Cyclodextrin Solutions

University of Rhode Island DigitalCommons@URI

Open Access Master's Theses

2020

METHODS IN THE DETECTION OF ALIPHATIC ALCOHOLS AND ANABOLIC STEROIDS IN β‑CYCLODEXTRIN SOLUTIONS

Anna Zocchi Haynes University of Rhode Island, [email protected]

Follow this and additional works at: https://digitalcommons.uri.edu/theses

Recommended Citation Haynes, Anna Zocchi, "METHODS IN THE DETECTION OF ALIPHATIC ALCOHOLS AND ANABOLIC STEROIDS IN β‑CYCLODEXTRIN SOLUTIONS" (2020). Open Access Master's Theses. Paper 1862. https://digitalcommons.uri.edu/theses/1862

This Thesis is brought to you for free and open access by DigitalCommons@URI. It has been accepted for inclusion in Open Access Master's Theses by an authorized administrator of DigitalCommons@URI. For more information, please contact [email protected].

METHODS IN THE DETECTION OF ALIPHATIC ALCOHOLS AND ANABOLIC STEROIDS IN β‑CYCLODEXTRIN SOLUTIONS

BY ANNA ZOCCHI HAYNES

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN CHEMISTRY

UNIVERSITY OF RHODE ISLAND

2020

MASTER OF SCIENCE

ANNA ZOCCHI HAYNES

APPROVED:

Thesis Committee:

Major Professor Mindy Levine

William Euler

Geoffrey Bothun

Nasser H. Zawia Dean of the Graduate School

UNIVERSITY OF RHODE ISLAND

2020

ABSTRACT

The sensitive, selective, and practical detection of aliphatic alcohols is a continuing technical challenge with significant impact in public health work and environmental remediation efforts. Reported in the first manuscript is the use of a β-cyclodextrin derivative to promote proximity-induced interactions between aliphatic alcohol analytes and a brightly colored organic dye, which result in highly analyte-specific color changes that enabled accurate alcohol identification. Linear discriminant analysis of the color changes enabled resulting in 100% differentiation of the colorimetric signals obtained from methanol, ethanol, and isopropanol in combination with BODIPY and Rhodamine dyes.

The resulting solution-state detection system has significant broad-based applicability because it uses only easily available materials to achieve such detection, with moderate limits of detection obtained. Future research with this sensor system will focus on decreasing limits of detection as well as on optimizing the system for quantitative detection applications.

Reported in Chapter 2 of this thesis are our efforts towards the development of similar cyclodextrin-based systems for the detection of anabolic steroids that are of interest in illegal doping scandals. These systems, in close analogy to other systems developed in our group, rely on the fact that combining the target analytes with a high quantum yield fluorophore results in a measurable, analyte-specific change in the fluorophore emission signal. Promising results were seen in our solution-state detection of anabolic steroids via the use of β-cyclodextrin derivatives as supramolecular scaffolds and Rhodamine 6G as the signal transducing element. Using linear discriminant analysis, arrays were generated, illustrating the differentiation and classification of the analytes with high selectivity in the system. Limits of detection falling below the micromolar range show a high sensitivity of

the system. Current efforts are focused on improving the reproducibility of these experiments to enable a high functioning detection system as well as efforts toward moving to a solid-state detection system.

ACKNOWLEGMENTS

I would first like to thank the University of Rhode Island for granting me the opportunity to follow my passion of chemistry into graduate school. I would specifically like to thank the faculty members of the chemistry department who have never failed to point me in the right direction toward my success and showing me endless possibilities for my education and career. To Dr. Mindy Levine, thank you for finding a space in such a great group, advising me from Israel and allowing me to take my ideas and run with them. I have become to trust my own thoughts on what my science is illustrating which has shaped and guided the research I have done these past two years. Thank you to all of the past members of the Levine group that have offered support and guidance no matter where they have found themselves. A special thank you to Dr. Teresa Mako and Joan Racicot for answering all of my never-ending questions, giving me the confidence to step out of my comfort zone and never failing to show their support for whatever my next step is. To all the friends that

I have made throughout my time at URI, thank you for reminding me to have fun every once in a while, even if it is just a trip to grab some ice cream or a Target run.

Lastly, I would like to thank my family. It is not an exaggeration to say that I would not have achieved this much without you. To my mum and dad, thank you for always picking up the phone to talk to me, even if you’re in the middle of making pasta. Thank you for showing me what true dedication and passion is and for never letting me forget that you’re proud of me. To Alex, thank you for always lending your ear to my smallest problems and giving me a hug when you know I need it most. To Heather, Paul and PJ, thank you for your constant reminder that I can do this and never failing to make me laugh (as long as I didn’t let the facts get in the way of the joke).

PREFACE

This thesis is presented in manuscript format according to the guidelines of the graduate school of the University of Rhode Island. Two manuscripts will be presented in this thesis.

Chapter 1 is published in ACS Omega with authors Anna Haynes, Priva Halpert, and Mindy

Levine. Chapter 2 is being submitted for publication to RSC Advances with authors Anna

Haynes and Mindy Levine.

TABLE OF CONTENTS

ABSTRACT ...... ii ACKNOWLEGMENTS ...... iv PREFACE ...... v TABLE OF CONTENTS ...... vi LIST OF TABLES ...... vii LIST OF FIGURES ...... x CHAPTER 1 ...... 1 ABSTRACT ...... 2 INTRODUCTION ...... 2 EXPERIMENTAL SECTION ...... 5 RESULTS AND DISCUSSION...... 9 CONCLUSIONS ...... 18 REFERENCES ...... 19 ELECTRONIC SUPPLEMENTARY INFORMATION ...... 23 CHAPTER 2 ...... 48 ABSTRACT ...... 49 INTRODUCTION ...... 49 EXPERIMENTAL SECTION ...... 51 RESULTS AND DISCUSSION...... 53 CONCLUSION ...... 64 ACKNOWLEDGEMENTS ...... 65 REFERENCES ...... 66 ELECTRONIC SUPPLEMENTARY INFORMATION ...... 70

LIST OF TABLES

Table 1: Concentration of cyclodextrins and dyes in solution.a ...... 6 Table 2: Cumulative proportions of total dispersion for each cyclodextrin with dye 4 and 5.a ...... 10 Table 3: Percent correct classification values obtained from Jackknifed Classification analysis of the arrays.a ...... 12 Table 5: LODs and LOQs of each alcohol with dyes 4 and 5 in the presence of 2-HP-β-CD...... 17 Table 6: LODs and LOQs of isopropanol with increased concentrations of dye 4 and 5 and 2-HP-β-CD...... 17 Table 7. Concentration of Dyes in Solution ...... 25 Table 8. Volume to Mass Conversions of Cyclodextrins and Final Concentrations ...... 25 Table 9. Amount of Alcohol Added in Each Solution ...... 29 Table 10. RGB Measures of Cyclodextrin-Analyte Combinations with Dyes 4 and 5 ... 34 Table 11. Standard Deviations of RGB Measures of All Cyclodextrin-Analyte Combinations with Dyes 4 and 5 ...... 35 Table 12. Percent Standard Deviations of RGB Measures of Cyclodextrin-Analyte Combinations with Dyes 4 and 5 ...... 36 Table 13. RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD Solution with Dyes 4 and 5 ...... 36 Table 14. Standard Deviations of RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD Solution with Dyes 4 and 5 ...... 37 Table 15. Percent Standard Deviations of RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD with Dyes 4 and 5 ...... 38 Table 16. Table of LODs and LOQs of Each Alcohol with Dyes 4 and 5 in the Presence of 2-HP-β-CD ...... 38 Table 17. Additions and Concentrations for Binding of Dye 4 in 2-HP-β-CD ...... 39 Table 18. Additions and Concentrations for Binding of Dye 5 in 2-HP-β-CD ...... 39 Table 19. Analytes with β-CD and Dye 4 ...... 39 Table 20. Analytes with Me-β-CD and Dye 4 ...... 39 Table 21. Analytes with 2-HP-β-CD and Dye 4 ...... 40 Table 22. Analytes with β-CD and Dye 5 ...... 40 Table 23. Analytes with Me-β-CD and Dye 5 ...... 40 Table 24. Analytes with 2-HP-β-CD and Dye 5 ...... 40 Table 25. 0.5 M Analyte solution in 2-HP-β-CD with Dye 4 ...... 40 vii

Table 26. 1.0 M Analyte solution in 2-HP-β-CD with Dye 4 ...... 40 Table 27. 2.0 M Analyte solution in 2-HP-β-CD with Dye 4 ...... 40 Table 28. 3.0 M Analyte solution in 2-HP-β-CD with Dye 4 ...... 40 Table 29. 0.5 M Analyte solution in 2-HP-β-CD with Dye 5 ...... 41 Table 30. 1.0 M Analyte solution in 2-HP-β-CD with Dye 5 ...... 41 Table 31. 2.0 M Analyte solution in 2-HP-β-CD with Dye 5 ...... 41 Table 32. 3.0 M Analyte solution in 2-HP-β-CD with Dye 5 ...... 41 Table 33. Analytes with Dye 4 in the absence of cyclodextrin ...... 41 Table 34. Analytes with Dye 5 in the absence of cyclodextrin ...... 41 Table 34. Fluorescence ratios obtained for the addition of analytes 1-5 with various cyclodextrins in the presence of fluorophore 6a ...... 55 Table 35. Percent (%) difference between the fluorescence emission without analyte and the fluorescence emission after the addition of analytea ...... 57 Table 36. Quantitative values calculated from the electrostatic potential maps of the steroids in Spartan 18’ ...... 59 Table 37. Limits of detection (µM) calculated for analytes 1-5 in the cyclodextrin host systemsa ...... 64 Table 38: Concentration of analytes and fluorophore in ...... 72 solution prior to dilution via sample preparation ...... 72 Table 39: Final concentrations (µM) of analytes and fluorophore ...... 74 in fluorescence modulation trials ...... 74 Table 40. Fluorescence modulation results obtained for analytes 1-5 with various cyclodextrins in the presence of fluorophore 6a ...... 78 Table 41. Summary table for limits of detection experiments with β-CDa ...... Error! Bookmark not defined. Table 42. Summary table for limits of detection experiments with Me-β-CDa ...... Error! Bookmark not defined. Table 43. Summary table for limits of detection experiments with 2-HPCDa ...... Error! Bookmark not defined. Table 45. Linear discriminant analysis results using β-CD and Me-β-CD as predictors . 80 Table 46. Linear discriminant analysis results using β-CD and 2-HPCD as predictors .. 80 Table 47. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 81 Table 48. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 81

viii

Table 49. Linear discriminant analysis results using β-CD and Me-β-CD as predictors . 81 Table 50. Linear discriminant analysis results using β-CD and 2-HPCD as predictors .. 82 Table 51. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 82 Table 52. Linear discriminant analysis results using β-CD, Me-β-CD, and 2-HPCD as predictors...... 82 Table 53. Linear discriminant analysis results using β-CD and Me-β-CD as predictors . 83 Table 54. Linear discriminant analysis results using β-CD and 2-HPCD as predictors .. 83 Table 55. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 83 Table 56. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 84 Table 57. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 84

LIST OF FIGURES

Figure 1: Structure of alcohol analytes (1-3) and highly colored dyes (4-5) ...... 5 Figure 2. Generated arrays for the detections of ethanol, isopropanol, and methanol with each cyclodextrin supramolecular host using dye 4: (A) β-cyclodextrin; (B) Methyl-β- cyclodextrin; and (C) 2-hydroxypropyl-β-cyclodextrin ...... 10 Figure 3. Generated arrays for the detections of ethanol, isopropanol, and methanol with each cyclodextrin supramolecular host using dye 5: (A) β-cyclodextrin; (B) Methyl-β- cyclodextrin; and (C) 2-hydroxypropyl-β-cyclodextrin ...... 10 Figure 4. Electrostatic potential maps of (A) analyte 1; (B) analyte 2; and (C) analyte 3. Red areas indicate regions of negative electron density and blue areas indicate regions of positive electron density. These computations were done using Spartan ‘18 software. ... 11 Figure 5. Linear discriminant analysis results obtained at 3.0 M analyte concentration for: (A) dye 4 and (B) dye 5. All results were obtained using SYSTAT version 13 and following the procedures detailed in the experimental section...... 13 Figure 6. Linear discriminant analysis results generated at lower analyte concentrations. (A) Dye 4 at 0.5 M concentration analyte; (B) Dye 4 at 1.0 M concentration analyte; (C) Dye 4 at 2.0 M concentration analyte; (D) Dye 5 at 0.5 M concentration analyte; (E) Dye 5 at 1.0 M concentration analyte; and (F) Dye 5 at 2.0 M concentration analyte. All results were obtained using Systat version 13 and following the procedures detailed in the experimental section...... 13 Figure 7: Lowest energy conformations of BODIPY (4) and Rhodamine (5) in 2-HP-β- CD. (A) Side view of the complex with BODIPY (4). (B) Aerial view of the complex with BODIPY (4). (C) Side view of the complex with Rhodamine (5). (D) Aerial view of the complex with Rhodamine (5). Color coding: for the cyclodextrin host, the dark blue color represents the carbon atoms, the red color represents the oxygen atoms, and the gray color represents hydrogen atoms. For BODIPY, the purple color represents carbon atoms, gray represents hydrogen, blue represents nitrogen, orange represents boron, and green represents fluorine. For Rhodamine, the teal color represents carbon, gray represents hydrogen, red represents oxygen, and dark blue represents nitrogen...... 15 Figure 8: Lowest energy conformations of BODIPY (4) in 2-HP-β-CD following the introduction of methanol. The methanol is modeled as teal stick figures, and the colors of the BODIPY and cyclodextrin are identical to the colors used in Figure 7...... 16 Figure 9. Colorimetric array of the 60 samples from the trial using dye 4 and analyte 3. The concentration of analyte increases from left to right...... 17 Figure 10: Structure of alcohol analytes and highly colored dyes ...... 24 Figure 11. Annotated dimensions of lightbox ...... 26 Figure 12. Annotated top-view of lightbox...... 26 Figure 13. Illuminated lightbox ...... 26

Figure 14. Illuminated lightbox...... 26 Figure 15. Cropping sample photos to 500x500 pixel ratio using software on https://www.birme.net...... 27 Figure 16. Processing of cropped sample photo using the ImageJ software RGB Measurement plug-in...... 28 Figure 17. Spartan dialogue box with the settings used for calculation of equilibrium geometry...... 33 Figure 18. MOE 2018 dialogue box with the settings used for the docking calculations...... 34 Figure 19. Analytes with β-CD and Dye 4 ...... 42 Figure 20. Analytes with Me-β-CD and Dye 4 ...... 42 Figure 21. Analytes with 2-HP-β-CD and Dye 4 ...... 42 Figure 22. Analytes with β-CD and Dye 5 ...... 42 Figure 23. Analytes with Me-β-CD and Dye 5 ...... 42 Figure 24. Analytes with 2-HP-β-CD and Dye 5 ...... 42 Figure 25. 0.5 M Analytes in 2-HP-β-CD with Dye 4 ...... 43 Figure 26. 1.0 M Analytes in 2-HP-β-CD with Dye 4 ...... 43 Figure 27. 2.0 M Analytes in 2-HP-β-CD with Dye 4 ...... 43 Figure 28. 3.0 M Analytes in 2-HP-β-CD with Dye 4 ...... 43 Figure 29. 0.5 M Analytes in 2-HP-β-CD with Dye 5 ...... 43 Figure 30. 1.0 M Analytes in 2-HP-β-CD with Dye 5 ...... 43 Figure 31. 2.0 M Analytes in 2-HP-β-CD with Dye 5 ...... 43 Figure 32. 3.0 M Analytes in 2-HP-β-CD with Dye 5 ...... 43 Figure 33. Analytes with Dye 4 in the absence of cyclodextrin ...... 44 Figure 34. Analytes with Dye 5 in the absence of cyclodextrin ...... 44 Figure 35. Analyte 1 – Dye 4 ...... 45 Figure 36. Analyte 2 – Dye 4 ...... 45 Figure 37. Analyte 3 – Dye 4 ...... 46 Figure 38. Analyte 1 – Dye 5 ...... 46 Figure 39. Analyte 2 – Dye 5 ...... 47 Figure 40. Analyte 3 – Dye 5 ...... 47 Figure 41: Structure of anabolic analytes [Mesterolone (compound 1); Oxandrolone (compound 2); Oxymetholone (compound 3); Stanozolol (compound 4); and Trenbolone (compound 5)] and Rhodamine 6G fluorophore (compound 6) ...... 54

Figure 42. Fluorescence modulation of fluorophore 6 in the presence of all hosts induced by 20 µL additions of (A) analyte 1 (B) analyte 2 (C) analyte 3 (D) analyte 4 and (E) analyte 5. Curves were normalized so that the highest fluorescence intensity for each panel was set to 1.0...... 56 Figure 43. The absolute value of the percent difference values from Table 2, grouped by (A) the various cyclodextrin hosts and (B) the analytes (compounds 1-5) in solution. .... 57 Figure 44. Spartan-calculated electrostatic potential maps of (A) analyte 1 (B) analyte 2 (C) analyte 3 (D) analyte 4 and (E) analyte 5. The areas in the dark blue shade corresponds to electron-deficient, non-polar regions of the model and scales to the red regions, corresponding to the electron-rich, polar regions of the molecule. Color code for the molecular models: dark grey: carbon (C), light grey: hydrogen (H), red: oxygen (O), and light purple: nitrogen (N)...... 59 Figure 45. Arrays generated using the three cyclodextrin hosts as the predictors for the (A) 5 µL addition trials, (B) 10 µL addition trials and (C) 20 µL addition trials...... 62 Figure 46. Arrays of fluorescence modulation data with three different addition volumes generated using the cyclodextrin hosts as predictors for all analytes (A) including THF as a control analyte; and (B) excluding THF as a control analyte ...... 63 Figure 47: Structure of anabolic analytes (compound 1: Mesterolone; compound 2: Oxandrolone; compound 3: Oxymetholone; compound 4: Stanozolol; compound 5: Trenbolone) and fluorophore Rhodamine 6G (compound 6) ...... 71 Figure 48. Fluorescence modulation of analyte 1 in the presence of β-CD ...... 85 Figure 49. Fluorescence modulation of analyte 2 in the presence of β-CD ...... 85 Figure 52. Fluorescence modulation of analyte 1 in the presence of Me-β-CD ...... 86 Figure 53. Fluorescence modulation of analyte 2 in the presence of Me-β-CD ...... 87 Figure 54. Fluorescence modulation of analyte 3 in the presence of Me-β-CD ...... 87 Figure 55. Fluorescence modulation of analyte 4 in the presence of Me-β-CD ...... 87 Figure 56. Fluorescence modulation of analyte 5 in the presence of Me-β-CD ...... 88 Figure 57. Fluorescence modulation of analyte 1 in the presence of 2-HPCD ...... 88 Figure 58. Fluorescence modulation of analyte 2 in the presence of 2-HPCD ...... 88 Figure 59. Fluorescence modulation of analyte 3 in the presence of 2-HPCD ...... 89 Figure 60. Fluorescence modulation of analyte 4 in the presence of 2-HPCD ...... 89 Figure 61. Fluorescence modulation of analyte 5 in the presence of 2-HPCD ...... 89 Figure 62. Fluorescence modulation of analyte 1 in the presence of no CD ...... 90 Figure 63. Fluorescence modulation of analyte 2 in the presence of no CD ...... 90 Figure 64. Fluorescence modulation of analyte 3 in the presence of no CD ...... 91 Figure 65. Fluorescence modulation of analyte 4 in the presence of no CD ...... 91

xii

Figure 66. Fluorescence modulation of analyte 5 in the presence of no CD ...... 91 Figure 67. Fluorescence modulation of analyte 1 in the presence of β-CD ...... 92 Figure 68. Fluorescence modulation of analyte 2 in the presence of β-CD ...... 92 Figure 69. Fluorescence modulation of analyte 3 in the presence of β-CD ...... 92 Figure 70. Fluorescence modulation of analyte 4 in the presence of β-CD ...... 92 Figure 71. Fluorescence modulation of analyte 5 in the presence of β-CD ...... 93 Figure 72. Fluorescence modulation of analyte 1 in the presence of Me-β-CD ...... 93 Figure 73. Fluorescence modulation of analyte 2 in the presence of Me-β-CD ...... 93 Figure 74. Fluorescence modulation of analyte 3 in the presence of Me-β-CD ...... 94 Figure 75. Fluorescence modulation of analyte 4 in the presence of Me-β-CD ...... 94 Figure 76. Fluorescence modulation of analyte 5 in the presence of Me-β-CD ...... 94 Figure 77. Fluorescence modulation of analyte 1 in the presence of 2-HPCD ...... 95 Figure 78. Fluorescence modulation of analyte 2 in the presence of 2-HPCD ...... 95 Figure 79. Fluorescence modulation of analyte 3 in the presence of 2-HPCD ...... 95 Figure 80. Fluorescence modulation of analyte 4 in the presence of 2-HPCD ...... 96 Figure 81. Fluorescence modulation of analyte 5 in the presence of 2-HPCD ...... 96 Figure 82. Fluorescence modulation of analyte 1 in the presence of no CD ...... 96 Figure 83. Fluorescence modulation of analyte 2 in the presence of no CD ...... 97 Figure 84. Fluorescence modulation of analyte 3 in the presence of no CD ...... 97 Figure 85. Fluorescence modulation of analyte 4 in the presence of no CD ...... 97 Figure 86. Fluorescence modulation of analyte 5 in the presence of no CD ...... 98 Figure 87. Fluorescence modulation of analyte 1 in the presence of β-CD ...... 98 Figure 88. Fluorescence modulation of analyte 2 in the presence of β-CD ...... 98 Figure 89. Fluorescence modulation of analyte 3 in the presence of β-CD ...... 99 Figure 90. Fluorescence modulation of analyte 4 in the presence of β-CD ...... 99 Figure 91. Fluorescence modulation of analyte 5 in the presence of β-CD ...... 99 Figure 92. Fluorescence modulation of analyte 1 in the presence of Me-β-CD ...... 100 Figure 93. Fluorescence modulation of analyte 2 in the presence of Me-β-CD ...... 100 Figure 94. Fluorescence modulation of analyte 3 in the presence of Me-β-CD ...... 100 Figure 107. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of THF (B) 10 µL of THF (C) 20 µL of THF in the presence of all hosts ...... 105 Figure 108. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 1 (B) 10 µL of analyte 1 (C) 20 µL of analyte 1 in the presence of all hosts ...... 105

xiii

Figure 109. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 2 (B) 10 µL of analyte 2 (C) 20 µL of analyte 2 in the presence of all hosts ...... 105 Figure 110. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 3 (B) 10 µL of analyte 3 (C) 20 µL of analyte 3 in the presence of all hosts ...... 106 Figure 111. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 4 (B) 10 µL of analyte 4 (C) 20 µL of analyte 4 in the presence of all hosts ...... 106 Figure 112. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 5 (B) 10 µL of analyte 5 (C) 20 µL of analyte 5 in the presence of all hosts ...... 106 Figure 113. LOD calibration curve of analyte 1 ...... 107 Figure 114. LOD calibration curve of analyte 2 ...... 107 Figure 115. LOD calibration curve of analyte 3 ...... 107 Figure 116. LOD calibration curve of analyte 4 ...... 108 Figure 117. LOD calibration curve of analyte 5 ...... 108 Figure 118. LOD calibration curve of analyte 1 ...... 108 Figure 119. LOD calibration curve of analyte 2 ...... 109 Figure 120. LOD calibration curve of analyte 3 ...... 109 Figure 121. LOD calibration curve of analyte 4 ...... 109 Figure 122. LOD calibration curve of analyte 5 ...... 110 Figure 123. LOD calibration curve of analyte 1 ...... 110 Figure 124. LOD calibration curve of analyte 2 ...... 110 Figure 125. LOD calibration curve of analyte 3 ...... 111 Figure 126. LOD calibration curve of analyte 4 ...... 111 Figure 127. LOD calibration curve of analyte 5 ...... 111 Figure 128. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 112 Figure 129. Linear discriminant analysis results using β-CD and Me-β-CD as predictors ...... 112 Figure 130. Linear discriminant analysis results using β-CD and 2-HPCD as predictors ...... 112 Figure 131. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 113 Figure 132. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 113 Figure 133. Linear discriminant analysis results using β-CD and Me-β-CD as predictors ...... 113

xiv

Figure 134. Linear discriminant analysis results using β-CD and 2-HPCD as predictors ...... 114 Figure 135. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 114 Figure 136. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 114 Figure 137. Linear discriminant analysis results using β-CD and Me-β-CD as predictors ...... 115 Figure 138. Linear discriminant analysis results using β-CD and 2-HPCD as predictors ...... 115 Figure 139. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors ...... 115 Figure 140. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 116 Figure 141. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors...... 116 Figure 142. Electrostatic potential map of analyte 1 in the gas phase at its most stable (i.e. “ground state” configuration) ...... 117 Figure 143. Electrostatic potential map of analyte 2 in the gas phase at its most stable (i.e. “ground state” configuration) ...... 117 Figure 144. Electrostatic potential map of analyte 3 in the gas phase at its most stable (i.e. “ground state” configuration) ...... 118 Figure 145. Electrostatic potential map of analyte 4 in the gas phase at its most stable (i.e. “ground state” configuration) ...... 118 Figure 146. Electrostatic potential map of analyte 5 in the gas phase at its most stable (i.e. “ground state” configuration) ...... 119 Figure 147. Stacked NMRs of (A) 2:1 analyte 1:β-cyclodextrin (B) 1:1 analyte 1:β- cyclodextrin (C) 1:2 analyte 1:β-cyclodextrin (D) analyte 1 and (E) β-cyclodextrin ..... 120 Figure 148. Stacked NMRs of (A) 2:1 analyte 2:β-cyclodextrin (B) 1:1 analyte 2:β- cyclodextrin (C) 1:2 analyte 2:β-cyclodextrin (D) analyte 2 and (E) β-cyclodextrin ..... 121 Figure 149. Stacked NMRs of (A) 2:1 analyte 3:β-cyclodextrin (B) 1:1 analyte 3:β- cyclodextrin (C) 1:2 analyte 3:β-cyclodextrin (D) analyte 3 and (E) β-cyclodextrin ..... 122 Figure 150. Stacked NMRs of (A) 2:1 analyte 4:β-cyclodextrin (B) 1:1 analyte 4:β- cyclodextrin (C) 1:2 analyte 4:β-cyclodextrin (D) analyte 4 and (E) β-cyclodextrin ..... 123 Figure 151. Stacked NMRs of (A) 2:1 analyte 5:β-cyclodextrin (B) 1:1 analyte 5:β- cyclodextrin (C) 1:2 analyte 5:β-cyclodextrin (D) analyte 5 and (E) β-cyclodextrin ..... 124

Figure 152. Stacked NMRs of (A) 2:1 analyte 1:β-cyclodextrin (B) 1:1 analyte 1:β- cyclodextrin (C) 1:2 analyte 1:β-cyclodextrin (D) analyte 1 and (E) β-cyclodextrin ..... 125 Figure 153. Stacked NMRs of (A) 2:1 analyte 2:β-cyclodextrin (B) 1:1 analyte 2:β- cyclodextrin (C) 1:2 analyte 2:β-cyclodextrin (D) analyte 2 and (E) β-cyclodextrin ..... 126 Figure 154. Stacked NMRs of (A) 2:1 analyte 3:β-cyclodextrin (B) 1:1 analyte 3:β- cyclodextrin (C) 1:2 analyte 3:β-cyclodextrin (D) analyte 3 and (E) β-cyclodextrin ..... 127 Figure 155. Stacked NMRs of (A) 2:1 analyte 4:β-cyclodextrin (B) 1:1 analyte 4:β- cyclodextrin (C) 1:2 analyte 4:β-cyclodextrin (D) analyte 4 and (E) β-cyclodextrin ..... 128 Figure 156. Stacked NMRs of (A) 2:1 analyte 5:β-cyclodextrin (B) 1:1 analyte 5:β- cyclodextrin (C) 1:2 analyte 5:β-cyclodextrin (D) analyte 5 and (E) β-cyclodextrin ..... 129

xvi

CHAPTER 1

Colorimetric Detection of Aliphatic Alcohols in

β-Cyclodextrin Solutions

Anna Haynes1; Priva Halpert2; Mindy Levine3

Published in ACS Omega in November of 2019

1 M.Sc. Student, Department of Chemistry, University of Rhode Island, 140 Flagg Road, Kingston, RI 02881. Email: [email protected] 2 High School Student, Stella K. Abraham High School for Girls, 291 Meadowview Ave, Hewlett, NY 11557. 3 Associate Professor, Department of Chemistry, Ari’el University, 65 Ramat HaGolan Street, Ari’el, Israel. Email: [email protected]; [email protected].

ABSTRACT

The sensitive, selective, and practical detection of aliphatic alcohols is a continuing technical challenge with significant impact in public health work and environmental remediation efforts. Reported herein is the use of a β-cyclodextrin derivative to promote proximity-induced interactions between aliphatic alcohol analytes and a brightly colored organic dye, which result in highly analyte-specific color changes that enabled accurate alcohol identification. Linear discriminant analysis of the color changes enabled resulting in 100% differentiation of the colorimetric signals obtained from methanol, ethanol, and isopropanol in combination with BODIPY and Rhodamine dyes. The resulting solution- state detection system has significant broad-based applicability because it uses only easily available materials to achieve such detection, with moderate limits of detection obtained.

Future research with this sensor system will focus on decreasing limits of detection as well as on optimizing the system for quantitative detection applications.

INTRODUCTION

Increased interest in using non-mass spectrometry-based techniques for the detection of small organic compounds has arisen due to practical challenges associated with the use of mass spectrometry that limit broad-based applicability.1–3 Such challenges include the fact that expensive, bulky instrumentation is often required in order to accomplish mass spectrometry-based detection combined with significant user training to operate the instrumentation, which prevents detection by relatively untrained citizen scientists.4 Many newly developed chemosensors have focused on systems that allow for portable, on-site testing of the target analytes, without requiring high-end, costly laboratory instrumentation.5,6 A challenging aspect of designing portable chemosensors is the need to

2 maintain high selectivity, sensitivity, and broad-based applicability in the efficient detection of various analytes, especially among analytes that might be similar in structure and size.

By utilizing the ability of cyclodextrin to act as a supramolecular scaffold that facilitates proximity-induced, highly analyte-specific interactions between an analyte of interest and a high quantum yield fluorophore, the Levine group has developed sensitive and selective fluorescence-based systems for toxicant detection.7–10 The systems utilize cyclodextrin- promoted fluorescence energy transfer from a toxicant to a high quantum yield fluorophore, for photophysically-active analytes,11 or cyclodextrin-promoted, toxicant-specific fluorescence modulation, for non-photophysically active analytes.8 In addition to monitoring the analyte-specific fluorescence changes, there are often analyte-specific color changes in the fluorophore, promoted through the cyclodextrin-assisted interaction of the brightly colored fluorophore and the target analyte.12 Advantages of colorimetric detection include the fact that the color changes can be easily detected using naked eye detection13 or RGB analysis.14 Significant literature precedent indicates that colorimetric analysis can be optimized for detection of very small concentrations of toxicants, both in solution-state as well as in solid-state detection devices.15–17

Colorimetric detection has potential utility in the detection of aliphatic alcohols, a class of analytes commonly found in commercial products that, in high concentrations, can cause health concerns.18–20 These alcohols, including isopropanol, ethanol, and methanol, are found in household cleaners,21 paints,22 self-care and beauty products,23 as well as in beverages.24 Moreover, the need for detection of aliphatic alcohols is rising with the increasing prevalence of at-home beer and alcohol production.25 With almost no regulation

3 of this process currently in place, there is significant potential for poorly regulated ethanol concentrations26 as well as the potential for methanol contamination and associated methanol toxicity.27 Furthermore, the brewing process can sometimes lead to the formation of other byproducts including n-propanol, isobutanol, and isoamyl alcohol.28 Because methods to detect these byproducts are not widely available, significant public health risks from their ingestion remain.29 Additional potential applications of colorimetric aliphatic alcohol detection include the use of a colorimetric device to detect alcohol intoxication in both medical30 and law-enforcement settings.31,32 Finally, forensic post-mortem analysis would benefit from the detection of a range of aliphatic alcohols that are byproducts of certain bacteria and could provide important information as in a forensic investigation. 31,33

While the fluorescence modulation method used previously in the Levine group provided good sensitivity and high selectivity among structurally similar analytes, it required laboratory-grade instrumentation, which severely limits widespread usage. Although portable fluorimetry has been accomplished using smartphone-based systems,15,34–38 these systems can be challenging for the user to implement, which means that portable colorimetric systems can have notable advantages. Reported herein is the development of an extremely practical colorimetric detection system for isopropanol, ethanol, and methanol, based on color changes in a dye-cyclodextrin complex upon addition of the aliphatic alcohol, with such color changes intimately dependent on the structure of each of the alcohols and its association with both the cyclodextrin scaffold and the colorimetric dye. This system is highly robust, with alcohol-induced color changes detectable even by a high school student) working with unpurified tap water solutions, and even in its optimized formulation uses no laboratory-grade instrumentation. Rather, the system uses a

4 spray-painted plastic box equipped with LED lights to facilitate consistent coloration and enable reproducible results.

EXPERIMENTAL SECTION

Figure 1: Structure of alcohol analytes (1-3) and highly colored dyes (4-5)

Materials and Methods: The alcohol analytes 1-3 and dyes 4 and 5 shown in Figure 1 were obtained from Millipore-Sigma chemical company and the cyclodextrins were obtained from Tokyo Chemical Industry (TCI) chemical company. All chemicals were used as received. All aqueous solutions were made in glass jars and transferred to 50 mL white, polypropylene cups that had previously been used in a Keurig machine and were washed thoroughly prior to usage. A plastic container (with dimensions 21 cm x 15 cm x 7 cm) was painted using Krylon Fusion Satin Black spray paint to limit ambient light, and a 1.5 cm x 1.5 cm hole was cut in the center of the lid to enable photography of the solution. An additional polypropylene cup previously used for a Keurig machine was positioned under the opening and secured to the bottom of the container with electrical tape. Two strips of

LED white light tape (purchased from The Home Depot) were placed on the interior of the container, on all sides of the container, to provide uniform sample illumination. An annotated figure of the lightbox can be found in the Electronic Supporting Information.

The cup that contained sample was placed into the secured cup prior to imaging. Photos of the solutions were taken from 2.0 cm above the top of the sample cup with a Samsung

Galaxy S8+ (model number: G950U) on manual mode with the following settings: ISO set to 100, aperture set to 1/350, macro focused (close-up focus), and the white balance set at

5500K. These settings were kept constant for all trials to avoid variation in color capture.

Images were processed with ImageJ software to measure the red, green and blue values

(RGB) of the solutions, following the procedures detailed below.

General procedure for making stock solutions: Three 250 mL solutions of β-cyclodextrin

(β-CD), methyl-β-cyclodextrin (Me-β-CD), and 2-hydroxypropyl-β-cyclodextrin (2-HP-β-

CD) cyclodextrin were made at relatively high concentrations (Table 1). The stock solutions for dyes 4 and 5 were made in isopropanol at concentrations of 3.80 mM and 2.08 mM respectively (1 mg/mL for each dye). Diluted dye solutions were prepared by adding

5.0 mL of the concentrated stock solutions to a 150 mL volumetric flask and diluting to the mark with water.

Table 1: Concentration of cyclodextrins and dyes in solution.a Solute Solvent Concentration (mM)

β-CD DI H2O 16.5

Me-β-CD DI H2O 9.69

2-HP-β-CD DI H2O 2.39 Dye 4 Isopropanol 3.82 Dye 5 Isopropanol 2.08 a The final solution concentrations were calculated based on the amounts of solute and solvent added. See text for more information.

General procedure for the optimization of supramolecular cyclodextrin host: In a glass sample jar, 10.00 mL of β-CD stock solution was combined with 10.00 mL of one of the diluted dye solutions. This mixture was manually shaken for one minute to ensure thorough

6 mixing. After mixing, 5.00 mL of alcohol was added. This mixture was transferred to the sample cup and placed in the lightbox. The cover was placed on and a photo was taken using the smartphone with the settings listed above. This procedure was repeated for Me-

β-CD and 2-HP-β-CD with both dyes and each of the three alcohols (18 total samples).

Four trials of each sample were completed with a calculated average standard deviation in red, green and blue values of 0.08%, 0.14%, and 1.99% respectively.

General procedure for the optimization of analyte concentration: Preparation of the cyclodextrin-dye solution was performed following the procedures detailed above, with 2-

HP-β-CD used as the host. A 0.5 M solution of the alcohol was made by adding alcohol to the cyclodextrin-dye solution in the glass jar, with additional samples tested for each alcohol at a variety of concentrations (0.5 M, 1.0 M, 2.0 M, and 3.0 M) using both dye 4 and 5 (8 samples, 3 trials each). These solutions were transferred to a sample cup, placed in the lightbox, and a photo was taken of every sample.

General procedure for calculating the limits of detection (LOD) and limits of quantification

(LOQ): The limit of detection (LOD), defined as the lowest concentration of the analyte that can be detected, was obtained using the calibration curve method, following procedures reported by Loock and co-workers.39 The limit of detection of the blank

(LODblank) is defined according to Equation 1, below:

LODblank = mblank + 3(SDblank) Eq. 1 where m is the average of the values obtained from the blank sample and SD is the standard deviation of those measurements. The limit of quantification (LOQ), the lowest concentration of analyte that can be quantified.40 The limit of quantification of the blank is defined according to Equation 2, below:

LOQblank = mblank + 10(SDblank) Eq. 2 where m is the average of the values obtained from the blank sample and SD is the standard deviation of those measurements.

The cyclodextrin-dye solution was prepared as mentioned above using only 2-HP-β-CD as a supramolecular host. This solution was transferred to the sample cup and placed in the spray-painted box. A photograph was taken in order to obtain the blank measurement of the solution, in the absence of any alcohol. Using a 20-200 μL Fisherbrand Elite micropipette, 100 μL of an alcohol was added and a picture was taken. These 100 μL additions continued until 6.0 mL of alcohol had been added to the solution. This process was repeated three times for each alcohol and in the presence of each dye. The RGB values of the solution were used to determine the level of detection of each alcohol in both dyes.

General procedure for obtaining RGB values: Photos were cropped to be the same

500x500 pixel ratio focused on the center of the sample (using https://www.birme.net) to ensure the area of the picture that was being measured was consistent across all samples.

These images were processed using the RBG measurement tool plug-in that is available for the ImageJ software. Figures of these procedures can be found in the Electronic

Supporting Information of this manuscript.

General procedure for conducting linear discriminant analyses: SYSTAT 13 statistical computing software was used to quantify the degree of separation of color change in the solutions using the following settings for linear discriminant analysis (LDA): (a) Classical

Discriminant Analysis; (b) Grouping Variable: Analytes (alcohols); (c) Predictors: Red,

Green, Blue; (d) Long-Range Statistics: Mahal.7

General procedure for computational modeling: Spartan ’18 was used to calculate the equilibrium values for the analytes in their ground-state electric potential surfaces using a semi-empirical PM3 model for each analyte. Molecular Operating Environment 2018

(MOE) was used to do the docking studies for each dye, alcohol analyte, and 2-HP-β-CD host. A general energy minimization was performed using the “quick prep” function on the software. For the docking studies, the set of atoms defined as the receptor was both 2-HP-

β-CD and the solvent so that the dye could move freely in the system. Placement was done using the Triangle Matches method with London Dispersion dG score in 30 poses.

Refinement was done using the Rigid Receptor method with GBVI/WSA dG score in 5 poses. This generated the docking of the dye-cyclodextrin complex with the lowest energy confirmation. Summary figures generated from these procedures can be found in the

Supporting information of this manuscript.

RESULTS AND DISCUSSION

Optimization of cyclodextrin

A variety of cyclodextrin hosts were screened, with the goal of determining which supramolecular host would provide maximum separation between the analyte-induced color changes, with such separation quantified as the “cumulative proportion of total dispersion.” An example of significant dispersion of analyte clusters is shown for 2-HP-β-

CD (Figure 2C), and a contrasting example with overlapping areas between clusters is shown for Me-β-CD (Figure 2B). Using linear discriminant analysis of the RGB data collected from the sample photos, it was determined that the 2-HP-β-CD had the highest dispersion using both dye 4 and 5 as color-changing elements, with cumulative proportion of total dispersion values of 1.000 and 0.998 respectively (Table 2). Similar trends in the

9 cyclodextrin host were seen with dye 5 (Figure 3), with 2-HP- β-CD showing the greatest dispersion (Figure 3C) compared to of β-CD and Me-β-CD (Figure 3A and 3B, respectively).

Figure 2. Generated arrays for the detections of ethanol, isopropanol, and methanol with each cyclodextrin supramolecular host using dye 4: (A) β-cyclodextrin; (B) Methyl-β- cyclodextrin; and (C) 2-hydroxypropyl-β-cyclodextrin

Figure 3. Generated arrays for the detections of ethanol, isopropanol, and methanol with each cyclodextrin supramolecular host using dye 5: (A) β-cyclodextrin; (B) Methyl-β- cyclodextrin; and (C) 2-hydroxypropyl-β-cyclodextrin

Table 2: Cumulative proportions of total dispersion for each cyclodextrin with dye 4 and 5.a

Dye β-CD Me-β-CD 2-HP-β-CD 4 0.915 0.986 1.000 5 0.778 0.749 0.998 a Values were generated after analysis using SYSTAT 13 LDA software.

The higher signal dispersion that was observed when using 2-HP-β-CD is likely due to strong binding between the dyes and 2-HP-β-CD, as well as the high flexibility and water solubility of 2-HP-β-CD compared to the other hosts investigated.41 High binding constants between the alcohol analytes and cyclodextrin hosts increase the strength of interactions between the analyte, host and dye, resulting in more sensitive analyte-induced signal changes, whereas greater flexibility and water solubility increase the availability of this host to participate in the desired interactions. 42–44

Further insight into the selectivity observed between the alcohol analytes was obtained from computational investigations. Electrostatic potential mapping of analytes 1-3, generated using Spartan ’18 software, showed significant similarities in the analyte structures, with areas of high polarity around the hydroxyl group (Figure 4). Differences between the analytes include noticeable size differences as well as a more concentrated region of positive electron density in analyte 1 compared to the other analytes, as shown by the dark blue color. Such differences contribute to differing binding affinities with the cyclodextrin, resulting in turn in high specificity in the analyte-induced color changes.

Figure 4. Electrostatic potential maps of (A) analyte 1; (B) analyte 2; and (C) analyte 3. Red areas indicate regions of negative electron density and blue areas indicate regions of positive electron density. These computations were done using Spartan ‘18 software.

Determining the optimal concentrations for testing

In order to determine the optimal analyte concentration to achieve high degrees of analyte- induced separation in the colorimetric responses, the dependence of the LDA plots on analyte concentration was explicitly investigated, and the results are summarized in Table

3. These results indicate that dye 4 was more effective in characterizing data at high analyte concentrations (3.0 M, Figure 5), whereas dye 5 provided more dispersed and accurate results between 0.5 M and 2.0 M concentrations (Figure 6). A plausible explanation for these observed results relates to the higher binding constant between dye 4 and 2-HP-β-

CD of 3.32x105 M-1 compared to 1.59 x 105 M-1 between dye 5 and the cyclodextrin. The lower binding constant of dye 5 allows for better detection at lower alcohol concentrations because the dye is being displaced easier. Because dye 4 binds with a slightly higher binding energy, a higher concentration of alcohol needs to be added to the system in order to displace the dye, causing an overall decrease in system performance.

Table 3: Percent correct classification values obtained from Jackknifed Classification analysis of the arrays.a Dye 0.5 M 1.0 M 2.0 M 3.0 M

4 56 22 67 89

5 67 100 100 78 a Values taken after linear discriminant analysis was conducted using SYSTAT version 13 software.

Figure 5. Linear discriminant analysis results obtained at 3.0 M analyte concentration for: (A) dye 4 and (B) dye 5. All results were obtained using SYSTAT version 13 and following the procedures detailed in the experimental section.

Figure 6. Linear discriminant analysis results generated at lower analyte concentrations. (A) Dye 4 at 0.5 M concentration analyte; (B) Dye 4 at 1.0 M concentration analyte; (C) Dye 4 at 2.0 M concentration analyte; (D) Dye 5 at 0.5 M concentration analyte; (E) Dye 5 at 1.0 M concentration analyte; and (F) Dye 5 at 2.0 M concentration analyte. All results were obtained using Systat version 13 and following the procedures detailed in the experimental section.

Additional computational studies conducted using MOE 2018 software provided important information about the lowest energy docking conformation of each dye with 2-HP-β-CD,

13 and results are shown in Figure 7. Of note, BODIPY (4) exhibited markedly more inclusion in the cyclodextrin host compared to Rhodamine (5), with a substantial portion of the dye remaining exposed to the solvent. This solvent-exposed area of Rhodamine has a greater ability to interact with the analyte in solution, resulting in more responsiveness at lower analyte concentrations compared to BODIPY.

This result implies that dye displacement by the alcohol may not be necessary to affect a color change if the alcohol and dye interact via the solvent-accessible portion. Such interactions are not dependent on the binding constant of the dyes in cyclodextrin and provide an additional mechanism by which the system can lead to analyte-specific color changes. Efforts to investigate the extent to which either or both mechanisms (i.e., dye displacement from the cavity and/or interactions between the dye and analyte through solvent exposed areas) are operative in this system are currently underway in our laboratory.

Moreover, the addition of the alcohol analyte to the solution of dye in cyclodextrin had measurable changes on the supramolecular complex. In particular, computational results indicated that adding methanol to a solution of BODIPY in 2-HP-β-CD resulted in the weakening of the association between BODIPY and 2-HP-β-CD and strengthening of the affinity of the BODIPY for the solvent (Figure 8). This result supports that colorimetric changes induced by the addition of the alcohol analyte are a result of decreased affinity of the dye for the hydrophobic cyclodextrin cavity.

Figure 7: Lowest energy conformations of BODIPY (4) and Rhodamine (5) in 2-HP-β- CD. (A) Side view of the complex with BODIPY (4). (B) Aerial view of the complex with BODIPY (4). (C) Side view of the complex with Rhodamine (5). (D) Aerial view of the complex with Rhodamine (5). Color coding: for the cyclodextrin host, the dark blue color represents the carbon atoms, the red color represents the oxygen atoms, and the gray color represents hydrogen atoms. For BODIPY, the purple color represents carbon atoms, gray represents hydrogen, blue represents nitrogen, orange represents boron, and green represents fluorine. For Rhodamine, the teal color represents carbon, gray represents hydrogen, red represents oxygen, and dark blue represents nitrogen.

Figure 8: Lowest energy conformations of BODIPY (4) in 2-HP-β-CD following the introduction of methanol. The methanol is modeled as teal stick figures, and the colors of the BODIPY and cyclodextrin are identical to the colors used in Figure 7.

Determining the LOD and LOQ for the analytes with dyes 4 and 5

In addition to measuring the ability of the system to differentiate between structurally similar analytes, the sensitivity of the system to low analyte concentrations was also investigated. These results are summarized in Table 4 (for 0.127mM BODIPY and

0.093mM Rhodamine) and Table 5 (for 0.382mM BODIPY and 0.280mM Rhodamine), and indicate that analyte concentrations as low as 0.2 M were detectable via this method

(for isopropanol using relatively high concentrations of Rhodamine). Compared to the

LODs reported in Table 3, there was no significant decrease in the LODs observed with

BODIPY. In contrast, the Rhodamine trial showed much better improvement at higher dye concentrations with a 48% decrease in the LOD and a decrease of 68% in the LOQ value.

These marked changes are in line with the higher solvent and analyte accessibility

16 displayed by Rhodamine (vide supra), and indicate substantial promise in the further optimization of sensitive alcohol sensors. An example of a color array that illustrates the visible color change of BODIPY in the presence of isopropanol is shown in Figure 9.

Table 5: LODs and LOQs of each alcohol with dyes 4 and 5 in the presence of 2-HP-β-CD.

Dye Alcohol LOD (M)a LOQ (M)b Isopropanol 0.319 0.805 4 Ethanol 0.491 1.74 Methanol 0.249 0.823 Isopropanol 0.386 1.05 5 Ethanol 0.216 0.442 Methanol 0.331 0.730 a Values calculated according to Equation 1 and the equation of the line of best fit for each dye- alcohol complex. b Values calculated according to Equation 2 and the equation of the line of best fit for each dye- alcohol complex.

Table 6: LODs and LOQs of isopropanol with increased concentrations of dye 4 and 5 and 2-HP-β-CD. Dye LOD (M) LOQ (M) 4 0.3170 (0.50%) a 0.7812 (3.0%) 5 0.2004 (48%) 0.3341(68%) a Number in parentheses represents the percent change, in all cases a decrease, from the LOD values obtained in Table 4 to the ones calculated using a higher concentration of dye in solution.

Figure 9. Colorimetric array of the 60 samples from the trial using dye 4 and analyte 3. The concentration of analyte increases from left to right.

CONCLUSIONS

Harnessing the highly specific complexation of small molecules guests inside supramolecular cyclodextrin hosts provides a fundamentally unique system for the detection of those guests. Reported herein is the application of such host-guest complexes for the colorimetric detection of alcohol, using highly practical, easily available materials to achieve excellent selectivity (100% differentiation) and moderate sensitivity (as low as

0.2 M). Computational experiments involving the cyclodextrin, analytes, and highly colored dyes are invoked to explain the underlying basis of this strong analyte specificity, as remarkably structurally similar analytes leading to noticeably different colorimetric read-out signals. Efforts to improve the sensitivity and broaden the scope of such detection are currently underway in our laboratory, and results of these and other investigations will be reported in due course.

REFERENCES (1) Yang, M.; Zhang, X.; Liang, Q.; Yang, B. Application of (LC/)MS/MS Precursor Ion Scan for Evaluating the Occurrence, Formation and Control of Polar Halogenated DBPs in Disinfected Waters: A Review. Water Research 2019, 158, 322–337. https://doi.org/10.1016/j.watres.2019.04.033. (2) Verma, A.; Ambatipudi, K. Challenges and Opportunities of Bovine Milk Analysis by Mass Spectrometry. Clinical Proteomics 2016, 13 (1). https://doi.org/10.1186/s12014-016-9110-4. (3) Mika, A.; Sledzinski, T.; Stepnowski, P. Current Progress of Lipid Analysis in Metabolic Diseases by Mass Spectrometry Methods. Current Medicinal Chemistry 2019, 26 (1), 60–103. https://doi.org/10.2174/0929867324666171003121127. (4) Háda, V.; Bagdi, A.; Bihari, Z.; Timári, S. B.; Fizil, Á.; Szántay, C. Recent Advancements, Challenges, and Practical Considerations in the Mass Spectrometry- Based Analytics of Protein Biotherapeutics: A Viewpoint from the Biosimilar Industry. Journal of Pharmaceutical and Biomedical Analysis 2018, 161, 214–238. https://doi.org/10.1016/j.jpba.2018.08.024. (5) Dutta, S. Point of Care Sensing and Biosensing Using Ambient Light Sensor of Smartphone: Critical Review. TrAC Trends in Analytical Chemistry 2019, 110, 393– 400. https://doi.org/10.1016/j.trac.2018.11.014. (6) Huang, X.; Xu, D.; Chen, J.; Liu, J.; Li, Y.; Song, J.; Ma, X.; Guo, J. Smartphone- Based Analytical Biosensors. The Analyst 2018, 143 (22), 5339–5351. https://doi.org/10.1039/C8AN01269E. (7) DiScenza, D. J.; Levine, M. Sensitive and Selective Detection of Alcohols via Fluorescence Modulation. Supramolecular Chemistry 2016, 28 (11–12), 881–891. https://doi.org/10.1080/10610278.2016.1140897. (8) DiScenza, D. J.; Levine, M. Selective Detection of Non-Aromatic Pesticides via Cyclodextrin-Promoted Fluorescence Modulation. New Journal of Chemistry 2016, 40 (1), 789–793. https://doi.org/10.1039/C5NJ02357B. (9) Serio, N.; Miller, K.; Levine, M. Efficient Detection of Polycyclic Aromatic Hydrocarbons and Polychlorinated Biphenyls via Three-Component Energy Transfer. Chemical Communications 2013, 49 (42), 4821. https://doi.org/10.1039/c3cc40534f. (10) Dana DiScenza; Ella Culton; Molly Verderame; Julie Lynch; Nicole Serio; Mindy Levine. Towards Rational Chemosensor Design through Improved Understanding of Experimental Parameter Variation and Tolerance in Cyclodextrin-Promoted Fluorescence Detection. Chemosensors 2017, 5 (4), 34. https://doi.org/10.3390/chemosensors5040034. (11) Serio, N.; Moyano, D. F.; Rotello, V. M.; Levine, M. Array-Based Detection of Persistent Organic Pollutants via Cyclodextrin Promoted Energy Transfer. Chemical Communications 2015, 51 (58), 11615–11618. https://doi.org/10.1039/C5CC04153H.

(12) Chaudhuri, S.; DiScenza, D. J.; Verderame, M.; Cromwell, B.; Levine, M. Colorimetric Detection of Polycyclic Aromatic Hydrocarbons Using Supramolecular Cyclodextrin Dimer-Squaraine Constructs. Supramolecular Chemistry 2019, 31 (3), 211–219. https://doi.org/10.1080/10610278.2019.1579332. (13) Ma, X.; He, S.; Qiu, B.; Luo, F.; Guo, L.; Lin, Z. Noble Metal Nanoparticle-Based Multicolor Immunoassays: An Approach toward Visual Quantification of the Analytes with the Naked Eye. ACS Sensors 2019, 4 (4), 782–791. https://doi.org/10.1021/acssensors.9b00438. (14) Kangas, M. J.; Burks, R. M.; Atwater, J.; Lukowicz, R. M.; Williams, P.; Holmes, A. E. Colorimetric Sensor Arrays for the Detection and Identification of Chemical Weapons and Explosives. Critical Reviews in Analytical Chemistry 2017, 47 (2), 138–153. https://doi.org/10.1080/10408347.2016.1233805. (15) Li, Z.; Zhang, S.; Yu, T.; Dai, Z.; Wei, Q. Aptamer-Based Fluorescent Sensor Array for Multiplexed Detection of Cyanotoxins on a Smartphone. Analytical Chemistry 2019. https://doi.org/10.1021/acs.analchem.9b00750. (16) Li, Y.; Wang, Z.; Sun, L.; Liu, L.; Xu, C.; Kuang, H. Nanoparticle-Based Sensors for Food Contaminants. TrAC Trends in Analytical Chemistry 2019, 113, 74–83. https://doi.org/10.1016/j.trac.2019.01.012. (17) Zhou, Y.; Ding, L.; Wu, Y.; Huang, X.; Lai, W.; Xiong, Y. Emerging Strategies to Develop Sensitive AuNP-Based ICTS Nanosensors. TrAC Trends in Analytical Chemistry 2019, 112, 147–160. https://doi.org/10.1016/j.trac.2019.01.006. (18) Woźniak, M. K.; Wiergowski, M.; Namieśnik, J.; Biziuk, M. Biomarkers of Alcohol Consumption in Body Fluids - Possibilities and Limitations of Application in Toxicological Analysis. Current Medicinal Chemistry 2019, 26 (1), 177–196. https://doi.org/10.2174/0929867324666171005111911. (19) Slaughter, R. J.; Mason, R. W.; Beasley, D. M. G.; Vale, J. A.; Schep, L. J. Isopropanol Poisoning. Clinical Toxicology 2014, 52 (5), 470–478. https://doi.org/10.3109/15563650.2014.914527. (20) Grzybowski, A.; Saari, J.; Aghdam, K.; Wasinska-Borowiec, W. An Updated Review on the Most Common Agents Causing Toxic Optic Neuropathies. Current Pharmaceutical Design 2017, 23 (4), 586–595. https://doi.org/10.2174/1381612823666170124113826. (21) Casini, B.; Righi, A.; De Feo, N.; Totaro, M.; Giorgi, S.; Zezza, L.; Valentini, P.; Tagliaferri, E.; Costa, A.; Barnini, S.; et al. Improving Cleaning and Disinfection of High-Touch Surfaces in Intensive Care during Carbapenem-Resistant Acinetobacter Baumannii Endemo-Epidemic Situations. International Journal of Environmental Research and Public Health 2018, 15 (10), 2305. https://doi.org/10.3390/ijerph15102305. (22) Cheng, W.-H.; Huang, H.-L.; Chen, K.-S.; Chang, Y.-J. Quantification of VOC Emissions from Paint Spraying on a Construction Site Using Solid Phase

Microextraction Devices. Journal of Environmental Science and Health, Part A 2017, 52 (12), 1158–1163. https://doi.org/10.1080/10934529.2017.1356208. (23) Rogula-Kopiec, P.; Rogula-Kozłowska, W.; Pastuszka, J. S.; Mathews, B. Air Pollution of Beauty Salons by Cosmetics from the Analysis of Suspensed Particulate Matter. Environmental Chemistry Letters 2019, 17 (1), 551–558. https://doi.org/10.1007/s10311-018-0798-4. (24) Gorinstein, S.; Trakhtenberg, S. Alcohol Beverages and Biochemical Changes in Blood. Addiction Biology 2003, 8 (4), 445–454. https://doi.org/10.1080/13556210310001646420. (25) Pradhan, B.; Hadengue, A.; Chappuis, F.; Chaudhary, S.; Baral, D.; Gache, P.; Karki, P.; Rijal, S. Alcoholic Liver Disease in Nepal: Identifying Homemade Alcohol as a Culprit. Clinical and Experimental Gastroenterology 2015, 183. https://doi.org/10.2147/CEG.S81321. (26) Janagama, H. K.; Mai, T.; Han, S.; Nadala, L.; Nadala, C.; Samadpour, M. Dipstick Assay for Rapid Detection of Beer Spoilage Organisms. Journal of AOAC International 2018, 101 (6), 1913–1919. https://doi.org/10.5740/jaoacint.17-0479. (27) Teshome, D. A.; Rainer, M.; Noel, J.-C.; Fuchs, D.; Bliem, H. R.; Günther, B. K. Chemical Compositions of Traditional Alcoholic Beverages and Consumers’ Characteristics, Ethiopia. Afr. J. Food Sci. 12. (28) Barreto, S. G.; Saccone, G. T. P. Alcohol-Induced Acute Pancreatitis: The ‘Critical Mass’ Concept. Medical Hypotheses 2010, 75 (1), 73–76. https://doi.org/10.1016/j.mehy.2010.01.036. (29) Renger, R. S.; Hateren, S. H. van; Luyben, K. Ch. A. M. THE FORMATION OF ESTERS AND HIGHER ALCOHOLS DURING BREWERY FERMENTATION; THE EFFECT OF CARBON DIOXIDE PRESSURE. Journal of the Institute of Brewing 1992, 98 (6), 509–513. https://doi.org/10.1002/j.2050-0416.1992.tb01137.x. (30) Liu, L.; Chen, M.; Zhao, L.; Zhao, Q.; Hu, R.; Zhu, J.; Yan, R.; Dai, K. Ethanol Induces Platelet Apoptosis. Alcoholism: Clinical and Experimental Research 2017, 41 (2), 291–298. https://doi.org/10.1111/acer.13295. (31) Moriya, F. FORENSIC SCIENCES | Alcohol in Body Fluids. In Encyclopedia of Analytical Science; 2005; pp 358–365. https://doi.org/10.1016/B0-12-369397- 7/00196-5. (32) Lindberg, L.; Grubb, D.; Dencker, D.; Finnhult, M.; Olsson, S.-G. Detection of Mouth Alcohol during Breath Alcohol Analysis. Forensic Science International 2015, 249, 66–72. https://doi.org/10.1016/j.forsciint.2015.01.017. (33) Boumba, V. A.; Kourkoumelis, N.; Gousia, P.; Economou, V.; Papadopoulou, C.; Vougiouklakis, T. Modeling Microbial Ethanol Production by E. Coli under Aerobic/Anaerobic Conditions: Applicability to Real Postmortem Cases and to Postmortem Blood Derived Microbial Cultures. Forensic Science International 2013, 232 (1–3), 191–198. https://doi.org/10.1016/j.forsciint.2013.07.021.

(34) Wang, H.; Yang, L.; Chu, S.; Liu, B.; Zhang, Q.; Zou, L.; Yu, S.; Jiang, C. Semiquantitative Visual Detection of Lead Ions with a Smartphone via a Colorimetric Paper-Based Analytical Device. Analytical Chemistry 2019. https://doi.org/10.1021/acs.analchem.9b02297. (35) Li, W.; Shi, Y.; Hu, X.; Li, Z.; Huang, X.; Holmes, M.; Gong, Y.; Shi, J.; Zou, X. Visual Detection of Nitrite in Sausage Based on a Ratiometric Fluorescent System. Food Control 2019, 106, 106704. https://doi.org/10.1016/j.foodcont.2019.06.030. (36) Iqbal, Z.; Bjorklund, R. B. Assessment of a Mobile Phone for Use as a Spectroscopic Analytical Tool for Foods and Beverages: Mobile Phone for Use as a Spectroscopic Analytical Tool. International Journal of Food Science & Technology 2011, 46 (11), 2428–2436. https://doi.org/10.1111/j.1365-2621.2011.02766.x. (37) Hossain, M. A.; Canning, J.; Yu, Z.; Ast, S.; Rutledge, P. J.; Wong, J. K.-H.; Jamalipour, A.; Crossley, M. J. Time-Resolved and Temperature Tuneable Measurements of Fluorescent Intensity Using a Smartphone Fluorimeter. The Analyst 2017, 142 (11), 1953–1961. https://doi.org/10.1039/C7AN00535K. (38) Hossain, A.; Yen, T. L.; Jamalipour, A. Lab-in-a-Phone: Smartphone-Based Portable Fluorometer for PH Field Measurements of Environmental Water. 7. (39) Loock, H.-P.; Wentzell, P. D. Detection Limits of Chemical Sensors: Applications and Misapplications. Sensors and Actuators B: Chemical 2012, 173, 157–163. https://doi.org/10.1016/j.snb.2012.06.071. (40) Belter, M.; Sajnóg, A.; Barałkiewicz, D. Over a Century of Detection and Quantification Capabilities in Analytical Chemistry – Historical Overview and Trends. Talanta 2014, 129, 606–616. https://doi.org/10.1016/j.talanta.2014.05.018. (41) Liu, J.; Liu, G.; Liu, W.; Wang, Y.; Xu, M.; Wang, B. Turn-on Fluorometric β- Carotene Assay Based on Competitive Host-Guest Interaction between Rhodamine 6G and β-Carotene with a Graphene Oxide Functionalized with a β-Cyclodextrin- Modified Polyethyleneimine. Microchimica Acta 2016, 183 (3), 1161–1168. https://doi.org/10.1007/s00604-016-1747-1. (42) Maitre, M. M.; Longhi, M. R.; Granero, G. G. Ternary Complexes of Flurbiprofen with HP-ß-CD and Ethanolamines Characterization and Transdermal Delivery. Drug Development and Industrial Pharmacy 2007, 33 (3), 311–326. https://doi.org/10.1080/03639040600901978. (43) Zhang, H.; Ge, C.; van der Spoel, D.; Feng, W.; Tan, T. Insight into the Structural Deformations of Beta-Cyclodextrin Caused by Alcohol Cosolvents and Guest Molecules. The Journal of Physical Chemistry B 2012, 116 (12), 3880–3889. https://doi.org/10.1021/jp300674d. (44) Boonyarattanakalin, K. Influence of Ethanol as Co-Solvent in Cyclodextrin Inclusion Complexation: A Molecular Dynamics Study. Scientia Pharmaceutica 2015, 83 (2), 387–399. https://doi.org/10.3797/scipharm.1412-08.

ELECTRONIC SUPPLEMENTARY INFORMATION

COLORIMETRIC DETECTION OF ALIPHATIC ALCOHOLS IN Β-CYCLODEXTRIN SOLUTIONS

Anna Haynes, Priva Halpert and Mindy Levine

MATERIALS AND METHODS

The alcohol analytes and dyes were obtained from Millipore Sigma chemical company and the cyclodextrins were obtained from Tokyo Chemical Industry chemical company, and all chemicals were used as received. Samples were illuminated using the homemade lightbox detailed on the following page. All photos were taken with a Samsung Galaxy S8+ (model number: G950U) on manual mode with the following settings: ISO set to 100, aperture set to 1/350, macro focused (close-up focus), and the white balance set at 5500K. Images were cropped using software from https://www.birme.net. Fluorescence measurements for the binding experiments were performed using a Shimadzu RF 6000 spectrophotometer. Both the excitation and emission slit widths were set to 3.0 nm. All fluorescence spectra were integrated vs. wavenumber on the X-axis using OriginPro 2019 Version 9.60. All arrays were generated using SYSTAT Version 13.1.

DETAILS OF ANALYTES AND DYES

Figure 10: Structure of alcohol analytes and highly colored dyes All analytes were used as received. The dyes were prepared at a concentration of 1.0 mg/mL in isopropanol. Diluted dye solutions were prepared by diluting 5.0 mL of the concentrated stock solutions with 150 mL of DI H2O, based on the dilution factor used by

24 high school student Priva Halpert. The final concentrations of the dyes are shown in the table below.

Table 7. Concentration of Dyes in Solution

Concentration before Final Concentration Dye Number Dilution (mM) (mM) 4 3.796 0.1265 5 2.079 0.0693

EXPERIMENTAL DETAILS FOR CYCLODEXTRIN SOLUTIONS

The method for making the cyclodextrin solutions was taken from the previous work of high school student Priva Halpert working under the supervision of Dr. Levine. As a high school student working remotely, Halpert based her measurements on volume of each component, using teaspoons, cups, etc. In her procedure, 0.5 teaspoon of cyclodextrin was added to 0.25 cups (59.15 mL). We wanted to replicate her procedure and converted her measurement of 0.5 teaspoons to grams for each cyclodextrin and scaled up to make a solution with a final volume of 250 mL. These conversions are summarized in the table below, together with the final concentrations of the cyclodextrins.

Table 8. Volume to Mass Conversions of Cyclodextrins and Final Concentrations

Final Mass of 0.5 Mass added to 250 Cyclodextrin Concentration teaspoons (g) mL of DI H2O (g) (mM)

β-Cyclodextrin 1.470 4.6806 16.49

Methyl-β- 0.750 3.1717 9.685 Cyclodextrin 2-hydroxypropyl- 0.778 3.2852 2.389 β-CD

EXPERIMENTAL DETAILS FOR CONSTRUCTION OF THE LIGHTBOX

A plastic container (with dimensions 21 cm x 15 cm x 7 cm) was painted using Krylon

Fusion Satin Black spray paint to limit ambient light, and a 1.5 cm x 1.5 cm hole was cut in the center of the lid to enable photography of the solution. An additional polypropylene cup previously used for a Keurig machine was positioned under the opening after thoroughly washing, and secured to the bottom of the container with electrical tape. Two strips of LED white light tape (purchased from The Home Depot) were placed on the interior of the container, on all sides, and turned on to provide uniform sample illumination.

Figure 11. Annotated dimensions of lightbox Figure 12. Annotated top-view of lightbox

Figure 13. Illuminated lightbox Figure 14. Illuminated lightbox

EXPERIMENTAL DETAILS FOR CROPPING PHOTOGRAPHS AND RGB

MEASUREMENTS

All photographs were cropped using the online tool found at https://www.birme.net.

Photos from the same trial were uploaded to the site then cropped to a 500x500 pixel ratio (Figure 15). The cropped photos were then opened into the ImageJ software. These measurements were recorded using the RGB measurement plug-in provided in the software (Figure 16).

Figure 15. Cropping sample photos to 500x500 pixel ratio using software on https://www.birme.net

Figure 16. Processing of cropped sample photo using the ImageJ software RGB Measurement plug-in.

EXPERIMENTAL DETAILS FOR THE OPTIMIZATION OF THE

SUPRAMOLECULAR CYCLODEXTRIN HOST

In a glass sample jar, 10.0 mL of β-cyclodextrin stock solution was combined with 10.0 mL of one of the diluted dye solutions. This mixture was manually shaken for 1 minute to ensure a homogeneous mixture. After mixing, 5.0 mL of alcohol was added. This mixture was transferred to the sample cup and placed in the lightbox. The cover was placed on and a photo was taken using the smartphone with the settings detailed above. This procedure was repeated for methyl-β-cyclodextrin and 2-hydroxypropyl-β-cyclodextrin with both dyes and each of the three alcohols (18 total samples). These samples were replicated 4 times in total with an average standard deviation in RGB values of 0.70%.

Full summary tables of standard deviation and standard deviation percentages can be seen

28 on page S15. As a control, experiments were also conducted in the absence of cyclodextrin but under otherwise identical conditions. Summary tables and figures from these experiments are included herein.

EXPERIMENTAL DETAILS FOR THE OPTIMIZATION OF ANALYTE

CONCENTRATION

In a glass sample jar, 10.0 mL of 2-hydroxypropyl-β-cyclodextrin stock solution was combined with 10.0 mL of one of the diluted dye solutions. This mixture was manually shaken for 1 minute to ensure a homogeneous mixture. A 0.5 M solution of the alcohol was made by adding the corresponding amount of alcohol to the cyclodextrin-dye solution in the glass jar, with additional samples tested for each alcohol at a variety of concentrations

(0.5 M, 1.0 M, 2.0 M, and 3.0 M) using both BODIPY and Rhodamine dyes (8 samples replicated 3 times each). The volume of alcohol necessary to obtain an 0.5 M solution is seen in the table below. These solutions were transferred to a sample cup, placed in the lightbox, and a photo was taken of each.

Table 9. Amount of Alcohol Added in Each Solution

Cyclodextrin 0.5 M 1.0 M 2.0 M 3.0 M Isopropanol 0.790 mL 1.650 mL 3.590 mL 5.915 mL Ethanol 0.600 mL 1.240 mL 2.640 mL 4.240 mL Methanol 0.415 mL 0.845 mL 1.765 mL 2.770 mL

EXPERIMENTAL DETAILS FOR LIMIT OF DETECTION EXPERIMENTS

29 analyte that can be quantified. The limit of detection1 and quantification2 experiments were conducted following literature-reported procedures.

To determine the LOD and LOQ, each dye-analyte combination in 2-hydroxypropyl-β- cyclodextrin solution was examined in the following manner:

1. In a glass sample jar, 10.0 mL of 2-hydroxypropyl-β-cyclodextrin stock solution was combined with 10.0 mL of one of the diluted dye solutions. This mixture was manually shaken for 1 minute to ensure a homogeneous mixture.

2. The solution was transferred to a sample cup and subsequently placed in the light box.

A photo was taken in order to obtain the blank measurement of the solution (i.e. in the absence of any alcohol, before any analyte had been added).

3. Using a 20-200 μL Fisherbrand Elite micropipette, 100 μL of an alcohol was added and a picture was taken. These 100 μL additions continued until 6.0 mL of alcohol was in solution.

4. Steps 1-3 were repeated 3 times for each dye-analyte combination (6 combinations, 18 total samples).

5. Photos were cropped to a 500x500 pixel ratio centered on the center of the sample cup using software from https://www.birme.net and the RGB values of the photos were measured using the RGB measurement plug-in tool in the ImageJ software.

6. The Green values (Y-axis) were chosen to be plotted verses the molarity (X-axis), because they exhibited the most consistent trends. Calibration curves were generated, fitted with an exponential function and an equation was determined.

7. The limit of the blank is defined according to the following equation:

퐿푂퐷퐵푙푎푛푘 = 푚퐵푙푎푛푘 + 3(푆퐷퐵푙푎푛푘), Equation 1. where m is the average of the values obtained from the blank sample and SD is the standard deviation of those measurements.

8. The limit of the blank was entered as the y-value in the equation from step 6, and the corresponding x-value was calculated. This value was the LOD of the system in M.

9. The LOQ was determined in a similar procedure to the LOD. The limit of quantification blank is defined according to the following equation:

퐿푂푄퐵푙푎푛푘 = 푚퐵푙푎푛푘 + 10(푆퐷퐵푙푎푛푘). Equation 2.

This value is then inputted as the y-value in the equation from step 6, and the corresponding x-value was calculated. This value is the LOQ for the system in M.

EXPERIMENTAL DETAILS FOR BINDING CONSTANT EXPERIMENTS

Fluorescence measurements were performed on a Shimadzu RF 6000 spectrophotometer.

Both the excitation and emission slit widths were set to 3.0 nm.

A 1.45 mM solution of 2-HP-β-CD was prepared in DI H2O. A solution of 2.09 mM Dye

5 was prepared in DI H2O and a 2.54 mM solution of Dye 4 was prepared in tetrahydrofuran, with the solvent selection and concentration optimized based on concentration. In both binding experiments, 2.50 mL of water was added to quartz cuvette.

In the Dye 5 trial, 8 μL of the dye was added. In the Dye 4 series, 10.5 μL was added. The tip of the micropipette was used to stir the solution in the cuvette to ensure a homogeneous mixture. Fluorescence measurements of these were taken 4 times. In each trial, 1 μL of the

2-HP-β-CD solution was added to the cuvette and the same scanning process was completed. This was repeated for the total addition amounts seen on page S19 in Tables

S10 and S11, the total addition amount representing the concentration at which the observed signal plateaued. All fluorescence spectra were integrated vs. wavenumber on the

X-axis using OriginPro 2019 Version 9.60. Binding constants (Ka) were determined using the equation shown below:

퐹/퐹0 = 1 + ((퐹/퐹0)푚푎푥 − 1)(퐾푎[퐺]0/(1 + 퐾푎[퐺]0), Equation 3.

where F is the fluorescence of the sample, F0 is the fluorescence of the blank, [G] is the concentration of the guest which in this case is 2-HP-β-CD, and Ka is the binding constant.

5 -1 5 -1 The average Ka was determined to be 3.32 x 10 M and 1.59 x 10 M for Dye 4 and 5 respectively.

EXPERIMENTAL DETAILS FOR ARRAY GENERATION EXPERIMENTS

Array analysis was performed using SYSTAT 13 statistical computing software with the following settings:

(a) Classical Discriminant Analysis

(b) Grouping Variable: Analytes

(d) Long-Range Statistics: Mahal

Arrays were generated for all analyte-dye-cyclodextrin combinations in the optimization experiments.

EXPERIMENTAL DETAILS FOR COMPUTTIONAL MODELING

Spartan ’18 was used to calculate the equilibrium at ground state in gas using a semi- empirical PM3 model for each analyte (Figure 8). This allowed an electrostatic potential map surface to be overlaid on the molecules. Molecular Operating Environment 2018

(MOE) was used to do the docking studies for each dye and 2-HP-β-CD. A general energy minimization was performed using the “quick prep” function and the default settings. For the docking studies, the set of atoms defined as the receptor was both 2-HP-β-CD and the solvent so that the dye could move freely in the system. Placement was done using the

Triangle Matches method with London Dispersion dG score in 30 poses. Refinement was done using the Rigid Receptor method with GBVI/WSA dG score in 5 poses (Figure 9).

This generated the docking of the dye-cyclodextrin complex with the lowest energy confirmation.

Figure 17. Spartan dialogue box with the settings used for calculation of equilibrium geometry.

Figure 18. MOE 2018 dialogue box with the settings used for the docking calculations.

SUMMARY TABLES FOR OPTIMIZATION OF CYCLODEXTRIN

EXPERIEMNTS

Table 10. RGB Measures of Cyclodextrin-Analyte Combinations with Dyes 4 and 5 Dye 4 Dye 5 Cyclodextrin Analyte Red Green Blue Red Green Blue 229.326 162.309 2.674 248.446 125.19 1.782 229.686 162.166 2.711 248.196 125.264 1.796 Isopropanol 229.522 162.683 2.708 248.35 125.206 1.791 229.631 161.993 2.696 248.282 125.225 1.792 228.556 164.062 3.058 245.065 125.688 1.8 228.86 163.756 3.086 244.803 125.694 1.851 β-CD Ethanol 228.839 164.402 2.95 244.971 125.699 1.808 228.928 163.684 2.902 244.848 125.665 1.834 228.012 162.721 2.156 246.063 126.866 1.569 228.335 162.413 2.136 245.885 126.996 1.788 Methanol 228.145 162.91 2.119 246.043 126.965 1.625 228.218 162.214 2.182 245.953 126.986 1.729 229.38 159.869 2.965 248.249 125.159 3.753 229.672 159.58 2.986 248.513 124.875 3.746 M-β-CD Isopropanol 229.307 159.911 2.972 248.176 125.165 3.736 229.49 159.772 2.983 248.353 124.985 3.717

229.054 159.4 2.789 247.118 124.484 4.095 229.35 159.669 2.846 247.392 123.999 4.05 Ethanol 229.012 158.958 2.826 247.011 124.875 4.086 229.109 159.563 2.771 247.155 124.981 4.09 229.02 159.398 2.81 248.061 124.455 3.459 229.946 159.065 2.559 248.378 124.153 3.405 Methanol 228.449 159.348 2.787 247.963 124.484 3.614 229.372 159.128 3.001 248.14 124.334 3.573 236.315 185.085 4.147 258.453 165.344 2.221 236.88 185.564 4.111 257.953 164.814 2.176 Isopropanol 236.55 185.294 4.103 258.259 165.094 2.182 236.641 185.368 4.129 258.174 165.003 2.184 233.448 158.496 3.736 259.583 167.58 2.677 233.62 158.678 3.735 259.392 167.38 2.599 2-HP-β-CD Ethanol 233.931 158.977 3.676 259.093 167.151 2.604 234.134 159.18 3.704 258.895 167.015 2.654 238.767 155.928 3.401 257.994 160.909 2.843 238.893 156.037 3.379 257.877 160.841 2.961 Methanol 238.56 155.729 3.384 258.159 161.11 2.372 238.496 155.653 3.365 258.098 161.219 3.22

Table 11. Standard Deviations of RGB Measures of All Cyclodextrin-Analyte Combinations with Dyes 4 and 5 Dye 4 Dye 5 Cyclodextrin Analyte Red Green Blue Red Green Blue Isopropanol 0.1589 0.2935 0.0168 0.1058 0.0319 0.0059 β-CD Ethanol 0.1643 0.3279 0.0873 0.1190 0.0150 0.0235 Methanol 0.1353 0.3107 0.0271 0.0826 0.0596 0.0990 Isopropanol 0.1588 0.1473 0.0097 0.1462 0.1413 0.0156 Me-β-CD Ethanol 0.1511 0.3132 0.0341 0.1607 0.4452 0.0205 Methanol 0.6278 0.1630 0.1810 0.1771 0.1504 0.0973 Isopropanol 0.2336 0.1979 0.0196 0.2072 0.2202 0.0205 2-HP-β-CD Ethanol 0.3076 0.3048 0.0287 0.3063 0.2496 0.0382 Methanol 0.1837 0.1768 0.0149 0.1238 0.1752 0.2407

Table 12. Percent Standard Deviations of RGB Measures of Cyclodextrin-Analyte Combinations with Dyes 4 and 5

Dye 4 Dye 5 Cyclodextrin Analyte Red Green Blue Red Green Blue Isopropanol 0.069% 0.181% 0.623% 0.043% 0.025% 0.330% β-CD Ethanol 0.072% 0.200% 2.911% 0.049% 0.012% 1.290% Methanol 0.059% 0.191% 1.262% 0.034% 0.047% 5.900% Isopropanol 0.069% 0.092% 0.327% 0.059% 0.113% 0.418% Me-β-CD Ethanol 0.066% 0.196% 1.216% 0.065% 0.357% 0.502% Methanol 0.274% 0.102% 6.489% 0.071% 0.121% 2.769% Isopropanol 0.099% 0.107% 0.476% 0.080% 0.133% 0.934% 2-HP-β-CD Ethanol 0.132% 0.192% 0.772% 0.118% 0.149% 1.450% Methanol 0.077% 0.113% 0.439% 0.048% 0.109% 7.766%

SUMMARY TABLES FOR CONCETRATION OPTIMIZATION EXPERIEMNTS

Table 13. RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD Solution with Dyes 4 and 5 Dye 4 Dye 5 Analyte Concentration Red Green Blue Red Green Blue 238.111 147.305 2.193 254.733 156.79 1.866 0.5 (M) 238.416 146.03 2.177 254.733 156.705 1.839 238.381 145.778 2.097 254.706 156.172 1.781 238.845 150.321 2.277 254.754 159.842 2.522 1.0 (M) 239.22 148.409 2.258 254.769 159.741 2.529 239.194 148.797 2.261 254.773 158.386 2.243 Isopropanol 238.872 164.365 2.656 254.667 164.322 2.479 2.0 (M) 240.322 159.668 2.625 254.666 164.221 2.542 239.51 156.642 2.554 254.683 163.836 2.64 235.774 184.183 3.744 254.65 164.844 2.401 3.0 (M) 237.338 187.3 3.54 254.662 164.798 2.418 236.972 181.636 3.29 254.65 164.483 2.443 238.143 146.809 2.198 254.725 156.132 1.76 0.5 (M) 237.959 146.661 2.206 254.735 156.577 1.885 238.281 148.452 2.257 254.682 154.636 1.694 238.783 148.084 2.215 254.752 157.888 2.139 Ethanol 1.0 (M) 238.571 147.951 2.262 254.762 158.104 2.21 239.398 150.663 2.331 254.738 157.438 2.098 239.414 152.039 2.434 254.718 161.873 2.658 2.0 (M) 238.579 150.438 2.327 254.7 162.154 2.702

239.706 153.549 2.493 254.714 162.028 2.555 239.551 157.139 2.594 254.656 164.283 2.502 3.0 (M) 238.263 157.975 2.623 254.667 164.842 2.409 239.108 159.737 2.593 254.693 164.471 2.493 238.6 148.121 2.255 254.625 154.415 1.642 0.5 (M) 240.228 149.071 2.255 254.662 154.439 1.688 240.219 148.381 2.263 254.666 154.457 1.713 238.891 149.142 2.281 254.727 156.382 1.829 1.0 (M) 240.041 148.84 2.236 254.7 155.996 1.801 240.045 148.386 2.246 254.704 155.403 1.85 Methanol 239.694 152.425 2.403 254.751 159.123 2.361 2.0 (M) 240.747 151.426 2.355 254.753 158.592 2.332 240.915 150.847 2.356 254.758 158.342 2.218 239.049 154.9 2.53 254.711 161.973 2.679 3.0 (M) 240.847 154.191 2.51 254.726 161.574 2.642 241.502 153.745 2.5 254.708 162.272 2.586

Table 14. Standard Deviations of RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD Solution with Dyes 4 and 5 Dye 4 Dye 5 Concentratio Analyte Red Green Blue Red Green Blue n 0.5 (M) 0.1669 0.8186 0.0514 0.0156 0.3350 0.0434 Isopropano 1.0 (M) 0.2094 1.0107 0.0102 0.0100 0.8130 0.0047 l 2.0 (M) 0.7267 3.8915 0.0523 0.0095 0.2565 0.0811 3.0 (M) 0.8181 2.8368 0.2274 0.0069 0.1965 0.0211 0.5 (M) 0.1615 0.9941 0.0320 0.0282 1.0168 0.0970 1.0 (M) 0.4296 1.5288 0.0583 0.0121 0.3398 0.0567 Ethanol 2.0 (M) 0.5849 1.5557 0.0841 0.0095 0.1407 0.0754 3.0 (M) 0.6544 1.3262 0.0170 0.0190 0.2844 0.0513 0.5 (M) 0.9373 0.4910 0.0046 0.0226 0.0211 0.0360 1.0 (M) 0.6651 0.3805 0.0236 0.0146 0.4931 0.0246 Methanol 2.0 (M) 0.6618 0.7983 0.0274 0.0036 0.3988 0.0756 3.0 (M) 1.2701 0.5825 0.0153 0.0096 0.3502 0.0468

Table 15. Percent Standard Deviations of RGB Measures of Varying Concentrations of Analyte in 2-HP-β-CD with Dyes 4 and 5

Dye 4 Dye 5 Analyte Concentration Red Green Blue Red Green Blue 0.5 (M) 0.070% 0.559% 2.386% 0.006% 0.214% 2.375% 1.0 (M) 0.088% 0.678% 0.451% 0.004% 0.510% 0.187% Isopropanol 2.0 (M) 0.303% 2.429% 2.002% 0.004% 0.156% 3.177% 3.0 (M) 0.346% 1.539% 6.451% 0.003% 0.119% 0.873% 0.5 (M) 0.068% 0.675% 1.441% 0.011% 0.653% 5.451% 1.0 (M) 0.180% 1.027% 2.571% 0.005% 0.215% 2.637% Ethanol 2.0 (M) 0.244% 1.023% 3.480% 0.004% 0.087% 2.860% 3.0 (M) 0.274% 0.838% 0.655% 0.007% 0.173% 2.078% 0.5 (M) 0.391% 0.331% 0.205% 0.009% 0.014% 2.142% 1.0 (M) 0.278% 0.256% 1.048% 0.006% 0.316% 1.346% Methanol 2.0 (M) 0.275% 0.527% 1.157% 0.001% 0.251% 3.281% 3.0 (M) 0.528% 0.378% 0.608% 0.004% 0.216% 1.776%

SUMMARY TABLES OF LOD AND LOQ EXPERIMENTS Table 16. Table of LODs and LOQs of Each Alcohol with Dyes 4 and 5 in the Presence of 2-HP-β-CD

Dye Analyte LOD (M) LOQ (M) Equation R2 Isopropano 0.3187 0.8053 y = 3.304ex/1.163 + 141.606 0.9994 l Dye 4 Ethanol 0.4911 1.7387 y = 1.531ex/1.351 + 145.250 0.9973 Methanol 0.2492 0.8235 y = 8.048ex/4.913 + 139.175 0.9986 Isopropano 0.3856 1.0539 y = -14.641ex/-1.665 + 167.917 0.9799 l Dye 5 Ethanol 0.2163 0.4416 y = -21.511ex/-4.405 + 175.206 0.9935 Methanol 0.331 0.7304 y = -26.284ex/-7.820 + 179.275 0.9964

SUMMARY TABLES FOR BINDING EXPERIMENTS Total addition amounts of 2-HP-β-CD added into the cuvette with the dye solution.

Additions were stopped when the fluorescence measurements begin to plateau. Full experimental details can be seen on page S11.

Table 17. Additions and Concentrations Table 18. Additions and Concentrations for Binding of Dye 4 in 2-HP-β-CD for Binding of Dye 5 in 2-HP-β-CD Volume Volume of 2- of 2- HP-β- [2-HP-β- [Dye 4] HP-β- [2-HP-β- [Dye 5] CD CD] (μM) (μM) CD CD] (μM) (μM) added added (μL) (μL) 0 0.0000 6.383 0 0.0000 6.659 1 0.5790 6.380 1 0.5790 6.656 2 1.158 6.378 2 1.158 6.654 3 1.736 6.375 3 1.736 6.651 4 2.313 6.373 4 2.313 6.648 5 2.890 6.370 5 2.890 6.646 6 3.467 6.368 6 3.467 6.643 7 4.043 6.365 7 4.043 6.640 8 4.619 6.362 8 4.619 6.638 12 6.918 6.352 12 6.918 6.627 16 9.209 6.342 16 9.209 6.617 20 11.49 6.332 20 11.49 6.606 24 13.77 6.322 24 13.77 6.596 28 16.04 6.312 28 16.04 6.585 32 18.32 6.575 36 20.58 6.565

40 22.83 6.554

SUMMARY TABLES OF COLORIMETRIC ARRAY EXPERIMENTS Summary Tables of Cyclodextrin-Analyte Combinations with Dyes 4 and 5 Arrays Table 19. Analytes with β-CD and Dye 4 Table 20. Analytes with Me-β-CD and Dye 4

Table 21. Analytes with 2-HP-β-CD and Dye 4 Table 23. Analytes with Me-β-CD and Dye 5

Table 22. Analytes with β-CD and Dye 5 Table 24. Analytes with 2-HP-β-CD and Dye 5

Summary Tables for RGB Measures of Varying Concentrations of Analyte in 2-

Hydroxypropyl-β-Cyclodextrin Solution with Dyes 4 and 5 Arrays

Table 25. 0.5 M Analyte solution in 2- Table 27. 2.0 M Analyte solution in 2- HP-β-CD with Dye 4 HP-β-CD with Dye 4

Table 26. 1.0 M Analyte solution in 2- Table 28. 3.0 M Analyte solution in 2- HP-β-CD with Dye 4 HP-β-CD with Dye 4

Table 29. 0.5 M Analyte solution in 2- Table 31. 2.0 M Analyte solution in 2- HP-β-CD with Dye 5 HP-β-CD with Dye 5

Table 30. 1.0 M Analyte solution in 2- Table 32. 3.0 M Analyte solution in 2- HP-β-CD with Dye 5 HP-β-CD with Dye 5

SUMMARY TABLES FOR CONTROL EXPERIMENTS Table 33. Analytes with Dye 4 in the absence of cyclodextrin

Table 34. Analytes with Dye 5 in the absence of cyclodextrin

SUMMARY FIGURES COLORIMETRIC ARRAY EXPERIMENTS

Figure 19. Analytes with β-CD and Dye 4 Figure 22. Analytes with β-CD and Dye 5

Figure 20. Analytes with Me-β-CD and Dye 4 Figure 23. Analytes with Me-β-CD and Dye 5

Figure 21. Analytes with 2-HP-β-CD and Dye 4 Figure 24. Analytes with 2-HP-β-CD and Dye 5

Figure 25. 0.5 M Analytes in 2-HP-β-CD with Dye 4 Figure 29. 0.5 M Analytes in 2-HP-β-CD with Dye 5

Figure 26. 1.0 M Analytes in 2-HP-β-CD with Dye 4 Figure 30. 1.0 M Analytes in 2-HP-β-CD with Dye 5

Figure 27. 2.0 M Analytes in 2-HP-β-CD with Dye 4 Figure 31. 2.0 M Analytes in 2-HP-β-CD with Dye 5

Figure 28. 3.0 M Analytes in 2-HP-β-CD with Dye 4 Figure 32. 3.0 M Analytes in 2-HP-β-CD with Dye 5

SUMMARY FIGURES FOR CONTROL EXPERIMENTS

Figure 33. Analytes with Dye 4 in the absence of cyclodextrin

Figure 34. Analytes with Dye 5 in the absence of cyclodextrin

44 SUMMARY FIGURES FOR ALL LOD EXPERIMENTS Red lines on each graph are representative of the lines of best fit of the equations given on page S18 in Table S9.

Figure 35. Analyte 1 – Dye 4

Figure 36. Analyte 2 – Dye 4

Figure 37. Analyte 3 – Dye 4

Figure 38. Analyte 1 – Dye 5

Figure 39. Analyte 2 – Dye 5

Figure 40. Analyte 3 – Dye 5 REFERENCES (1) Loock, H.-P.; Wentzell, P. D. Detection Limits of Chemical Sensors: Applications and Misapplications. Sensors and Actuators B: Chemical 2012, 173, 157–163. https://doi.org/10.1016/j.snb.2012.06.071. (2) Belter, M.; Sajnóg, A.; Barałkiewicz, D. Over a Century of Detection and Quantification Capabilities in Analytical Chemistry – Historical Overview and Trends. Talanta 2014, 129, 606–616. https://doi.org/10.1016/j.talanta.2014.05.018.

47 CHAPTER 2

Detection of Anabolic Steroids Using Cyclodextrin

Promoted Fluorescence Modulation

Anna Haynes1; Mindy Levine2

in preparation for submission to RSC Advances

1 Masters of Science Student, Department of Chemistry, University of Rhode Island, 140 Flagg Road, Kingston, RI 02881. Email: [email protected] 2 Associate Professor, Department of Chemistry, Ari’el University, 65 Ramat HaGolan Street, Ari’el, Israel. Email: [email protected]; [email protected].

48 ABSTRACT

The detection of anabolic steroids, particularly those that are commonly used in illegal doping scandals, remains a high priority research objective with applications in public health, law enforcement, and a variety of sporting disciplines. Reported herein are our efforts to address this objective, through the use of cyclodextrin-promoted interactions between the analyte of interest and a high quantum yield fluorophore, which lead to measurable, analyte-specific changes in the fluorophore emission signal. By using a variety of β-cyclodextrin derivatives (unmodified β-cyclodextrin, methyl-β-cyclodextrin, and 2- hydroxypropyl-β-cyclodextrin) in combination with high quantum yield Rhodamine 6G as a fluorophore, we were able to detect a variety of anabolic steroid analytes with 100% differentiation between structurally similar analytes and micromolar level limits of detection. Overall, these results show significant potential in the development of practical, fluorescence-based steroid detection devices.

INTRODUCTION

The detection of steroids using sensitive, selective, and portable methods is of significant interest in a variety of contexts, particularly in athletic and sporting scenarios in which illegal anabolic steroid use has been reported.1 Such illegal use has been increasing in recent years at all levels, including in youth athletic programs, and the lack of effective methods for steroid detection means that the usage is likely to increase unless a rapid, sensitive, selective, and easy-to-use method is developed.2 Detection methods that currently exist suffer from a variety of drawbacks,3 including the need for expensive laboratory instrumentation and significant sample preparation prior to analysis, that make them impractical for on-site usage at sporting events.4 As such, a new method is needed.

49 One approach to a fundamentally new detection method has been developed by the Levine group in recent years, and relies on the use of cyclodextrin as a supramolecular scaffold to promote proximity-induced interactions between the analyte of interest and a high quantum yield fluorophore that leads to effective fluorescence detection.5 Such detection has been demonstrated for a variety of analytes, including polycyclic aromatic hydrocarbons,6 bisphenols,7 pesticides,8,9 and alcohols,10 and in a variety of contexts, including in the complex matrices of human breast milk,11 urine,7 and saliva.12 High detection sensitivity and selectivity has also been demonstrated, and is maintained even in the aforementioned complex biological matrices. The fact that the read-out signal of such sensors is a rapid and measurable change in the fluorescence emission signal means that such a system is easily translatable for on-site measurements. The use of such a fluorescence-based detection system for steroid detection has not been reported to date, despite the advantages of the system and the known challenges with effective steroid detection.

Reported herein is the cyclodextrin-promoted fluorescence detection of five anabolic steroids: mesterolone, oxandrolone, oxymesterolone, trenbolone, and stanozolol

(compounds 1-5, Figure 1). When these analytes are combined with cyclodextrin hosts and

Rhodamine 6G (compound 6), highly sensitive, analyte-specific changes in the fluorescence emission of the fluorophore results. Particularly promising results were seen using β-cyclodextrin (β-CD) and its derivatives, methyl-β-cyclodextrin (Me-β-CD) and 2- hydroxypropyl-β-cyclodextrin (2-HP-β-CD), as the supramolecular hosts with Rhodamine

6G as the fluorophore signaling element, with limits of detection as low as 0.05 μM obtained and 100% success in separating the signals obtained from the five steroid analytes using linear discriminant analysis. Overall, these results provide a promising proof-of-

50 concept for successful steroid detection using cyclodextrin-promoted fluorescence changes and indicate that such a system can eventually be used for on-site detection of illegal doping in competitive sporting scenarios.

EXPERIMENTAL SECTION

Materials and Methods. The anabolic steroid analytes, buffer chemicals, Rhodamine 6G, tetrahydrofuran (THF), and dimethyl sulfoxide (DMSO)-d6 were obtained from Sigma-

Aldrich Chemical company and the cyclodextrins were obtained from Tokyo Chemical

Industry. All chemicals were used as received without further purification. All fluorescence measurements were performed using a Shimadzu RF 6000 spectrophotometer. The excitation and emission slit widths were set to 3.0 nm. All fluorescence spectra were integrated vs. wavenumber on the X-axis using OriginPro 2019 Version 9.60. All arrays were generated using SYSTAT Version 13.1. NMR studies were conducted using a Bruker

400 MHz NMR and spectra were analyzed using MestReNova 14.1 software. Electrostatic potential map models were generated using Spartan ’18 software.

Fluorescence Modulation Experiments In six 15 mL glass vials, 100.0 µL of fluorophore

6 solution (0.1 mg/mL) in THF, 2.00 mL of a 10 mM cyclodextrin solution in citrate buffer and 0.400 mL 0.1 M citrate buffer were combined. These solutions were left to stabilize for 48 hours in a dark drawer and then transferred into a quartz cuvette. To the first cuvette,

5.00 µL of analyte 1 was added, in the second cuvette, 5.00 µL of analyte 2 was added, and so on until analyte 5 was added to a cuvette. 5.00 µL of THF was added to the last cuvette as a control. The solutions were excited at 490 nm and the fluorescence spectra were recorded from 500-800 nm. This was repeated for each cyclodextrin solution as well as in a cyclodextrin-free control, where citrate buffer was used instead of a cyclodextrin

51 solution. The entire procedure was then repeated using 10.00 µL and 20.00 µL additions of analyte instead of 5.00 µL additions. All captured fluorescence spectra were integrated vs. wavenumber on the X-axis. The fluorescence modulation of each analyte was determined using Equation 1, below:

Fluorophore Ratio = Flanalyte / Flblank (Eq. 1) where Flanalyte represents the integrated fluorescence emission of the fluorophore in the presence of the analyte, and Flblank represents the integrated fluorescence emission of the fluorophore in the absence of analyte. All trials were repeated four times and the reported modulation values represent the average of those repeated trials with the standard deviation values from those trials included as well.

Limit of Detection Experiments The limit of detection (LOD), defined as the lowest concentration of the analyte that can be detected, and the limit of quantification (LOQ), defined as the lowest concentration of analyte that can be reliably quantified were also calculated. The limit of detection and quantification experiments were obtained using the calibration curve method, following literature-reported procedures.13,14 Limit of detection and limit of quantification experiments were done with sequential 5 µL additions of analyte, to the same initial solution matrix described in the fluorescence modulation experiments (see ESI for more details).

Array Generation Experimental Details Arrays were generated using SYSTAT 13 statistical computing software with the following settings9: (a) Classical discriminant analysis, (b) Grouping variable: Analytes, (c) Predictors: β-cyclodextrin (β-CD), methyl-

β-cyclodextrin (Me-β-CD), and 2-hydroxypropyl-β-cyclodextrin (2-HP-β-CD), and (d)

Long-range statistics: Mahal. These experiments were then repeated using only two

52 predictors instead of all three, and the results of array-based analysis for each pair of predictors are also reported herein.

Computational Experiments Spartan ’18 was used to calculate the equilibrium values of the analytes in their ground-state electric potential surfaces using a semi-empirical PM3 model for each analyte.

1H NMR Titration Experiments All analytes and cyclodextrin solutions were prepared at a concentration of 10 mM in deuterated DMSO (dimethyl sulfoxide-d6). Trials with a total volume of 0.50 mL were prepared in Wilmad precision NMR tubes by combining varying ratios of analyte to the β-CD solution (see ESI for more details). A 1H NMR spectrum was taken of each using a Bruker 400 MHz NMR. These NMR spectra were analyzed for shifting and broadening of peaks using MestReNova software.

RESULTS AND DISCUSSION

Selection of Analytes and Fluorophore The selected anabolic steroids were selected because they are included in the lists of banned and misused substances in athletic competitions,15 and because of their significant structural similarities derived from their synthesis from testosterone.16 Fluorophore 6 was chosen due to its efficiency in fluorescence based detection systems, which has been proven and reported both by this group8,17 and others.18,19

Figure 41: Structure of anabolic analytes [Mesterolone (compound 1); Oxandrolone (compound 2); Oxymetholone (compound 3); Stanozolol (compound 4); and Trenbolone (compound 5)] and Rhodamine 6G fluorophore (compound 6)

Selection of Cyclodextrin Hosts Cyclodextrins were chosen as the supramolecular hosts in this study because of their known ability to interact with various guest molecules, including steroids,20–22 due to their ability to promote non-covalent intermolecular interactions including hydrogen bonding, hydrophobic interactions in the interior cavity, and van der

Waals interactions.23–26 β-cyclodextrin in particular has a well-documented ability to bind a variety of hydrophobic analytes, including steroids with significant structural similarity to analytes 1-5.10,27–30 α-Cyclodextrin and γ-cyclodextrin, by contrast, with average cavity diameters of 5.2 Å and 8.4 Å, respectively, were determined to have non-ideal size matches with analytes of average diameters of 5.6 Å.31 The two derivatives of β-cyclodextrin selected, methyl-β-cyclodextrin and 2-hydroxypropyl-β-cyclodextrin, have significant structural similarities to the unmodified analogue, but their notable differences in hydrophobicity, solubility, and steric accessibility means that the binding of analytes 1-5 in these derivatives is expected to differ from binding in the unmodified β-cyclodextrin host.32–37 As a result, the use of three different supramolecular hosts was expected to lead

54 to selectivity in array-based linear analysis, an expectation that was successfully borne out in experiments (vide infra).6,38

Fluorescence Modulation Experiments In order to show that each steroid is capable of inducing a measurable change in fluorescence emission that is unique for each host-steroid- fluorophore combination, fluorescence modulation experiments were performed with each analyte-cyclodextrin combination, together with control trials run in the absence of a cyclodextrin host. Small amounts of the steroids in THF were added to a solution of cyclodextrin and fluorophore 6 that had been left to stabilize for 48 hours. The fluorescence emission of the fluorophore was measured at an excitation wavelength of 490 nm and compared to the fluorescence emission spectra of the fluorophore after the addition of 5

µL, 10 µL, and 20 µL of the analytes. The degree of fluorescence modulation of the curves is the ratio of the fluorescence emission without analyte to the fluorescence emission of the fluorophore with analyte, calculated according to Equation 1. The results of the analyte- induced fluorescence modulation obtained after adding 20 µL of steroid solution are summarized in Table 1 and Figure 2, below.

Table 34. Fluorescence ratios obtained for the addition of analytes 1-5 with various cyclodextrins in the presence of fluorophore 6a

Analyte β-CD Me-β-CD 2-HPCD No CD 1 0.948 ± 0.001 0.932 ± 0.000 0.892 ± 0.000 0.994 ± 0.001 2 0.974 ± 0.001 0.966 ± 0.001 0.947 ± 0.000 0.987 ± 0.000 3 0.978 ± 0.001 0.954 ± 0.000 0.955 ± 0.000 1.004 ± 0.003 4 0.975 ± 0.000 0.986 ± 0.001 0.952 ± 0.000 0.972 ± 0.001 5 0.975 ± 0.001 0.969 ± 0.001 0.988 ± 0.001 1.036 ± 0.001 a All values were obtained after the addition of 20 µL of steroid solution. The results were calculated using Equation 1 and represent an average of at least four trials.

These results show that adding various steroid analytes to a cyclodextrin-fluorophore solution leads to measurable, analyte-specific fluorescence modulation. In the trials where cyclodextrin was present, each analyte induces a unique response, showing a key role for the cyclodextrin in enabling fluorescence modulation. In contrast, the cyclodextrin-free controls showed minimal fluorescence modulation, with less differences between the analytes observed. Moreover, in the vast majority of cases, the addition of steroid analytes caused a decrease in the observed fluorophore emission, represented by a fluorescence modulation value less than 1. This decrease is likely due to the fact that the steroids cause a displacement of the fluorophore from the cyclodextrin cavity, which increases the availability of non-radiative decay pathways and in turn leads to a decrease in fluorescence.

Analyte 5 in the absence of cyclodextrin represents a notable exception to this trend, and

56 may be a result of the reported fluorescence activity of analyte 5,39–42 which can interfere with the observed fluorophore emission signal.

Another way of viewing these analyte-specific changes is through calculating the percent differences induced by each analyte when compared to the blank (i.e. analyte-free sample) in the same host solution. These values were calculated according to Equation 2, below:

Percent (%) difference = (1 – Flratio) x 100% (Eq. 2) where Flratio is the value found in Table 1. The percent differences obtained for each analyte in each host-fluorophore solution were calculated, and the results are summarized in Table

2 and Figure 3, below.

Table 35. Percent (%) difference between the fluorescence emission without analyte and the fluorescence emission after the addition of analytea Analyte β-CD Me-β-CD 2-HP-β-CD No CD 1 5.235 6.834 10.76 0.566 2 2.555 3.391 5.265 1.346 3 2.157 4.639 4.529 -0.444b 4 2.481 1.362 4.752 2.759 5 2.542 3.114 1.233 -3.609b a Percent difference determined after the addition of 20 µL of analyte, using the fluorescence modulation ratios determined in Table 1 and entered into Equation 2 b Negative values represent a situation where the fluorescence signal with analyte was greater than the signal without analyte

Figure 43. The absolute value of the percent difference values from Table 2, grouped by (A) the various cyclodextrin hosts and (B) the analytes (compounds 1-5) in solution.

57 Figure 3A groups the percent difference values by hosts, allowing for better visualization of which cyclodextrin host facilitates the most modulation of fluorescence. Among the various hosts that were investigated, 2-HP-β-CD showed the greatest analyte-induced fluorescence modulation overall. This result is likely due to the high aqueous solubility of

2-HP-β-CD, which provides increased interaction between the cyclodextrin and the analyte and increased opportunities for analyte-induced fluorescence modulation. Variability between analyte-induced responses was also seen using Me-β-CD, which is likely due to the increased hydrophobicity of the cyclodextrin host that facilitates increased analyte- cyclodextrin interactions and concomitant variations in the observed modulation values. In contrast to the two substituted β-cyclodextrin hosts, unmodified β-cyclodextrin showed the least analyte-induced fluorescence modulation, which is likely due to a combination of limited aqueous solubility and flexibility and a less hydrophobic internal cavity, all of which combine to limit the intermolecular interactions between the cyclodextrin and the analyte and the resultant analyte-induced fluorescence modulation. As stated previously, the control trials without cyclodextrin displayed the lowest average fluorescence modulation, highlighting the critical role of cyclodextrin in promoting highly analyte- specific interactions with the high quantum yield fluorophore.

Among the various analytes investigated, analyte 1 was able to induce the highest degree of fluorescence modulation in the cyclodextrin hosts, whereas analyte 5 induced the lowest degree of modulation (Figure 3B). Moreover, analytes 2-4 have a similar, intermediate level of analyte-induced modulation when introduced in a cyclodextrin system. Differences in analyte-induced responses can be explained with the aid of computed electrostatic potential maps, which showing the molecular charge distributions (Figure 4), and the

58 quantitative electrostatic surface values (Table 3), which show minimum and maximum electrostatic potentials, dipole moments and polarized surface area.

Figure 44. Spartan-calculated electrostatic potential maps of (A) analyte 1 (B) analyte 2 (C) analyte 3 (D) analyte 4 and (E) analyte 5. The areas in the dark blue shade corresponds to electron-deficient, non-polar regions of the model and scales to the red regions, corresponding to the electron-rich, polar regions of the molecule. Color code for the molecular models: dark grey: carbon (C), light grey: hydrogen (H), red: oxygen (O), and light purple: nitrogen (N).

Table 36. Quantitative values calculated from the electrostatic potential maps of the steroids in Spartan 18’ Minimum Maximum Polar Dipole Analyte Electrostatic Electrostatic Surface Moment (D)c Potential (kJ/mol)a Potential (kJ/mol)b Area (Å2)d 1 -262.3 114.2 2.11 34.117 2 -301.9 114.4 5.14 41.129 3 -262.4 114.1 2.69 48.988 4 -348.4 123.7 3.69 43.486 5 -257.9 100.7 2.23 34.214 a Corresponds to the red, electron-rich regions of the electrostatic potential maps b Corresponds to the dark blue, electron-poor regions of the electrostatic potential maps c D: abbreviation of the unit debye, dipole moment resulting from two charges of opposite sign but an equal magnitude of 10−10 statcoulomb d Polar areas that occur due to electronegative elements and hydrogen atoms attached to them in a molecule

59 Overall, the main differences in the analyte structures occurs in the polar, electron-rich region on the left side of the molecule, the region where the minimum electrostatic potential fluctuates. This potential is used to gauge how reactive different regions of the molecule will be in the presence of varying species of molecules (electrophiles, nucleophiles, etc.) as well as provide insight into non-covalent intermolecular interactions.43 For example,

Figure 4A shows the structure of mesterolone (analyte 1) which has one hydroxyl group extending from the carbon ring on the left-hand side while Figure 4B illustrates oxandrolone (analyte 2), which has an ester in the same region. The minimum electrostatic potential energy increases in magnitude from analyte 1 to analyte 2, suggesting that analyte

2 has greater reactivity towards other species. In analyte 2, this change in structure also results in an increased polar surface area, defined as the area on the surface of a molecule that is affected by the charge of electronegative atoms and elements such as nitrogen and oxygen atoms as well as any hydrogens that are bonded to these elements.44–46 There is also a much higher dipole moment, the magnitude of the sum of net charge in a molecule based on the combination of nuclear and electron charges, which provides substantial insight into chemical reactivities of the species.47,48 The increase in these characteristics from analyte 1 to analyte 2 in turn decreases the interaction of analyte 2 with the nonpolar binding pocket of the cyclodextrin hosts while the interactions between analyte 1 and the nonpolar binding pocket increase. Analytes 3 and 4 have similarly high polar surface areas relative to analyte 2, which explains why the analyte-induced modulation caused by analyte

1 is noticeably higher than those of the other analytes.

In contrast to analytes 2-4 which show markedly higher dipole moments and polar surface areas that explain their decrease modulation values relative to analyte 1, analyte 5 has

60 computed molecular properties that are remarkably similar to those of analyte 1.

Nonetheless, the differences in how analytes 1 and 5 induce fluorescence modulation are substantial, with analyte 1 leading to markedly higher degrees of fluorescence modulation

(vide supra). Such differences may be explained not by the Spartan computed structural information, but by the literature-reported photophysical activity of analyte 5,39–42 which can lead to interference with the analyte-induced fluorescence modulation and complicate the observed results. NMR was used to investigate the interaction profiles of the analytes with β-CD. The only interaction seen was between the analytes and the out rim of the cyclodextrin and was indicated by a very slight shift in the two hydroxyl peaks located downfield of the aliphatic groups. There was not a large enough shift in spectra to confirm covalent bonding, only illustrating the analytes do you not readily enter the cavity of the cyclodextrin. Full NMR spectra can be seen in the ESI of this manuscript.

Array Generation Experiments

The ability of the analyte-induced fluorescence modulation to provide unique, highly selective signals for each analyte was determined through the use of linear discriminant analysis (LDA). By using the various cyclodextrin hosts as predictors for the system, we determined that each analyte generates an extremely well-separated signal in the LDA plots, regardless of whether 5 µL, 10 µL, or 20 µL, a range from 6 µM to just under 30

µM, of the analyte solution was used (Figure 5). The final concentrations of these analytes in the fluorescence modulation trials can be found in the ESI of this manuscript. Of note, these plots include tetrahydrofuran (THF) as a control analyte, because that was the solvent used in the steroid solutions, and the well-separated signal of the THF indicates that the

61 analytes all induce fluorescence responses that are unique from the response generated by their solvent alone.

Figure 45. Arrays generated using the three cyclodextrin hosts as the predictors for the (A) 5 µL addition trials, (B) 10 µL addition trials and (C) 20 µL addition trials. In addition to the well-separated signals between each analyte (apparent via visual inspection of Figure 5), introduction of analytes as unknowns into the system after classification resulted in 100% accurate classification of the analytes (see ESI for more details). Interestingly, as the additions increased in volume, the average cumulative proportion of total dispersion, a measure of the separation between signals, also increased.

The 20 µL trials had an average of 96.7% dispersion while the 5 µL and 10 µL trials had an average dispersion of 86.4% and 96.2% dispersion, respectively. This increase in dispersion is easily understandable, as higher concentrations of analytes lead to increased interactions with the fluorophore-cyclodextrin system and concomitant greater separation between signals.

Due to the differences in the observed cumulative proportions of total dispersion, we decided to further investigate system responsiveness to the analytes at the variety of concentrations investigated. Linear discriminant analysis of the signals generated from all analytes at all concentrations are summarized in Figure 6A (with a THF control signal) and

62 Figure 6B (without the THF control signal included). Of note, although well-separated signals were generated for all analytes, the signal for THF in Figure 6A has some overlap with the analytes investigated. When the THF signal was eliminated from the analysis,

100% separation was observed. Moreover, the closely grouped but still well-separated points on the plots represent the same analyte at different concentrations, which strongly suggests that quantitative analyte determination can also occur.

Figure 46. Arrays of fluorescence modulation data with three different addition volumes generated using the cyclodextrin hosts as predictors for all analytes (A) including THF as a control analyte; and (B) excluding THF as a control analyte Limit Detection and Quantification Experiments

In addition to determining the ability of the system to selectively distinguish between different analytes, the ability of the system to sensitively select analytes at low concentrations is critically important in practical detection devices. To that end, the limits of detection and limits of quantification were calculated for all analytes in each of the cyclodextrin solutions, following literature reported procedures (see ESI for more details), and key data is summarized in Table 4.

63 Table 37. Limits of detection (µM) calculated for analytes 1-5 in the cyclodextrin host systemsa Analyte β-CD Me-β-CD 2-HP-β-CD 1 17.03 5.300 2.335 2 11.39 3.886 3.352 3 8.914 0.148 1.886 4 7.665 5.981 0.775 5 6.108 0.049 3.587 a Values calculated using procedure found in the ESI of this manuscript. All results represent an average of at least four trials. Overall, the limits of detection using unmodified β-cyclodextrin were higher than the limits of detection obtained using the other supramolecular hosts, and reflects both the lower modulation values as well as the presumed weaker interactions between β-cyclodextrin and the steroid analytes. Of note, all limits of detection reported in Table 4 are markedly lower than the known concentrations of steroids found in urine testing following illegal doping activities, 49,50 which highlights the potential of this system to be used in practical detection applications.

CONCLUSION

As illegal doping scandals become more prevalent in practice for athletes in high level competition, the need for a rapid on-site detection system grows. The fluorescence modulation system introduced herein shows significant promise as a tool for the detection of illegal steroids. This system, which relies on cyclodextrin-promoted interactions between the target analytes and a high-quantum yield fluorophore, enabled 100% selective differentiation between response patterns of the structurally similar anabolic steroids as well as different concentrations of the analytes, as well as limits of detection that were lower than concentrations reported in illegal doping scenarios. Future research in this area will focus on incorporating more analytes as well as transitioning these promising results

64 to a solid-state detection system. Work towards these goals is currently underway in our laboratory, and the results of these and other investigations will be reported in due course.

ACKNOWLEDGEMENTS

Funding in support of this research was provided by the Rhode Island Foundation through their Clean Competition Fund. The authors greatly acknowledge the assistance of Mr.

Benjamin Cromwell and Mr. Carson Hasselbrink in setting up the computational experiments. The authors would also like to greatly acknowledge the assistance of Ms.

Kassie Picard in the development of all NMR experiments.

SUPPLEMENTARY INFORMATION

Details of fluorescence modulation experiments, limit of detection experiments, NMR experiments, array generation experiments, and computational experiments; summary tables and figures for all experiments. This information is available online.

65 REFERENCES (1) Patel, A.; Quint, R. H.; Fuchs, G. J. Applications of Electrical Energy to the Prostate: An Evolutionary Perspective. J. Urol. 1998, 159 (6), 1802–1809. (2) Vlad, R. A.; Hancu, G.; Popescu, G. C.; Lungu, I. A. Doping in Sports, a Never-Ending Story? Advanced Pharmaceutical Bulletin 2018, 8 (4), 529–534. (3) Ponzetto, F.; Giraud, S.; Leuenberger, N.; Boccard, J.; Nicoli, R.; Baume, N.; Rudaz, S.; Saugy, M. Methods for Doping Detection. In Frontiers of Hormone Research; Lanfranco, F., Strasburger, C. J., Eds.; S. Karger AG, 2016; Vol. 47, pp 153–167. (4) Martinez Brito, D.; Torre, X.; Botrè, F. Detection of Urinary Metabolites of Arimistane in Humans by Gas Chromatography Coupled to High‐accuracy Mass Spectrometry for Antidoping Analyses. Rapid Communications in Mass Spectrometry 2019, 33 (24), 1894–1905. (5) Mako, T. L.; Racicot, J. M.; Levine, M. Supramolecular Luminescent Sensors. Chemical Reviews 2019, 119 (1), 322–477. (6) Serio, N.; Moyano, D. F.; Rotello, V. M.; Levine, M. Array-Based Detection of Persistent Organic Pollutants via Cyclodextrin Promoted Energy Transfer. Chemical Communications 2015, 51 (58), 11615–11618. (7) DiScenza, D. J.; Lynch, J.; Feder, E.; Levine, M. Detection of Bisphenol A and Derivatives in Human Urine via Cyclodextrin-Promoted Fluorescence Modulation. Analytical Methods 2018, 10 (30), 3783–3790. (8) DiScenza, D. J.; Levine, M. Selective Detection of Non-Aromatic Pesticides via Cyclodextrin-Promoted Fluorescence Modulation. New Journal of Chemistry 2016, 40 (1), 789–793. (9) DiScenza, D. J.; Lynch, J.; Miller, J.; Verderame, M.; Levine, M. Detection of Organochlorine Pesticides in Contaminated Marine Environments via Cyclodextrin- Promoted Fluorescence Modulation. ACS Omega 2017, 2 (12), 8591–8599. (10) Haynes, A.; Halpert, P.; Levine, M. Colorimetric Detection of Aliphatic Alcohols in β-Cyclodextrin Solutions. ACS Omega 2019, 4 (19), 18361–18369. (11) DiScenza, D. J.; Lynch, J.; Verderame, M.; Serio, N.; Prignano, L.; Gareau, L.; Levine, M. Efficient Fluorescence Detection of Aromatic Toxicants and Toxicant Metabolites in Human Breast Milk. Supramolecular Chemistry 2018, 30 (4), 267–277. (12) DiScenza, D. J.; Smith, M. A.; Intravaia, L. E.; Levine, M. Efficient Detection of Phthalate Esters in Human Saliva via Fluorescence Spectroscopy. Analytical Letters 2019, 52 (3), 479–495. (13) Loock, H.-P.; Wentzell, P. D. Detection Limits of Chemical Sensors: Applications and Misapplications. Sensors and Actuators B: Chemical 2012, 173, 157–163.

66 (14) Belter, M.; Sajnóg, A.; Barałkiewicz, D. Over a Century of Detection and Quantification Capabilities in Analytical Chemistry – Historical Overview and Trends. Talanta 2014, 129, 606–616. (15) Sjöqvist, F.; Garle, M.; Rane, A. Use of Doping Agents, Particularly Anabolic Steroids, in Sports and Society. The Lancet 2008, 371 (9627), 1872–1882. (16) Maravelias, C.; Dona, A.; Stefanidou, M.; Spiliopoulou, C. Adverse Effects of Anabolic Steroids in Athletes. Toxicology Letters 2005, 158 (3), 167–175. (17) Chaudhuri, S.; DiScenza, D. J.; Verderame, M.; Cromwell, B.; Levine, M. Colorimetric Detection of Polycyclic Aromatic Hydrocarbons Using Supramolecular Cyclodextrin Dimer-Squaraine Constructs. Supramolecular Chemistry 2019, 31 (3), 211–219. (18) Breul, A. M.; Hager, M. D.; Schubert, U. S. Fluorescent Monomers as Building Blocks for Dye Labeled Polymers: Synthesis and Application in Energy Conversion, Biolabeling and Sensors. Chemical Society Reviews 2013, 42 (12), 5366. (19) Wang, L.; Liu, Y.; Chen, F.; Zhang, J.; Anpo, M. Manipulating Energy Transfer Processes between Rhodamine 6G and Rhodamine B in Different Mesoporous Hosts. The Journal of Physical Chemistry C 2007, 111 (14), 5541–5548. (20) Fenyvesi, É.; Puskás, I.; Szente, L. Cyclodextrin-Steroid Interactions and Applications to Pharmaceuticals, Food, Biotechnology and Environment. In Cyclodextrin Applications in Medicine, Food, Environment and Liquid Crystals; Fourmentin, S., Crini, G., Lichtfouse, E., Eds.; Springer International Publishing: Cham, 2018; Vol. 17, pp 19–57. (21) Ammeraal, R. N.; Nagle, S. CYCLODEXTRN STEROD COMPLEX. 9. (22) Fenyvesi, É.; Puskás, I.; Szente, L. Applications of Steroid Drugs Entrapped in Cyclodextrins. Environmental Chemistry Letters 2019, 17 (1), 375–391. (23) Bhopate, S. B.; Dhole, S. N. Preparation and Characterization of β-Cyclodextrin Nebivolol Inclusion Complex. IJPSR 2018, 6 (5), 2205–2213. (24) Liu, L.; Guo, Q.-X. The Driving Forces in the Inclusion Complexation of Cyclodextrins. 14. (25) Schneider, H.-J. Dispersive Interactions in Solution Complexes. Accounts of Chemical Research 2015, 48 (7), 1815–1822. (26) Plass, W. Chiral and Supramolecular Model Complexes for Vanadium Haloperoxidases: Host–Guest Systems and Hydrogen Bonding Relays for Vanadate Species. Coordination Chemistry Reviews 2011, 255 (19–20), 2378–2387.

67 (27) Martín, L. Spectrofluorimetric Determination of Stoichiometry and Association Constants of the Complexes of Harmane and Harmine with β-Cyclodextrin and Chemically Modified β-Cyclodextrins. Talanta 2003, 60 (2–3), 493–503. (28) Bru, R.; López-Nicolás, J. M.; Núñez-Delicado, E.; Nortes-Ruipérez, D.; Sánchez- Ferrer, A.; Garciá-Carmona, F. Cyclodextrins as Hosts for Poorly Water-Soluble Compounds in Enzyme Catalysis. Applied Biochemistry and Biotechnology 1996, 61 (1–2), 189–198. (29) Breslow, R.; Yan, J.; Belvedere, S. Catalytic Hydroxylation of Steroids by Cytochrome P-450 Mimics. Hydroxylation at C-9 with Novel Catalysts and Steroid Substrates. Tetrahedron Letters 2002, 43 (3), 363–365. (30) Yang, J.; Gabriele, B.; Belvedere, S.; Huang, Y.; Breslow, R. Catalytic Oxidations of Steroid Substrates by Artificial Cytochrome P-450 Enzymes †. The Journal of Organic Chemistry 2002, 67 (15), 5057–5067. (31) Connors, K. A. The Stability of Cyclodextrin Complexes in Solution. Chemical Reviews 1997, 97 (5), 1325–1358. (32) Khuntawee, W.; Karttunen, M.; Wong-ekkabut, J. A Molecular Dynamics Study of Conformations of Beta-Cyclodextrin and Its Eight Derivatives in Four Different Solvents. Physical Chemistry Chemical Physics 2017, 19 (35), 24219–24229. (33) Frankewich, R. P.; Thimmaiah, K. N.; Hinze, W. L. Evaluation of the Relative Effectiveness of Different Water-Soluble .Beta.-Cyclodextrin Media to Function as Fluorescence Enhancement Agents. Analytical Chemistry 1991, 63 (24), 2924–2933. (34) Albers, E.; Müller, B. W. Cyclodextrin Derivatives in Pharmaceutics. Crit Rev Ther Drug Carrier Syst 1995, 12 (4), 311–337. (35) Fourmentin, S.; Ciobanu, A.; Landy, D.; Wenz, G. Space Filling of β-Cyclodextrin and β-Cyclodextrin Derivatives by Volatile Hydrophobic Guests. Beilstein Journal of Organic Chemistry 2013, 9, 1185–1191. (36) Loftsson, T.; Jarho, P.; Másson, M.; Järvinen, T. Cyclodextrins in Drug Delivery. Expert Opinion on Drug Delivery 2005, 2 (2), 335–351. (37) Christian, A. E.; Byun, H. S.; Zhong, N.; Wanunu, M.; Marti, T.; Fürer, A.; Diederich, F.; Bittman, R.; Rothblat, G. H. Comparison of the Capacity of Beta-Cyclodextrin Derivatives and Cyclophanes to Shuttle Cholesterol between Cells and Serum Lipoproteins. J. Lipid Res. 1999, 40 (8), 1475–1482. (38) Serio, N.; Miller, K.; Levine, M. Efficient Detection of Polycyclic Aromatic Hydrocarbons and Polychlorinated Biphenyls via Three-Component Energy Transfer. Chemical Communications 2013, 49 (42), 4821. (39) Li, Y. Determination of Trenbolone Acetate in Cattle Feeds by a Flow Injection Chemiluminescence Method. Analytical Letters 2011, 44 (1–3), 58–66.

68 (40) Oehrle, K.-L.; Vogt, K.; Hoffmann, B. Determination of Trenbolone and Trenbolone Acetate® by Thin-Layer Chromatography in Combination with a Fluorescence Colour Reaction. Journal of Chromatography A 1975, 114 (1), 244–246. (41) Durhan, E. J.; Lambright, C. S.; Makynen, E. A.; Lazorchak, J.; Hartig, P. C.; Wilson, V. S.; Gray, L. E.; Ankley, G. T. Identification of Metabolites of Trenbolone Acetate in Androgenic Runoff from a Beef Feedlot. Environmental Health Perspectives 2006, 114 (Suppl 1), 65–68. (42) Yoshioka, N.; Akiyama, Y.; Takeda, N. Determination of Alpha- and Beta-Trenbolone in Bovine Muscle and Liver by Liquid Chromatography with Fluorescence Detection. J. Chromatogr. B Biomed. Sci. Appl. 2000, 739 (2), 363–367. (43) Murray, J. S.; Politzer, P. The Electrostatic Potential: An Overview. WIREs Comput Mol Sci 2011, 1 (2), 153–163. (44) van de Waterbeemd, H.; Kansy, M. Hydrogen-Bonding Capacity and Brain Penetration. CHIMIA International Journal for Chemistry 1992, 46, 299–303. (45) Hehre, W.; Ohlinger, S. Spartan ’18 Manual for Windows, Macintosh and Linux: Tutorial and User’s Guide; Wavefunction, Inc: Irvine, CA, 2019. (46) Comprehensive Medicinal Chemistry II, 2nd ed.; Taylor, J. B., Triggle, D. J., Eds.; Elsevier: Amsterdam, 2007; Vol. 5. (47) Bader, R. F. W.; Larouche, A.; Gatti, C.; Carroll, M. T.; MacDougall, P. J.; Wiberg, K. B. Properties of Atoms in Molecules: Dipole Moments and Transferability of Properties. The Journal of Chemical Physics 1987, 87 (2), 1142–1152. (48) Gao, J.; Luque, F. J.; Orozco, M. Induced Dipole Moment and Atomic Charges Based on Average Electrostatic Potentials in Aqueous Solution. The Journal of Chemical Physics 1993, 98 (4), 2975–2982. (49) Lood, Y.; Eklund, A.; Garle, M.; Ahlner, J. Anabolic Androgenic Steroids in Police Cases in Sweden 1999–2009. Forensic Science International 2012, 219 (1–3), 199– 204. (50) Andersen, D. W.; Linnet, K. Screening for Anabolic Steroids in Urine of Forensic Cases Using Fully Automated Solid Phase Extraction and LC–MS-MS. Journal of Analytical Toxicology 2014, 38 (9), 637–644.

ELECTRONIC SUPPLEMENTARY INFORMATION

DETECTION OF STEROIDS VIA CYCLODEXTRIN-PROMOTED FLUORESCENCE MODULATION

Anna Z. Haynes and Mindy Levine

70 MATERIALS AND METHODS

The anabolic steroid analytes, chemicals required to make buffer solutions, fluorophore

Rhodamine 6G, and solvent tetrahydrofuran were obtained from Sigma-Aldrich Chemical company and the cyclodextrins were obtained from Tokyo Chemical Industry (TCI). All chemicals were used as received without further purification. All fluorescence measurements were performed using a Shimadzu RF 6000 spectrophotometer. The excitation and emission slit widths were set to 3.0 nm. All fluorescence spectra were integrated vs. wavenumber on the X-axis using OriginPro 2019 Version 9.60. All arrays were generated using SYSTAT Version 13.1. All NMR spectra were taken using a

Bruker 400 MHz NMR and analyzed with MestReNova 14.1 software.

DETAILS OF ANALYTES AND FLUOROPHORES

Figure 47: Structure of anabolic analytes (compound 1: Mesterolone; compound 2: Oxandrolone; compound 3: Oxymetholone; compound 4: Stanozolol; compound 5: Trenbolone) and fluorophore Rhodamine 6G (compound 6) All analyte samples were prepared at a concentration of 1.0 mg/mL in THF. The fluorophore solution was prepared at a concentration of 0.1 mg/mL in THF. An 0.1 M citrate buffer was prepared by combining 2.409 grams of sodium citrate and 0.347 grams

71 of citric acid in a 1.0 L volumetric flask and diluting to the mark with distilled water. The pH of the buffer was measured at 6.1, and remained consistent throughout the experimental procedures. Cyclodextrin solutions were prepared at a concentration of 10 mmol in the citrate buffer. The final concentrations of the analytes and fluorophore are shown in Table S1, below:

Table 38: Concentration of analytes and fluorophore in solution prior to dilution via sample preparation Compound Number Concentration (mM) 1 3.28 2 3.26 3 3.01 4 3.04 5 3.62 6 2.09

EXPERIMENTAL PROCEDURES

Experimental Procedure for Fluorescence Modulation Experiments

Fluorescence modulation experiments were done with 5 µL, 10 µL, and 20 µL sequential additions of analyte. The fluorescence modulation values for each analyte-cyclodextrin combination was determined in the following procedure:

1. 100 µL of fluorophore 6 solution (0.1 mg/mL) in THF was measured into six 15

mL glass vials (vial 1 for THF, vial 2 for analyte 1, vial 3 for analyte 2, etc.). 2.00

mL of a 10 mM cyclodextrin in citrate buffer and 0.40 mL 0.1 M citrate buffer were

added to each (citrate buffer was at pH 6.1). The vials were capped and left to

stabilize for 48 hours in a dark drawer.

72 2. After the 48 hours, the contents of one vial and 5.0 µL of analyte were added to the

cuvette and stirred thoroughly to ensure homogeneity. The solution was excited at

490 nm and recorded from 500-800 nm. Four repeat measurements were taken. This

step was repeated for each analyte and an additional time, adding 5.0 µL of THF

instead of an analyte solution to use as a control.

3. Step 1 and 2 were repeated for each analyte-cyclodextrin combination (18 trials in

total). In all cases, the solution was excited at the same wavelength (490 nm) and

the emission spectra from 500 to 800 nm was recorded four times.

4. To conduct an experiment with no cyclodextrin present, step 1 was repeated but

citrate buffer solution was substituted in place of the cyclodextrin solution. The

contents of the vials were then 100 µL of fluorophore 6 solution and 2.4 mL of 0.1

M citrate buffer.

5. After 48 hours, step 2 was repeated using this set of solutions containing no

cyclodextrin for each analyte. The solutions were excited at the same wavelength

(490 nm) and the emission spectra from 500 to 800 nm was recorded four times.

6. Emission spectra were integrated versus wavenumber on the X-axis using

OriginPro software and fluorescence modulation ratios were determined according

to Equation 1, below:

Fluorophore Modulation Ratios = Flanalyte / Flblank (Eq.1)

where Flanalyte represents the integrated fluorescence emission of the fluorophore in

the presence of the analyte and Flblank represents the integrated fluorescence

emission of the fluorophore in the absence of the analyte.

7. These ratios were recorded and are show in Table S4.

73 Step 2 and 5 were repeated with sequential additions for total addition amounts of 10 µL and 20 µL of analyte. The final concentrations of the analytes and the fluorophore in solution using each of the three addition protocols are summarized in Table S2, below:

Table 39: Final concentrations (µM) of analytes and fluorophore in fluorescence modulation trials Compound 5 µL 10 µL 20 µL 1 6.419 12.81 25.53 2 6.419 12.81 25.53 3 5.890 11.76 23.42 4 5.949 11.88 23.67 5 7.226 14.42 28.73 6 8.334 8.317 8.284

Experimental Procedure for Limit of Detection Experiments

The limit of detection (LOD), defined as the lowest concentration of the analyte that can be detected, was obtained using the calibration curve method, following procedures reported by Loock et. al.1 The limit of quantification (LOQ) is the lowest concentration of analyte that can be reliably and accurately quantified. The limit of detection and quantification experiments were conducted following literature-reported procedures.1,2

LOD experiments were done with sequential 5 µL additions of analyte, according to the procedures listed below:

1. 100 µL of fluorophore 6 solution (0.1 mg/mL) in THF was measured into a 15 mL

glass vial. 2.00 mL of a 10 mM cyclodextrin in citrate buffer and 0.40 mL 0.1 M

citrate buffer were added (citrate buffer was at pH 6). The vial was capped and left

to stabilize for 48 hours.

74 2. The solution was transferred to a quartz cuvette and then excited at 490 nm and the

fluorescence emission spectra was recorded from 500 to 800 nm. Each fluorescence

measurement was repeated six times

3. 5.0 µL of the analyte solution in THF was added to the cuvette and stirred

thoroughly to ensure homogeneity. The solution was excited at the same

wavelength (490 nm) and the emission was measured between 500 nm and 800 nm.

Six repeat measurements were taken.

4. Step 2 was repeated four times for total addition volumes of 10 µL, 15 µL, 20 µL,

and 25 µL of the analyte solution. In all cases, the solution was excited at the same

wavelength (490 nm) and the emission spectra from 500 to 800 nm was recorded

six times.

5. Emission spectra were integrated versus wavenumber on the X-axis using

OriginPro software, and were used to generate calibration curves with analyte

concentration on the X-axis and integrated fluorescence emission on the Y-axis.

The curve was fitted with a linear trendline and the equation of the line was

determined.

6. The measurements from Step 1, the emission spectra of the combination of the

Rhodamine solution and β-cyclodextrin solution with no addition of analyte, are

referred to as the blank in the following calculations.

7. The limit of the blank (LODblank) is defined according to the following equation:

LODblank = mblank – 3(SDblank) (Eq. 2)

where mblank is the mean of the blank integrations and SDblank is the standard

deviation of those measurements.

8. The LODblank was then entered into the equation determined in Step 4 as the y-value.

The corresponding x-value was calculated. This value is the LOD of the analyte in

µM in the system.

9. The LOQ (LOQblank) was calculated in a similar manner to the LOD. The limit of

quantification of the blank is defined according to the following equation:

LOQblank = mblank – 10(SDblank). (Eq. 3)

10. This value is then entered as the y-value from step 4 and the corresponding x-value

was calculated. This is the value of the LOQ of the analyte for the system in µM.

All steps were repeated for each analyte-cyclodextrin combination:

(5 analytes x 3 cyclodextrin solutions = 15 trials).

The summary tables of these results for each analyte-cyclodextrin combination are shown in Tables S4-S6 (vide infra).

Experimental Procedure for Array Generation Experiments

Linear discriminant analysis was performed using SYSTAT 13 statistical computing software with the following software settings3:

(a) Classical Discriminant Analysis

(b) Grouping Variable: Analytes

(d) Long-Range Statistics: Mahal.

These experiments were then repeated using only two predictors (i.e. cyclodextrins) instead of all three, and the results of array-based analysis for each pair of predictors is reported herein as well.

76 Experimental Procedure for Computational Experiments

Spartan version ’18 was used to calculate the equilibrium molecular conformations of each analyte in their ground states in the gas phase using a semi-empirical PM3 model for each analyte. This allowed an electrostatic potential map surface to be overlaid over the molecules with the mesh overlay function.

Experimental Procedure for 1H NMR Titration Experiments

All analytes and β-cyclodextrin solutions were prepared at a concentration of 10 mM in deuterated dimethyl sulfoxide (DMSO). 1H NMR samples with total volumes of 0.50 mL were prepared in Wilmad precision NMR tubes according to the details listed below:

Sample 1: β-CD only

Sample 2: Analyte only

Sample 3: A molar ratio of 1:2 of analyte to β-CD

Sample 4: A molar ratio of 1:1 of analyte to β-CD

Sample 5: A molar ratio of 2:1 of analyte to β-CD.

These samples were unheated and were at room temperature when spectra were taken. A

1H NMR spectrum was taken of each using a Bruker 400 MHz NMR. These NMR spectra were analyzed using MestReNova 14.1 software.

77 SUMMARY TABLES Summary Tables for Fluorescence Modulation Experiments Table 40. Fluorescence modulation results obtained for analytes 1-5 with various cyclodextrins in the presence of fluorophore 6a Addition Analyte β-CD Me-β-CD 2-HPCD No CD amount 5 μL 1 0.958 ± 0.001 0.956 ± 0.000 0.905 ± 0.000 1.001 ± 0.001 2 0.997 ± 0.001 0.971 ± 0.000 0.957 ± 0.000 0.973 ± 0.001 3 1.004 ± 0.001 0.956 ± 0.000 0.966 ± 0.000 1.006 ± 0.001 4 0.989 ± 0.003 0.998 ± 0.001 0.906 ± 0.000 0.979 ± 0.001 5 0.997 ± 0.001 0.978 ± 0.001 0.999 ± 0.001 1.047 ± 0.001 10 μL 1 0.950 ± 0.000 0.938 ± 0.000 0.900 ± 0.000 1.004 ± 0.001 2 0.990 ± 0.001 0.967 ± 0.000 0.953 ± 0.000 0.990 ± 0.000 3 0.988 ± 0.001 0.955 ± 0.000 0.962 ± 0.000 1.009 ± 0.001 4 0.985 ± 0.002 0.995 ± 0.000 0.901 ± 0.000 0.969 ± 0.002 5 0.992 ± 0.001 0.975 ± 0.001 0.992 ± 0.000 1.040 ± 0.000 20 μL 1 0.948 ± 0.001 0.932 ± 0.000 0.892 ± 0.000 0.994 ± 0.001 2 0.974 ± 0.001 0.966 ± 0.001 0.947 ± 0.000 0.987 ± 0.000 3 0.978 ± 0.001 0.954 ± 0.000 0.955 ± 0.000 1.004 ± 0.003 4 0.975 ± 0.000 0.986 ± 0.001 0.952 ± 0.000 0.972 ± 0.001 5 0.975 ± 0.001 0.969 ± 0.001 0.988 ± 0.001 1.036 ± 0.001 a All values were calculated using Equation S1, and results reported represent an average of at least four trials.

78 Summary Tables for Limit of Detection Experiments Table 41. Summary table for limits of detection experiments with β-CDa,b Analyte LOD (μM) LOQ (μM) Equation R² 1 17.03 46.58 y = -1292.6693x + 2133565.2492 0.8369 2 11.39 35.32 y = -2079.0320x + 2229038.8629 0.9589 3 8.914 21.42 y = -2559.6708x + 2233461.8386 0.9308 4 7.665 30.49 y = -2237.8468x + 2220576.1935 0.9776 5 6.108 24.35 y = -2308.8616x + 2241047.0782 0.9498 a All LOD values were calculated using Equation 2. a All LOQ values were calculated using Equation 3. Table 42. Summary table for limits of detection experiments with Me-β-CDa,b Analyte LOD (μM) LOQ (μM) Equation R² 1 5.300 11.72 y = -27374.1167x + 31229827.0035 0.8451 2 3.886 18.34 y = -11127.1906x + 31855034.7259 0.9193 3 0.148 4.102 y = -9482.3445x + 31383949.2786 0.8929 4 5.981 18.21 y = -16056.2718x + 32744010.5956 0.9379 5 0.049 4.586 y = -21626.3876x + 32220132.7800 0.9755 a All LOD values were calculated using Equation 2. a All LOQ values were calculated using Equation 3. Table 43. Summary table for limits of detection experiments with 2-HPCDa,b Analyte LOD (μM) LOQ (μM) Equation R² 1 2.335 7.129 y = -18245.1311x + 26534054.7777 0.9956 2 3.352 14.66 y = -12790.4792x + 28037979.4349 0.9657 3 1.886 8.361 y = -21071.8113x + 28366093.4431 0.9945 4 0.775 3.371 y = -10283.4719x + 26474157.9172 0.8972 5 3.587 11.53 y = -11788.1436x + 29202530.0773 0.9438 a All LOD values were calculated using Equation 2. a All LOQ values were calculated using Equation 3.

79 Summary Tables for Array Generation Experiments With 5 µL analyte additions Table 44. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

Table 45. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

Table 46. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

80 Table 47. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors

With 10 µL analyte additions Table 48. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

Table 49. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

81 Table 50. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

Table 51. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors

With 20 µL analyte additions Table 52. Linear discriminant analysis results using β-CD, Me-β-CD, and 2-HPCD as predictors

82 Table 53. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

Table 54. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

Table 55. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors

83 All Additions with THF Table 56. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

All Additions excluding THF Table 57. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

84 SUMMARY FIGURES Summary Figures for Individual Fluorescence Modulation Experiments using fluorophore 6 With 5 µL analyte addition β-CD

Figure 48. Fluorescence modulation of analyte 1 in the presence of β-CD

Figure 49. Fluorescence modulation of analyte 2 in the presence of β-CD

Figure 50. Fluorescence modulation of analyte 3 in the presence of β-CD

Figure 51. Fluorescence modulation of analyte 4 in the presence of β-CD

Figure 52. Fluorescence modulation of analyte 5 in the presence of β-CD

Me-β-CD

Figure 52. Fluorescence modulation of analyte 1 in the presence of Me-β-CD

Figure 53. Fluorescence modulation of analyte 2 in the presence of Me-β-CD

Figure 54. Fluorescence modulation of analyte 3 in the presence of Me-β-CD

Figure 55. Fluorescence modulation of analyte 4 in the presence of Me-β-CD

Figure 56. Fluorescence modulation of analyte 5 in the presence of Me-β-CD

2-HPCD

Figure 57. Fluorescence modulation of analyte 1 in the presence of 2-HPCD

Figure 58. Fluorescence modulation of analyte 2 in the presence of 2-HPCD

Figure 59. Fluorescence modulation of analyte 3 in the presence of 2-HPCD

Figure 60. Fluorescence modulation of analyte 4 in the presence of 2-HPCD

Figure 61. Fluorescence modulation of analyte 5 in the presence of 2-HPCD

No CD

Figure 62. Fluorescence modulation of analyte 1 in the presence of no CD

Figure 63. Fluorescence modulation of analyte 2 in the presence of no CD

90 Figure 64. Fluorescence modulation of analyte 3 in the presence of no CD

Figure 65. Fluorescence modulation of analyte 4 in the presence of no CD

Figure 66. Fluorescence modulation of analyte 5 in the presence of no CD

With 10 µL analyte addition

β-CD

91 Figure 67. Fluorescence modulation of analyte 1 in the presence of β-CD

Figure 68. Fluorescence modulation of analyte 2 in the presence of β-CD

Figure 69. Fluorescence modulation of analyte 3 in the presence of β-CD

Figure 70. Fluorescence modulation of analyte 4 in the presence of β-CD

Figure 71. Fluorescence modulation of analyte 5 in the presence of β-CD

Me-β-CD

Figure 72. Fluorescence modulation of analyte 1 in the presence of Me-β-CD

Figure 73. Fluorescence modulation of analyte 2 in the presence of Me-β-CD

Figure 74. Fluorescence modulation of analyte 3 in the presence of Me-β-CD

Figure 75. Fluorescence modulation of analyte 4 in the presence of Me-β-CD

Figure 76. Fluorescence modulation of analyte 5 in the presence of Me-β-CD

94 2-HPCD

Figure 77. Fluorescence modulation of analyte 1 in the presence of 2-HPCD

Figure 78. Fluorescence modulation of analyte 2 in the presence of 2-HPCD

Figure 79. Fluorescence modulation of analyte 3 in the presence of 2-HPCD

Figure 80. Fluorescence modulation of analyte 4 in the presence of 2-HPCD

Figure 81. Fluorescence modulation of analyte 5 in the presence of 2-HPCD No CD

Figure 82. Fluorescence modulation of analyte 1 in the presence of no CD

Figure 83. Fluorescence modulation of analyte 2 in the presence of no CD

Figure 84. Fluorescence modulation of analyte 3 in the presence of no CD

Figure 85. Fluorescence modulation of analyte 4 in the presence of no CD

Figure 86. Fluorescence modulation of analyte 5 in the presence of no CD

With 20 µL analyte addition

β-CD

Figure 87. Fluorescence modulation of analyte 1 in the presence of β-CD

Figure 88. Fluorescence modulation of analyte 2 in the presence of β-CD

Figure 89. Fluorescence modulation of analyte 3 in the presence of β-CD

Figure 90. Fluorescence modulation of analyte 4 in the presence of β-CD

Figure 91. Fluorescence modulation of analyte 5 in the presence of β-CD

99 Me-β-CD

Figure 92. Fluorescence modulation of analyte 1 in the presence of Me-β-CD

Figure 93. Fluorescence modulation of analyte 2 in the presence of Me-β-CD

Figure 94. Fluorescence modulation of analyte 3 in the presence of Me-β-CD

100

Figure 95. Fluorescence modulation of analyte 4 in the presence of Me-β-CD

Figure 96. Fluorescence modulation of analyte 5 in the presence of Me-β-CD

2-HPCD

Figure 97. Fluorescence modulation of analyte 1 in the presence of 2-HPCD

101

Figure 98. Fluorescence modulation of analyte 2 in the presence of 2-HPCD

Figure 99. Fluorescence modulation of analyte 3 in the presence of 2-HPCD

Figure 100. Fluorescence modulation of analyte 4 in the presence of 2-HPCD

102

Figure 101. Fluorescence modulation of analyte 5 in the presence of 2-HPCD

No CD

Figure 102. Fluorescence modulation of analyte 1 in the presence of no CD

Figure 103. Fluorescence modulation of analyte 2 in the presence of no CD

103

Figure 104. Fluorescence modulation of analyte 3 in the presence of no CD

Figure 105. Fluorescence modulation of analyte 4 in the presence of no CD

Figure 106. Fluorescence modulation of analyte 5 in the presence of no CD

104 Summary Figures for Combined Fluorescence Modulation Experiments using Fluorophore 6

A B C

Figure 107. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of THF (B) 10 µL of THF (C) 20 µL of THF in the presence of all hosts

A B C

Figure 108. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 1 (B) 10 µL of analyte 1 (C) 20 µL of analyte 1 in the presence of all hosts

A B C

Figure 109. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 2 (B) 10 µL of analyte 2 (C) 20 µL of analyte 2 in the presence of all hosts

105 A B C

Figure 110. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 3 (B) 10 µL of analyte 3 (C) 20 µL of analyte 3 in the presence of all hosts

A B C

Figure 111. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 4 (B) 10 µL of analyte 4 (C) 20 µL of analyte 4 in the presence of all hosts

A B C

Figure 112. Fluorescence modulation of fluorophore 6 induced by: (A) 5 µL of analyte 5 (B) 10 µL of analyte 5 (C) 20 µL of analyte 5 in the presence of all hosts

106 Summary Figures for Limit of Detection (LOD) Experiments Experiments were carried out with 5 µL sequential additions of analytes with fluorophore 6 in the presence of all cyclodextrin hosts. LOD calibration curves of analytes in presence of β-CD and fluorophore 6

Figure 113. LOD calibration curve of analyte 1

Figure 114. LOD calibration curve of analyte 2

Figure 115. LOD calibration curve of analyte 3

107

Figure 116. LOD calibration curve of analyte 4

Figure 117. LOD calibration curve of analyte 5

LOD calibration curves of analytes in presence of Me-β-CD and fluorophore 6

Figure 118. LOD calibration curve of analyte 1

108

Figure 119. LOD calibration curve of analyte 2

Figure 120. LOD calibration curve of analyte 3

Figure 121. LOD calibration curve of analyte 4

109

Figure 122. LOD calibration curve of analyte 5

LOD calibration curves of analytes in presence of 2-HPCD and fluorophore 6

Figure 123. LOD calibration curve of analyte 1

Figure 124. LOD calibration curve of analyte 2

110

Figure 125. LOD calibration curve of analyte 3

Figure 126. LOD calibration curve of analyte 4

Figure 127. LOD calibration curve of analyte 5

111 Summary Figures for Array Generation Experiments With 5 µL analyte additions

Figure 128. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

Figure 129. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

Figure 130. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

112

Figure 131. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors With 10 µL analyte additions

Figure 132. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

Figure 133. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

113

Figure 134. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

Figure 135. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors With 20 µL analyte additions

Figure 136. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

114

Figure 137. Linear discriminant analysis results using β-CD and Me-β-CD as predictors

Figure 138. Linear discriminant analysis results using β-CD and 2-HPCD as predictors

Figure 139. Linear discriminant analysis results using Me-β-CD and 2-HPCD as predictors

115 All Additions with THF

Figure 140. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

All Additions excluding THF

Figure 141. Linear discriminant analysis results with β-CD, Me-β-CD, and 2-HPCD as predictors

116 Summary Figures for Computational Experiments Spartan ’18 Electrostatic Potential Map Diagrams

Figure 142. Electrostatic potential map of analyte 1 in the gas phase at its most stable (i.e. “ground state” configuration)

Figure 143. Electrostatic potential map of analyte 2 in the gas phase at its most stable (i.e. “ground state” configuration)

117

Figure 144. Electrostatic potential map of analyte 3 in the gas phase at its most stable (i.e. “ground state” configuration)

Figure 145. Electrostatic potential map of analyte 4 in the gas phase at its most stable (i.e. “ground state” configuration)

118

Figure 146. Electrostatic potential map of analyte 5 in the gas phase at its most stable (i.e. “ground state” configuration)

119 Summary Figures for NMR Experiments Stacked NMR spectra (Full spectra from 0 ppm to 6 ppm)

Figure 147. Stacked NMRs of (A) 2:1 analyte 1:β-cyclodextrin (B) 1:1 analyte 1:β- cyclodextrin (C) 1:2 analyte 1:β-cyclodextrin (D) analyte 1 and (E) β-cyclodextrin

120

Figure 148. Stacked NMRs of (A) 2:1 analyte 2:β-cyclodextrin (B) 1:1 analyte 2:β- cyclodextrin (C) 1:2 analyte 2:β-cyclodextrin (D) analyte 2 and (E) β-cyclodextrin

121

Figure 149. Stacked NMRs of (A) 2:1 analyte 3:β-cyclodextrin (B) 1:1 analyte 3:β- cyclodextrin (C) 1:2 analyte 3:β-cyclodextrin (D) analyte 3 and (E) β-cyclodextrin

122

Figure 150. Stacked NMRs of (A) 2:1 analyte 4:β-cyclodextrin (B) 1:1 analyte 4:β- cyclodextrin (C) 1:2 analyte 4:β-cyclodextrin (D) analyte 4 and (E) β-cyclodextrin

123

Figure 151. Stacked NMRs of (A) 2:1 analyte 5:β-cyclodextrin (B) 1:1 analyte 5:β- cyclodextrin (C) 1:2 analyte 5:β-cyclodextrin (D) analyte 5 and (E) β-cyclodextrin

124 Stacked NMR spectra (Zoomed-in spectra from 5.63 ppm to 5.77 ppm)

Figure 152. Stacked NMRs of (A) 2:1 analyte 1:β-cyclodextrin (B) 1:1 analyte 1:β- cyclodextrin (C) 1:2 analyte 1:β-cyclodextrin (D) analyte 1 and (E) β-cyclodextrin

125

Figure 153. Stacked NMRs of (A) 2:1 analyte 2:β-cyclodextrin (B) 1:1 analyte 2:β- cyclodextrin (C) 1:2 analyte 2:β-cyclodextrin (D) analyte 2 and (E) β-cyclodextrin

126

Figure 154. Stacked NMRs of (A) 2:1 analyte 3:β-cyclodextrin (B) 1:1 analyte 3:β- cyclodextrin (C) 1:2 analyte 3:β-cyclodextrin (D) analyte 3 and (E) β-cyclodextrin

127

Figure 155. Stacked NMRs of (A) 2:1 analyte 4:β-cyclodextrin (B) 1:1 analyte 4:β- cyclodextrin (C) 1:2 analyte 4:β-cyclodextrin (D) analyte 4 and (E) β-cyclodextrin

128

Figure 156. Stacked NMRs of (A) 2:1 analyte 5:β-cyclodextrin (B) 1:1 analyte 5:β- cyclodextrin (C) 1:2 analyte 5:β-cyclodextrin (D) analyte 5 and (E) β-cyclodextrin

129 REFERENCES (1) Loock, H.-P.; Wentzell, P. D. Detection Limits of Chemical Sensors: Applications and Misapplications. Sensors and Actuators B: Chemical 2012, 173, 157–163. (2) Belter, M.; Sajnóg, A.; Barałkiewicz, D. Over a Century of Detection and Quantification Capabilities in Analytical Chemistry – Historical Overview and Trends. Talanta 2014, 129, 606–616. (3) DiScenza, D. J.; Lynch, J.; Miller, J.; Verderame, M.; Levine, M. Detection of Organochlorine Pesticides in Contaminated Marine Environments via Cyclodextrin- Promoted Fluorescence Modulation. ACS Omega 2017, 2 (12), 8591–8599.

130