View Is Primarily on Addressing the Issues of Non-Permanently Charged Reactivators and the Development of Treatments for Aged Ache

Home , Demeton, Diisopropyl paraoxon, Dimethylamine, Echothiophate, Fenamiphos, Gerhard Schrader

Design, Synthesis, and Evaluation of Therapeutics for the Treatment of

Organophosphorus Poisoning by Nerve Agents and Pesticides

Dissertation

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

in the Graduate School of The Ohio State University

Andrew Joseph Franjesevic

Graduate Program in Chemistry

The Ohio State University

2019

Dissertation Committee

Professor Christopher M. Hadad, Advisor

Professor Thomas J. Magliery

Professor David Nagib

Professor Jonathan R. Parquette

Copyrighted by

Andrew Joseph Franjesevic

2019

Abstract

Organophosphorus (OP) compounds, both pesticides and nerve agents, are some of the most lethal compounds known to man. Although highly regulated for both military and agricultural use in Western societies, these compounds have been implicated in hundreds of thousands of deaths annually, whether by accidental or intentional exposure through agricultural or terrorist uses. OP compounds inhibit the function of the enzyme acetylcholinesterase (AChE), and AChE is responsible for the hydrolysis of the neurotransmitter acetylcholine (ACh), and it is extremely well evolved for the task.

Inhibition of AChE rapidly leads to accumulation of ACh in the synaptic junctions, resulting in a cholinergic crisis which, without intervention, leads to death.

Approximately 70-80 years of research in the development, treatment, and understanding of OP compounds has resulted in only a handful of effective (and approved) therapeutics for the treatment of OP exposure.

The search for more effective therapeutics is limited by at least three major problems: (1) there are no broad scope reactivators of OP-inhibited AChE; (2) current therapeutics are permanently positively charged and cannot cross the blood-brain barrier efficiently; and (3) current therapeutics are ineffective at treating the aged, or dealkylated, form of AChE that forms following inhibition of of AChE by various OPs.

iii

Herein we report initial computational investigations, the synthesis, and then the biological evaluation of a new class of therapeutic quinone methide precursors (QMPs) for the purpose of treating OP exposure. We have thus far been able to demonstrate that our compounds bind with AChE, reactivate OP-inhibited AChE, and resurrect (restore) native function of the aged-AChE. These compounds are non-oxime structures and address at least two of the three major problems with current therapeutics.

The testing of a structure-activity library of 40 substituted QMPs for the resurrection of methylphosphonate-aged AChE yielded great success. In total, 20 of the

40 compounds showed resurrection of the methylphosphonate-aged AChE, with the best compound achieving ~14.5% recovery in 24 hours at a pH of 7.5. Further testing of the top seven compounds revealed that there is a significant dependence of resurrection on the medium’s pH and the ratio of [QMP]:[AChE] in solution. Interestingly, at varying concentrations of [QMP]:[AChE], one compound is shown to be the most effective, recovering 54.9% and 62.6% of the original native activity at pH 7.5 after 24 hours when the [QMP]:[AChE] ratio is increased by 5 and 10 fold, respectively.

Members of this structure-activity relationship library were even more effective with organophosphate pesticides, and many of these compounds recovered >50% in 24 hours. With methyl paraoxon as the pesticide, our QMPs could recover >20% in 1 hr, and with an effective concentration (EC50) of about 100 µM.

The vast improvement in therapeutic efficiency over the course of this research demonstrates the potential of these compounds for the treatment of OP exposure and warrants significant continued investigation.

Acknowledgments

Absolutely none of this would have been possible without the continued support of my family and friends. There were points in times when I no longer wished to continue pursuit of my doctorate degree but the continued support of my immediate family and

Professor Christopher Hadad gave me the motivation and strength to continue forward.

At times this program left me distraught but the love and support of my mother, father, and brother never waivered. I would like to personally thank Chris for his continued guidance both in my chemistry endeavors as well as personal experiences during the course of this program. His belief in my abilities as both a chemist and mentor drove me to continue my career in organic chemistry and the sciences. I also wish to thank Dr.

Christopher Callam for his mentorship in regards to both chemistry as well as graduate school and career pursuits. To Dr. Qinggeng (Albert) Zhuang and Dr. William (Bill)

Coldren, thanks for making lab interactive. I have greatly enjoyed our various conversations whether chemistry related or not. To Dr. Thomas (Tom) Corrigan for his assistance in helping me develop my synthetic skills in the time we managed to work with one another. And to all the current and future group members, I share such a passion for the research conducted in the Hadad lab. Each year the group has grown closer and has proven to be more fun. I wish the best in your future research and I’m excited for the progress to come.

Vita

2014 ………………………………………………...B.S. Chemistry, Wittenberg

University

2014 to present ……………………………………..Graduate Research Associate,

Department of Chemistry and

Biochemistry, The Ohio State

University

Publications

Franjesevic, A. J.; Sillart, S. B.; Beck, J. B.; Vyas, S.; Callam, C. S.; Hadad, C. M.

Resurrection and Reactivation of Acetylcholinesterase and Butyrylcholinesterase.

Chem. Eur. J. 2019, 25, 5337-5371.

Zhuang, Q.; Franjesevic, A. J.; Corrigan, T. S.; Coldren, W. H.; Dicken, R.; Sillart, S.;

DeYong, A.; Yoshino, N.; Smith, J.; Fabry, S.; Fitzpatrick, K.; Blanton, T. G.; Joseph, J.;

Yoder, R. J.; McElroy, C. A.; Dogan Ekici, O.; Callam, C. S.; Hadad, C. M.

Demonstration of In Vitro Resurrection of Aged Acetylcholinesterase after Exposure to

Organophosphorus Chemical Nerve Agents. J. Med. Chem. 2018, 61, (16), 7034–7042.

Yoder, R. J.; Zhuang, Q.; Beck, J. M.; Franjesevic, A.; Blanton, T. G.; Sillart, S.; Secor,

T.; Guerra, L.; Brown, J. D.; Reid, C.; McElroy, C. A.; Dogan Ekici, O.; Callam, C. S.;

Hadad, C. M. Study of para-Quinone Methide Precursors toward the Realkylation of

Aged Acetylcholinesterase. ACS Med. Chem. Lett. 2017, 8, (6), 622–627.

Fields of Study

Major Field: Chemistry

vii

Table of Contents

Abstract ...... iii Acknowledgments ...... v Vita ...... vi List of Tables ...... xi List of Figures ...... xii Chapter 1. INTRODUCTION TO ORGANOPHOSPHORUS COMPOUNDS AND THEIR INHIBITION AND AGING OF ACETYLCHOLINESTERASE ...... 1 1.1 Historical Development of Organophosphorus Pesticides and Nerve Agents ...... 1 1.2 Recent Uses of Organophophorus Pesticides and Nerve Agents for Terrorism...... 7 1.3 Acetylcholinesterase Structure and Function in the Native, Inhibited, and Aged States...... 9 1.4 Butyrylcholinesterase Structure and Function in the Native, Inhibited, and Aged States...... 19 1.5 Reactivation of Acetylcholinesterase...... 24 1.6 Recent Results from Efforts to Develop Broad Scope Reactivators of Acetylcholinesterase...... 25 1.7 Developing Non-Permanently Charged Oxime Reactivators of OP-Inhibited Acetylcholinesterase...... 43 1.8 Developing Non-Oxime Based Reactivators of Inhibited Acetylcholinesterase. ... 54 1.9 Difficulties with the Evaluation of Developed Reactivators...... 62 1.10 Reactivation of Butyrylcholinesterase...... 64 1.11 Recent Results for the Reactivation of Butyrylcholinesterase...... 66 1.12 Resurrection of Aged Acetylcholinesterase...... 77 1.13 Early Attempts at the Realkylation of Aged Acetylcholinesterase...... 78 1.14. Quinone Methides as Alkylating Agents of Model Nucleophiles...... 89 1.15 Mannich Bases for the Alkylation of Aged Acetylcholinesterase...... 95 1.16 Problems with Realkylation of Aged AChE...... 101 1.17 Perspective and Future Work...... 103

viii

Chapter 2. COMPUTATIONAL INSIGHTS INTO THE ALKYLATION REACTIONS OF PYRIDINE AND PYRIDINIUM QUINONE METHIDE PRECURSORS: AN EFFORT TO REVERSE THE EFFECTS OF ACETYLCHOLINESTERASE FOLLOWING AGING ...... 107 2.1 Introduction...... 107 2.2 Experimental...... 115 2.3 Mechanistic Study ...... 116 2.4 Docking Study...... 117 2.5 Results and Discussion ...... 120 2.6 Conclusions ...... 137 2.7 Future Work and Adjustments...... 137 Chapter 3. DEVELOPMENT AND TESTING OF QUINONE METHIDE PRECURSORS FOR THE RESURRECTION OF AGED ACETYLCHOLINESTERASE USING A 3-HYDROXYPYRIDINE FRAMEWORK 140 3.1 Introduction ...... 140 3.2 Computationally Guided Selection of realkylator Candidates...... 143 3.3 Screening of Reakylator Library...... 145 3.4 Kinetics of Resurrection...... 149 3.5 Bottom-Up Proteomics ...... 150 3.6 Reactivation of Inhibited AChE by Five QMPs...... 152

3.7 pH Effect and EC50 for C8 Activity...... 153 3.8 Resurrection of Aged Human AChE after exposure to DFP...... 156 3.9 Discussion and Conclusions...... 156 3.10 Experimental Section...... 158 Chapter 4. STRUCTURE-ACTIVITY RELATIONSHIPS FOR SUBSTITUTED 3- PYRIDINOL MANNICH BASES IN THE RESURRECTION OF METHYLPHOSPHONATE-AGED HUMAN ACETYLCHOLINESTERASE ...... 165 4.1 Introduction ...... 165 4.2 Resurrection of rhuAChE after aging with CMP, a methylphosphonate nerve agent analogue ...... 169

4.3 pH Dependenc and EC50 determination of the best QMPs ...... 174

4.4 IC50 Determination of rhuAChE ...... 177 4.5 QMP to AChE Ratio Dependence on the Efficacy of Resurrection ...... 178 4.6 Computational Investigation of the pH Dependence and Thermodynamics for QM Formation ...... 180 ix

4.7 Important Interactions Leading to Resurrection Efficacy ...... 184 4.8 Conclusions ...... 189 4.9 Experimentals ...... 192 Chapter 5. SUBSTITUTED 3-PYRIDINOL MANNICH BASES FOR THE RESURRECTION AND REACTIVATION OF HUMAN ACETYLCHOLINESTERASE ...... 198 5.1 Introduction ...... 198 5.2 Resurrection of rhuAChE After Aging With DFP ...... 202 5.3 Resurrection of rhuAChE After Aging With Methyl Paraoxon ...... 204 5.4 pH and Time Dependence on the Resurrection of Methyl Paraoxon-Aged rhuAChE ...... 206 5.5 ADME-Tox Determination Using Computational Techniques ...... 207 5.6 Experimentals ...... 208 Bibliography ...... 211

List of Tables

Table 1. Aging half-times for various organophosphorus compounds.35 ...... 15 Table 2. Protonated and deprotonated energies and dihedrals of the PAM library...... 121 Table 3. Energies and bond distances resulting from the energy profile scans with defined stationary points for an SN1 mechanism...... 124 Table 4. The energy and bond distances associated with an energy profile scan assuming an SN2 mechanism...... 125 Table 5. Docking percentages for the PAM library...... 133 Table 6. Percentages of eeAChE species after resurrection by C8, as determined by LC- MS/MS...... 151 Table 7. pH dependence of EC50 (mM) and the maximum percentage (%) at the plateau for the top seven compounds as shown in Figure 82...... 175 Table 8. Relative free energy (ΔG298, kcal/mol) of the net neutral states at the B3LYP(SMD,water)/6-31+G* level of theory.a ...... 182 Table 9. Relative free energy (ΔG298, kcal/mol) of the net positive charge states at the B3LYP(SMD,water)/6-31+G* level of theory.a ...... 183 Table 10. Center of mass distance (Å) between QMP and phosphylated-Ser averaged over the 10 ns MD trajectories and percentage of time that the benzylic carbon – oxyanion distance is less than 4 Å (3 Å in parentheses) for single trajectories with the greatest time spent under 3 Å for each protonation state.a ...... 185

List of Figures

Figure 1. Select examples of G-series organophosphorus agents...... 4 Figure 2. Select examples of V-series organophosphorus nerve agents...... 5 Figure 3. Select examples of organophosphorus pesticides in their oxon form...... 7 Figure 4. Catalytic acive site of AChE (S203, H447, E334), the oxyanion hole (G121, G122, A204), and cation-binding pocket (W86, E202) without ACh being bound (left) and with ACh (right)...... 10 Figure 5. Hydrolysis of acetylcholine by native AChE...... 12 Figure 6. Aromatic residues (left) composing the gorge bottleneck and a space filling model (right) showing the width of the gorge...... 13 Figure 7. Inhibition and aging of AChE with a model phosphonate nerve agent...... 16 Figure 8. Examples of oximes administered for treatment of organophosphorus poisoning...... 18 Figure 9. Visual comparison of the different widths and volumes of the (left) AChE and (right) BChE gorges...... 21 Figure 10. Novel lipohilic pyridinium oximes.95 ...... 27 Figure 11. Novel oxime structures synthesized using the Ugi multicomponent reaction.96 ...... 30 Figure 12. Tetroxime structure as studied by Musilek and coworkers.97 ...... 31 Figure 13. bis-Pyridinium compounds capable of reactivating AChE and BChE with (ethyl) paraoxon.98 ...... 32 Figure 14. (E)-2-Butene linked bis-pyridinium oximes for the reactivation of tabun inhibited AChE.99 ...... 34 Figure 15. Two novel K-oximes showing some efficiency for the reactivation of (ethyl) paraoxon inhibited hAChE.100 ...... 35 Figure 16. ZINC database oximes determined from virtual screening which are capable of reactivating multiple OP pesticides.101 ...... 37 Figure 17. Hydroxyiminoacetamide nucleophiles using structural components of HI-6 for reactivation of sarin and VX-inhibited AChE.102 ...... 38 Figure 18. Imidazolium oxime for the reactivation of sarin and VW-inhibited AChE.103 39 Figure 19. Hydroxyiminoacetamide nucleophiles using structural components of HI-6 for the reactivation of sarin- and VX-inhibited AChE.104 ...... 43 Figure 20. Novel oxime structures based on salicylaldoximes.106 ...... 45 Figure 21. Novel oxime structure that shows significant tabun reactivation.107 ...... 46 Figure 22. Novel oxime structure that shows some soman reactivation.108 ...... 47 Figure 23. Novel oxime structures based on the Alzheimer's Diseas drug donepezil by Renard and coworkers.109 ...... 48 xii

Figure 24. Uncharged reactivators of cholinesterases with the potential to cross the blood- brain barrier.110 ...... 51 Figure 25. Tacrine-linked reactivators designed on the basis of X-ray crystallographic data.111 ...... 53 Figure 26. Vitamin-B6 based oximes for the reactivation of hAChE and hBChE.112 ...... 54 Figure 27. Non-oxime reactivators.113,114 ...... 55 Figure 28. Additional reactivator classes.91 ...... 60 Figure 29. More reactivat Mannich base for the reactivation of AChE.116 ...... 62 Figure 30. Imidazolium-based oximes for the reactivation of OP-inhibited BChE.139 .... 68 Figure 31. Cinchona based oxime showing broad scope reactivation potential.140 ...... 70 Figure 32. Mannich base for the selective reactivation of paraoxon-inhibited BChE.91 .. 71 Figure 33. Edrophonium derived derivative TAB2OH for the reactivation of hBChE.141 ...... 73 Figure 34. Imidazole based oximes found by conducting structural modifications from the originally tested imidazole aldoximes by Radic and coworkers.142 ...... 75 Figure 35. K Oximes capable of reactivating tabun-inhibited BChE.143 ...... 76 Figure 36. Sulfonate alkylator in first attempt to alkylate phosphonate anions.150 ...... 79 Figure 37. Strong alkylator groups used unsuccessfully in vitro to realkylate soman-aged AChE.149...... 81 Figure 38. N-methyl-2-methoxypyridnium structure and model phosphonate anion used to study methyl transfer by Quinn and coworkers.147 ...... 82 Figure 39. Various core frameworks and peripheral site ligands employed for hAChE inhibition by Quinn and coworkers.148 ...... 84 Figure 40. Computationally investigated sulfonium realkylators of aged AChE.146 ...... 86 Figure 41. Computationally investigated aminoalcohol realkylators of aged AChE.145 .. 89 Figure 42. Initial reaction showing the potential of QM alkylation of phosphodiesters.154 ...... 91 Figure 43. Subsequent study showing the potential of QM alkylation of phosphodiesters.155 ...... 92 Figure 44. QMP structures used to study the aspects of QM reaction by Rokita and coworkers.156 ...... 95 Figure 45. QMP structures used in reaction with nucleophiles for proof of principle alkylation.159 ...... 97 Figure 46. Nerve agent analogues used in biological assays.160 ...... 99 Figure 47. Lead QMP structure for resurrection of aged AChE.160 ...... 101 Figure 48. The entire acetylcholinesterase enzyme with a quinone methide precursor bound within the active site...... 109 Figure 49. A model phosphonate (ball and stick, left, line structure, right) used to represent the aged phosphonic-Ser residue of the AChE active site following the dealkylation reaction...... 113 Figure 50. The proposed formation of a QM from a QMP that would then result in the realkylation of the phosphonic anion of the aged-Ser residue...... 115 Figure 51. Examples of quinone methides (ortho, left, and para, right) and the known AChE inhibitor edrophonium...... 115 xiii

Figure 52. Library of 2-PAM derivatives evaluated based on molecular docking and thermodynamics of reaction...... 116 Figure 53. Example docking histogram showing the population of multiple binding modes with a potential maximum of 200. The number of poses satisfying the distance and hinderance criteria were then used to determine a percentage of 'binding.' ...... 119 Figure 54. A pose satisfying the distance criteria and showing no steric hinderance between reactive centers (A) and a pose only satisfying the distance criteria (B)...... 120 Figure 55. A representative measurement of the dihedral angles reported with PAM9 121 Figure 56. An asymptotic energy profile for PAM1...... 123 Figure 57. An energy profile scan for PAM2 with a more well defined stationary point...... 124 Figure 58. Energy diagram for an SN1 mechanism of neutral pyridine PAM scaffolds. 127 Figure 59. Energy diagram for an SN1 mechanism of protonated pyridine PAM scaffolds...... 128 Figure 60. Energy diagram for an SN1 mechanism of methylated pyridine PAM scaffolds...... 129 Figure 61. Energy diagram for an SN2 mechanism of neutral pyridine PAM scaffolds. 130 Figure 62. Energy diagram for an SN2 mechanism of protonated pyridine PAM scaffolds...... 131 Figure 63. Energy diagram for an SN2 mechanism of methylated pyridine PAM scaffolds...... 132 Figure 64. Most commonly observed docking pose of the studied PAM library. The compound pictured is PAM3...... 134 Figure 65. Second most commonly observed docking pose of the studied PAM library. The compound pictured is PAM3...... 135 Figure 66. And additional commonly observed docking pose of the studied PAM library. The compound pictured is PAM9...... 136 Figure 67. Structures and substitution reaction of QMs and QMPs. Typical structures of QMs (top) and QMPs (bottom). Nucleophiles can substitute the leaving group of QMP by either an SN2 reaction or formation of the corresponding QM...... 143 Figure 68. A snapshot obtained from a 1 ns MD simulation, demonstrating a protonated QMP (shown in cyan, later called C8) near the active site of aged AChE (wall-eyed stereo). The electrophilic carbon is ~4.2 Å from the phosphylated oxyanion. Hydrogen bonds with short contact distances are shown (green dashed lines)...... 144 Figure 69. Structures of realkylator candidates...... 144 Figure 70. Structures of OPs used for aging of AChE and the corresponding aged adducts...... 146 Figure 71. Screening of three concentrations of various C series QMPs against (a) methylphosphonate-aged eeAChE and (b) isopropyl phosphate-aged eeAChE. The horizontal solid and the dashed lines mark the negative controls and 2-PAM controls, respectively. The error bars reflect standard deviations from four replicate efforts...... 147 Figure 72. Kinetics of realkylation of aged eeAChE by 4 mM C8 at pH 8 over 4 days for (a) isopropyl phosphate-aged (DFP-treated) eeAChE and (b) methylphosphonate-aged

xiv

(PiMP-treated) eeAChE. The blue dashed lines illustrate the result of linear regression...... 150 Figure 73. Reactivation activity test of four QMPs (4 mM) against eeAChE inhibited by EMP. (a) Structure of EMP. (b) Relative activity of eeAChE as reactivation proceeded...... 152 Figure 74. Resurrection efficiency of 4 mM C8 against aged eeAChE increased with pH within the range of pH 6~9. (a) Methylphosphonate-aged (PiMP-treated) eeAChE. (b) Isopropyl phosphate-aged (DFP-treated) eeAChE...... 153 Figure 75. Influence of C8 protonation states on spectra. (a) 1H NMR spectra of the aromatic protons and (b) UV-vis spectra of C8 as pH is varied from 6~9...... 154 Figure 76. (a) The four most probable protonation states of C8 at pH 8-9. (b) A representative snapshot in the MD simulation of C8c (cyan, wall-eyed stereo). Hydrogen bonds with short contact distances are shown (green dashed lines)...... 155 Figure 77. Kinetics of resurrection of isopropyl phosphate-aged huAChE by 4 mM C8 (pH 9). The green dashed line illustrates the result of linear regression...... 156 Figure 78. (A) Structures of various authentic organophosphorus (OP) nerve agents and CMP, an OP simulant for GF. (B) General schematic of the inhibition and aging process of AChE, following exposure to a methylphosphonate OP...... 167 Figure 79. Mannich bases capable of reactivation or resurrection of AChE after exposure to OP nerve agents...... 168 Figure 80. (A) The numbering scheme used to refer to substituted quinone methide precursors (QMPs) and the hypothesized route of re-alkylation of methylphosphonate- aged AChE with the quinone methide (QM) intermediate. (B) QMP library of compounds synthesized for testing of resurrection of methylphosphonate-aged human AChE. (C) A general scheme for the Mannich reaction that was used to synthesize the QMPs from the 3-hydroxypyridine frameworks. The three-step synthetic route used to synthesize the 5- methyl-3-hydroxypyridine framework for compounds 11a-e...... 170 Figure 81. Resurrection activity of QMPs against a methylphosphonate-aged (CMP) rhuAChE (37°C, pH 7.5, 24 hours). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figures 2 and 3 for structures. Two different 2-PAM concentrations were used: 100 µM and 1 mM, respectively, left to right. The dashed line is the activity recovered by 1 mM 2-PAM of an aged control that was not treated with a QMP and is representative of oxime reactivation of any residual OP-inhibited rhuAChE; thus, any entries above this level demonstrate recovery of CMP-aged rhuAChE activity. The error bars reflect a standard deviation of 4 replicate measurements...... 173 Figure 82. EC50 evaluation for a methylphosphonate-aged rhuAChE (1.4 U for all samples, 0.14 U for positive control) after resurrection. Selected QMP concentrations (0– 5 mM) were incubated for 24 hours and 37 °C at (A) pH 7.5 and (B) pH 9.0. The error bars reflect a standard deviation of 4 replicate measurements...... 177 Figure 83. EC50 evaluation against a methylphosphonate-aged rhuAChE (5-fold rhuAChE is 0.028 U and 10-fold is 0.014 U). Selected QMP concentrations (0–5 mM) were incubated for 24 hours and at 37 °C for (A) 5-fold diluted rhuAChE and (B) 10-fold xv diluted rhuAChE. The error bars reflect a standard deviation of 4 replicate measurements...... 179 Figure 84. Examples of the protonation states analyzed for the formation of the QM intermediate using compound 6d as a representative molecule...... 181 Figure 85. At the B3LYP(SMD,water)/6-31+G* level of theory, the free energy (ΔG298) for formation of the quinone methide (QM) intermediate in (kcal/mol) from the two most stable/favorable net neutral states: non-zwitterionic neutral and zwitterionic aminium (Z- Am) species, and the two most stable/favorable net positive states: aminium (Am) and zwitterionic pyridinium-aminium (Z-Pyr-Am) species...... 183 Figure 86. Active site showing van der Waals surfaces of Tyr124 and Ser125 hydroxyl groups and methyl group in blue for (A) modified 11b (in orange) with addition of 4- methyl (11b*), (B) modified 8b with addition of 4-methyl (8b*) and (C) unmodified 6b. All images are taken from the same timepoints as the snapshots shown in in the supporting information...... 187 Figure 87. (A) Structure of generic methylphosphonate OPs, accounting for most of the G- and V-series nerve agents and the generic structure of an organophosphate OP accounting for the pesticides. (B) Structures of various OP pesticides...... 200 Figure 88. Resurrection activity of QMPs against a phosphate-aged (DFP) rhuAChE (37°C, pH 7.5, 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figure 80 for structures. The 2-PAM concentration was 1 mM. The error bars reflect a standard deviation of 4 replicate measurements...... 203 Figure 89. Resurrection activity of QMPs against a phosphate-aged (methyl paraoxon) rhuAChE (37°C, pH 7.5, 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figure 80 for structures. The 2-PAM concentration was 1 mM. The error bars reflect a standard deviation of 4 replicate measurements...... 204 Figure 90. Resurrection activity of QMPs (1 mM) against a phosphate-aged (methyl paraoxon) rhuAChE (37°C, pH 7.5 (A) and pH 9.0 (B), at 1, 3, and 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP was used at a concentration of 1 mM – see Figure 80 for structures. The error bars reflect a standard deviation of 4 replicate measurements...... 206

xvi

Chapter 1. INTRODUCTION TO ORGANOPHOSPHORUS COMPOUNDS AND THEIR INHIBITION AND AGING OF ACETYLCHOLINESTERASE

The research presented in this chapter is adapted from the manscript that is published as Franjesevic, A. J.; Sillart, S. B.; Beck, J. M.; Vyas, S.; Callam, C. S.; Hadad, C. M. Chem. Eur. J. 2019, 25, 5337–5371. The majority of the content is due to the first author, but this publication does contain very limited components from the Ph.D. theses of Dr. Jeremy Beck and Dr. Shubham Vyas as well as content from Ms. Sydney Sillart’s undergraduate research thesis.

1.1 Historical Development of Organophosphorus Pesticides and Nerve Agents

Following the First World War, Germany found itself in a state of complete distress and devastation. The Third Reich strategists desired to become more self- sufficient during this time of desolation and strove to reduce Germany’s reliance on imported food. Gerhard Schrader, a chemist working at the I.G. Farben chemical company, had been assigned the task of designing and synthesizing new insecticides in efforts to protect food production.1 The first pesticides were based on fluorine and sulfur, but proved to be ineffective, thus Schrader moved his focus to phosphorus and cyanide derivatives.

On December 23, 1936, Schrader synthesized “Preparation 9/91”, which proved to be extremely toxic and deadly. In fact, Schrader himself was hospitalized after working with small quantities of the sample, exhibiting a variety of symptoms including difficulty breathing, impaired vision, and dizziness.2 Schrader’s coworkers were also inadvertently exposed to the compound and displayed similar symptoms – all who were

1 exposed took weeks to recover. The sample to which they were exposed was ethyl-(N,N- dimethylamido)-phosphoro-cyanidate, known today as tabun (Figure 1). The name is derived from the German word for taboo, and indeed, tabun affected the nervous system in a way that the victim’s bodily functions were no longer under the brain’s control. An early sample was administered as a vapor to apes, and resulted in lethal effects. This lethality was primarily observed in mammals and not insects, making tabun a poor insecticide, despite Schrader’s initial intent. However, due to its obvious toxicity to humans, I.G. Farben alerted the German military concerning the potential of this compound to be weaponized. Schrader had inadvertently discovered a new class of toxic chemicals that are now more commonly referred to as nerve agents. Tabun (GA) became the first in the G-series, “G” standing for “German” (Figure 1).

Despite the peace accord via the Treaty of Versailles, German scientists continued weapon development. During WWI, some of the utilized chemical warfare agents, such as phosgene and mustard gas, took hours to days to cause lethality. The potential of this new organophosphorus nerve agent was recognized, as lethality occurred within 20 minutes. Scientists began studying the physiological effects of tabun and how to further increase its lethality. In June 1939, Schrader developed Substance 146, or isopropyl- methyl-phosphono-fluoridate. This compound is now more commonly referred to as sarin, an acronym for its four creators: Schrader, Ambos, Rüdiger, and Van der Linde

(Figure 1). Although more difficult to synthesize, sarin (GB) was found to be 500 times more lethal than cyanide.

In 1943, Richard Kuhn was hired to research the mechanism of action of these compounds. Kuhn determined that these compounds inhibit acetylcholinesterase (AChE), resulting in the buildup of acetylcholine in synapses and the subsequent prevention of electrical termination of signals to muscles in the body, due to the muscle cells being overstimulated by excess neurotransmitter. Kuhn screened a wide variety of organophosphorus agents to test the various levels of inhibition of AChE. Upon replacing the isopropyl group with a pinacolyl group, he discovered an even more potent nerve agent than tabun, one that was even twice as lethal as sarin. Compound 25075 or 3,3- dimethyl-butan-2-yl-methyl-phosphono-fluoridate (soman, GD), deactivated AChE within two minutes and readily penetrated the skin to further increase lethality (Figure

1).3

Fortunately, Germany did not deploy these nerve agents during WWII, although they had significant stockpiles.4 Shortly after the war concluded, Russia found evidence of these chemical agents by uncovering old lab notebooks that detailed the synthesis of sarin. During the Cold War in the 1950’s, the United States and United Kingdom collaborated to develop analogous nerve agents. They screened nerve agents, like sarin and cyclohexyl-methyl-phosphono-fluoridate (cyclosarin, GF), and synthesized new sarin-like derivatives to determine which compounds would function as the best weapons

(Figure 1). The most expensive agent, soman, was determined to be the most lethal agent, but as sarin was still sufficiently toxic and cheaper to synthesize, sarin was developed and stockpiled instead.

O O P P N O O F N tabun (GA) sarin (GB)

O O P P O O F F

soman (GD) cyclosarin (GF) Figure 1. Select examples of G-series organophosphorus agents.

Following the conclusion of WWII, the field of insecticide and pesticide research became increasingly popular. In 1952, a researcher from the Imperial Chemical

Industries, Ranajit Gosh, discovered a new class of nerve agents. Gosh and Newman were investigating the use of organophosphorus agents as potential pesticides by synthesizing compounds containing nitrogen and sulfur. One such compound was able to effectively kill lice and was placed on the market under the trade name Amiton.5

Eventually, the company had to remove Amiton (O,O-diethyl-S-(2-(diethylamino)ethyl)- phosphoro-thioate) from the market due to its toxicity. This compound became part of what is known as the V-series, “V’ meaning “venomous”, and is referred to as VG

(Figure 2). Currently, structurally similar organophosphates, such as echothiophate, are used as anti-gluacoma treatments, despite having some toxicity.6

A similar compound, ethyl-N-2-diisopropyl-aminoethyl-methyl-phosphono- thiolate, was developed in the partner company in the US, later renamed VX (Figure 2).

Some other well-known isomers include S-(diethylamino)ethyl-O-ethyl-ethyl-phosphono- thioate (deemed VE) and S-(2-(diethylamino)ethyl-O-ethyl-methyl-phosphono-thioate

(VM, Figure 2). Studies then showed the V-series agents to be far deadlier than previously discovered agents. Under average temperature conditions, these compounds persist for days, have no odor, can be administered as a vapor or gas, and only require 20

µg/kg to be lethal, rendering these agents to be the mostly deadly known to mankind at the time of their discovery.7 VX proved to be stable for months in cold temperatures, so it was selected as the US’s weapon of choice and thus stockpiled during the Cold War

(Figure 2).

The Soviet Union began their chemical weapon development to try and remain on an equal level to the US. At the Scientific Research Institute No. 42, Ivin, Soborovsky, and Shilakova developed Substance 33, an isomer of VX (Figure 2).8 N,N-Diethyl-2-

(methyl-(2-methylpropoxy)phosphoryl)-sulfanyl-ethanamine (VR or “Russian VX”,

Figure 2) was found to have a similar lethal dose to VX, while reducing the treatment window due to its rapid inhibition of AChE.

O O O N P S N P N P O S S O O VX VR CVX

O O O N P N P N P S S S O O O O

VM VE VG (Amiton) Figure 2. Select examples of V-series organophosphorus nerve agents.

After WWII, the US began to develop organophosphorus pesticides in large quantities. Diethyl-(dimethoxy-phosphino-thioyl)-thio-butanedioate, also known as malathion, was discovered in 1950. In 1951, Schrader’s continued development of new insecticides resulted in the synthesis of the insecticide demeton. Demeton is a mixture of the thiono- and thiolo-isomers of O,O-diethyl-2-ethylmercaptoethyl-phosphoro-thioate, thereby introducing a new class of insecticides possessing a thioether group. Today, a wide range of organophosphorus compounds is still available as pesticides. Subsequently, other compounds, such as chlorpyrifos, paraoxon, and tebupirimphos, were synthesized, showing lower environmental persistence as compared to previously synthesized pesticides. Their popularity increased after the ban of organochlorine insecticides in the

1970s. Organophosphorus pesticides are used in very low concentrations in order to pose less harm to users and food. Though dilute, there is still potential for these pesticides to cause harm to agricultural workers who handle these chemicals. The US Food and Drug

Administration has lowered the limits of pesticides in foods sold to consumers in the US, but the world has varying levels of regulation, thus leading to 3 million global cases of pesticide poisoning per year, with an estimated 220,000 deaths annually.9 The majority of these cases are in developing countries, and most cases are a result of poor working conditions, improper handling, inadequate regulation, or intentional self-harm.

O N 2 O O O P P S P O O F O S O O O O

Ethyl Paraoxon DFP Demeton

O N Cl Cl O O 2 O O O O P P P O O Cl N O O O S O O O O

Methyl Paraoxon Chlorpyrifos Oxon Malaoxon Figure 3. Select examples of organophosphorus pesticides in their oxon form.

1.2 Recent Uses of Organophophorus Pesticides and Nerve Agents for Terrorism.

During the Iran-Iraq War in the 1980s, Iraq used chemical warfare agents against

Iran. Iraq claims to have used 600 tons of sarin and 140 tons of tabun against enemy forces.10 The onslaught killed nearly 5,000 Iranians and over 100,000 were hospitalized.

In 1988, Iraq even attacked its own citizens in Halabja, killing over 5,000 and injuring over 7,000.

On March 20, 1995, cult members from the Aum Shinrikyo sect punctured bags of homemade sarin on a subway in Tokyo, Japan.11 Five plastic bags of liquid sarin were deployed during rush hour in order to “speed up the pending apocalypse”. Though only a dozen people were killed, over 5,500 were hospitalized. If the sarin was deployed in a different manner, it is hypothesized that the release could have killed thousands.

Syria has made use of organophosphorus compounds as part of its civil war over the past few years. Damascus in Syria was the location of a sarin attack on the morning of

August 21, 2013, when rockets filled with the agent struck the rebel suburbs of the 7 capital, killing and injuring thousands.12 Around 1,429 were found dead, including 426 children. Syria was the site of another attack on April 4, 2017 – more than 89 people were killed and another 541 injured in the rebel-held town of Khan Sheikhoun after the

Syrian government’s air strike.13 Traces of one of the decomposition products of sarin, isopropyl-methyl-phosphonic acid, were detected in the urine and blood of some of the victims.

On February 13, 2017, Kim Jong-nam was murdered in an airport in Kuala

Lumpur. Kim is the half-brother of Kim Jong-un, the current leader of North Korea.14

Kim Jong-nam was attacked by two women who smeared VX on his face, and he died shortly thereafter, while on the way to the hospital.

On March 4, 2018, a Russian spy, Sergei Skripal and his daughter were found unconscious on a bench in the UK.15 Both were hospitalized and exhibited symptoms of organophosphorus nerve agent exposure. Reports detail that the Skripals were exposed to a nerve agent called Novichok, a compound that, at present, has an unreleased structure.

The substance was reportedly placed on the front door of the Skirpal’s home using a modified perfume bottle where the Skripals later contacted the substance. An additional

21 doctors, first responders, and bystanders had to be treated for exposure as well. Later, on June 30, 2018, a second Novichok poisoning was determined that left Dawn Sturgess dead and hospitalized Charlie Rowley in Amesbury. A perfume bottle was again the mode of exposure. Rowley was exposed to some while putting the applicator and bottle together and Sturgess sprayed the perfume (agent) directly on her wrists.15

Although nerve agents receive the most publicity, organophosphorus pesticides still present significant risk for terrorist use, but they also continue to result in incidental ingestion or exposure. For instance, in 2013, 23 school children in India were killed after ingesting monocrotophos, which had contaminated their school lunch. The cooking oil used to prepare the meals was contaminated by the pesticide through storage of the cooking oil in a container that previously had been used for pesticide storage, resulting in the death of the children.16 This problem seems to be of major concern as the diet of children in India has been found to contain almost 40 times higher concentrations of pesticides as compared to the US, based on the identification of metabolites in children’s urine.17 The ease of access to such pesticides and the incidents caused by contamination present significant risk to civilian populations.

Additional incidental exposures have been noted in air travel professionals, in addition to passengers, and is now referred to as aerotoxic syndrome. The proposed cause of the toxicity is the exposure to tri-o-cresyl phosphate (TOCP), a component of jet hydraulic fluids and engine oils, which is converted physiologically to the toxic metabolite 2-(o-cresyl)-4H-1,3,2-benzodioxaphosphoran-2-one (CBDP). CBDP is a potent inhibitor of cholinesterases and is believed to lead to the neurological signs that are characteristic of aerotoxic syndrome.18,19

1.3 Acetylcholinesterase Structure and Function in the Native, Inhibited, and Aged States.

Acetylcholinesterase (AChE) is an extremely efficient enzyme located throughout the body, namely at the synapses of the central nervous system, neuromuscular junctions in the peripheral nervous system, and bound to erythrocyte membranes in blood.20 AChE 9 is involved in the neurosynaptic communication process, specifically the process of maintaining proper levels of the neurotransmitter acetylcholine (ACh) at the synaptic cleft. The enzyme accomplishes this regulation via catalytic hydrolysis of ACh, forming acetate and choline, which are then used to regenerate ACh in the peripheral nerve. The catalytic efficiency (kcat/KM) for ACh hydrolysis in humans, a measure of the catalytic

9 – hydrolysis rate (kcat) corrected for binding affinity of the substrate (KM), is 1.50×10 M

1s–1, and the turnover rate of the enzyme equates to hydrolyzing 25,000 molecules of ACh per second.21 This equates to a rate that is near the diffusion-cntrolled limit, or the rate at which the enzyme and substrate can collide in aqueous solution. Meaning the enzyme is extremely well evolved and the binding step has a near zero effect on the rate of diffusion of the pair to one another.

Figure 4. Catalytic acive site of AChE (S203, H447, E334), the oxyanion hole (G121, G122, A204), and cation-binding pocket (W86, E202) without ACh being bound (left) and with ACh (right).

The active site of AChE was confirmed by crystallographers in 1991, and consists of a catalytic triad of serine (S203), histidine (4H47), and glutamate (E334) (Figure 4).

This catalytic triad is common to a variety of enzymes classified as serine hydrolases.22,23 10

In addition to the catalytic triad, two subsites aid in the hydrolysis of acetylcholine: the oxyanion hole and the cation-binding pocket (Figure 4). The oxyanion hole is comprised of the backbone peptide N–H groups of G121, G122, and A204. These residues are aligned to form strong hydrogen bonds with the carboxyl oxygen of ACh and aid in stabilizing the negative charge generated during the hydrolysis reaction. The cation- binding pocket is comprised of W86 and E202, and stabilizes the choline moiety in the active site via cation-π interactions. Acetylcholine, upon reaching the catalytic site, undergoes nucleophilic attack at the carbonyl carbon by S203, forming a covalent bond with the enzyme. Through subsequent reactions with water activated in the active site,

ACh is cleaved into acetate and choline, and the native AChE enzyme is regenerated

(Figure 5). An additional peripheral binding site is located at the entrance of the AChE gorge and is capable of binding cationic and aromatic substrates. This site of AChE has inspired a great deal of research as a location by which a ligand can be tethered to a drug candidate to increase overall binding affinity.

O N N O HO H O O O N HN N NH NH NH O H O O

AChE AChE AChE

O O OH HO N H N OH NH O NH

AChE AChE

Figure 5. Hydrolysis of acetylcholine by native AChE.

Given the rapid hydrolysis rates of ACh, the cholinesterase community was surprised when researchers reported that the active site does not reside on the protein surface, but is in fact buried in the interior of the protein, connected to the exterior by a narrow gorge roughly 20 Å deep.23 Interestingly, the gorge is so narrow that in the crystallographic orientation, the narrowest point, referred to as the gorge bottleneck, is too compact to accommodate the movement of the substrate from the exterior of the protein to the buried catalytic site (Figure 6).24,25

In order for the enzyme to operate at near-diffusion limits with an active site buried at the bottom of a narrow 20 Å deep gorge, substrate trafficking into the enzyme’s interior must be extremely efficient. Although in the crystallographic orientation, the gorge is too narrow for a substrate to pass from the protein surface into the catalytic site, molecular dynamics simulations have suggested a ‘breathing’ motion that occurs on a 10

12 ns period, a frequency that is sufficient for the hydrolysis of 25,000 molecules of ACh per second.26 This breathing motion is a combination of residue side chain motions as well as concerted motion of two loops in the AChE structure: the Ω-loop and acyl loop.

The Ω-loop and acyl loops make up two of the walls that define the AChE gorge, spanning from the mouth of the gorge down to the catalytic site. As the two loops move, the gorge radius has been observed to fluctuate between 0.75 Å and 2.5 Å, although in the absence of substrate, the gorge is rarely wide enough to accommodate ACh transfer.26

Figure 6. Aromatic residues (left) composing the gorge bottleneck and a space filling model (right) showing the width of the gorge.

As the gorge bottleneck can fluctuate to allow substrate into the catalytic site of

AChE, the mechanism to transfer substrate from the bulk solution to the protein interior has been a subject of significant investigation. Association of ACh with the enzyme is an obvious requirement for rapid catalysis and occurs via a binding site at the mouth of the gorge, which has been named the peripheral anionic site (PAS).27,28 The PAS is a region of the protein with an abundance of aromatic residues that forms cation-π interactions with the choline portion of ACh, and includes W286, Y72, Y124 and the anionic D74. 13

Previous studies have identified a large electrostatic dipole, calculated at 505 Debye, along the gorge axis of AChE that essentially pulls ACh from the PAS at the enzyme surface, past the gorge bottleneck, and into the catalytic site in the protein interior.29

However, this electrostatic dipole would also seem to prohibit exit of the choline hydrolysis product, trapping it in the gorge.30 Researchers have long debated the presence of a “back door” to the catalytic site, with most interest being focused on a thin area of the gorge wall along the Ω-loop region (C65–C92) and with W86 acting as a gate- keeper.31 A computational study that modeled the behavior of the reaction product thiocholine in the AChE gorge observed three separate paths for thiocholine to exit the catalytic site. In the majority of simulations, totaling 27 out of 40 trajectories, thiocholine exited through the backdoor when the residues W86, V132, and G448 opened the pathway via concerted motions.32 Recent work involving kinetic crystallography experiments have also confirmed the fluctuations leading to backdoor opening in the

AChE gorge, thereby corroborating the computational hypothesis.33,34

In understanding the catalytic cycle of acetylcholine hydrolysis by AChE, the reason for the toxicity of organophosphorus (OP) compounds becomes apparent. OPs, upon reacting with the catalytic serine and dissociation of their leaving group, form a tetrahedral phosphonate (or phosphate) which resides in the catalytic site of AChE and prevents any further ACh hydrolysis (Figure 7). As nucleophilic attack at phosphorus is significantly slower than at carbon, water in the active site is not sufficiently nucleophilic to react with the phosphylated serine’s P center in order to cleave the covalently bound

OP from the catalytic site. Thus, covalent modification of the catalytic S203 residue

14 results in the inability of AChE to bind and hydrolyze ACh, thereby leading to a rapid buildup of ACh and subsequent overstimulation of ACh receptors.

Table 1. Aging half-times for various organophosphorus compounds.35

OP Aging Half-time (h) Tabun (GA) 19.2 Sarin (GB) 3.0 Soman (GD) 0.07 Cyclosarin (GF) 7.0 VX 36.5 VR 138.6 CVX 32.2 Ethyl Paraoxon 31.5 Methyl Paraoxon 3.7

OP nerve agent deactivation of AChE occurs in two separate stages. The first stage, referred to simply as inhibition, is the formation of a covalent P–O(Ser) bond between the OP and catalytic serine of AChE (Figure 7). At this point, introduction of strong nucleophiles to the active site, mainly in the form of pyridinium oximes (Figure 8), can cleave the P–O(Ser) bond, regenerating the catalytic activity of the enzyme, a process called reactivation (Figure 7). The second inhibitory state was observed in early experimental studies, when the efficacy of oximes at cleaving the P–O(Ser) bond would decrease over time, and the enzyme was therefore referred to as “aged” (Figure 7).36

Aging was later determined to correspond to a secondary reaction of the OP-AChE adduct which for the common nerve agents comprises O-dealkylation of the phosphylated center.37 The O-dealkylation results in an anionic phosphylated serine residue, rendering

15 the aged form to be resistant to nucleophilic attack by oximes. Strong hydrogen-bonding interactions with H447 have been postulated for stabilizing this anionic phosphylated serine residue.38,39 After the enzyme has undergone aging, it was considered to be un- reactivatable (until 2018, more on that later). Aging rates (Table 1) vary between OPs, but the range of aging half-times can span from roughly 37 hours (VX) to a mere 4 minutes (GD).40,41

X O X R O P O 2 R2 O R O R2 O 1 P P O P O H N HN HN N NH R1 NH NH NH O H O R1 O R1 O dealkylation

AChE AChE AChE AChE

R R N R O O N N P O O O O O R2 R2 R1 R 2 P O P O N H N reactivation HN R OH NH 1 O NH R1 O NH

AChE AChE AChE Figure 7. Inhibition and aging of AChE with a model phosphonate nerve agent.

Wilson first reported the concept of reactivation in 1951 for OP-inhibited AChE, as he showed that incubation of two day old OP-inhibited AChE with concentrated solutions of choline or hydroxylamine for a few days regenerated about 75–90% native

AChE.42 Wilson further showed that hydroxamic acids are better nucleophiles and can regenerate 96% of the di-isopropyl fluorophosphate (DFP, Figure 3) bound AChE in 24 hours, while hydroxylamine could only generate 19% of the enzyme in the same period

16 of time.43 Following initial studies by Wilson, several other groups utilized different nucleophiles to further explore this concept.44–50 However, for this approach to work effectively as a therapeutic, one needs to make a potent nucleophile that is selective for the OP-inhibited cholinesterase and shows fast reactivation, while minimizing any cross reactions and ensuring that extremely high concentrations are not required, so as to avoid severe immunological responses.

From the aforementioned reports on the reactivations of OP-bound AChE, it could be concluded that other N–OH or oxime nucleophiles might also act as reactivators, and most oximes are mildly acidic. Based on this hypothesis, Rutland et al. evaluated several acidic oximes against GB-inhibited AChE, which led to the discovery of pyrimidine-2- hydroxamic acids and monoisonitrosoacetone (MINA) (Figure 8) that showed about 90% reactivation of AChE after 15 minutes with modest concentrations.49 Simultaneously,

Holmes and Robins showed that treatment with MINA and pyridine-2-aldoxime methiodide (2-PAM, Figure 8) showed reactivation of AChE in rats after OP exposure.50

O NH2

NOH N NOH I N N 2 Cl O 2-PAM HI-6

OH OH OH OH N N N N O N OH N N N N 2 Br 2 Cl O Trimedoxime Obidoxime MINA Figure 8. Examples of oximes administered for treatment of organophosphorus poisoning.

Although these early results opened an exciting avenue for finding better oxime- based reactivators, it should be noted that the OP compounds utilized in some of these experiments for AChE inhibition were pesticides, and not authentic nerve agents. It turned out to be very challenging to develop a selective nucleophile that can reactivate nerve agent inhibited AChE before the rapid aging step takes place. Another challenge in designing an oxime is that the considered candidates should be able to cross the blood- brain barrier (BBB) so as to reach OP-inhibited AChE in the central nervous system

(CNS). Nonetheless, many attempts have been made in the last few decades to develop oximes that will work against a wide spectrum of nerve agent inhibited AChE, and in a timely manner.51–55

Nerve agents possess a stereogenic center at phosphorus, with GD possessing a second stereogenic center at the pinacolyl-carbon. It has been observed experimentally 18

– + that one stereoisomer, generally the SP (P ) isomer, is much more toxic than the RP (P ) isomer.56 This specificity is imparted through the arrangement of groups on the OP in relation to the protein structure in the active site of AChE, leading to more favorable binding of the P– stereoisomers.

In addition to the immediate symptoms of OP toxicity that are tied to the increase in ACh concentrations at the neuromuscular junction, exposure to AChE inhibitors has also demonstrated long-term side effects, especially in cases of chronic exposure such as agricultural settings.57,58 One such study has suggested the cause of these long-term symptoms may be due to removal of the inhibited enzyme from the synapse following prolonged inhibition.59 The exact implication of chronic exposure and their effects are still being actively investigated.

1.4 Butyrylcholinesterase Structure and Function in the Native, Inhibited, and Aged States.

Butyrylcholinesterase (BChE), first discovered as pseudocholinesterase, is a native enzyme that is closely related to AChE.60 Both enzymes belong to the α/β- hydrolase family, have the same catalytic triad located inside a ~20 Å deep gorge; have similar structural features, such as the oxyanion hole and choline binding pocket; have almost identical number of residues and very similar tertiary structure; possess very high catalytic efficiency (almost at the diffusion-controlled limit); and are inhibited by OP nerve agents through a similar inhibition mechanism.61 In addition to having almost the same number of amino acid residues, their sequence similarity is also quite high at 54%.

Critical components of the respective active sites of AChE and BChE have almost the same structure. 19

Despite the mentioned consistencies, there are significant differences among the two enzymes. AChE has fewer glycosylation sites than BChE has, thereby affecting the enzymes’ circulatory lifetime in the body as well as the folding, stability and many other properties. BChE is so heavily glycosylated that four of the glycosylation sites were deleted in order to get a good X-ray crystal structure.61 A major difference in their quaternary structure is the formation of a homo-dimer in the case of AChE at high concentrations whereas BChE is homo-tetrameric.61–63 Furthermore, the subunits of the homo-dimer in AChE are anti-parallel, while the openings of BChE subunits are parallel.

Thus, the orientation of the helices forming the oligomerization motif are at an angle of about 45° in the case of BChE, rather than being anti-parallel for AChE. Recently, cryo-

EM techniques have been used to determine the three-dimensional structure of a hBChE tetramer demonstrating that the base of the tetramers are oriented like a propeller rather than being situated flat as has been previously proposed.64

The most important difference at the molecular level is that the BChE lacks six aromatic amino acids out of the fourteen that line the catalytic gorge of AChE. This disparity makes the gorge of BChE nearly double the width (~13 Å) of AChE’s gorge (~6

Å), thereby accounting for almost 300 Å3 of extra available volume in BChE (Figure

9).24,25 For this reason, the gorge and active site domain for BChE is more accessible for a wide range of substrates and inhibitors. For example, the catalytic turnover for ACh is much higher for AChE than BChE.65,66 However, for a larger substrate, such as butyrylthiocholine (BCh), the catalytic efficiency for BChE is about 100 times greater than for AChE.67

Figure 9. Visual comparison of the different widths and volumes of the (left) AChE and (right) BChE gorges.

There have been several suggestions regarding the physiological role of BChE such as neuronal function, hydrolysis of γ-amino-butyrylcholine, morphogenesis, cytogenesis and tumorigenesis, hydrolysis of ACh at central nervous system synapses replacing AChE function, and converting the ß-amyloid form into more toxic forms in

Alzheimer patients. However, none of these roles have been conclusively determined.68–

74 In fact, a very early study showed that dogs treated with selective inhibitors of BChE

(about 95% serum BChE inhibition) had no signs of toxicity, and this report was followed by the discovery of a genetic variation named silent BChE where it was shown that 1 out of 100,000 people in Europe and America do not have BChE in their body.75,76

Further, discovery of the Vysya community in Coimbatore, India showed that about 1 in

24 people have this genetic variation.77 Despite the absence of BChE in these individuals, there were no signs of any physical or brain disability in this population, supporting the 21 suggestion that BChE may not have a direct physiological role.78 Yet, recently it has been determined that BChE may serve to regulate the hydrolysis of the peptidic hormone ghrelin. A study evaluated the increased plasma levels of BChE in mice for cocaine hydrolysis and the mice showed less aggressive behavior than BChE knockout mice and control mice. The study concluded that the reason for this decrease in aggression, stress, and anxiety was lower levels of ghrelin in the plasma as a result of the hydrolysis facilitated by BChE.79

BChE is inhibited by OP compounds in a similar manner as AChE and also undergoes a similar aging process after OP inhibition (Figure 7).80 The aging process is even faster for BChE than for AChE.81,82 And, aged BChE is similarly recalcitrant to oxime reactivation.

To design oximes that can reactivate OP-bound BChE at a desirable rate, it is crucial to understand how an oxime binds to BChE, and which residues are the key players in this process. We carried out parallel studies in which oxime binding to AChE and BChE was studied by the same molecular docking protocol.83,84 We utilized several of the known oximes and BChE crystal structures with and without an OP molecule being bound in the active site for these simulations and identified important interactions between the different components of various oximes and BChE.

We found that monopyridinium oximes, such as 2-PAM (Figure 8), could have two binding sites with native BChE, that is, in the absence of an OP molecule at the catalytic serine. (1) The first binding site is located at the choline binding pocket, aligned parallel to the W82 residue, while (2) the second binding site utilizes hydrogen-bonding

22 interactions at the oxyanion hole and simultaneously face-to-edge interactions with

W231. Although in the presence of an OP in BChE, the second binding site is blocked, we found two binding sites for 2-PAM. Relatively more stable binding utilizes π-π interactions with W82 residue sitting close to the bound OP molecule, while the second binding site is located at the mouth of the gorge. These data were in agreement with some experimental data obtained by Cerasoli and coworkers at the US Army Medical Research

Institute of Chemical Defense that showed that two molecules of 2-PAM interacts with

BChE at the same time.85 Furthermore, the first binding site of the 2-PAM-BChE complex is in excellent agreement with previous experimental findings from Lockridge and coworkers.86

These docking simulations also predicted that bis-pyridinium oximes (Figure 8) have relatively high affinities of binding to BChE when compared with monopyridinium oximes. Indeed, bis-pyridinium oximes utilize two π-π stacking interactions in the active site of BChE, with W82 and with F329 or Y332 residues, which suggests that these aromatic residues are essential for binding of the oximes with BChE. Out of all of the oximes, HI-6 was predicted to have very good binding affinity to OP-inhibited BChE and with an appropriate orientation of the oxime unit to reactivate the OP-bound BChE. We also found that the D70 residue often interacts with the positively charged oximes at the mouth of the gorge and is essential for binding, which is also in accordance to extant experimental results.86

As part of the reactivation process of the OP-bound cholinesterases, phosphylated oximes (POXs) are generated that are known to re-inhibit the free cholinesterases.87

Alternatively, POXs can undergo a decomposition process in which a corresponding cyano compound is generated. Thus, it is important to minimize the re-inhibition process of cholinesterases after reactivation and to accelerate the decomposition of POXs in order to provide effective oxime therapy. We studied the POX formation and decomposition pathways using ab initio quantum chemical calculations and atomistic charge analysis. It was found that the location of the oxime group should be ortho to the pyridinium center as it increases the charge separation at the oxime bond that may enhance the efficiency of the reactivation. Additionally, POXs formed from an ortho substituted pyridinium oxime were found to be least stable, i.e., more susceptible to decomposition. As a result, our calculations suggest that ortho oximes would be most efficient for the reactivation process, while minimizing the re-inhibition of the cholinesterases.

1.5 Reactivation of Acetylcholinesterase.

Since the discovery of chemical nerve agents, significant efforts have been placed to develop novel and efficient therapeutics for their treatments. Most medical countermeasures for the “treatment” of OP exposure aim to minimize the cholinergic crisis upon deactivation of acetylcholinesterase (AChE) at the neurosynaptic and neuromuscular junctions or to remove the OP toxicant by some scavenging process.88

Clinically, OP poisoning is currently “treated” by a combination of anti-cholinergic drugs

(e.g. atropine) and oximes (e.g. 2-PAM). Atropine acts as an antagonist of muscarinic acetylcholine receptors, while oximes (more likely, their deprotonated forms, the oximates) substitute the phosphylated serine in a nucleophilic manner, thereby reactivating the OP-inhibited form of AChE.89

Although significant progress has been made to date in the development of such treatments, researchers are still plagued by three main problems: (1) there is no broad scope reactivator of OP-inhibited cholinesterases, (2) current small-molecule therapeutics possess a permanent charge that limits their effectiveness in the central nervous system, and (3) to date, no therapeutics treat the aged form of AChE (or BChE).35,90,91

1.6 Recent Results from Efforts to Develop Broad Scope Reactivators of Acetylcholinesterase.

Currently, for effective treatment of a patient who has been exposed to an OP, one needs to know the OP toxicant. This problem is extremely difficult to address in a mass- casualty situation because by the time patients develop symptoms, the therapeutic window has significantly decreased. Patients can be tested in efforts to determine the OP exposure, while racing the clock to provide adequate treatment. Therefore, significant effort has been placed into developing novel oximes and reactivators that are capable of treating multiple OPs, and in an effective manner, so that a single treatment can be provided that will provide coverage against a range of OPs. In theory this solution sounds trivial but the complexities of the enzyme’s active site after inhibition by various OPs has proven troublesome with the different classes of OPs, thereby requiring different therapeutic approaches due to the size and orientation of the attached O- or N-alkyl groups (Figure 7).35,90,92 Despite these complexities, significant progress has been made in this field.

The focus of this review is primarily on addressing the issues of non-permanently charged reactivators and the development of treatments for aged AChE. Earlier in this review, we noted the seminal reports by Wilson and by Rutland on reactivation of OP- 25 inhibited AChE.43,49 An extensive review of reactivators by Kuca et al. covering the significant developments since the discovery of 2-PAM in 1955 to 2016 has been published which covers this topic in great detail.93 It has been estimated that a therapeutic

-1 with some effectiveness should have a kr > 1 min and a KD < 100 µM based on models representative of OP exposure.90

In order to facilitate reactivation in the central nervous system (CNS), Chambers et al. took the unique approach of continuing to use pyridinium-based oximes but lengthened the phenoxylalkyl chains such that the lipophilicity increases, thereby providing for anticipated penetration into the CNS.94,95 Although previously outlined by

Kuca et al., recent work using lethal doses of nerve agent analogues have been conducted in which two compounds show an increase in survival rates and earlier cessation of symptoms compared to those of 2-PAM.93,95 The Chambers study made use of two nerve agent surrogates – one for sarin and one for VX – and evaluated four novel lipophilic oximes, along with 2-PAM, to test survival rates after LD99 challenges. It was found that the use of Oxime 1 and Oxime 20 (Figure 10) increased the survival rate to 65% and

55%, respectively, and when combined with 2-PAM, the survival rates increased to 73% and 80%, respectively, relative to the 40% 2-PAM control for a sarin analogue. When the same study was repeated using a VX analogue, recovery was determined to be 33% with

2-PAM alone, 75% and 65% for Oxime 1 and Oxime 20 alone, respectively, and interestingly, only 53% and 53% when the novel oximes were used in combination with

2-PAM. Additionally, all animals treated with only 2-PAM displayed seizure-like symptoms after 8 hours, whereas all of the animals treated with Oxime 1 or Oxime 20,

26 both alone and with 2-PAM, showed cessation of seizure-like symptoms after 6 hours.

Brain cholinesterase levels were also measured but no statistically significant differences were determined between the 2-PAM and novel oxime treated samples. These results provide only indirect evidence for the penetration of the novel oximes into the brain.

OMs

O N HON R

Oxime 1: R = Cl Oxime 20: R = OBn Figure 10. Novel lipohilic pyridinium oximes.95

The work of de Bruijn et al. made use of a multicomponent reaction, the Ugi reaction, to generate a small diverse library of both charged and uncharged oxime reactivators of OP-inhibited AChE (Figure 11).96 The Ugi multicomponent reaction is a condensation reaction that combines an aldehyde or ketone, carboxylic acid, amine, and an isocyanide in a single reaction. Such a highly efficient reaction then allows the synthesis of a diverse library of compounds rapidly and easily, as was exploited by de

Bruijn et al. In their work, the carboxylic acid moiety was attached to various oximes and possessed various linker lengths. The other components, i.e., the amine, aldehyde/ketone, and isocyanide, were then used to generate a moiety to bind in the peripheral site of

AChE. The oxime moieties were chosen to cover a broad range of nucleophiles and included pyridinium oximes, imidazole oximes, ketooximes, and hydroxyiminoacetamides. Each oxime-containing compound was initially combined with 27 methoxybenzylamine, isopropyl isocyanide, and formaldehyde to generate the peripheral site ligand; special attention was paid to the amine, and an aromatic amine was specifically selected to try and maximize peripheral site interactions. The initial library of compounds was tested against human AChE (hAChE), via erythrocyte ghosts,81 after inhibition by sarin, cyclosarin, and tabun at various concentrations. The library was ineffective at the reactivation of the cyclosarin- and tabun-inhibited hAChE; however, the pyridinium and imidazole oximes, in addition to the hydroxyiminoacetamides, were able to reactivate sarin-inhibited AChE. In agreement with previous research efforts, the compounds that were the most efficient at reactivation of sarin-inhibited AChE were those that possessed a pyridinium oxime. The pyridinium oxime Ugi product with the longest linker length showed superior reactivation to that with a shorter linker length since it allows for more appropriate orientation of the peripheral site ligand in the enzyme. The influence of the peripheral site ligand was further tested by analyzing the reactivation ability of just the oxime functionality which showed much more limited reactivation than the Ugi-linked product, further supporting the benefit of the peripheral site ligand. Interestingly, the uncharged imidazole oxime Ugi products also showed sarin reactivation potential, whereas just the imidazole oxime showed no reactivation.

In the interest of further exploring optimization of the peripheral site ligand, a representative imidazole Ugi product was carried through to the next phase of testing by expanding the Ugi multicomponent library to include additional imidazole oxime compounds as well as three different amines and two isocyanides. In addition to the normal characterization methods, pKa values were determined for representative Ugi

28 products. The pKa for the imidazole oxime Ugi products were determined to be 10.2, for the pyridinium Ugi products to be 8.5, and for 4-PAM, the 4-substituted regioisomer of

2-PAM, to be 8.6. The two-unit pKa difference between the products was addressed by alkylating the imidazole oxime Ugi products, thereby reducing the pKa from 10.2 in the non-alkylated product to 8.3–8.4 in the alkylated product. Despite the now permanent positive charge on the imidazole Ugi products, the chain length was anticipated to increase the BBB penetration, which was supported by in silico calculations. All newly synthesized compounds were screened for their reactivation potential versus sarin, cyclosarin, and tabun inhibited hAChE. It was determined that the longer linker length, ranging from 1 to 3 to 5, the greater the reactivation efficacy of the tested oximes. The imidazolium compounds were found to reactivate faster than their imidazole counterparts, while demonstrating reactivation efficacy against cyclosarin-, sarin-, and tabun-inhibited hAChE. They also reported that aromatic amines and isocyanides have a positive effect on the reactivation efficacy of the oximes, and they asserted that binding affinity for the peripheral site was increased. Although none of the synthesized reactivators outperformed known oxime treatments, four of the synthesized Ugi products

(12c, 12d, 15a, and 15c, Figure 11) were carried through to in vivo testing. These studies were conducted using rats challenged with sarin, and notably, 15c showed signs of toxicity at the original dosage and was then lowered to prevent lethality. Even so, compound 15c was the only oxime capable of preventing seizures in all test animals. Of the tested oximes, 15c also showed higher brain cholinesterase levels than that of 2-PAM and the other tested oximes. Although none of the developed Ugi oxime products showed

29 superior reactivation to that of the reference oximes, the utility of the Ugi reaction to rapidly generate large libraries of structurally diverse oximes was effectively demonstrated. The Ugi products, even when permanently positively charged, did show increased brain cholinesterase activity and a reduction in seizures after sarin exposure.

The synthetic utility of the Ugi multicomponent reaction can be further used to develop and modify reactivation efficacy of oxime derivatives.

O O N H H N N N N N N N HON O HON O

MeO

12c 12d

I I O H O N N N H N N N N N HON O HON O MeO MeO

15a 15c Figure 11. Novel oxime structures synthesized using the Ugi multicomponent reaction.96

Musilek and coworkers recently evaluated the ability of tetroxime, a bis- pyridinium compound with four nucleophilic oximes (Figure 12),97 for its reactivation ability with AChE inhibited by VX, sarin, cyclosarin, and tabun. The in vitro study was conducted using rat brain homogenate and at two different concentrations, 1 mM and 10

µM. Unfortunately, tetroxime proved to not be a viable broad scope reactivator, and only showed reactivation above 2-PAM in the case of cyclosarin- and VX-inhibited AChE.

For sarin-inhibited AChE, 2-PAM was shown to be a more effective reactivator 30 especially at higher concentrations. Both oximes, even at high concentrations, were determined to be very poor reactivators of tabun-inhibited AChE. Their in silico studies showed that at the tested pH of 8.0, the 2-positioned oxime is 88% deprotonated and the

4-positioned oxime is only 6% deprotonated, suggesting that the 2-position is the nucleophilic site in tetroxime. However, on the basis of molecular docking studies, the 4- positioned oxime was shown to be in closer proximity to the reaction center suggesting that binding may be the reason for the poor performance of tetroxime. This study concluded that simply increasing the number of nucleophilic moieties on the reactivator compounds does not directly result in increased reactivation potential but may need to be further balanced by additional structural modifications.

2 Br HON

N N HON NOH NOH

tetroxime Figure 12. Tetroxime structure as studied by Musilek and coworkers.97

Franca and coworkers designed and synthesized novel oximes for the reactivation of (ethyl) paraoxon-inhibited AChE and evaluated them both in vitro as well as in silico.98 In total, three different oximes (K131, K142, and K153, Figure 13) were evaluated against the reference oximes (2-PAM, obidoxime, and HI-6, Fig 8) for their reactivation ability of OP-inhibited hAChE. Unfortunately, the best performing oxime

K153, and with the shortest aromatic peripheral site linker, provided only 9.8% of

31 reactivation, and with lower efficacy than all of the reference oximes. Indeed, obidoxime showed a near complete reactivation of 96.9% in just 10 minutes when applied at a concentration of 100 µM. These results were analyzed in silico as well. The analysis method was unique in that not only was the energy of binding analyzed but also what was referred to as the near-attack conformations (NAC). The NAC was determined to be those docked structures with a POP–Oox distance of less than 10 Å and an angle of Oox–

POP–OSer203 being 180 ± 60°. When the docking data were analyzed strictly on the structure (pose) with the best docking energy, the only conclusion that could be drawn was that the oximes have affinity for the AChE active site. However, when they analyzed the NAC poses, there was a correlation between reactivator potency and the percentage of poses that satisfy the NAC criteria. Thus, although the reactivators themselves did not outperform the reference compounds, the NAC analysis showed some promise as an in silico diagnostic that may correlate to reactivator potency. Further investigation of in silico criteria, other than just qualitative interactions of binding, may further enhance the predictability of computational models.

NOH NOH NOH

N N N N N N

2 Br 2 Br 2 Br

K131 K142 K153 Figure 13. bis-Pyridinium compounds capable of reactivating AChE and BChE with (ethyl) paraoxon.98

Ghosh and coworkers synthesized and evaluated bis-pyridinium oximes and evaluated their efficacy versus sarin, VX, tabun, and paraoxon inhibited human and electric eel AChE (hAChE and eeAChE).99 The synthesized oximes were somewhat novel in that the spacer between the pyridinium moieties was an (E)-2-butene linker and the carbonyl groups at the 4-position of the pyridinium ring were varied from being a carboxylic acid, a methylketone, and an ethyl ester (K250, K252 and K255, respectively,

Figure 14). Unfortunately, their oximes were only better than the reference oximes (2-

PAM and obidoxime) for tabun-inhibited hAChE. More intriguing, however, was that reactivation efficacy between OP-inhibited hAChE and eeAChE did not appear to follow any clear trend. For instance, they determined that for paraoxon-inhibited hAChE, obidoxime was substantially more effective than 2-PAM; however, when the same OP was used for electric eel AChE (eeAChE), 2-PAM and obidoxime had nearly the same reactivation efficacy, albeit the efficacy for obidoxime was reduced. If one compares the previous results with tabun-inhibited hAChE and eeAChE, the reactivation efficacy increases in the electric eel isoform. Thus, differences in the very well conserved AChE enzyme are nevertheless critical when evaluating oxime reactivation.

2 Br R

N N NOH

K250 R = COOH K252 R = COMe K255 R = COOEt Figure 14. (E)-2-Butene linked bis-pyridinium oximes for the reactivation of tabun inhibited AChE.99

Continuing to adapt the K-oxime structures, Kuca and coworkers used the details from study of other K-oximes to design 15 new oximes for the reactivation of ethyl paraoxon, methyl paraoxon, tabun, and DFP inhibited hAChE. The authors noted that the but-1,4-diyl linkage worked the best in previous studies and focused instead on varying the aromatic group attached in the bis-pyridinium oximes.100 The novel oximes were tested relative to a reference set of K-oximes in addition to known oximes: 2-PAM, HI-6, obidoxime, trimedoxime, and methoxime. Unfortunately, none of the newly synthesized compounds was effective at tabun reactivation, but the previously observed affinity of

K203 (Figure 21) was again observed. Of the reference oximes, trimedoxime was the best performer at 31.5% and 100 µM. Of the newly prepared oximes, only K15 and K22 showed any activity above that of the reference oximes and previously studied compounds for hAChE inhibited by ethyl paraoxon (Figure 15). In the case of methyl paraoxon and DFP, reference oximes were much better or all compounds were ineffective, respectively. The novel oximes in the study proved to be potent inhibitors of 34 hAChE with IC50 values in the low micromolar range (6–90 µM). From the synthesized variants, they determined that the pyridinium ring with both hydrophobic and hydrophilic groups at the 4-position form more favorable interactions in the peripheral site owing to their increased reactivation potency. Unfortunately, none of these modifications resulted in any compound that had broad spectrum activity with multiple OP pesticides.

2 Br R

N N NOH

K15 R = tBut K22 R = CN Figure 15. Two novel K-oximes showing some efficiency for the reactivation of (ethyl) paraoxon inhibited hAChE.100

Madura and coworkers made use of computational tools and the ZINC database to select a diverse library of commercial oximes to test for OP reactivation.101 The authors started their study with a search of the ZINC database yielding thousands of compounds.

Using a similarity and dissimilarity search, the authors narrowed the number of compounds down to just 60 compounds which was further cut to 18 based on visual structural evaluation, yielding a final library of 18 untested oxime compounds. To ensure that all of the compounds could fit within the AChE active site, docking studies were conducted in which each compound was docked with mouse AChE inhibited by methamidophos, ensuring a docked structure was found with a P–oxime bond distance of less than 6 Å. All 18 of the compounds satisfied this criterion and were continued

35 forward. However, despite being extracted from the ZINC database, the authors could only purchase six of the 18 compounds for biological evaluation. All six of the oximes were compared relative to 2-PAM activity for their reactivation of ethyl paraoxon, DFP, henamiphos, and methamidophos inhibited AChE. It is of interest that of the six oximes tested for all the OP pesticides, the reactivation potency was similar or superior to that of

2-PAM in all cases. The most promising of the compounds is #2 (Figure 16) which showed 41% reactivation of fenamiphos and 58% reactivation versus DFP in one hour; these OPs are two pesticides that are notoriously difficult to reactivate.

Unfortunately, #1 and #2 (Figure 16) showed high inhibitory potency at the tested

100 µM concentration, and 25 and 15% relative activity, respectively. Despite the high inhibition potency, this study shows the utility of database searches for provision of structural modifications that may not otherwise be made in drug design, offering a method for scaffold elucidation in future reactivation design.

O O

N HN N N HN N NOH NOH

#1 #2

O NOH NH H HN NOH N N N N O

#3 #4

O H NOH N N N HN N N NOH O O

#5 #6 Figure 16. ZINC database oximes determined from virtual screening which are capable of reactivating multiple OP pesticides.101 Acharya and coworkers noted the efficiency of HI-6 and four different isonicotinamide derivatives of pyridinium substituted hydroxyiminoacetamides for the reactivation after sarin and VX exposure.102 The four compounds were screened for their reactivation potency with sarin- and VX-inhibited hAChE, via erythrocyte ghosts, relative to the control oximes 2-PAM and obidoxime. Oximes 4a and 4b (Figure 17) showed superior reactivation potency of sarin-inhibited AChE with effective rates of

29.77 and 23.94 mM-1 min-1, compared to 2-PAM and obidoxime at 2.33 and 7.91 mM-1 min-1. However, for VX-inhibited AChE, 4c (Figure 17) performed the best with an effective rate of 25.47 compared to 2-PAM and obidoxime at 3.50 and 8.23 mM-1 min-1, respectively. The oximes differed only in the length of the linker chain and the results 37 suggest that steric effects of the inhibited serine play a role in the reactivation potency.

This reiterates the difficulty in designing broad-spectrum reactivators. This study did prove that the oxime moiety itself does not need to be directly linked to the aromatic core of reactivators and can be pendent. Additionally, it confirms the utility of hydroxyiminoacetamides as potential nucleophiles for reactivation.

O NH2 NOH 2 Br O HON O NH2 NH N NH 2 Br N O N N

4a 4b

NH2 NOH O O 2 Br N NH N

Figure 17. Hydroxyiminoacetamide nucleophiles using structural components of HI-6 for reactivation of sarin and VX-inhibited AChE.102

Ghosh and coworkers followed up the findings that imidazole-based reactivators are very effective for BChE reactivation by synthesizing and testing imidazole-based reactivators for AChE reactivation with sarin and VX.103 In total, eight imidazolium compounds were tested for their reactivation potency with erythrocyte ghost hAChE compared to the reference oximes: 2-PAM, obidoxime, and HI-6. The pKa values of the

38 compounds were determined to be within the expected 7–9 range as is usually observed for oxime nucleophiles. However, unfortunately the oximes were no more potent than the reference oximes HI-6 and obidoxime for both sarin- and VX-inhibited hAChE.

However, one compound, 9a (Figure 18), showed greater reactivation potency than 2-

PAM for sarin and VX. This compound is an imidazolium linked to a 4-PAM moiety. It was determined from the reactivation studies that alkylimidazoliums are less effective reactivators when compared to pyridylimidazoliums, as may be expected based on aromatic interactions of the pyridinium ring with the peripheral site. For this set of compounds, it was observed that a three-carbon methylene linker between the imidazole and the pyridine was the best and extension of the chain length out to five carbons reduced reactivation potency. Thus, it appears as though the affinity and utility of imidazole-based nucleophiles for AChE is not the same as that for BChE and continued structural modifications may need to be made to achieve an imidazole-based reactivator of AChE.

NOH N

N NOH N 2 Br

9a Figure 18. Imidazolium oxime for the reactivation of sarin and VW-inhibited AChE.103

Amitai et al. used a bifunctional drug concept to link together oximes with known ability to reactivate AChE with hydroxamic acids which showed a greater ability to 39 scavenge and hydrolyze OPs in solution as a therapeutic capable of protection and reactivation.104 The authors first investigated the ability of the conjugates to detoxify sarin, cyclosarin, soman, and VX in solution. It was determined that all of the conjugates were faster at detoxifying the OP cyclosarin compared to 2-PAM; however, for sarin and soman, the rates of detoxification were slightly slower for two of the three conjugates. Of the tested compounds, 2PAMMeBHA (Figure 19) was by far the best at detoxifying all of the OPs compared to the reference compounds, having half-lives between 1.9–3.6 minutes. For all tested compounds, oximes and conjugates, VX detoxification was very slow. Next, the reactivation potency of the compounds was tested using the OPs sarin,

VX, and cyclosarin and the reference oxime 2-PAM again. In the case of sarin, 2-PAM recovered the most activity of AChE and at the fastest rate when compared to the conjugates. For VX, 2PAMMeBHA showed the highest rate of reactivation but only reached a maximum of 56% compared to 2-PAM which reached a maximum of 94%

AChE activity. This may be caused from re-inhibition of the phosphylated-oxime. For cyclosarin-inhibited AChE, the fastest reactivator was 2PAMPr4PHA (Figure 19), reaching a maximum of 70% AChE activity relative to the maximum of 65% for 2-PAM.

Further analysis was conducted using HI-6 as a reference and comparing reactivation potency of 2PAMPr4PHA with sarin, cyclosarin, and VX. In all cases HI-6 had superior reactivation ability. The synthesized conjugates proved to be relatively strong inhibitors with IC50 values of 34–90 µM, yet they were still weaker inhibitors than HI-6 with a measured IC50 of 11 µM. The ability of these compounds to serve as pre-treatments for

OP exposure was explored on the premise of their high reversible inhibition. What was

40 found was that 2PAMMeBHA and 2,4DiPAMMeBHA (Figure 19) were capable of significantly slowing down the inhibition of AChE by VX, sarin, cyclosarin, and soman – and significantly greater than by 2-PAM. This finding then suggested to these researchers that they can protect from OP inhibition much like has been observed for pyridostigmine.

Prior to moving to in vivo testing, the LD50 values of the conjugates were determined.

The conjugates had values between 179–200 mg/kg in rats, similar to 2-PAM, and 80–

200 mg/kg in guinea pigs, with 2-PAM being ~170 mg/kg. To determine the appropriate dosing of the conjugates for in vivo studies, cholinesterases in blood serum was measured following challenges with sublethal doses of sarin along with toxicology studies. At the highest dose of 142 µmol/kg, 2-PAM recovered 15% of blood ChE activity, while

2PAMPr4PHA recovered 42% at the same dose and 2PAMMeBHA recovered 22% when test animals were pre-treated. The results for post-treatment were slightly different with

2PAMPr4PHA at 50% and 2-PAM and 2PAMMeBHA at 20%, again at 142 µmol/kg.

The differences in results could be caused by reactivity with BChE in the blood, but it was demonstrated that the degradation and reversible binding did increase protection in the blood for these conjugates. The in vivo protective ratios were determined for each compound when mice were exposed to sarin, cyclosarin, VX, and soman. The best compound in all cases was 2PAMPr4PHA. For sarin-treated mice, good protective ratios were observed for both 2PAMPr4PHA and 2PAMMeBHA in a pre-treatment regime, compared to 2-PAM, but post-treatment was similar to 2-PAM. The same results hold true for cyclosarin except that 2PAMPr4PHA was superior to other conjugates. In the case of VX, all of the conjugates and 2-PAM had similar protective ratios for both pre-

41 and post-treatment regimes. Finally, the protective ratios for 2PAMPr4PHA was better for soman exposure in both the pre- and post-treatment cases relative to 2-PAM. These same studies were repeated in a guinea pig model as well. Interestingly, for guinea pigs exposed to sarin, all of the conjugates have lower protective ratios than 2-PAM for both pre- and post-treatment methods, in contrast to what was observed in mice; this may be attributed to having to lower the doses based on the findings that these conjugates were more toxic in the guinea pig model. For the cyclosarin-exposed guinea pigs, the best protective ratios were observed for 2PAMMeBHA and 2PAMPr4PHA in pre-treatment and for post-treatment just 2PAMPr4PHA. Unfortunately for VX exposure, all of the conjugates performed worse than 2-PAM in terms of protection. Finally, for soman- exposed guinea pigs, 2,4DiPAMMeBHA showed the best pre-treatment protective ratios and 2PAMPr4PHA and 2,4DiPAMMeBHA showed the same protective ratios in post- treatment methods. Overall, the best conjugate for in vivo protection of both animals and multiple agents was 2PAMPr4PHA, given its ability to reactivate, scavenge, and in vivo protection ratios, this conjugate method seems to present a unique way to treat and protect from OP exposure.

NOH NOH

N N N H H N N OH OH 2 Br O Br O

2PAMPr4PHA 2PAMMeBHA

NOH

N H HON N OH Br O

2,4PAMMeBHA Figure 19. Hydroxyiminoacetamide nucleophiles using structural components of HI-6 for the reactivation of sarin- and VX-inhibited AChE.104

1.7 Developing Non-Permanently Charged Oxime Reactivators of OP-Inhibited Acetylcholinesterase.

The pursuit to find an uncharged reactivator of OP-inhibited AChE has taken multiple different paths. One approach has been to use a pro-drug for the desired reactivator and to generate the active drug in the desired tissue, such as for pro-2-PAM.105

Some have tried to develop novel non-oxime reactivators, and some have attempted to take the existing oxime functionality from clinically approved therapeutics and to remove the need for the positive charge of the pyridinium moiety by tethering the oxime to a linker and peripheral site ligand that sufficiently enhances binding. The AChE active site has selectivity for positively charged species that closely resemble the native substrate, acetylcholine. Thus, the originally designed oximes, such as 2-PAM, trimedoxime, obidoxime, and HI-6 (Figure 8), share the positively charged pyridinium moiety that

43 significantly increases the binding affinity. However, to develop a therapeutic that is capable of crossing the BBB, the permanent positive charge needs to be “removed”, and thus the affinity for the active site of AChE needs to be recovered in some manner, such as attachment to peripheral site ligands (PSL). A great deal of variability arises with selection of the peripheral site ligand, including linker type, linker length, and nucleophilic moiety. The difficulty in such an approach is that changing a single element, nucleophile, linker, and PSL, can have a profound impact on the reactivation potential requiring a complete structure-activity evaluation for each class of molecule to find the optimum combinations, not to mention the further complexities arising from stereochemistry, regiochemistry, and pKa. Thus, years of research have been conducted, and to date, no new therapeutics have been approved in decades. Nevertheless, the continued threat and worldwide crisis caused by OP compounds necessitates the continued efforts and developments of new therapeutics.

Li et al. took a unique approach of using an oxime, based on salicylaldoxime derivatives, a linker unit derived from the known indanone framework of anti-

Alzheimer’s disease treatment donepezil, a piperidine linkage, and tetrahydroisoquinoline peripheral site ligands to construct new oxime reactivators, and then tested against OP- inhibited hAChE.106 The peripheral site ligand alone showed no reactivation potential of

OP-inhibited AChE as was expected. Additionally, the untethered oximes reactivated poorly, confirming the hypothesis that tethering an oxime to an appropriate PSL and linker can sufficiently enhance reactivation potential. Three of the studied oximes (3c–5c,

Figure 20) were shown in vitro to outperform the reference oximes HI-6 and obidoxime

44 in the reactivation of VX- and tabun-inhibited AChE and had similar reactivation potential in the case of sarin-inhibited AChE. They concluded that piperidine-linked conjugates are more effective than methylene-linked analogues. Furthermore, the linkage at the ortho position, relative to the phenol, outperformed those oriented para. Finally, increasing the hydrophobicity of the oxime rings by introduction of methyl, chloro, and bromo substituents also had a positive effect on the reactivation potential of the oximes.

The study also conducted in silico evaluation of the BBB permeability using an ADMET program and showed that the newly synthesized conjugates have a much better probability of crossing the BBB compared to the reference oximes, though no other experimental data were presented.

OH O N NOH N O R

3c: R = Me 4c: R = Cl 5c: R = Br Figure 20. Novel oxime structures based on salicylaldoximes.106

Kuca et al. recently revisited the testing of K203 (Figure 21) which had shown promising tabun reactivation results in rodent models; however, no human data had been reported until recently. The study was unique in that they evaluated human AChE from human brain homogenate. K203 reactivated over 60% of tabun-inhibited hAChE, while for the three reference oximes (2-PAM, obidoxime, and HI-6), obidoxime was the best of

45 three and only reached 33% reactivation, and at a concentration higher than can be achieved in vivo. This same experiment was repeated in a rat model as verification of previous studies and the trends were consistent in both models. However, the effective reactivation rate was reduced by a factor of 7.5 for K203 but by a factor of 149 for obidoxime when comparing the ratio of reactivation rates between human brain homogenate and rat brain homogenate.107

O 2 Br NH2 N N HON

K203

Figure 21. Novel oxime structure that shows significant tabun reactivation.107

Li and coworkers investigated the reactivation potential of novel ortho- hydroxylbenzaldoximes linked to phenols against soman-inhibited hAChE which is extremely difficult to reactivate as the competing aging process is very rapid, on the order of just a few minutes.108 Surprisingly, in their report, all of the synthesized oximes reactivated soman-inhibited hAChE to levels higher than that of 2-PAM, but not as high as HI-6. Despite being tested at multiple time points, the reactivation by the oximes was fully achieved within the first 2 hours, and then remained constant out to 24 hours. The best performing compounds (7b–9b, Figure 22), showed reactivation percentages of

15.5%, 19.7%, and 13.4%, respectively, compared to 36.9% by HI-6 at 24 hours. From

46 the results in this study, the presence of inhibited and aged hAChE after soman exposure needs clarification.

R2 HO R3 N NOH

7b: R1 = H, R2 = H, R3 = OH 8b: R1 = OH, R2 = Me, R3 = H 9b: R1 = H, R2 = Me, R3 = OH

Figure 22. Novel oxime structure that shows some soman reactivation.108

Renard and coworkers synthesized and tested uncharged oxime reactivators making use of the anti-Alzheimer’s Disease drug donepezil as a peripheral site ligand.109

The benzyl moiety of donepezil was replaced with a 3-hydroxypyridinaldoxime as the nucleophilic portion of the molecule (Figure 23). In total, four variants (Don1 – Don4) were synthesized and evaluated for their reactivation ability versus VX, sarin, paraoxon, and tabun inhibited hAChE. Don1–3 performed better than 2-PAM but are not as efficient for reactivation of VX-inhibited AChE as obidoxime and HI-6. Comparing

Don1 to Don2, they determined that the indanone carbonyl moiety increases binding affinity but does not improve the overall reactivation efficacy. Comparing Don3 and

Don4 showed that the positioning of the linkage has a significant effect on reactivation efficacy, with the 6-linkage having a better reactivation efficacy. Interestingly, replacement of the piperidine ring in the linkage seems to have no effect on the reactivation efficacy between Don1 and Don3. Sarin-inhibited AChE was tested with HI- 47

6 and Don1, but unfortunately the new oxime was not as efficient as HI-6. Oximes Don2 and Don3 were demonstrated to be better reactivators of paraoxon-inhibited AChE than

HI-6. When tabun-inhibited AChE was analyzed, no reactivation was observed with the donepezil-based oximes. The strategy of linking the donepezil-based fragment to known oximes was shown to achieve binding affinity similar to that of mono- and bis- pyridinium oximes for VX and sarin even though these synthesized oximes were uncharged, lending support to the peripheral site ligand strategy.

O O HO

N HON N O N O

O O Donepezil Don1

HO O

HON N HO O N O O HON O N

Don2 Don3

O O N O HON

Don4 Figure 23. Novel oxime structures based on the Alzheimer's Diseas drug donepezil by Renard and coworkers.109

Renard and coworkers subsequently published non-permanently charged oxime reactivators of various OP compounds using amine or tetrahydroisoquinoline moieties as

48 peripheral site ligands.110 All of the synthesized oximes (15–19, Figure 24) were shown to have moderate binding affinities in the µM range and similar to that of HI-6. The oximes were tested for their reactivation efficacy against VX, sarin, cyclosarin, tabun, and paraoxon. Compound 18 was significantly better at reactivating VX-inhibited AChE than HI-6, while the rest of the oximes were similar to HI-6. Unfortunately, for sarin- inhibited AChE, none of the oximes were shown to outperform HI-6, but their reactivation potencies were similar to HI-6. For cyclosarin-inhibited AChE, the oximes performed rather poorly compared to HI-6, and were 70–100 times less effective.

Interestingly, tabun-inhibited AChE was reactivated more effectively for all of the new oximes when compared to HI-6, and importantly, HI-6 showed very limited reactivation.

Additionally, all of the new oximes also outperformed HI-6 when tested against (ethyl) paraoxon-inhibited AChE. Although the reactivation potential for some of these compounds were below that of the reference, it is worth noting that at different concentrations, these newly synthesized oximes were capable of reactivating 50–100% of the inhibited forms of each enzyme, and at various time points. The most notable is the recovery of 50–90% of the activity of tabun-inhibited AChE, with tabun being known as an OP that is difficult to reactivate.

These results are very promising and demonstrate a new class of compounds that are relatively broad scope in comparison to known oximes. This study further investigated the ability of the compounds to recover AChE activity in whole blood through both reactivation as well as scavenging of the OPs in the blood. Oxime 16 showed very promising results, recovering more than 80% of the cholinesterase activity

49 in whole blood in 5–15 minutes after VX, sarin, and paraoxon exposure and 60 min after cyclosarin exposure, thereby showing significant promise as a treatment, especially since whole blood cholinesterase activity is both a measure of the AChE and BChE activity.

The pKa values of the oximes were tested using both in silico as well as in vitro methods, and for all oximes, the pKa of the oxime and amine moieties were determined to be ~8 and 10.5, respectively. These pKa results suggest that these compounds with protonation states near physiological pH should be able to penetrate the BBB. The ability to penetrate the BBB was further investigated using both in silico and in vitro approaches. The CNS multi-parameter optimization score for each of the oximes was determined and accounted for six physiochemical parameters and showed that all of the new oximes are well within the high BBB permeation scores with the exception of 19 which is below the CNS cutoff.

These results were confirmed using an in vitro parallel artificial membrane permeation assay PAMPA-BBB using lipid extract of porcine brain and a reference set of 14 commercial drugs. All of the synthesized oximes were categorized based on the reference set to have high BBB permeation. Altogether, these newly synthesized and uncharged oximes reactivators of AChE are extremely promising, showing broad scope reactivation with V-agents, G-agents, and pesticides and with an ability to penetrate the BBB and be active in the CNS. Future in vivo testing is needed to confirm their utility.

O O O O

N N N N HO n N HON R N 15 (n = 3) 16 (n = 2) 18 (n = 3) 19 (n = 3)

17 (n = 3) Figure 24. Uncharged reactivators of cholinesterases with the potential to cross the blood- brain barrier.110

Understanding the limitation of having a potent inhibitor of native AChE as a reactivator, Nachon and coworkers proposed structural modifications for a known tacrine-linked reactivator in efforts to reduce its interaction with the native enzyme using knowledge derived from crystal structures (Figure 25).111 First, three X-ray crystal structures were solved after soaking Tac1 into one of native T. californica AChE

(TcAChE), a tabun analogue inhibited TcAChE, and an inhibited TcAChE with a rivastigmine analogue. For only the uninhibited TcAChE crystal structure was a molecule of Tac1 observed within the active site; for the other two, the ligand was observed to bind in the PAS. Interestingly, the uninhibited TcAChE structure has two molecules of

Tac1 bound: one in the catalytic active site (CAS) and one in the PAS. The authors hypothesized that the interactions within the CAS are what leads to the high potency and they desired to interfere with these interactions present within the CAS by introducing a chlorine atom at position 7 on the tacrine rings to generate a steric interaction with nearby residues, leading to the design of Tac2. To test the hypothesis of steric interference, molecular docking was performed with both molecules. The results supported their

51 hypothesis as Tac2 was only observed in the PAS, and not the CAS. To determine how the reactivation efficiency was affected, in vitro studies were conducted with hAChE and compared to 2-PAM, HI-6, and obidoxime. The IC50 values showed that the introduction of a Cl atom was successful in affecting the binding affinity of Tac1 (IC50 value of 0.25

µM) as compared to Tac2 with an IC50 of 2.3 µM. Interestingly, in addition to reducing the affinity for the native enzyme, the reactivation potency for tabun-inhibited hAChE was actually increased, while for VX, a sarin analogue (NIMP), and paraoxon, the potency remained approximately the same. Tac1 and Tac2 outperformed all three reference oximes for the reactivation of VX, tabun, and paraoxon and were very similar to HI-6 for the reactivation of the sarin analogue NIMP. To determine the new binding mode that Tac2 was adopting with AChE, the previous X-ray crystallography approach was repeated for native TcAChE, tabun analogue inhibited TcAChE, and rivastigmine inhibited TcAChE. All of the structures showed binding in only the PAS, confirming the disruption of interactions within the CAS. The decreased inhibition of native hAChE, combined with the similar reactivation profile and increased reactivation potency with tabun, suggests that Tac2 is a viable candidate for continued investigation. Additionally, this work supports the utility of using X-ray crystal structures of inhibited AChE to guide structure-based design of reactivators and inhibitors of AChE.

H N N NOH N OH

Tac1 R = H Tac2 R = Cl Figure 25. Tacrine-linked reactivators designed on the basis of X-ray crystallographic data.111

Using the structural scaffold of vitamin B6, Gaso-Sokac and workers designed a library of pyridoxal oxime derivatives and tested their reactivation potency with VX-, tabun-, and paraoxon-inhibited AChE and BChE.112 Unfortunately, of the nine synthesized oximes (Figure 26), none performed better than the reference oximes

(obidoxime, TMB-4, and HI-6). However, at slower rates of reaction, some compounds were able to recover 85% of the hAChE activity in 23 hours. The results for hBChE are slightly more promising but still not successful overall in comparison to the observed rates by the reference oximes. Novel oximes B6-6 and B6-8 (Figure 26) showed the best results, recovering 90 and 95% in 8 and 5 hours, respectively. The binding affinities of all of the compounds were tested in addition to the rates of reaction. They determined that for all compounds, the Ki is in the µM region, including the reference oximes, for recombinant hAChE and hBChE. It was postulated that the reason for this inactivity is due to the increase in steric hindrance around the oxime moiety, which is reflected in the slower rate of oximolysis in the absence of enzyme.

NOH NOH

HO HO OH OH O N N

Br Br O O

B6-6 B6-8 Figure 26. Vitamin-B6 based oximes for the reactivation of hAChE and hBChE.112

1.8 Developing Non-Oxime Based Reactivators of Inhibited Acetylcholinesterase.

The continued need to develop non-permanently charged oximes for the treatment of OP exposure has started to see an increase in the investigation of non-oxime based therapeutics. Pyridine-based therapeutics are effective, but their binding and efficacy are significantly increased when the therapeutics are alkylated to form the pyridinium. The positive charge increases affinity for the active site through cation-π interactions with

Trp86.113 Recently, more and more nucleophilic compounds, which are non-oximes, have been studied in order to find structural elements that provide similar binding affinities and rates of reactivation without the need to change the charge of the compound to facilitate binding and/or efficacy.

OH N O O OH N N H O Cl amodiaquine (ADQ) scopoletin

H O N N N OH

N H2N

Br ADOC SP138 Figure 27. Non-oxime reactivators.113,114

Stojanovic et al. took a unique approach of screening a library of 2000 bioactive and approved drugs for their ability to reactivate OP-inhibited AChE.113 The screen resulted in two initial leads: the antimalarial drug amodiaquine and scopoletin (Figure

27). Amodiaquine was chosen as the compound of initial focus as scopoletin was effective against paraoxon-inhibited AChE, but not DFP-inhibited AChE. Investigation of the active structural features of amodiaquine led to the discovery that Mannich phenols were effective reactivators of organophosphate-inhibited AChE and in fact 4-amino-2-

(diethylaminomethyl)phenol (ADOC, Figure 27), which is amodiaquine without the hydrophobic component, is the effective portion of the molecule. Further investigation of

ADOC found that the para-relationship between the amine and hydroxyl group as well as the hydroxyl group itself are essential to reactivation ability. ADOC was shown to be an effective reactivator of multiple forms of both OP pesticide and nerve agent inhibited

AChE. Mechanistic investigations postulated two separate mechanisms: (1) The phenol is

55 participating in direct nucleophilic attack of the phosphylated serine, and (2) the Mannich base is providing an external base for reactivation within the active site. Further studies led to the discovery of chloroquine being a reactivator of inhibited AChE but lacking the

Mannich phenol. Instead, chloroquine possessed a hydrophobic anchoring moiety and a basic moiety. In fact, swapping of the linker length and the basic moiety led to the discovery of additional classes of AChE reactivators, such as SP138 (Figure 27), all of which are non-oxime based. Further in vivo studies with both ADOC as well as SP138 showed excellent protective effects against lethal doses of DFP. Most interestingly, when comparing cholinesterase levels in various tissues, the authors reported that tissues beyond the blood-brain barrier (BBB) in the CNS were more active when treated with

ADOC and SP138 than when mice were treated with 2-PAM or left untreated with any reactivator. This result suggests that these newly discovered classes of reactivators address directly the issue of permanent charge and the ability to penetrate the BBB. This discovery of two new classes of non-oxime reactivators of inhibited AChE opens new avenues of continued research and the potential development of more effective and BBB- penetrating therapeutics.

Amodiaquine (Figure 27) was further investigated by Worek and coworkers for its reactivation potential, not only against DFP as was observed by Stojanovic and coworkers, but for its reactivation potential against tabun, sarin, cyclosarin, and VX.115

Amodiaquine was determined to be a fully reversible inhibitor of erythrocyte ghost hAChE but it is a highly potent one with a measured IC50 value of 669 nM. It was determined that amodiaquine is a mixed competitive/non-competitive inhibitor of hAChE

56 in contrast to the results for eeAChE which showed amodiaquine to be a competitive inhibitor. As a reactivator, amodiaquine was able to recover small percentages of activity of sarin, cyclosarin and VX inhibited hAChE, but was ineffective versus tabun-inhibited

AChE. Although reactivation was observed for these nerve agents, the rate of reactivation was significantly reduced in comparison to HI-6. Amodiaquine was shown not to scavenge OPs directly. These studies again confirm that amodiaquine has some functional moiety that is capable of reactivation of inhibited AChE.

Cerasoli et al. followed up on the study of ADOC with a focus on the importance of each structural element and the effectiveness versus OP nerve agents.114 First, removal of the aniline and benzylamine moieties of ADOC significantly reduced the reactivation efficacy, but did not prevent reactivation. However, elimination of the hydroxyl functionality or blocking of the hydroxyl group (as a methyl ether) eliminated or significantly reduced reactivation, respectively. These results suggest that for ADOC, the portion of the molecule responsible for reactivation is the phenol acting as a nucleophile.

It was found that altering the positioning of the benzylamine reduced reactivation ability.

Interestingly, both decreasing and increasing the steric bulk of the benzylamine decreased inhibitory potential. If the size of the benzylamine was further increased reactivation potential was decreased. These results then suggest that the active mechanism for reactivation of inhibited AChE is unrelated to formation of the corresponding quinone methide as an intermediate, since all three variants would form the same intermediate − yet the results varied by structure.

Manipulation of the aniline to increase and decrease electron-donating effects resulted in decreased reactivation rates in both cases. The determined dissociation constants for the modification of the aniline moiety shows that the loss of reactivation is not related to binding, but instead to a reduced rate of reactivation. Changing the positioning of the aniline group from para- to meta- resulted in decreased reactivation potential, but not complete elimination, leading to the conclusion that the para-oriented aniline aids in proper positioning of ADOC within the AChE active site. The authors also conducted in vivo experiments with ADOC using an established guinea pig model; however, the results were contradictory to those previously reported in that there was no significant reactivation observed for ADOC. Notably, the authors report that the protocol used was based on pyridinium oximes and that administration time and method may have affected the results and also that there may be species differences between the in vitro human AChE, the previously studied mouse AChE, and guinea pig AChE. This species difference in reactivation continues to be a problem noted by multiple authors and studies.99 Nonetheless, ADOC’s in vitro reactivation potential and its discovery as a new class of reactivator of OP-inhibited AChE can be explored by conventional medicinal chemistry techniques to develop optimized therapeutics.

Stojanovic and coworkers followed up their discovery of two classes of non- oxime reactivators in an additional expansion study. The first portion of the study focused on analyzing an analogue of amodiaquine called isoquine (Figure 28).91 Isoquine was shown to have reduced toxicity as it does not form the iminoquinone; however, the structure still remains a Mannich base. Isoquine, like amodiaquine, showed reactivation

58 potential versus paraoxon-inhibited AChE and actually performed at a level equal to that of amodiaquine. However, the results for DFP-inhibited AChE were reduced in comparison to amodiaquine. While Mannich bases are able to reactivate OP-inhibited

AChE, the orientation of the Mannich base does have an effect on reactivation efficiency

– this effect was also observed by Cerasoli and coworkers.114

The Stojanovic study evaluated structure-activity relationships on the SP138

(Figure 27) class of reactivators. The main modification was the removal of the p- bromophenyl ring resulting in deceased reactivation potency. Additional analogues were investigated where the para-bromo moiety was replaced by a fluoro, chloro, phenyl, and hydroxyl group, all of which resulted in little change on reactivation potency leading to the conclusion that the para position can be used to improve physical properties and formulation in future investigations. Unfortunately, none of the synthesized analogues outperformed SP138. A similar set of compounds to that of SP138 was tested drawing from the structure of SP134 which was reported in their original study.113 The new compounds replaced the imidazole moiety of SP134 with a pyridine moiety, and three analogues were synthesized with methylene linkers ranging from 1–3 units. The newly synthesized compounds did show reactivation potential, but no increase in effectiveness, and the new compounds were all outperformed by SP134. One of the reactivators (C47,

Figure 28) from the initial screen was noted for its features that were similar to donepezil and its reactivation potency was similar to SP138 and SP134. Additional donepezil- inspired analogues, such as SP111, were synthesized and possessed either a piperidinyl or piperazinyl ring attached to the pyridyl ring. These compounds were evaluated for their

59 reactivation potency but were ruled out as they, like donepezil itself, proved to be extremely potent inhibitors of native AChE.

N O N HN OH Cl N

Cl N Cl

isoquine SP134

O N H O N O N O N

SP111 C47 Figure 28. Additional reactivator classes.91

The interest in reactivators with a Mannich phenol base continued as de Koning et al. expanded on the work conducted by Cerasoli and coworkers in synthesizing additional derivatives of ADOC and then did testing with human erythrocyte AChE.114,116 A total of

14 derivatives of ADOC were synthesized, with focus on variations in the amine leaving group. IC50 values were determined for all compounds and showed that simple modifications can have great effects. Compounds with 4-6 carbons attached to the amine had some affinity for human erythrocyte AChE ranging from 6.3 – 482.5 µM.

Lineweaver-Burk analysis showed that compounds 3h (ADOC) and 3l are mixed noncompetitive inhibitors (Figure 29). All of the synthesized compounds were tested for their ability to reactivate human erythrocyte AChE at varying concentrations and after

60 inhibition by VX, paraoxon, tabun, cyclosarin, and sarin. As with most reactivator studies, none of the tested compounds were able to reactivate tabun-inhibited AChE. Of the tested compounds, the best were those of ADOC and 3l. All other compounds showed negligible to some reactivation ability at high concentrations, but not in comparison to ADOC and 3l. Most interestingly, compound 3l showed superior reactivation efficiency to ADOC for all tested OPs. Compound 3l displayed rapid reactivation as VX-inhibited AChE was reactivated in 20 min with only 10 µM 3l. Some strange observations were made regarding some of the Mannich bases. For six of the compounds at 1 mM, they observed that there was initial rapid reactivation, followed by a slow decrease in reactivity over time. It was hypothesized that this was due to re- inhibition by the phosphylated phenol after reactivation (perhaps, akin to POX re- inhibition,117 as noted earlier in Section 1.4 of this review). The adduct was synthesized and found to be stable and a weak inhibitor of hAChE. Additional experiments using 1 mM reactivator and hAChE in the absence of OP or inhibition showed the same decrease in enzyme activity – the two offered explanations were that these high concentrations lead to denaturation of the enzyme or that the Mannich bases are forming quinone methides as intermediates and are then alkylating and altering the enzyme’s function. A second unusual finding was that some of the compounds, excluding 3l, require up to 20 min to begin reactivation, in particular with VX- and paraoxon-inhibited AChE. No explanation was offered for these results. Given the multiple studies based on Mannich phenols and their rapid reactivation for a number of OPs, it appears that significant focus should be shifted toward the investigation of these non-oxime, non-permanently charged

61 reactivators, particularly trying to address some of the strange, or interesting, findings reported by van Grol and coworkers.

N H2N

3l Figure 29. More reactivat Mannich base for the reactivation of AChE.116

1.9 Difficulties with the Evaluation of Developed Reactivators.

To date, thousands of oximes have been synthesized for the reactivation of various types of OP-inhibited AChE; however, the treatment methods still being used are those that were discovered in the 1950s. Outlined above is the current advancement in the field of oxime and reactivator development, but it is worth noting that all current and future studies encounter the same problems in terms of advancing therapeutic development.

These difficulties begin with the structure-based design approach making use of

X-ray crystal structures of known inhibitors and therapeutics with AChE. To date 178 structures have been reported with a majority being for T. Californica and M. Musculus and fewer for human forms. Of those structures reported 146 are co-crystallized with various ligands.118 X-ray crystallography is conducted at low temperatures and may not sample the appropriate conformations exhibited by AChE for binding under physiological conditions. This is particularly related to binding of some large ligands in which the crystallographic structure gives no inclination as to how the ligand can make its way into or out of the narrow AChE gorge, despite its high binding affinity. Additionally,

62 crystallization additives have been shown to bind within and on the periphery of AChE, which can alter positioning of the enzyme structure itself or the binding within the enzyme. It has also been noted that significant structural changes occur on ligand binding, most notably in the Ω loop. These structural changes cannot be predicted directly from the crystal structures of native AChE compared to the ligand-bound AChE due to the above-mentioned conditions.92 In fact, a 20 ns simulation of hAChE shows upwards of 1400 Å3 variation in the pocket volume resulting from significant structural movements.118

The most troubling issue is the comparison of data from various research groups with one another. Experimental details are critical, and different results can be obtained based on conditions including: AChE sources, in particular human compared to non- human, and the chosen experimental conditions including time of reactivation, buffer choice, pH, temperature, concentration, dilution, and calculation method.35,99,107,119 These experimental parameters are only those associated with in vitro experiments and the issues become even more widespread when moving to in vivo models. Issues of particular note for in vivo studies include animal species (additionally whether these are knockout or knock-in variants), OP exposure routes, dosage of OP, comparison to oximes or additional therapeutics, combination of therapeutic administration (whether atropine or benzodiazepines are included), time intervals in the study, and different methods for evaluating success (survival rate, AChE tissue levels, symptoms, etc.).35 The only solution to these extensive problems seems to be a single testing protocol, although the implementation of such an idea is difficult as each individual laboratory has different

63 access and resources to synthesize, express, or purchase sources of OPs, AChE, and animal models. At a minimum, these issues need to be acknowledged when comparing data from different sources when trying to determine how to further one’s own research.

1.10 Reactivation of Butyrylcholinesterase.

Since BChE is a native human enzyme, has no assigned physiological role, and has high affinity for OP compounds, it has been suggested that BChE can act as a drug that can scavenge OP nerve agents.120–122 Additionally, the availability of BChE in blood serum with high concentrations (~5 mg L–1, about 700 times more than AChE) makes it a suitable candidate to be used as a therapeutic against OP exposure.78,123 Indeed, Ashani and coworkers carried out in vivo experiments in mice and other models to show that stoichiometric quantities of BChE provide protection against nerve agent exposure.120,121,124 Lenz et al. reported that a recombinant human form of BChE called

Protexia™, which was expressed in the milk of transgenic goats, can be a potential bioscavenger against nerve agent poisoning.122,125,126 A recent report by Lenz and coworkers showed that plasma-derived hBChE is a better candidate to be a therapeutic, and this formulation was granted investigational new drug status by the US Food and

Drug Administration in 2006.127–129

Although several reports suggested BChE as a therapeutic approach against nerve agents, stoichiometric quantities of the enzyme are needed. Thus, an enormous quantity of BChE is needed to protect a large human population. Furthermore, recombinant hBChE does not have a sufficient serum lifetime as compared to plasma-derived hBChE, and subsequent dosages are needed for continuous protection, potentially causing several

64 side effects. Regarding post-treatment, i.e., BChE administration post-exposure to an OP agent, it was shown that the post-treatment of BChE, 120 minutes after exposure, provided protection against several dosages of the nerve agent VX.130,131 However, if one waits until the signs of OP poisoning are visible, only 1 out of 6 animals survived after

BChE treatment.132 As a result, despite many favorable properties of BChE in its native form, further improvements are needed for hBChE to be an effective remedy for OP poisoning. Ideally, the OP scavenger needs to have a long serum life that sufficiently overlaps with the therapeutic circulatory lifetime and should be able to hydrolyze the nerve agents catalytically so that large quantities of therapeutic are not needed for administration.

Once an OP inhibits BChE, there is no potent nucleophile available in the active site that can attack the (Ser)O–P bond to hydrolyze the OP molecule and reactivate the enzyme. However, if a potent nucleophile could reactivate the OP-inhibited BChE, then

BChE and the nucleophile can act as a pseudo-catalytic bioscavenger to hydrolyze an OP.

However, none of the developed oximes for AChE demonstrated desirable efficiency of reactivation of nerve agent inhibited BChE. Since most of the oximes were optimized to interact with AChE, they did not have very effective binding affinities for

BChE. Taylor et al. observed that reactivation of phosphylated BChE is not as successful with two common oximes, 2-PAM and HI-6, when compared to phosphylated AChE.133

Furthermore, Jun and coworkers tested more than 20 oximes against AChE and BChE inhibited by the pesticide paraoxon.134 Their work showed that none of the oximes were able to reactivate BChE more effectively than AChE. Similar results were reported

65 recently for several other OPs bound to AChE and BChE.135,136 These results suggest that currently available oximes are not able to reactivate OP-bound BChE, even though AChE and BChE have extremely similar tertiary structure and active site surroundings.

The rate limiting step for a pseudo-catalytic bioscavenger is dephosphylation which must be more rapid then the competing aging reaction in BChE.137 The ability of treatment is also dependent on the method and agents to which an individual is exposed.

For instance, the inhalation of a G-agent causes rapid distribution into the target tissues, so as a post-exposure treatment, bioscavengers will likely not suffice. However, the slow distribution of V-agents from tissue after exposure offers a longer and more viable treatment window.138 An additional issue with regard to catalytic bioscavengers is their specificity for classes, and even specific OPs. Thus, a mixture (cocktail) of multiple bioscavengers or mutants of the same enzyme might be needed to cover a broad spectrum of current and potential OPs.138

1.11 Recent Results for the Reactivation of Butyrylcholinesterase.

Kovarik and coworkers developed a library of non-pyridinium based oximes for the generation of a BChE pseudo-catalytic bioscavenger.139 A total of 39 imidazole and benzoimidazole oximes were synthesized and screened for reactivation of BChE and

AChE, inhibition potency of BChE, cytotoxicity as well as ex vivo and in vivo efficacy.

The reactivators were designed to have a methylene linkage from the imidazole to which a hydrophobic group was attached as a means by which to increase the binding affinity of the reactivators for BChE. To determine the amount of the nucleophilic oximate form present in solution, pKa studies were conducted to determine the acidity of the oxime

66 functional group. Of the compounds, neutral imidazoles were determined to have pKa values of 10.3–10.4, neutral benzimidazoles of 9.7–9.8, cationic imidazoles of 8.0–8.3, and cationic benzimidazoles of 7.4–7.6. These compounds were then screened with

BChE after inhibition by tabun, VX, and (ethyl) paraoxon. Unfortunately, the only recovery of activity was demonstrated for VX-inhibited BChE. From the initial screen, a total of five compounds showed significant potency: I, III, V, VI, X (Figure 30). These compounds were further tested to acquire detailed kinetic parameters to further refine their selection. These compounds were ineffective at recovering the activity of paraoxon- and tabun-inhibited BChE, in comparison to obidoxime. However, for VX-inhibited

BChE, all of the compounds showed effective reactivation constants greater than that of

HI-6 and obidoxime. All of the tested compounds had reactivation maximum of ≥85% in two hours or less. From the determined IC50 values, all of the compounds had a high affinity for the enzyme with values ≤20 µM, significantly lower than traditional oximes.

The imidazolium ring, especially with 1,3-substituted aromatic moieties, showed the highest affinity, and additional modifications may lead to improved oximes. One intriguing result is that the KOX and IC50 for the best oxime VI were similar, suggesting that the size of the bound OP has no effect on the fit of the oxime, potentially allowing for structural design to be determined from un-inhibited BChE structures. AChE screening was also conducted, but none of the five lead compounds showed potency above that of the reference oximes. Toxicity studies were then conducted using HepG3 and THP-1 cell lines using the MTS assay. They determined that these imidazolium oximes possessed higher cytotoxicity than traditional pyridinium oximes. In particular the

67 compounds with 1,3-aromatic substituents were the most toxic. However, given the promise and lower toxicity of compound VI, it was carried through to ex vivo and in vivo tests. The ex vivo study was conducted using human whole blood and showed that 0.1 or

0.5 mM VI and 0.3 µM BChE could degrade 1.5 µM of VX in 25 minutes. Following a few dosing trials, they determined that 2 mg/kg VI and 0.4 mg/kg BChE was capable of protecting animals up to 6.3 LD50 of VX. These results are very promising and are great proof of principle that BChE can be paired with an appropriate oxime to be pseudo- catalytic. However, as has been seen in AChE oxime research, there may be difficulty in finding a broad scope reactivator.

Cl N N Cl N N Br Br NOH NOH

I III Cl

Cl N N N N Br NOH Br NOH V VI

N N Br NOH

X Figure 30. Imidazolium-based oximes for the reactivation of OP-inhibited BChE.139

Kovarik and coworkers took the known affinity of cinchonidine for inhibiting cholinesterases and synthesized and tested three novel oximes for their AChE and BChE reactivation potential (Figure 31).140 Three different Chichona oximes were synthesized, and the resulting products were characterized fully but left in their appropriate epimeric ratios, C1–C3. The compounds were initially screened after human AChE inhibition by

VX, sarin, cyclosarin, tabun, and paraoxon. Of the tested compounds, none of them reactivated more than 25%, and there was no obvious trend based on structure of oxime.

However, for hBChE reactivation, efficacy reached as high as 75% and the best of the synthesized compounds for all OPs was C2. A derivative of C2, lacking the quaternary ammonium, performed the worst of all of the compounds, suggesting that like AChE,

BChE has preference for positively charged groups. Most interestingly, C2 showed relatively broad scope reactivation with reactivation maxima of 60, 70, 45, 30, and 60% for VX, sarin, cyclosarin, paraoxon, and tabun, respectively. When compared to the reference oximes for all of the OPs, the reactivation maxima are lower than the reference; however, this broad scope reactivation with all OPs is unique to this set of compounds.

Additional studies to get more detailed kinetic parameters were conducted as the compounds proved to be potent inhibitors of hBChE. Further investigation of the Ki values for these compounds showed selectivity for hBChE over hAChE and that the values for hAChE were higher than literature values for pyridinium oximes and lower for hBChE. Hunter-Downs analysis showed that these compounds are mixed inhibitors of cholinesterases and showed a preference for the C2 derivative with an N- benzylammonium, suggesting preference for increased aromaticity. Toxicity studies

69 using HepG2 and SH-SY5Y cell types showed that the compounds have an IC50 of ~800

µM suggesting they are not highly toxic. Given their broad scope reactivation potency and their moderate binding affinity and toxicity, this class of oximes requires further development in an attempt to generate hBChE as a pseudo-catalytic therapy.

N HON I

C2 Figure 31. Cinchona based oxime showing broad scope reactivation potential.140

Along with testing their non-oxime reactivators with AChE, Stojanovic and coworkers tested the same library of compounds with paraoxon-inhibited BChE.91

Indeed, some of the compounds they had tested with AChE were capable of reactivating

OP-inhibited BChE, most notably #17 (Figure 32), SP134 and amodiaquine (ADQ)

(Figure 27). Compound #17 is unique in that it only shows activity for BChE approaching 75% recovery, yet it is inactive against AChE. More interestingly, compounds SP134 and ADQ are active for both the reactivation of AChE and BChE, approaching 75% and 70% reactivation, respectively. These initial findings are very promising since as has been demonstrated with AChE, Mannich bases or other general bases can reactivate both cholinesterases, thereby opening up a number of new frameworks outside those of traditional oximes. These new routes present increased

70 opportunity as the exhaustive search for oximes has thus far proven unfruitful for BChE reactivation.

HN O

Cl N

#17

Figure 32. Mannich base for the selective reactivation of paraoxon-inhibited BChE.91

The very efficient and broad spectrum reactivators developed by Renard and coworkers proved to be efficient in the reactivation of BChE as well.110 The synthesized compounds were tested with hBChE after inhibition by VX, sarin, cyclosarin, tabun, and paraoxon (the same OPs as the AChE study, Figure 18). The compounds showed good reactivation potential with all OPs, except sarin, and only moderately for tabun. Of the tested compounds, 18 (Figure 24) had the highest effective reactivation rate in the study.

Also, all of the synthesized compounds outperformed HI-6 for reactivation of paraoxon- inhibited hBChE. Despite their reactivation potential, it is worth noting that all of the tested compounds displayed higher dissociation constants than that for hAChE, suggesting further optimization with regard to binding may further increase their potency.

Using the molecule edrophonium as a scaffold for development, Taylor and coworkers synthesized a library of novel oximes in efforts to develop a pseudo-catalytic bioscavenger of OPs.141 Structural adjustments on the molecules included changes to the position of the oxime, the presence or absence of a hydroxyl group, and the size of the alkyl substituents on the aniline group. All compounds were tested for their reactivation efficiency with both hAChE and hBChE. They determined that for hAChE, most of these compounds were inactive. In fact, only a single oxime showed any activity relative to 2-

PAM for cyclosarin-inhibited hAChE, yet still performed significantly worse than the reference oxime HI-6. On the contrary, one molecule, TAB2OH (Figure 33), showed faster reactivation for OP inhibition by paraoxon, cyclosarin and VX, when compared to the reference oximes, but not sarin. Any structural modification of TAB2OH reduced the reactivation potency. Analysis of the reactivation efficiency showed that TAB2OH was superior for paraoxon, cyclosarin, and VX, but slightly lower for sarin and tabun when compared to 2-PAM. The pKa of TAB2OH was determined through three different methods each giving a slightly different result. The hydroxyl pKa was determined to be

7.50, 7.65, or 7.21 dependent on the method used. For the oxime, the pKa was determined to be 10.6, 10.1, or 9.3 dependent on the method, showing that the oxime is significantly protonated at physiological pH. They also confirmed that 2-PAM was a better nucleophile than TAB2OH as demonstrated by oximolysis studies. Oxime-assisted degradation of cyclosarin, VX, sarin, and soman were all found to be rapid, within the first few minutes of reaction, and in the case of cyclosarin and VX to be even more rapid.

Similar experiments were conducted with human whole blood in ex vivo studies, showing

72 near complete cholinesterase activity recovery, 80%, after 100 minutes. Within the first

10 minutes following TAB2OH addition up to 60% activity for VX, 30% activity for cyclosarin, and 15% activity for sarin was recovered. To continue toward in vivo testing, the toxicity of the compound was determined to be 100 mg/kg in near direct agreement with that of 2-PAM. The in vivo protective indices showed that they were actually higher for 2-PAM rather than for TAB2OH, in contradiction to the results of the in vitro tests.

When mice were dosed with BChE, TAB2OH, and atropine however, protective indices increased for paraoxon and sarin. Further tests show that small amounts of hBChE and

TAB2OH when administered together with atropine pre- and post-exposure could increase protection for paraoxon further demonstrating the utility of the combination in generating a pseudo-catalytic system. Unfortunately, it was determined that TAB2OH was 18-fold less concentrated in the brain than in the blood plasma, suggesting a poor ability to penetrate the BBB. In total, the results of the study are promising as the toxicity of the compound is similar to known therapeutics and shows potential as a reactivator with a catalytic bioscavenger.

NOH

TAB2OH Figure 33. Edrophonium derived derivative TAB2OH for the reactivation of hBChE.141

Radic and coworkers followed up their discovery of the efficient hBChE reactivator TAB2OH by investigating additional derivatives of N-alkylimidazole derivatives of similar compounds.142 The investigation began with review of the screening of six imidazole aldoximes for their potency of reactivation of hBChE inhibited by low-toxicity analogues of sarin, cyclosarin, and VX in addition to (ethyl) paraoxon.

The most efficient oxime was identified as RS-113B for all four OPs (Figure 34). Due to its universal nature of reactivation, additional derivatives of the RS-113B scaffold were prepared and evaluated with the same set of four OPs. Of the new oximes synthesized, compound Imd3 and Imd4 were identified as having good broad-spectrum capacity as a reactivator (Figure 34). The trends identified by the investigators were that the enzyme seems to have a preference for hydrophobic groups positioned four methylene units away from the imidazole ring, which may be used in future oxime development. The most interesting trend noted, and one that is contradictory to most previous findings, is that introduction of charge, by converting the imidazole to an imidazolium, reduced reactivation efficiency. The best compounds from the structural screen were then further subjected to additional modification of the imidazole group. Again, the finding was confirmed that a positive charge on these structures does not increase reactivation efficiency, and the only exception was for a single oxime in which its reactivation potency with a cyclosarin-surrogate inhibited hBChE was increased 50-fold by the charge. Though no benefit was gained in terms of a universal reactivator at the time of this report, Imd3 and Imd4 are the fastest reactivators of inhibited hBChE. All of this work was repeated with hAChE but unfortunately these structural modifications provided

74 no benefit and all compounds were either comparable or slower than the reference oxime

2-PAM. Ultimately, this study shows that imidazole-based oximes can have very rapid rates of reactivation of hBChE inhibited by a variety of OPs. Given the lack of permanent positive charge on these molecules, it is hypothesized that these oximes will better cross biological barriers such as the BBB, making in vivo trials of great interest.

HON HON HON

N N N N N N

RS-113B Imd3 Imd4

Figure 34. Imidazole based oximes found by conducting structural modifications from the originally tested imidazole aldoximes by Radic and coworkers.142

Kovarik et al. continued the investigation of K-oximes for the application to generate pseudo-catalytic bioscavengers.143 They determined that both oximes K117 and

K127 (Figure 35) were capable of reactivating tabun-inhibited hBChE, with K117 having a maximum of 80% reactivation in 50 minutes. They noted that reactivation kinetics deviated towards the end of the reactivation period, and this was attributed to three possibilities: (1) tabun-inhibited BChE is aging, (2) re-inhibition by the phosphylated oxime, and (3) different reactivation kinetics of the two enantiomers of tabun-inhibited

BChE. Molecular modeling of the docked oximes within AChE and BChE noted that the major difference in affinity was attributed to the lack of aromatic groups in the peripheral site leading to differential binding. Overall, they concluded that introduction of an

75 oxygen into the linker chain generates advantageous interactions in the BChE enzyme leading to reactivation.

2 Br HON NOH N N O

K117

HON 2 Br NH2 N N O

K127

Figure 35. K Oximes capable of reactivating tabun-inhibited BChE.143

Worek and coworkers approached the pseudo-catalytic bioscavenger problem from a different angle, taking the currently available oximes, obidoxime and HI-6, and testing their ability to scavenge VX using different blood sources.144 An interim solution was proposed that uses augmentation of AChE and BChE using blood products and currently approved oximes to scavenge OPs in the blood as more effective BChE reactivators have not yet been approved and are still in development. The samples of blood used were whole blood, human erythrocyte ghosts, plasma, packed red blood cells, and fresh frozen plasma. The whole blood system proved to be a better scavenger when paired with the oxime HI-6, and this trend was also observed for human erythrocyte ghosts. On the other hand, plasma showed a comparable effectiveness for obidoxime and

HI-6. When looking at storable blood sources, they determined that HI-6 and packed red 76 blood cells worked better than with obidoxime. The observed rate of detoxification for freshly frozen plasma, however, proved to be slower and incomplete. All of these studies were conducted ex vivo and further in vivo trials are required.

1.12 Resurrection of Aged Acetylcholinesterase.

While the reactivation of inhibited AChE has proven to be a challenging task for the past 70 years, the process is only more complicated by the competition of aging with the rate of reactivation. Since aging is the spontaneous dealkylation of the alkyl (R) group on the (inhibited) phosphylated serine residue, leading to an anionic phosphylated serine

(Figure 7), aging of AChE removes the alkyl side chain of the original OP. Thus, the aging process leads to only three unique aged AChE structures – phosphonates (G and V agents, with the exception of GA), phosphates (paraoxon, DFP, and other pesticides), and phosphoramidates (GA).

Despite many attempts and across decades of effort, aging was considered irreversible, and until 2018, there were no in vitro or in vivo reports of any recovery of the activity of aged AChE. The enzyme was hypothesized to adopt a stabilized aged conformation, perhaps by forming a strong salt bridge between the phosphylated oxyanion at the serine residue and the histidinium ion of the catalytic triad, and perhaps also due to charge-charge repulsion between the phosphylated oxyanion and typical reactivating nucleophiles, such as pyridinium oximates or their neutral forms. Whether the nucleophile possesses a full negative charge or not, the increase in electron density leading to its nucleophilicity generates a repulsive interaction.145,146 Multiple attempts were made to convert the phosphylated oxyanion aged form to a realkylated (i.e.,

77 inhibited) form of AChE using a variety of alkylating reagents, including sulfonates, haloketones, methoxypyridniums, and sulfoniums.88,145–151 However, while some of these alkylating strategies were effective in model systems, these attempts proved to be unsuccessful when they were attempted in vitro with aged AChE.

In 2018, our research group published the first report that demonstrated the in vitro recovery of activity from aged AChE by the use of a small drug-like compound, a process that is termed “resurrection” as the aged (dead) form of AChE is brought back to life as the active, native form.147 In the following, we will outline some of the major difficulties for the resurrection process with the aged form of cholinesterases.

1.13 Early Attempts at the Realkylation of Aged Acetylcholinesterase.

One of the first reports to attempt realkylation of phosphonate anions, as models of aged AChE after exposure to authentic chemical nerve agents, was conducted by

Stevens and coworkers.150,151 A library of alkylsulfonates was synthesized with the inclusion of a quaternary ammonium moiety to aid in the binding to AChE. Structural modifications were made with regards to the linker length between the alkylating moiety and the binding portion in addition to changing the alkylating groups and the binding moiety to triethylammoniums and pyridiniums. Although a report of their synthetic success was detailed, the only compound that appears to have been published, due to its poor reactivity with a model phosphonate anion, is 1-methyl-3-

(methylsulfonate)pyridinium, PyrSulf (Figure 36). The compound was tested for the alkylation of a number of anions including isopropylmethylphosphonate and p- nitrophenylmethylphosphonate. Unfortunately, the results of the model study were

78 limited as the compounds reacted slower with phosphonates than with nearly any other anion, except nitrate. These initial results were troubling given the structural similarity to known binders but more troublesome due to their clear propensity for off-target alkylations. This aspect is very important – many alkylating agents are possible, but the selective re-alkylation and eventual resurrection of aged AChE is critical, and off-target activity must be minimized or preferably eliminated.

N ClO4 O S O O

PyrSulf Figure 36. Sulfonate alkylator in first attempt to alkylate phosphonate anions.150

A second 20th century attempt to realkylate a model phosphonate and aged AChE showed only success with model nucleophiles, but the authors did draw some interesting conclusions based on rate changes for hydrolysis and alkylation due to distant carbonyl groups.149 Boldt and coworkers used sodium p-nitrophenylmethylphosphonate as a model nucleophile for kinetic studies of alkylation and hydrolysis by strong alkylators composed of various functional groups including α-haloketones, α-halodiketones, alkyl and arylhalides, alkyl and arylmesylates, α-haloamides, α-haloximes, chloromethylphosphonic acid, α-halocarboxylic acids, α-haloalcohols, and epoxides.

(Many of these “warheads” are often used for suicide inhibition of select biological targets.152) As was anticipated, compounds with electron-withdrawing groups resulted in both faster alkylation reactions due to an increase in electrophilicity but also the rate of 79 hydrolysis was increased for the generated mixed phosphoryl esters. The same trend holds true for hydrolysis of the alkylating agents themselves. One of the more interesting findings was the intramolecular participation of carbonyl groups in the hydrolysis of the generated alkylated phosphylated products. By placing either direct nucleophiles, such a phenols or oximes, on the ring or using activated carbonyl groups, with electron-deficient character favoring hydrate formation, or amides, the authors were able to promote hydrolysis, and likely through intramolecular catalysis. Following initial kinetic studies, in vitro reactivation of aged AChE was attempted with seven representative compounds:

VII, VIII, XXV, XXXIII, XXXIV, XXV, XXXVI (Figure 37). The compounds were incubated with soman-aged AChE, but none proved to be effective in vitro. Control studies did show that all of the compounds were progressive inhibitors of the native enzyme, owing to their very high electrophilic character. Although no hits were identified by this study, their efforts did present some key themes regarding carefully balancing of the electrophilic character so as to not achieve off-target activity but also the inclusion of chemical groups to aid in the cleavage of the “realkylated” phosphylated product that is generated following an effective reaction.

Br Na NO O O 2 N O Br P Br O O O Br O VII VIII XXV Model aged Ser

Br OMs N O O S N O Br Br O O XXXIV XXXV XXXVI

Figure 37. Strong alkylator groups used unsuccessfully in vitro to realkylate soman-aged AChE.149

One of the first modern approaches to alkylate aged AChE mimics was conducted by Quinn and coworkers through the use of N-methyl-2-methoxypyridiniums as methyl- transfer reagents.147 Quinn was trying to recover what has long been treated as a dead enzyme and thus coined the phrase “resurrection” in which the aged adduct is reverted back to the native form of AChE. The selection of N-methyl-2-methoxypyridiniums was made based on the predicted balancing of selectivity, toxicity, and susceptibility to hydrolysis in addition to the similar structural features to that of the FDA approved oxime, 2-PAM. In total, nine different compounds were synthesized and their methyl- transfer rates were determined using the aged AChE model of sodium methyl

1 methanephosphonate in DMSO-d6 using H NMR. The presence of small amounts of water also allowed for the competitive formation of methanol via the hydrolysis reaction.

The rates of reaction were determined using numerical integration and displayed various

81 methyl-transfer rates. They noted that the rate of methyl transfer, methoxy NMR shift, and Swain-Lupton field and resonance parameters allowed for estimation of the rate of reaction based on the substituent and the chemical shift of the methylmethoxypyridiniums. The effects of each parameter on the rate of reaction showed that the methyl transfer is sensitive to both resonance and field effects and can be tuned as such to increase or decrease the rate of reaction appropriately. The highest rate of methyl transfer was observed for the N-methyl-2-methoxypyridinium, 1b (Figure 38), with a 3-F substituent, achieving 40% methyl transfer in less than 10 min. Analysis of the rates of hydrolysis in D2O showed that some compounds were stable to hydrolysis for an entire week. In general, the rate of hydrolysis in D2O was two orders of magnitude slower than methyl transfer. However, when hydrolysis was measured in pH 7.2 phosphate buffer, the hydrolysis rates were faster, but still slower than methyl transfer in DMSO, and compound 1b showed methyl transfer to the phosphate buffer. Both the rates of methyl transfer and hydrolysis in all solvent systems showed that these compounds were stable under near physiological conditions and unselective alkylation should be minimal, making these compounds tunable alkylating agents of model phosphonate anions.

BF4 Na O N O P O O F

1b Model aged Ser Figure 38. N-methyl-2-methoxypyridnium structure and model phosphonate anion used to study methyl transfer by Quinn and coworkers.147

In a subsequent study, Quinn and coworkers synthesized a larger library of methoxypyridinium compounds and drew inspiration from investigated and functional oximes and determined the inhibitory potency of these methoxypyridiniums to ensure that inhibition was reversible with AChE.148 In total, 32 different compounds, representing seven (Pyr I–VII, Figure 39) different classes, were synthesized and their

IC50 values were then determined with hAChE. For the Pyr I framework, potency varied from 7–70 µM, being slightly lower than structurally similar 2-PAM. The bis- methoxypyridiniums of the Pyr II framework showed increased binding affinity with

IC50 values below 1 µM, consistent with previous observations from numerous oxime studies. Interestingly, bis-substituted compounds for Pyr III and Pyr IV did not show a reduction in IC50 when linked at the pyridinium, suggesting that the linking position can have a significant effect on potency. Other peripheral site linked compounds such as those to Pyr V, the dimethoxyindanone core from donepezil, and Pyr VI, a tetrahydroisoquinoline, reduced the IC50 values to the nM range, further demonstrating the effectiveness of tethering the potential warhead to peripheral site ligands. The reduction was not as dramatic when using coumarin-linked compounds (Pyr VII) with an

83 observed reduction of only one order of magnitude. Although the binding affinities and realkylation efforts with model nucleophiles were successful, preliminary in vitro evaluations showed no recovery of activity for aged AChE. Yet, these compounds present a core framework and ideology to continue to investigate methods of realkylation of aged

AChE.

OTf F BF O 4 O O R R N N O F n O O O O Pyr I Pyr III Pyr V

OTf OTf O O OTf Pyr VII

R R N N N O CF3 O n n

Pyr II Pyr IV Pyr VI Figure 39. Various core frameworks and peripheral site ligands employed for hAChE inhibition by Quinn and coworkers.148

An approach making use of a sulfonium-based alkylating agent, selected for its observed stability, selectivity, and toxicity, for the realkylation and subsequent reactivation of methylphosphonate AChE was investigated by Ganguly and coworkers.146

The authors made use of molecular docking studies followed by molecular dynamics simulations (MD), and subsequent steered molecular dynamics (SMD) to investigate the binding and egress pathways of three novel sulfonium alkylators (Sulf 1–3, Figure 40).

The final structure, Sulf 3, actually possessed a nucleophilic oxime moiety and was investigated to see if it adopted a pose that would allow for additional reactivation with

84 the aged Ser residue. The first step was quantum mechanical investigation of the alkylation reaction between a model soman-aged phosphonyl adduct with the proposed alkylating agents using the M05-2X/6-31G* level of density functional theory. For Sulf

1–3, the activation barriers were determined to be 31.4, 27.9, and 26.9 kcal/mol with all of the reactions being exothermic with energies of –6.5, –9.9, and –11.4 kcal/mol, respectively. The same method was used to determine the activation barrier for the aging of a model soman-inhibited Ser and found to be 37.8 kcal/mol, suggesting that for these three compounds, the aging process is predicted to be reversible.153

Using the same methodology, an unsubstituted 2-methoxypyridinium was also evaluated and determined to have a larger barrier to realkylation than all of the sulfonium compounds investigated at 32.3 kcal/mol, only being exothermic by 0.6 kcal/mol. These energetic results suggest why in vitro realkylation was not observed. There was an energetic gain by moving from Sulf 1 to Sulf 2 and incorporating a pyridine ring. This observation then inspired the design of Sulf 3 which still had the pyridine ring but was additionally tethered to a nucleophilic oxime as a means by which to accomplish the realkylation by the sulfonium and then subsequent reactivation by the oxime. The energetic study showed that Sulf 3 was capable of the realkylation so an additional investigation of the nucleophilicity was undertaken. The determined nucleophilicity index of 2-PAM was 0.0027 eV and that of Sulf 3 was 0.0108 eV, thus implying that the molecule Sulf 3 is capable of accomplishing both electrophilic and nucleophilic tasks.

The affinity of Sulf 3 was then determined by analyzing the docking with both AChE and

BChE followed by SMD simulations with TcAChE, a reference reaction was determined

85 to be in good agreement with the crystal structure used for BChE and a 1-methyl-2-

(pentafluorobenzyloxyimino)pyridinium. Compared to the reference compound, Sulf 3 was computed to have a shorter distance to the reactive phosphonate anion, being 3.840

Å from the alkylator. To further investigate those interactions contributing to the binding and egress of Sulf 3, the ligand and enzyme were subjected to SMD. It was shown that during the course of the simulation, as Sulf 3 is pulled away from the enzyme, it breaks a number of hydrophobic interactions over the course of egress, pointing to significant stabilization within the enzyme active site. Most notably however, is the rotation of the drug candidate within the enzyme as Sulf 3 is pulled out as measured by multiple metrics.

This is significant as it presents the possibility of reactivation by the oxime moiety as the molecule clearly rotates toward the phosphylated Ser residue. However, the compound used in the SMD simulation is still a sulfonium and the enzyme itself is aged and not inhibited, yet according to the authors, the size of the enzyme active site does allow for rotational freedom. These results coupled together then offer computational evidence for the continued investigation of sulfonium-oxime compounds for the reactivation of aged

AChE.

S N N N HON

Sulf 2 Sulf 3 Figure 40. Computationally investigated sulfonium realkylators of aged AChE.146

Wallqvist and coworkers used computational tools to develop novel reactivators of methylphosphonate-aged AChE in which the reactivation was direct and did not proceed through a realkylated intermediate.145 Drawing inspiration from their research on

CapD, which functions similar to serine proteases but having a single threonine residue rather than a catalytic triad, they noted that β-aminoalcohols may activate the catalytic threonine residue. Since β-aminoalcohols are neutral and become charged late into the nucleophilic attack, it was hypothesized that this functionality would serve as a sufficient nucleophile for aged AChE reactivation due to its proton-transfer ability to quench the negative charge of the phosphylated Ser while being neutral should allow for BBB permeability. To achieve selectivity for the AChE active site, two molecules were then designed, Molecule 1 placed the β-aminoalcohol group onto a pyridinium ring, drawing inspiration from the oxime 2-PAM, and Molecule 2 was a constitutional isomer removing the nitrogen from the aromatic ring and placing it on the periphery (Figure 41).

To determine the ability to reactivate aged AChE, the molecules were first docked within the enzyme and then molecular dynamics simulations in solution were performed to allow for flexibility of the enzyme. Following these initial calculations, QM/MM studies were conducted in which the catalytic triad, oxyanion hole, E199, Y121, five water molecules, and the ligand were all modeled using quantum mechanics, while the remainder of the enzyme was then evaluated using a molecular mechanics force field.

Despite the structural similarity to 2-PAM, Molecule 1 was determined to have a 50 kcal/mol activation barrier with the more reactive S enantiomer. The authors concluded that for this particular molecule, the reaction proceeded in a single step; however, there

87 was insufficient disruption of the salt bridge between the phosphylated oxyanion and the histidinium to allow for proton transfer from the ligand. It was thus concluded that

Molecule 1 would not be an effective reactivator of aged AChE. In the case of Molecule

2, the overall reaction was exothermic by 1.1 kcal/mol and the reaction proceeded through an intermediate with an energy of 7.2 kcal/mol. The overall activation barrier for this reaction was 25.3 kcal/mol, proceeding in two steps: the first step being the activation of the alcohol by an intramolecular proton transfer and the second step being the nucleophilic displacement of the phosphylated moiety. The studied protonation state was not predicted to be the one observed in solution. The authors determined the pKa values of the groups to be 9.4 for the β-amine, then 5.1 and 2.8 sequentially for the anilines. The authors, however, justified these differences, noting that enzyme active sites can have significant effects on pKa values and also that structural modifications can further alter the pKa values of the studied molecule. Initial estimations of the rate of reaction were predicted to result in 50% reactivation of aged AChE in approximately 4 minutes. This estimation makes multiple approximations and does not account for any binding energies but rather just activation barriers and lacking entropic contributions.

Such β-aminoalcohol functional groups have not been tested experimentally, to our knowledge.

OH OH

H2N H2N

NH2 H2N NH2

Molecule 1 Molecule 2 Figure 41. Computationally investigated aminoalcohol realkylators of aged AChE.145

1.14. Quinone Methides as Alkylating Agents of Model Nucleophiles.

Initial attempts to find alkylating agents for aged AChE have proven difficult due to the primary issues regarding binding, selectivity, stability, and toxicity. During the

1990s to the 2000s, there was a number of publications noting the utility of quinone methides (QMs) as alkylating agents of various biologically relevant molecules including amino acids, DNA and nucleic bases, and structural modifications could be used to modulate the selectivity and reactivity of the alkylating agents to promote or repress specific parts of the reaction. In 1999 it was reported that these QMs could alkylate phosphodiesters which are structural analogues of pesticide-aged AChE. The investigation of such molecules and reactions could then be used as a method to develop effective therapeutics for in vitro and in vivo realkylation of aged AChE.154–156

The first report of alkylation of a phosphodiester was published by Turnbull and coworker in 1999.154 This study made use of a p-quinone methide generated from mixing

2,4,6-trimethylphenol with silver oxide in organic solution (Figure 42). Upon the generation of the QM, an organic solution of tetrabutylammonium dibenzyl phosphate was added, but no product was observed by 1H NMR until the addition of MsOH. Studies

89 showed that the reaction was completely dependent upon the addition of MsOH, and was in fact reversible upon manipulation of the pH through addition of potassium carbonate.

Changes in acid strength, ensuring that the conjugate base was not nucleophilic enough to compete, proved to be unsuccessful. Through changes in the concentration of the phosphate nucleophile, it was determined that there was a pre-equilibrium with the added acid reducing nucleophilicity and lowering overall conversion. Additional experiments were conducted with dibenzyl and diethylphosphoric acid serving as both the acid and the nucleophile to observe that the equivalents of the phosphoric acid pushed the reaction to a high conversion of 83% for 2.7 eq of dibenzyl and 3.9 eq of diethyl. These same reactions showed that the rate of reaction was dependent on both the acidity and nucleophilicity of the phosphoric acid, a finding that needs to be considered and evaluated when designing a therapeutic for realkylation of aged AChE. The authors concluded that the QM intermediate had to be activated to allow for attack by the weak phosphates and thus overall, the reaction was acid catalyzed. This initial study was an important and first demonstration of the potential for alkylation of aged AChE model compounds with QMs, but they did not address aqueous solubility, competitive hydrolysis or the generation of the QMs in solution.

CDCl3/CD3CN O OH O P HO OR OR O P O OR OR

R = CH2Ph, Et Figure 42. Initial reaction showing the potential of QM alkylation of phosphodiesters.154

The initial study was followed up two years later and addressed the competitive hydrolysis reactions by moving to mixed aqueous solutions (Figure 43).155 The initial study was conducted in 10% aqueous buffered solutions with the same p-QM used in the initial studies and diethylphosphate as the choice of nucleophile. The maximum conversion to the phosphodiester was 16% within 9 min, and by 10 min, the conversion to the benzyl alcohol had reached 75%. However, kinetic determination of the reaction showed that the hydrolysis rate was only 2.2% that of the alkylation rate despite the concentration of water being 220 times that of the phosphate nucleophile. Additional studies were carried out in which 28.5% aqueous buffered solutions were used and the nucleophile was varied between an inorganic phosphate and dibenzyl, dibutyl, and diethyl phosphodiesters at a pH of 4.0 and 7.0. It was determined for inorganic phosphate, that the hydrolysis of the QM is a specific acid catalyzed process. For the tested phosphodiesters, the authors reported that dibenzylphosphate alkylation is five times faster than dibutyl and 10 times faster than diethyl at pH 4.0, again showing that the nucleophile has a significant effect on the rate of reaction which will be key to drug

91 design. For the pH 4.0 studies, dibenzylphosphate reacted 3700 times faster than water and even the slowest phosphate (diethyl) still reacts 370 times faster than water.

Unfortunately, when they used pH 7.0, it was determined that hydrolysis was the dominant reaction and almost no rate of change was seen when varying the concentrations of the phosphates in solution.

Additional studies with phosphate nucleophiles were also reported by Turnbull and coworkers, using a cyclization reaction to increase the synthetic yield of phosphate reaction with the QM intermediate.157,158

CDCl3/CD3CN O OH OH O P HO OR OR O P OH O OR OR R = CH2Ph, Et, Bu Figure 43. Subsequent study showing the potential of QM alkylation of phosphodiesters.155 Thus QMs are capable of making similar transformations to the one desired for the realkylation of aged AChE, yet the selectivity and reactivity still require substantial adjustment given high conversions to the benzyl alcohol previously reported. Rokita and coworkers reported on the ability to appropriately modify formation and stability of QMs using substituent effects.156 Their study tested the effects that substituents have on product distribution as well as reversibility by changing the groups attached meta and para to the QM methylene position, while intentionally avoiding the steric complications caused by ortho substitution. The initial study used a silyl-protected phenol as a quinone

92 methide precursor (QMP) which, upon treatment with a fluoride source, would cleave the

Si–O bond, and then would be followed by the expulsion of the acetate leaving group, in order to generate the appropriate QM. Three different QMPs (QMP1–3, Figure 44) were used with para substitution, relative to the methylene, and substituted by either a methyl, hydrogen, or methyl ester. Each of these QMPs was then reacted with deoxycytidine (dC) and the rates of formation and decomposition of the product were monitored. For QMP2, the more electron-rich precursor, initial formation of the product was rapid and was complete within 5 hours, yet the decomposition was also rapid with near complete decomposition, half-life of 5 hours. When this is compared to QMP1, product formation is slower, taking approximately 10 hours with a slower rate of decomposition as well, half-life 24 hours. Finally, QMP3 slowly formed product reaching a maximum at about

40 hours and showed no decomposition. These results then suggested that compounds with increased electron density will more rapidly form the QM than those that are electron deficient. These results were further supported when the same set of compounds was incubated with four different deoxynucleosides in a competition experiment. It was again noted that the electron-deficient QMP3 formed product at a slower rate; for example, for a dA complex with QMP3, it took 10 hours to reach its maximum but for

QMP2 only 0.5 hours. The more interesting of the findings was that the overall product distribution changed based on the electronic substituents of the ring, reflecting that some adducts had such poor leaving groups that the QM was never regenerated in the case of

QMP3 but was for QMP1–2. To further study the effects that electron-donating and electron-withdrawing groups have on QM generation, a new set of QMPs were used to

93 ensure that the reaction was on a detectable time scale. To this end methoxy, carboxymethyl, and unsubstituted phenols were used with a morpholino- leaving group,

QMP4–8. The stability of these QMPs was then monitored in aqueous solution for the generation of their benzyl alcohol derivatives and the results emphasized the previous observations even further.

The rate of formation of the QM is significantly affected by the ring’s electronic effects, and substitution is more sensitive to resonance effects as compared to field effects. To determine the effects that the same substituents had on nucleophile addition, a final set of QMP9–14 was used where para-substituted Mannich bases with trimethylammonium leaving groups were converted to their QMs via photolysis and the rate of product formation determined based on competition between three nucleophiles: water, morpholine, and 2-mercaptoethanol. It was determined by these studies that electron-withdrawing groups greatly increase the electrophilicity of the QM as is demonstrated by multiple orders of magnitude increase in the rate of reaction. The opposite is observed for electron-rich substituents in which the rate of reaction for nucleophiles is reduced by approximately a factor of four. Additionally, selectivity was reduced as stronger electron-withdrawing groups are included as is expected by the reactivity-selectivity principle.

Ultimately, this study showed that quinone methide precursors can generate their

QMs in solution via multiple routes: thermally, by deprotection, and photochemically; moreover, the properties related to reactivity, selectivity, and reversibility can all be altered by changing the substituents on the aromatic ring. With these conclusions and the

94 results with alkylation of phosphodiesters, QMPs may serve as viable options for the design of small-molecule realkylators for aged AChE.

OH OH OTBS N NMe OAc 3 O Z X Y W

QMP1 X = H QMP4 Y = H; Z = COOMe QMP9 W = OMe QMP2 X = Me QMP5 Y = COOMe; Z = H QMP10 W = H QMP3 X = COOMe QMP6 Y = H; Z = H QMP11 W = Cl QMP7 Y = OMe; Z = H QMP12 W = COOMe QMP8 Y = H; Z = OMe QMP13 W = CN QMP14 W = NO2 Figure 44. QMP structures used to study the aspects of QM reaction by Rokita and coworkers.156

1.15 Mannich Bases for the Alkylation of Aged Acetylcholinesterase.

Given the proven potential of QMs and QMPs, Hadad and coworkers began an initial study to determine if these types of compounds would be sufficient to recover the activity of aged AChE.159 A total of eight different compounds were synthesized on the basis of a vanillin framework with various amine leaving groups in both the protonated and deprotonated forms, Van1–4 and VanCl1–4 (Figure 45). First, the compounds were tested with model nucleophiles (4-methylbenzenethiol, piperidine, and benzyl alcohol) for their realkylation abilities. The compounds were incubated with the nucleophiles at

100 °C for three hours and the conversion was observed. For the strong sulfur-based nucleophile, conversion was ~85%, the weaker piperidine nucleophile was 70–80%, and the benzyl alcohol was very limited, only 0–16%. Thus, the compounds were determined to be sufficient alkylators with modest activity and notably low hydrolysis. The

95 alkylation with a model phosphonate was then attempted with Van1 and VanCl1 and monitored by HPLC/MS; however, only trace amounts of alkylation were observed.

To determine if the QMPs have any affinity for the active site of hAChE, molecular docking studies were conducted in which the percentage of docking poses with the benzylic carbon within 5 Å of the phosphylated oxyanion was used as a metric of a reactive conformation. The docking studies suggested a preference for the pyrrolidine and piperidine leaving groups. This result was further confirmed for pyrrolidine when the compounds were subjected to MD simulations, again using the benzylic distance to the oxyanion of the aged phosphonate as a metric of reactivity. The MD simulations showed that of the amines, the pyrrolidine compound spent the most time within the enzyme active site. The MD simulations also showed a strong preference for those compounds which are protonated at the amine, likely due to the similarity to ACh. To try and confirm the accuracy of the docking and MD results, IC50 values were determined for native

AChE, as an assay to determine affinity for the aged enzyme is unknown. For the native enzyme, the preference for binding was determined to be piperidine > pyrrolidine > dimethylamine > morpholine, and IC50 values were on the order of 0.227 – 1.83 mM.

Although the ordering is not the same as that concluded for the docking and MD, this is reasonable as the active sites of native and aged AChE are substantially different. All of the compounds were shown to be competitive or near competitive mixed-type inhibitors and confirmed a preference for the aged active site. A realkylation assay of the HCl salts of all of the compounds was conducted with eeAChE, but unfortunately no recovery of activity was observed in the initial study. Thus, it was proven that QMPs are capable of

96 serving as moderate alkylating agents with model nucleophiles and the structures do have affinity for the AChE active site as was demonstrated by inhibition studies, molecular docking, and MD simulations. Given the tunabilty of the reactivity presented previously, the concept was further carried forward for additional optimization.

Cl X

O OH

Van1 X = N(CH3)2 Van2 X = N(CH2)4 Van3 X = N(CH2)5 Van4 X = N(CH2)4O + VanCl1 X = HN(CH3)2 + VanCl2 X = HN(CH2)4 + VanCl3 X = HN(CH2)5 + VanCl4 X = HN(CH2)4O Figure 45. QMP structures used in reaction with nucleophiles for proof of principle alkylation.159

Inspired by the positive results of the nucleophilic alkylation and affinity for the

AChE active site, Hadad and coworkers continued their investigation of computational libraries of QMPs.160 These computational investigations showed that the AChE active site had a higher affinity for pyridyl QMPs over phenyl frameworks, and these results are certainly consistent with effective AChE reactivators being mono- or bis-pyridinium compounds. A library of compounds was synthesized including 13 derivatives or 2- alkylamino-3-hydroxypyridines with variation of the amines. To test whether these compounds were capable of reactivation of aged AChE, both PiMP (Figure 46), a 3- cyano-4-methylcoumarin analogue of soman (GD),161,162 and a model phosphate (DFP)

97 were selected to generate methylphosphonate-aged and isopropylphosphate-aged forms of AChE. The OPs were incubated with eeAChE for a sufficient time to ensure the enzyme was completely aged. In the case of PiMP, a double aging cycle was employed in which the enzyme was aged and then reactivated with 2-PAM and aged again with the same OP, and this procedure ensured a low baseline for detecting eventual resurrection of

AChE activity. Using various concentrations, the 13 compounds were then incubated with aged eeAChE for 24 hours at pH 8 in the presence of a nonselective reactivator

NH4F, and to ensure nothing was trapped in the inhibited state following realkylation, a second 2-PAM treatment was conducted. Native AChE activity was then detected by

Ellman’s assay. Surprisingly, a number of molecules showed activity above baseline and the results followed relatively similar trends. Interestingly, the methylphosphonate-aged

AChE showed higher levels of recovered activity than that of the isopropylphosphate- aged AChE. The results for the methylphosphonate-aged AChE show a trend related to steric effects of the amine. The N-methyl-N-ethyl substituted amine recovered ~9.5% of native activity and decreased until it reached the 5-membered heterocycle pyrrolidine C8

(Figure 47) which was the most active at 12%. The isopropylphosphate-aged AChE showed similar trends, but the results were not as pronounced and showed some tolerance for N-methyl-N-isopropylamine, N,N-diethylamine, and piperidine. The compound with the pyrrolidine ring was chosen as a lead compound and was pursued in subsequent studies.

Kinetics studies were conducted on an extended timeframe, although in this perturbation of the assay, an external reactivator was not included. Activity of the aged

98 samples were recovered at linear rates out to four days and successfully resurrected 32.7 and 20.4% of the activity for methylphosphonate-aged and isopropylphosphate-aged eeAChE, respectively. But importantly, these results also indicated that these compounds are not only resurrectors of aged to the native state but also reactivators of the inhibited

(or realkylated) form. To determine the reactivation potential of these compounds, four of them were selected and incubated with a VX analogue, EMP (Figure 46), for 24 hours at high concentrations. Two of the compounds completely recovered the activity of the enzyme up to the level of the positive control and the other two reached 60–70%. These results are in good agreement with a number of studies which found that Mannich bases are reactivators of OP-inhibited AChE.91,113,114,116

CN O P O O O O

PiMP

O P O O O O

EMP Figure 46. Nerve agent analogues used in biological assays.160

While the kinetic assays confirmed recovery of the native enzyme, bottom-up proteomics were conducted to try and isolate the enzyme in the native, inhibited

(realkylated), and aged states. A solution of the lead compound C8 was incubated with aged eeAChE for 11 days before being digested with trypsin. Positive and negative 99 controls were prepared in parallel. The negative control showed only aged AChE, the positive control revealed only native AChE, and the sample treated with a QMP showed both aged and native AChE following digestion, but not realkylated – and in a ratio consistent with the kinetic assay for resurrected activity. These results confirmed the kinetic recovery of activity of the aged enzyme to the native state, clearly showing resurrection to the active serine residue.

Since these molecules showed clear reactivation, a limited study was performed with various pH values in order to determine the effect on resurrection. Surprisingly, a very large pH dependence was noted for methylphosphonate-aged resurrection. The difference between pH 7 and 9 for resurrection was approximately 20%. C8 was analyzed

1 by UV-vis and H NMR at various pH values in order to determine the pKa of between 7–

8 and implicated various protonation and zwitterionic states of C8. Computational techniques were used to investigate the highest affinity for the enzyme active site, and the

MD results of the study suggest both zwitterionic forms of C8 have affinity for the active site over other protonation states. The kinetic profile of C8 was finally compared for its ability to reactive hAChE aged with DFP and similarly reached about 20% resurrection, confirming that these results are not just exclusive to eeAChE.

After almost 70 years of effort, the goal to recover activity from the aged form of

AChE has been achieved, and resurrection was demonstrated after aging by both organophosphate pesticides and authentic organophosphonate nerve agents. But further work remains to create a therapeutic drug with all of the desired activity, efficacy and selectivity.

100

N N

C8 Figure 47. Lead QMP structure for resurrection of aged AChE.160

1.16 Problems with Realkylation of Aged AChE.

Despite the recent success in resurrection by Hadad and coworkers, the reactivation or realkylation of aged AChE is challenging; a substrate has to bind selectively and efficiently in the active site, to bind in an active conformation so as to generate the critical reactive intermediate and/or to facilitate the critical transition state for the desired realkylation, and then to allow the realkylated phosphylated serine to be reactivated by a good nucleophile. And, all of this has to occur for an active site that is relatively compact and constricted.

A computational investigation conducted by Liu and coworkers evaluated the attempts to realkylate aged AChE with 2-methoxypyridniniums by Quinn and coworkers.163 Liu evaluated two different mechanisms of action as proposed by Quinn as the mechanism by which realkylation occurs, and in total, the reaction of nine different 2- methoxypyridiniums were investigated for their mechanism of action both in solution using density functional theory (DFT) and bound within the enzyme using QM/MM techniques. They determined that the mechanism of action is an SN2 displacement of the methyl moiety from the methoxy group. The calculated activation barriers in solution were in good agreement with experiment. The QM/MM methods confirmed that as in

101 solution, the preferred reaction mechanism in the enzyme is an SN2 process, but with substantial energetic barriers. The activation barriers for the reaction have a mean of 30.4 kcal/mol and a Boltzmann-weighted average of 26.6 kcal/mol, suggesting that the reaction will not proceed in vitro. The authors reported that the significant energy barrier is due to the strong π-π interaction between W86 and the pyridinium substrates. The authors proposed a few solutions, such as introducing a linker between the pyridinium and the electrophile, increasing the size and reactivity of the atom to S or Se, or increasing the size of the aromatic ring to use steric effects to push the compound away from W86, or increasing the electrophilicity of the methyl group via substitution. These results may suggest an additional reason why oximes have found limited use in reactivation of aged AChE since most are pyridinium species.

The aged state of AChE is a thermodynamically stable form of AChE, and resurrection presents some difficulties in overcoming the conformational changes and hydrogen-bonding networks within the active site of aged AChE. Quinn notes that the aged state of the enzyme makes four hydrogen bonds with the aged Ser residue, one between the phosphylated oxyanion and the histidinium residue of the catalytic triad and three additional hydrogen bonds with the oxyanion hole. Since the enzyme has evolved to turn over acetylcholine at near diffusion-controlled rates, the ability of the active site to stabilize the tetrahedral or negatively charged intermediate is high. Quinn and coworkers estimated the stabilization of the tetrahedral intermediate in the deacylation step to be at least 11 kcal/mol.164 Quinn and coworkers estimated that the strong H-bonding network drops the pKa of the phosphonic acid to –2 due to stabilization of negative charge that has

102

38 evolutionarily developed for catalytic turnover. This vast reduction in pKa would then have a significant effect on the anion’s ability to serve as a nucleophile, making the realkylation reaction to be more difficult than those conducted in bulk solution. A calculated estimate places the rate of reaction at 100 times slower than what would be observed with a model phosphonate in solution.164

Thus, a successful resurrector of aged AChE has to bind in order to disrupt this hydrogen-bonding network or to cause significant conformational changes within the active site, thereby reducing the strength of the hydrogen bonding to allow for reactivity of the phosphylated oxyanion, while also facilitating the desired transition state for electrophilic realkylation.

1.17 Perspective and Future Work.

Despite extensive quantities of work with regard to reactivation of OP-inhibited and aged AChE and BChE, a new therapeutic solution has yet to be reached and we continue to rely on old oximes and management of the cholinergic crisis as the standard of care. However, recently significant discoveries have been made that have the potential to change the landscape for the reactivation and resurrection of cholinesterases.

Newly synthesized oximes that are uncharged have shown activity for a number of OP compounds with AChE and BChE, potentially addressing the issue of being broad spectrum therapeutics for OP poisoning and the permanent positive charge of most treatments preventing BBB permeability.110 The linking strategy of oximes to peripheral site ligands has proven to be advantageous for increasing overall reactivity in a number of circumstances, yet the broad spectrum issue often continues to arise, in particular with

103 efficiency for reactivation of phosphoramidates, such as after tabun (or Novichok) exposure, as has been noted in a number of studies. The approach of using peripheral site ligands to increase binding affinity continues to serve its purpose but the issue may lie in the nucleophile of choice that is being attached. For nearly 70 years, the nucleophile has been chosen as an oxime, and this trend has continued with some limited variation.

Attention needs to be placed on oximes that are outside those traditionally studied. The inhibition and aging of OP compounds such as CBDP with hAChE and hBChE has two different pathways and in the case of hBChE, even two different aging steps.19 Although not traditionally considered as a target, the toxicity and challenge of treating aerotoxic syndrome is unique and requires further development. Moreover, entirely new and highly toxic classes of OP compounds, Novichoks, have emerged as a threat, yet to date, very little is known with regards to treatment, modeling, and even structure as much of the information is still classified. This presents a unique problem in that only those in the military or defense sector may have access to the appropriate information required to try and develop effective therapeutics for this class of OPs.

The recent discovery of novel non-oxime nucleophiles presents entirely new chemical frameworks to create improvement in standard of care. Two promising families of compounds have been most notable, including those with a pendant base attached to an aryl anchor and then Mannich bases.91,113,114,116,160 While all of these compounds were discovered independently, the structural features continue to be similar. What is conserved between both types of compounds is the basic nitrogen attached to an aromatic core; for the Mannich bases, ortho substitution relative to a phenol is the most effective.

104

Continued efforts are needed to better understand the mechanism of action for these particular types of molecules to aid in future drug design, whether those efforts are computational and crystallographic studies or structure-activity relationships. Another major benefit of both classes of compounds is that they exist as non-permanently charged compounds and this should allow for BBB penetration; however, this needs to be confirmed as these are only predicted or hypothesized.

The other major advantage of Mannich bases is that they address the additional issue of aging, having been shown to recover small quantities of activity from aged hAChE.160 This may be slightly expected as computational investigations of the aged enzyme have noted that the H-bonding network needs to be altered in order for these compounds to function efficiently.145,164 Of the computationally proposed molecules,

Mannich bases are simply aromatic β-aminoalcohols which were reported to disrupt this

H-bonding network.145 The very specific orientation of H-bond donors and acceptors may then explain the high affinity for the active site of AChE and the rapid rate of reactivation in addition to the ability to resurrect the anionic aged Ser residue.

Given the computational, crystallographic, and synthetic techniques developed during the course of these reported investigations, many of these answers can be determined and continued efforts toward the treatments and developments of OP exposed

AChE can continue. All of the work that goes into the development of these non-oxime reactivators though can then be mirrored in the reactivation and development of BChE as a bioscavenger. Continued efforts support the potential of BChE to be an effective endogenous or therapeutically applied scavenger, but we are lacking means to express

105 larger quantities of the enzyme and methods to make BChE to be catalytic or pseudo- catalytic. The most ideal situation would be to use endogenous BChE and to make this efficient scavenger to be pseudo-catalytic by administration of a therapeutic but this solution may take longer than using doses of BChE combined with a reactivator in the more immediate future. Nonetheless, the field seems to be at a turning point as the exhaustive search for oxime-based therapeutics has proven unfruitful, but newer scaffolds have demonstrated great potential for addressing the continued problems of BBB permeability, reactivation after inhibition and also resurrection of the aged forms.

106

Chapter 2. COMPUTATIONAL INSIGHTS INTO THE ALKYLATION REACTIONS OF PYRIDINE AND PYRIDINIUM QUINONE METHIDE PRECURSORS: AN EFFORT TO REVERSE THE EFFECTS OF ACETYLCHOLINESTERASE FOLLOWING AGING

2.1 Introduction.

A continuing threat in this modern era to both civilians and military personnel alike is that of organophosphorous nerve agents and pesticides. Organophosphorous nerve agents in particular, are some of the most toxic agents ever synthesized by humans for the purpose or intention of killing or harming other human beings.165

The risk associated with nerve agents is their use on human populations for nefarious purposes. This continues to be a concern as stockpiles of these agents are still known. During WWII Germany stockpiled nerve agents with the intent to weaponize them, fortunately never doing so during the WWII conflict. Again, during the Cold War, both the U.S. and Soviet Union stockpiled nerve agents, even though most countries had signed the 1925 Geneva Convention banning the use of chemical weapons. Even so, members of the of the Studies and Observation Group, a U.S. group formed during the

Vietnam War, claim they were ordered to use sarin on a Laotian village in an undercover operation.166 Several Middle-Eastern nations are believed to have stockpiles of these weapons, most notably Libya and Iraq. It was confirmed in 1984 that Iraq had in fact used nerve agents, GA and GB, against Iran in the Iraq-Iran War.166–169 The first documented use of nerve agents in a terrorist attack occurred in 1994 when nearly 600 107 people in the Japanese city of Matsumoto became ill, with 7 individuals losing their lives.

The cause of these deaths and illness arose from the contamination of a pond with sarin.166 The following year, 1995, a second terrorist attack was aimed at the Tokyo

Subway system affecting 5,000 and killing 12. Those responsible for the attack were the religious cult Aum Shinrikyo, whom used a crude form of sarin to execute the attack. A container of the nerve agent was placed underneath a seat and punctured allowing for the evaporation of the nerve agent and subsequent inhalation by the passengers. Although many were affected by the attack, the method of dispersion was crude; had an aerosol of the nerve agent been used, then many more deaths may have likely been observed.166,167,170 The risk of other terrorist attacks is still possible as recent news stories have been concerned with the possible possession of VX by terrorist cells and rogue states.170

The lethality of nerve agents arises from how the compounds inhibit the function of acetylcholinesterase (AChE). AChE is an enzyme that functions by hydrolyzing the neurotransmitter acetylcholine (ACh) at neurosynaptic junctions. 171,172 The hydrolysis of

ACh allows for transmission of action potentials across neuro-neuro and neuromuscular synapses within the central and peripheral nervous systems.165 The reaction between

AChE and ACh is remarkably well catalyzed with rates of reaction approaching that of a purely diffusion-controlled reaction.165,171,172 The ACh must first make its way down a 20

Å gorge in the enzyme toward two active sites. Within the active sites is a catalytic triad consisting of Ser203, His447, and Glu334, numbering from human AChE. His447 deprotonates the Ser203 oxygen atom allowing for Ser203 to participate in nucleophillic

108 attack of the carbonyl carbon in ACh. The hydrolysis of the ACh results in choline and acetic acid.165,166,171–175 Nerve agents function by inhibiting this reaction resulting in the accumulation of ACh at synaptic junctions, leading to a variety of life threatening symptoms.167,168,175 In the inhibition process, the Ser203 attacks an organophosphorus nerve agent at its phosphorous center, generating a complex that is no longer able to be hydrolyzed by water, as is the normal function of the enzyme.175

Figure 48. The entire acetylcholinesterase enzyme with a quinone methide precursor bound within the active site.

The inhibition of AChE causes severe problems for the human body. The latency, or time interval between observation of symptoms and exposure, varies based on the

109 nerve agent and route of exposure. For exposure to vaporous or aerosol forms the latency is 30-120 seconds and for percutaneous exposure, the time interval increases to 1 minute or up to 18 hours.166,167 The recognition of nerve agent exposure is a difficult process for medical professionals because laboratory testing takes much too long for the diagnosis of exposure. Therefore, patients exposed to nerve agents must be treated based on symptoms rather than laboratory tests, although GC-MS is an effective tool for the identification of nerve agents in blood yet in a forensic rather than medical setting.166,176 The diagnosis of nerve agent exposure is determined from the SLUDGE syndrome, or observation of excessive excretions. Some of these symptoms include salivation, lacrimation (tear production), urination, defecation, diaphoresis (increased sweating), and gastric emesis

(vomiting).166,167,176 Although unpleasant, these symptoms are not those that lead to the lethality of nerve agents. The lethal symptoms are known as the killer “B’s”: bronchorhea

(production of a large quantity of mucus), bronchoconstriction, and bradycardia (low resting heart rate). Additional consequences include fasciculations (muscle contraction or twitching), weakness, and flaccid paralysis, with paralysis of the diaphragm having deadly consequences.166,167,176 These symptoms of nerve agent exposure are observed shortly after exposure, yet complications continue to arise after survival of such an exposure. Such further complications were reported by military recruits whom had been exposed to a sarin-like compound and include: increasing depression, “jittery”-ness, increased anxiety, intellectual impairment, and unusual dreams.166,176

There currently are treatment methods available for nerve agent exposure, once detected. The methods of treatment depend on whether exposure is a risk or has occurred.

110

To prevent the inhibition of AChE when there is a risk of exposure, military forces have been pretreated with pyridostigmine bromide, a carbamate compound that inhibits AChE, rather than the nerve agents that soldiers may come in contact with.166,167 In this case however, the carbamate inhibitor binds reversibly, thereby blocking nerve agent binding without completely inhibiting the function of AChE. A second treatment is currently the topic of a great deal of research, that being the design of bioscavengers that would be able to catalytically hydrolyze nerve agents. One such bioscavenger receiving a great deal of interest is the human enzyme paraoxonase 1 (huPON1).177,178 These bioscavengers offer an advance in the treatment of nerve agent exposure however, they have yet to be integrated into the medical system. The most common form of treatment occurs after exposure and diagnosis of nerve agent exposure via treatment with oximes. Oximes such as obidoxime and pralidoxime (2-PAM) are commercially available for nerve agent treatment yet their efficiency against the treatment of select nerve agents is questionable, nerve agents such as soman and cyclosarin. Other oximes, namely the bis-pyridinium

Hagedorn oximes (H oximes) such as HI-6 and Hlö-7, are more effective at treating soman, sarin, cyclosarin, and VX. 2-PAM, however, is the only oxime approved by the

FDA and has seen a long record of safe use despite being less effective than H oximes.166–169,172,174,179 These oxime treatments are distributed to at risk populations, such as in Israel during the first Persian Gulf War, in the form of auto injectors. An additional treatment of atropine is distributed in auto injector form as well to treat the SLUDGE symptoms previously discussed, while the oximes reactivate the AChE enzyme and restore it to its normal function. U.S. military personnel are given MARK I kits

111 containing both injectors and multiple doses. The key to treatment with atropine is re- administering the drug every 5-10 minutes until secretions dry up.166,167

There are known methods of reactivation of AChE; however, these processes are complicated by the “aging” of an inhibited Ser203-nerve agent complex. The aging process is the spontaneous dealkylation of the organophosphorous nerve agent-Ser complex.166,169,172,174,176,180 This dealkylation process leaves the inhibited complex immune to reactivation by commercial oximes, not allowing for any reactivation of the enzyme unless more AChE is synthesized by the body. This process is particularly problematic due to the time periods by which inhibited complexes age. The aging half- life for soman is 2 minutes, sarin 5 hours, tabun 13 hours, and VX 40 hours.166,176,180 This makes the treatment of soman very difficult by conventional oxime methods. Patients may be treated with oximes that will function properly for only minutes while those complexes that have aged are untreatable, thus any symptoms such as paralysis that are observed may be permanent.

The focus of this chapter is to investigate the reactivation of an aged phosphonate model representative of the aged-complex formed within AChE. Computational methods were chosen to significantly reduce the risk of working with nerve agents and because studies of human AChE cannot be undertaken without ethical dilemmas surrounding human testing.168,170 Additionally, experimental testing conducted on animals is difficult to extrapolate to humans due to differences in the toxicokinetics and pharmacokinetics of different species.168 Computational experiments similar to those reported here have been found to be consistent with experimental findings regarding docking within the AChE

112 enzyme and the energy barriers determined for reactions of nucleophiles and organophosphorous nerve agents.173,174 Computational studies allow for both the prediction and confirmation of experimental results of enzymatic reaction mechanisms, as is such with the current study.177

O P O O

Figure 49. A model phosphonate (ball and stick, left, line structure, right) used to represent the aged phosphonic-Ser residue of the AChE active site following the dealkylation reaction.

The class of molecules being investigated are known as quinone methides (QMs).

QMs are methylene cyclohexadiones that can exist in multiple isometric forms. A precursor of a QM (QMP) can be administered as a prodrug allowing for the QM to form before realkylating the aged inhibited complex at the bottom of the AChE gorge (Figure

50). These structures are very promising since QMPs resemble the known AChE inhibitor edrophonium (Figure 51). Thus, it seems likely these QMPs may enter the catalytic site. Based on previously conducted studies, the QMPs were found to be highly tunable via variation of substituents and positioning. The progression of this study has led to the investigation of pyridine and pyridinium QMPs as a potential therapeutic for the realkylation of aged AChE. To determine how effective these compounds are, isomers of

113 pyridine and pyridinium molecules will be compared based on the predicted thermodynamic of the realkylation reaction, potential being SN2 or SN1, as well as with molecular docking studies to affinity for aged-AChE. With such large numbers of structural derivatives of QMPs resulting from the variability of regioisomers, leaving groups, substituents, and protonation or methylation states, it is unreasonable to synthesize every compound. Thus, molecular docking studies can be performed as a means to determine which frameworks or structural modifications should be further pursued based on the hypothesis that those realkylators that bind the best with the aged-

AChE active site, having the greatest potential for realkylation. This hypothesis is based on the idea that due to proximity to the phosphylated oxygen, the entropy of the reaction is reduced, thereby facilitating a lowered free energy of reaction. This procedure of down-selection of compounds has been demonstrated in the literature with AChE.181–183

The major variation between docking studies is how to interpret the docking data. At present, conclusions will be based on the fraction of poses within the defined active site relative to all of those analyzed. Later refinement of such criteria can be based more specifically on binding energy, amino acid interactions, or predetermined scoring functions as demonstrated by other groups.146,181–187

114

N O O P O O O OH N N P R ⎼R-H AChE O OH O +R-H AChE QMP QM

Figure 50. The proposed formation of a QM from a QMP that would then result in the realkylation of the phosphonic anion of the aged-Ser residue.

The first library of study was inspired by the oxime reactivator 2-PAM, with a particular focus on the regioisomeric effects that the phenol has on binding, as well as the effects observed for methylation and protonation of the leaving group, dimethylamine

(Figure 52).

O OH O

o-QM p-QM edrophonium

Figure 51. Examples of quinone methides (ortho, left, and para, right) and the known AChE inhibitor edrophonium.

2.2 Experimental.

QMP permutations (18) were analyzed based on the preferred reaction mechanism and docking patterns. Each QMP was reacted with a model phosphonate, representative of an aged inhibited Ser residue (Figure 49).

115

HO N N N LG

2-PAM Library 1 Scaffold N N HO HO N N OH H N H N N N H NH N NH NH OH HO OH PAM1 PAM4 PAM7 PAM10 PAM13 PAM16

H N H N H H H H HO HO N N OH H N H N N N H NH N NH NH OH HO OH

PAM2 PAM5 PAM8 PAM11 PAM14 PAM17

N N HO HO N N OH H N H N N N H NH N NH NH OH HO OH PAM3 PAM6 PAM9 PAM12 PAM15 PAM18

Figure 52. Library of 2-PAM derivatives evaluated based on molecular docking and thermodynamics of reaction.

2.3 Mechanistic Study

All geometry optimizations, ModRedundant scans, transition state optimizations, frequency calculations and internal reaction coordinates (IRCs) were calculated using

B3LYP/6-31G*.188–190 The contribution of the solvent was included by the polarized continuum model (PCM) for water.191 The inclusion of solvent then is an electrostatic permutation on the wavefunction rather than explicit water molecules. All calculations were performed using Gaussian09 and constructed using GaussView 4.1.2.192,193 The geometry of each structure was first optimized. Conformers for the QMPs and model 116 phosphonate were determined by multiple geometry optimizations and manual displacement of the dihedral angles rather than a full conformational search. Relaxed potential energy scans, ModRedundant scans, were used to determine the likely transition state structures for each reaction mechanism by scanning the bond distances between of each step. For the SN1 mechanism the appropriate C-N or C-O bond was broken over a series of 25 steps of 0.1 Å per step. The SN2 mechanism was found to require the approach of the amine group displacing the model phosphonate to function properly.

These calculations were conducted using 15 steps of 0.1 Å each. The geometry associated with the highest energy of each scan was then used as a starting point to determine the appropriate transition state structure. Confirmation of the transition state geometry was determined using a vibrational frequency calculation to ensure a single imaginary frequency, indicative of a saddle point in the potential energy surface. If a structure had a single imaginary frequency, an IRC was then used to trace the reaction, ensuring it was on the correct reaction path. The resultant energies from each structure were then used to calculate the overall enthalpy of each reaction. Difficulties arose with the ModRedundant scans regarding the SN1 reaction scans, these are discussed below.

2.4 Docking Study.

The computational protocol began with conformational searches of each QMP.

The conformational search was conducted manually by rotating around individual bonds in 60° increments. Each of the resulting conformer geometries was optimized at the

B3LYP/6-31G* level of theory and basis set for all PAM compounds.194–197 Additional frequency calculations were performed to confirm that all optimized structures were

117 minimums along the potential energy surface. The atomic partial charges for all structures were calculated for the optimized geometries using the Merz-Singh-Kollman methodology at the B3LYP/6-311+G** level of theory.198 All calculations were performed using Gaussian 09.199

The AChE receptor was previously prepared for docking by Dr. Jeremy Beck.83

The human AChE structure was constructed using the available 1B41 crystal structure.

The original Protein Data Bank (PDB) structure, 1B41, had fasiculin bound in the active site of the enzyme. The fasiculin molecule was removed and an aged organophosphorous compound generated by overlay of an aged non-human AChE enzyme. Thus, the cavity caused by the removal of fasiculin was rebuilt using a non-human structure with an aged complex within. A model QMP was then docked within the newly generated structure and molecular dynamics (MD) simulations run, generating the frames previously mentioned.

Docking studies were conducted using Autodock 4 using the Lamarckian genetic algorithm with a sampling of 200 simulations per realkylator per receptor frame (13), resulting in 2,600 simulations for each realkylator.200 The bond distance between the benzylic carbon and the phosphonic oxygen is measured for all simulations. Zoning criteria has been used to determine what constitutes being within the active site of AChE.

All docked poses with a bond distance less than 5.0 Å were considered to be within the active site of the enzyme based on numerous measurements to surrounding amino acid residues. Additionally, all poses were sorted into either a “reactive” category, in which the direct approach of the phosphonic oxygen was unhindered by the amine leaving

118 group, and a “bound” category in which only bond rotation will result in reaction. The

“bound” category also includes all those determined to be “reactive”.

Figure 53. Example docking histogram showing the population of multiple binding modes with a potential maximum of 200. The number of poses satisfying the distance and hinderance criteria were then used to determine a percentage of 'binding.'

Although docking studies are conducted at a lower level of theory than the quantum mechanical mechanistic studies, the results are still supported in experimental work. A study conducted by Sanan et al. found that the docking of nerve agents within huPON1 were consistent with known modulation of substrate specificity, meaning docking protocols sample realistic binding modes.177

119

Figure 54. A pose satisfying the distance criteria and showing no steric hinderance between reactive centers (A) and a pose only satisfying the distance criteria (B).

2.5 Results and Discussion

The geometry optimizations were originally conducted with the phenol in its neutral protonation state; however, difficulties with the ModRedundant scans resulted in having to scan the phenol as being deprotonated. The reported values here are of those that have the phenol deprotonated, these structures were used for the remainder of the study with the exception of docking calculations in which the phenol remained in its neutral state.

120

Figure 55. A representative measurement of the dihedral angles reported with PAM9

Table 2. Protonated and deprotonated energies and dihedrals of the PAM library.

Protonated Energy Deprotonated Energy Dihedral Angle Name (Hartrees) (Hartrees) (º) PAM1 -497.265 -496.797 -123.63 PAM2 -497.714 -497.267 -101.71 PAM3 -537.024 -536.576 -102.44 PAM4 -497.266 -496.805 -101.24 PAM5 -497.724 -497.282 -100.73 PAM6 -537.033 -536.591 -99.89 PAM7 -497.27 -496.808 -100.97 PAM8 -497.721 -497.289 -99.26 PAM9 -537.031 -536.594 80.62 PAM10 -497.274 -496.798 -100.43 PAM11 -497.725 -497.284 -107.17 PAM12 -537.034 -536.593 -109.62 PAM13 -497.267 -496.791 -162.74 PAM14 -497.716 -497.259 -97.48 PAM15 -537.022 -536.561 -89.92 PAM16 -497.265 -496.787 -81.05 PAM17 -497.716 -497.256 -69.61 PAM18 -537.021 -536.564 -94.31

121

The dihedral angles of all the optimized geometries are consistent across most of the 18 QMPs observed, with notable exceptions of 80.62 and -162.74º. The 80.62º angle for the PAM9 molecule is not entirely surprising when considering the steric hindrance caused by the alkyl chain, methyl substituent, and oxygen all being on adjacent carbons

(Figure 52). A similar observation is made for PAM13-16 compounds where the amino methyl rotation is more restricted in the protonated and methylated pyridinium forms.

The more restricted rotation of PAM14 and PAM15 then results in similar dihedral angles while the unrestricted PAM13 can rotate more freely to a lower energy conformation.

ModRedundant scans are relaxed potential energy surface scans that allow for tracking the energy changes of a molecule as the result of a geometric perturbation (angle or distance change), which includes a re-optimization of the geometry of the resulting structure. The scans shown herein then can be viewed reaction energy profile representation. The first attempts at conducting ModRedundant scans, with the phenol protonated, resulted in a proton transfer between either the amine leaving group or model phosphonate rather than the desired SN1 or SN2 reaction. Thus the structures were reconstructed and re-optimized in their phenolate forms. The justification for this is that general base catalysis can deprotonate the QMP upon binding within the active site. Since the reactions are being modeled in a bulk solvent in these calculations we had to change our computational approach. The intention of these relaxed potential energy scans was to find an energy maximum during either bond breaking or forming indicative of a transition state. The calculations in for SN2 reaction mechanism were successfully able to

122 do so. During the SN2 scans, the amine group was moved in steps toward the reactive carbon center while displacing the model phosphonate, effectively modeling the reverse reaction but by the principle of microscopic reversibility allowing for determination of the transition state. However, complications arose during the SN1 scans. For 12 of the 18 compounds of study, a well-defined energy peak was unobservable by the SN1 mechanism (Figure 56). Instead, the potential energy surface appeared asymptotic. The compounds that had clear energy peaks for the SN1 ModRedundant scans were all pyridine, not pyridinium, compounds (Figure 57). This result is to be expected as the resonance stabilized carbocation, QM, would be much higher in energy with a permanent positive charge. Thus, inclusion of an additional positive charge then prevents the formation of a QM intermediate, which is why it cannot be identified on the potential energy surface.

Figure 56. An asymptotic energy profile for PAM1.

123

Figure 57. An energy profile scan for PAM2 with a more well defined stationary point. Table 3. Energies and bond distances resulting from the energy profile scans with defined stationary points for an SN1 mechanism.

Peak 1 Peak 2 Peak 1 C-N Peak 1 C-O Name Energy Energy Length (Å) Length (Å) (Hartrees) (Hartrees) PAM1 -496.744 2.62 -1047.967 2.13 PAM2 - - - - PAM3 - - - - PAM4 - - -1047.978 2.23 PAM5 - - - - PAM6 - - - - PAM7 -496.759 2.72 -1047.983 2.13 PAM8 - - - - PAM9 - - - - PAM10 -496.763 2.53 -1047.984 2.13 PAM11 - - - - PAM12 - - - - PAM13 -496.752 2.52 -1047.973 2.13 PAM14 - - - - PAM15 - - - - PAM16 -496.75 2.60 -1047.972 2.03 PAM17 - - - - PAM18 - - - -

124

The ModRedundant scans were all capable of finding an energy maxima along the potential energy surface when represented by a SN2 mechanism. These results then suggest that the SN2 mechanism is the only mechanism by which pyridinium compounds can realkylate a phosphonate anion.

Table 4. The energy and bond distances associated with an energy profile scan assuming an SN2 mechanism.

Peak C-N Length C-O Length Name Energy (Å) (Å) (Hartrees) PAM1 -1183.134 2.30 1.61 PAM2 -1183.603 2.30 2.80 PAM3 -1222.919 2.30 1.58 PAM4 -1183.141 2.30 1.64 PAM5 -1183.625 2.30 1.59 PAM6 -1222.933 2.30 1.59 PAM7 -1183.143 2.30 1.66 PAM8 -1183.634 2.30 1.58 PAM9 -1222.944 2.30 1.59 PAM10 -1183.145 2.40 1.60 PAM11 -1183.631 2.30 1.60 PAM12 -1222.940 2.40 1.54 PAM13 -1183.131 2.30 1.65 PAM14 -1183.600 2.20 1.68 PAM15 -1222.912 2.30 1.57 PAM16 -1183.136 2.30 1.61 PAM17 -1183.608 2.30 1.57 PAM18 -1222.918 2.30 1.57

The results found for the SN2 ModRedundant scans show a very consistent pattern among all of the compounds analyzed. The C-N bond length appears between 2.2-2.4 Å

125 and the C-O bond generally 1.55-1.65 Å. The clear exception, PAM2, has a rather long

C-O bond length. Despite this irregularity, all of the calculated structures were used to determine the actual transition states using an Opt=TS calculation. The observed energy pattern for transition state stability is methylated pyridiniums > protonated pyridiniums > pyridines.

Transition states were determined for compounds with appropriate starting geometries from the ModRedundant scans. All transition states determined were confirmed by conducting a subsequent frequency calculation. These additional frequency calculations helped to ensure a true transition state. An additional test of the accuracy of the transition state was conducted in which an IRC was used to trace the reaction in both directions of the transition state. If the correct products and reactants were generated by the IRC the accuracy of the transition state was further confirmed.

The thermal energy of each reaction was calculated relative to the starting reactant energies for each compound and reaction mechanism. The energy plotted along the reaction coordinate is the sum of the optimized energies. For instance, the intermediate energies for the SN1 mechanism are often greater than the transition states, this is because the intermediate energy is the sum of the optimized energies: QM, phosphonate, and amine.

126

Neutral Pyridine 40 PAM1 PAM4 30 PAM7

20 PAM10 PAM13

10 PAM16 Energy (kcal/mol)

Int TS1 TS2 Product

Starting Material

Figure 58. Energy diagram for an SN1 mechanism of neutral pyridine PAM scaffolds.

With the pyridine compounds the reaction with the lowest energy barriers are those with the alkyl chain para to the oxygen. Both the PAM10 and PAM16 compounds have very similar energetics, with a near complete overlap. Of the compounds with the alkyl chain ortho to the oxygen the lowest energy reaction path is the PAM13.

127

Protonated Pyridine 60

PAM2 40 PAM5 PAM8 PAM11 20 PAM14 PAM17 Energy (kcal/mol)

Int TS1 TS2 Product

Starting Material Figure 59. Energy diagram for an SN1 mechanism of protonated pyridine PAM scaffolds.

With a protonated ring-nitrogen, the energetics of the reaction seem to change considerably. It is again true that the PAM14 compound has the lowest energy barrier for the ortho compounds. When evaluating pyridinium compounds the ortho relation between the phenol and alkyamine is preferred over the para. PAM11 and PAM17 have a considerable separation in comparison to the near overlap in Figure 59. All six compounds in this case have higher reaction barriers than those in Figure 58. This further confirms that the mechanism of reaction of the pyridiniums is an SN2 rather than SN1.

128

Methylated Pyridine 50

40 PAM3 30 PAM6 20 PAM9 PAM12

Energy (kcal/mol) 10 PAM15 PAM18 0

Int TS1 TS2 Product

Starting Material Figure 60. Energy diagram for an SN1 mechanism of methylated pyridine PAM scaffolds.

The methylated pyridinium compounds follow the same exact ordering as that observed for protonated pyridiniums in Figure 60. However, when a methylated pyridiniumm the reaction is slightly more favorable in comparison to a protonated pyridinium. Both the protonated and methylated pyridinium reactions are considerably higher in energy compared to the pyridine compounds, ~50 kcal/mol compared to ~35 kcal/mol respectively. This suggests that the pyridine compounds are considerably more likely to react via an SN1 mechanism.

129

Neutral Pyridine 40 PAM1 PAM7 30 PAM10

10 Energy (kcal/mol)

Product

Starting Material Figure 61. Energy diagram for an SN2 mechanism of neutral pyridine PAM scaffolds.

It has previously been mentioned that it was difficult to find transition states for the pyridinium compounds via an SN1 mechanism, the opposite holds true for the SN2 mechanism. Although explicit peaks were determined from the ModRedundant scans, 3 of the 6 compounds could not be reduced to a single imaginary frequency. This suggests that the preferential reaction mechanism for the pyridinine-based compounds is again an

SN1 mechanism. This conclusion is further supported by the fact the SN2 reaction barriers are higher than the SN1 barriers, with the exception of PAM7, likely due to the difference in electronics cause by the potential for PAM7 to be in its pyridone form.

130

Protonated Pyridinium 50

40 PAM2 PAM5 30 PAM8 PAM11 20 PAM14 PAM17

Energy (kcal/mol) 10

Product

Starting Material Figure 62. Energy diagram for an SN2 mechanism of protonated pyridine PAM scaffolds.

When the aryl core is a protonated pyridinium, the SN2 reaction mechanism is preferential to the SN1 mechanism, with the exception of PAM14. All of the compounds, except, PAM14, react with a very similar energy barriers making the ordering less conclusive, with five of the transition state energies being within ~3 kcal/mol of each other.

131

Methylated Pyridinium 40

30 PAM3 PAM6 20 PAM9 PAM12 10 PAM15 Energy (kcal/mol) PAM18

Product

Starting Material Figure 63. Energy diagram for an SN2 mechanism of methylated pyridine PAM scaffolds.

As observed for Figures 59 and 60, the order for the pyridinium SN2 reaction barriers follow a very similar partner. The reaction energy barriers for the SN2 mechanism of the pyridinium compounds are similar, with the methylated compounds being slightly lower in all cases.

Overall, the lowest energy barriers for the SN1 reaction mechanism are observed for compounds pyridine-based compounds. The lowest energy barriers observed for the

SN2 mechanism are for the methylated pyridinium compounds, as is to be expected based on the electronic effects the protonation state has on the QM intermediate. It can conclusively be stated that the pyridine comppiunds proceed via SN1 and the pyridnium compounds an SN2 reaction.

132

Each of the 18 compounds were docked within a modified AChE enzyme and scored according to the scoring system outlined above.

Table 5. Docking percentages for the PAM library.

Structure Reactive % Bound % PAM1 7 19 PAM2 4 32 PAM3 30 38 PAM4 10 20 PAM5 11 31 PAM6 22 38 PAM7 7 22 PAM8 11 32 PAM9 14 27 PAM10 11 22 PAM11 11 33 PAM12 19 32 PAM13 24 41 PAM14 13 24 PAM15 39 46 PAM16 19 41 PAM17 14 47 PAM18 25 53

The results of the docking studies for the PAM library provide structural patterns that lend to better QMP bidning affinity (Table 5). For all studied structures the pyridinium compounds bind better than the pyridine derivatives when analyzing the bound category, with the exception of PAM14. Methylated pyridiniums tend to bind better with the exception of PAM9, and PAM12 binding approximately the same. The

PAM18 compound, in which the hydroxyl group is at the 5-position and the alkyl amine

133 at the 2-position, binds the best of the studied library. The second best binder is a methylated pyridinium, PAM15, having a similar meta relation between the pyridyl-N and the hydroxyl group, except with the hydroxyl at the 3- position and the alkyl amine at the 2- position (Figure 52). This relation suggests there are cooperative active site binding interactions interactions caused by a pyridinium and a meta-hydroxyl group.

Figure 64. Most commonly observed docking pose of the studied PAM library. The compound pictured is PAM3.

The first and most preferential pose adopted by binding ligands is shown in

Figure 24. In this pose the amine functional group is situated down in a pocket at the bottom of the AChE gorge. This pose allows for a completely unhindered reaction between the benzylic carbon. The methyl group on the ring nitrogen extends back into the opposite side of the gorge. It appears as though there may be hydrogen-bonding interactions between the phenol group and the two oxygen residues. The pose seems particularly promising if these observed hydrogen-bonds position the molecule for reaction and act as a basic site to deprotonate the compound and form the QM. Although 134 deprotonation may occur in this pose the displacement of the amine is hindered. The reaction of this compound, was previously concluded to be SN2 direct displacement of the amine occurs although, this displacement may be hindered as the amine already being forced into the bottom of the gorge.

Figure 65. Second most commonly observed docking pose of the studied PAM library. The compound pictured is PAM3.

This secondary pose is a near complete reflection of the one observed in Figure

64. Here, the methyl group is still situated back of the active site, opposite the aged-Ser residue. The amine group faces up into the actual gorge, which may facilitate leaving.

The above mechanistic studies concluded the reaction of PAM3 is SN2, thus as the benzylic carbon is attacked by the aged-Ser nucleophile the amine will be displaced into the gorge opening and out of the enzyme. The secondary binding pose in Figure 65 is higher in energy as the hydrogen-bond stabilizations are not present with the phenolic

135 oxygen. With the phenol situated into the gorge, a water molecule may be able to deprotonate the phenol.

Figure 66. And additional commonly observed docking pose of the studied PAM library. The compound pictured is PAM9.

The last pose observed situates the ring in the bottom of the gorge with the amine placed into the gorge opening. For reasons outlined above having the amine in the gorge opening is preferential for reaction. This binding pose has the lowest energy of the three shown, likely caused by the hydrogen-bonding between the aged-Ser residue. In this pose it may be difficult to deprotonate the phenol as it is placed against the wall of the enzyme.

Additionally, the angle at which the reaction would take place seems unusual. Although an uncommon angle the displacement of the amine may deprotonate the phenol simultaneously, as observed in ModRedundant scans before deprotonation of the phenol in the thermodynamic calculations.

136

Overall the docking studies provide evidence that the compounds of study will bind in reasonable poses within the AChE binding pocket. Additional docking studies were conducted with a native AChE enzyme and the compounds never docked within the gorge. These results suggest that the docking of the compounds is selective for the aged enzyme.

Combining both results, pyridinium QMPs with a meta relation between the pyridine-N and the hydroxyl group and a sterically small charged amine leaving group have the best potential as a realkylator as demonstrated by the docking studies.

2.6 Conclusions

After the examination of the reaction mechanisms and docking results for 18

QMPs it has been determined the reaction proceeds via an SN1 mechanism for pyridine

QMPs and via SN2 mechanism for pyridnium QMPs. Protonated pyridnium QMPs have higher energy barriers whether modeled as SN1 or SN2 reactions. The energetics of reaction via SN1 of pyridine QMPs and via SN2 of pyridnium QMPs are comparable; however, docking studies suggest that pyridiniums dock much better within the AChE binding site. Given he inability of permanently positively charged compounds to penetrate the BBB, and the easier synthetic approach to making pyridine QMPs the results suggest that the primary synthetic effort should be target on the 2,3 (PAM13) and

2,5 (PAM16) frameworks.

2.7 Future Work and Adjustments.

The results and methodology outlined above were part of initial computational efforts for the design and optimization of QMP therapeutics. The results are interesting

137 but they are complicated by the specific set of protonation states evaluated. It was determined after this study that the protonation states of the QMPs is extremely important for effective resurrection of aged AChE, see Chapters 3-5. The computational techniques currently being used focus primarily on molecular dynamics (MD) studies rather than docking studies. MD studies allow for a better sampling of the protein dynamics in addition to the outputs of the reactions being easier to analyze using computational tools and programs.

This initial work did show that computational development and optimization is accurate and effective. As will be presented in Chapter 3-5, the 2,5 related pyridine

QMPs are in fact effective resurrectors of aged AChE. Further refinement of these and additional studies need to consider the protonation states of the QMPs and AChE, various enzyme forms such as inhibited and aged with various OP pesticides and nerve agents, and more widely varied QMP frameworks, such as being substituted or linked. Each of these options provides an opportunity to quickly direct synthetic efforts. All of these efforts are currently on going by other members of the Hadad group.

The mechanistic studies resulted in interesting, but relatively predictable results.

The difficulty with the mechanistic study conducted herein is that the reaction was modeled in bulk solution, even using and implicit rather than explicit water model. The difficulties noted in the reaction energy scans are to be expected. The more accurate way to determine the reaction mechanism of the QMPs is to use quantum mechanical (QM) molecular mechanical (MM) methods, QM/MM. This methodology allows for MM treatment of a large section of the AChE enzyme and QM treatment of the QMP and

138 select amino acid residues. This methodology is computationally expensive and would require exhaustive analysis of hundreds of compounds. Even so, efforts are being made to develop these protocols by current group members.

139

Chapter 3. DEVELOPMENT AND TESTING OF QUINONE METHIDE PRECURSORS FOR THE RESURRECTION OF AGED ACETYLCHOLINESTERASE USING A 3-HYDROXYPYRIDINE FRAMEWORK

The research presented in this chapter is adapted from the manuscript that is published as J. Med. Chem. 2018, 61, 16, 7034-7042. The biological screening and proteomics in this chapter was conducted by Dr. Qinggeng Zhuang. The computational work was completed by Dr. William Coldren. The synthesis was a collaborative effort between a number of undergraduate students with guidance and participation of Dr. Christopher Callam, Dr. Thomas Corrigan, and myself.

The OP compounds, both purchased pesticides as well as synthesized nerve agent analogues, are extremely hazardous and should be handled with the utmost care. The byproducts of reactions with the enzyme, fluoride, and oximes additionally produce compounds of unknown toxicity that should be regarded also as being extremely hazardous. When handling these reagents and solutions, appropriate personal protective equipment (PPE), including a lab coat, gloves, safety glasses and a well-ventilated fume hood, should be utilized at all times. When isolating or purifying synthesized analogues, additional butyl gloves and a chemical faceshield should be warn to prevent exposure to any potentially exposed skin. All solutions, glassware, caps, spatulas, or any other laboratory equipment that comes into contact with the OP compounds or solutions should be decontaminated with concentrated sodium hydroxide solution, preferably soaking for at least 24 hours. The washing of these contaminates should be collected and clearly labeled before being disposed of as hazardous material through the appropriate environmental health and safety guidelines.

3.1 Introduction

Acetylcholinesterase (AChE) is a serine hydrolase found in brain synapses, neuromuscular junctions and erythrocytes. AChE selectively hydrolyzes the neurotransmitter acetylcholine with its Glu-His-Ser catalytic triad. Organophosphorus

(OP) compounds phosphylate the catalytic serine of AChE, and inhibition of AChE results in the accumulation of acetylcholine. OP exposure may lead to death due to

140 seizures or respiratory failure caused by persistent nerve impulses.201–203 Thus, OPs are toxic and have been used as pesticides and chemical warfare agents. Every year OP-based pesticides kill approximately 200,000 people worldwide, especially in rural areas of developing countries.201 OP-based chemical warfare agents have been used in armed conflicts such as the Iran-Iraq war204 and the recent Syrian civil war,205 as well as terrorist attacks such as the Tokyo subway sarin attack.206 OP-inhibited AChE can be reactivated by oximes (more likely, their deprotonated form, the oximates), which nucleophilically substitute the phosphylated serine in the active site.207

Exposure of AChE to OP compounds is complicated by an aging process in which loss of the alkyl side chain of the phosphylated serine produces an oxyanion of OP- poisoned AChE.208,209 Oximes, such as 2-pralidoxime (2-PAM), are ineffective against aged AChE. Some OP compounds, such as soman, with an aging half-time (t1/2) of only several minutes, provide only a minimal chance for medical treatment.210 After decades of research, no clinical treatment has been developed to resurrect aged AChE.

To reverse aging, realkylation of the phosphylated oxyanion has been proposed as a strategy against this dealkylation process,151 including unsuccessful efforts from 1970 by Steinberg et al.149 as well as more recent efforts led by Quinn.147 With the negative charge on the phosphylated serine being neutralized by some sort of electrophilic realkylation process, oximes should reactivate AChE again. Several types of electrophilic alkylating agents including sulfonates,151 haloketones,149 sulfoniums,146 and methoxypyridiniums147 were evaluated as potential AChE realkylators. Recently,

Khavrutskii and Wallqvist suggested the in silico possibility to resurrect aged AChE by

141

β-aminoalcohols, by a direct process without proceeding through a realkylation event.145

However, prior to this report, no experimental evidence has been reported for the efficacy of any drug to realkylate aged AChE to the best of our knowledge.88 Herein, we report the first family of compounds that demonstrate in vitro efficacy.

Quinone methides (QMs, Figure 61) can be regarded as carbocations stabilized by resonance delocalization. Quinone methide precursors (QMPs, Figure 61) are derivatives of QMs with a leaving group attached to the partially positively charged carbon. QMPs can be attacked by nucleophiles either directly via SN2 substitution, or via the corresponding QMs as reactive intermediates (Figure 61). Protein and nucleic acid alkylation by QMPs has been reported for years.158,211–215 Phosphodiesters and dibutyl phosphate, which structurally resemble the phosphyl group of aged AChE, have been successfully alkylated by QMs, implying the possibility to realkylate and resurrect aged

AChE with QMPs and QMs.154,155,158

Herein we report a series of QMPs intended to be realkylators of aged AChE.

Guided by in silico studies, a library of candidate compounds was synthesized. Their activities were characterized by Ellman’s assay, providing up to 32.7% in vitro resurrection of electric eel AChE (eeAChE) for a methylphosphonate and up to 20.4% for a phosphate after 4 days. Resurrection of AChE was confirmed by bottom-up proteomics.

142

Figure 67. Structures and substitution reaction of QMs and QMPs. Typical structures of QMs (top) and QMPs (bottom). Nucleophiles can substitute the leaving group of QMP by either an SN2 reaction or formation of the corresponding QM.

3.2 Computationally Guided Selection of realkylator Candidates.

We previously conducted molecular docking and molecular dynamics (MD) simulations to evaluate the potential orientation of QMPs at the active site of methylphosphonate-aged human AChE (huAChE).159 In this work a larger library of

QMPs was studied through this modeling approach using an in silico model of aged huAChE.

143

Figure 68. A snapshot obtained from a 1 ns MD simulation, demonstrating a protonated QMP (shown in cyan, later called C8) near the active site of aged AChE (wall-eyed stereo). The electrophilic carbon is ~4.2 Å from the phosphylated oxyanion. Hydrogen bonds with short contact distances are shown (green dashed lines).

Figure 69. Structures of realkylator candidates.

We determined that pyridyl compounds had a higher propensity to be bound in the active site and close to the phosphylated oxyanion as compared to their phenyl analogues.159 Moreover, 3-hydroxypyridine-derived QMPs with the reactive benzylic carbon attached at the 2-position displayed promising interactions. Of the 72 compounds modeled, six of the top-10 compounds were members of that specific 3-hydroxypyridine

144 framework. The top compound had a pyrrolidine leaving group attached to the reactive benzylic carbon (Figure 68).

Thirteen 3-hydroxypyridine-derived QMPs (C series, Figure 69) were thus synthesized via Mannich reactions, and then evaluated by screening against aged AChE.

The electrophilic benzylic methylene, hypothesized to be the site of attack by nucleophiles (Figure 69), is attached to the 2-position of the pyridine ring, and the leaving groups were various secondary amines.

3.3 Screening of Reakylator Library.

Two representative OPs (Figure 70) were used. PiMP (a pinacolyl methylphosphonate ester), a soman analogue, was synthesized as reported by Amitai et al.216 The resulting methylphosphonate-aged AChE is the aging product of any methylphosphonate nerve agent (e.g. sarin, soman, VX as well as many other V-agents and G-agents except for tabun (GA)). We also used the commercially available pesticide

DFP (diisopropyl fluorophosphate) to evaluate a phosphate at the serine residue (Figure

70). We chose eeAChE as the target enzyme of these studies, considering its commercial availability and affordable cost.

145

Figure 70. Structures of OPs used for aging of AChE and the corresponding aged adducts.

The aged enzyme was reacted with various concentrations (0.2 – 4 mM) of each

QMP at 37°C, pH 8 for 1 d. Without knowing the re-aging t1/2 of the possible realkylated

217–220 adducts, NH4F (4 mM) was added as a mild and nonselective reactivator. We posited that once the aged AChE was realkylated, the newly formed “inhibited” AChE could be reactivated by fluoride to the native AChE. The samples were eventually treated with 4 mM of 2-PAM for 1 h to ensure that any realkylated AChE was completely reactivated. Ellman’s assay,221 with acetylthiocholine as a substrate, was carried out to determine AChE activity. Three controls were prepared and analyzed in parallel: a negative control without realkylator; a 2-PAM control with the realkylator replaced by 2-

PAM; and a positive control with native AChE, rather than aged, and without realkylator.

146

Figure 71. Screening of three concentrations of various C series QMPs against (a) methylphosphonate-aged eeAChE and (b) isopropyl phosphate-aged eeAChE. The horizontal solid and the dashed lines mark the negative controls and 2-PAM controls, respectively. The error bars reflect standard deviations from four replicate efforts.

147

The screening results are shown in Figure 71a. The percentage relative activity is based on the positive control. The negative and 2-PAM controls showed negligible background signals, confirming the completion of aging (solid and dashed horizontal lines in Figure 71). For this small library, the only difference was the amino leaving group. Compound C2 with an N-(methyl)ethylamino leaving group showed the highest efficacy among C1 to C7, with each of these compounds having a noncyclic amine.

Efficacy was compromised when the N-alkyl groups were shortened (C1 vs C2), lengthened (C3, C5 and C7 vs C2), or branched (C4 vs C2, and C6 vs C3). Overall, the pyrrolidinyl compound C8 was the most potent candidate. The other candidates with cyclic leaving groups showed lower or even no activity. The screening against isopropyl phosphate-aged AChE (DFP-treated) showed a similar result (Figure 71b), but the trend of varying activity with the N-alkyl groups was less obvious.

These comparisons demonstrate that pyrrolidine is the optimum among all tested leaving groups. Hence we compared a variety of realkylator candidates with the pyrrolidinyl leaving group (D series in Figure 71). The screenings were carried out following the same procedures as for the 3-hydroxypyridine derivatives (C series). C8 and unsubstituted 3-hydroxypyridine were used for comparison. D5 is the only active compound of these candidate compounds. D5 is active only against isopropyl phosphate- aged AChE, and is less active than C8. This observation indicates the superiority of the

3-hydroxypyridine framework over other tested scaffolds. C8 was therefore chosen as a lead compound for subsequent investigations.

148

3.4 Kinetics of Resurrection.

In order to further confirm the reactivity of C8 and determine the rate of reaction, we monitored the resurrection of aged AChE by this compound for days. Out of curiosity, neither 2-PAM nor fluoride was added to the QMP-treated samples for reactivation of realkylated AChE. As the reaction progressed, linearly increasing activity was seen in the isopropyl phosphate-aged AChE sample treated with 4 mM C8 (Figure 72a). The negative control and 2-PAM control remained inactive. After four days, the resurrected relative activity was as high as 20.4%. The apparent reaction rate was 0.20% per hour (r2

= 0.9958).

Resurrection kinetics of methylphosphonate-aged AChE was monitored similarly.

The C8-treated sample displayed resurrected activity, which reached 32.7% after four days (Figure 72b). The apparent reaction rate was 0.32% per hour (r2 = 0.9185) and faster than against isopropyl phosphate-aged AChE.

The absence of extra reactivators (NH4F or 2-PAM, etc.) in this reaction suggests the inherent reactivation activity of C8, that is, reactivating the realkylated AChE. This is in agreement with reports of reactivation activity of Mannich phenols by Katz et al.,113

Cadieux et al.,114 and Bierwisch et al.115

149

a b

120 140

100 120 100 80 80 Negative Control 60 Negative Control 2-PAM Control 60 2-PAM Control QMP 40 QMP 40 Positive Control Positive Control Relative activity (%) activity Relative (%) activity Relative 20 20

0 0 0 20 40 60 80 100 0 20 40 60 80 100 Time (h) Time (h) Figure 72. Kinetics of realkylation of aged eeAChE by 4 mM C8 at pH 8 over 4 days for (a) isopropyl phosphate-aged (DFP-treated) eeAChE and (b) methylphosphonate-aged (PiMP-treated) eeAChE. The blue dashed lines illustrate the result of linear regression.

3.5 Bottom-Up Proteomics

Besides determining the resurrected AChE activity by Ellman’s assay, confirmation of the reaction between the QMP and aged AChE can also be revealed by mass spectrometry. Bottom-up proteomics222,223 was used to sequence the peptides and differentiate enzyme species with variable modifications at the catalytic serine. We treated isopropyl phosphate-aged eeAChE with C8 for 11 d, and digested it with trypsin.

2-PAM was not applied to the C8-treated sample. The digest was analyzed with LC-

MS/MS. The positive, negative and 2-PAM controls were also prepared in parallel, and percentages of AChE species were obtained from the LC peak areas (Table 6).

150

Table 6. Percentages of eeAChE species after resurrection by C8, as determined by LC- MS/MS.

Percentage (%)

Sample Isopropyl phosphate-aged Methylphosphonate-aged

Native Aged Realkylated Native Aged Realkylated

Positive Control 100 0 0 100 0 0

Negative Control 0 100 0 0 100 0

2-PAM Control 0 100 0 0 100 0

C8-treated 15.4 84.6 0 2.1 97.9 0

Peptide QVTIFGESAGAASVGMHLLSPDSRPK (residues 195–220), was observed in all samples. The catalytic serine (underscored) in the positive control was completely unmodified because the enzyme was native. By contrast, the modification observed in the negative and 2-PAM controls indicated that they were completely aged.

The wild-type serine or otherwise modified serine was not observed at this position.

Compared to the unmodified peptide, there was a mass shift of 122.0133 Da (C3H7O3P added), which matches the added isopropyl phosphyl moiety. In the C8-treated sample, the unmodified catalytic serine was observed, indicating resurrection induced by the

QMP. Realkylated AChE, however, was not directly observed. This again suggests that our QMPs can resurrect aged AChE independent of extra reactivators such as oximes due to their inherent reactivation activity. Another possibility is that the realkylated phosphyl

151 group was lost under used MS conditions. Similar results were obtained with methylphosphonate-aged AChE (Table 6).

3.6 Reactivation of Inhibited AChE by Five QMPs. a b

140

120

100 Positive Control 80 2-PAM (0.04 mM) 60 3-Hydroxypyridine (0.4 mM) 40 D5

Relative activity (%) activity Relative C8 C2 20 C1 Negative Control 0 0 4 8 12 16 20 24 Time (h) Figure 73. Reactivation activity test of four QMPs (4 mM) against eeAChE inhibited by EMP. (a) Structure of EMP. (b) Relative activity of eeAChE as reactivation proceeded. Four representative QMPs, namely C1, C2, C8, D5 and unsubstituted 3- hydroxypyridine, were reacted with eeAChE inhibited with a VX analogue (EMP, an ethyl methylphosphonate compound, Figure 73a) to determine whether they are reactivators. We chose this OP because the aging by VX is slow.81 After inhibition by

EMP, the inhibited AChE was cleaned by Sephadex spin columns, then incubated with the chosen QMPs (4 mM) at pH 8.0 and 37°C. Aliquots were taken and cleaned also with

Sephadex spin columns, and monitored by Ellman’s assay (Figure 73b).

All tested QMPs exhibited obvious and similar reactivation activity, regardless of their different performances in the aforementioned realkylation screening. After 24 h of

152 reaction, they reactivated 64%~80% of inhibited AChE. This may explain why C8 can efficiently resurrect aged AChE in the absence of 2-PAM and fluoride. All tested compounds reacted slower than 3-hydroxypyridine and 2-PAM, which reached their respective plateaus within 1 h of reaction, although they were used at only 0.4 and 0.04 mM concentrations, respectively.

3.7 pH Effect and EC50 for C8 Activity.

We performed the resurrection of methylphosphonate-aged eeAChE by 4 mM C8 at four different pH values (6~9) at 37°C for 1 d, with neither 2-PAM nor fluoride being used. The relative activity of C8-resurrected eeAChE increased dramatically with pH

(Figure 74a) – over 20% for 1 d at pH 9. Similar effects, but less pronounced, were also observed with isopropyl phosphate-aged eeAChE (Figure 74b).

25 8

Negative Control 7 Negative Control 20 QMP QMP 6

15 5

4 10 3 Relative activity (%) activity Relative Relative activity (%) activity Relative 2 5 1

0 0 6 7 8 9 6 7 8 9 pH pH a b

Figure 74. Resurrection efficiency of 4 mM C8 against aged eeAChE increased with pH within the range of pH 6~9. (a) Methylphosphonate-aged (PiMP-treated) eeAChE. (b) Isopropyl phosphate-aged (DFP-treated) eeAChE.

153

Seven concentrations of C8 (0 – 20 mM), were therefore compared at pH 9 against methylphosphonate-aged and isopropyl phosphate-aged eeAChE, in order to determine the EC50. The results (1.21 and 1.02 mM, respectively) illustrate that further optimization in binding affinity is needed for a more effective therapeutic.

There are three heteroatoms in C8, forming multiple possible protonation states.

Each of them may interact with the aged AChE active site with different orientations, affinities and rates. 1H NMR spectra of C8 displayed shifting signals as the pH changed

(Figure 75a). The UV-vis spectra also dramatically changed when pH was increased from

6 to 9 (Figure 75b), indicating a pKa between 7 and 8.

a b

3.0 pH6 2.5 pH7 2.0 pH8 pH9 1.5

Absorbance 1.0 0.5 0.0 240 280 320 360 Wavelength (nm)

Figure 75. Influence of C8 protonation states on spectra. (a) 1H NMR spectra of the aromatic protons and (b) UV-vis spectra of C8 as pH is varied from 6~9.

154 a b

Figure 76. (a) The four most probable protonation states of C8 at pH 8-9. (b) A representative snapshot in the MD simulation of C8c (cyan, wall-eyed stereo). Hydrogen bonds with short contact distances are shown (green dashed lines).

The strong pH dependence motivated us to study which isomers might be the active component(s). We compared the four most probable states at pH 8 and 9, namely the neutral, anionic and two zwitterionic forms (C8n, C8a, C8b and C8c, respectively,

Figure 76a). Each protonation state was individually docked into a number of geometries of aged AChE and subsequent MD simulations were run on the docked conformations; the location of the exchangeable proton has a dramatic effect on the affinity of the ligand toward different zones in the AChE active site. The simulations suggest that both zwitterionic forms (C8b and C8c) have stable interactions and well-defined proximity to the anionic oxygen of the phosphylated serine. The neutral form (C8n) also stably binds to the active site, but the benzylic carbon is not oriented for realkylation. The anionic form (C8a) gradually moves away from the active site during the simulation.

Calculations at the B3LYP/6-311+G** (SMD, water) level of theory suggest that C8c

(pyrrolidinium) has a free energy that is lower than C8b (pyridinium) by 3.1 kcal/mol, hence is a more probable species. Ser203, Tyr337, Tyr341 residues, along with some 155 hydrogen-bonded water molecules, contribute significantly to the binding of C8c by hydrogen bonding and electrostatic attractions (Figure 76b).

3.8 Resurrection of Aged Human AChE after exposure to DFP.

120

100

60 Negative Control 40 QMP Positive Control Relative activity (%) activity Relative 20

0 0 1 2 3 4 5 6 7 8 Reaction time (d) Figure 77. Kinetics of resurrection of isopropyl phosphate-aged huAChE by 4 mM C8 (pH 9). The green dashed line illustrates the result of linear regression.

The efficacy of C8 was examined against aged huAChE. The activity of isopropyl phosphate-aged (DFP-treated) recombinant huAChE was monitored in the presence of 4 mM of C8 at pH 9 for a week by Ellman’s assay. After 7 d of reaction (Figure 77), 18% of the aged enzyme was resurrected, and the relative activity of huAChE was still increasing. The apparent rate of resurrection was 2.5% per day, as determined by linear regression. Neither 2-PAM nor fluoride were used in this test, again confirming that C8 alone can resurrect aged AChE to the native form.

3.9 Discussion and Conclusions.

With in silico guidance, we synthesized 20 QMPs with various scaffolds and leaving groups, and then evaluated them for their in vitro resurrection activity against methylphosphonate-aged and isopropyl phosphate-aged eeAChE. Multiple QMPs with a 156

3-hydroxypyridine scaffold, with the reactive “benzylic” carbon attached at the 2- position, proved efficacious. C8 was found to be the lead compound in the screening.

Used at 4 mM without the assistance of 2-PAM, C8 successfully resurrected the relative activity of isopropyl phosphate-aged eeAChE to 20.4% and methylphosphonate-aged eeAChE to 32.7% after 4 d of observation (Figure 72). This activity for C8 is also reproduced for aged human AChE (Figure 77).

The activities of C8 against isopropyl phosphate-aged and methylphosphonate- aged eeAChE were also confirmed with bottom-up proteomics. The peptide containing the catalytic serine was sequenced to reveal the modification. In the samples treated with

C8, partial resurrection was observed as indicated by the presence of unmodified catalytic serine from native AChE. Realkylated AChE was not directly observed, though no extra reactivator was added. These observations suggest the activity of C8 against both aged AChE and inhibited/realkylated AChE. This hypothesis was confirmed as C8 and three other QMPs were able to reactivate inhibited AChE (Figure 73).

At this stage, C8 is not a practical drug for clinical post-aging treatment of OP poisoning or in vivo tests. The rate of reaction is slower than ideal, although the concentration was on the millimolar scale. Whether this concentration will cause toxicity also remains unknown. And, 2-4 d of reaction was needed to resurrect ~20% of activity.

However, the resurrection rate is increased dramatically when pH is increased to 9, suggesting that structure-activity relationships and a deeper mechanistic understanding may be effective in affording better activity. Calculations suggest that the zwitterionic

157 form of C8, with the pyrrolidinyl nitrogen being protonated, plays a critical role in activity.

Finally, this report provides the first in vitro success to substantially resurrect aged AChE via the use of an appropriate realkylator. After almost 70 years of effort, this goal has been achieved, but further work remains to create a therapeutic drug with all of the desired activity and selectivity.

3.10 Experimental Section.

Computation-guided selection of realkylator candidates

Optimizations, molecular docking, and MD simulations were all performed with the initial library of compounds. The three lowest-energy docking poses of each flexible ligand across 13 rigid aged AChE structures were used as starting points for subsequent 1 ns MD simulations. The QMPs were evaluated based on the time throughout the MD simulations in which the reactive benzylic carbon was within close proximity to the anionic O-(P=O) of the aged serine.

LC-MS characterization of realkylator candidates

Mass spectra were obtained using Thermo LTQ Orbitrap (ESI) mass spectrometer. HPLC was performed with Agilent 1100 HPLC system equipped with

Phenomenex Luna 5 µm C18 column (150 × 4.6 mm), eluted with mixtures of water and methanol (5%~95% methanol, containing 0.1% formic acid, 0.6 mL/min), and detected by UV-vis detector at 254 nm as well as ESI-orbitrap. The purity of each realkylator candidate was confirmed to be ≥ 95%.

158

Preparation of Aged AChE

Some OPs (mainly the phosphonates, which are commonly seen in G- and V-type warfare agents) are chiral, and the stereoisomers can inhibit and/or age at different rates.224 Inhibited AChE may remain un-aged even after reacting for longer than the apparent t1/2, if the enzyme is inhibited by the slower aging OP stereoisomer.

Reactivation of the residual un-aged, but inhibited, AChE can interfere with the observation of the resurrection of aged AChE and may lead to artifacts.

Thus, AChE must be thoroughly aged and free of inhibited AChE. To ensure complete aging, the methylphosphonate-aged AChE (treated with PiMP) was prepared via two rounds of aging, considering the chirality of PiMP. We treated AChE with 2-

PAM after the first round of aging, in order to reactivate any residual inhibited AChE.

Then PiMP was added to inhibit and age the enzyme again. The fast-aging isomers thus had a second chance to compete with other isomers and react with the enzyme. The amount of residual inhibited or native AChE in the sample was significantly minimized, often <1% residual activity of a native AChE control. By contrast, to age AChE with achiral DFP, only one round of aging was needed.

Ellman’s Assay

Ellman’s assay221 was carried out on clear flat-bottom 96-well microplates. The assay solution was 180 µL × 0.56 mM of acetylthiocholine in pH 8.0 40 mM buffer, containing 0.1 g/L of BSA and 1.1 mM of 5,5’-dithio-bis-(2-nitrobenzoic acid) (DTNB).

159

20 µL of tested AChE sample was added to initiate the reaction. Samples were diluted if the activity was very high. The absorption at 412 nm was monitored at 25°C with

Molecular Devices SpectraMax i3 microplate reader. Each sample was tested in four replicate wells. The average initial slope of the four absorbance-time plots of each sample and the standard deviation were divided by that of the control to obtain the relative activity and error.

Screening of Realkylator Library

Each realkylator (4 µL × 5, 25 or 100 mM) was mixed with methylphosphonate- aged AChE (2 µL), phosphate buffer (90 µL × 200 mM, pH 8.0 containing 1 g/L BSA and 0.02% NaN3), and NH4F (4 µL × 100 mM). After reacting at 37°C for 1 d, each sample was treated with 2-PAM for 1 h to reactivate the realkylated AChE. Three controls were prepared in parallel. In the positive and negative controls, the realkylator solution was replaced with blank buffer. In the 2-PAM control, it was replaced with 4 mM 2-PAM solution. In the positive control, the aged AChE was replaced with the same amount of native AChE. After the reactions, all reagents were removed with a Zeba desalting spin plate (purchased from Thermo Fisher, USA). The AChE activity of each sample was determined with Ellman’s assay.

Kinetics of Alkylation Induced by C8

Aged AChE (5 µL) was mixed with 955 µL × 200 mM of pH 8.0 phosphate buffer (200 mM, containing 1 g/L BSA and 0.02% NaN3) and C8 solution (40 µL × 100

160 mM) and incubated at 37°C. The negative control, positive control and 2-PAM control were also prepared in parallel as aforementioned. Aliquots of 95 µL were taken at various time intervals (9~24 h), all reagents were then removed by Sephadex size exclusion spin columns (filled with 0.1 g dry weight of superfine Sephadex G-25). AChE activity was then determined with Ellman’s assay as aforementioned. 100% was set at the t = 0 h point of the positive control.

Bottom-up proteomics

Isopropyl phosphate-aged AChE (60 µL, aged with DFP as described above without use of BSA) was mixed with pH 8.0 phosphate buffer (36 µL × 40 mM, no BSA or NaN3) and C8 solution (4 µL × 100 mM). BSA was not used to minimize complication in LC-MS/MS. The positive control, negative control and 2-PAM control were prepared in parallel as aforementioned. After reacted for 11 d at 37°C, 100 µL acetonitrile was added to denature the enzyme. 10 min later, the solution was washed through an Amicon

Ultra centrifugal filter (3 kDa cut-off molecular weight, purchased from EMD Millipore) and with 3 × 400 µL ammonium acetate buffer (40 mM, pH7.5) to remove organic solvent and reagents. 40 µL × 0.1 g/L modified porcine trypsin (Arg and Lys methylated, sequencing grade, purchased from Promega) was added. After reacted at 37°C for digestion for 7 h, the reaction was terminated by the addition of 10 µL × 1 M acetic acid.

Methylphosphonate-aged AChE in the absence of BSA as well as the controls were similarly tested with bottom-up proteomics. Methylphosphonate-aged AChE (50 µL, aged with PiMP as described above without use of BSA) was mixed with pH 8.0

161 phosphate buffer (45 µL × 40 mM, no BSA or NaN3) and C8 solution (12 µL × 100 µL).

The samples were reacted for 5 d.

Reactivation of Inhibited AChE by QMPs

Electric eel AChE (~4 units in 1 µL × 50% glycerin) was mixed with BSA (96.5

µL × 1g/L in 200 mM phosphate buffer, pH 8.0, containing 0.02% NaN3) and EMP (2.5

µL × 0.2 mM in 2% DMSO). After reacted at 37°C for 1 h, the solution was cleaned with a Sephadex spin column. Controls with EMP replaced by blank solvents were also prepared.

Each tested compound (4 µL × 100 mM, final concentrations 4 mM) was mixed with 1 µL of the freshly prepared inhibited AChE solution and BSA (95 µL × 1 g/L in

200 mM phosphate buffer, pH 8.0, containing 0.02% NaN3). Six replicates were made for each compound. One replicate was taken at various time intervals and cleaned again with

Sephadex columns, then analyzed with Ellman’s assay as aforementioned. For comparison, 1 mM 2-PAM and 10 mM 3-hydroxypyridine were tested in parallel (final concentrations 0.04 and 0.4 mM). The negative and positive controls had the reagent solution replaced with blank water. The AChE in the positive control was native and not treated with EMP.

pH Effect in Resurrection of Aged AChE

162

To 94 µL × 1 g/L BSA solution (in 200 mM phosphate buffer, containing 0.02%

NaN3), 2 µL of methylphosphonate-aged AChE solution (treated with DFP or PiMP as described above) and 4 µL × 100 mM C8 solution were added. A negative control was prepared with C8 solution replaced with blank water. A positive control was similarly prepared in parallel with the aged AChE replaced with native AChE. Buffers at four pH values were compared: 6, 7, 8 and 9. Samples were incubated at 37°C for 1 d. AChE activity was then determined with Ellman’s assay.

Measurement of EC50 of C8

Seven concentrations of C8, ranging from 0 to 20 mM, were compared against methylphosphonate-aged, combined with 2-PAM as the reactivator. NH4F was not added in this test to see whether rapid reactivation is necessary to suppress re-aging. The specific procedures are as follows:

To 94 µL of pH 9.0 phosphate buffer (200 mM, containing 1 g/L BSA and 0.02%

NaN3), 2 µL as obtained aged AChE (treated with PiMP) and 4 µL of C8 solution

(neutral, 0, 25, 50, 100, 200, 300 and 500 mM) were mixed, and allowed to react at 37°C for 1 d. The final concentrations of C8 were hence 0, 1, 2, 4, 8, 12 and 20 mM, respectively. The positive control and 2-PAM control were also prepared in parallel following a procedure similar to that aforementioned. The reagents were removed in the end by size exclusion spin columns. Ellman’s assay was carried out following the aforementioned procedure.

163

Resurrection Kinetics of Aged Recombinant Human AChE

Recombinant human AChE was provided by Dr. Douglas Cerasoli (USAMRICD, purchased from Chesapeake PERL Inc) or by Dr. Zoran Radić (UCSD, expressed with

HEK-293 cells).225

To 97 µL of buffer at pH 7.0 (200 mM phosphate, containing 1 g/L bovine serum albumin and 0.02% NaN3), recombinant human AChE (0.5 µL, ~50U/µL in 50% glycerol) and DFP (2.5 µL, 0.2 mM in 2% DMSO) were added. After reacting for 3 d, the solution was washed with 3 × 400 µL pH 8.0 buffer in a centrifugal ultrafilter (Amicon

Ultra 0.5 mL, cut-off molecular weight 30 kDa). 1 µL × 2% NaN3 was added and the solution was stored at 4°C for further use. A control of native AChE was prepared in parallel by replacing DFP solution with blank 2% DMSO.

Aged AChE (20 µL) was mixed with 172 µL × 200 mM phosphate buffer (pH

9.0, containing 1 g/L BSA and 0.02% NaN3) and C8 solution (neutral, 8 µL × 100 mM) and incubated at 37°C. The negative control and positive control were also prepared in parallel as aforementioned. Aliquots of 20 µL (10 µL from the positive control due to high activity) were taken at various time intervals (1 or 2 d), and diluted to 100 µL.

AChE activity was then determined with Ellman’s assay as aforementioned. 50% was set at the t = 0 d point of the positive control (not 100% due to difference in dilution).

164

Chapter 4. STRUCTURE-ACTIVITY RELATIONSHIPS FOR SUBSTITUTED 3- PYRIDINOL MANNICH BASES IN THE RESURRECTION OF METHYLPHOSPHONATE-AGED HUMAN ACETYLCHOLINESTERASE

The research presented in this chapter is adapted from the manuscript that has been submitted to the Journal of Medicinal Chemistry. The computational work was completed by Joseph Fernandez and Ola Nosseir. The synthesis was a collaborative effort between a number of undergraduate students with guidance and participation of Dr. Christopher Callam and myself.

The OP compounds, both purchased pesticides as well as synthesized nerve agent analogues, are extremely hazardous and should be handled with the utmost care. The byproducts of reactions with the enzyme, fluoride, and oximes additionally produce compounds of unknown toxicity that should be regarded also as being extremely hazardous. When handling these reagents and solutions, appropriate personal protective equipment (PPE), including a lab coat, gloves, safety glasses and a well ventilated fume hood, should be utilized at all times. When isolating or purifying synthesized analogues, additional butyl gloves and a chemical faceshield should be warn to prevent exposure to any potentially exposed skin. All solutions, glassware, caps, spatuales, or any other laboratory equipment that comes into contact with the OP compounds or solutions should be decontaminated with concentrated sodium hydroxide solution, preferably soaking for at least 24 hours. The washing of these contaminates should be collected and clearly labeled before being disposed of as hazardous material through the appropriate environmental health and safety guidelines.

4.1 Introduction

Organophosphorus (OP) nerve agents (Figure 78A) are potent inhibitors of the critical biological function of acetylcholinesterase (AChE).20,226,227 OP nerve agents, originally discovered in 1936, have been highlighted recently in the media due to nefarious uses in the assassination of Kim Jong-nam, multiple terrorist attacks in the

Syrian civil war, and the most recent exposure to Novichok OP nerve agents in Salisbury and Amesbury.226 When OP nerve agents inhibit the function of AChE, accumulation of 165 the neurotransmitter acetylcholine (ACh) occurs at neurosynaptic junctions, and without medical intervention, can lead to a cholinergic crisis and eventually death. OP nerve agent exposure is treated by the administration of atropine (an antimuscarinic), diazepam

(an antiseizure agent) and a nucleophilic reactivator (an oxime), such as 2-PAM, HI-6, or obidoxime.90 Atropine and diazepam manage the symptoms of the cholinergic crisis, while the oxime, in a process called reactivation, recovers the activity of the OP-inhibited form of AChE (Figure 1B).228 There is also an aging process in which the OP-inhibited form of AChE is de-alkylated to a phosphylated oxyanionic serine residue (Figure 78B).

Despite years of research, three main problems still remain unsolved. (1) Some oximes have limited effectiveness against select OP compounds, specifically phosphoramidates, as demonstrated by the limited effectiveness of 2-PAM at reactivating tabun-inhibited AChE.81,229 (2) To date, the only approved oximes are positively charged pyridinium compounds, thereby limiting their effectiveness in the central nervous system due to low blood-brain barrier penetration.230 (3) Oximes, and other reactivators, are not capable of recovering the activity of aged AChE.35,90,93

Recent efforts have noted the utility of non-oxime based reactivators for the recovery of OP-inhibited AChE,91,113,115,116,231 and in select cases, also recovery of OP- aged AChE.160,232 Katz et al. reported that hydrophobic cores can serve as a binding element and, when tethered with Mannich bases, can serve as reactivators of OP-inhibited

AChE.91,113 In particular, the molecular fragment referred to as ADOC (4-amino-2-(N,N- diethylaminomethyl)phenol, 1, Figure 79) was particularly effective at reactivation of the

OP-inhibited form of recombinantly generated human AChE. Subsequent investigations

166 by Cadieux et al.231 and de Koning et al.116 found that minor modifications of the ADOC structure had significant effects on potency and efficacy. During the course of their studies, de Koning et al. discovered that the pyrrolidine version of ADOC (2, Figure 79) was even more effective in the reactivation of OP-inhibited human, erythrocyte-ghost

AChE.116

We have recently reported that 3-hydroxypyridine-based Mannich phenols (3,

Figure 79) are capable of recovering the activity of AChE that has been aged to the phosphylated oxyanion form.160 Such recovery of activity from the aged (dead) form to the native (active) form of AChE is a process referred to as resurrection by Quinn (Figure

1B).148 Our initial studies made use of primarily electric eel AChE, but we have also confirmed that resurrection of aged AChE was possible with a recombinant human form of AChE (rhuAChE), as reported by Radić et al.,225 after exposure to OP nerve agent analogues.160 Our initial study showed that a pyrrolidine ring as the amine provided the best efficacy, and in a similar manner to that reported by de Koning et al.116

A O O O P P P N O O O F F N tabun (GA) sarin (GB) soman (GD)

O O O P N P N P B O S S R F O O inhibition O O aging O O VR P P cyclosarin (GF) VX O OH P R O O O LG CN O reactivation P native AChE inhibited AChE aged AChE O O O O 2-PAM

CMP resurrection

Figure 78. (A) Structures of various authentic organophosphorus (OP) nerve agents and CMP, an OP simulant for GF. (B) General schematic of the inhibition and aging process of AChE, following exposure to a methylphosphonate OP. 167

More recently, we observed that minor modifications of the structure of 3- hydroxypyridyl-based Mannich bases can result in very large differences in OP resurrection, specifically with use of the enantiopure (R)-2-methylpyrrolidine group as the amine (4, Figure 79).232 Noting the significant efficacy afforded by the enantiopure amine, we sought to conduct a structure-activity relationship study for the resurrection of methylphosphonate-aged human AChE, as derived from exposure to methylphosphonate chemical nerve agents (Figure 78), and with a primary focus of substituent effects on the aromatic ring. We made use of five amines: dimethylamine, diethylamine, pyrrolidine,

(R)-2-methylpyrrolidine, and piperidine, while we also varied the groups attached to the aromatic core from methyl, chloro, bromo, and methoxy as well as the position of attachment, 4-, 5- and 6-positions (Figure 80A and B). These modifications resulted in a number of active compounds for resurrection of OP-aged human AChE, and further in silico studies have helped elucidate the key interactions responsible for their increased reactivity. Our results are reported herein.

OH OH

N N H2N H2N

1 2

N N N N

OH OH

3 4

Figure 79. Mannich bases capable of reactivation or resurrection of AChE after exposure to OP nerve agents.

168

4.2 Resurrection of rhuAChE after aging with CMP, a methylphosphonate nerve agent analogue

A total of 40 substituted quinone methide precursors (QMPs) were synthesized for testing of their resurrection efficiencies (Figure 80B). These QMPs are hypothesized to generate a reactive quinone methide (QM) intermediate for re-alkylation of the aged form of AChE (Figure 80A). The compounds were synthesized from commercially available 3-hydroxypyridines via a Mannich reaction,233 except for compounds 11a-e.

Compounds 11a-e were prepared by a chemical sequence including a benzyl protection,

Stille coupling,234 deprotection, and then Mannich reaction (Figure 80C). As reported previously,160 we made use of a cyclosarin analogue (CMP, Figure 78) that had been enriched in the more toxic enantiomer through the enzymatic resolution with the 3B3 variant of paraoxonase-1 (PON1).232 This resolution aids in minimizing the issues arising from different diastereomeric pairs resulting from the inhibition and aging process, thereby resulting in a lower baseline for our resurrection studies. It is worth noting that G and V series nerve agents, with the exception of tabun, age to the same methylphosphonate anion in the active site of aged AChE (Figure 78); thus, CMP is representative of the whole series of OP chemical nerve agents, except phosphoramidates such as tabun (and Novichok agents).

169

N O O P A O O O OH 6 N N P R ⎼R-H AChE O 5 OH O 4 +R-H AChE QMP QM

N N Br N N B R R R R

OH OH OH OH Br 5a-e 6a-e 7a-e 8a-e

N N N N R R R R

Cl OH Br OH OH O OH

9a-e 10a-e 11a-e 12a-e

R = N N N N N

a b c d e C

N HNR1R2, paraformaldehyde N benzene NR1R2 OH OH X X

BnBr, K CO SnMe , Pd(PPh ) Pd(OH) /C, H N 2 3 N 4 3 4 N 2 2 N DMF DMF EtOAc, hexanes

Br OH Br OBn H3C OBn H3C OH

42% 76% 91%

Figure 80. (A) The numbering scheme used to refer to substituted quinone methide precursors (QMPs) and the hypothesized route of re-alkylation of methylphosphonate- aged AChE with the quinone methide (QM) intermediate. (B) QMP library of compounds synthesized for testing of resurrection of methylphosphonate-aged human AChE. (C) A general scheme for the Mannich reaction that was used to synthesize the QMPs from the 3-hydroxypyridine frameworks. The three-step synthetic route used to synthesize the 5- methyl-3-hydroxypyridine framework for compounds 11a-e.

To determine the resurrection efficacy of each of the compounds, CMP-aged recombinant human AChE (rhuAChE) was prepared by incubating an excess of the enantio-enriched CMP with native rhuAChE for 24 hours at a pH of 6.0 and 37 °C in order to accelerate the aging reaction.235 The OP was removed by ultrafiltration and the

170 buffer was adjusted to pH 7.5. An excess of 2-PAM, a known AChE reactivator, was then added and incubated for 1-2 hours at 37 °C. The reactivator was removed by ultrafiltration and the isolated enzyme was analyzed by Ellman’s assay, using acetylthiocholine and DTNB,221 to ensure complete aging before moving to the incubation with the QMP compounds.236 Then, after aging was ensured, each QMP (final concentration of 1 mM) was added separately to each tube, and importantly, no additional reactivator was included in the samples from this point on. The QMP samples with OP- aged rhuAChE were incubated for 24 h at 37 °C and diluted to a final QMP concentration of <10 µM prior to a final analysis by Ellman’s assay with acetylthiocholine and DTNB.

The recovered activity was compared to a rhuAChE control that was not exposed to the

OP, in order to determine the amount of rhuAChE that had been resurrected. In all of the following figures, the y-axis shows the percentage of rhuAChE activity, relative to the positive control, as recovered from the aged state by the QMP, and in the absence of any additional reactivators.

As shown in Figure 81, 20 different compounds can recover some activity of the

OP-aged rhuAChE above that of 1 mM 2-PAM, with the highest recovered activity being

~14.5% in 24 hours. Seven of the compounds were above 5% in 24 hours. (The 2-PAM control (~1%) provides a measure of the recovery of any residual OP-inhibited, but not aged, rhuAChE.)

For the structures in this library, compounds 5 (4-Br), 6 (4-Me) and 7 (6-Br), along with the various amines, are all ineffective at recovering any activity of aged rhuAChE. These results suggest that 4-substitution on the pyridine ring of the QMP

171 eliminates resurrection efficacy. In the case of the 6-substituted families 7 (6-Br) and 8

(6-Me), only the 6-methyl (8) compounds are active. Perhaps this is due to steric issues resulting from the size of the Br atom and an inability to create an effective binding pose.

Interestingly, all of the 5-substitued compounds, i.e., 9–12 are active, with the exception of 10d, suggesting that substituents are well tolerated at the 5-position and perhaps even preferential over the 4- and 6-positions. Both electron-withdrawing and electron- donating, including large and small, substituents are tolerated at the 5-position for resurrection of methylphosphonate-aged rhuAChE.

172

100

5 Relative activity (%)

0 5 6 7 8 9 4 10 11 12 2PAM Controls Substitution

Figure 81. Resurrection activity of QMPs against a methylphosphonate-aged (CMP) rhuAChE (37°C, pH 7.5, 24 hours). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figures 2 and 3 for structures. Two different 2-PAM concentrations were used: 100 µM and 1 mM, respectively, left to right. The dashed line is the activity recovered by 1 mM 2-PAM of an aged control that was not treated with a QMP and is representative of oxime reactivation of any residual OP-inhibited rhuAChE; thus, any entries above this level demonstrate recovery of CMP-aged rhuAChE activity. The error bars reflect a standard deviation of 4 replicate measurements.

Between families 9−11, there appears to be a general trend associated with the steric effects of the attached amine leaving groups. As the steric bulk increases up to (R)-

2-methylpyrrolidine, the activity also increases; however, expansion to piperidine then decreases the activity. Compound 10d is an exception to this trend. Interestingly, family

12 (5-OMe) does not follow this trend and seems to prefer smaller amines (dimethyl and diethyl). The origin of this effect may be a different binding orientation within the

173 enzyme active site as a result of the 5-methoxy substituent being a hydrogen-bond acceptor.

These results suggest that there are multiple contributors to resurrection efficiency including electronic effects of the aromatic core, steric effects around the aromatic core, and also the size/basicity of the amine leaving group. Together, these effects change both the rate of formation and reaction of the hypothesized quinone methide (QM) intermediate (Figure 3A) as well as the binding mode to be adopted within the enzyme’s active site.

4.3 pH Dependenc and EC50 determination of the best QMPs

We tested the phenomenological effects of the concentration (EC50) of various substituted compounds at both pH 7.5 and 9.0, as shown in Figure 82. We selected the seven compounds with activity above 5% resurrection efficiency. The observed plateaus did vary when compared to the single-point screening results presented in Figure 81, likely as a result of a different method of preparation prior to Ellman’s assay; dilution in the case of the single-point analysis and ultrafiltration in the case of the EC50. This filtration step, coupled with the steep concentration dependence for a majority of the tested compounds around 500 µM (Figure 5), can account for the variability in the observed percentages. For methylphosphonate-aged rhuAChE, the EC50 values at pH 7.5 span a range from 86 µM for 8d (plateau maximum at 7.7%) to 730 µM for 12b (plateau maximum at 13.1%) (Figure 82A). These values are compiled in Table 7. These results suggest that even at lower QMP concentrations (~400 µM), significant activity (>5%) can

174 be recovered. Some QMPs, such as 8b and 8d, have a steeper onset and a high plateau for resurrection activity.

When the pH is increased to 9.0, the EC50 curves change quite significantly

(Table 7 and Figure 82B). For compounds in the 8 (6-Me) family, the plateau maxima for resurrection increases, yet for families 11 (5-Me) and 12 (5-OMe), the plateaus decrease, with the exception of 12b which remains approximately the same. When comparing the groups attached as well as the regioisomer being tested, both 11 (5-Me) and 12 (5-OMe) families, with an electron-donating group situated para relative to the aminomethyl group, demonstrate a decrease in activity, while family 8 (6-Me), with the electron- donating group para to the phenol, shows an increase in activity.

Table 7. pH dependence of EC50 (mM) and the maximum percentage (%) at the plateau for the top seven compounds as shown in Figure 82. pH 8b 8c 8d 11b 11d 12a 12b

EC50 Plateau EC50 Plateau EC50 Plateau EC50 Plateau EC50 Plateau EC50 Plateau EC50 Plateau 7.5 210 10.7 988 10.9 86 7.7 491 11.7 150 6.9 389 11.2 730 13.1 9.0 310 17.0 301 17.0 202 14.4 398 6.2 165 6.5 407 5.6 916 13.1

In our hypothesized mechanism of action, we posit that when a QMP is bound in the active site of methylphosphonate-aged AChE, the QMP forms the quinone methide

(QM) reactive intermediate in the vicinity of the phosphylated oxyanion (Figure 80A).

Then, after re-alkylation of the phosphylated oxyanion, a number of paths can then ensue, including re-aging to a similar (or new) structure of the aged form and/or interaction of the re-alkylated form with a reactivator (e.g., a water molecule as a reactivator, an

175 intramolecular reactivation of the re-alkylated QM intermediate or with a second QMP for reactivation). Moreover, with changing pH, different charge states of our QMPs may be formed in different amounts, and each may bind differentially with the aged form of

AChE. With changing pH, intermolecular and intramolecular reactions, including general base catalysis, can vary for the formation of the QM intermediate as well as potential re- aging and reactivation steps. We have also recently demonstrated that the byproducts of resurrection can subsequently re-inhibit the newly recovered native rhuAChE, leading to a reduction in efficacy.232 Higher pH can possibly hydrolyze these byproducts, thereby increasing resurrection efficiency, a situation that is more reflective of the re-alkylation efficacy of these QMPs but not the overall therapeutic efficacy.

We thus hypothesize that at higher pH, the rate of reactivation of the (re)inhibited complex, resulting from re-alkylation by the QM intermediate, is accelerated by the para- methyl group leading to a higher efficiency (plateau) and that higher pH minimizes re- inhibition by any formed byproducts. It is also possible that the substitution at the position para to the phenol changes the protonation states being populated at pH 9.0, thus altering the binding of the species between families 8, 11 and 12.

176

A B

pH 7.5 8b (6-Me) pH 9.0 8b (6-Me) 15 10 8c (6-Me) 8c (6-Me)

8d (6-Me) 10 8d (6-Me) 11b (5-Me) 11b (5-Me) 5 % activity 11d (5-Me) % activity 5 11d (5-Me) 12a (5-OMe) 12a (5-OMe)

0 12b (5-OMe) 0 12b (5-OMe) 0.0 0.5 1.0 0.0 0.5 1.0 Concentration (mM) Concentration (mM)

Figure 82. EC50 evaluation for a methylphosphonate-aged rhuAChE (1.4 U for all samples, 0.14 U for positive control) after resurrection. Selected QMP concentrations (0– 5 mM) were incubated for 24 hours and 37 °C at (A) pH 7.5 and (B) pH 9.0. The error bars reflect a standard deviation of 4 replicate measurements.

4.4 IC50 Determination of rhuAChE

The compounds presented herein have shown resurrection efficacy at multiple concentrations and pH values as demonstrated by the single-point screen and the EC50 determinations. We additionally wanted to ensure that our methodology for screening was not affected by inhibition of the resurrected native enzyme, which would give rise to inaccurate results. We took the same top seven compounds screened during the EC50 study and evaluated their inhibition of the native rhuAChE enzyme to determine the IC50 of each compound. The lowest IC50 was determined to be 404 µM for compound 8d. This value is much greater than the dilution achieved during our single-point methodology where all QMPs are at a concentration of <10 µM in the final Ellman’s solution, verifying that that the results are unaffected by inhibition of the native enzyme. The range of IC50 values were between 404 to 1154 µM, showing a very broad range of weak inhibition. Addition of amine d to the studied compounds seems to significantly lower the

177

IC50 values with native AChE, and the (R)-2-methylpyrrolidine group plays an important role in binding and suggests why 8d and 11d are so effective at resurrection.

For the 12 (5-OMe) family, this selectivity for amine d is not observed. We hypothesize that the 12 family may have a different binding mode, which is further supported by the inverse amine selectivity as mentioned above. These high (unfavorable) values for the IC50 with native rhuAChE suggest that these compounds are promising for further therapeutic development as inhibition of the native form of the enzyme would be detrimental to the therapeutic effect observed following reactivation or resurrection of both forms of the enzyme.

4.5 QMP to AChE Ratio Dependence on the Efficacy of Resurrection

As has been demonstrated by our group previously, increasing the ratio of

[QMP]:[AChE] leads to an increase in the resurrection plateau.232 Here, we tested the same hypothesis for this new library’s effective compounds, specifically that decreasing the amount of AChE in solution will decrease the generation of a resurrection byproduct that can re-inhibit the resurrected native AChE.232 Changing this ratio allows for better determination of therapeutic efficiency that is independent of the varying potency of the formed resurrection byproducts.

178

A B

60 60 5 fold dilution 8b (6-Me) 10 fold dilution 8b (6-Me) 8c (6-Me) 8c (6-Me) 40 8d (6-Me) 40 8d (6-Me) 11b (5-Me) 11b (5-Me) 11d (5-Me) 11d (5-Me) % activity % activity 20 20 12a (5-OMe) 12a (5-OMe) 12b (5-OMe) 12b (5-OMe) 0 0 0 2 4 0 2 4 Concentration (mM) Concentration (mM)

Figure 83. EC50 evaluation against a methylphosphonate-aged rhuAChE (5-fold rhuAChE is 0.028 U and 10-fold is 0.014 U). Selected QMP concentrations (0–5 mM) were incubated for 24 hours and at 37 °C for (A) 5-fold diluted rhuAChE and (B) 10-fold diluted rhuAChE. The error bars reflect a standard deviation of 4 replicate measurements.

As was previously observed,232 increasing the ratio of [QMP]:[AChE] leads to an increase in the plateau of resurrection (Figure 83). However, the observed increases are much more dramatic than what has previously been observed,232 and we further posit that this is due to the increase in electron density of the pyridinol core and the steric size resulting from substituents on the pyridinol ring. Both of these factors may then increase the inhibitory potency of the formed resurrection byproducts, thereby accounting for such a dramatic change in the observed plateaus. The ordering and percentage of resurrection are similar at both rhuAChE concentrations (Figure 83) with the notable differences being a switch between 12b and 11b in the maximum resurrection and an overall line shape change for 8d (with an EC50 at 10-fold dilution of 159 µM), further emphasizing its dramatic concentration dependence.

179

4.6 Computational Investigation of the pH Dependence and Thermodynamics for QM Formation

We were first interested in understanding if different substituents favored different protonation states of active vs inactive compounds. Thus, as considered in our previous study of 4,232 several geometries of the different protonation states of 13 of the active and inactive compounds (6a-6e, 8b-8d, 11b, 11d, 12a, 12b and 12e) in the current study were optimized using the B3LYP/6-31+G(d) density functional theory method188,190,237,238 with consideration of implicit solvation (SMD,239 water) using

Gaussian 16.192

As shown in Figure 84 for compound 6d, as an example, we considered the net neutral states for each QMP which include the non-zwitterionic neutral isomer, the zwitterionic aminium (Z-Am) and the zwitterionic pyridinium (Z-Pyr) species. For the net positive (+1) states, we considered the aminium (Am), the pyridinium (Pyr) and the zwitterionic pyridinium-aminium (Z-Pyr-Am) species. For the compounds containing an

(R)-2-methylpyrrolidine group (d), such as 6d, 8d and 11d, both the syn and anti conformations of the zwitterionic aminium (Z-Am), the aminium (Am) and the zwitterionic pyridinium-aminium (Z-Pyr-Am) species were considered. The geometry of the lowest energy conformer of each charge state was then selected for comparisons of the relative enthalpies and free energies at various levels of theory. We also computed the free energy (∆G298) and the enthalpy (∆H298) of the proposed QM formation reactions

(Figure 80A) from the various protonation states.

180

N N H H N N N N OH O O

6d Neu 6d Z-Am 6d Z-Pyr (syn or anti) (syn or anti)

H N H N H N N N H N OH OH O

6d Am 6d Pyr 6d Z-Pyr-Am (syn or anti) (syn or anti)

Figure 84. Examples of the protonation states analyzed for the formation of the QM intermediate using compound 6d as a representative molecule.

A comparison between the relative free energy (∆G298) of the net neutral states and the net positive (+1) states, and between the QM formation free energy (∆G298) of the

13 active and inactive compounds was carried out (Tables 8 and 9). For compounds 8d and 12b, we also evaluated different solvation methods (SMD239 and PCM240) along with multiple basis sets241,242 with B3LYP and M06-2X237,238,243 functionals as well as the

CBS-QB3 method.244

181

Table 8. Relative free energy (ΔG298, kcal/mol) of the net neutral states at the B3LYP(SMD,water)/6-31+G* level of theory.a

Compound State Species 6a 6b 6c 6d 6e 8b 8c 8d 11b 11d 12a 12b 12e

Neu 0.42 1.28 0.71 0.64 1.06 1.17 0.71 0.49 1.89 1.13 1.26 2.36 2.09

Z-Am 0.00 0.00 0.00 0.00b 0.00 0.00 0.00 0.00b 0.00 0.00b 0.00 0.00 0.00

Net Neutral Net Z-Pyr 2.08 3.82 2.61 2.66 3.08 3.00 1.65 2.37 4.31 3.47 4.34 5.41 5.33 a See Scheme 1 for representative structures of each protonation state. b The syn conformation for the protonated (R)-2-methylpyrrolidine was preferred over the anti conformation.

For the net neutral states, the zwitterionic aminium (Z-Am) species, with the syn conformation of the protonated (R)-2-methylpyrrolidine derivatives, is the most favored species, in general, regardless of the type of the substituent on the pyridine ring and the amine leaving group (Table 8). There are two dominant forms of the net protonated state of the different QMPs: the aminium (Am) and the zwitterionic pyridinium-aminium (Z-

Pyr-Am) form (Figure 84, Table 9). In all cases, the difference in energy between favored states is ~ 1 kcal/mol. And, of course, this relative energy could depend on the environment in the active site.

182

Table 9. Relative free energy (ΔG298, kcal/mol) of the net positive charge states at the B3LYP(SMD,water)/6-31+G* level of theory.a

Compound State Species 6a 6b 6c 6d 6e 8b 8c 8d 11b 11d 12a 12b 12e

0.6 Am 0.88 0.78 1.01b 0.79 1.19 0.63 0.38b 0.06 0.00b 0.00 0.00 0.00 8 3.3 Pyr 4.23 3.65 3.99 3.95 4.35 3.90 2.84 4.24 4.55 4.86 6.37 5.77 5

Z-Pyr- 0.0 b b b Net Positive Net 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.55 0.93 1.09 0.66 Am 0 a See Scheme 1 for representative structures. b The syn conformation for the protonated (R)-2-methylpyrrolidine was preferred over the anti conformation.

Since the six forms shown in Figure 84 are predicted to be the more stable structures, we considered all of them in the molecular docking efforts discussed below.

4 (kcal/mol) G ∆ 2

-2

-4 6a 6b 6c 6d 6e 8b 8c 8d 11b 11d 12a 12b 12e Compound

Neutral Zamin Amin Zpyriamin (1)

Figure 85. At the B3LYP(SMD,water)/6-31+G* level of theory, the free energy (ΔG298) for formation of the quinone methide (QM) intermediate in (kcal/mol) from the two most stable/favorable net neutral states: non-zwitterionic neutral and zwitterionic aminium (Z- Am) species, and the two most stable/favorable net positive states: aminium (Am) and zwitterionic pyridinium-aminium (Z-Pyr-Am) species. 183

The quinone methide (QM) formation calculations for the different protonation states reveal that the diethylamine leaving group (compounds 6b, 8b, 11b and 12b) greatly decreases the energy to form the proposed QM intermediate (Figure 80A) as compared to the dimethylamine, pyrrolidine, (R)-2-methylpyrrolidine and piperidine leaving group. The 5-methoxy substituent (compounds 12a, 12b and 12e) significantly decreases both the ∆G298 and ∆H298 of QM formation compared to the 5-methyl

(compounds 11b and 11d) and 6-methyl (compounds 8b-8d) groups. Thus, compound

12b, with both the diethylamine and the 5-methoxy groups, displays the lowest ∆G298 and

∆H298 of QM formation for all of its different species as compared to the other 12 compounds (Figure 85). And, 12b is one of our most effective compounds, except at pH

4.7 Important Interactions Leading to Resurrection Efficacy

Using an in silico model of methylphosphonate-aged human AChE derived from the 5FPQ crystal structure prepared in a previous study,232 the active site of hAChE was analyzed for interactions between the phosphylated catalytic serine, nearby residues, and as shown in Scheme 1, the docked net neutral, zwitterionic neutral (Z-Am), net protonated (Am), and zwitterionic protonated (Z-Pyr-Am) states of 6b, 11b, 8b, and

12b.245 The optimized structures for each of these protonation states of the QMPs were docked into the active site with Autodock Vina 1.1.2.246 The docking poses were extracted and prepared for use in molecular dynamics (MD) simulations with AMBER

18.247 MD production trajectories were generated for a 10 ns time period. Distances were extracted and plotted in color coded distance maps.

184

Table 10. Center of mass distance (Å) between QMP and phosphylated-Ser averaged over the 10 ns MD trajectories and percentage of time that the benzylic carbon – oxyanion distance is less than 4 Å (3 Å in parentheses) for single trajectories with the greatest time spent under 3 Å for each protonation state.a

Compound Neu Z-Am Am Z-Pyr-Am distance time distance time distance time distance time 6b 14.3 8 (0) 15.6 0 (0) 13.6 76 (3) 12.4 64 (0) 8b 15.0 0 (0) 15.2 0 (0) 11.9 76 (33) 13.0 76 (4) 11b 14.9 0 (0) 15.7 0 (0) 12.7 100 (36) 13.3 0 (0) 12b 15.0 2 (0) 14.8 65 (0) 13.0 39 (0) 14.4 62 (0) a See Figure X for representative structures of each protonation state.

Considering the number of docked poses and for multiple conformations, selected interatomic distance data were extracted, plotted, and analyzed for the approximately 300

MD trajectories that were each run for 10 ns. Longer simulations were previously conducted to ensure that a 10 ns production was sufficient to account for the dominant features of the protein dynamics.232

The results make clear that the net protonated state populated a more reactive conformation in the active site than the other protonation states. For the compounds shown to be active in vitro (11b, 8b, 12b), the smallest average distance, between the center-of-mass (COM) of the QMP and the phosphylated serine residue, is 12.7, 11.0 and

13.0 Å (Table 10), respectively, which are all exhibited by the net protonated state (Am).

For the inactive 6b, the smallest average is 12.4 Å and is found for the zwitterionic protonated state (Z-Pyr-Am). These data suggest that the net protonated state of the QMP plays a dominant role in facilitating the binding to a reactive distance. Furthermore, the relatively high values of all of these COM averages (12–16 Å) reflects the observation that for the majority of trajectories the ligand does not come in close proximity to the 185 phosphylated serine. When analyzing the percentage of time that the re-alkylation distance is less than 4 Å, a similar trend is found in which the net protonated species

(Am) of both the inactive (6b), and the active (11b, 8b), display the greatest preference for interaction with the phosphylated serine. In particular, the active compounds (11b and

12b) spend 36% and 33% of time, respectively, with the re-alkylation distance under 3 Å.

In contrast, 12b exhibits a greater persistence of re-alkylation distance under 4 Å for the zwitterionic neutral (Z-Am) and zwitterionic protonated (Z-Pyr-Am). This incongruence, despite 12b being an active compound, may be explained by the different electronic properties with respect to quinone methide formation, as noted above.

In trying to understand the reactivity differences between the inactive (6b), and active (11b and 8b), the snapshots of the active site were compared to illustrate the differences and similarities in conformational poses. One major difference of note arises from the interaction between the amine leaving group and Trp86. For the inactive compound 6b, this interaction is largely absent. In contrast, the active compounds 11b and 8b display a significant and close interaction, often persisting into the 2–3 Å range.

The lack/presence of these interactions are illustrated in both the active site snapshots as well as distance maps. Additionally, it is observed that the 6b pose is oriented in the active site differently from 8b and 12b. Specifically, it appears that the phenolic hydrogen may be serving to position the pyridinol ring of 8b and 12b through a hydrogen bond with the phosphylated serine, while the same interaction is absent for 6b.

186

A) B)

C) Figure 86. Active site showing van der Waals surfaces of Tyr124 and Ser125 hydroxyl groups and methyl group in blue for (A) modified 11b (in orange) with addition of 4- methyl (11b*), (B) modified 8b with addition of 4-methyl (8b*) and (C) unmodified 6b. All images are taken from the same timepoints as the snapshots shown in in the supporting information.

To probe the feasibility of the 4-methyl adopting the same orientation, the 11b and 8b models that correspond to the snapshots of other compounds were modified by the

187 addition of a methyl group in the 4 position, yielding 11b* and 8b*. In these constructed models, van der Waals surfaces for the 4-methyl group (orange) are shown to overlap with nearby hydroxyl groups (blue) of Tyr124 and Ser125 (Figure 86). Thus, it appears that a 4-methyl substituted derivative would exhibit a drastic steric interaction in the active site by adopting the same conformation that is exhibited for the given snapshots of effective 11b and 8b. Furthermore, it is observed that the van der Waals overlap is avoided for 6b through the change in ligand orientation. This same steric interaction can account for the lack of activity amongst all 4-substituted compounds

The 5-methoxy compound, 12b, does not exhibit as persistent of a close realkylation interaction. However, there are some noteworthy observations. First, there is a strong interaction between the phenolic hydrogen and (anionic) Glu202, which may serve to both position the ligand as well as catalyze the quinone methide formation.

Interestingly, this interaction is also seen in another trajectory, specifically that the hydrogen-bond stability of the phosphylated serine by His447 is interrupted by the presence of 12b. This hydrogen-bond disruption can serve to destabilize the oxyanion of the phosphylated serine, rendering it more susceptible to reaction with an electrophile. In conjunction with this increased reactivity, it is also shown in the previous section that the

5-methoxy substituted QMPs exhibit more favorable quinone methide formation thermodynamics (Figure 85). The combination of these two factors may account for the efficacy of these 5-methoxy substituted compounds, despite the apparent decreased affinity for close proximity to the phosphylated serine.

188

4.8 Conclusions

We synthesized and tested a structure-activity library of 40 compounds and compared their resurrection capabilities with a methylphosphonate-aged recombinant human AChE to determine which properties lead to greater reaction efficacy. From a single-point, 24-hour resurrection screen, 20 various substituted compounds are capable of recovering the activity of aged acetylcholinesterase, with seven of these compounds being greater than 5% and being more effective than our previously reported lead compound, 4.232 Of the seven more active compounds all of them had electron-donating groups, either methyl- or methoxy-, attached at the 5- or 6-position of the 3-pyridinol ring. More electron-rich QMPs appear to be more effective at resurrecting the activity of aged acetylcholinesterase to the native state in the absence of any external reactivator.

The EC50 determination for the top compounds demonstrated the pH dependence of the methyl- and methoxy-substituted compounds. The EC50 values range over one order of magnitude from 86 – 730 µM. The features controlling the reactivity of the compounds were tested by increasing the pH from 7.5 to 9.0 resulting in increased resurrection for the 6-methyl substituted compounds (8) and decreased activity for the 5- methyl and 5-methoxy substituted compounds (11 and 12, respectively). Our current rationalization for this trend is that changes in the protonation state population affect binding and/or changes in the rate-determining step of the mechanism from reactivation to re-alkylation.

Additional determination of the IC50 values of the same top seven compounds with native rhuAChE revealed that their values lie between 404 – 1154 µM. This

189 confirms that the QMPs themselves are weak inhibitors of native AChE, an advantage in the development of resurrection therapeutics as further inhibition of the recovered enzyme activity would be detrimental. In five out of the seven compounds the EC50 values determined are smaller than that of the IC50 values, suggesting significant therapeutic opportunity.

We tested the effect of the [QMP]:[AChE] ratio of the plateau of recovery.

Notably, we observed an increase in the plateau for recovery, reaching very high percentages of recovery of greater than 60% at 24 hours in some cases. These results then further confirm the ability of QMPs to effectively resurrect AChE, but the maximum percentage is limited by the re-inhibition of the generated resurrection byproduct, an issue that requires further investigation.

To further determine the driving force for this increase in activity, we turned to computational methods to determine the effects that substituents had on the binding, protonation states, and thermodynamics of QM formation. The thermodynamic calculations show that the preferred protonation state for the neutral compounds is zwitterionic with the proton on the amine (Z-Am), and no matter the substitution. In the case of the net positive structures, those with 6- and 4-substituents (6 and 8) prefer to be zwitterionic with both the amine and pyridine being protonated (Z-Pyr-Am), although 5- substituted compounds prefer to have the amine protonated (Am), i.e., 11 and 12, with the exception of 11b which is nearly identical in stability. Investigation of the thermodynamics of QM formation for these protonation states show that the net positive isomers form the QM intermediate more favorably in comparison to the neutral

190 structures. Substitution on the ring favors electron-donating groups at the 5-position and the most energetically preferred has a diethylamine leaving group.

From the molecular dynamics studies, the dominant protonation state has affinity for the active site as the net protonated state (Am). Considering that the native enzyme hydrolyzes the permanently charged acetylcholine and that the CMP-aged form has an oxyanion, such a preference for charged substrates seems reasonable. Furthermore, the 4- substituted compounds may be largely inactive due to their inability to access a specific conformation. This inability appears to result from a steric interaction that would exist between the 4-substituent and the hydroxyl group of nearby residues. Finally, the 5- methoxy substituted compounds may owe more to their reactivity from electronic effects, rather than favorable binding in the active site. In addition to enhanced intrinsic reactivity, the 5-methoxy substituted QMPs may increase the reactivity of the phosphylated serine through the destabilization of a significant hydrogen bond.

Overall, the results of this study suggest that substitution at the 4-position of the aromatic ring has a negative effect on resurrection activity, as can be explained by the steric effects observed in the MD simulations. Substituents at the 5- and 6- positions are well tolerated, and the overall resurrection potency is increased when substituted with electron-rich substituents at those positions, believed to be due to the better thermodynamics of reaction. Interestingly, the methoxy-substituted compounds are effective at resurrection but are predicted to have a lower affinity for the enzyme active site based on the MD simulations. These contradicting results emphasize the complexity

191 of therapeutic development as both factors effect efficacy; however, the importance of each can vary based on the substitution of the compound.

Beyond these 20 active compounds, these results may allow for more directed synthesis towards even more efficient resurrectors of aged human AChE. Those efforts are on-going.

4.9 Experimentals

Computational Procedures

The following describes, in general, the methodology for the computational work.

The geometries of the different protonation states of all of the molecules were manually generated using the starting geometries for similar compounds from our previous publication232 and further optimized using density functional theory (DFT).232

Vibrational frequency analyses were performed to confirm the stationary points as minima on the potential energy surface as well as to derive thermodynamic corrections to the enthalpy and free energy. Furthermore, relative energies were compared within each set of stoichiometrically equivalent protonation states.

A human AChE crystal structure (PDB ID: 5FPQ) was modified in silico to produce a CMP-aged structure of AChE. Protonation states net neutral (Neu), zwitterionic neutral (Z-Am), net protonated (Amin), and zwitterionic protonated (Z-Pyr-

Am) were docked into these prepared receptors. Molecular dynamics (MD) trajectories were produced by using the docked poses as initial geometries. From the resulting trajectories, several interatomic distances were extracted and plotted for each time point.

192

Synthesis of QMPs

All of the QMPs were synthesized via a Mannich reaction using commercially available starting materials. Family 11 was synthesized by a protection, Stille coupling, deprotection, and Mannich reaction sequence. Detailed synthesis, purification, and characterization of all of the included compounds can be found appendix A

Enzymatic resolution of CMP

The racemic nerve agent analogue, CMP, was enzymatically resolved following a modified literature procedure from Ashani and coworkers.248 CMP (46.2 mg) was dissolved in 7.5 mL of MeOH, and the resulting solution was added dropwise to 100 mL of 0.1 mM 3B3 PON1 in pH 8.0 50 mM Tris buffer with 1 mM CaCl2. An aliquot of the solution was taken and loss of the coumarin leaving group was monitored by absorbance using a Molecular Devices SpectraMax i3 microplate reader at 400 nm. After 45 minutes the absorbance had plateaued and the reaction was determined to be complete. The reaction was quenched by adding 20 g of NaCl before extracting with dichloromethane

(DCM) (3x35 mL). The organic layer was further washed with saturated potassium carbonate (3x35 mL) before drying over sodium sulfate. The organic layer was then removed and the isolated solid (13.8 mg) was used to prepare all CMP solutions.

Aging of AChE by OP nerve agent analogues

193

2 µL of 7.94 g/L recombinant human AChE (69,586 U) was mixed with 96 µL of

1 g/L bovine serum albumin (BSA) in 200 mM sodium citrate buffer, pH 6.0, containing

0.02% NaN3 and 2 µL of 10 mM 3B3 PON1 resolved and extracted CMP (Figure 1). The mixed solution was reacted at 37°C for 24 h. The sample was then washed with 3 × 400

µL of 200 mM sodium phosphate buffer (pH 7.5) using 30 kDa Amicon centrifugal ultrafilters. Recovered protein concentrate (~50 µL) was mixed with 45 µL phosphate buffer and 5 µL of 10 mM 2-PAM. The solution was further incubated for 1-2 h. The sample was then washed with 3 × 400 µL of 200 mM sodium phosphate buffer (pH 7.5) using 30 kDa Amicon centrifugal ultrafilters. Recovered protein concentrate (~50 µL) was mixed with 50 µL of 200 mM pH 7.5 phosphate buffer. A single-point aging check was conducted using Ellman’s assay to ensure the protein was completely aged, if so the sample was used in the following experiments. A positive control was prepared in parallel by replacing the OP solution with blank acetonitrile.

Determining resurrection of aged rhuAChE with QMPs

1 µL of CMP-aged rhuAChE (see above) was mixed with 89 µL of 1 g/L BSA in

200 mM sodium phosphate buffer, pH 7.5, containing 0.02% NaN3 and 10 µL of QMP

(10 mM), 2-PAM (10 mM and 1 mM for 2-PAM control), or blank water, for negative control. The positive control was prepared in parallel with native AChE, BSA and blank water. The mixed solutions were reacted at 37 °C for 24 h. Following incubation the samples were diluted by 10-fold with 200 mM pH 7.5 sodium phosphate buffer prior to their use in Ellman’s assay.

194

EC50 of compounds with CMP-aged rhuAChE

10 µL of QMP solution at eight concentrations, 0–50 mM, was added to 89 µL of

1 g/L BSA 200 mM sodium phosphate buffer at pH 7.5 containing 0.02% NaN3, and mixed with 1 µL of aged rhuAChE solution (see above). The final concentrations were 0–

5 mM. A positive control was prepared in parallel with the aged rhuAChE being replaced with native rhuAChE solution and the QMP solution being replaced with blank water.

The mixed solutions were reacted at 37 °C for 24 h. The protein was then washed with 3

× 400 µL of pH 7.5 200 mM sodium phosphate buffer using 30 kDa AcroPrep Advance centrifugal filter plates. Retrieved protein concentrate was diluted to 200 µL prior to

Ellman’s assay with 200 mM pH 7.5 phosphate buffer. The positive controls were further diluted by 10-fold prior to use in Ellman’s assay with 200 mM pH 7.5 phosphate buffer.

Data were analyzed using Prism 7 with a best-fit comparison between a three- and four- parameter fits using the dose-response [agonist] vs response feature.

IC50 of compounds

1.5 µL of 7.94 g/L rhuAChE (69586 U) was diluted to 300000 µL with pH 7.5

200 mM phosphate buffer. Varying concentrations of QMP were prepared between 0 and

100 mM. To a 96-well plate was added 134 µL of 40 mM sodium phosphate buffer (pH

7.5), 18 µL of 1 g/L BSA, in the same buffer containing 0.02% NaN3, 10 µL of 10 mM acetylthiocholine and 10 µL of 20 mM 5,5’-dithio-bis(2-nitrobenzoic acid) (DTNB), and

195

8 µL of the 0-100 mM QMP solutions. To this was then added 20 µL of the diluted native rhuAChE stock. The absorption at 412 nm was monitored for 10.5 min with a Molecular

Devices SpectraMax i3 microplate reader. Four replicates were measured. The average slope and standard deviation of the initial pseudo-linear phase of the absorbance-time plot was divided by the average slope of the corresponding positive control for relative activity and relative standard deviation. Data were analyzed using Prism 7 with a best-fit comparison between a three- and four-parameter fit using the dose-response [inhibitor] vs response feature.

EC50 of compounds at different rhuAChE concentrations

10 µL of compound QMP solution at eight concentrations, 0–50 mM, was added to 89 µL of 1 g/L BSA 200 mM sodium phosphate buffer at pH 7.5 containing 0.02%

NaN3, and mixed with 1 µL of aged rhuAChE solution which was diluted by a factor of 5 or 10 with 200 µL of 200 mM pH 7.5 phosphate buffer previously. The final concentrations were 0–5 mM. A positive control was prepared in parallel with the aged rhuAChE replaced with native rhuAChE solution, diluted just as the aged sample was) and the QMP solution replaced with blank water. The mixed solutions were reacted at 37

°C for 24 h. The protein was then washed with 3 × 400 µL of pH 7.5 200 mM sodium phosphate buffer using 30 kDa Amicon centrifugal ultrafilters. Retrieved protein concentrate was diluted to 200 µL prior to Ellman’s assay. The positive controls were further diluted by 10-fold prior to use in Ellman’s assay. Data were analyzed using Prism

196

7 with a best-fit comparison between a three- and four-parameter fit using the dose- response [agonist] vs response feature.

Ellman’s Assay

20 µL of rhuAChE sample was added to a well of a 96-well microplate and mixed with 142 µL of 40 mM sodium phosphate buffer (pH 7.5), 18 µL of 1 g/L BSA, in the same buffer containing 0.02% NaN3, 10 µL of 10 mM acetylthiocholine and 10 µL of 20 mM 5,5’-dithio-bis(2-nitrobenzoic acid) (DTNB). 1 equivalent of Na3PO4 was added to aid in the dissolution of DTNB. The absorption at 412 nm was monitored for 10.5 min with a Molecular Devices SpectraMax i3 microplate reader. Four replicates were measured. The average slope and standard deviation of the initial pseudo-linear phase of the absorbance-time plot was divided by the average slope of the corresponding positive control for relative activity and relative standard deviation.

197

Chapter 5. SUBSTITUTED 3-PYRIDINOL MANNICH BASES FOR THE RESURRECTION AND REACTIVATION OF HUMAN ACETYLCHOLINESTERASE

The research presented in this chapter is adapted from a manuscript that will be submitted to the Journal of the American Chemical Society. The ADME-Tox work was conducted by Ms. Rachel Hopper. The synthesis was a collaborative effort between a number of undergraduate students with guidance and participation of Dr. Christopher Callam and myself.

5.1 Introduction

Organophosphorus (OP) compounds are toxic molecules that exhibit their toxicity through the inhibition or deactivation of the enzyme acetylcholinesterase (AChE). There are two main types of OP compounds (Figure 87A): chemical nerve agents (generally, methylphosphonate structures) and OP pesticides (phosphates with widely varying alkoxy or phenoxy substituents). The media lends much attention toward the use of OP nerve agents in armed conflicts and terrorist attacks, yet OP pesticides are a much

198 broader toxicological problem, given their use in agricultural applications that can lead to exposure.201,249 A more troubling problem is the issue of intentional ingestion for self– harm or suicide, especially in rural Asia. It is estimated that 200,000 deaths occur annually as the result of direct ingestion of OP pesticides, this accounts for 60% of all self-harm cases in the region.201 There is the additional problem of degradation products from other common sources such as jet fuels. Recently, the metabolite CBDP, an organophosphate, has been cited as the potential cause of what has come to be known as aerotoxic syndrome in airline professionals and maintenance workers.18,19

All OP compounds, whether nerve agent or pesticide, inhibit the function of

AChE. AChE is an enzyme present in both the peripheral as well as the central nervous systems where the enzyme hydrolyzes the neurotransmitter acetylcholine (ACh).165,250 OP inhibition of AChE then leads to accumulation of ACh in the synaptic clefts resulting in a cholinergic crisis, that without medical intervention, leads to death. The current standard of care for OP exposure consists of a nucleophilic oxime (e.g., 2-PAM), atropine, and diazepam. However, the current standard of care is still limited in its effectiveness, failing to address three critical issues. The first of which is that currently approved oximes are not broad scope reactivators of all inhibited forms of AChE; second, currently utilized oximes are permanently positively charged and have limited blood-brain barrier penetrability; and finally, oximes are ineffective at recovering AChE from what is referred to as the aged state.35,90,250

199

A O O P P LG OR LG OR2 3 OR1

generic methylphosphonate B generic phosphate

O N 2 O O O P P S P O O F O S O O O O

Ethyl Paraoxon DFP Demeton

O N Cl Cl O O 2 O O O O P P P O O Cl N O O O S O O O O

Methyl Paraoxon Chlorpyrifos Oxon Malaoxon

Figure 87. (A) Structure of generic methylphosphonate OPs, accounting for most of the G- and V-series nerve agents and the generic structure of an organophosphate OP accounting for the pesticides. (B) Structures of various OP pesticides.

AChE’s aged form is the result of a spontaneous dealkylation reaction that generates, for organophosphate exposures, a phosphate anion in the enzyme’s active site.

This negatively charged, phosphylated serine residue is then resistant to nucleophilic oxime treatment as a result of charge-charge repulsions as well as conformational changes adopted by the charged residue, namely hydrogen-bonding to the neighboring histidinium residue forming a very strong salt bridge.38,153,164,251,252 Pesticides vary greatly in their aging half-lives based on their leaving groups, and alkoxy substituents. Herein we choose to study methyl paraoxon which has a rapid aging half-life, 3.7 h,35 and DFP due to it slow rate of spontaneous reaction and fast aging.81,88

Our research group has been able to demonstrate, using multiple compounds, that pyridinol-based quinone methide precursors (QMPs) are capable of resurrecting147 OP- aged AChE, as derived from human160,232,253 (recombinant and erythrocyte ghost 200 isoforms) as well as electric eel sources.160 The most notable point about these therapeutic QMPs is their ability to convert the aged state to the native state of the enzyme, without the use of any other external reactivators, such as an oxime. We have demonstrated that not only are these QMPs capable of the resurrection process but they are also capable of reactivating the OP-inhibited, akin to a re-alkylated, form of AChE, albeit at a slow rate.160 This reactivation propensity of Mannich bases has been noted by a number of other groups in the past few years, resulting in new and effective non-oxime based reactivators of OP-inhibited AChE (Figure 79).91,113,116,231

We recently focused on analyzing the structure-activity relationships present for the resurrection of methylphosphonate aged-AChE and found some compounds in this library that were particularly effective at recovering activity from the methylphosphonate-aged form of recombinant human AChE.253 Noting the structural differences in the aged states of methylphosphonate-aged AChE and pesticide-aged

AChE, we sought to analyze the same structure-activity library for the reactivation and resurrection of pesticide-inhibited and pesticide-aged AChE. We hypothesize that the trends observed for resurrection efficiency will not follow with those observed for reactivation, given the different roles that the QMPs are serving in each step (i.e., electrophile versus nucleophile) and the structural and charged differences between the inhibited and aged states. Additionally, given the promising results observed for resurrection, we wanted to begin analyzing the pharmacokinetic (ADME-Tox) properties of these QMPs as we continue to try and advance to in vivo studies. The results of these studies are reported herein.

201 5.2 Resurrection of rhuAChE After Aging With DFP

We screened the previously reported structure-activity library253 (Figure 88) to determine their resurrection efficiencies for DFP-aged recombinant human AChE

(rhuAChE), as generously provided by Dr. Zoran Radić (UC-San Diego). DFP (Figure

87B) is a symmetric phosphate that, upon inhibition of AChE, does not result in diastereomers; thus, there was no need to conduct an enzymatic resolution of the OP prior to the aging process, as was done in our previous study with methylphosphonates.253 DFP has a fluoride leaving group, similar to that of most G-series nerve agents. The OP and rhuAChE were incubated together for 24 hours at 37 °C prior to ultrafiltration. The known reactivator 2-PAM was then used in a single-point verification to ensure complete aging. The activity of the aged rhuAChE and positive control were then analyzed by

Ellman’s assay.221 If the enzyme was determined to be completely aged and the positive control stable, then the QMP compounds were then added, reaching a final concentration in the incubation vessels of 1 mM. We remind the reader that from this point on, no additional external reactivator was used. With DFP, the samples were then incubated for

24 h at 37 °C before the samples were diluted by a factor of 100, in an effort to ensure that the AChE activity would be within a reasonable window for detection using

Ellman’s assay as well as to dilute the QMP to a concentration (< 10 µM) that would not interfere with any regenerated native enzyme. All of the recovered enzyme activity was compared versus a positive control that was never treated with the OP and water being used in place of QMP.

202 100

40 % activity 20

0 5 6 7 8 9 - + 10 11 12 2-PAM

Figure 88. Resurrection activity of QMPs against a phosphate-aged (DFP) rhuAChE (37°C, pH 7.5, 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figure 80 for structures. The 2-PAM concentration was 1 mM. The error bars reflect a standard deviation of 4 replicate measurements. The results for the DFP resurrection show a clear preference for families 8 and 11, with a very clear preference for amine d, (R)-2-methylpyrrolidine. Both of these compounds have a methyl substituent at the 6- or 5- position, respectively. Unlike in our previous methylphosphonate study,253 resurrection of DFP-aged rhuAChE has an extremely clear preference for amine d. All other compounds in the library perform quite modestly in comparison resurrecting ~ 8% in most cases, with the exception of the 5 and

6 families which are completely inactive. The inactivity of 5 and 6 was also observed in our methylphosphonate study and further confirms that 4-substituents are not tolerated in the resurrection process, probably as a result of steric interaction in the enzyme active site that would prevent orientation for a productive reaction.253 The compounds 8d and 11d were also leads in the previous study with methylphosphonate-aged rhuAChE and this suggests that these compounds may be broad scope resurrectors, although additional

203 testing is required. Here, the maxima of resurrection is significantly higher than what has been previously observed at 51% and 45% for 8d and 11d, respectively, after 24 h. In the case of pesticides, these QMP derivatives are well suited for the resurrection reaction, with even those that are performing poorly with a pesticide still recovering approximately the same amount of activity of our top compounds in the previous methylphosphonate study.253

5.3 Resurrection of rhuAChE After Aging With Methyl Paraoxon

Seeing as how the results for the resurrection of DFP differed dramatically from the previously tested methylphosphonate-aged AChE, we sought to attempt resurrection with an additional pesticide, methyl paraoxon. Like DFP, methyl paraoxon is a symmetric OP pesticide, which has a p-nitrophenol leaving group rather than a fluoride leaving group (Figure 87B). The procedure for the study was the exact same as that conducted for DFP, and made use of the same library of compounds.

100

50 % activity

0 5 6 7 8 9 - + 10 11 12 2-PAM

Figure 89. Resurrection activity of QMPs against a phosphate-aged (methyl paraoxon) rhuAChE (37°C, pH 7.5, 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP (5–12) was used at a concentration of 1 mM, and the 5 amines in each family are listed a-e from left to right – see Figure 80 for structures. The

204 2-PAM concentration was 1 mM. The error bars reflect a standard deviation of 4 replicate measurements. The results for the resurrection of methyl paraoxon-aged rhuAChE are extremely promising, with multiple compounds exceeding the 50% mark for resurrected activity in

24 h and with the best compound (8b) reaching a maximum of 85%. For methyl paraoxon, the selectivity for amine d and compound 8 and 11, specifically, are not present. Rather, there is good conversion for many members of the 8–12 families. This is interesting as both electron-donating (methyl- and methoxy--) as well as electron- withdrawing (bromo- and chloro-) facilitate resurrection. Even so, the best conversions are for those with electron-donating groups: specifically, 8 (6-Me), 11 (5-Me), and 12 (5-

OMe).

Compared to the DFP results, compounds from family 12 (5-OMe) outperform family 11 (5-Me). It is also worth noting that the selectivity for the amine changes based on the group attached: b and d for 8 and 11 but a, b, and e for the 12 family. This difference in amine selectivity was observed in our methylphosphonate study where we investigated this difference in selectivity using computational methods, and we determined that a different binding mode of 12 accounts for this difference in amine slectivity.253 Once again, for methyl paraoxon, the percentage of resurrection is much higher (>50%) than was previously observed for methylphosphonate-aged AChE, suggesting our QMP therapeutics are more efficacious for treatment of pesticide-aged

AChE, although additional testing is required and ongoing.

205 5.4 pH and Time Dependence on the Resurrection of Methyl Paraoxon-Aged rhuAChE

In our previous studies, we have noted the dependence of resurrection on the time of reaction as well as the pH of the incubation media. Previously, it was observed that for the 8 family, a pH of 9.0 increased the observed resurrection maxima, as we hypothesized was due to the hydrolysis of a byproduct of reaction that is capable of re-inhibiting native

AChE.232 We decided to test this pH dependence on the methyl paraoxon-aged rhuAChE and included three timepoints (1, 3, and 24 h) in our study in order to prove that our compounds are effective in a more relevant therapeutic window of time. We analyzed all of the compounds using this methodology but only the top seven compounds (8b, 8d,

11b, 11d, 12a, 12b, 12e) are reported here.

A pH 7.5 1 h B pH 9.0 1 h 3 h 3 h 100 24 h 100 24 h

50 50

0 0 - + - + 8b 8d 8b 8d 11b 11d 12a 12b 12e 11b 11d 12a 12b 12e

Figure 90. Resurrection activity of QMPs (1 mM) against a phosphate-aged (methyl paraoxon) rhuAChE (37°C, pH 7.5 (A) and pH 9.0 (B), at 1, 3, and 24 h). The rhuAChE was diluted to a final activity of 0.14 U in the Ellman’s assay. Each QMP was used at a concentration of 1 mM – see Figure 80 for structures. The error bars reflect a standard deviation of 4 replicate measurements.

The results show that unlike methylphosphonate-aged AChE, there is not as significant of a pH dependence; however, in general, a higher pH increases resurrection.

Interestingly, compared to methylphosphonate-aged AChE, all of the compounds 206 increased in conversion at pH 9.0 after 24 h, with the best compound (8d) surpassing

100%, potentially due to allosteric modulation.254 The rates of each transformation seem to be relatively similar with the exception of the 8 compounds where there appears to be an overall rate acceleration, based on relative percentages. Notably after one hour, we still observe a significant amount of aged-AChE recovery, at approximately 10% for all compounds shown. This high of a conversion over a smaller time period then begins to approach numbers that may be valuable for an in vivo therapeutic, but of course, the concentration needs to be reduced. We therefore evaluated the effective concentration for resurrection.

5.5 ADME-Tox Determination Using Computational Techniques

Two computational interfaces (SwissADME255 and ACD/Percepta256) were used as a screening method for the pharmacokinetic properties of the investigated therapeutics and we will summarize the results. The pharmacological criteria that were considered in this analysis included blood-brain barrier (BBB) permeability, plasma protein binding

(PPB), solubility and cytochrome P450 interactions.

Comparison of the predicted BBB permeability of both programs are in good agreeance with one another, and only with the exception of 9a. Compound 9a is listed as not penetrant according to ACD/Percepta, while SwissADME indicates that it can permeate the BBB. All other compounds are predicted to permeate the BBB with both programs, a particularly promising result given the difficulty of current oxime therapeutics at crossing the BBB. Also, only ACD/Percepta predicts plasma protein binding (PPB), and scaffolds 8, 11, 12 (methyl- or methoxy- substituents) have much lower percentages of PPB and with greater reliability of the calculations. ACD/Percepta 207 calculated the probability of these compounds being human ether-a-go-go (hERG) inhibitors, and all were indicated as non-hERG inhibitors.

The solubility determination differs between the two software packages.

SwissADME offers three different solubility models, each being variable within the

SwissADME package. There seems to be a loose trend consistent among the different solubility models in that the highest solubility values are for methoxy-substituted compounds, which may be expected as they have an additional hydrogen-bond accepting substituent. The solubility for the rest of the compounds decreases from methyl- > chloro- > bromo-. While the values are skewed among the different models, the trend appears to be relatively consistent.

Both software packages differ significantly in regards to CYP450 inhibition.

SwissADME indicates that only CYP1A2 will be inhibited by the tested compounds, contrary to ACD/Percepta, which indicates that almost all of the compound are non- inhibitors, with the exception of a few undefined cases (Br substituent). It appears that the in silico evaluation suggests that compounds in the 8, 11 and 12 families will have good drug-like properties.

5.6 Experimentals

Biological Assays

To assist in preparing samples for the biochemical assays an Integra-Biosciences

Assist Plus automated pipetting system was utilized. This system uses an Integra-

Biosciences Voyager electronic pipette.

Aging of AChE by OP pesticides 208 2 µL of 7.94 g/L recombinant human AChE (69,586 U) was mixed with 96 µL of

1 g/L bovine serum albumin (BSA) in 200 mM sodium citrate buffer, pH 6.0, containing

0.02% NaN3 and 2 µL of 10 mM DFP or methyl paraoxon in acetonitrile (Figure 1). The mixed solution was reacted at 37°C for 24 h. The sample was then washed with 3 × 400

µL of 200 mM sodium phosphate buffer (pH 7.5) using 30 kDa Amicon centrifugal ultrafilters. Recovered protein concentrate (~50 µL) was mixed with 45 µL phosphate buffer and 5 µL of 10 mM 2-PAM. The solution was further incubated for 1-2 h. The sample was then washed with 3 × 400 µL of 200 mM sodium phosphate buffer (pH 7.5) using 30 kDa Amicon centrifugal ultrafilters. Recovered protein concentrate (~50 µL) was mixed with 50 µL of 200 mM pH 7.5 phosphate buffer. A single point aging check was conducted using Ellman’s assay to ensure the protein was completely aged, if so the sample was used in the following experiments. A positive control was prepared in parallel by replacing the OP solution with blank acetonitrile. In the case of larger screens the ratio was lept constant and the volumes increased by a factor of 2.

Determining resurrection of aged rhuAChE with QMPs

1 µL of CMP-aged rhuAChE (see above) was mixed with 89 µL of 1 g/L BSA in

200 mM pH 7.5 sodium phosphate buffer or pH 9.0 200 mM ammonia buffer, containing

0.02% NaN3 and 10 µL of QMP (10 mM), or blank water, for the negative control. The positive control was prepared in parallel with native AChE, BSA containing buffer matching what was used for the QMP samples and blank water. The mixed solutions

209 were reacted at 37 °C for 1-24 h. Following incubation the samples were diluted by 10- fold with 200 mM pH 7.5 sodium phosphate buffer prior to their use in Ellman’s assay.

Ellman’s Assay

210 Bibliography

(1) Everts, S. The Nazi Origins of Deadly Nerve Gases. Chem. Eng. News 2016, 26– 28. (2) Christianson, S. Fatal Airs. In Fatal Airs: The Deadly History and Apocalyptic Future of Lethal Gases That Threaten Our World; Praeger, 2010; pp 73–76. (3) Coulston Gillispie, C.; Lawrence Holmes, F.; Koertge, N.; Thomson Gale. Kuhn, R. Complete Dictionary of Scientific Biography; 2008. (4) Pruitt, S. The Nazis Developed Sarin Gas During WWII , But Hitler Was Afraid to Use It. History.com. 2017, pp 18–21. (5) Minelle, B. VX Nerve Agent : What Is It, Where Did It Come from and What Does It Do to You ? Sky News. February 25, 2017, pp 1–12. (6) Nachon, F.; Asojo, O. A.; Borgstahl, G. E. O.; Masson, P.; Lockridge, O. Role of Water in Aging of Human Butyrylcholinesterase Inhibited by Echothiophate: The Crystal Structure Suggests Two Alternative Mechanisms of Aging. Biochemistry 2005, 44 (4), 1154–1162. https://doi.org/10.1021/bi048238d. (7) Cotton, S. Handle with Care – the World’s Five Deadliest Poisons. The Conversation. April 12, 2016, pp 1–5. (8) VR ( nerve agent ) www.revolvy.com/page/VR-%28nerve-agent%29 (accessed Aug 1, 2018). (9) Cornell. Parathion http://pmep.cce.cornell.edu/profiles/extoxnet/metiram- propoxur/parathion-ext.html (accessed Aug 1, 2018). (10) Stone, R. Seeking Answers for Iran’s Chemical Weapons Victims — Before Time Runs Out. Science. January 2018. (11) Alfred, C. 20 Years Ago, A Shadowy Cult Poisoned The Tokyo Subway. Huffington Post. April 2015, pp 1–9. (12) Amarasingam, A. A History of Sarin as a Weapon. The Atlantic. April 5, 2017. (13) Syria Chemical ’attack’ : What We Know. BBC News. April 26, 2017, pp 24–26. (14) Sephton, C. Nerve Agent Classed as ’ Weapon of Mass Destruction ’ Killed Kim. Sky News. February 24, 2017. (15) Russian Spy: What We Know so Far. BBC News. April 10, 2018. (16) Reuters. Pesticide Found in Meals That Killed Indian Children, Official Says. New York Times. 2013. (17) Nilesh, V. Pesticide in Food 40 Times More than in US , UK. The New Indian Express. 2018, pp 24–26. (18) Carletti, E.; Schopfer, L. M.; Colletier, J. P.; Froment, M. T.; Nachon, F.; Weik, M.; Lockridge, O.; Masson, P. Reaction of Cresyl Saligenin Phosphate, the Organophosphorus Agent Implicated in Aerotoxic Syndrome, with Human

211 Cholinesterases: Mechanistic Studies Employing Kinetics, Mass Spectrometry, and X-Ray Structure Analysis. Chem. Res. Toxicol. 2011, 24 (6), 797–808. https://doi.org/10.1021/tx100447k. (19) Carletti, E.; Colletier, J. P.; Schopfer, L. M.; Santoni, G.; Masson, P.; Lockridge, O.; Nachon, F.; Weik, M. Inhibition Pathways of the Potent Organophosphate CBDP with Cholinesterases Revealed by X-Ray Crystallographic Snapshots and Mass Spectrometry. Chem. Res. Toxicol. 2013, 26 (2), 280–289. https://doi.org/10.1021/tx3004505. (20) Quinn, D. M. Acetylcholinesterase: Enzyme Structure, Reaction Dynamics, and Virtual Transition States. Chem. Rev. 1987, 87 (5), 955–979. https://doi.org/10.1021/cr00081a005. (21) Bajgar, J.; Kuca, K.; Fusek, J.; Karasova, J. Inhibition of Blood Cholinesterases by Nerve Agents in Vitro. Appl. Biomed. 2009, 7, 201–206. (22) Botos, I.; Wlodawer, A. The Expanding Diversity of Serine Hydrolases. Curr. Opin. Struct. Biol 2007, 17, 683–690. https://doi.org/10.1016/j.sbi.2007.08.003. (23) Sussman, J.; Harel, M.; Frolow, F.; Oefner, C.; Goldman, A.; Toker, L.; Silman, I. Atomic Structure of Acetylcholinesterase from Torpedo Californica: A Prototypic Acetylcholine-Binding Protein. Science (80-. ). 1991, 253 (5022), 872–879. (24) Kovarik, Z.; Bosak, A.; Latas, T. Exploring the Active Sites of Cholinesterases by Inhibition with Bambuterol and Haloxon. Croat. Chem. Acta 2003, 76 (1), 63–67. (25) Saxena, A.; Redman, A. M. G.; Jiang, X.; Lockridge, O.; Doctor, B. P. Differences in Active Site Gorge Dimensions of Cholinesterases Revealed by Binding of Inhibitors to Human Butyrylcholinesterase †. Biochemistry 1997, 36, 14642– 14651. (26) Tai, K.; Shen, T.; Bo, U.; Philippopoulos, M.; Mccammon, J. A. Analysis of a 10- Ns Molecular Dynamics Simulation of Mouse Acetylcholinesterase. Biophys. J. 2001, 81, 715–724. (27) Taylor, P.; Lwebuga-Mukasa, J.; Lappi, S.; Rademacher, J. Propidium-a Fluorescence Probe for a Peripheral Anionic Site on Acetyicholinesterase. Mol. Pharmacol. 1974, 10 (4), 703–708. (28) Changeux, J.-P. Responses of Acetylcholinesterase from to Salts and Curarizng Drugs. Mol. Pharmacol. 1966, 2, 369–392. (29) Ripoll, D. R.; Faerman, C. H.; Axelsen, P. H.; Silmanii, I.; Sussman, J. L. An Electrostatic Mechanism for Substrate Guidance down the Aromatic Gorge of Acetylcholinesterase. Proc. Natl. Acad. Sci. USA 1993, 90, 5128–5132. (30) Botti, S. A.; Felder, C. E.; Lifson, S.; Sussman, J. L.; Silman, I. A Modular Treatment of Molecular Traffic Through the Active Site of Cholinesterase. Biophys. J. 1999, 77, 2430–2450. (31) Gilson, M. K.; Straatsma, T. P.; McCammon, J. A.; Ripoll, D. R.; Faerman, C. H.; Axelsen, P. H.; Silman, I.; Sussman, J. L. Open “Back Door” in a Molecular Dynamics Simulation of Acetylcholinesterase. Science (80-. ). 1994, 263 (5151), 1276–1278. (32) Xu, Y.; Colletier, J.; Weik, M.; Qin, G.; Jiang, H.; Silman, I.; Sussman, J. L. Long Route or Shortcut ? A Molecular Dynamics Study of Traffic of Thiocholine within

212 the Active-Site Gorge of Acetylcholinesterase. Biophys. J. 2010, 99 (12), 4003– 4011. https://doi.org/10.1016/j.bpj.2010.10.047. (33) Colletier, J.; Royant, A.; Masson, P.; Zaccai, G.; Joel, L.; Goeldner, M.; Silman, I.; Weik, M. Use of a ‘ Caged ’ Analogue to Study the Traffic of Choline within Acetylcholinesterase by Kinetic Crystallography Research Papers. Acta Cryst. 2007, D63, 1115–1128. https://doi.org/10.1107/S0907444907044472. (34) Colletier, J.; Bourgeois, D.; Sanson, B.; Fournier, D.; Sussman, J. L.; Silman, I.; Weik, M. Shoot-and-Trap : Use of Specific x-Ray Damage Temperature- Controlled Cryo-Crystallography. Proc. Natl. Acad. Sci. USA 2008, 105, 11742– 11747. (35) Worek, F.; Thiermann, H. The Value of Novel Oximes for Treatment of Poisoning by Organophosphorus Compounds. Pharmacol. Ther. 2013, 139 (2), 249–259. https://doi.org/10.1016/j.pharmthera.2013.04.009. (36) Hobbiger, F. EFFECT OF NICOTINHYDROXAMIC ACID METHIODIDE ON HUMAN PLASMA CHOLINESTERASE INHIBITED BY ORGANOPHOSPHATES CONTAINING A DIALKYLPHOSPHATO GROUP. Br. J. Pharmacol. 1955, 10, 356–359. (37) Wilson, B. W.; Hooper, M. J.; Hansen, M. E.; Neiberg, P. S. Organophosphates, Chemistry, Fate and Effects; 1992. (38) Quinn, D. M. Resurrection Biology: Aged Acetylcholinesterase Brought Back to Life. J. Med. Chem. 2018, 61 (16), 7032–7033. https://doi.org/10.1021/acs.jmedchem.8b01122. (39) Wolthuis, O. L.; Kepner, L. A. Successful Oxime Therapy One Hour after Soman Intoxication in the Rat. Eur. J. Pharmacol. 1978, 49 (4), 415–425. https://doi.org/10.1016/0014-2999(78)90316-3. (40) Sidell, F. R.; Groff, W. A. The Reactivatibility of Cholinesterase Inhibited by VX and Sarin in Man. Toxicol. Appl. Pharmacol. 1974, 27, 241–252. (41) Mage, P. P. Multidimensional Pharmacochemistry; 1984. (42) Wilson, B. Acetylcholinesterase: Reversibility of Tetraethyl Pyrophosphate Inhibition. J. Bio. Chem. 1951, 190 (1), 111–118. (43) Wilson, I. B.; Meislich, E. K. Reactivation of Acetylcholinesterase Inhibited by Alkylphosphates. J. Am. Chem. Soc. 1953, 75 (1), 4628–4629. (44) Aldridge, W. N.; Davison, A. N. The Mechanism of Inhibition of Cholinesterases by Organophosphates. Biochem. J. 1953, 55, 763. (45) Hackley, B. E.; Plapinger, R.; Stolberg, M.; Wagner-Jauregg, T. Acceleration of the Hydrolysis of Organic Fluorophosphates and Fluorophosphonates with Hydroxamic Acids. J. Am. Chem. Soc. 1955, 77 (13), 3651–3653. (46) Wilsom, I. B.; Ginsburg, S. Reactivation of Acetylcholinesterase Inhibited by Alkylphosphates. Arch. Biochem. Biophys. 1955, 55, 569–571. (47) Jandorf, B. J.; Crowell, E. A.; Lebin, A. P. No Title. Fed. Proc. 1955, 14, 231. (48) Michel, H. O. No Title. Fed. Proc. 1955, 14, 255. (49) Childs, A. F.; Davies, D. R.; Green, A. L.; Rutland, J. P. THE REACTIVATION BY OXIMES AND HYDROXAMIC ACIDS OF CHOLINESTERASE INHIBITED BY ORGANO-PHOSPHORUS COMPOUNDS Activity of

213 Reactivated Enzyme-Activity of Inhibited Enzyme Activity of Enzyme Control - Activity of Inhibited Enzyme X1. Brit. J. Pharmacol. 1955, 10, 462–465. (50) Holmes, R.; Robins, E. L. The Reversal by Oximes of Neuromuscular Block Produced by Anticholinesterases. Brit. J. Pharmacol 1955, 10 (4), 490–495. https://doi.org/10.1111/j.1476-5381.1955.tb00110.x. (51) Dunn, M. A.; Sidell, F. R. Progress in Medical Defense Against Nerve Agents. J. Am. Med. Assoc. 1989, 262 (5), 649–652. https://doi.org/10.1001/jama.1989.03430050065028. (52) Clement, J. G. Efficacy of Various Oximes against GF (Cyclohexyl Methylphosphonofluoridate) Poisoning in Mice. Arch. Toxicol. 1992, 66, 143–144. (53) Dawson, R. M. Review of Oximes Available for Treatment of Nerve Agent Poisoning. J. Appl. Toxicol. 1994, 14 (5), 317–331. https://doi.org/10.1002/jat.2550140502. (54) Eddleston, M.; Szinicz, L.; Eyer, P.; Buckley, N. Oximes in Acute Organophosphate Pesticide Poisoning : A Systematic Review of Clinical Trials. Q. J. Med. 2006, 95 (5), 275–283. (55) Jokanović, M.; Stojiljković, M. P. Current Understanding of the Application of Pyridinium Oximes as Cholinesterase Reactivators in Treatment of Organophosphate Poisoning. Eur. J. Pharmacol. 2006, 553 (1–3), 10–17. https://doi.org/10.1016/j.ejphar.2006.09.054. (56) Ordentlich, A.; Barak, D.; Kronman, C.; Benschop, H. P.; De Jong, L. P. A.; Ariel, N.; Barak, R.; Segall, Y.; Velan, B.; Shafferman, A. Exploring the Active Center of Human Acetylcholinesterase with Stereomers of an Organophosphorus Inhibitor with Two Chiral Centers. Biochemistry 1999, 38 (10), 3055–3066. https://doi.org/10.1021/bi982261f. (57) Rosenstock, L.; Keifer, M.; Daniell, W. E.; McConnell, R.; Claypoole, K. Chronic Central Nervous System Effects of Acute Organophosphate Pesticide Intoxication. Lancet 1991, 338 (8761), 223–227. https://doi.org/10.1016/0140-6736(91)90356- T. (58) London, L.; Nell, V.; Thompson, M. Lou; Myers, J. E. Effects of Long-Term Organophosphate Exposures on Neurological Symptoms, Vibration Sense and Tremor among South African Farm Workers. Scand. J. Work. Environ. Heal. 1998, 24 (1), 18–29. https://doi.org/10.5271/sjweh.274. (59) Krejci, E.; Martinez-Pena y Valenzuela, I.; Ameziane, R.; Akaaboune, M. Acetylcholinesterase Dynamics at the Neuromuscular Junction of Live Animals. J. Biol. Chem. 2006, 281 (15), 10347–10354. https://doi.org/10.1074/jbc.M507502200. (60) Mendel, B.; Rudney, H. Studies on Cholinesterase: 1. Cholinesterase and Pseudo- Cholinesterase. Biochem. J. 1943, 37 (1), 59–63. (61) Nachon, F.; Masson, P.; Nicolet, Y.; Lockridge, O.; Fonticilla-Camps, J. C. Comparison of the Structures of Butyrylcholinesterase and Acetylcholinesterase. In Butyrylcholinesterase: Its functions and inhibitors; 2003; pp 39–54. (62) Bourne, Y.; Taylor, P.; Radić, Z.; Marchot, P. Structural Insights into Ligand Interactions at the Acetylcholinesterase Peripheral Anionic Site. EMBO J. 2003,

214 22 (1), 1–12. https://doi.org/10.1093/emboj/cdg005. (63) Dvir, H.; Silman, I.; Harel, M.; Rosenberry, T. L.; Sussman, J. L. Acetylcholinesterase: From 3D Structure to Function. Chem. Biol. Interact. 2010, 187 (1–3), 10–22. https://doi.org/10.1016/j.cbi.2010.01.042. (64) Boyko, K. M.; Baymukhametov, T. N.; Chesnokov, Y. M.; Hons, M.; Lushchekina, S. V; Konarev, P. V; Lipkin, A. V; Vasiliev, A. L.; Masson, P.; Popov, V. O.; et al. 3D Structure of the Natural Tetrameric Form of Human Butyrylcholinesterase as Revealed by CryoEM , SAXS and MD. Biochemie 2019, 156, 196–205. (65) Rosenberry, T. L. Catalysis by Acetylcholinesterase: Evidence That the Rate- Limiting Step for Acylation with Certain Substrates Precedes General Acid-Base Catalysis. Proc. Natl. Acad. Sci. U.S.A 1975, 72 (10), 3834–3838. https://doi.org/10.1073/pnas.72.10.3834. (66) Lockridge, O. Genetic Variants of Human Serum Cholinesterase Influence Metabolism of the Muscle Relaxant Succinylcholine. Pharmacol. Ther. 1990, 47 (1), 35–60. https://doi.org/0163-7258(90)90044-3 [pii]. (67) Kaplan, D.; Ordentlich, A.; Barak, D.; Ariel, N.; Kronman, C.; Velan, B.; Shafferman, A. Does “Butyrylization” of Acetylcholinesterase through Substitution of the Six Divergent Aromatic Amino Acids in the Active Center Gorge Generate an Enzyme Mimic of Butyrylcholinesterase? Biochemistry 2001, 40, 7433–7445. https://doi.org/10.1021/bi010181x. (68) Desmedt, J. E.; La Grutta, G. Control of Brain Potentials by Pseudo- Cholinesterase. Proc Physiol. Soc. 1955, 129, 46–47. https://doi.org/10.1016/S0140-6736(00)96392-X. (69) Giacobini, E. Selective Inhibitors of Butyrylcholinesterase: A Valid Alternative for Therapy of Alzheimer’s Disease? Drugs Aging 2001, 18, 891–898. (70) Holmstedt, B.; Sjoqvist, F. Distribution of Acetocholinesterase i n the Ganglion Cells of Various Sympathetic Ganglia. Acta Physiol. Scand. 1959, 47, 284–296. (71) Soreq, H.; Lapidot-Lifson, Y.; Zakut, H. A Role for Cholinesterases in Tumorigenesis? Cancer Cells 1991, 3 (12), 511–516. (72) Giacobini, E. From Molecular Structure to Alzheimer Therapy. Jpn. J. Pharmacol. 1997, 74 (3), 225–241. https://doi.org/10.1254/jjp.74.225. (73) Li, B.; Stribley, J. A.; Ticu, A.; Xie, W.; Schopfer, L. M.; Hammond, P.; Brimijoin, S.; Hinrichs, S. H.; Lockridge, O. Abundant Tissue Butyrylcholinesterase and Its Possible Function in the Acetylcholinesterase Knockout Mouse. J. Neurochem. 2000, 75 (6), 1320–1331. https://doi.org/10.1046/j.1471-4159.2000.751320.x. (74) Guillozet, A. L.; Smiley, J. F.; Mash, D. C.; Mesulam, M. M. Butyrylcholinesterase in the Life Cycle of Amyloid Plaques. Annals of Neurology. 1997, pp 909–918. https://doi.org/10.1002/ana.410420613. (75) Hawkins, R. D.; Gunter, J. M. Studies on Cholinesterase in Obstetrics. Biochem. J. 1946, 40, 192–197. https://doi.org/10.14930/jsma1939.19.11. (76) Primo-Parmo, S. L.; Bartels, C. F.; Wiersema, B.; van der Spek, A. F.; Innis, J. W.; La Du, B. N. Characterization of 12 Silent Alleles of the Human

215 Butyrylcholinesterase (BCHE) Gene. Am. J. Hum. Genet. 1996, 58, 52–64. (77) Manoharan, I.; Wieseler, S.; Layer, P. G.; Lockridge, O.; Boopathy, R. Naturally Occurring Mutation Leu307Pro of Human Butyrylcholinesterase in the Vysya Community of India. Pharmacogenet. Genomics 2006, 16, 461–468. https://doi.org/10.1097/01.fpc.0000197464.37211.77. (78) Manoharan, I.; Boopathy, R.; Darvesh, S.; Lockridge, O. A Medical Health Report on Individuals with Silent Butyrylcholinesterase in the Vysya Community of India. Clin. Chim. Acta 2007, 378 (1), 128–135. https://doi.org/10.1016/j.cca.2006.11.005. (79) Chen, P.; Gao, Y.; Geng, L.; Ping, V.; Gao, Y.; Geng, L.; Parks, R. J.; Pang, Y.; Brimijoin, S. Plasma Butyrylcholinesterase Regulates Ghrelin to Control Aggression. Proc. Natl. Acad. Sci. 2015, 112 (7), 2251–2256. https://doi.org/10.1073/pnas.1503913112. (80) Lenz, D. E.; Yeung, D.; Smith, J. R.; Sweeney, R. E.; Lumley, L. A.; Cerasoli, D. M. Stoichiometric and Catalytic Scavengers as Protection against Nerve Agent Toxicity: A Mini Review. Toxicology 2007, 233 (1–3 SPEC. ISS.), 31–39. https://doi.org/10.1016/j.tox.2006.11.066. (81) Worek, F.; Thiermann, H.; Szinicz, L.; Eyer, P. Kinetic Analysis of Interactions between Human Acetylcholinesterase, Structurally Different Organophosphorus Compounds and Oximes. Biochem. Pharmacol. 2004, 68 (11), 2237–2248. https://doi.org/10.1016/j.bcp.2004.07.038. (82) Carletti, E.; Aurbek, N.; Gillon, E.; Loiodice, M.; Nicolet, Y.; Fontecilla-Camps, J.-C.; Masson, P.; Thiermann, H.; Nachon, F.; Worek, F. Structure–Activity Analysis of Aging and Reactivation of Human Butyrylcholinesterase Inhibited by Analogues of Tabun. Biochem. J. 2009, 421 (1), 97–106. https://doi.org/10.1042/BJ20090091. (83) Beck, J. M. Organophosphorus Nerve Agent Chemistry; Interactions of Chemical Warfare Agents and Their Therapeutics with Acetylcholinesterase, The Ohio State University, 2011. (84) Shubham, V. Computational and Experimental Studies Towards the Development of Novel Therapeutics Against Organophosphorus Nerve Agents: Butyrylcholinesterase and Paraoxonase, 2011. (85) Cerasoli, D.; Cadieux, C. L. ICD Preparation; 2018. (86) Masson, P.; Froment, M. T.; Bartels, C. F.; Lockridge, O. Importance of Aspartate- 70 in Organophosphate Inhibition, Oxime Re-Activation and Aging of Human Butyrylcholinesterase. Biochem. J. 1997, 325, 53–61. https://doi.org/10.1042/bj3250053. (87) Ashani, Y.; Bhattacharjee, A. K.; Leader, H.; Saxena, A.; Doctor, B. P. Inhibition of Cholinesterases with Cationic Phosphonyl Oximes Highlights Distinctive Properties of the Charged Pyridine Groups of Quaternary Oxime Reactivators. Biochem. Pharmacol. 2003, 66 (2), 191–202. https://doi.org/10.1016/S0006- 2952(03)00204-1. (88) Zhuang, Q.; Young, A.; Callam, C. S.; Mcelroy, C. A. Efforts toward Treatments against Aging of Organophosphorus- Inhibited Acetylcholinesterase Efforts

216 toward Treatments against Aging of Organophosphorus-Inhibited Acetylcholinesterase. Ann. N. Y. Acad. Sci. 2016, 1–11. https://doi.org/10.1111/nyas.13124. (89) Mercey, G.; Verdelet, T.; Renou, J.; Kliachyna, M.; Baati, R.; Nachon, F.; Jean, L.; Renard, P. Y. Reactivators of Acetylcholinesterase Inhibited by Organophosphorus Nerve Agents. Acc. Chem. Res. 2012, 45 (5), 756–766. https://doi.org/10.1021/ar2002864. (90) Worek, F.; Thiermann, H.; Wille, T. Oximes in Organophosphate Poisoning: 60 Years of Hope and Despair. Chem. Biol. Interact. 2016, 259, 93–98. https://doi.org/10.1016/j.cbi.2016.04.032. (91) Katz, F. S.; Pecic, S.; Schneider, L.; Zhu, Z.; Hastings, A.; Luzac, M.; Macdonald, J.; Landry, D. W.; Stojanovic, M. N. New Therapeutic Approaches and Novel Alternatives for Organophosphate Toxicity. Toxicol. Lett. 2018, 291, 1–10. https://doi.org/10.1016/j.toxlet.2018.03.028. (92) Kovalesky, A.; Blumenthal, D. K.; Cheng, X.; Taylor, P.; Radic, Z. Limitations in Current Acetylcholinesterase Structure-Based Design of Oxime Antitodes for Oganophosphate Poisoning. Ann. N. Y. Acad. Sci. 2016, 1378 (1), 41–49. https://doi.org/10.5588/ijtld.16.0716.Isoniazid. (93) Gorecki, L.; Korabecny, J.; Musilek, K.; Malinak, D.; Nepovimova, E.; Dolezal, R.; Jun, D.; Soukup, O.; Kuca, K. SAR Study to Find Optimal Cholinesterase Reactivator against Organophosphorous Nerve Agents and Pesticides. Arch. Toxicol. 2016, 90 (12), 2831–2859. https://doi.org/10.1007/s00204-016-1827-3. (94) Chambers, J. E.; Chambers, H. W.; Meek, E. C.; Pringle, R. B. Testing of Novel Brain-Penetrating Oxime Reactivators of Acetylcholinesterase Inhibited by Nerve Agent Surrogates. Chem. Biol. Interact. 2013, 203 (1), 135–138. https://doi.org/10.1016/j.cbi.2012.10.017. (95) Chambers, J. E.; Meek, E. C.; Chambers, H. W. Novel Brain-Penetrating Oximes for Reactivation of Cholinesterase Inhibited by Sarin and VX Surrogates. Ann. N. Y. Acad. Sci. 2016, 1374 (1), 52–58. https://doi.org/10.1111/nyas.13053. (96) De Koning, M. C.; Joosen, M. J. A.; Worek, F.; Nachon, F.; Van Grol, M.; Klaassen, S. D.; Alkema, D. P. W.; Wille, T.; De Bruijn, H. M. Application of the Ugi Multicomponent Reaction in the Synthesis of Reactivators of Nerve Agent Inhibited Acetylcholinesterase. J. Med. Chem. 2017, 60 (22), 9376–9392. https://doi.org/10.1021/acs.jmedchem.7b01083. (97) Kuca, K.; Korabecny, J.; Dolezal, R.; Nepovimova, E.; Soukup, O.; Gorecki, L. Tetroxime : Reactivation Potency – in Vitro and in Silico Study. RSC Adv. 2017, 7, 7041–7045. https://doi.org/10.1039/c6ra16499d. (98) Kuca, K.; Jun, D.; Junova, L.; Musilek, K.; Hrabinova, M.; Alberto, J.; Ramalho, T. C.; Valko, M.; Wu, Q.; Nepovimova, E.; et al. Synthesis, Biological Evaluation, and Docking Studies of Novel Bisquaternary Aldoxime Reactivators on Acetylcholinesterase and Butyrylcholinesterase Inhibited by Paraoxon. Molecules 2018, 23 (1103), 1–11. https://doi.org/10.3390/molecules23051103. (99) Sahu, A. K.; Sharma, R.; Gupta, B.; Musilek, K.; Kuca, K.; Acharya, J.; Ghosh, K. K. Oxime-Mediated in Vitro Reactivation Kinetic Analysis of Organophosphates-

217 Inhibited Human and Electric Eel Acetylcholinesterase. Toxicol. Mech. Methods 2016, 6516 (5), 319–326. https://doi.org/10.3109/15376516.2016.1143070. (100) Malinak, D.; Nepovimova, E.; Jun, D.; Musilek, K.; Kuca, K. Novel Group of AChE Reactivators—Synthesis, In Vitro Reactivation and Molecular Docking Study. Molecules 2018, 23 (2291), 1–11. https://doi.org/10.3390/molecules23092291. (101) Berberich, J. A.; Stouch, T. R.; Manepalli, S.; Esposito, E. X.; Madura, J. D. Biological Testing of Organophosphorus-Inactivated Acetylcholinesterase Oxime Reactivators Identified via Virtual Screening. Chem. Res. Toxicol. 2016, 29 (9), 1534–1540. https://doi.org/10.1021/acs.chemrestox.6b00198. (102) Karade, H. N.; Raviraju, G.; Acharya, B. N.; Valiveti, A. K.; Bhalerao, U.; Acharya, J. Synthesis and in Vitro Reactivation Study of Isonicotinamide Derivatives of 2-(Hydroxyimino)-N-(Pyridin-3-Yl)Acetamide as Reactivators of Sarin and VX Inhibited Human Acetylcholinesterase (HAChE). Bioorganic Med. Chem. 2016, 24 (18), 4171–4176. https://doi.org/10.1016/j.bmc.2016.07.005. (103) Sharma, R.; Gupta, B.; Sahu, A. K.; Acharya, J.; Satnami, M. L.; Ghosh, K. K. Synthesis and In-Vitro Reactivation Screening of Imidazolium Aldoximes as Reactivators of Sarin and VX-Inhibited Human Acetylcholinesterase (HAChE). Chem. Biol. Interact. 2016, 259, 85–92. https://doi.org/10.1016/j.cbi.2016.04.034. (104) Amitai, G.; Gez, R.; Raveh, L.; Bar-Ner, N.; Grauer, E.; Chapman, S. Novel Bifunctional Hybrid Small Molecule Scavengers for Mitigating Nerve Agents Toxicity. Chem. Biol. Interact. 2016, 259, 187–204. https://doi.org/10.1016/j.cbi.2016.04.036. (105) Bodor, N.; Shek, E.; Higuchi, T. Improved Delivery Through Biological Membranes. 1. Synthesis and Properties of 1-Methyl-1,6-Dihydropyridine-2- Carbaldoxime, a Pro-Drug of N-Methylpyridinium-2-Carbaldoxime Chloride. J. Med. Chem. 1976, 19 (1), 102–107. https://doi.org/10.1021/jm00223a017. (106) Wei, Z.; Liu, Y. qin; Wang, S. zheng; Yao, L.; Nie, H. fang; Wang, Y. an; Liu, X. Y.; Zheng, Z. bing; Li, S. Conjugates of Salicylaldoximes and Peripheral Site Ligands: Novel Efficient Nonquaternary Reactivators for Nerve Agent-Inhibited Acetylcholinesterase. Bioorganic Med. Chem. 2017, 25 (16), 4497–4505. https://doi.org/10.1016/j.bmc.2017.06.041. (107) Kuca, K.; Musilek, K.; Jun, D.; Zdarova-Karasova, J.; Nepovimova, E.; Soukup, O.; Hrabinova, M.; Mikler, J.; Franca, T. C. C.; Da Cunha, E. F. F.; et al. A Newly Developed Oxime K203 Is the Most Effective Reactivator of Tabun-Inhibited Acetylcholinesterase. BMC Pharmacol. Toxicol. 2018, 19 (1), 1–10. https://doi.org/10.1186/s40360-018-0196-3. (108) Wei, Z.; Liu, Y.; Wang, Y.; Li, W.; Zhou, X.; Zhao, J.; Huang, C.; Li, X.; Liu, J.; Zheng, Z.; et al. Novel Nonquaternary Reactivators Showing Reactivation Efficiency for Soman-Inhibited Human Acetylcholinesterase. Toxicol. Lett. 2016, 246, 1–6. https://doi.org/10.1016/j.toxlet.2016.01.015. (109) Julien, R.; Jose, D.; Mercey, G.; Verdelet, T.; Rousseau, C.; Gastellier, A.-J.; Melanie, A.; Touvrey-Loiodice, M.; Baati, R.; Jean, L.; et al. Synthesis and in Vitro Evaluation of Donepezil Based Reactivators and Analogues for Nerve

218 Agent-Inhibited Human Acetylcholinesterase. RSC Adv. 2016, 6, 17929–17940. https://doi.org/10.1039/C5RA25477A. (110) Zorbaz, T.; Braïki, A.; Marakovic, N.; Renou, J.; de la Mora, E.; Marek Hrvat, N.; Katalinic, M.; Silman, I.; Sussman, J. L.; Mercey, G.; et al. Potent 3-Hydroxy-2- Pyridine Aldoxime Reactivators of Organophosphate-Inhibited Cholinesterases with Predicted Blood-Brain Barrier Penetration. Chem. - A Eur. J. 2018. https://doi.org/10.1002/chem.201801394. (111) Santoni, G.; de Sousa, J.; De la Mora, E.; Dias, J.; Jean, L.; Sussman, J. L.; Silman, I.; Renard, P.-Y.; Brown, R. C. D.; Weik, M.; et al. Structure-Based Optimization of Non-Quaternary Reactivators of Acetylcholinesterase Inhibited by Organophosphorus Nerve Agents. J. Med. Chem. 2018, 61, 7630–7639. https://doi.org/10.1021/acs.jmedchem.8b00592. (112) Bušić, V.; Katalinić, M.; Šinko, G.; Kovarik, Z.; Gašo-Sokač, D. Pyridoxal Oxime Derivative Potency to Reactivate Cholinesterases Inhibited by Organophosphorus Compounds. Toxicol. Lett. 2016, 262, 114–122. https://doi.org/10.1016/j.toxlet.2016.09.015. (113) Katz, F. S.; Pecic, S.; Tran, T. H.; Trakht, I.; Schneider, L.; Zhu, Z.; Ton-That, L.; Luzac, M.; Zlatanic, V.; Damera, S.; et al. Discovery of New Classes of Compounds That Reactivate Acetylcholinesterase Inhibited by Organophosphates. ChemBioChem 2015, 16 (15), 2205–2215. https://doi.org/10.1002/cbic.201500348. (114) Cadieux, C. L.; Wang, H.; Zhang, Y.; Koenig, J. A.; Shih, T.; Mcdonough, J.; Koh, J.; Cerasoli, D. Probing the Activity of a Non-Oxime Reactivator for Acetylcholinesterase Inhibited by Organophosphorus Nerve Agents. Chem. Biol. Interact. 2016, 259, 133–141. https://doi.org/10.1016/j.cbi.2016.04.002. (115) Bierwisch, A.; Wille, T.; Thiermann, H.; Worek, F. Kinetic Analysis of Interactions of Amodiaquine with Human Cholinesterases and Organophosphorus Compounds. Toxicol. Lett. 2016, 246, 49–56. https://doi.org/10.1016/j.toxlet.2016.02.004. (116) de Koning, M. C.; Horn, G.; Worek, F.; van Grol, M. Discovery of a Potent Non- Oxime Reactivator of Nerve Agent Inhibited Human Acetylcholinesterase. Eur. J. Med. Chem. 2018, 157, 151–160. https://doi.org/10.1016/j.ejmech.2018.08.016. (117) Chunyuan, L.; Saxena, A.; Smith, M.; Garcia, G.; Radić, Z.; Taylor, P.; Doctor, B. P. Phosphoryl Oxime Inhibition of Acetylcholinesterase during Oxime Reactivation Is Prevented by Edrophonium. Biochemistry 1999, 38 (31), 9937– 9947. https://doi.org/10.1021/bi9905720. (118) Ochoa, R.; Rodriguez, C. A.; Zuluaga, A. F. Perspectives for the Structure-Based Design of Acetylcholinesterase Reactivators. J. Mol. Graph. Model. 2016, 68, 176–183. https://doi.org/10.1016/j.jmgm.2016.07.002. (119) Macek Hrvat, N.; Zorbaz, T.; Sinko, G.; Kovarik, Z. The Estimation of Oxime Efficiency Is Affected by the Experimental Design of Phosphylated Acetylcholinesterase Reactivation. Toxicol. Lett. 2018, No. 293, 222–228. (120) Ashani, Y.; Shapira, S.; Levy, D.; Wolfe, A. D.; Doctor, B. P.; raveh, lily. Butyrylcholinesterase and Acetylcholinesterase Prophylaxis against Soman Poisoning in Mice. Biochem. Pharmacol. 1991, 41 (1), 37–41.

219 https://doi.org/10.1016/0006-2952(91)90008-S. (121) Raveh, L.; Grauer, E.; Grunwald, J.; Cohen, E.; Ashani, Y. The Stoichiometry of Protection against Soman and VX Toxicity in Monkeys Pretreated with Human Butyrylcholinesterase. Toxicol. Appl. Pharmacol. 1997, 145, 43–53. https://doi.org/10.1006/taap.1997.8160. (122) Cerasoli, D. M.; Griffiths, E. M.; Doctor, B. P.; Saxena, A.; Fedorko, J. M.; Greig, N. H.; Yu, Q. S.; Huang, Y.; Wilgus, H.; Karatzas, C. N.; et al. In Vitro and in Vivo Characterization of Recombinant Human Butyrylcholinesterase (Protexia) as a Potential Nerve Agent Bioscavenger. Chem. Biol. Interact. 2005, 157–158, 363– 365. https://doi.org/10.1016/j.cbi.2005.10.052. (123) ØStergaard, D.; Viby–Mogensen, J.; Hanel, H. K.; Skovgaard, L. T. Half–Life of Plasma Cholinesterase. Acta Anaesthesiol. Scand. 1988, 32 (3), 266–269. https://doi.org/10.1111/j.1399-6576.1988.tb02727.x. (124) Raveh, L.; Grunwald, J.; Marcus, D.; Papier, Y.; Cohen, E.; Ashani, Y. AS A GENERAL PROPHYLACTIC ANTIDOTE FOR NERVE AGENT TOXICITY. Biochem. Pharmacol. 1993, 45 (12), 2465–2474. (125) Huang, Y.-J.; Huang, Y.; Baldassarre, H.; Wang, B.; Lazaris, A.; Leduc, M.; Bilodeau, A. S.; Bellemare, A.; Cote, M.; Herskovits, P.; et al. Recombinant Human Butyrylcholinesterase from Milk of Transgenic Animals to Protect against Organophosphate Poisoning. Proc. Natl. Acad. Sci. 2007, 104 (34), 13603–13608. https://doi.org/10.1073/pnas.0702756104. (126) Huang, Y. J.; Lundy, P. M.; Lazaris, A.; Huang, Y.; Baldassarre, H.; Wang, B.; Turcotte, C.; Côté, M.; Bellemare, A.; Bilodeau, A. S.; et al. Substantially Improved Pharmacokinetics of Recombinant Human Butyrylcholinesterase by Fusion to Human Serum Albumin. BMC Biotechnol. 2008, 8, 1–11. https://doi.org/10.1186/1472-6750-8-50. (127) Saxena, A.; Luo, C.; Chilukuri, N.; Maxwell, D. M.; Doctor, B. P. Novel Approaches to Medical Protection Against Chemical Warfare Agents. In Chemical warfare agents chemistry, pharmacology, toxicology and therapeutics; 2007; pp 145–173. (128) Lenz, D. E.; Broomfield, C. A.; Yeung, D. T.; Masson, P.; Maxwell, D. M.; Cerasoli, D. M. Nerve Agent Bioscavengers: Progress in Development of a New Mode of Protection against Organophosphate Exposure. In Chemical warfare agents chemistry, pharmacology, toxicology and therapeutics; 2007; pp 175–202. (129) Lenz, D. E.; Clarkson, E. D.; Schulz, S. M.; Cerasoli, D. M. Butyrylcholinesterase as a Therapeutic Drug for Protection against Percutaneous VX. Chem. Biol. Interact. 2010, 187 (1–3), 249–252. https://doi.org/10.1016/j.cbi.2010.05.014. (130) Saxena, A.; Viragh, C.; Frazier, D. S.; Kovach, I. M.; Maxwell, D. M.; Lockridge, O.; Doctor, B. P. The PH Dependence of Dealkylation in Soman-Inhibited Cholinesterases and Their Mutants: Further Evidence for a Push-Pull Mechanism. Biochemistry 1998, 37 (43), 15086–15096. https://doi.org/10.1021/bi980917z. (131) Duysen, E. G.; Bartels, C. F.; Lockridge, O. Wild-Type and A328W Mutant Human Butyrylcholinesterase Tetramers Expressed in Chinese Hamster Ovary Cells Have a 16-Hour Half-Life in the Circulation and Protect Mice from Cocaine

220 Toxicity. J. Pharmacol. Exp. Ther. 2002, 302 (2), 751–758. https://doi.org/10.1124/jpet.102.033746. (132) Mumford, H.; Price, M. E.; Cerasoli, D. M.; Teschner, W.; Ehrlich, H.; Schwarz, H. P.; Lenz, D. E. Efficacy and Physiological Effects of Human Butyrylcholinesterase as a Post-Exposure Therapy against Percutaneous Poisoning by VX in the Guinea-Pig. Chem. Biol. Interact. 2010, 187 (1–3), 304–308. https://doi.org/10.1016/j.cbi.2010.02.020. (133) Kovarik, Z.; Ciban, N.; Radić, Z.; Simeon-Rudolf, V.; Taylor, P. Active Site Mutant Acetylcholinesterase Interactions with 2-PAM, HI-6, and DDVP. Biochem. Biophys. Res. Commun. 2006, 342 (3), 973–978. https://doi.org/10.1016/j.bbrc.2006.02.056. (134) Musilova, L.; Kuca, K.; Jung, Y. S.; Jun, D. In Vitro Oxime-Assisted Reactivation of Paraoxon-Inhibited Human Acetylcholinesterase and Butyrylcholinesterase. Clin. Toxicol. 2009, 47 (6), 545–550. https://doi.org/10.1080/15563650903058914. (135) Jun, D.; Musilova, L.; Pohanka, M.; Jung, Y. S.; Bostik, P.; Kuca, K. Reactivation of Human Acetylcholinesterase and Butyrylcholinesterase Inhibited by Leptophos- Oxon with Different Oxime Reactivators in Vitro. Int. J. Mol. Sci. 2010, 11 (8), 2856–2863. https://doi.org/10.3390/ijms11082856. (136) Jun, D.; Musilova, L.; Musilek, K.; Kuca, K. In Vitro Ability of Currently Available Oximes to Reactivate Organophosphate Pesticide-Inhibited Human Acetylcholinesterase and Butyrylcholinesterase. Int. J. Mol. Sci. 2011, 12 (3), 2077–2087. https://doi.org/10.3390/ijms12032077. (137) Masson, P.; Nachon, F. Cholinesterase Reactivators and Bioscavengers for Pre- and Post-Exposure Treatments of Organophosphorus Poisoning. J. Neurochem. 2017, 142, 26–40. https://doi.org/10.1111/jnc.14026. (138) Worek, F.; Thiermann, H.; Wille, T. Catalytic Bioscavengers in Nerve Agent Poisoning: A Promising Approach? Toxicol. Lett. 2016, 244, 143–148. https://doi.org/10.1016/j.toxlet.2015.07.012. (139) Katalinić, M.; Maček Hrvat, N.; Baumann, K.; Morasi Piperčić, S.; Makarić, S.; Tomić, S.; Jović, O.; Hrenar, T.; Miličević, A.; Jelić, D.; et al. A Comprehensive Evaluation of Novel Oximes in Creation of Butyrylcholinesterase-Based Nerve Agent Bioscavengers. Toxicol. Appl. Pharmacol. 2016, 310, 195–204. https://doi.org/10.1016/j.taap.2016.09.015. (140) Katalinić, M.; Zandona, A.; Ramić, A.; Zorbaz, T.; Primožič, I.; Kovarik, Z. New Cinchona Oximes Evaluated as Reactivators of Acetylcholinesterase and Butyrylcholinesterase Inhibited by Organophosphorus Compounds. Molecules 2017, 22 (7). https://doi.org/10.3390/molecules22071234. (141) Radić, Z.; Dale, T.; Kovarik, Z.; Berend, S.; Garcia, E.; Zhang, L.; Amitai, G.; Radić, B.; Duggan, B. M.; Ajami, D.; J, J. R.; Taylor, P.; Green, C. Catalytic Detoxification of Nerve Agent and Pesticide Organophosphates by Butyrylcholinesterase Assisted with Non-Pyridinium Oximes. Biochem. J. 2013, 242, 231–242. https://doi.org/10.1042/BJ20121612. (142) Sit, R. K.; Fokin, V. V.; Amitai, G.; Sharpless, K. B.; Taylor, P.; Radić, Z. Imidazole Aldoximes Effective in Assisting Butyrylcholinesterase Catalysis of

221 Organophosphate Detoxification. J. Med. Chem. 2014, 57 (4), 1378–1389. https://doi.org/10.1021/jm401650z. (143) Kovarik, Z.; Katalinić, M.; Šinko, G.; Binder, J.; Holas, O.; Jung, Y. S.; Musilova, L.; Jun, D.; Kuča, K. Pseudo-Catalytic Scavenging: Searching for a Suitable Reactivator of Phosphorylated Butyrylcholinesterase. Chem. Biol. Interact. 2010, 187 (1–3), 167–171. https://doi.org/10.1016/j.cbi.2010.02.023. (144) Wille, T.; von der Wellen, J.; Thiermann, H.; Worek, F. Pseudocatalytic Scavenging of the Nerve Agent VX with Human Blood Components and the Oximes Obidoxime and HI-6. Arch. Toxicol. 2017, 91 (3), 1309–1318. https://doi.org/10.1007/s00204-016-1776-x. (145) Khavrutskii, I. V; Wallqvist, A. B-Aminoalcohols as Potential Reactivators of Aged Sarin- / Soman-Inhibited Acetylcholinesterase. Chem. Sel. 2017, 2, 1885– 1890. https://doi.org/10.1002/slct.201601828. (146) Chandar, N. B.; Lo, R.; Ganguly, B. Quantum Chemical and Steered Molecular Dynamics Studies for One Pot Solution to Reactivate Aged Acetylcholinesterase with Alkylator Oxime. Chem. Biol. Interact. 2014, 223, 58–68. https://doi.org/10.1016/j.cbi.2014.08.015. (147) Topczewski, J. J.; Quinn, D. M. Kinetic Assessment of N-Methyl-2- Methoxypyridinium Species as Phosphonate Anion Methylating Agents. Org. Lett. 2013, 15 (5), 1084–1087. https://doi.org/10.1021/ol400054m. (148) Topczewski, J. J.; Lodge, A. M.; Yasapala, S. N.; Payne, M. K.; Keshavarzi, P. M.; Quinn, D. M. Reversible Inhibition of Human Acetylcholinesterase by Methoxypyridinium Species. Bioorganic Med. Chem. Lett. 2013, 23 (21), 5786– 5789. https://doi.org/10.1016/j.bmcl.2013.09.008. (149) Steinberg, G. M.; Lieske, C. N.; Boldt, R.; Goan, J. C.; Podall, H. E. Model Studies for the Reactivation of Aged Phosphonylated Acetylcholinesterase. Use of Alkylating Agents Containing Nucleophilic Groups. J. Med. Chem. 1970, 13 (3), 435–446. https://doi.org/10.1021/jm00297a024. (150) Ash, A. B.; Blumbergs, P.; Stevens, C. L.; Michel, H. O.; Hackley, B. E.; Epstein, J. W. Relative Nucleophilicity. Methylation of Anions in Aqueous Media. J. Org. Chem. 1969, 34 (12), 4070–4072. https://doi.org/10.1021/jo01264a068. (151) Blumbergs, P.; Ash, A. B.; Daniher, F. A.; Stevens, C. L.; Michel, H. O.; Hackley, B. E.; Epstein, J. Alkylating Agents Containing a Quaternary Nitrogen Group. J. Org. Chem. 1969, 34 (12), 4065–4070. https://doi.org/10.1021/jo01264a067. (152) Powers, J. C.; Asgian, J. L.; Ekici, Ö. D.; James, K. E. Irreversible Inhibitors of Serine, Cysteine, and Threonine Proteases. Chem. Rev. 2002, 102 (12), 4639– 4750. https://doi.org/10.1021/cr010182v. (153) Chandar, N. B.; Ganguly, B. A First Principles Investigation of Aging Processes in Soman Conjugated AChE. Chem. Biol. Interact. 2013, 204 (3), 185–190. https://doi.org/10.1016/j.cbi.2013.05.013. (154) Zhou, Q.; Turnbull, K. D. Phosphodiester Alkylation with a Quinone Methide. J. Org. Chem. 1999, 64 (8), 2847–2851. https://doi.org/10.1021/jo9823745. (155) Zhou, Q.; Turnbull, K. D. Quinone Methide Phosphodiester Alkylations under Aqueous Conditions. J. Org. Chem. 2001, 66 (21), 7072–7077.

222 https://doi.org/10.1021/jo015792+. (156) Weinert, E. E.; Dondi, R.; Colloredo-Melz, S.; Frankenfield, K. N.; Mitchell, C. H.; Freccero, M.; Rokita, S. E. Substituents on Quinone Methides Strongly Modulate Formation and Stability of Their Nucleophilic Adducts. J. Am. Chem. Soc. 2006, 128 (36), 11940–11947. https://doi.org/10.1021/ja062948k. (157) Zhou, Q.; Turnbull, K. D. Trapping Phosphodiester-Quinone Methide Adducts through in Situ Lactonization. J. Org. Chem. 2000, 65 (7), 2022–2029. https://doi.org/10.1021/jo9915168. (158) Bakke, B. A.; McIntosh, M. C.; Turnbull, K. D. Improved Alkylation and Product Stability in Phosphotriester Formation through Quinone Methide Reactions with Dialkyl Phosphates. J. Org. Chem. 2005, 70 (11), 4338–4345. https://doi.org/10.1021/jo050050s. (159) Yoder, R. J.; Zhuang, Q.; Beck, J. M.; Franjesevic, A.; Blanton, T. G.; Sillart, S.; Secor, T.; Guerra, L.; Brown, J. D.; Reid, C.; et al. Study of Para -Quinone Methide Precursors toward the Realkylation of Aged Acetylcholinesterase. J. Med. Chem. Lett. 2017, 8 (6), 622–627. https://doi.org/10.1021/acsmedchemlett.7b00037. (160) Zhuang, Q.; Franjesevic, A. J.; Corrigan, T. S.; Coldren, W. H.; Dicken, R.; Sillart, S.; Deyong, A.; Yoshino, N.; Smith, J.; Fabry, S.; et al. Demonstration of in Vitro Resurrection of Aged Acetylcholinesterase after Exposure to Organophosphorus Chemical Nerve Agents. J. Med. Chem. 2018. https://doi.org/10.1021/acs.jmedchem.7b01620. (161) Timperley, C. M.; Casey, K. E.; Notman, S.; Sellers, D. J.; Williams, N. E.; Williams, N. H.; Williams, G. R. Synthesis and Anticholinesterase Activity of Some New Fluorogenic Analogues of Organophosphorus Nerve Agents. J. Fluor. Chem. 2006, 127 (12), 1554–1563. https://doi.org/10.1016/j.jfluchem.2006.07.017. (162) Briseño-Roa, L.; Hill, J.; Notman, S.; Sellers, D.; Smith, A. P.; Timperley, C. M.; Wetherell, J.; Williams, N. H.; Williams, G. R.; Fersht, A. R.; et al. Analogues with Fluorescent Leaving Groups for Screening and Selection of Enzymes That Efficiently Hydrolyze Organophosphorus Nerve Agents. J. Med. Chem. 2006, 49 (1), 246–255. https://doi.org/10.1021/jm050518j. (163) An, Y.; Zhu, Y.; Yao, Y.; Liu, J. Is It Possible to Reverse Aged Acetylcholinesterase Inhibited by Organophosphorus Compounds? Insight from the Theoretical Study. Phys. Chem. Chem. Phys. 2016, 18 (14), 9838–9846. https://doi.org/10.1039/C5CP07991H. (164) Quinn, D. M.; Topczewski, J.; Yasapala, N.; Lodge, A. Why Is Aged Acetylcholinesterase so Difficult to Reactivate? Molecules 2017, 22 (9), 1–6. https://doi.org/10.3390/molecules22091464. (165) Quinn, D. M. Acetylcholinesterase: Enzyme Structure, Reaction Dynamics, and Virtual Transition States. Chem. Rev. 1987, 87 (5), 955,979. https://doi.org/10.1021/cr00081a005. (166) Wiener, S. W.; Hoffman, R. S. Nerve Agents : A Comprehensive Review. J. Intensive Care Med. 2004, 19 (1), 22–37. https://doi.org/10.1177/0885066603258659.

223 (167) Chauhan, S.; Chauhan, S.; Cruz, R. D.; Faruqi, S.; Singh, K. K.; Varma, S.; Singh, M.; Karthik, V. Chemical Warfare Agents. Environ. Toxicol. Pharmacol. 2008, 26, 113–122. https://doi.org/10.1016/j.etap.2008.03.003. (168) Worek, F.; Szinicz, L.; Eyer, P.; Thiermann, H. Evaluation of Oxime Efficacy in Nerve Agent Poisoning : Development of a Kinetic-Based Dynamic Model. Toxicol. A 2005, 209, 193–202. https://doi.org/10.1016/j.taap.2005.04.006. (169) Worek, F.; Reiter, G.; Eyer, P.; Szinicz, L. Reactivation Kinetics of Acetylcholinesterase from Different Species Inhibited by Highly Toxic Organophosphates. Arch. Toxicol. 2002, 76, 523–529. https://doi.org/10.1007/s00204-002-0375-1. (170) Seckute, J.; Menke, J. L.; Emnett, R. J.; Patterson, E. V; Cramer, C. J. Ab Initio Molecular Orbital and Density Functional Studies on the Solvolysis of Sarin and O , S -Dimethyl Methylphosphonothiolate , a VX-like Compound. J. Org. Chem 2005, 70 (22), 8649–8660. (171) Vasilyev, V. V. Tetrahedral Intermediate Formation in the Acylation Step of Acetylcholinesterases . A Combined Quantum Chemical and Molecular Mechanical Model. J. Mol. Struct. THEOCHEM 1994, 304, 129–141. (172) Beck, J. M.; Hadad, C. M. Hydrolysis of Nerve Agents by Model Nucleophiles : A Computational Study. Chem. Biol. Interact. 2008, 175 (1–3), 200–203. https://doi.org/10.1016/j.cbi.2008.04.026. (173) Barakat, N. H.; Zheng, X.; Gilley, C. B.; Macdonald, M.; Okolotowicz, K.; Cashman, J. R.; Vyas, S.; Beck, J. M.; Hadad, C. M.; Zhang, J. Chemical Synthesis of Two Series of Nerve Agent Model Compounds and Their Stereoselective Interaction with Human. Chem. Res. Toxicol. 2009, 22 (10), 1669– 1679. (174) Beck, J. M.; Hadad, C. M. Reaction Profiles of the Interaction between Sarin and Acetylcholinesterase and the S203C Mutant : Model Nucleophiles and QM / MM Potential Energy Surfaces. Chem. Biol. Interact. 2010, 187 (1–3), 220–224. https://doi.org/10.1016/j.cbi.2010.02.012. (175) Hurley, M. M.; Wright, J. B.; Lushington, G. H.; White, W. E. Regular Article Quantum Mechanics and Mixed Quantum Mechanics / Molecular Mechanics Simulations of Model Nerve Agents with Acetylcholinesterase. Theoretical 2003, 109 (3), 160–168. https://doi.org/10.1007/s00214-002-0424-0. (176) Shenoi, R. Chemical Warfare Agents. Clin. Pediatr. Emerg. Med. 2002, 3 (4), 239–247. (177) Sanan, T. T.; Muthukrishnan, S.; Beck, J. M.; Tao, P.; Hayes, C. J.; Otto, T. C.; Cerasoli, D. M.; Lenzy, D. E.; Hadad, C. M. Computational Modeling of Human Paraoxonase 1: Preparation of Protein Models, Binding Studies, and Mechanistic Insights. J. Phys. Org. Chem. 2013, 23 (4), 357–369. https://doi.org/10.1002/poc.1678.Computational. (178) Muthukrishnan, S.; Shete, V. S.; Sanan, T. T.; Vyas, S.; Oottikkal, S.; Porter, L. M.; Magliery, T. J.; Hadad, C. M. Mechanistic Insights into the Hydrolysis of Organophosphorus Compounds by Paraoxonase-1 : Exploring the Limits of Substrate Tolerance in a Promiscuous Enzyme. J. Phys. Org. Chem. 2012, 25 (12),

224 1247–1260. https://doi.org/10.1002/poc.3002. (179) Vyas, S.; Hadad, C. M. Reactivation of Model Cholinesterases by Oximes and Intermediate Phosphyloximes : A Computational Study. Chem. Biol. Interact. 2008, 175 (1–3), 187–191. https://doi.org/10.1016/j.cbi.2008.05.006. (180) Michel, H. O.; Hackley, B. E.; Berkowitz, L.; List, G.; Hackley, E. B.; Gillilan, W.; Pankau, M. Ageing and Dealkylation of Soman (Pinacolylmethylphosphonofluoridate)-Inactivated Eel Cholinesterase. Arch. Biochem. Biophys. 1967, 121 (1), 29–34. (181) Bhattacharjee, A. K.; Marek, E.; Le, H. T.; Ratcliffe, R.; Demar, J. C.; Pervitsky, D.; Gordon, R. K. Discovery of Non-Oxime Reactivators Using an in Silico Pharmacophore Model of Reactivators for DFP-Inhibited Acetylcholinesterase. Eur. J. Med. Chem. 2015, 90, 209–220. https://doi.org/10.1016/j.ejmech.2014.11.013. (182) Valasani, K. R.; Chaney, M. O.; Day, V. W.; Yan, S. S. Acetylcholinesterase Inhibitors : Structure Based Design , Synthesis , Pharmacophore Modeling , and Virtual Screening. 2013. (183) Mizutani, M. Y.; Itai, A. Efficient Method for High-Throughput Virtual Screening Based on Flexible Docking: Discovery of Novel Acetylcholinesterase Inhibitors. J. Med. Chem. 2004, 47 (20), 4818–4828. https://doi.org/10.1021/jm030605g. (184) Kapková, P.; Stiefl, N.; Sürig, U.; Engels, B.; Baumann, K.; Holzgrabe, U. Synthesis, Biological Activity, and Docking Studies of New Acetylcholinesterase Inhibitors of the Bispyridinium Type. Arch. Pharm. (Weinheim). 2003, 336 (11), 523–540. https://doi.org/10.1002/ardp.200300795. (185) Pejchal, V.; Štěpánková, Š.; Pejchalová, M.; Královec, K.; Havelek, R.; Růžičková, Z.; Ajani, H.; Lo, R.; Lepšík, M. Synthesis, Structural Characterization, Docking, Lipophilicity and Cytotoxicity of 1-[(1R)-1-(6-Fluoro- 1,3-Benzothiazol-2-Yl)Ethyl]-3-Alkyl Carbamates, Novel Acetylcholinesterase and Butyrylcholinesterase Pseudo-Irreversible Inhibitors. Bioorg. Med. Chem. 2016, 24 (7), 1560–1572. https://doi.org/10.1016/j.bmc.2016.02.033. (186) Morris, G. M.; Green, L. G.; Radić, Z.; Taylor, P.; Sharpless, K. B.; Olson, A. J.; Grynszpan, F. Automated Docking with Protein Flexibility in the Design of Femtomolar “Click Chemistry” Inhibitors of Acetylcholinesterase. J. Chem. Inf. Model. 2013, 53 (4), 898–906. https://doi.org/10.1021/ci300545a. (187) Geromichalos, G. D.; Lamari, F. N.; Papandreou, M. A.; Trafalis, D. T.; Margarity, M.; Papageorgiou, A.; Sinakos, Z. Saffron as a Source of Novel Acetylcholinesterase Inhibitors: Molecular Docking and in Vitro Enzymatic Studies. J. Agric. Food Chem. 2012, 60 (24), 6131–6138. https://doi.org/10.1021/jf300589c. (188) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti Correlation- Energy Formula into a Functional of the Electron Density. Phys. Rev. B 1988, 37 (2), 785–789. https://doi.org/10.1103/PhysRevB.37.785. (189) Hariharan, P. C.; Pople, J. A. The Influence of Polarization Functions on Molecular Orbital Hydrogenation Energies. Theor. Chim. Acta 1973, 28 (3), 213– 222.

225 (190) Becke, A. D. Density‐functional Thermochemistry. III. The Role of Exact Exchange. J. Chem. Phys. 1993, 98 (7), 5648–5652. https://doi.org/10.1063/1.464913. (191) Tomasi, J.; Mennucci, B.; Cancès, E. The IEF Version of the PCM Solvation Method: An Overview of a New Method Addressed to Study Molecular Solutes at the QM Ab Initio Level. J. Mol. Struct. THEOCHEM 1999, 464 (1–3), 211–226. https://doi.org/10.1016/S0166-1280(98)00553-3. (192) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Petersson, G. A.; Nakatsuji, H.; et al. Gaussian 16. Gaussian Inc.: Wallingfor, CT 2016. (193) Dennington, R.; Keith, T. A., Millam, J. M. GaussView. Semichem Inc.: Shawnee Mission, KS 2016. (194) Becke, A. D. A New Mixing of Hartree-Fock and Local-Density-Functional Theories. J. Chem. Phys. 1993, 98 (2), 1372–1377. https://doi.org/10.1063/1.464304. (195) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti Correlation- Energy Formula into a Functional of the Electron Density. Phys. Rev. B 1988, 37 (2), 785–789. https://doi.org/10.1103/PhysRevB.37.785. (196) Hariharan, P. C.; Pople, J. A. The Effect of D-Functions on Molecular Orbital Energies for Hydrocarbons. Chem. Phys. Lett. 1972, 16 (2), 217–219. https://doi.org/10.1016/0009-2614(72)80259-8. (197) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. Universal Solvation Model Based on Solute Electron Density and on a Continuum Model of the Solvent Defined by the Bulk Dielectric Constant and Atomic Surface Tensions. J. Phys. Chem. B 2009, 113 (18), 6378–6396. https://doi.org/10.1021/jp810292n. (198) Singh, U. C.; Kollman, P. a. An Approach to Computing Electrostatic Charges for Molecules. J. Comput. Chem. 1984, 5 (2), 129–145. https://doi.org/10.1002/jcc.540050204. (199) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; et al. Gaussian 09, Revision A.02. Gaussian Inc Wallingford CT. 2009, p Wallingford CT. https://doi.org/10.1159/000348293. (200) Morris, G.; Huey, R. AutoDock4 and AutoDockTools4: Automated Docking with Selective Receptor Flexibility. J. Comput. Chem. 2009, 30 (16), 2785–2791. https://doi.org/10.1002/jcc.21256.AutoDock4. (201) Eddleston, M.; Buckley, N. A.; Eyer, P.; Dawson, A. H. Management of Acute Organophosphorus Pesticide Poisoning. Lancet 2008, 371 (9612), 597–607. https://doi.org/10.1016/S0140-6736(07)61202-1. (202) Lotti, M. Clinical Toxicology of Anticholinesterase Agents in Humans. In Handbook of Pesticide Toxicology; Krieger, R. I., Krieger, W. C., Eds.; Academic Press: San Diego, 2001; pp 1043–1085. (203) Holmstedt, B. PHARMACOLOGY OF ORGANOPHOSPHORUS INHIBITORS. In Pharmacological Reviews; 1959; pp 567–688. (204) Macilwain, C. Study Proves Iraq Used Nerve Gas. Nature 1993, 363, 3.

226 (205) United Nations Mission to Investigate Allegations of the Use of Chemical Weapons in the Syrian Arab Republic; Hague, 2013. (206) Tu, A. T. Basic Information on the Nerve Gas and the Use of Sarin by Aum Shinrikyo. J. Mass Spectrom. Soc. Jpn. 1996, 44 (3), 293–320. (207) Sanson, B.; Nachon, F.; Colletier, J.; Toker, L.; Greenblatt, H. M.; Sussman, J. L.; Ashani, Y.; Masson, P.; Silman, I.; Weik, M. Crystallographic Snapshots of Nonaged and Aged Conjugates of Soman with Acetylcholinesterase , and of a Ternary Complex of the Aged Conjugate with Pralidoxime †,‡. J. Med. Chem. 2009, 52 (23), 7593–7603. https://doi.org/10.1021/jm900433t. (208) Fleisher, J. H.; Harris, L. W. DEALKYLATION AS A MECHANISM FOR AGING OF CHOLINESTERASE AFTER POISONING WITH. Biochem. Pharmacol. 1965, 14, 641–650. (209) Berends, F.; Posthumus, C. H.; Sluys, I. V. D.; Deirerkauf, F. A. The Chemical Basis of the “Ageing Process” of DFP-Inhibited Pseudocholinesterase. Biochem. Biophys. Acta 1959, 34, 576–578. (210) Shafferman, A.; Ordentlich, A.; Barak, D.; Stein, D.; Ariel, N.; Velan, B. Aging of Phosphylated Human Acetlcholinesterase: Catalytic Processes Mediated by Aromatic and Polar Residues of the Active Centre. Biochem. J. 1996, 318 (3), 833–840. (211) Thompson, D. C.; Perera, K.; London, R. Spontaneous Hydrolysis of 4- Trifluoromethylphenol to a Quinone Methide and Subsequent Protein Alkylation. Chem. Biol. Interact. 2000, 126 (1), 1–14. (212) Bolton, J. L.; Turnipseed, S. B.; Thompson, J. A. Influence of Quinone Methide Reactivity on the Alkylation of Thiol and Amino Groups in Proteins : Studies Utilizing Amino Acid and Peptide Models. Chem. Biol. Interact. 1997, 107 (3), 185–200. (213) McCracken, P. G.; Bolton, J. L.; Thatcher, G. R. J. Covalent Modification of Proteins and Peptides by the Quinone Methide from 2- Tert -Butyl-4,6- Dimethylphenol: Selectivity and Reactivity with Respect to Competitive Hydration. J. Org. Chem. 1997, 62 (6), 1820–1825. https://doi.org/10.1021/jo962088y. (214) Reboud-Ravaux, M.; Wakselman, M. Quinone Methides and Aza‐Quinone Methides as Latent Alkylating Species in the Design of Mechanism‐Based Inhibitors of Serine Proteases and Β‐Lactamases. In Quinone Methides; Rokita, S. E., Ed.; John Wiley & Sons Inc: Hoboken, 2009; pp 357–283. (215) Zhou, Q.; Qu, Y.; Mangrum, J. B.; Wang, X. DNA Alkylation with N - Methylquinolinium Quinone Methide to N 2 -DG Adducts Resulting in Extensive Stops in Primer Extension with DNA Polymerases and Subsequent Suppression of GFP Expression in A549 Cells. Chem. Res. inToxicology 2011, 24 (3), 402–411. (216) Amitai, G.; Adani, R.; Yacov, G.; Yishay, S.; Teitlboim, S.; Tveria, L.; Limanovich, O.; Kushnir, M.; Meshulam, H. Asymmetric Fluorogenic Organophosphates for the Development of Active Organophosphate Hydrolases with Reversed Stereoselectivity. Toxicology 2007, 233 (1–3 SPEC. ISS.), 187– 198. https://doi.org/10.1016/j.tox.2006.09.020.

227 (217) Heilbronn, E. Action of Fluoride on Cholinesterase-II. Biochem. Pharmacol. 1965, 14 (1), 1363–1373. (218) Clement, J. G.; Filbert, M. ANTIDOTE EFFECT OF SODIUM FLUORIDE AGAINST ORGANOPHOSPHATE POISONING IN MICE. Life Sci. 1983, 32 (16), 1803–1810. (219) Albanus, L.; Heilbronn, E.; Sundwall, A. ANTIDOTE EFFECT OF SODIUM FLUORIDE IN ORGANOPHOSPHORUS ANTICHOLINESTERASE POISONING. Biochem. J. 1965, 14, 1375–1381. (220) Hagstrom, D.; Zhang, S.; Ho, A.; Tsai, E. S.; Radić, Z.; Jahromi, A.; Kaj, K. J. Planarian Cholinesterase : Molecular and Functional Characterization of an Evolutionarily Ancient Enzyme to Study Organophosphorus Pesticide Toxicity. Arch. Toxicol. 2018, 92 (3), 1161–1176. (221) Ellman, G. L.; Courtney, K. D.; Andres, V.; Feather-Stone, R. M. A New and Rapid Colorimetric Determination of Acetylcholinesterase Activity. Biochem. Pharmacol. 1961, 7, 88–95. (222) Sun, J.; Lynn, B. C. Development of a LC / MS / MS Method to Analyze Butyrylcholinesterase Inhibition Resulting from Multiple Pesticide Exposure ଝ. J. Chromatogr. B 2009, 877 (29), 3681–3685. https://doi.org/10.1016/j.jchromb.2009.09.009. (223) Marsillach, J.; Richter, R. J.; Kim, J. H.; Stevens, R. C.; Maccoss, M. J.; Tomazela, D.; Suzuki, S. M.; Schopfer, L. M.; Lockridge, O.; Furlong, C. E. NeuroToxicology Biomarkers of Organophosphorus ( OP ) Exposures in Humans. Neurotoxicology 2011, 32 (5), 656–660. https://doi.org/10.1016/j.neuro.2011.06.005. (224) Keijer, J. H.; Wolring, G. Z. Stereospeciflc Aging of Phosphonylated Cholinesterases. Biochem. Biophys. Acta 1969, 6 (2), 465–468. (225) Radic, Z.; Kalisiak, J.; Fokin, V. V.; Sharpless, K. B.; Tatlor, P. Interaction Kinetics of Oximes with Native, Phosphylated and Aged Human Acetylcholinesterase. Chem. Biol. Interact. 2010, 187 (1–3), 163–166. https://doi.org/10.1158/0008-5472.CAN-10-4002.BONE. (226) Costanzi, S.; Machado, J.; Mitchell, M. Nerve Agents: What They Are, How They Work, How to Counter Them. ACS Chem. Neurosci. 2018, 9, 873–885. https://doi.org/10.1021/acschemneuro.8b00148. (227) Worek, F.; Wille, T.; Koller, M.; Thiermann, H. Toxicology of Organophosphorus Compounds in View of an Increasing Terrorist Threat. Arch. Toxicol. 2016, 90 (9), 2131–2145. https://doi.org/10.1007/s00204-016-1772-1. (228) Cannard, K. The Acute Treatment of Nerve Agent Exposure. 2006, 249, 86–94. https://doi.org/10.1016/j.jns.2006.06.008. (229) Luo, C.; Tong, M.; Chilukuri, N.; Brecht, K.; Maxwell, D. M.; Saxena, A. An In Vitro Comparative Study on the Reactivation of Nerve Agent-Inhibited Guinea Pig and Human Acetylcholinesterases by Oximes. 2007, 11771–11779. https://doi.org/10.1021/bi701002f. (230) Thompson, C. M.; Gerdes, J. M. Positron Emission Tomography Studies of Organophosphate Chemical Threats and Oxime Countermeasures. Neurobiol. Dis.

228 2019, No. March. (231) Cadieux, C. L.; Wang, H.; Zhang, Y.; Koenig, J. A.; Shih, T.-M.; McDonough, J.; Koh, J.; Cerasoli, D. Probing the Activity of a Non-Oxime Reactivator for Acetylcholinesterase Inhibited by Organophosphorus Nerve Agents. Chem. Biol. Interact. 2016, 1–9. https://doi.org/10.1016/j.cbi.2016.04.002. (232) Acetylcholinesterase, H.; Zhuang, Q.; Franjesevic, A. J.; Coldren, W. H.; Fernandez, J. P.; Nosseir, O.; Sillart, S.; Deyong, A.; Scarpitti, B. T.; Joseph, J.; et al. Novel Chiral Compound Capable of Resurrecting Organophospho-. (233) Roman, G. Mannich Bases in Medicinal Chemistry and Drug Design. 2015. https://doi.org/10.1016/j.ejmech.2014.10.076. (234) Stille, J. K. The Palladium-Catalyzed Cross-Coupling Reactions of Organotin Reagents with Organic Electrophiles. Angew. Chemie - Int. Ed. 1986, 25, 508–524. (235) Keijer, J. H.; Wolring, G. Z.; De Jong, L. P. A. Effect of PH, Temperature and Ionic Strength on the Aging of Phosphonylated Cholinesterases. Biochem. Biophys. Acta 1974, 334, 146–155. (236) Ellman, G. L. Tissue Sulfhydryl Groups. Arch. Biochem. Biophys. 1959, 82 (1), 70–77. https://doi.org/10.1016/0003-9861(59)90090-6. (237) Rassolov, V. A.; Ratner, M. A.; Pople, J. A.; Redfern, P. C.; Curtiss, L. A. 6-31G* Basis Set for Third-Row Atoms. J. Comput. Chem. 2001, 22 (9), 976–984. https://doi.org/10.1002/jcc.1058. (238) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. R. Efficient Diffuse Function-Augmented Basis Sets for Anion Calculations. III. The 3-21+G Basis Set for First-Row Elements, Li-F. J. Comput. Chem. 1983, 4 (3), 294–301. https://doi.org/10.1002/jcc.540040303. (239) Marenich, A. V; Cramer, C. J.; Truhlar, D. G. Universal Solvation Model Based on Solute Electron Density and on a Continuum Model of the Solvent Defined by the Bulk Dielectric Constant and Atomic Surface Tensions. 2009, 6378–6396. (240) Tomasi, J.; Mennucci, B.; Cammi, R. Quantum Mechanical Continuum Solvation Models. Chem. Rev. 2005, No. 105, 2999–3094. https://doi.org/10.1021/cr9904009. (241) Krishnan, R.; Binkley, J. S.; Seegar, R.; Pople, J. A. Self-Consistent Molecular Orbital Methods . XX . A Basis Set for Correlated Wave Functions. J. Chem. Phys. 1980, 72, 650–654. https://doi.org/10.1063/1.438955. (242) Frisch, M. J.; Pople, J. A.; Binkley, J. S. Self-Consistent Molecular Orbital Methods 25 . Supplementary Functions for Gaussian Basis Sets. J. Chem. Phys. 1984, 80, 3265–3269. https://doi.org/10.1063/1.447079. (243) Zhao, Y.; Truhlar, D. G. The M06 Suite of Density Functionals for Main Group Thermochemistry , Thermochemical Kinetics , Noncovalent Interactions , Excited States , and Transition Elements : Two New Functionals and Systematic Testing of Four M06-Class Functionals and 12 Other Fun. Theor. Chem. Acc. 2008, No. 120, 215–241. https://doi.org/10.1007/s00214-007-0310-x. (244) Montgomery, J. A.; Frisch, M. J.; Ochterski, J. W.; Peterson, G. A. A Complete Basis Set Model Chemistry . VII . Use of the Minimum Population Localization Method. J. Chem. Phys. 2000, No. 112, 6532–6542.

229 https://doi.org/10.1063/1.481224. (245) Allgardsson, A.; Berg, L.; Akfur, C.; Hörnberg, A.; Worek, F.; Linusson, A.; Ekström, F. J. Structure of a Prereaction Complex between the Nerve Agent Sarin, Its Biological Target Acetylcholinesterase, and the Antidote HI-6. Proc. Natl. Acad. Sci. 2016, 113 (20), 5514–5519. https://doi.org/10.1073/pnas.1523362113. (246) Trott, O.; Olson, A. J. AutoDock Vina: Improving the Speed and Accuracy of Docking with a New Scoring Function, Efficient Optimization, and Multithreading. J. Comput. Chem. 2009, 31 (2), 455–461. https://doi.org/10.1002/jcc.21334. (247) Case, D. A.; Ben-Shalom, I. Y.; Brozell, S. R.; Cerutti, D. S.; Cheatham, T. E. I.; Cruzeiro, V. W. D.; Darden, T. A.; Duke, R. E.; Ghoreishi, D.; Gilson, M. K.; et al. AMBER 2018. 2018. (248) Ashani, Y.; Gupta, R. D.; Goldsmith, M.; Silman, I.; Sussman, J. L.; Tawfik, D. S.; Leader, H. Stereo-Specific Synthesis of Analogs of Nerve Agents and Their Utilization for Selection and Characterization of Paraoxonase (PON1) Catalytic Scavengers. Chem. Biol. Interact. 2010, 187 (1–3), 362–369. https://doi.org/10.1016/j.cbi.2010.02.039. (249) Colovic, M. B.; Krstic, D. Z.; Lazarevic-Pasti, T. D.; Bondzic, A. M.; Vasic, V. M. Acetylcholinesterase Inhibitors: Pharmacology and Toxicology. Curr. Neuropharmacol. 2013, 11 (3), 315–335. https://doi.org/10.2174/1570159X11311030006. (250) Franjesevic, A. J.; Sillart, S. B.; Beck, J. M.; Vyas, S.; Callam, C. S.; Hadad, C. M. Resurrection and Reactivation of Acetylcholinesterase and Butyrylcholinesterase. 2019, 5337–5371. https://doi.org/10.1002/chem.201805075. (251) Carletti, E.; Li, H.; Li, B.; Ekström, F.; Nicolet, Y.; Loiodice, M.; Gillon, E.; Froment, M. T.; Lockridge, O.; Schopfer, L. M.; et al. Aging of Cholinesterases Phosphylated by Tabun Proceeds through O-Dealkylation. J. Am. Chem. Soc. 2008, 130 (47), 16011–16020. https://doi.org/10.1021/ja804941z. (252) Li, Y.; Du, L.; Hu, Y.; Sun, X.; Hu, J. Theoretical Study on the Aging and Reactivation Mechanism of Tabun-Inhibited Acetylcholinesterase by Using the Quantum Mechanical / Molecular Mechanical Method. Can. J. Chem. 2012, 90 (4), 376–383. https://doi.org/10.1139/v2012-007. (253) Franjesevic, A. J.; Zhuang, Q.; Nosseir, O.; Fernandez, J. P.; Ward, N. A.; Allen, S. K.; Weaver, J. D.; Kode, R.; Yang, D. M.; Davis, S. N.; et al. Structure-Activity Relationships for Substituted 3-Pyridinol Man- Nich Bases in the Resurrection of Methylphosphonate-Aged Human Acetylcholinesterase. 2019. (254) Roca, C.; Requena, C.; Sebastian-Perez, V.; Malhotra, S.; Radoux, C.; Perez, C.; Martinez, A.; Antonio Paez, J.; Blundell, T. L.; Campillo, N. E. Identification of New Allosteric Sites and Modulators of AChE through Computational and Experimental Tools. J. Enzyme Inhib. Med. Chem. 2018, 33 (1), 1034–1047. https://doi.org/10.1080/14756366.2018.1476502. (255) Daina, A.; Michielin, O.; Zoete, V. SwissADME : A Free Web Tool to Evaluate Pharmacokinetics , Drug- Likeness and Medicinal Chemistry Friendliness of Small Molecules. Sci. Rep. 2017, 7 (42717), 1–13. https://doi.org/10.1038/srep42717.

230 (256) ACD/Percepta. Advanced Chemistry Development, Inc.: Toronto 2019.

231