalose *isoleucine NA NA NA Glycerol * isoleucine −0.06422 −0.01725 0.12257

118

(a) (b)

(c)

119

7.3.5 Differential Scanning Calorimetry (DSC)

The thermophysical properties of both DMSO-free and DMSO-containing cryoprotectant solutions including melting temperature, enthalpy of melting, glass transition temperature

(Tg) and softening temperature (Ts) were measured via DSC as listed in Table 3. TGI was found to have the lowest melting temperature and lowest enthalpy of melting. In contrast,

DMSO had the highest melting temperature as well as enthalpy of melting, and the lowest glass transition temperature. There was no significant difference between DMSO- free cryoprotectants in terms of Tg and Ts.

Table 7.3 The thermophysical properties of both DMSO and DMSO-free cryoprotectants

Melting Temperature (C) Enthalpy of Melting (J/g) Tg (C) Ts (C) DMSO −9.63 195.7 −120.77 N/A SGI −11.58 191.5 −93.90 −65.08 TGI −13.35 132.6 −94.55 −64.90 MGI −12.43 184.9 −93.76 −65.94 LGI −11.99 190.4 −95.42 −67.10

120

7.4 Discussion and Conclusion

There has been great interest in finding nontoxic alternatives to DMSO that can cryopreserve cells effectively. Previous work has shown the potential of combining osmolytes including sugar, glycerol and amino acids for the cryopreservation of Jurkat cells [68,69]. This work establishes this approach for the preservation of primary human

PBMCs as well as specific subsets contained therein.

Various research groups attempted to analyze the freezing responses of PBMC subsets with different DMSO-containing cryoprotectants [155–159]. This work first cryopreserved human primary PBMCs with DMSO-free cryoprotectants and analyzed the freezing responses of PBMC subsets. DMSO-free cryoprotectants showed competitive post-thaw recovery of PBMCs (CD45+) compared to DMSO-containing solutions, and higher preservation capability on several specific subpopulations including B cells, lymphocytes and subsets of lymphocytes. Cytotoxic T-cells showed lower post-thaw recovery for DMSO-free solutions compared to helper T-cells. DMSO-free cryoprotectants composed of osmolytes might be not sufficient to preserve all immune cells. The cytotoxic T-cell may require other compounds to enhance preservation quality, as well as other immune cells such as NK and NK T-cells. IL-2, HSA and other cytokines or metabolic factors for immune cells may be required to increase post-thaw recovery. In addition, some subsets, such as cytotoxic T-cells, might be separated and cryopreserved individually in order to eliminate regulatory signaling from other cell types.

121

DEMO has been applied to solve engineering problems such as reservoir system optimization, industrial chemical reactor, even hypersonic vehicle and rocket engine

[160–163]. The formulations of DMSO-free cryoprotectants were optimized using

DEMO and localized several compatible formulations including the original formulations.

The results of the DEMO algorithm confirmed our previous works [68–70,85], which started from an immortalized CD3+ T-cell line and found one optimal formulation for each group of osmolytes. These particular formulations of DMSO-free cryoprotectants presented similar post-thaw recoveries for human T-cells and the Jurkat cell line, as well as higher post-thaw recovery of helper T-cell compared to DMSO. These results not only confirm that osmolyte mixtures can serve as an effective alternative to DMSO with human primary cells but also show that the methodology we previously developed for immortalized cell lines can be directly applied to primary cells.

The DMSO-free cryoprotectants can be used to achieve high post-thaw recovery of many subpopulations of PBMCs but developing customized cryoprotectants will likely still be necessary to optimize recovery of particular subsets of interest. The optimization of dual subsets might result in bias to another subset, which indicates the difficulty of cryopreserving all subsets in a heterogeneous population with a universal solution.

However, this phenomenon indicates that cryopreservation could become an alternative label-free method to automatically sort out targeted subsets. In other words, undesired subsets can be naturally removed in cryopreservation with specific formulations to avoid the process of cell sorting.

122

DSC was utilized to measure heat releases as a function temperature. This instrument is ideally suited for a variety of measurements of the state of water (particularly liquid to solid phase changes) in biological systems. [152–154]. The DSC data demonstrated the melting profile of TGI, which addresses the different performance between three DMSO- free cryoprotectants. Raman images previously published [68,69,85,113] also provided insightful information to understand the mechanism behind the interactions of osmolytes.

The post-thaw proliferation of PBMC showed DMSO-free cryoprotectants had higher survival rates long term after thawing, which indicated DMSO-free cryoprotectants may accelerate the enrichment process in manufacturing cellular therapy products.

123

Chapter 8: Conclusions and Future Work

8.1 Conclusions

The objective of this research is to develop DMSO-free cryoprotectants to preserve human immune cells, understand the mechanisms of cryoprotectant molecules using

Raman spectroscopy and statistical modeling, and understand the influences of control parameters in differential evolution algorithm to improve the optimization and analyze the freezing responses of PBMC subset cryopreserved in DMSO-free cryoprotectants. To fulfill this goal, we proposed the following hypotheses and aims:

Hypothesis 1: Osmolytes act in concert to improve cell viability and interactions between different osmolytes is critical for the overall performance of the solutions

We confirmed this hypothesis by fulfilling the following aims in Chapter 4-5.

Aim 1.1: Measure post-thaw recovery of Jurkat cells cryopreserved with single osmolytes and combinations of osmolytes using high-throughput screening.

In Chapter 4-5, the post-thaw recoveries of Jurkat cells cryopreserved with all formulations (216) of five different combinations (SGI, SGC, SMC, TGI, and TGC) of sugars, sugar alcohols and amino acids under one cooling profile were measured via high-throughput screening. The incubation time and cooling rate were determined before the screening. The experimental data directly indicated that the post-thaw recovery depended on the formulation and each combination had optimal concentrations of each osmolyte. The data also showed a single osmolyte was not enough to cryopreserve cells

124 and osmolyte combinations exhibited much higher post-thaw recoveries. The observed data explicitly showed a complex interaction between osmolytes.

Post-thaw recovery of Jurkat cells in a single osmolyte solution was low. Combinations of osmolytes displayed distinct trends of post-thaw recovery and a non-linear relationship between compositions and post-thaw recovery was observed, suggesting interactions not only between different solutes but also between solutes and cells. The post-thaw recovery for optimized cryoprotectants in different combinations of osmolytes at a cooling rate of

1C/min was comparable to that measured with 10% dimethyl sulfoxide.

125

Aim 1.2: Characterize the osmolyte distribution of lymphocytes and ice morphology during freezing using Raman spectroscopy

Raman spectroscopy was used to quantify the freezing response of cells to various formulations of cryoprotectants and presented different protective properties of osmolytes to frozen cells in both Chapter 4 and 5. Raman images indicated that sucrose was non- penetrating and overlapped with the cellular membrane and glycerol penetrated the cell membrane. The interaction between the cell membrane and sugars has been a long-term hypothesis and was supported by Raman imaging. Raman images clearly demonstrated that damaging intracellular ice formation was observed more often in the presence of single osmolytes as well as non-optimized multi-component solution compositions.

Raman images also explained how concentrations of osmolytes affected cryopreservation.

High concentration sucrose was able to damage the cell membrane and low concentration glycerol could not penetrate the cell membrane. Both resulted in poor post-thaw recovery.

In addition, blebbing morphology, big chunk of intracellular ice, and non-concentrated signal of Amide I were seen in Jurkat cells cryopreserved using either a single osmolyte or suboptimal formulation of multiple-osmolyte solutions.

Raman images showed the influence of osmolytes and combinations of osmolytes on ice crystal shape in terms of size and ellipticity, which reflected the interactions between osmolytes and water. Differences in composition also influenced the presence or absence of intracellular ice formation, which could also be detected by Raman spectroscopy.

126

Aim 1.3: Analyze the main effects and interactions of osmolytes using statistical modeling

The statistical modeling provides another way to establish a numerical function for inferencing the experimental data. It was used to understand the importance of individual osmolytes as well as interactions between osmolytes on post-thaw recovery. Statistical modeling suggested both higher concentrations of glycerol and certain interactions between sugars and glycerol were found to typically increase post-thaw recovery. The statistical models supported the observations made via Raman spectroscopy.

127

Hypothesis 2: Different controlling parameters influence the ability of the differential evolution algorithm to optimize the compositions of DMSO-free cryoprotectants

We confirmed this hypothesis by fulfilling the aims in Chapter 6.

Aim 2: The experimental data is used to test several types of differential evolution algorithms and their control parameters including mutation, crossover, and population size.

In Chapter 6, this study presents the influence of control parameters including population size (NP), mutation factor (F), crossover (Cr) and four types of differential evolution algorithms including random, best, local-to-best and local-to-best with self-adaptive modification for the purpose of optimizing the compositions of DMSO-free cryoprotectants. Post-thaw recovery of Jurkat cells cryopreserved with two DMSO-free cryoprotectants at a cooling rate of 1C/min displayed a non-linear, 4-dimensional structure with multiple saddle nodes, which was a suitable training model to tune the control parameters and select the most appropriate type of differential evolution algorithm. Self-adaptive modification presented a better performance in terms of optimization accuracy and sensitivity of mutation factor and crossover among the four different types of algorithms tested. Specifically, the classical type of differential evolution algorithm exhibited a wide acceptance of mutation factor and crossover. The optimization performance is more sensitive to mutation than crossover and the optimization accuracy is proportional to the population size. Increasing population size also reduces the sensitivity of the algorithm to the value of the mutation factor and

128 crossover. The analysis of optimization accuracy and convergence speed suggests a larger population size with F>0.7 and Cr>0.3 are well suited for use with cryopreservation optimization purposes. The tuned differential evolution algorithm is validated through finding global maximums of other two DMSO-free cryoprotectant formulation data sets.

The results of these studies can be used to more efficiently determine the optimal composition of multicomponent DMSO-free cryoprotectants in the future.

129

Hypothesis 3: The DMSO-free cryoprotectants developed based on immortalized cell line provide equivalent performance to various types of human immune cells in comparison to conventional DMSO-containing cryoprotectant

We confirmed this hypothesis by fulfilling the aims in Chapter 7.

Aim 3: Measure the compositions and viabilities of peripheral blood mononuclear cells subsets with DMSO-free cryoprotectants as well as DMSO

The screening of various formulations using differential evolution showed the high post- thaw recoveries of all white blood cells (CD45+) with osmolyte-based cryoprotectants, which exhibited the potential to cryopreserve a heterogeneous population. The phenotypes and viabilities of PBMC subsets were characterized using flow cytometry.

Osmolyte-based cryoprotectants displayed high post-thaw recoveries for both T-cell and

B-cell subsets. Significant differences between the post-thaw recovery for helper T-cell

(CD3+CD4+) and cytotoxic T-cell (CD3+CD8+) were observed. A differential evolution algorithm was used to optimize cryoprotectant composition using multiple objectives in order to improve the post-thaw recoveries of these two cell types simultaneously. Several formulations with similar post-thaw recovery for both subsets were determined.

Statistical models were used to analyze the importance of individual osmolytes and interactions between post-thaw recoveries of three subsets of T-cell including helper T- cell, cytotoxic T-cell and natural killer T-cell (CD3+CD56+). The statistical model indicated that the preferred concentration levels of osmolytes and interaction modes were distinct between these three subsets. As post-thaw apoptosis is a significant concern for lymphocytes, the stability of frozen and thawed PBMC was observed for osmolyte-

130 containing solutions and solutions containing dimethyl sulfoxide (DMSO). Improved post-thaw stability of the cells was observed. This study helped us to understand the freezing responses of different subsets in human PBMCs using combinations of osmolytes.

131

8.2 Future Work

8.2.1 Understanding the role of amino acids using novel microscopy

The statistical models suggest that interactions between sugar and amino acids exist.

However, amino acids cannot be detected with Raman spectroscopy due to their small molecular size and overlapped peak with other osmolytes. Previous works proposed several hypotheses about the mechanism of amino acids as a cryoprotectant [164,165], but have not been proved experimentally. New instruments could be used to detect the amino acid such as two-photon microscopy and scanning tunneling microscopy [166].

8.2.2 Simulating the interactions between cell, ice and osmolytes using molecular

dynamic simulation

Raman images interpreted the roles of sugar and sugar alcohol during freezing, and statistical models quantified the interactions between osmolytes. However, the interactions between cells, ice and osmolytes are still unclear in the molecular prospect.

The molecular dynamics simulation has been intensively used to understand the interactions between water and osmolytes [167–171], and interactions between cryoprotectants and cell membranes [63,172–174]. The molecular dynamics simulation might be able to model the triple interactions between the cell (membrane and cytoskeleton), water molecules and osmolytes for proving insightful information in cryobiology.

132

8.2.3 Optimizing cryoprotectant formulations using other computational

algorithms

In this work, the control parameters including mutation, crossover and population size have been tuned to optimize the formulations of DMSO-free cryoprotectants.

However, there are many optimization algorithms, such as the artificial bee colony algorithm, that might be used to optimize the formulations with higher accuracy and convergence speed. In addition, deep learning (i.e., neural network) is a rising technology in many fields. One might build up a neural network model to optimize the formulations of DMSO-free cryoprotectants.

8.2.4 Improving the formulations to cryopreserve cytotoxic T-cell and other

subsets

The unpaired post-thaw recoveries between helper T-cells (CD3+CD4+) and cytotoxic T- cells (CD3+CD8+) were presented in Chapter 7. That phenomenon is independent of formulations in various combinations of osmolytes. The mechanism of freezing responses and the biophysical difference between helper T-cell and cytotoxic T-cells are still unknown and should be understood in the future. The DMSO-free cryoprotectants composed of osmolytes might not be enough to cryopreserve all immune cells. Other additives such as human serum albumin or cytokines such as IL-2 are might be used. NK cells (CD3-CD56+) have been a rising star in immune cell therapy [175–177]. However, the post-thaw recovery of NK cells is not high enough for current DMSO-free cryoprotectants in this work. A specific DMSO-free cryoprotectant should be developed for NK cells in the future.

133

Chapter 7 showed the results of cryopreserving a heterogeneous population. The interactions were much more complicated than cryopreserving a homogeneous population because of the interactions between different cell types. The statistical models suggested different cell types had their own preferences and interactions with the osmolytes. The interaction mechanism and signal pathway between different cell types during freezing and thawing should be characterized in the future.

134

Bibliography

[1] Pettitt D, Arshad Z, Smith J, Stanic T, Holländer G, Brindley D. CAR-T Cells: A Systematic Review and Mixed Methods Analysis of the Clinical Trial Landscape. Mol Ther 2018;26:342–53. doi:10.1016/j.ymthe.2017.10.019. [2] Piscopo NJ, Mueller KP, Das A, Hematti P, Murphy WL, Palecek SP, et al. Bioengineering Solutions for Manufacturing Challenges in CAR T Cells. Biotechnol J 2018;13:1700095. doi:10.1002/biot.201700095. [3] Xu H, Cao W, Huang L, Xiao M, Cao Y, Zhao L, et al. Effects of cryopreservation on chimeric antigen receptor T cell functions. Cryobiology 2018;83:40–7. doi:10.1016/j.cryobiol.2018.06.007. [4] Yong KW, Wan Safwani WKZ, Xu F, Wan Abas WAB, Choi JR, Pingguan- Murphy B. Cryopreservation of Human Mesenchymal Stem Cells for Clinical Applications: Current Methods and Challenges. Biopreserv Biobank 2015;13:231– 9. doi:10.1089/bio.2014.0104. [5] Elliott GD, Wang S, Fuller BJ. Cryoprotectants: A review of the actions and applications of cryoprotective solutes that modulate cell recovery from ultra-low temperatures. Cryobiology 2017;76:74–91. doi:10.1016/j.cryobiol.2017.04.004. [6] Scott KL, Lecak J, Acker JP. Biopreservation of Red Blood Cells: Past, Present, and Future. Transfus Med Rev 2005;19:127–42. doi:10.1016/j.tmrv.2004.11.004. [7] Alencar S, Garnica M, Luiz RR, Nogueira CM, Borojevic R, Maiolino A, et al. Cryopreservation of peripheral blood stem cell: the influence of cell concentration on cellular and hematopoietic recovery. Transfusion 2010;50:2402–12. doi:10.1111/j.1537-2995.2010.02743.x. [8] Hubel A. Parameters of cell freezing: Implications for the cryopreservation of stem cells. Transfus Med Rev 1997;11:224–33. doi:10.1053/tmrv.1997.0110224. [9] Hanna J, Hubel A. Preservation of stem cells. Organogenesis 2009;5:134–7. doi:10.4161/org.5.3.9585. [10] Lovelock JE, Bishop MWH. Prevention of Freezing Damage to Living Cells by Dimethyl Sulphoxide. Nature 1959;183:1394–5. doi:10.1038/1831394a0. [11] Windrum P, Morris TCM, Drake MB, Niederwieser D, Ruutu T. Variation in dimethyl sulfoxide use in stem cell transplantation: a survey of EBMT centres. Bone Marrow Transplant 2005;36:601–3. doi:10.1038/sj.bmt.1705100. [12] Hubel A. Preservation of Cells. 1st ed. Hoboken, NJ, USA: John Wiley & Sons, Inc.; 2018. doi:10.1002/9781118989883. [13] Areman EM, Sacher RA, Deeg HJ. Processing and storage of human bone marrow: a survey of current practices in North America. Bone Marrow Transplant 1990;6:203–9. [14] de Abreu Costa L, Henrique Fernandes Ottoni M, dos Santos M, Meireles A, Gomes de Almeida V, de Fátima Pereira W, et al. Dimethyl Sulfoxide (DMSO) Decreases Cell Proliferation and TNF-α, IFN-γ, and IL-2 Cytokines Production in Cultures of Peripheral Blood Lymphocytes. Molecules 2017;22:1789. doi:10.3390/molecules22111789. [15] Yancey P, Clark M, Hand S, Bowlus R, Somero G. Living with water stress: evolution of osmolyte systems. Science (80- ) 1982;217:1214–22. doi:10.1126/science.7112124.

135

[16] Schneider U, Schwenk H-U, Bornkamm G. Characterization of EBV-genome negative “null” and “T” cell lines derived from children with acute lymphoblastic leukemia and leukemic transformed non-Hodgkin lymphoma. Int J Cancer 1977;19:621–6. doi:10.1002/ijc.2910190505. [17] Dong J, Malsam J, Bischof JC, Hubel A, Aksan A. Spatial Distribution of the State of Water in Frozen Mammalian Cells. Biophys J 2010;99:2453–9. doi:10.1016/j.bpj.2010.08.035. [18] Sherman JK. Survival of higher animal cells after the formation and dissolution of intracellular ice. Anat Rec 1962;144:171–89. doi:10.1002/ar.1091440302. [19] Chang A, Dantzig JA, Darr BT, Hubel A. Modeling the interaction of biological cells with a solidifying interface. J Comput Phys 2007;226:1808–29. doi:10.1016/j.jcp.2007.05.039. [20] Ishiguro H, Rubinsky B. Mechanical Interactions between Ice Crystals and Red Blood Cells during Directional Solidification. Cryobiology 1994;31:483–500. doi:10.1006/cryo.1994.1059. [21] Lipp G, Galow S, Körber C, Rau G. Encapsulation of Human Erythrocytes by Growing Ice Crystals. Cryobiology 1994;31:305–12. doi:10.1006/cryo.1994.1036. [22] Hubel A, Darr TB, Chang A, Dantzig J. Cell partitioning during the directional solidification of trehalose solutions. Cryobiology 2007;55:182–8. doi:10.1016/j.cryobiol.2007.07.002. [23] Mao L, Udaykumar HS, Karlsson JOM. Simulation of micro-scale interaction between ice and biological cells. Int J Heat Mass Transf 2003;46:5123–36. doi:10.1016/S0017-9310(03)00364-8. [24] Carin M, Jaeger M. Numerical simulation of the interaction of biological cells with an ice front during freezing. Eur Phys J Appl Phys 2001;16:231–8. doi:10.1051/epjap:2001205. [25] Pegg DE. The relevance of ice crystal formation for the cryopreservation of tissues and organs. Cryobiology 2010;60:S36–44. doi:10.1016/j.cryobiol.2010.02.003. [26] Zhang M, Gao C, Ye B, Tang J, Jiang B. Effects of four disaccharides on nucleation and growth of ice crystals in concentrated glycerol aqueous solution. Cryobiology 2019;86:47–51. doi:10.1016/j.cryobiol.2018.12.006. [27] Kobayashi A, Golash HN, Kirschvink JL. A first test of the hypothesis of biogenic magnetite-based heterogeneous ice-crystal nucleation in cryopreservation. Cryobiology 2016;72:216–24. doi:10.1016/j.cryobiol.2016.04.003. [28] Toner M, Cravalho EG, Karel M. Thermodynamics and kinetics of intracellular ice formation during freezing of biological cells. J Appl Phys 1990;67:1582–93. doi:10.1063/1.345670. [29] John Morris G, Acton E. Controlled ice nucleation in cryopreservation - A review. Cryobiology 2013;66:85–92. doi:10.1016/j.cryobiol.2012.11.007. [30] Dowgert MF, Steponkus PL. Effect of Cold Acclimation on Intracellular Ice Formation in Isolated Protoplasts. Plant Physiol 1983;72:978–88. doi:10.1104/pp.72.4.978. [31] Lovelock JE. The haemolysis of human red blood-cells by freezing and thawing. Biochim Biophys Acta 1953;10:414–26. doi:10.1016/0006-3002(53)90273-X. [32] Steponkus PL, Dowgert MF, Gordon-Kamm WJ. Destabilization of the plasma membrane of isolated plant protoplasts during a freeze-thaw cycle: The influence

136

of cold acclimation. Cryobiology 1983;20:448–65. doi:10.1016/0011- 2240(83)90035-4. [33] Steponkus PL, Lynch D V. Freeze/thaw-induced destabilization of the plasma membrane and the effects of cold acclimation. J Bioenerg Biomembr 1989;21:21– 41. doi:10.1007/BF00762210. [34] Meryman HT. Freezing Injury and its Prevention in Living Cells. Annu Rev Biophys Bioeng 1974;3:341–63. doi:10.1146/annurev.bb.03.060174.002013. [35] Leibo SP, Mazur P. The role of cooling rates in low-temperature preservation. Cryobiology 1971;8:447–52. doi:10.1016/0011-2240(71)90035-6. [36] Buhl T, Legler TJ, Rosenberger A, Schardt A, Schön MP, Haenssle HA. Controlled-rate freezer cryopreservation of highly concentrated peripheral blood mononuclear cells results in higher cell yields and superior autologous T-cell stimulation for dendritic cell-based immunotherapy. Cancer Immunol Immunother 2012;61:2021–31. doi:10.1007/s00262-012-1262-0. [37] Shivakumar SB, Bharti D, Jang S-J, Hwang S-C, Park J-K, Shin J-K, et al. Cryopreservation of Human Wharton’s Jelly-derived Mesenchymal Stem Cells Following Controlled Rate Freezing Protocol Using Different Cryoprotectants; A Comparative Study. Int J Stem Cells 2015;8:155–69. doi:10.15283/ijsc.2015.8.2.155. [38] Roach KL, King KR, Uygun K, Hand SC, Kohane IS, Yarmush ML, et al. High- throughput single cell arrays as a novel tool in biopreservation. Cryobiology 2009;58:315–21. doi:10.1016/j.cryobiol.2009.03.001. [39] Mazur P. Freezing of living cells: mechanisms and implications. Am J Physiol Physiol 1984;247:C125–42. doi:10.1152/ajpcell.1984.247.3.C125. [40] Karlsson JO, Cravalho EG, Borel Rinkes IH, Tompkins RG, Yarmush ML, Toner M. Nucleation and growth of ice crystals inside cultured hepatocytes during freezing in the presence of dimethyl sulfoxide. Biophys J 1993;65:2524–36. doi:10.1016/S0006-3495(93)81319-5. [41] Narduche L, Hamond C, Martins GMS, Medeiros MA, Lilenbaum W. Recovery of Leptospires in Short- and Medium-Term Cryopreservation Using Different Glycerol and Dimethyl Sulfoxide Concentrations. Biopreserv Biobank 2016;14:81–3. doi:10.1089/bio.2015.0046. [42] de Paz P, Alvarez-Rodriguez M, Nicolas M, Alvarez M, Chamorro C, Borragán S, et al. Optimization of Glycerol Concentration and Freezing Rate in the Cryopreservation of Ejaculate From Brown Bear (Ursus arctos). Reprod Domest Anim 2012;47:105–12. doi:10.1111/j.1439-0531.2011.01808.x. [43] Dijkstra-Tiekstra MJ, Setroikromo AC, Kraan M, Gkoumassi E, de Wildt-Eggen J. Optimization of the freezing process for hematopoietic progenitor cells: effect of precooling, initial dimethyl sulfoxide concentration, freezing program, and storage in vapor-phase or liquid nitrogen on in vitro white blood cell quality. Transfusion 2014;54:3155–63. doi:10.1111/trf.12756. [44] Marsella T, Sena P, Xella S, La Marca A, Giulini S, De Pol A, et al. Human ovarian tissue cryopreservation: effect of sucrose concentration on morphological features after thawing. Reprod Biomed Online 2008;16:257–67. doi:10.1016/S1472-6483(10)60583-4. [45] Ragoonanan V, Less R, Aksan A. Response of the cell membrane-cytoskeleton

137

complex to osmotic and freeze/thaw stresses. Part 2: The link between the state of the membrane-cytoskeleton complex and the cellular damage. Cryobiology 2013;66:96–104. doi:10.1016/j.cryobiol.2012.10.008. [46] Ragoonanan V, Hubel A, Aksan A. Response of the cell membrane-cytoskeleton complex to osmotic and freeze/thaw stresses. Cryobiology 2010;61:335–44. doi:10.1016/j.cryobiol.2010.10.160. [47] Laster SM, Mackenzie JM. Bleb formation and F-actin distribution during mitosis and tumor necrosis factor-induced apoptosis. Microsc Res Tech 1996;34:272–80. doi:10.1002/(SICI)1097-0029(19960615)34:3<272::AID-JEMT10>3.0.CO;2-J. [48] Gores GJ, Herman B, Lemasters JJ. Plasma membrane bleb formation and rupture: A common feature of hepatocellular injury. Hepatology 1990;11:690–8. doi:10.1002/hep.1840110425. [49] Mazur P. Life in the Frozen State. 1st ed. Boca Raton, FL: CRC Press; 2004. doi:10.1201/9780203647073. [50] Van Venrooij GEPM, Aertsen AMHJ, Hax WMA, Ververgaert PHJT, Verhoeven JJ, Van Der Vorst HA. Freeze-etching: Freezing velocity and crystal size at different locations in samples. Cryobiology 1975;12:46–61. doi:10.1016/0011- 2240(75)90040-1. [51] Fiser PS, Fairfull RW, Marcus GJ. The effect of thawing velocity on survival and acrosomal integrity of ram spermatozoa frozen at optimal and suboptimal rates in straws. Cryobiology 1986;23:141–9. doi:10.1016/0011-2240(86)90005-2. [52] van den Abbeel E, Van Der Elst J, Van Steirteghem AC. The Effect of Temperature at Which Slow Cooling Is Terminated and of Thawing Rate on the Survival of One-Cell Mouse Embryos Frozen in Dimethyl Sulfoxide or 1,2- Propanediol Solutions. Cryobiology 1994;31:423–33. doi:10.1006/cryo.1994.1052. [53] Rall WF, Polge C. Effect of warming rate on mouse embryos frozen and thawed in glycerol. Reproduction 1984;70:285–92. doi:10.1530/jrf.0.0700285. [54] Polge C, Smith AU, Parkes AS. Revival of Spermatozoa after Vitrification and Dehydration at Low Temperatures. Nature 1949;164:666–666. doi:10.1038/164666a0. [55] Anchordoguy TJ, Rudolph AS, Carpenter JF, Crowe JH. Modes of interaction of cryoprotectants with membrane phospholipids during freezing. Cryobiology 1987;24:324–31. doi:10.1016/0011-2240(87)90036-8. [56] Crowe JH, Leslie SB, Crowe LM. Is Vitrification Sufficient to Preserve Liposomes during Freeze-Drying? Cryobiology 1994;31:355–66. doi:10.1006/cryo.1994.1043. [57] Crowe JH, Crowe LM, Carpenter JF, Rudolph AS, Wistrom CA, Spargo BJ, et al. Interactions of sugars with membranes. Biochim Biophys Acta - Rev Biomembr 1988;947:367–84. doi:10.1016/0304-4157(88)90015-9. [58] Leslie SB, Israeli E, Lighthart B, Crowe JH, Crowe LM. Trehalose and sucrose protect both membranes and proteins in intact bacteria during drying. Appl Environ Microbiol 1995;61:3592–7. [59] Wolfe J, Bryant G. Freezing, Drying, and/or Vitrification of Membrane– Solute– Water Systems. Cryobiology 1999;39:103–29. doi:10.1006/cryo.1999.2195. [60] Bischof JC, Rubinsky B. Large Ice Crystals in the Nucleus of Rapidly Frozen Liver Cells. Cryobiology 1993;30:597–603. doi:10.1006/cryo.1993.1062. [61] Arakawa T, Carpenter JF, Kita YA, Crowe JH. The basis for toxicity of certain

138

cryoprotectants: A hypothesis. Cryobiology 1990;27:401–15. doi:10.1016/0011- 2240(90)90017-X. [62] Gurtovenko A a, Anwar J. Modulating the Structure and Properties of Cell Membranes: The Molecular Mechanism of Action of Dimethyl Sulfoxide. J Phys Chem B 2007;111:10453–60. doi:10.1021/jp073113e. [63] Sum AK, de Pablo JJ. Molecular Simulation Study on the Influence of Dimethylsulfoxide on the Structure of Phospholipid Bilayers. Biophys J 2003;85:3636–45. doi:10.1016/S0006-3495(03)74781-X. [64] Notman R, Noro M, O’Malley B, Anwar J. Molecular Basis for Dimethylsulfoxide (DMSO) Action on Lipid Membranes. J Am Chem Soc 2006;128:13982–3. doi:10.1021/ja063363t. [65] Thaler R, Spitzer S, Karlic H, Klaushofer K, Varga F. DMSO is a strong inducer of DNA hydroxymethylation in pre-osteoblastic MC3T3-E1 cells. Epigenetics 2012;7:635–51. doi:10.4161/epi.20163. [66] Fleming Glass KK, Longmire EK, Hubel A. Optimization of a microfluidic device for diffusion-based extraction of DMSO from a cell suspension. Int J Heat Mass Transf 2008;51:5749–57. doi:10.1016/j.ijheatmasstransfer.2008.04.018. [67] Hanna J, Hubel A, Lemke E. Diffusion-based extraction of DMSO from a cell suspension in a three stream, vertical microchannel. Biotechnol Bioeng 2012;109:2316–24. doi:10.1002/bit.24499. [68] Pi C-H, Yu G, Dosa PI, Hubel A. Characterizing modes of action and interaction for multicomponent osmolyte solutions on Jurkat cells. Biotechnol Bioeng 2019;116:631–43. doi:10.1002/bit.26880. [69] Pi C-H, Yu G, Petersen A, Hubel A. Characterizing the “sweet spot” for the preservation of a T-cell line using osmolytes. Sci Rep 2018;8:16223. doi:10.1038/s41598-018-34638-7. [70] Pollock K, Yu G, Moller-Trane R, Koran M, Dosa PI, McKenna DH, et al. Combinations of Osmolytes, Including Monosaccharides, Disaccharides, and Sugar Alcohols Act in Concert During Cryopreservation to Improve Mesenchymal Stromal Cell Survival. Tissue Eng Part C Methods 2016;22:999–1008. doi:10.1089/ten.tec.2016.0284. [71] Pasley S, Zylberberg C, Matosevic S. Natural killer-92 cells maintain cytotoxic activity after long-term cryopreservation in novel DMSO-free media. Immunol Lett 2017;192:35–41. doi:10.1016/j.imlet.2017.09.012. [72] Katkov II, Kan NG, Cimadamore F, Nelson B, Snyder EY, Terskikh A V. DMSO- Free Programmed Cryopreservation of Fully Dissociated and Adherent Human Induced Pluripotent Stem Cells. Stem Cells Int 2011;2011:1–8. doi:10.4061/2011/981606. [73] Nazarpour R, Zabihi E, Alijanpour E, Abedian Z, Mehdizadeh H, Rahimi F. Optimization of Human Peripheral Blood Mononuclear Cells (PBMCs) Cryopreservation. Int J Mol Cell Med 2012;1:88–93. [74] Dong Q, Hill D, VandeVoort CA. Interactions among pre-cooling, cryoprotectant, cooling, and thawing for sperm cryopreservation in rhesus monkeys. Cryobiology 2009;59:268–74. doi:10.1016/j.cryobiol.2009.08.002. [75] Freimark D, Sehl C, Weber C, Hudel K, Czermak P, Hofmann N, et al. Systematic parameter optimization of a Me2SO- and serum-free cryopreservation protocol for

139

human mesenchymal stem cells. Cryobiology 2011;63:67–75. doi:10.1016/j.cryobiol.2011.05.002. [76] Rusco G, Di Iorio M, Gibertoni PP, Esposito S, Penserini M, Roncarati A, et al. Optimization of Sperm Cryopreservation Protocol for Mediterranean Brown Trout: A Comparative Study of Non-Permeating Cryoprotectants and Thawing Rates In Vitro and In Vivo. Animals 2019;9:304. doi:10.3390/ani9060304. [77] Conrad PB, Miller DP, Cielenski PR, de Pablo JJ. Stabilization and Preservation of Lactobacillus acidophilus in Saccharide Matrices. Cryobiology 2000;41:17–24. doi:10.1006/cryo.2000.2260. [78] Kearney JN, Wheldon LA, Gowland G. Effects of cryobiological variables on the survival of skin using a defined murine model. Cryobiology 1990;27:164–70. doi:10.1016/0011-2240(90)90008-R. [79] Storn R, Price K. Differential Evolution – A Simple and Efficient Heuristic for global Optimization over Continuous Spaces. J Glob Optim 1997;11:341–59. doi:10.1023/A:100820282. [80] Qin AK, Suganthan PN. Self-adaptive Differential Evolution Algorithm for Numerical Optimization. 2005 IEEE Congr. Evol. Comput., vol. 2, IEEE; 2005, p. 1785–91. doi:10.1109/CEC.2005.1554904. [81] Brest J, Greiner S, Boskovic B, Mernik M, Zumer V. Self-Adapting Control Parameters in Differential Evolution: A Comparative Study on Numerical Benchmark Problems. IEEE Trans Evol Comput 2006;10:646–57. doi:10.1109/TEVC.2006.872133. [82] Dietz AB, Bulur PA, Emery RL, Winters JL, Epps DE, Zubair AC, et al. A novel source of viable peripheral blood mononuclear cells from leukoreduction system chambers. Transfusion 2006;46:2083–9. doi:10.1111/j.1537-2995.2006.01033.x. [83] de Veij M, Vandenabeele P, De Beer T, Remon JP, Moens L. Reference database of Raman spectra of pharmaceutical excipients. J Raman Spectrosc 2009;40:297– 307. doi:10.1002/jrs.2125. [84] Mendelovici E, Frost RL, Kloprogge T. Cryogenic Raman spectroscopy of glycerol. J Raman Spectrosc 2000;31:1121–6. doi:10.1002/1097- 4555(200012)31:12<1121::AID-JRS654>3.0.CO;2-G. [85] Yu G, Yap YR, Pollock K, Hubel A. Characterizing Intracellular Ice Formation of Lymphoblasts Using Low-Temperature Raman Spectroscopy. Biophys J 2017;112:2653–63. doi:10.1016/j.bpj.2017.05.009. [86] Eroglu A, Russo MJ, Bieganski R, Fowler A, Cheley S, Bayley H, et al. Intracellular trehalose improves the survival of cryopreserved mammalian cells. Nat Biotechnol 2000;18:163–7. doi:10.1038/72608. [87] Layne Jr. JR, Lee Jr. RE. Freeze tolerance and the dynamics of ice formation in wood frogs ( Rana sylvatica ) from southern Ohio. Can J Zool 1987;65:2062–5. doi:10.1139/z87-315. [88] Holthauzen LMF, Bolen DW. Mixed osmolytes: The degree to which one osmolyte affects the protein stabilizing ability of another. Protein Sci 2006;16:293–8. doi:10.1110/ps.062610407. [89] Yancey PH. Organic osmolytes as compatible, metabolic and counteracting cytoprotectants in high osmolarity and other stresses. J Exp Biol 2005;208:2819– 30. doi:10.1242/jeb.01730.

140

[90] Pollock K, Sumstad D, Kadidlo D, McKenna DH, Hubel A. Clinical mesenchymal stromal cell products undergo functional changes in response to freezing. Cytotherapy 2015;17:38–45. doi:10.1016/j.jcyt.2014.06.008. [91] Pollock K, Budenske JW, McKenna DH, Dosa PI, Hubel A. Algorithm-driven optimization of cryopreservation protocols for transfusion model cell types including Jurkat cells and mesenchymal stem cells. J Tissue Eng Regen Med 2017;11:2806–15. doi:10.1002/term.2175. [92] Pollock K, Samsonraj RM, Dudakovic A, Thaler R, Stumbras A, McKenna DH, et al. Improved Post-Thaw Function and Epigenetic Changes in Mesenchymal Stromal Cells Cryopreserved Using Multicomponent Osmolyte Solutions. Stem Cells Dev 2017;26:828–42. doi:10.1089/scd.2016.0347. [93] Roozen MJGW, Hemminga MA. Molecular motion in sucrose-water mixtures in the liquid and glassy state as studied by spin probe ESR. J Phys Chem 1990;94:7326–9. doi:10.1021/j100381a068. [94] Vanderkooi JM, Dashnau JL, Zelent B. Temperature excursion infrared (TEIR) spectroscopy used to study hydrogen bonding between water and biomolecules. Biochim Biophys Acta - Proteins Proteomics 2005;1749:214–33. doi:10.1016/j.bbapap.2005.03.008. [95] McGrath JJ. Membrane transport properties. In: McGrath JJ, Diller KR, editors. Low Temp. Biotechnol. Emerg. Appl. Eng. Contrib., New York: ASME; 1988, p. 273–330. [96] Carpenter TAKF. Protein–Solvent Interactions in Pharmaceutical Formulations. Pharm Res 1991;8:285–291. doi:10.1023/A:1015825027737. [97] Wen X, Wang S, Duman JG, Arifin JF, Juwita V, Goddard WA, et al. Antifreeze proteins govern the precipitation of trehalose in a freezing-avoiding insect at low temperature. Proc Natl Acad Sci 2016;113:6683–8. doi:10.1073/pnas.1601519113. [98] Abraham RT, Weiss A. Jurkat T cells and development of the T-cell receptor signalling paradigm. Nat Rev Immunol 2004;4:301–8. doi:10.1038/nri1330. [99] Castro VIB, Craveiro R, Silva JM, Reis RL, Paiva A, Ana AR. Natural deep eutectic systems as alternative nontoxic cryoprotective agents. Cryobiology 2018;83:15–26. doi:10.1016/j.cryobiol.2018.06.010. [100] Petrenko YA, Rogulska OY, Mutsenko V V, Petrenko AY. A sugar pretreatment as a new approach to the Me2SO- and xeno-free cryopreservation of human mesenchymal stromal cells. Cryo Letters 2014;35:239–46. [101] Matsumura K, Bae JY, Hyon SH. Polyampholytes as Cryoprotective Agents for Mammalian Cell Cryopreservation. Cell Transplant 2010;19:691–9. doi:10.3727/096368910X508780. [102] Gläfke C, Akhoondi M, Oldenhof H, Sieme H, Wolkers WF. Cryopreservation of platelets using trehalose: The role of membrane phase behavior during freezing. Biotechnol Prog 2012;28:1347–54. doi:10.1002/btpr.1600. [103] Martinetti D, Colarossi C, Buccheri S, Denti G, Memeo L, Vicari L. Effect of trehalose on cryopreservation of pure peripheral blood stem cells. Biomed Reports 2017;6:314–8. doi:10.3892/br.2017.859. [104] Rao W, Huang H, Wang H, Zhao S, Dumbleton J, Zhao G, et al. Nanoparticle- Mediated Intracellular Delivery Enables Cryopreservation of Human Adipose- Derived Stem Cells Using Trehalose as the Sole Cryoprotectant. ACS Appl Mater

141

Interfaces 2015;7:5017–28. doi:10.1021/acsami.5b00655. [105] Meryman HT, Hornblower M. A Method for Freezing and Washing Red Blood Cells Using a High Glycerol Concentration. Transfusion 1972;12:145–56. doi:10.1111/j.1537-2995.1972.tb00001.x. [106] Rowe AW, Eyster E, Kellner A. Liquid nitrogen preservation of red blood cells for transfusion. Cryobiology 1968;5:119–28. doi:10.1016/S0011-2240(68)80154-3. [107] Arakawa T, Timasheff SN. The stabilization of proteins by osmolytes. Biophys J 1985;47:411–4. doi:10.1016/S0006-3495(85)83932-1. [108] Singh LR, Poddar NK, Dar TA, Kumar R, Ahmad F. Protein and DNA destabilization by osmolytes: The other side of the coin. Life Sci 2011;88:117–25. doi:10.1016/j.lfs.2010.10.020. [109] Yancey PH. Water Stress, Osmolytes and Proteins. Am Zool 2001;41:699–709. doi:10.1093/icb/41.4.699. [110] Strauss G, Hauser H. Stabilization of lipid bilayer vesicles by sucrose during freezing. Proc Natl Acad Sci 1986;83:2422–6. doi:10.1073/pnas.83.8.2422. [111] Crowe JH, Hoekstra F a, Crowe LM. Membrane phase transitions are responsible for imbibitional damage in dry pollen. Proc Natl Acad Sci 1989;86:520–3. doi:10.1073/pnas.86.2.520. [112] Koster KL, Webb MS, Bryant G, Lynch D V. Interactions between soluble sugars and POPC (1-palmitoyl-2-oleoylphosphatidylcholine) during dehydration: vitrification of sugars alters the phase behavior of the phospholipid. Biochim Biophys Acta - Biomembr 1994;1193:143–50. doi:10.1016/0005-2736(94)90343-3. [113] Yu G, Li R, Hubel A. Interfacial Interactions of Sucrose during Cryopreservation Detected by Raman Spectroscopy. Langmuir 2019;35:7388–95. doi:10.1021/acs.langmuir.8b01616. [114] Liu J, Tanrikut C, Wright DL, Lee GY, Toner M, Biggers JD, et al. Cryopreservation of human spermatozoa with minimal non-permeable cryoprotectant. Cryobiology 2016;73:162–7. doi:10.1016/j.cryobiol.2016.08.004. [115] Shiraga K, Adachi A, Ogawa Y. Characterization of the hydrogen-bond network of water around sucrose and trehalose: H-O-H bending analysis. Chem Phys Lett 2017;678:59–64. doi:10.1016/j.cplett.2017.04.023. [116] Lerbret A, Bordat P, Affouard F, Guinet Y, Hédoux A, Paccou L, et al. Influence of homologous disaccharides on the hydrogen-bond network of water: complementary Raman scattering experiments and molecular dynamics simulations. Carbohydr Res 2005;340:881–7. doi:10.1016/j.carres.2005.01.036. [117] Lerbret A, Affouard F, Bordat P, Hédoux A, Guinet Y, Descamps M. Slowing down of water dynamics in disaccharide aqueous solutions. J Non Cryst Solids 2011;357:695–9. doi:10.1016/j.jnoncrysol.2010.05.092. [118] Bailey TL, Wang M, Solocinski J, Nathan BP, Chakraborty N, Menze MA. Protective effects of osmolytes in cryopreserving adherent neuroblastoma (Neuro- 2a) cells. Cryobiology 2015;71:472–80. doi:10.1016/j.cryobiol.2015.08.015. [119] Lovelock JE. The protective action of neutral solutes against haemolysis by freezing and thawing. Biochem J 1954;56:265–70. doi:10.1042/bj0560265. [120] Timasheff SN. The Control of Protein Stability and Association by Weak Interactions with Water: How Do Solvents Affect These Processes? Annu Rev Biophys Biomol Struct 1993;22:67–97. doi:10.1146/annurev.bb.22.060193.000435.

142

[121] Dashnau JL, Nucci N V., Sharp KA, Vanderkooi JM. Hydrogen Bonding and the Cryoprotective Properties of Glycerol/Water Mixtures. J Phys Chem B 2006;110:13670–7. doi:10.1021/jp0618680. [122] Silverstein KAT, Haymet ADJ, Dill KA. The Strength of Hydrogen Bonds in Liquid Water and Around Nonpolar Solutes. J Am Chem Soc 2000;122:8037–41. doi:10.1021/ja000459t. [123] Cocks FH, Brower WE. Phase diagram relationships in cryobiology. Cryobiology 1974;11:340–58. doi:10.1016/0011-2240(74)90011-X. [124] Redmond J, Bolin R, Cheney B. Glycerol-glucose cryopreservation of platelets. In vivo and in vitro observations. Transfusion 1983;23:213–4. doi:10.1046/j.1537- 2995.1983.23383224897.x. [125] Sei T, Gonda T, Arima Y. Growth rate and morphology of ice crystals growing in a solution of trehalose and water. J Cryst Growth 2002;240:218–29. doi:10.1016/S0022-0248(02)00875-8. [126] Uchida T, Nagayama M, Shibayama T, Gohara K. Morphological investigations of disaccharide molecules for growth inhibition of ice crystals. J Cryst Growth 2007;299:125–35. doi:10.1016/j.jcrysgro.2006.10.261. [127] Petzold G, Aguilera JM. Ice Morphology: Fundamentals and Technological Applications in Foods. Food Biophys 2009;4:378–96. doi:10.1007/s11483-009- 9136-5. [128] Gómez-Fernández J, Gómez-Izquierdo E, Tomás C, Mocé E, de Mercado E. Effect of different monosaccharides and disaccharides on boar sperm quality after cryopreservation. Anim Reprod Sci 2012;133:109–16. doi:10.1016/j.anireprosci.2012.06.010. [129] Hahn MB, Uhlig F, Solomun T, Smiatek J, Sturm H. Combined influence of ectoine and salt: spectroscopic and numerical evidence for compensating effects on aqueous solutions. Phys Chem Chem Phys 2016;18:28398–402. doi:10.1039/C6CP05417J. [130] Jena S, Suryanarayanan R, Aksan A. Mutual Influence of Mannitol and Trehalose on Crystallization Behavior in Frozen Solutions. Pharm Res 2016;33:1413–25. doi:10.1007/s11095-016-1883-7. [131] Lu X, Pikal MJ. Freeze‐Drying of Mannitol–Trehalose–Sodium Chloride‐Based Formulations: The Impact of Annealing on Dry Layer Resistance to Mass Transfer and Cake Structure. Pharm Dev Technol 2004;9:85–95. doi:10.1081/PDT- 120027421. [132] Lawson A, Mukherjee IN, Sambanis A. Mathematical modeling of cryoprotectant addition and removal for the cryopreservation of engineered or natural tissues. Cryobiology 2012;64:1–11. doi:10.1016/j.cryobiol.2011.11.006. [133] Anderson DM, Benson JD, Kearsley AJ. Foundations of modeling in cryobiology—I: Concentration, Gibbs energy, and chemical potential relationships. Cryobiology 2014;69:349–60. doi:10.1016/j.cryobiol.2014.09.004. [134] Das S, Suganthan PN. Differential Evolution: A Survey of the State-of-the-Art. IEEE Trans Evol Comput 2011;15:4–31. doi:10.1109/TEVC.2010.2059031. [135] Lichtblau D. Differential Evolution in Discrete Optimization. Int J Swarm Intell Evol Comput 2012;1:1–10. doi:10.4303/ijsiec/Z110301. [136] Jingqiao Zhang, Avasarala V, Sanderson AC, Mullen T. Differential evolution for

143

discrete optimization: An experimental study on Combinatorial Auction problems. 2008 IEEE Congr. Evol. Comput. (IEEE World Congr. Comput. Intell., IEEE; 2008, p. 2794–800. doi:10.1109/CEC.2008.4631173. [137] Das S, Mullick SS, Suganthan PN. Recent advances in differential evolution – An updated survey. Swarm Evol Comput 2016;27:1–30. doi:10.1016/j.swevo.2016.01.004. [138] Tsutsui H, Valamehr B, Hindoyan A, Qiao R, Ding X, Guo S, et al. An optimized small molecule inhibitor cocktail supports long-term maintenance of human embryonic stem cells. Nat Commun 2011;2:167. doi:10.1038/ncomms1165. [139] Brest J, Maučec MS. Self-adaptive differential evolution algorithm using population size reduction and three strategies. Soft Comput 2011;15:2157–74. doi:10.1007/s00500-010-0644-5. [140] Wong I, Liu W, Ho C-M, Ding X. Continuous Adaptive Population Reduction (CAPR) for Differential Evolution Optimization. SLAS Technol Transl Life Sci Innov 2017;22:289–305. doi:10.1177/2472630317690318. [141] Lin C, Qing A, Feng Q. A comparative study of crossover in differential evolution. J Heuristics 2011;17:675–703. doi:10.1007/s10732-010-9151-1. [142] Sreedhar D, Rajan M .R B. Differential Evolution based Multiobjective Optimization-A Review. Int J Comput Appl 2013;63:14–9. doi:10.5120/10541- 5019. [143] Robič T, Filipič B. DEMO: Differential Evolution for Multiobjective Optimization. Int. Conf. Evol. Multi-Criterion Optim., Springer, Berlin, Heidelberg; 2005, p. 520–33. doi:10.1007/978-3-540-31880-4_36. [144] Babu BV, Gujrathi AM. Multi-Objective Differential Evolution (MODE) Algorithm for Multi-Objective Optimization: Parametric Study on Benchmark Test Problems. I-Manager’s J Futur Eng Technol 2007;3:47–59. doi:10.26634/jfet.3.1.697. [145] Badu B V, Anabarasu B. Multi-Objective Differential Evolution (MODE): An Evolutionary Algorithm for Multi-Objective Optimization Problems (MOOPs). Evolution (N Y) 2006;31:1–7. [146] Hønge BL, Petersen MS, Olesen R, Møller BK, Erikstrup C. Optimizing recovery of frozen human peripheral blood mononuclear cells for flow cytometry. PLoS One 2017;12:e0187440. doi:10.1371/journal.pone.0187440. [147] Kety Pratt Riccio E, Neves I, Maria Banic D, Corte-Real S, das Graças Alecrim M, Morgado M, et al. Cryopreservation of peripheral blood mononuclear cells does not significantly affect the levels of spontaneous apoptosis after 24-h culture. Cryobiology 2002;45:127–34. doi:10.1016/S0011-2240(02)00121-9. [148] Angel S, von Briesen H, Oh Y-J, Baller MK, Zimmermann H, Germann A. Toward Optimal Cryopreservation and Storage for Achievement of High Cell Recovery and Maintenance of Cell Viability and T Cell Functionality. Biopreserv Biobank 2016;14:539–47. doi:10.1089/bio.2016.0046. [149] Germann A, Oh Y-J, Schmidt T, Schön U, Zimmermann H, von Briesen H. Temperature fluctuations during deep temperature cryopreservation reduce PBMC recovery, viability and T-cell function. Cryobiology 2013;67:193–200. doi:10.1016/j.cryobiol.2013.06.012. [150] Best A, Hidalgo G, Mitchell K, Yannelli JR. Issues concerning the large scale

144

cryopreservation of peripheral blood mononuclear cells (PBMC) for immunotherapy trials. Cryobiology 2007;54:294–7. doi:10.1016/j.cryobiol.2007.02.002. [151] Mathur A, Tripathi A, Kuse M. Scalable system for classification of white blood cells from Leishman stained blood stain images. J Pathol Inform 2013;4:15. doi:10.4103/2153-3539.109883. [152] Ablett S, Izzard MJ, Lillford PJ. Differential scanning calorimetric study of frozen sucrose and glycerol solutions. J Chem Soc Faraday Trans 1992;88:789. doi:10.1039/ft9928800789. [153] Wang GM, Haymet a. DJ. Trehalose and Other Sugar Solutions at Low Temperature: Modulated Differential Scanning Calorimetry (MDSC). J Phys Chem B 1998;102:5341–7. doi:10.1021/jp980942e. [154] Devireddy R V, Raha D, Bischof JC. Measurement of Water Transport during Freezing in Cell Suspensions Using a Differential Scanning Calorimeter. Cryobiology 1998;36:124–55. doi:10.1006/cryo.1997.2071. [155] Lemieux J, Jobin C, Simard C, Néron S. A global look into human T cell subsets before and after cryopreservation using multiparametric flow cytometry and two- dimensional visualization analysis. J Immunol Methods 2016;434:73–82. doi:10.1016/j.jim.2016.04.010. [156] Liang X, Hu X, Hu Y, Zeng W, Zeng G, Ren Y, et al. Recovery and functionality of cryopreserved peripheral blood mononuclear cells using five different xeno-free cryoprotective solutions. Cryobiology 2019;86:25–32. doi:10.1016/j.cryobiol.2019.01.004. [157] Germann A, Schulz JC, Kemp-Kamke B, Zimmermann H, von Briesen H. Standardized Serum-Free Cryomedia Maintain Peripheral Blood Mononuclear Cell Viability, Recovery, and Antigen-Specific T-Cell Response Compared to Fetal Calf Serum-Based Medium. Biopreserv Biobank 2011;9:229–36. doi:10.1089/bio.2010.0033. [158] Tijssen MR, Woelders H, de Vries-van Rossen A, van der Schoot CE, Voermans C, Lagerberg JWM. Improved postthaw viability and in vitro functionality of peripheral blood hematopoietic progenitor cells after cryopreservation with a theoretically optimized freezing curve. Transfusion 2008;48:893–901. doi:10.1111/j.1537-2995.2008.01650.x. [159] Berens C, Heine A, Müller J, Held SAE, Mayer K, Brossart P, et al. Variable resistance to freezing and thawing of CD34-positive stem cells and lymphocyte subpopulations in leukapheresis products. Cytotherapy 2016;18:1325–31. doi:10.1016/j.jcyt.2016.06.014. [160] Reddy MJ, Kumar DN. Multiobjective differential evolution with application to reservoir system optimization. J Comput Civ Eng 2007;21:136–46. doi:10.1061/(ASCE)0887-3801(2007)21:2(136). [161] Gujarathi AM, Babu BV. Multi-objective optimization of industrial styrene reactor: Adiabatic and pseudo-isothermal operation. Chem Eng Sci 2010;65:2009–26. doi:10.1016/j.ces.2009.11.041. [162] Oyama A, Liou M-S. Multiobjective Optimization of Rocket Engine Pumps Using Evolutionary Algorithm. J Propuls Power 2002;18:528–35. doi:10.2514/2.5993. [163] Deepak NR, Ray T, Boyce RR. Evolutionary Algorithm Shape Optimization of a

145

Hypersonic Flight Experiment Nose Cone. J Spacecr Rockets 2008;45:428–37. doi:10.2514/1.33826. [164] Mohammed AR, Coombes AGA, Perrie Y. Amino acids as cryoprotectants for liposomal delivery systems. Eur J Pharm Sci 2007;30:406–13. doi:10.1016/j.ejps.2007.01.001. [165] Souba WW, Pacitti AJ. Review: How Amino Acids Get Into Cells: Mechanisms, Models, Menus, and Mediators. J Parenter Enter Nutr 1992;16:569–78. doi:10.1177/0148607192016006569. [166] Hachtel JA, Huang J, Popovs I, Jansone-Popova S, Keum JK, Jakowski J, et al. Identification of site-specific isotopic labels by vibrational spectroscopy in the electron microscope. Science (80- ) 2019;363:525–8. doi:10.1126/science.aav5845. [167] Egorov A V, Lyubartsev AP, Laaksonen A. Molecular Dynamics Simulation Study of Glycerol–Water Liquid Mixtures. J Phys Chem B 2011;115:14572–81. doi:10.1021/jp208758r. [168] Cottone G, Giuffrida S, Ciccotti G, Cordone L. Molecular dynamics simulation of sucrose- and trehalose-coated carboxy-myoglobin. Proteins Struct Funct Bioinforma 2005;59:291–302. doi:10.1002/prot.20414. [169] Lerbret A, Affouard F, Bordat P, Hédoux A, Guinet Y, Descamps M. Molecular dynamics simulations of lysozyme in water/sugar solutions. Chem Phys 2008;345:267–74. doi:10.1016/j.chemphys.2007.09.011. [170] Bordat P, Lerbret A, Demaret J-P, Affouard F, Descamps M. Comparative study of trehalose, sucrose and maltose in water solutions by molecular modelling. Europhys Lett 2004;65:41–7. doi:10.1209/epl/i2003-10052-0. [171] Ekdawi-Sever NC, Conrad PB, de Pablo JJ. Molecular Simulation of Sucrose Solutions near the Glass Transition Temperature. J Phys Chem A 2001;105:734– 42. doi:10.1021/jp002722i. [172] Malajczuk CJ, Hughes ZE, Mancera RL. Molecular dynamics simulations of the interactions of DMSO, mono- and polyhydroxylated cryosolvents with a hydrated phospholipid bilayer. Biochim Biophys Acta - Biomembr 2013;1828:2041–55. doi:10.1016/j.bbamem.2013.05.010. [173] Hughes ZE, Mancera RL. Molecular Mechanism of the Synergistic Effects of Vitrification Solutions on the Stability of Phospholipid Bilayers. Biophys J 2014;106:2617–24. doi:10.1016/j.bpj.2014.05.012. [174] Sum AK, Faller R, De Pablo JJ. Molecular Simulation Study of Phospholipid Bilayers and Insights of the Interactions with Disaccharides. Biophys J 2003;85:2830–44. doi:10.1016/S0006-3495(03)74706-7. [175] Suck G, Linn YC, Tonn T. Natural Killer Cells for Therapy of Leukemia. Transfus Med Hemotherapy 2016;43:89–95. doi:10.1159/000445325. [176] Fujisaki H, Kakuda H, Shimasaki N, Imai C, Ma J, Lockey T, et al. Expansion of Highly Cytotoxic Human Natural Killer Cells for Cancer Cell Therapy. Cancer Res 2009;69:4010–7. doi:10.1158/0008-5472.CAN-08-3712. [177] Eguizabal C, Zenarruzabeitia O, Monge J, Santos S, Vesga MA, Maruri N, et al. Natural Killer Cells for Cancer Immunotherapy: Pluripotent Stem Cells-Derived NK Cells as an Immunotherapeutic Perspective. Front Immunol 2014;5:1–10. doi:10.3389/fimmu.2014.00439.

146

Appendix

The osmolarities of SGC and TGC were displayed in Figure A.1. Over a range of osmolarity from 200 to 1600mOsm/kg, there were little correlations between post-thaw recovery and osmolarity to both SGC and TGC as SGI discussed in Chapter 4. The post- thaw recoveries of Jurkat cells cryopreserved with TGI and TGC were presented in

Figure A.2 and Figure A.4, respectively. The statistical modeling of Jurkat cells cryopreserved with TGI and TGC was presented in Figure A.3 and Figure A.5, respectively. The comparisons of post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with SGI, SGC and TGC under cooling rate 1ºC/min were presented in

Figure A.6, Figure A.7 and Figure A.8, respectively. It showed that cell types had different freezing responses to the same DMSO-free cryoprotectants and cooling profile.

The post-thaw recoveries of MSCs were higher than Jurkat cells for all concentration levels of glycerol in SGI, SGC and TGI. The detailed mechanism is unclear, but the differences between MSC and Jurkat in terms of size, membrane properties, and cytoskeletons may affect the freezing responses. The molar ratio of sugar alcohol and sugars was proportional to the post-thaw recovery of Jurkat cells as shown in Figure A.9.

This also displayed that amino acids played an important role to improve the post-thaw recovery under the same molar ratio. For example, the post-thaw recovery varied from

30% to 90% under molar ratio was10 to SGI. The effects of amino acids could be observed in comparing the trends of post-thaw recoveries of Jurkat cells between SGI and

SGC as well as TGI and TGC.

147

(a) (b)

Figure A.1 Post-thaw recovery of Jurkat cryopreserved with (a) SGC and (b) TGC at 1ºC/min as a function of cryoprotectant osmolarities

148

(a) (b)

(e) (f)

149

(a) (b)

(c)

150

(a) (b)

(e) (f)

Figure A.4 Post-thaw recoveries of Jurkat cells cryopreserved at a cooling rate of 1C/min and plotted to show (a) the effect of trehalose with coloring by level of glycerol, (b) the effect of trehalose with coloring by level of creatine, (c) the effect of glycerol with coloring by level of trehalose, (d) the effect of glycerol with coloring by level of creatine, (e) the effect of creatine with coloring by level of trehalose, and (f) the effect of creatine with coloring by level of glycerol. Each solid line demonstrates the effect of the x-axis osmolyte on post-thaw recovery for fixed levels of the other two osmolytes. The dashed lines indicate the post-thaw recoveries for the single-component solutions. 151

(a) (b)

(c)

152

(a) (b)

(c)

Figure A.6 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with SGI for (a) G=0 and 1, (b) G=2 and 3, and (c) G=3 and 5 under cooling rate 1ºC/min (n=10).

153

(a) (b)

(c)

Figure A.7 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with SGC for (a) G=0 and 1, (b) G=2 and 3, and (c) G=4 and 5 under cooling rate 1ºC/min.

154

(a) (b)

(c)

Figure A.8 The post-thaw recoveries of Jurkat cells and mesenchymal stem cells (MSC) with TGC for (a) G=0 and 1, (b) G=2 and 3, and (c) G=4 and 5 under cooling rate 1ºC/min.

155

(a) (b)

Figure A.9 The post-thaw recoveries of Jurkat cells as a function of the molar ratio between sugar alcohol and sugar to (a) SGI, (b) SGC, (c) TGI, and (d) TGC

156