Synthesis and Bioactivity Investigation of Bridged Bicyclic Compounds and A

Synthesis and Bioactivity Investigation of Bridged Bicyclic Compounds and a

Mechanistic Investigation of a Propargyl Hydrazine Cycloaddition Catalyzed by an

Ammonium Salt

Elijah St.Germain

A Dissertation Submitted to the Faculty of the

Charles E. Schmidt College of Science

In Partial Fulfilment of the Requirements for the Degree of

Doctor of Philosophy

Florida Atlantic University

Boca Raton, FL

August 2018

Acknowledgements

I would like to express sincere gratitude to my committee members for their invaluable critiques and advice, and especially to my advisor Dr. Salvatore Lepore during the preparation of this manuscript. I would like to thank the National Institute of Health for financial support, the Dawson-Scully research group for their collaboration, and my lab mates in the Lepore group for their support and camaraderie.

iv Abstract

Author: Elijah J. St.Germain

Title: Synthesis and Bioactivity Investigation of Bridged Bicyclic Compounds and a Mechanistic Investigation of a Propargyl Hydrazine Cycloaddition Catalyzed by an Ammonium Salt

Institution: Florida Atlantic University

Thesis Advisor: Dr. Salvatore D. Lepore

Degree: Doctor of Philosophy

Year: 2018

We report the development of a general route to the synthesis of [4.3.1], [3.3.1], an especially [3.2.1] bicyclic compounds structurally related to vitisinol D, a natural product.

This allows for diastereoselective synthesis of bicyclic compounds with five adjacent chiral centers. This route was employed in a preliminary SAR investigation into the neuroprotectant effect of small molecules in an in vivo experiment measuring the degree of restorative effect of synaptic transmission in the neuromuscular junction of Drosophila melanogaster larvae under acute oxidative stress. One of the compounds exhibited intriguing potential as a neuroprotectant and outperformed resveratrol in restoring synaptic function under oxidative stress. The hypothesis that bridged bicyclic compounds may hold promise as drug scaffolds due to their conformational rigidity and ability to orient functional appendages in unique orientations is developed.

v The second focus is a mechanistic investigation into a tetrabutylammonium-catalyzed cycloaddition as evidence of a novel ammonium-alkyne interaction. A carbamate nitrogen adds to a non-conjugated carbon–carbon triple bond under the action of an ammonium catalyst leading to a cyclic product. Studies in homogeneous systems suggest that the ammonium agent facilitates cyclitive nitrogen–carbon bond formation through a cation–π interaction with the alkyne unit. Using Raman spectroscopy, this cation–π interaction is directly observed for the first time. DFT modeling elucidated the mechanistic factors in this cycloaddition.

A teaching experiment was developed based on this mechanistic investigation. Control experiments were employed to demonstrate the testing of two alternative mechanistic hypotheses. Cyclization reactions were performed with a soluble base (sodium phenoxide) with and without tetrabutylammonium bromide under homogeneous conditions. Students observed that ammonium salt accelerates the reaction. They were encouraged to develop a testable hypothesis for the role of the ammonium salt in the cyclization mechanism: typical phase transfer or other. IR spectroscopy was used to directly observe a dose dependent shift of the alkyne stretching mode due to a cation−π interaction. Undergraduates were able to employ the scientific method on a contemporary system and see how data are generated and interpreted to adjudicate between rival hypotheses in a way that emulates authentic and current research in a lab setting.

vi Dedication

This work is dedicated to my wife Tatiana and my daughter Phaedra.

Synthesis and Bioactivity Investigation of Bridged Bicyclic Compounds and a

Mechanistic Investigation of a Propargyl Hydrazine Cycloaddition Catalyzed by an

Ammonium Salt

List of Tables ...... xii

List of Figures ...... xiii

List of Schemes ...... xvi

Chapter 1 ...... 1

Annulation of allenoates to form [4.3.1], [3.3.1], and [3.2.1] bridged bicycles and

medium-sized carbocycles ...... 1

1.1 Introduction ...... 1

1.1.1 Biological significance of bridged bicyclic compounds...... 1

1.1.2 Synthesis of [4.3.1] systems ...... 3

1.1.3 Synthesis of [3.3.1] systems ...... 4

1.1.4 Synthesis of [3.2.1] bicyclic systems...... 10

1.1.5 Synthesis of medium sized ring compounds ...... 13

1.2 Method Development ...... 14

1.2.1 Synthetic strategy ...... 14

1.2.2 Synthesis of bicyclo[4.3.1]decane-2,10-diones ...... 15

1.2.3 Generality study and synthesis of medium-sized rings ...... 17

1.3 Medicinal investigation of compounds ...... 20

1.4 Conclusion ...... 22

viii 1.5 Experimental section ...... 23

Chapter 2 ...... 36

Bioactivity Investigation of [3.2.1] bicyclic compounds as modified functional

isomorphs of resveratrol and vitisinol D in a neuroprotective model ...... 36

2.1 Introduction ...... 36

2.1.1 Biological relevance ...... 36

2.1.2 Phenotypic models of anoxic shock and oxidative stress ...... 36

2.1.3 Small molecule neuroprotectants ...... 38

2.1.4 Possible mechanisms of neuroprotection ...... 40

2.1.5 Vitisinol D hypothesis ...... 43

2.1.6 Resveromorph conception for potential sirtuin activation ...... 47

2.1.7 Other design considerations...... 49

2.2 Synthesis of vitisinol D analogues ...... 53

2.2.1 Synthesis of cyclic β-ketoester ...... 53

2.2.2 Synthesis of allenoates...... 53

2.2.3 Convergent synthesis of bicyclo[3.2.1]oct-2-ene-4,8-diols...... 56

2.2.4 Diastereoselectivity and mechanistic hypothesis ...... 57

2.2.5 End stage synthesis ...... 61

2.3 Biological testing...... 63

2.4 Conclusion ...... 66

2.5 Experimental Section ...... 67

Chapter 3 ...... 76

Ammonium catalyzed cycloadditions and evidence for a cation-π ...... 76

ix interaction with alkynes ...... 76

3.1 Introduction ...... 76

3.1.1 Cation-π Interaction ...... 76

3.1.2 Cyclization reactions ...... 78

3.1.3 Synthesis of propargyl hydrazines ...... 80

3.2 Mechanistic Investigation ...... 81

3.2.1 Initial Transition State Hypotheses ...... 81

3.2.2 Rate experiments ...... 83

3.2.3 Raman evidence ...... 85

3.2.4 DFT calculations...... 89

3.2.5. Role of the tetraalkylammonium salt ...... 95

3.3 Discussion ...... 98

3.4 Conclusion ...... 103

3.5 Experimental Section ...... 105

Chapter 4 ...... 106

Teaching Lab Module to Elucidate a Possible Cation-π Interaction in an

Intramolecular Ammonium-Alkyne Cyclization ...... 106

4.1 Introduction ...... 106

4.1.1 The cation-π interaction in chemical education ...... 106

4.1.2 Evidence supporting a hypothesis ...... 109

4.1.3 The role of spectroscopy ...... 110

4.2 Experimental design and results ...... 111

4.3 Discussion ...... 115

x 4.3.1 Outcome of teaching module ...... 115

4.3.2 Evaluation of pedagogical goals ...... 119

4.4. Conclusion ...... 123

Appendixes ...... 124

Appendix A: Selected Spectra...... 125

Appendix B: Raman data (Table 5)...... 162

xi List of Tables

Table 1. Generality Study...... 17

Table 2. Ring opened compounds ...... 18

Table 3. Selected assay results of simple [3.n.1] bicycles ...... 20

Table 4. Role of base in TBAB-catalyzed cyclization of 50a under homogeneous

conditions...... 84

Table 5. Raman measurements of alkynes with added TBAB ...... 88

Table 6. Selected NBO changes due to presence of TBAB ...... 93

Table 7. Student Rate Experiments ...... 117

xii List of Figures

Figure 1. Select bridged bicyclic diketone natural products ...... 2

Figure 2. Concentration response curve for compound in mSTC GLP-1 secretion

assay ...... 211

Figure 3. Compounds used in a previous in vivo neuroprotection assay ...... 38

Figure 4. Small molecule neuroprotectant drug edaravone ...... 39

Figure 5. Example of free radical scavenging mechanism in representative

compound resveratrol...... 400

Figure 6. Natural polyphenols that activate SIRT1 in vitro ...... 411

Figure 7. Representative examples of highly conjugated, ‘flat’ STAC design...... 433

Figure 8. Vitisinol D analogues as all-carbon scaffolds with dense architectural

complexity...... 444

Figure 9. Comparison of structures and activities of selected resveratrol dimers ...... 46

Figure 10. Similarity in distance between key binding moieties of fisetin to

hypothetical vitisinol D analogue ...... 47

Figure 11. Resveramorph and isoresveramorph concept ...... 48

Figure 12. All carbon bridged bicyclic scaffold in acetylcholinesterase inhibitors ...... 500

Figure 13. All carbon bridged bicyclic scaffold in an antiviral compound ...... 500

Figure 14. All carbon bridged bicyclic scaffold in an improved estrogen receptor

reguator ...... 511

xiii Figure 15. General scaffold of vitisinol D analogues with quantified architectural

complexity...... 522

Figure 16. Assignment of observed relative stereochemistry ...... 611

Figure 17. Initial screening of compounds against resveratrol for neuroprotection

against oxidative stress in Drosophila melanogaster (vertical bars shown

as mean  S.E.M, 2-4 replicates each)...... 633

Figure 18. Dose-response plot of most active compound against resveratrol (vertical

bars shown as mean  S.E.M, 2-4 replicates each)...... 65

Figure 19. Cation-π Interaction of Arenes, and Proposed Ammonium-Alkyne

Interaction ...... 778

Figure 20. Complexation of cation in crown ether involving alkynyl sidearms ...... 78

Figure 21. Transition state analysis to account for the role of ammonium...... 822

Figure 22. Kinetic resolution catalyzed by a chiral ammonium catalyst ...... 83

Figure 23. Raman titration of alkynyl hydrazine with TBAB ...... 86

Figure 24. Raman spectra showing blue shift of N-H stretch with added TBAB ...... 89

Figure 25. Computational evidence for a conformational change in the presence of

TBAB ...... 91

Figure 26. Computational results in deprotonated system with and without TBAB ...... 92

Figure 27. Computational results in deprotonated system with and without TBAB ...... 94

Figure 28. Relative thermodynamic stability of reactive intermediates ...... 94

Figure 29. Tetralkyl bromide linker used to preorganize and stabilize reaction

complex ...... 97

xiv Figure 30. Representation of cation-π and anion-π interaction with an alkyne’s

quadrupole moment ...... 98

Figure 31. Baldwin’s rules modification for 5-endo-dig cycloadditions and aromatic

TS ...... 99

Figure 32. Comparison of theoretical predictions with experimental results ...... 101

Figure 33. Mechanistic hypothesis for tetraalkylammonium assisted 5-endo-dig

cycloaddition ...... 102

Figure 34. Cation-π interactions involving arenes and alkynes as π-systems ...... 107

Figure 35. Evidence of ammonium salt having a catalytic role in the cyclization ...... 111

Figure 36. Phase transfer mechanism as a hypothesis of TBAF catalysis ...... 112

Figure 37. Presentation of competing hypotheses and falsifiable predictions ...... 113

Figure 38. Representative Student IR Titration of 116a with TBAB ...... 118

xv List of Schemes

Scheme 1. Platinum catalyzed enolate arylation approach to the [4.3.1] bicyclic core ..... 4

Scheme 2. Indolyne cyclization approach to the [4.3.1] bicyclic core ...... 4

Scheme 3. Nicolaou’s early approach to bicyclo[3.3.1]nonanes ...... 5

Scheme 4. Tandem Michael/Aldol approach to bicyclo[3.3.1]nonanes and

[3.2.1]octanes ...... 6

Scheme 5. Mehta’s oxidative cyclization approach to bicyclo[3.3.1]nonane ...... 7

Scheme 6. Iodocarbocyclization approach to bicyclo[3.3.1]nonan-diones ...... 7

Scheme 7. Shibasaki’s ring closing metathesis approach to bicyclo[3.3.1]nonane ...... 8

Scheme 8. Alkylative dearomatization/annulation ...... 9

Scheme 9. Synthesis of bicyclo[3.3.1]nona-9-one via reductive aldol rearrangement ...... 9

Scheme 10. Dixon’s synthesis of bicyclo[3.2.1]alkenediones via medium-sized rings ... 11

Scheme 11. Diastereoselective Michael/aldol synthesis of 2-hydroxy-

bicyclo[3.2.1]nonan-8-ones ...... 11

Scheme 12. Michael-Stork entry into bridged bicyclo[3.2.1]octenones ...... 12

Scheme 13. Previous model synthesis of vitisinol D in the Lepore Group ...... 13

Scheme 14. Synthesis of hydroazulene medium sized rings ...... 14

Scheme 15. Reductive Aldol approach to bicyclic molecules ...... 15

Scheme 16. Synthetic approach to [4.3.1] bicycles and medium sized ring

derivatives ...... 16

xvi Scheme 17. Wittig Approach to Allenoates and Generality...... 19

Scheme 18. Synthesis of cyclic β-ketoester ...... 53

Scheme 19. Alkynylogous addition leading to thermodynamic γ-carbinol product ...... 54

Scheme 20. TBAF mediated double addition leading to bis-γ-carbinol quaternary

allenoates...... 55

Scheme 21. MBH-Brooks reaction to form γ-carbinol ...... 55

Scheme 22. Direct and regioselective substitution at the γ-carbon ...... 56

Scheme 23. Convergent synthesis of mono and bis aryl lactone intermediates ...... 57

Scheme 24. Diastereoselective mechanism of addition/annulation of allenoates to β-

aryl cyclopentyl ketoester ...... 58

Scheme 25. Proposed diastereoselective mechanism in reductive aldol rearrangement .. 59

Scheme 26. End stage synthesis ...... 62

Scheme 27. Direct deprotection of diols ...... 62

Scheme 28. Selected reactions stabilized by cation-π interactions ...... 77

Scheme 29. Previous discovery from the Lepore Group ...... 79

Scheme 30. Precedents for TBAF catalyzed cycloadditions ...... 79

Scheme 31. Synthesis of β-alkynyl hydrazines and azaprolines ...... 81

Scheme 32. Tandem cycloaddition/aldehyde addition trapping vinyl anion ...... 95

Scheme 33. Cation templated synthesis of potassium 18-crown-6 ether ...... 96

xvii Chapter 1

Annulation of allenoates to form [4.3.1], [3.3.1], and [3.2.1] bridged bicycles and

medium-sized carbocycles

1.1 Introduction

1.1.1 Biological significance of bridged bicyclic compounds

Efforts to synthesize natural products featuring highly substituted bridged bicycles with all-carbon skeletons have yielded a wide array of synthetic approaches. Bridged bicyclic natural products have received considerable attention in recent years owing to their architectural complexity and biological activity. Among this class, vitisinol D (1), containing a bicyclo[3.2.1]octane system, exhibits modest blood clot inhibitory activity.1

This natural product has never succumbed to total synthesis and thus attracted our interest as a challenging synthesis target with potentially interesting biological activity (Figure 1).

Vitisinol D, a bicyclo[3.2.1]non-2,9-dione isolated from the roots of Vitis thunbergii is interesting in that it is a dimer of resveratrol. Typically, dimerization does not proceed with dearomatization of one of the phenyl rings. Vitisinol D is unique in that the dearomative dimerization results in a bridged bicyclic compound. To date, there is not a proposed biosynthesis of vitisinol D in the literature.

Figure 1. Select bridged bicyclic diketone natural products The synthesis of oxygenated bicyclo[3.3.1]alkanes has been the focus of considerable synthetic efforts due mainly to the promising biological activity of many members of the large family of natural products known as polycyclic polyprenylated acylphloroglucinols

(PPAPs).2 With [3.3.1] bicyclic natural products of medicinal interest being more numerous than [3.2.1] or [4.3.1] systems, there have been a number of successful total syntheses of [3.3.1] bicyclic phloroglucinols, with some recent examples being the work of Mehta,3 Plietker, Bieber, and Horeischi.4 Several general strategies have been deployed towards the synthesis of general classes of these bicyclic compounds.5

These PPAP compounds, isolated from plants in the family Guttiferae, are biosynthesized from monocyclc polyprenylated acylphloroglucinols (MPAPs), which occur in plants from the families Myrtaceae and Cannabinaceae. The majority of PPAPs are bicyclo[3.3.1]nonanes with a ketone on the 2-carbon (3-carbon bridge) and the 9-carbon

(see generic carbon skeleton, Figure 1). Often the 4-carbon is oxygenated as well. Although

2 less common, there are also PPAPs with a bicyclo[3.2.1]octane backbone. Notable examples of PPAPs include hyperforin (4, Figure 1), the active antidepressant compound in St. John’s Wort (Hypericum perforatum), and the potential anti-Alzheimer’s compound garsubellin A (2, Figure 1). Compounds biologically unrelated to the PPAPs but structurally similar with respect to the oxidized carbon skeleton include vitisinol D (1) and

N-methylwelwitindolinone D isonitrile (3), a bicyclo[4.3.1]deca-2,10-dione compound isolated from blue green algae that is part of a family of alkaloids of interest as potential reversers of drug resistance in cancer models.

1.1.2 Synthesis of [4.3.1] systems

Just as many of the strategies for the synthesis of [3.2.1] and [3.3.1] bridged bicyclic compounds have been developed in the service of efforts to access natural products such as vitisinol D and the polyprenylated acylphloroglucinols, so too have many of the recent strategies for the synthesis of [4.3.1] bridged bicyclic compounds been developed in the pursuit of the total synthesis of the welwitindolidine family of natural products (see Figure

1). The distinctive structural feature of most welwitindolidines is the fusion of an oxindole group at the 3 and 4 position with a cyclohexanone at the α and α’ positions of the latter to form the bicyclo[4.3.1]decanone skeleton. A large amount of synthetic effort has been devoted to the synthesis of these compounds6. Because the chemistry involved in the fusion of an aromatic indole to a cyclic ketone is often very different from the reaction mechanisms used to synthesize hypothetical [4.3.1] versions of non-welwitindolonone bicyclic compounds which are the concern of this thesis, only a few representative examples from the recent literature of the formation of the [4.3.1] bicycle in the synthesis of the welwitindolidines will be considered here.

Scheme 1. Platinum catalyzed enolate arylation approach to the [4.3.1] bicyclic core In the ﬁrst total synthesis of N-methylwelwitindolinone D isonitrile, Rawal reported a palladium catalyzed approach to building the [4.3.1] bicyclic core of intermediate 6 from bromoindole 5 (Scheme 1).7 An approach to the total synthesis of (−)-N- methylwelwitindolinone C isonitrile by Garg involved a similar bromoindole intermediate.8 In this approach, the critical second fusion of the indole to the cyclohexanone ring was accomplished through an indolyne cyclization of intermediate 7 yielding compound 8 (Scheme 2).

Scheme 2. Indolyne cyclization approach to the [4.3.1] bicyclic core 1.1.3 Synthesis of [3.3.1] systems

The synthetic approaches to the construction of [3.3.1] bicycles almost invariably involve the installation of a 3-carbon bridge, joining a 3-carbon fragment to a cyclic ketone at the

α and α’ positions. Depending on the functional groups present on the synthetic intermediates, this raises a number of problems that have been confronted by researchers.

An early example by Nicolaou9 (Scheme 3), the first to synthesize a PPAP, featured a selenium-mediated cyclization on substrate 9 using SnCl4 and N-

4 (phenylseleno)phthalimide at -23 °C, leading to the bicyclo[3.3.1]nonane 10 in good yield

(95%).

Scheme 3. Nicolaou’s early approach to bicyclo[3.3.1]nonanes An exclusive preference for the exo diastereomer pictured was observed. In bridged bicyclic systems, exo denotes orientation of a substituent facing the shorter of the other two bridges, while endo substituents face the longer of the remaining bridges (see insert in

Scheme 3).

The problem that Nicolaou set out to solve is the introduction of sterically crowding substitution on the C1-C5 positions of garsubellin A, namely a quaternary bridgehead at

C1, a gem-dimethyl on C2, a tertiary prenylated carbon on C3, and a quaternary carbon on the other bridgehead at C5. Due to the steric crowding of these substituents, and due critically to the necessity of having two quaternary bridgeheads, Nicolaou avoided an α,α’- annulation approach. Instead, he employed a selenium-mediated endocyclization of a pendent prenyl olefin through enolate attack at the α-position of the β-keto-ester (or β- diketone). The ensuing phenylselenium moiety at C3 would then lend itself to radical addition of the necessary prenyl group at that carbon. Although this method is tolerant of fused aromatic rings on the substrate, it does not lend itself to the construction of the C-C unsaturated 3-carbon bridge bearing the second oxygen (on C8) of the bicyclic skeleton of garsubellin A and related natural products. There were five steps in total leading up to this bicyclic dione intermediate.

5 Nicolaou would later advance a different approach to the bicyclo[3.3.1]nonane PPAP hyperforin10 in which an acid-catalyzed tandem Michael/aldol addition of α,β unsaturated aldehyde 11 (Scheme 4) at the α, α’ positions affords the bridged bicycle, but a further oxidation step is needed to produce diketone 13 with α,β-unsaturation on the installed 3- carbon bridge. Furthermore, this unsaturation is lost upon fragmentation giving the medium-sized ring (see section 1.1.6). This approach is generalizable for cyclopentandiones such as 12, leading to bicyclo[3.2.1]octa-2,8-diones, and also to larger bridged bicyclic compounds (see section 1.1.2).

Scheme 4. Tandem Michael/Aldol approach to bicyclo[3.3.1]nonanes and [3.2.1]octanes Mehta’s palladium-mediated oxidative cyclization was the first enantioselective approach to synthesis of a PPAP. This approach to bicyclo[3.3.1]nonanes for a model synthesis of nemorosome11 involves a palladium-mediated oxidative cyclization of silyl enol intermediate 14 to give 15 in modest yield (Scheme 5). His model synthesis featured the key step in which a vinyl pendant of enantiopure cyclic ketone was endocyclized at the adjacent α-position by way of a silyl enol ether intermediate. Oxidative cyclization using

Pd(OAc)2 led to the [3.2.1] bicycle, albeit in modest yields. An advantage of this cyclization is the direct installation of a 3-carbon bridge with C-C unsaturation, but a disadvantage is the absence of oxygen on the bridge, requiring more steps in an already lengthy synthesis (8 linear steps to (final) intermediate).

Scheme 5. Mehta’s oxidative cyclization approach to bicyclo[3.3.1]nonane In his total synthesis of garsubellin A, Danishefsky, building on the earlier work of

Nicolaou, employed an iodocarbocyclization. Intermediate 16 (Scheme 6) was treated with iodine and potassium iodide under basic conditions to give intermediate 17.12 The bicyclo[3.2.1]nonan-dione 17 was then further functionalized. This approach, specifically tailored towards the synthesis of garsubellin A, relies on the installation of a prenyl group at the α-position, and results in a gem-dimethyl moiety on the installed 3-carbon bridge without C-C unsaturation. The difference here (from Mehta’s approach for example) is that it is the unsaturated and more heavily alkyl substituted bridge that is being installed to form the [3.3.1] bicycle.

Scheme 6. Iodocarbocyclization approach to bicyclo[3.3.1]nonan-diones Iodocarbocyclization also results in iodine moieties on the bicycle, including one on the bridgehead, that are useful for functionalizing the molecule at those positions. A drawback of this approach in terms of general bridged bicyclo[3.3.1]nonane synthesis is that the construction of the bicyclic skeleton is very specifically tailored to garsubellin A and compounds that similarly feature a tetrahydrofuran ring fused to the enone bridge and adjacent bridgehead .

7 Shibasaki utilized the Grubbs-Hoveyda ring closing metathesis in the key step of his total synthesis of garsubellin A, the first racemic total synthesis of the natural product.13 In this relatively late-stage construction of the 3-carbon bridge leading to the bicyclo[3.3.1]nonane carbon skeleton, the α and α’ positions of cyclic ketone 18 (Scheme

7) are quaternary carbons bearing an allylic and vinyl group that are metathesized.

Oxidation and removal of the MOMO protecting group led to intermediate 19. Although the metathesis proceeds in excellent yield, the drawback of this approach is the difficulty of installing the pendant vinyl and allyl groups to begin with. These efforts required a 15- step synthesis up to the point of the cross metathesis substrate, and then three steps to install the three carbon unsaturated bridge and convert it into an enone bridge.

Scheme 7. Shibasaki’s ring closing metathesis approach to bicyclo[3.3.1]nonane

Recently, Porco has demonstrated a tandem alkylative dearomatization/annulation protocol that allowed the direct construction of the bicyclo[3.3.1]nonane1,3,5-trione core of their target (±)-clusianone and other PPAPs.14 Phloroglucinol 20 (Scheme 8) reacted with 21 through a tandem Michael-elimination/intramolecular Michael addition sequence to produce model target 22 as a mixture of enol-ethers shown, with mild (4:1) diastereoselectivity of the methyl ester in the exo orientation as shown.(regioselectvity)

The Porco group went on to show that the use of chiral Cinchona alkaloid-derived phase transfer catalysts can allow similar dearomative annulations to proceed enantioselectively,

8 as demonstrated by their total synthesis and assignment of absolute configuration to hyperibone K. 15.

Scheme 8. Alkylative dearomatization/annulation In an important precedent for our work, Mehta described in their study towards the synthesis of garsubellin A: a bicyclization involving a reductive aldol rearrangement mediated by DIBAL-H.16

Scheme 9. Synthesis of bicyclo[3.3.1]nona-9-one via reductive aldol rearrangement To achieve the necessary relative stereochemistry in this case, with prenyl groups in the exo orientation, methyl acrylate was added via a Michael addition to 24 after the second prenyl group was added to 23, forming the quaternary center (Scheme 9). This allowed steric facial selectivity of addition of the three-carbon chain, leading to a cis-orientation of the two prenyl units of six membered ring 25. Ester hydrolysis gave carboxylic acid 26, 9 which was subsequently converted to enol lactone 27. Perhaps the critical step in Mehta’s route was an aldol reaction where the reactive intermediate was created by the action of a reducing agent (DIBAL-H) to ultimately afford bicycle 28. As discussed in subsequent paragraphs, a similar cascade of diastereomeric induction is at play in the current approach towards the total synthesis of vitisinol D by Lepore and coworkers. One of the unique synthesis challenges of vitisinol D relative to garsubellin is the diastereoselective installation of four consecutive stereocenters on a more strained [3.2.1] bicycle. One drawback of this method is the lack of C-C unsaturation on the installed 3-carbon bridge

To achieve the necessary enone moiety in garsubellin A and many other similar target compounds, a Saegusa-Ito Pd-mediated eliminative dehydrosilylation had to be employed after oxidation of the hydroxyl group.

1.1.4 Synthesis of [3.2.1] bicyclic systems

Recently, Dixon put forward an approach to medium sized ring compounds and bicyclo[3.2.1]alkenediones.17 In an interesting contrast to the work presented in this thesis, the bicyclo[3.2.1]alkenedione is obtained as the thermodynamic product beginning with the homologous medium sized ring compound. In our approach, the synthesis of bicyclo[3.N.1]alkanones from cyclic ketones with an EWG at the α-position is followed by a ring opening leading to medium sized ring compounds. Dixon and coworkers synthesized ketodiester 29, a seven-membered ring, from 30 with cyclic α-ketoester 31 under basic conditions (Scheme 10). Krapcho dealkyloxycarbonylation led to the cyclic ketoester 32 that then underwent an acid-catalyzed intramolecular Dieckmann-type condensation to form the bicyclo[3.2.1]octanone 33. This method is functional group tolerant for a variety of alkyl and aromatic substituents forming bicyclo[3.2.1]octanones in

10 fair to good yield. It was demonstrated to be moderately effective at producing several bicyclo[3.3.1]nonanones in modest-to-fair yields.

Scheme 10. Dixon’s synthesis of bicyclo[3.2.1]alkenediones via medium-sized rings In an early precedent by Rodriguez,18 2-hydroxy-bicyclo[3.2.1]octan-8-ones 36 were assembled from simple intermediates (α-cyclopentanoic esters 34 and α,β-unsaturated aldehydes 35) through a base-promoted (K2CO3, Cs2CO3, or DBU) tandem Michael/Aldol reaction (Scheme 11). A general preference was noted for equatorial (endo) configuration of the hydroxyl on C2. Diastereoselectivity of prochiral aldehydes was demonstrated, and three-center diastereoselectivity was influenced by the orientation of the β-substituent of the endocyclic enolate.19

Scheme 11. Diastereoselective Michael/aldol synthesis of 2-hydroxy-bicyclo[3.2.1]nonan-8-ones

11 Lepore developed an entry to [3.2.1] bicyclic systems based on a Michael-Stork addition of cyclopentyl enamine to allenic ketones and esters.20 Taking advantage of the dense functionality of allenylcarbonyl compounds, this work showed that allenyl esters and ketones can be used in the α, α’ addition-type strategies discussed above to synthesize

[3.2.1]alkenones with exocyclic unsaturation on the 3-carbon bridge. The initial addition of enamine 38 to the β-position of the allenyl carbonyl 40 is selective for E double bond geometry with respect to the R group as shown (Scheme 12). The second, intramolecular addition of the enamine in 39 to the allenyl pendant, leading to annulation product 40, and to the product 41 upon morpholine removal, also proceeds diastereoselectively, with preference for the R1 group in the endo position. Although this work showed the potential of substituted alkenyl compounds in stereoselectively building bridged bicycles, it was not suitable for adaptation into a total synthesis of natural products such as vitisinol D because the method would not tolerate substitution on the cyclopentane ring.

Scheme 12. Michael-Stork entry into bridged bicyclo[3.2.1]octenones

12 Other efforts in the Lepore Group to create [3.2.1] bicycles involved a double enamine addition to an allenyl ketone.20 This led to an approach using phenyl allenyl esters for the synthesis of the dione system of vitisinol D anticipating that this approach would allow access to bicyclic diketone core A directly from phenyl allenyl esters 42 and cyclic ketones

43 in one step (Scheme 1). Maity and Lepore discovered that the preferred double addition product was the enol-lactone.21 However, this unsaturated intermediate 44 was successfully transformed to the corresponding [3.2.1] bicyclic structure via a two-step reductive aldol similar to that of Mehta22 using two equivalents of a reducing agent (Scheme 13). A subsequent oxidation step aﬀorded desired bicyclic diketone 45 in modest two-step yields.

With this precedent, we pursued various analogues of [3.2.1] bicycles, and began working on a general approach to [3.n.1] bicycles that would expand the generality of this approach.

Scheme 13. Previous model synthesis of vitisinol D in the Lepore Group 1.1.5 Synthesis of medium sized ring compounds

The ring expansion of carbocyclic β-ketoesters with acetylenic esters is one method of producing functionalized medium sized rings with endocyclic unsaturation. This expansion was pioneered by Frew and Proctor23, and is adaptable to form seven, eight, nine or even ten membered ring products starting from five, six, seven, or eight membered cyclic β- ketoesters, respectively. This is a method that has been more recently employed by Dixon

13 to access the medium sized ring compounds he went on to form into bridged bicyclic compounds (see section 1.1.4). One limitation of this method is that because it proceeds through a formal [2+2] addition of the alkyne moiety to the enolate or silyl enol, only the two alkynyl carbons can be incorporated into the ring.

Other earlier precedents include work by Miesch24, who developed a route to hydroazulene derivative 48 (Scheme 14) by similar cycloaddition of DMAD and ethyl propionate to the silyl enol ether of morpholino enamine 47 derived from bicycle[3.3.0]octane-2-one 46, followed by photocatalyzed ring expansion.

Scheme 14. Synthesis of hydroazulene medium sized rings

1.2 Method Development

1.2.1 Synthetic strategy

A robust method for constructing bicyclo[3.n.1]alkenediones through annulation of allenyl esters with cyclic ketones bearing an electron withdrawing group on the α-position was necessary for our synthetic efforts aimed at vitisinol D and medicinal analogues of that and similar natural products. We sought to develop a method that was general and allowed for various ring sizes to be annulated with allenyl esters that are variously substituted. For our methodology we chose β-phenylsulphonyl cycloalkanones with ring sizes from five to seven carbons. Phenyl allenyl esters proved necessary for the initial formation of the enol- lactonate, as with the Maity precedent.

Scheme 15. Reductive Aldol approach to bicyclic molecules

1.2.2 Synthesis of bicyclo[4.3.1]decane-2,10-diones

Initial studies on the synthesis of [4.3.1] bicyclic compounds began with the readily accessible 2-(phenylsulphonyl)-cycloheptanone 51 (Scheme 16).25 In the presence of

K2CO3, 51 reacted with phenyl allenyl ester 52 in acetone to afford enol lactone 53. This transformation likely proceeds via Michael addition to the reactive sp-hybridized allene- carbon, followed by ester-enolate cyclization and double bond isomerization. The reactions of allenyl esters with cycloheptanone substrates to produce lactones 53a and 53b were slow and of modest yield. Reflux conditions and longer reaction times needed for annulation of the initial α-addition product suggest that the lactone is the thermodynamic product.

Reaction times needed for annulation of the cycloheptanones (1-3days) were longer than

15 those needed for the cyclohexanone or cyclopentanone homologues made by others.

Scheme 16. Synthetic approach to [4.3.1] bicycles and medium sized ring derivatives This is plausibly due to an entropic penalty imposed by the larger ring size. Due to the longer reaction times needed, the allenoate was prone to being lost to dimerization, reducing yield. Two equivalents of allenoate were found to be conducive to a successful reaction. Lactones were further treated with LiAlH4 (LAH) at −50 °C to afford a diol intermediate as a mixture of diastereomers. Subsequent oxidation with Dess−Martin periodinane (DMP) resulted in bicyclic diketones 54a and 54b as the exclusive products in

43% and 36% two-step yields, respectively. NMR evidence suggests that various allene addition byproducts account for the low yield of desired lactone products with the 7- membered ring analogues, perhaps due to an entropic penalty from the larger ring size.

16 1.2.3 Generality study and synthesis of medium-sized rings

With a route to bridged bicycle 54 established, we next sought to convert it into a medium- sized ring structure. As part of this effort, we uncovered an interesting Grob-type fragmentation to convert 54 into functionalized monocyclic rings under mild conditions.

Previously we in treated 38 with TBAF and unexpectedly found that decarboxylative ring opening led to the ring opened product 39. We treated 54a with TBAF·3H2O in THF at room temperature, expecting to obtain an α-carboxylic acid ring-opened product. However, this reaction led to a rapid decarboxylation to afford compound 55a in 90% yield. We discovered that Et3N in MeOH led to highly conjugated medium-sized ring 56b in 95% yield, which retains the carbonyl from the original carbon framework of 54.

Table 1. Generality Study In collaboration with others in the Lepore group, we evaluated the scope of this annulation strategy using 2-(phenylsulphonyl)-cycloalkanones (57, n=0,1) and variously substituted phenyl allenyl esters 52 (Table 1). These reactions delivered lactones 58-59 in good yields

2 3 except with an α-substituted allenyl ester 52c (R = H 3;R = CH3). Speciﬁcally, the formation of lactone 58c with an α-substituted allene gave only a modest yield (48%) even after an extended reaction time (24 h). With few exceptions, the reduction of all lactones

17 (53, 58, 59) using LAH and their subsequent DMP oxidation led to the formation of bridged bicyclic diones (54, 60, 61) in reasonable two-step yields.

Table 2. Ring opened compounds Ring-opening reactions of bridged bicyclic are precedented and often an excellent entry into functionalized medium-sized ring systems.26 Medium sized rings are often difficult to synthesize, due in part to the entropic penalty working against ring-closing reactions. In direct contrast to the work of Dixon, in which unsaturated medium sized ring compounds are converted into bridged bicycles as the thermodynamically preferred products, in our case the bridged bicyclic alkenediol intermediates are formed in the reductive aldol step as the likely kinetic products. Thus it was not entirely unexpected when we discovered that treatment of the intermediates following mild oxidation to the alkenedione resulted in ring opening through a decarboxylation reaction upon treatment with TBAF. This was first observed in a model synthesis of vitisinol D analogue compounds from the ketoester

18 leading to intermediate 49 which ring opened to give 50 (Scheme 16). The same reactivity proved true of the sulfone compounds, and ring opening with TBAF was afforded under mild conditions in good to excellent yield. The use of TBAF·3H2O converted bicycles 54 and 60 into monocycles 55 and 63 with concomitant decarboxylation in excellent yields

(Table 2). Using MeOH/Et3N, a rapid ring-opening of diketones 54, 60 and 61 led to esters

56, 64 and 65 in good yields. This ring-opening reaction likely occurs via an attack of hydroxide or methanol on the carbonyl bridge, resulting in intermediate 62. Subsequent collapse of this tetrahedral intermediate leads to a sulphonyl-stabilized carbanion that is rapidly protonated to give monocyclic products. In the case of TBAF·3H2O-mediated fragmentation, decarboxylation is sometimes followed by a simple double bond migration to aﬀord more substituted double bond products. For reactions in methanol, the carbanion is immediately quenched by a proton to result in the formation of resonance stabilized dienol-containing rings 56, 64 and 65 exclusively.

Scheme 17. Wittig Approach to Allenoates and Generality 19 We employed a Wittig olefination approach to the synthesis of the phenyl allenoates needed for the annulation (Scheme 17). The dehydrative method developed by others in the group is not useful for synthesizing allenoates with the critical phenyl ester group needed as a leaving group. This is due in part to the lack of inexpensive, commercially available phenyl acetoacetate, and moreover because the phenyl ester is too reactive for the stepwise synthesis involving highly nucleophilic intermediates. For more detail on this alternate route, see the work of Maity and Lepore.21, 99

In this case, we were restricted in our choice of inexpensive, commercially available starting material. We tested the generality of the Wittig approach and found that alpha and gamma substituted products are obtainable, but attempts to synthesize a quaternary allenoate failed.

1.3 Medicinal investigation of compounds

Table 3. Selected assay results of simple [3.n.1] bicycles

20 To investigate the usefulness of simple [3.N.1] bicyclic compounds as medicinal probes, we partnered with Eli Lilly’s Open Innovation Drug Discovery program and our bicyclic and medium sized carbocycles were screened for a variety of therapeutic targets (Table 3).

The most significant results returned were from assays for glucagon-like peptide (GLP-1) secretogogues.

Glucagon-like peptide (GLP-1) is an important biomolecule related to diabetes therapies.

It is produced in the small intestine by the proglucagon gene in the presence of nutrients27.

GLP-1 then stimulates the release of insulin from the pancreatic islets in a nutrient dependent manner. In subjects with type 2 diabetes, GLP-1 secretion is decreased28.

Stimulating its secretion is therefore a logical goal of medicinal chemical research (other approaches include administration of the native or mimetic peptides or small molecules, or regulation of the degradation of GLP-1).

Figure 2. Concentration response curve for compound in mSTC GLP-1 secretion assay One target for small molecule mediated GLP-1 secretion is the G protein-coupled receptor

40 (GRP40). This receptor, when expressed in the gastric track, is associated with the

21 secretion of GLP-1. It is therefore a possible mechanism for small molecule secretagogues of GLP-1. A variety of compounds have been investigated as GRP40 agonists. Other compounds have been investigated more generally as GLP-1 secretagogues, including several generations of sulfonylurea drugs.29 These drugs often carry the risk of hypoglycemia, so there is a present need to continue to search for new compounds.

Several of our [3.n.1] diketone compounds as well as the medium sized ring derivatives returned intriguing activities in the initial rounds of screening for diabetes therapeutics via

GLP-1 secretion assays. Of those compounds that returned significant single point responses in the stimulation of GLP-1 from mouse intestinal cells (mSTC), dose-response curves were established, yielding the EC50 values: concentrations at which the test compounds stimulated half of the maximum stimulatory response. The best results (Table

3) were in the low double digit to single digit micromolar range. Considering the relative paucity of functionalization of these compounds, and the possibility of accessing more functionalized analogues through this method (the phenyl rings could bear a number of polar substituents, for example), these modest results seem to argue for the further exploration of this family of compounds as GLP-1 secretogogues.

1.4 Conclusion

The synthesis of these bicyclic compounds under mild conditions using inexpensive reagents (K2CO3 was chosen as the base to catalyze the addition and lactone formation of the allenoates and the cyclic ketosulphones with this consideration in mind) showcases the utility and potential diversifiability of this synthetic startagy for the construction of similar

[3.N.1] bicyclic compounds. Although our purpose was mainly to develop a reliable method for the construction of bicyclic[3.2.1]nonendiones and related compounds for

22 efforts towards the synthesis of vitisinol D and its analogues, the success of synthesizing even the difficult bicyclic[4.3.1]decendiones hints at the potential of this allenoate approach for diverse targets. One of the strengths of our method is that unlike most other

α, α’ double addition approaches to constructing bicyclic compounds from cyclic ketones, the use of allenoates as addition partners furnishes immediate unsaturation on the added three-carbon bridge. Another important advantage is that the reductive aldol rearrangement mechanism initially produces a diol diastereoselectively. Although this is not an advantage in the synthesis of vitisinol D, the following chapter will show that this feature of the method could prove useful in the design of medicinal compounds.

1.5 Experimental section

Reactions were carried out under an argon atmosphere (unless otherwise stated) in oven- dried glassware with magnetic stirring. Puriﬁcation of reaction products was carried out using ﬂash silica gel 40−63 μm. Analytical thin-layer chromatography was performed on

200 μm silica gel 60 F-254 plates. Visualization of TLC plates was accomplished with UV light, followed by staining with vanillin or potassium permanganate and drying with a heat gun. 1H NMR were recorded on a 400 MHz spectrometer and are reported in ppm (parts per million) using solvent as an internal standard (CDCl3 at 7.26 ppm). Data are reported as b = broad, s = singlet, d = doublet, t = triplet, q = quartet, p = pentet, m = multiplet; coupling constants in hertz (Hz). 13C NMR were recorded on a 100 MHz spectrometer.

Chemical shifts are reported in ppm, with solvent resonance employed as the internal standard (CDCl3 at 77.0 ppm). High-resolution mass spectra were recorded by an ESI-TOF

MS spectrometer (DART ion source). All reagents were purchased from commercially

23 available sources and were used without further puriﬁcation. All solvents were dried over activated 3 Å molecular sieves.

Compounds 51, 57a-b were synthesized according to previously published methods.30

Synthesis of Phenyl 5-Phenylpenta-2,3-dienoate (52b).

A solution of phenyl (triphenylphosphoranylidene) acetate (5.9 g, 15 mmol) and triethylamine (2.2 mL,15.6 mmol) in dichloromethane (40mL) was charged to a round- bottom flask. To it was added hydrocinnamoyl chloride (2.4 mL, 16.2 mmol) in dichloromethane (15 mL) using an addition funnel over a period of 5−10 min. The reaction was stirred at room temperature until it turned into a dark orange color solution. This solution was concentrated under reduced pressure to about a quarter of its volume, and then diethyl ether was added to precipitate the triphenylphosphine oxide formed during the reaction. Triphenylphosphine oxide was removed by filtration, and the filtrate was concentrated under reduced pressure. The crude residue was purified using silica gel chromatography to give phenyl 5-phenylpenta-2,3-dienoate (52b) (the minor product, an alkyne isomer, was also formed as an inseparable impurity) using ethyl acetate/hexanes(1:20). The product was obtained as a clear oil in 88% yield (3.278 g): 1H

NMR (400 MHz,CDCl3) δ3.54 (ddd, 2H, J = 7.4, 4.7, 2.7 Hz), 5.78−5.81 (m,1H), 5.90 (td,

1H, J=7.4, 6.3 Hz), 7.11−7.16 (m, 2H), 7.24−7.27 (m, 2H), 7.30−7.33 (m, 3H), 7.37−7.43

(m, 3H); 13C NMR (100 MHz, CDCl3) δ 34.0, 88.3, 95.4, 121.5, 125.8, 126.7, 128.5, 128.5,

24 129.4, 138.3, 150.8, 164.4, 213.8; HRMS calc. for C17H14O2 [M + NH4]+: 268.1332.

Found 268.1334.

Synthesis of Phenyl 2-Methylbuta-2,3-dienoate (52c).

A solution of phenyl 2-(triphenylphosphanylidene) propanoate (6.16g, 15mmol) and triethylamine (2.2 mL, 15.6 mmol) in dichloromethane (40 mL) was charged to a round- bottom ﬂask. To it was added acetyl chloride (1.15 mL, 16.2 mmol) in dichloromethane

(15 mL) through an addition funnel over a periodof5−10min.The reaction was stirred at room temperature until it turned into a dark orange color solution. The solution was concentrated under reduced pressure to about a quarter of its volume, and diethyl ether was added to precipitate the triphenylphosphine oxide. Triphenylphosphine oxide was then removed through filtration, and the filtrate was concentrated under reduced pressure. The crude residue was purified using silica gel chromatography to give phenyl 2- methylbuta2,3-dienoate (52c) (the minor alkyne isomer was also formed as an inseparable impurity) using ethyl acetate/hexanes (1:20). The product was obtained as a clear oil in

1 45% yield (0.901g): H NMR (400 MHz, CDCl3) δ1.98 (t, 3H, J=3.1Hz), 5.20 (q, 2H, J =

3.1 Hz), 7.12(d, 2H, J = 7.4 Hz), 7.22 (t, 1H, J = 7.4 Hz), 7.38 (t, 2H, J = 7.4 Hz; 13C NMR

(100 MHz, CDCl3) δ 14.7, 78.4, 95.0, 121.5, 125.7, 129.3, 151.0, 166.2, 214.7; HRMS calc. for C11H10O2 [M + NH4]+: 192.1019. Found 192.1023.

General Procedure for the Preparation of Compounds 53, 58-59.

To a solution of 57a (0.37−2.0 mmol, 1 equiv) in acetone (5 mL mmol−1) was added

K2CO3 (0.37−2.0 mmol, 1 equiv) and 52 (0.44−2.4 mmol, 1.2equiv). The reaction mixture was stirred at room temperature for 0.25−2 h. Upon reaction completion (as monitored by

TLC), the crude reaction mixture was filtered and the filtrate was evaporated and purified using silica gel column chromatography with ethyl acetate/ hexanes (3:7) to give pure 59a- b in 56−80% yields.

4-Methyl-4a-(phenylsulfonyl)-5,6-dihydrocyclopenta[b]pyran-2(4aH)-one (59a).

1 Obtained in 69% yield (402 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 2.08−2.34

(m, 3H), 2.23 (d, 3H, J = 1.2 Hz), 2.99 (dd, 1H, J = 14.1, 7.0 Hz), 5.50 (dd, 1H, J = 3.5,

2.3 Hz), 5.96 (q, 1H, J = 1.2 Hz), 7.54 (t, 2H, J = 7.8 Hz), 7.69 (t, 1H, J = 7.8 Hz), 7.86 (d,

13 2H, J = 7.8 Hz); C NMR (100 MHz, CDCl3) δ 19.3, 25.9, 29.6, 75.4, 114.9, 121.2, 128.9,

130.3, 133.9, 134.8, 146.0, 153.0, 159.0;

HRMS calc. for C15H14O4S [M + NH 4]+: 308.0951. Found 308.0959.

4-Phenethyl-4a-(phenylsulfonyl)-5,6-dihydrocyclopenta[b]pyran2(4aH)-one (59b).

1 Obtained in 80% yield (112 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.99−2.09

(m, 1H), 2.13−2.26 (m, 2H), 2.62−2.70 (m, 1H), 2.84−2.99 (m, 3H), 3.07−3.16 (m, 1H),

5.47 (s, 1H), 6.07 (t, 1H, J = 1.6 Hz), 7.21−7.28 (m, 3H), 7.34 (t, 2H, J = 7.4 Hz), 7.50 (t,

2H, J = 7.8 Hz), 7.67 (t, 1H, J = 7.4 Hz), 7.78 (d, 2H, J = 7.4 Hz); 13C NMR (100 MHz,

CDCl3) δ 25.8, 29.4, 32.4, 33.0, 75.5, 114.8, 119.8, 126.5, 128.4, 128.6, 128.9, 130.3,

26 133.7, 134.7, 139.7, 145.8, 155.6, 159.2; HRMS calc. for C22H20O4S [M + H] +: 381.1155.

Found 381.1158.

3,4-Dimethyl-4a-(phenylsulfonyl)-5,6-dihydrocyclopenta[b]pyran-2(4aH)-one (59c).

1 Obtained in 56% yield (99 mg) as a white solid. H NMR (400MHz, CDCl3) δ1.86 (d, 3H,

J = 1.2 Hz), 1.99−2.07 (m, 1H), 2.09−2.21 (m, 1H), 2.12 (d, 3H, J = 1.2 Hz), 2.25−2.33

(m, 1H), 2.95 (dd, 1H, J = 14.5, 7.0 Hz), 5.42 (dd, 1H, J = 2.7, 2 Hz), 7.51 (t, 2H, J = 7.4

13 Hz), 7.66 (t, 1H, J = 7.4 Hz), 7.80 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 13.9,

16.6, 25.6, 30.1, 75.7, 113.8, 128.2, 128.8, 130.0, 134.2, 134.6, 144.7, 145.3, 160.4;

HRMS calc. for C16H16O4S [M + H] +: 305.0842. Found 305.0835.

4-Methyl-4a-(phenylsulfonyl)-4a, 5, 6, 7-tetrahydro-2H-chromen2-one (58a). Obtained in

1 82% yield (158 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.72−1.79 (m, 1H),

1.88 (td, 1H, 13.7, 3.9 Hz), 2.17 (s, 3H), 2.19−2.30 (m, 3H), 3.01 (dt, 1H, J = 14.5, 3.5Hz),

5.76 (d, 1H, J = 1.2 Hz), 5.80 (t, 1H, J = 3.9 Hz), 7.51 (t, 2H, J = 7.8 Hz), 7.66 (t, 1H,J =

13 7.8Hz), 7.83 (d, 2H, J = 7.8 Hz); CNMR (100 MHz,CDCl3) δ 18.0, 19.2, 22.6, 27.4, 67.9,

117.1, 121.8, 128.8, 130.2, 134.6, 135.5, 142.6, 153.1, 159.1; HRMS calc. for C16H16O4S

[M+NH 4]+: 322.1108. Found 322.1108.

4-Phenethyl-4a-(phenylsulfonyl)-4a, 5, 6, 7-tetrahydro-2H-chromen-2-one (58b).

Obtained in 82% yield (137 mg) as a white solid.

1 H NMR (400 MHz, CDCl3) δ 1.66−1.75 (m, 1H), 1.86 (td, 1H, 12.5, 3.5 Hz), 2.09−2.31

(m, 3H), 2.62−2.71 (m, 1H), 2.75−2.88 (m, 2H), 2.93−3.05 (m, 2H), 5.80 (dd, 1H, J = 4.5,

3.3 Hz), 5.87 (t, 1H, J = 1.6 Hz), 7.23−7.36 (m, 5H), 7.48 (t, 2H, J = 7.4 Hz), 7.64 (t, 1H,

13 J = 7.4 Hz), 7.74 (d, 2H, J = 7.4Hz); C NMR (100MHz, CDCl3) δ18.0, 22.5, 27.2, 32.4,

27 32.5, 68.1, 117.2, 120.2, 126.6, 128.4, 128.7, 128.8, 130.2, 134.5, 135.3, 139.6, 142.5,

155.7, 159.3;

HRMS calc. for C23H22O4S [M + NH 4]+: 412.1577. Found 412.1584.

3,4-Dimethyl-4a-(phenylsulfonyl)-4a, 5, 6, 7-tetrahydro-2H-chromen-2-one (58c).

1 Obtained in 48% yield (78 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.69 (s,

3H), 1.72−1.79, (m, 1H), 1.85 (t, 1H, 13.7 Hz), 2.07 (s, 3H), 2.11−2.36 (m, 3H), 3.03 (d,

1H, 14.5 Hz), 5.74 (s, 1H), 7.49 (t, 2H, J = 7.4 Hz), 7.63 (t, 1H, J = 7.4 Hz), 7.76 (d, 2H, J

13 = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 13.5, 15.6, 18.2, 22.6, 27.8, 68.4, 116.1, 128.5,

128.8, 129.8, 134.4, 135.9, 142.2, 145.2, 160.5;

HRMS calc. for C17H18O4S [M + NH 4]+: 336.1264. Found 336.1259.

4-Methyl-4a-(phenylsulfonyl)-5,6,7,8-tetrahydrocyclohepta[b]pyran-2(4aH)-one (53a).

1 Obtained in 34% yield (228 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.35−1.46

(m, 1H), 1.51−1.68 (m, 2H), 1.73−1.88 (m, 2H), 1.88−1.98 (m, 1H), 2.17 (d, 3H, J = 1.6

Hz), 2.44 (dt, 2H, J = 5.8, 2.6), 5.70 (dd, 1H, J = 9.8, 5.8), 6.10 (s, 1H), 7.51 (t, 2H, J =

7.4Hz), 7.65 (t, 1H, J = 7.4Hz), 7.85 (d, 2H, J = 7.4Hz); 13C NMR (100 MHz, CDCl3) δ

19.2, 19.5, 20.7, 23.5, 24.0, 74.6, 118.8, 123.9, 128.7, 130.5, 134.0, 134.5, 143.7, 151.4,

159.5; HRMS calc. for C17H18O4S [M + NH 4]+: 336.1264. Found 336.1274.

4-Phenethyl-4a-(phenylsulfonyl)-5, 6, 7, 8-tetrahydrocyclohepta[b]pyran-2(4aH)-one

(53b). Obtained in 44% yield (71 mg) as a white solid. 1H NMR (400 MHz, CDCl3) δ

1.37−1.57 (m, 1H), 1.55−1.74, (m, 2H), 1.77−2.02 (m, 3H), 2.42 (m, 3H), 2.90 (t, 2H, J =

7.6Hz), 3.00− 3.09 (m, 1H), 5.76 (dd, 1H, J = 9.8, 5.5 Hz), 6.18 (s, 1H), 7.23−7.29 (m,

3H), 7.35 (t, 2H, J = 7.8 Hz), 7.51 (t, 2H, J = 7.4Hz), 7.67 (t, 1H, J = 7.4 Hz), 7.81 (d, 2H,

28 13 J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 19.46, 19.48, 23.2, 23.6, 32.7, 33.7, 75.0,

118.8, 112.4, 126.6, 128.4, 128.7, 128.8, 130.6, 134.0, 134.6, 139.7, 143.8, 153.7, 159.6;

HRMS calc. for C24H24O4S [M + NH 4]+: 426.1734. Found: 426.1722.

General Procedure for the Preparation of Compounds 54, 60-61.

To a solution of lactone (53, 58-59) (0.041−1.1 mmol, 1 equiv) in anhydrous THF (5 mL mmol−1) at−78 °C was added 2.4 M LiAlH4 solution (0.041−1.1 mmol, 1 equiv), and the reaction mixture was stirred at −50 °C for 1−2 h. The reaction was quenched by adding acetone (2 mL) and stirred for an additional 1 h. The reaction mixture was then passed through a Celite pad and washed with CH2Cl2 (50mL). The combined solvent was evaporated to dryness, and the crude product was subjected to oxidation without further purification. The crude product thus obtained was dissolved in CH2Cl2 (5mLmmol−1), and to it was added DMP (2.0 equiv, 0.3 M solution in CH2Cl2). The mixture was stirred overnight at room temperature. Upon reaction completion (as monitored by TLC), the reaction mixture was filtered through Celite. The filtrate was evaporated under reduced pressure, and the residue was purified using silica gel column chromatography with ethyl acetate/hexane (3:7) to give pure 54, 60-61.

(1R*,5S*)-4-Methyl-5-(phenylsulfonyl)bicyclo [3.2.1] oct-3-ene2,8-dione (61a).

Obtained in 57% yield (17 mg) over two steps as a white solid.

1 H NMR (400 MHz, CDCl3) δ 1.73 (ddd, 1H, J = 13.7, 10.6, 5.1 Hz), 2.20−2.38 (m, 2H),

2.62 (s, 3H), 2.76 (ddd, 1H, J = 13.4, 10.7, 5.7 Hz), 3.39 (d, 1H, J = 7.4 Hz), 6.18 (s, 1H),

29 7.59 (t, 2H, J = 7.4 Hz), 7.70 (t, 1H, J = 7.8 Hz), 8.16 (d, 2H, J = 7.4 Hz); 13C NMR (100

MHz, CDCl3) δ 20.6, 21.4, 29.7, 62.6, 78.2, 128.9, 130.0, 131.4, 134.8, 136.4, 160.2, 194.2,

195.2;

HRMS calc. for C15H14O4S [M+NH 4]+: 308.0951. Found: 308.0949.

(1R*,5S*)-4-Phenethyl-5-(phenylsulfonyl)bicyclo[3.2.1]oct-3-ene2,8-dione(61b).

1 Obtained in 27% yield (11 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3)

δ 1.63−1.70 (m, 1H), 2.17−2.32 (m, 2H), 2.78 (ddd, 1H, J = 13.5, 10.4, 5.5 Hz), 2.91−2.98

(m, 1H), 3.02−3.09 (m, 1H), 3.23−3.31 (m, 1H), 3.39 (d, 1H, J = 7.4 Hz), 3.47− 3.55 (m,

1H), 6.26 (s, 1H), 7.24 (d, 3H, J = 7.4 Hz), 7.32 (t, 2H, J = 7.4 Hz), 7.58 (t, 2H, J = 7.4

13 Hz), 7.70 (t, 1H, J = 7.8 Hz), 8.16 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 20.6,

30.1, 34.3, 34.8, 62.8, 78.5, 126.5, 128.4, 128.6, 128.7, 128.9, 131.5, 134.8, 136.5, 139.7,

163.0, 194.3, 195.3;

HRMS calc. for C22H20O4S [(M−CO+H2)+NH 4]+: 372.1628. Found: 372.1628.

(1R*,5S*)-3,4-Dimethyl-5-(phenylsulfonyl)bicyclo[3.2.1]oct-3ene-2,8-dione (61c).

1 Obtained in 20% yield (19 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3)

δ 1.57−1.71 (m, 1H)), 1.89 (s, 3H), 2.17−2.35 (m, 2H), 2.59 (s, 3H), 2.80 (ddd, 1H, J =

13.2, 10.7, 5.5 Hz), 3.46 (d, 1H, J = 7.4Hz), 7.58 (t, 2H, J = 7.4 Hz), 7.69 (t, 1H, J = 7.8

13 Hz), 8.17 (d, 2H, J = 7.4 Hz); C NMR(100 MHz, CDCl3) δ12.2, 17.2, 20.6, 29.4, 62.8,

79.4, 128.8, 131.4, 134.6, 135.3, 136.8, 153.5, 194.9,195.5;

HRMS calc. for C16H16O4S [M+NH 4]+:322.1108.Found 322.1112.

(1R*,5S*)-4-Methyl-5-(phenylsulfonyl)bicyclo[3.3.1]non-3-ene2,9-dione (60a). Obtained

1 in 60% yield (182 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3) δ

1.59−1.87 (m, 3H), 1.99− 2.09 (m, 1H), 2.24 (td, 1H, J = 12.9, 4.3 Hz), 2.55 (s, 3H),

30 2.57−2.68 (m,1H), 3.26 (t, 1H, J = 3.5 Hz), 6.50 (s, 1H), 7.56 (t, 2H, J = 7.4 Hz), 7.67

13 (t,1H, J = 7.4 Hz), 8.15 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 18.2, 22.3,

30.8, 31.2, 62.6, 79.2, 128.5, 131.7, 134.3, 135.1, 138.2, 154.5, 194.6, 199.1; HRMS calc. for C16H16O4S [M + NH 4]+: 322.1108. Found 322.1110.

(1R*,5S*)-4-Phenethyl-5-(phenylsulfonyl)bicyclo[3.3.1]non-3ene-2,9-dione (60b).

1 Obtained in 58% yield (87 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3)

δ 1.39−1.50 (m, 1H), 1.69− 1.78 (m, 2H), 1.99−2.09 (m, 1H), 2.24 (td, 1H, J = 12.5, 4.3

Hz), 2.60− 2.66 (m, 1H), 3.05 (t, 2H, J = 7.4 Hz), 3.20−3.31 (m, 2H), 3.42−3.50 (m, 1H),

6.60 (s, 1H), 7.19−7.35 (m, 5H), 7.57 (t, 2H, J = 7.4 Hz), 7.68 (t, 1H, J = 7.4 Hz), 8.16 (d,

13 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 18.0, 30.6, 31.8, 34.2, 35.0, 62.6, 79.6,

126.5, 128.38, 128.44, 128.6, 131.8, 134.0, 134.3, 138.4, 139.8, 157.2, 194.8, 199.0;

HRMS calc. for C23H22O4S [M + NH 4]+: 412.1577. Found 412.1584.

(1R*,5S*)-3,4-Dimethyl-5-(phenylsulfonyl)bicyclo[3.3.1]non-3ene-2,9-dione (60c).

1 Obtained in 54% yield (26 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3)

δ 1.44−1.53 (m, 1H), 1.70− 1.84 (m, 2H), 1.95−2.07 (m, 1H), 1.99 (s, 3H), 2.27 (td, 1H, J

= 13.3, 3.9 Hz), 2.54 (s, 3H), 2.61−2.71 (m, 1H), 3.32 (t, 1H, J = 3.9 Hz), 7.55 (t, 2H, J =

13 7.4 Hz), 7.65 (t, 1H, J = 7.4 Hz), 8.14 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3)

δ12.5, 17.4, 18.4, 30.6, 30.9, 62.6, 80.4, 128.4, 131.6, 134.1, 138.9, 140.6, 148.3, 194.7,

199.2; HRMS calc. for C17H18O4S [M + NH 4]+: 336.1264. Found 336.1277

(1S*,6R*)-9-Methyl-1-(phenylsulfonyl)bicyclo[4.3.1]dec-8-ene7,10-dione (54a).

1 Obtained in 43% yield (17 mg) over two steps as a white solid. H NMR (400 MHz, CDCl3)

δ 1.15−1.29 (m, 1H), 1.33−1.48 (m, 1H), 1.67 (s, 1H), 1.78−1.89 (m, 1H), 1.90−2.01 (m,

1H), 2.21 (t, 2H, J = 13.3 Hz), 2.49 (s, 3H), 2.68 (td, 1H, J = 13.7, 3.1 Hz), 3.36 (dd, 1H,

31 J = 11.3, 3.9 Hz), 6.38 (s, 1H), 7.52 (t, 2H, J = 7.4 Hz),7.64 (t, 1H, J = 7.4 Hz), 7.84 (d,

13 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 21.4, 25.3, 25.6, 29.5, 32.0, 62.1, 81.2,

128.6, 130.5, 133.3, 134.2, 138.2, 153.6, 195.3, 199.8; HRMS calc. for C17H18O4S [M +

NH4]+: 336.1264. Found 336.1269.

(1S*,6R*)-9-Phenethyl-1-(phenylsulfonyl)bicyclo[4.3.1]dec-8-ene7,10-dione (54b).

Obtained in 36% yield (6.0 mg) over two steps as a white solid. 1H NMR (400 MHz,

CDCl3) δ 1.14−1.19 (m, 2H), 1.30− 1.40 (m, 1H), 1.67−1.85 (m, 2H), 2.11−2.22 (m, 2H),

2.63 (td, 1H, J = 13.7, 3.7 Hz), 2.85−3.00 (m, 2H), 3.12−3.15 (m, 1H), 3.34−3.46 (m, 2H),

6.58 (s, 1H), 7.22−7.36 (m, 5H), 7.50 (t, 2H, J = 7.4 Hz), 7.63 (t, 1H, J = 7.4 Hz), 7.80 (d,

13 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 25.42, 25.46, 29.4, 32.3, 33.9, 34.6, 62.0,

81.6, 126.5, 128.6, 128.62, 128.67, 130.6, 131.8, 124.1, 128.4, 140.2, 156.6, 195.4, 199.9;

HRMS calc. for C24H24O4S [M + NH 4]+: 426.1734. Found 426.1715.

General Procedure for the Preparation of Medium Rings 55, 63.

To a stirred solution of compound 54 or 60 (0.047−0.065 mmol, 1 equiv) dissolved in

THF (2 mL mmol−1) was added TBAF·3H20 (10 mol%), and the reaction was stirred at room temperature for 1h.Upon reaction completion (as monitored by TLC), solvent was evaporated and the residue was puriﬁed using silica gel column chromatography with ethyl acetate/hexanes (3:7) to aﬀord pure 55 or 63. (E)-3-Methyl-4-

(phenylsulfonyl)cyclooct-3-en-1-one (63a). Obtained in 95% yield (17 mg) as a gummy

1 compound. H NMR (400 MHz, CDCl3) δ 1.73−1.86 (m, 4H), 2.23 (s, 3H), 2.42−2.48

32 (m, 2H), 2.66−2.73 (m, 2H), 3.32 (s, 2H), 7.54 (t, 2H, J = 7.4 Hz), 7.60 (t, 1H, J = 7.4

13 Hz), 7.87 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 21.7, 23.4, 28.5, 29.1, 42.6,

52.9, 127.2, 129.2, 133.2, 137.6, 141.7, 144.8, 208.1; HRMS calc. for C15H18O3S [M +

NH 4]+: 296.1315. Found 396.1327.

(E)-3-Phenethyl-4-(phenylsulfonyl)cyclooct-3-en-1-one(63b).Obtained in 96% yield (18

1 mg) as a gummy compound. H NMR (400 MHz, CDCl3) δ1.71−1.85 (m, 4H), 2.44−2.50

(m, 2H), 2.63−2.69 (m, 2H), 2.71−2.79 (m, 2H), 2.90−2.98 (m, 2H), 3.39 (s, 2H),

7.18−7.33 (m, 5H), 7.53 (t, 2H, J = 7.4 Hz), 7.61 (t, 1H, J = 7.4 Hz), 7.86 (d, 2H, J = 7.4

13 Hz); C NMR (100 MHz, CDCl3) δ 23.6, 28.2, 29.4, 34.4, 37.5, 42.6, 51.2, 126.3, 127.2,

128.4, 128.5, 129.3, 133.3, 138.0, 140.7, 141.6, 148.1,208.6;

HRMS calc. for C22H24O3S [M+NH4]+: 386.1784. Found 386.1794.

(E)-3-Methyl-4-(phenylsulfonyl)cyclonon-3-en-1-one (55a). Obtained in 95% yield (14

1 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.67−1.81 (m, 6H), 2.27 (s, 3H), 2.49

(dd, 2H, J = 7.4, 5.8 Hz), 2.78 (t, 2H, J = 6.6 Hz), 3.26 (s, 2H), 7.52 (t, 2H, J = 7.4 Hz),

13 7.59 (t, 1H, J = 7.4 Hz), 7.84 (d, 2H, J = 7.4 Hz); C NMR (100 MHz, CDCl3) δ 22.8,

23.0, 23.7, 26.2, 28.0, 44.4, 49.4, 127.0, 128.3, 129.0, 133.0, 142.0, 145.0, 208.7;

HRMS calc. for C16H20O3S [M + NH 4]+: 310.1471. Found 310.1480.

General Procedure for the Preparation of Compounds 56, 64-65.

Compounds 54, 60, or 61 (0.040−0.060 mmol, 1 equiv) were dissolved in an anhydrous methanol:triethylamine solution (5:1 ratio, 2 mL mmol−1), and there actions were allowed to stir at room temperature for 1h.Upon reaction completion (as monitored by TLC),the 33 solvent was evaporated and the crude residue was puriﬁed using silica gel column chromatography with ethyl acetate/hexanes (3:7) to aﬀord pure 56, 64-65.

Methyl (1Z, 3Z) 2-Hydroxy-4-methyl-5-(phenylsulfonyl)cyclohepta-1,3diene-1- carboxylate (64). Obtained in 55% yield (8.0 mg) as a white solid. 1H NMR (400 MHz,

CDCl3) δ 2.01 (d, 3H, J = 1.6 Hz), 2.21−2.31 (m, 2H), 2.51−2.65 (m, 2H), 3.74 (s, 3H),

3.98 (dd, 1H, J = 8.2, 5.5 Hz), 6.03 (d, 1H, J = 1.6 Hz), 7.53 (t, 2H, J = 7.4 Hz), 7.64 (t,

13 1H, J = 7.4 Hz), 7.85 (d, 2H, J = 7.4 Hz), 12.07 (s, 1H); C NMR (100 MHz, CDCl3) δ

21.5, 27.4, 32.3, 51.8, 69.9, 101.9, 128.8, 129.0, 129.4, 133.9, 137.6, 141.4, 166.5, 172.4;

HRMS calc. for C16H18O5S [M + NH 4]+: 340.1213. Found 340.1220.

Methyl (1Z, 3Z)-2-Hydroxy-4-methyl-5-(phenylsulfonyl) cycloocta1,3-diene-1- carboxylate (65a).

Obtained in 98% yield (22 mg) as a white solid. 1H NMR (400 MHz, CDCl3) δ 1.10−1.27

(m, 1H), 1.43− 1.56 (m, 1H), 1.70 (dd, 1H, J=14.5, 12.1 Hz), 1.84−1.95 (m, 1H), 2.07 (s,

3H), 2.30 (d, 1H, J = 12.9 Hz), 2.66 (dd, 1H, J = 14.7, 8.0 Hz), 3.78, (s, 3H), 4.12 (d, 1H,

J = 9.4 Hz), 5.91 (s, 1H), 7.54 (t, 2H, J = 7.4 Hz), 7.64 (t, 1H, J = 7.4 Hz), 7.84 (d, 2H, J =

13 7.4 Hz), 12.16 (s, 1H); C NMR (100 MHz, CDCl3) δ 19.3, 22.2, 24.5, 24.9, 52.0, 66.5,

102.7, 124.7, 128.2, 129.2, 133.7, 138.4, 139.5, 167.5, 173.0;

HRMS calc. for C17H20O5S [M + NH 4]+: 354.1370. Found 354.1383.

Methyl (1Z, 3Z)-2-Hydroxy-4-phenethyl-5-(phenylsulfonyl) cycloocta-1,3-diene-1- carboxylate (65b).

1 Obtained in 90% yield (19 mg) as a white solid. H NMR (400 MHz, CDCl3) δ 1.04−1.18

(m, 1H), 1.38−1.51(m, 2H),1.76−1.91 (m, 1H), 2.29 (d, 1H, J = 12.1 Hz), 2.57 (dd, 1H, J

= 14.7, 7.6 Hz), 2.69−2.85 (m, 4H), 3.78 (s, 3H), 4.16 (d, 1H, J = 9.4 Hz), 6.01 (s, 1H),

34 7.14−7.24 (m, 3H), 7.26−7.35 (m, 2H), 7.55 (t, 2H, J = 7.4 Hz), 7.64 (t, 1H, J = 7.4 Hz),

13 7.83 (d, 2H, J = 7.4 Hz), 12.21 (s, 1H); C NMR (100 MHz, CDCl3) δ 22.3, 24.2, 24.8,

32.4, 33.9, 52.0, 66.9, 102.8, 123.7, 126.2, 128.2, 128.40, 128.44, 129.2, 133.7, 138.4,

140.6, 142.5, 167.8, 173.0; HRMS calc. for C24H26O5S [M + NH4]+: 444.1839. Found

444.1860.

Methyl (1R, 5R, Z)-2-Oxo-4-phenethyl-5-(phenylsulfonyl)cyclonon3-ene-1-carboxylate

(56).

Obtained in 90% yield (19 mg) as a white solid. 1H NMR (400 MHz, CDCl3) δ 1.29−1.41

(m, 1H), 1.47−1.66 (m, 2H), 1.78−1.90 (m, 2H), 1.91−2.02 (m, 1H), 2.17 (dd, 1H, J = 15.7,

10.6 Hz), 2.26−2.34 (m, 1H), 2.80 (t, 1H, J = 13.3 Hz), 3.06 (dd, 1H, J = 16.0, 8.6 Hz),

3.21 (dd, 1H, J = 16.4, 7.0 Hz), 3.34 (dd, 1H, J = 16.0, 7.4 Hz), 3.47 (d, 2H, J = 3.9 Hz),

3.72 (s, 3H), 5.27 (t, 1H, J = 6.5 Hz), 6.94 (d, 2H, J = 7.4 Hz), 7.16−7.25 (m, 3H), 7.36 (t,

2H, J = 7.4 Hz), 7.52 (t, 1H, J = 7.4 Hz), 7.92 (d, 2H, J = 7.4 Hz), 13C NMR (100 MHz,

CDCl3) δ 25.0, 27.1, 28.3, 30.6, 33.2, 35.2, 42.0, 52.3, 84.7, 126.3, 126.4, 127.8, 128.3,

128.6, 131.7, 133.4, 136.5, 138.2, 140.0, 172.0, 205.7;

HRMS calc. for C25H28O5S [M + NH 4]+: 458.1996. Found 458.2001.

35 Chapter 2

Bioactivity Investigation of [3.2.1] bicyclic compounds as modified functional isomorphs

of resveratrol and vitisinol D in a neuroprotective model

2.1 Introduction

2.1.1 Biological relevance

Small molecule-mediated neuroprotection is an area of research that has important implications for several disease and injury models such as Alzheimer’s disease and stroke.

Biological targets for small molecule interaction are diverse, and there are often questions about the mechanism of neuroprotection that many small molecule neuroprotectants act through, as will be discussed in subsequent sections. Many neuroprotectants are known to be potent antioxidants, and oxidative stress is often the primary source of damage to neurons in anoxic models (see subsequent section). The possibility of these molecules acting as other than free radical scavengers, that is, interacting directly at the neurons or with a protein capable of causing an upstream neuroprotectant effect, is an intriguing reason to investigate the subtle relationships between the functionality and molecular skeletal structure of these small molecule neuroprotectants in the hope of gaining the ability to design targeted drug therapies for a range of brain injury events and neurodegenerative diseases.

2.1.2 Phenotypic models of anoxic shock and oxidative stress

A phenotypic model for measuring the neuroprotective effects of small molecules against oxidative stress using Drosophila melanogaster has been developed. This species of fly, 36 one of the most genetically studied and well-understood species in biological research, has evolved with the capacity to survive environmental anoxic shock due to low oxygen environments through an adaptation in which the animal enters a protective coma that prevents anoxia-induced neurodegradation.31 This adaptation makes the species a useful subject32 for various protocols that study the effect of different therapeutic regimens against the effects of oxidative stress on the brain.

Dawson-Scully has shown that the anoxic coma response of D melanogaster is dependent upon the protein kinase G (PKG) cascade.33 One implication from these data is that inhibition of the PKG pathway may protect against oxidative stress by preventing the shutdown of synaptic transmission associated with the anoxic coma of D melanogaster.34

One promising method to measure the neuroprotective effects of various medicinal compounds against oxidative stress in D. melanogaster has been developed and utilized recently by Dawson-Scully and others31. In this phenotypic model, fly larvae have their synaptic transmission measured via electrical probes in the post-synaptic neuromuscular junction (NMJ). The time until synaptic failure is measured, and then the time to synaptic failure is measured when a solution of 2.25 mM hydrogen peroxide is introduced to the tissue. Hydrogen peroxide, a reactive oxygen species, greatly reduces the time to synaptic failure relative to the control by damaging the neurons through oxidative stress. Once the time to synaptic failure has been established with hydrogen peroxide alone, then different compounds can be introduced at varying concentrations along with the hydrogen peroxide solution, and their effects on the time to synaptic failure are measured and compared to the two controls. Restoration of the time to synaptic failure seen in the baseline control

(without hydrogen peroxide) is taken as consistent with the hypothesis that the medicinal

37 compound is acting as a neuroprotectant. Unfortunately, with a phenotypic model, neither the exact biomolecular target nor the mechanism of neuroprotection can be easily inferred, but, in combination with other experiments, a case may be made to suggest certain modes of action as warranting more direct study. A note on terminology: at least one author distinguishes between the clinical effects of neuroprotection and restoration, with neuroprotectants preventing damage and restoratives restore lost neurological function, but as even this source admits that the term neuroprotectant is often used to describe restoratives, the term neuroprotectant will be used throughout this work to describe small molecules that are purported to have either a neuroprotectant or neurological restorative effect.35

2.1.3 Small molecule neuroprotectants

The natural product pseudopterosin A, a marine diterpene glycoside isolated from

Pseudopterogorgia elisabethae, was evaluated33 as a neuroprotectant on the phenotypic model described above.

Figure 3. Compounds used in a previous in vivo neuroprotection assay Pseudopterosin A extended the time to synaptic failure when administered at 1 μM with 1 mM H2O2 relative to the control time to synaptic failure without H2O2 (from 50 min to 63 min). The compound Trolox (an antioxidant analogue of vitamin E known to reduce oxidative stress) was also used as a control at 5 μM for a comparison of neuroprotection,

38 and it afforded a slightly higher extension of the time to synaptic failure with 1 mM H2O2 than pseudopterosin A, extending the time to failure from 50 min to 80 min.

Figure 4. Small molecule neuroprotectant drug edaravone An example of a small molecule neuroprotectant that has found medical use and entered the market is the drug edaravone36 (73 Figure 4), which is used in the treatment of stroke and amyotrophic lateral sclerosis. Edaravone has been shown to inhibit the formation of linoleic acid-conjugated dienes that are formed from hydroxyl radical with an activity of

37 IC50 = 32.0 μM. Although this compound is considered a free radical scavenger due to its ability to donate an electron, it may have other mechanisms of biochemical action as well.38

In investigating whether a compound’s neuroprotectant effect is due to free radical scavenging or through interaction with a biological target, it should first be established that the molecule is a potent free radical scavenger. In the case of edaravone, DFT analysis corroborates experimental evidence that edaravone is a potent free radical scavenger.39 To investigate whether a protectant effect is due to more than free radical scavenging, careful target-oriented experiments are necessary, as were used to show that edaravone was shown to augment a regulator of antioxidant enzymes.40

Resveratrol 74 (Figure 6), a polyphenolic stilbene produced as a secondary metabolite in the skins of plants such as grapes, has long been of interest as an antioxidant and anti- atherosclerotic agent. This natural product is believed to be responsible for the purported health benefits of red wine.41 As discussed in subsequent paragraphs, resveratrol was shown to convey neuroprotection in a NMJ phenotypic assay.

39 2.1.4 Possible mechanisms of neuroprotection

The current understanding of in vivo mechanistic pathways of neurodegradation caused by hydrogen peroxide induced oxidative stress is incomplete. However, it is understood that antioxidants, often called free radical scavengers, mitigate this damage, at least in part, by reacting with the various radicals generated under oxidative stress, including fatty acid radicals and fatty acid peroxyradicals.42 Such reactions result in highly resonance stabilized and thus relatively unreactive radicals or serve as terminators of free radical chain reactions initiated by hydrogen peroxide or other initiators. Typically, phenolic compounds serve as antioxidants due to their ability to donate a proton homolytically to a reactive oxygen species or other radical intermediate

Figure 5. Example of free radical scavenging mechanism in representative compound resveratrol The use of Trolox as a control in the pseudopterosin A experiment is significant because it establishes a level of neuroprotection from the effects of oxidative stress plausibly accomplished mainly through a free radical scavenging mechanism. Resveratrol, along with many other neuroprotectants, is a free radical scavenger in addition to its other known

40 biological effects, and its mechanism of free radical scavenging is through homolytic loss of a phenol proton to a reactive oxygen species or other radical intermediate.43 The high degree of conjugation in these mainly flat sp2-hybridized molecules results in a low energy radical that is relatively unreactive, effectively terminating the radical chain reaction (see

Figure 5). Due to this free radical scavenging possibility, bioassays that measure neuroprotective effects against oxidative stress should be accompanied by control studies to determine if a test compound acts by a non-radical mechanisms. The para-hydroxyl group of trans-resveratrol has been experimentally shown to be the more reactive moiety towards free radicals, but homolytic deprotonation also occurs at the 3 or 5 position (Figure

5). The phenoxyl radicals shown dismutate into nonradical products, but the keto radical can react at the radical carbon with oxygen.43

Figure 6. Natural polyphenols that activate SIRT1 in vitro Neuroprotection from oxidative stress is possible through small molecule mediated effects other than free radical scavenging. Recently, evidence that a class of small molecules may convey neuroprotection against oxidative stress and other benefits through activation of sirtuins has been accumulating.44 The sirtuins, or Silent Information Regulator (SIR) genes, are conserved in organisms as diverse as yeast, roundworms, flies and mammals, and overexpression of these genes has been observed to lead to longer lifespans in worms and, notably, fruit flies.45 The proteins for which the SIR genes code are Class II histone deacetylases, known as SIRT enzymes, which are dependent upon nicotinamide adenine

41 dinucleotide (NAD) as a coenzyme.46 Nicotinamide is released as a byproduct of the deacylation of lysines in the target histones, and deactivates the sirtuins, setting up a negative feedback mechanism.47 Although SIRT2 inhibition is being explored as a potential method of combating certain types of cancer,48,49 the activation of the various isoforms of SIRT (1-7) is of interest as they are suspected to be involved in a broad range of beneficial effects ranging from lifespan extension to protection against cardiovascular disease, neurodegeneration, metabolic disorders, inflammation, and cancer. Several natural products have been identified that can activate sirtuins in vitro (74-76, Figure 6).44

The mechanism of SIRT activation has been a controversial topic, but following much investigation, allosteric binding mechanisms seem to be gaining support.50 The hypothesis that these compounds directly activate SIRT1 via allosteric binding, thus lowering the peptide substrate Km, was tested against the alternative hypothesis that in vivo SIRT activation was an incidental consequence of some action of the compounds elsewhere. The

“direct activation hypothesis” was initially supported by fluorometric assays using an acetylated peptide substrate conjugated to aminomethylcoumarin, and by fluorescence polarization assays using a substrate tagged with carboxytetramethylrhodamine analyzed by mass spectroscopy. However, it was found that the fluorescent moiety tags were themselves involved in the mechanism of in vitro SIRT1 activation, leading some to doubt altogether that resveratrol and other STACs are direct activators of SIRT1. Subsequent studies have redeemed the “direct activation hypothesis” by showing that hydrophobic amino acids will suffice to allow for SIRT1 activation in the absence of the fluorescent moieties, and that resveratrol and synthetic STACs can activate SIRT1 in vitro in the

42 presence of native peptides such as FOXO3a and PGC-1α51. The exact mechanism of in vivo activation remains an area of ongoing investigation.

2.1.5 Vitisinol D hypothesis

More recently, resveratrol was discovered to be the most active natural product in vitro activator of human recombinant SIRT1.52 There has been much effort towards synthesizing small molecule sirtuin-activating compounds (STACs).

Figure 7. Representative examples of highly conjugated, ‘flat’ STAC design A review of the literature reveals that most synthetic STACs are chemically distinct from their polyphenolic natural product predecessors (Figure 7). \The initial STAC scaffolds were based on imidothiazole 7753 and then on benzimidazole and urea 78.54 Many STACs have share their high degree of conjugation and genral ‘L’ shape.

Figure 8. Vitisinol D analogues as all-carbon scaffolds with dense architectural complexity The hypothesis we are investigating is that there may be a link between the SIRT activating mechanism of resveratrol and its neuroprotective effect on the CNS towards stresses such as stroke or traumatic brain injury (TBI).55 We sought to investigate if [3.2.1] bicyclic analogues of vitisinol D, could serve as effective STACs. A general similarity in the relative positioning of several phenol moieties is noted when comparing vitisinol D to the known STAC fisetin (79, Figure 8). Our synthesis route used in ongoing efforts by others in our group to synthesize vitisinol D allowed us to access a number of analogs and provides ample future target compounds.

Vitisinol D, a natural product dimer of resveratrol, is one of several oligomers of resveratrol found in the roots of the grapevine Vitis thunbergii, a plant that has been used as a traditional medicine for a range of ailments. Vitisinol D was found to have a very modest activity (IC50 = 15.0±4.8 μM) in inhibiting arachidonic acid induced platelet aggregation

56 in vitro, not much more potent than aspirin (IC50 = 32.7±6.4 μM). This led to the proposal

44 to investigate vitisinol D and its analogues as potential selective inhibitors of enzymes responsible for the blood clotting cascade such as thrombin.

The structure of vitisinol D makes it somewhat unique among the many known dimers of resveratrol. The bridged[3.2.1]bicyclic core of vitisinol D appears to have been constructed through some kind of dearomative addition of one of the 3,5-dihydroxy phenyl rings of one monomer to the former alkene carbons of the other. Although vitisinol D is not the only bridged bicyclic dimer of resveratrol, to our knowledge it is the only one where one of the phenyl rings of a monomer has been incorporated into the bicyclic skeleton with loss of aromaticity and saturation. Many resveratrol dimers, such as pallidol and ε-viniferin, appear to have been conjoined through an enzyme mediated nucleophilic substitution.

Resveratrol dimer ε-viniferin (Figure 9. 80) showed activity on an anticancer assay at IC50

= 16 μM.57 It also inhibits cyctochrome p450 activity with Ki = 0.5 μM. 58 Compound 81,

59 pallidol, is a selective singlet oxygen quencher and has an activity of EC50 = 25 μm against herpes simplex virus HSV-1.60 The bicyclic resveratrol dimer (-)-ampelopsin F, compound 82, features a [3.2.1] bicyclic core that is constructed from the former alkene carbons of the monomers and two of the 3,5-dihydroxy phenyl rings of the monomers that are fused the the [3.2.1] bicyclic core of the molecule. (-)Ampelopsin F was shown to have

61 promising antimalarial activity with IC50 = 0.001 μg/mL.

Figure 9. Comparison of structures and activities of selected resveratrol dimers Initial efforts with our synthetically derived vitisinol D analogues was to explore their potential as blood clot inhibitors. The hypothesis was that the bridged bicyclic core of vitisinol and its analogs provides a rigid scaffold that holds its phenolic appendages in a well-defined relative orientation. This hypothesized ‘L-shape’ is similar to that found in many known small molecule inhibitors of thrombin and factor Xa. However, this project was halted when an opportunity arose to examine the neuroprotective effects of vitisinol D analogs.

The initial hypothesis developed by Lepore was that vitisinol D may be acting upon an allosteric site of sirtuins in a similar manner to known STACs such as fisetin. Initial computational studies (using force field methods) reveal that fisetin and vitisinol D analogs possess a similar relative orientation of the two phenolic appendages that presumably bind to the allosteric site (Figure 10).

The sirtuin activation hypothesis will be corroborated or ruled out by in vitro assays that measure the direct activation of sirtuins. Some of these assays rely on fluorescence measurements that detect sirtuin activity as sirtuin removes a fluorescent tag from a peptide in the presence of potential activators or inhibitors. With the aforementioned controversy surrounding the use of fluorescence assays to detect sirtuin activation having been

46 addressed,51 sensitive assays can be employed62 to narrow the possibilities of the mechanism of neuroprotection for analogs of vitisinol D and resveratrol.

Figure 10. Similarity in distance between key binding moieties of fisetin to hypothetical vitisinol D analogue 2.1.6 Resveromorph conception for potential sirtuin activation

As described subsequently, one of our mono-aryl analogs of vitisinol D analogs has proven to be an effective neuroprotective agent. Given that resveratrol has been demonstrated to activate sirtuin, we considered that the neuroprotective activity of our analog of vitisinol D might also be a result of sirtuin activation. Firstly, our bioactive mono-aryl analogue of vitisinol D is nearly isomeric to resveratrol. Secondly, it is possible that resveratrol and our mono-aryl analogue of vitisinol D assume similar active conformations. In a thought experiment, one could consider that an aryl unit of resveratrol twists to form a bond (no electrocyclization is intended here) with the ethylene bridge leading to the concave shape of a monoaryl analogue of vitisinol D. In this conception, the 2-carbon bridge of the bridged bicycle serves as the structural equivalent of the trans alkene in resveratrol (Figure 11).

Figure 11. Resveramorph and isoresveramorph concept However, our envisioned mono aryl vitisinol analogues are not perfect structural equivalents to resveratrol. A true resveramorph has the hydroxyls on the bicyclic scaffold connected at the 1 and 3-positions. By contrast, our bioactive compounds possess these hydroxyls at the 4 and 8-positions and therefore are better considered as isoresveramorphs

(Figure 11). The compound that our mono-aryl diol analogues of vitisinol D are more isomorphic to in terms of its substitution pattern is the 2,6-dihydroxy isomer of resveratrol

(a natural product of Senna didymobotrya),63 but again with 3’’,5’’-disubstitution on the

48 aryl ring attached to the bicyclic core rather than 4’’ substitution as in he natural 2,6- dihydroxy resveratrol. The isoresveramorphs should be considered patchwork variations on resveratrol isomers with respect to their aryl substitution patterns

2.1.7 Other design considerations

There are several promising examples in the literature of medicinal compounds featuring all-carbon bridged bicyclic scaffolds finding success. The use of bridged bicyclic scaffolds to ensure the relative steric orientation of two or more functionalities due to the conformational restriction of the scaffold is being explored in the class of related compounds known as bicyclic conformationally restricted diamines (CRDA).64 The concept of conformational restriction as a strategy to enhance such factors as receptor selectivity, potency, pharmacophore identification and metabolic stabilization, is an established principle in drug design.65 In addition to advantages discussed in subsequent sections, bridged bicyclic compounds restrict conformational flexibility. However, it should be noted that none of the launched drugs featuring a bicyclic conformationally restricted diamine subunit featured in this review contained an all-carbon bicycle.64

Additionally, an analysis of FDA approved drugs from 1983-2012 revealed that 40% of drugs do not contain any sp3 carbons in a ring system. 66 Taken together, these findings suggest that all-carbon bridged bicyclic scaffolds are promising yet underutilized in drug design.

There are good precedents for carbon bridged bicyclic scaffolds correlating with good biological activity. Huperizine A (Figure 12, 83) is a natural product isolated from the club moss Huperzia serrata that has long since succumbed to total synthesis67 and has been of interest as a treatment for Alzheimer’s disease.68 Many related and derivative compounds

49 such as Huprine Y show high potency as selective inhibitors of acetylcholinesterase, with huprine Y possessing one of the highest known affinities for the active site of

69 acetylcholinesterase, with an IC50 of 0.78nm for human ACE.

Figure 12. All carbon bridged bicyclic scaffold in acetylcholinesterase inhibitors A pyrrolo[2,3-d]thiazole Figure 13, 85) has been developed as an antiviral to combat the

Chikungunya virus, and this compound features an all carbon bicyclo[3.3.1]nonane subunit incorporated into its scaffold. This compound showed micromolar activity against the virus in vitro, and has good in vivo pharmacokinetic properties.70

Figure 13. All carbon bridged bicyclic scaffold in an antiviral compound Another example is the group of bicyclo[3.3.1]nonane analogues of selective estrogen receptor down-regulators (Figure 14, 86) that showed improved binding affinities for ERα and ERβ compared to the non-bicyclic version of the drug 85 from GSK.71

Figure 14. All carbon bridged bicyclic scaffold in an improved estrogen receptor reguator The bridged bicyclic scaffold confers several potential benefits including relative lack of reactivity, especially with highly saturated all-carbon scaffolds. The other potential benefit is due to the more ‘three dimensional’ molecular designs possible with sp3-carbon- containing scaffolds and the potentially chiral nature of the molecules.72 Following

Lipinski’s well-known rule of five73 for molecular drug design, other general trends has been observed between the physical properties of small molecules and clinical efficacy as drug candidates. One such trend is between ‘architectural complexity’ and biological activity.74 While architectural molecular complexity is a difficult criterion to definitively parameterize,75 the degree of saturation and number of chiral centers are two important components according to Lovering and coworkers.72 They observed that more sp3 centers in a molecule allow for placement of appendages in an out-of-plane orientation towards one another. These factors could be expected to lead to more finely tuned complementarity between the molecule and the active site, and thus to greater selectivity.

Based on the criterion of architectural complexity, our [4.3.1] and [3.N.1] bicyclic motifs hold promise as scaffolds in the design of potentially bioactive compounds (Figure 15). A crucial advantage of the bicyclic scaffold is that the advantages conferred by increased saturation and chiral centers are not offset by a concomitant gain in fully rotatable bonds.

51 Indeed, a later addition to Lipinski’s rules of five states that medicinal compounds should have less than ten rotatable bonds.76

Figure 15. General scaffold of vitisinol D analogues with quantified architectural complexity Through both the bridged bicyclic connectivity and the steric congestion of vitisinol D like compounds, the conformational degrees of freedom are greatly reduced relative to a hypothetical open-chain sp3–carbon containing structure of comparable molecular weight.

In the case of vitisinol D, the parent natural product whose ongoing total synthesis provided the synthetic impetus for our research into [4.3.1] and [3.N. 1] bicyclic compounds synthesized via allene intermediates, there are four consecutive stereocenters on an all carbon [3.2.1] bicyclic scaffold that bears three phenolic appendages (compare Figure 8).

The absolute configuration has yet to be determined, but the relative stereochemistry is known. The two phenolic substituents on the 2-carbon bridge are trans to one another, with the 3,5-diphenol substituent syn to the 1-carbon bridge (exo) and the 4-phenol substituent anti to the 1-carbon bridge (endo). Additionally, there are two prochiral carbons: the ketones carbonyls on the 1 and 3-carbon bridges. Our route to simplified analogues of vitisinol D allows us to access the corresponding diols resulting from the reduction of these ketone carbonyls with a high degree of diastereoselectivity. Although one stereocenter is lost in our synthetic approach (along with the third aryl group of vitisinol D) two

52 stereocenters are gained with the hydroxyl carbons, giving our analogues five consecutive stereocenters and an overall chirality.

2.2 Synthesis of vitisinol D analogues

2.2.1 Synthesis of cyclic β-ketoester

As described in Chapter 1, a useful synthesis of an all-carbon [3.2.1]bicycle proceeds through an α,α’-addition strategy (Chapter 1, Scheme 13). The first stage of our synthesis is the construction of the 5-carbon cyclic ketoester with the appropriate pendant group(s) on the carbons that will become the 2-carbon bridge of the bicyclic target. In an extension of the work of others in the Lepore group to synthesize vitisinol D,77 the aryl moiety and the β-ester were installed on commercially available and inexpensive cyclopent-2-enone

87 via 1,4-conjugate addition of the aryl group through higher order cuprate addition

(Scheme 18). This is followed by quenching by ethyl cyanoformate to furnish the β- ketoester 88. Steric hindrance ensures that the ester is added trans to the aryl unit. In order that it can be deprotected to reveal the carboxylic acid later in the synthesis, the next step is the transesterification of the ethyl ester to TMS-protected comound 89.

Scheme 18. Synthesis of cyclic β-ketoester 2.2.2 Synthesis of allenoates

The ability to synthesize diversely functionalized vitisinol D analogues through our annulation method is predicated on having a flexible and efficient method of synthesizing

γ-carbinol allenoates in acceptable yields. Several approaches to this class of compounds

53 include Mukiama Aldol methods78 and the Morita Bayless Hilmann method developed in the Lepore group.21 The latter approach, through a vinylogous aldol mechanism, favors the thermodynamic γ-addition over the kinetically favored α-addition through a brooks rearrangement of the γ-silyl group onto the oxygen of the carbinol. Hammond developed an alkynologous approach in which allenoate or its alkyne tautomer lacking a γ-silyl group, but possessing one alkyl substituent at the γ-position (90 Scheme 19), undergoes aldol addition via a base mediated alkynylenolate intermediate.79 Hammond showed that the addition of TBAF promoted γ-addition of a range of aldehydes. The probable role of TBAF in promoting the formation of the thermodynamic γ-carbinol 91 is that it plausibly better chelates the intermediate of the retro aldol, thus allowing equilibration to occur between the kinetic α-addition product and the γ-carbinol, the latter of which becomes the single product if and only if two equivalents of TBAF are added.

Scheme 19. Alkynylogous addition leading to thermodynamic γ-carbinol product Building on this study of regioselectivity and the use of TBAF, Lepore has recently published a method in which γ-silyl allenoates with alkyl substitution at the α-position (92

Scheme 20) exclusively yield the α-addition products with a quaternary carbon center via nucleophile catalyzed vinylogous aldol addition (MBH) when the reaction occurs in the presence of TBAF. A hypothesized hypervalent silicon intermediate is thought to block γ- addition in this case. They also discovered that after forming a γ-carbinol via the MBH method, a subsequent one-pot second γ-addition to another aldehyde could be achieved by

54 adding stoichiometric TBAF, providing a double addition protocol for the synthesis of bis-

γ-carbinol quaternary allenoate 93.

Scheme 20. TBAF mediated double addition leading to bis-γ-carbinol quaternary allenoates Although the MBH-Brooks method is an excellent method for synthesizing γ-carbinol alkyl allenoates, we found that scaling up the synthesis of γ-carbinol 95 from γ-silyl phenyl allenoate 94 was difficult. This is perhaps due to the lability of the phenyl ester toward the nucleophilic organocatalyst, which may have quenched the catalytic cycle prematurely.

Because of the reactivity of the phenyl allenoates, due to the stability of the phenoxide leaving group in nucleophilic acyl substitution, is critical to the success of the addition and annulation of the γ-carbonol allenoates with cyclic ketoesters, their preparation required an alternative method that avoided nucleophilic reagents.

Scheme 21. MBH-Brooks reaction to form γ-carbinol In the course of the present research, it was discovered that direct deprotonation of an α- substituted, γ-unsubstituted allenoate by strong base (LiHMDS) leads to the one-step formation of the γ-carboxy silyl ether. In the absence of a nucleophilic catalyst, it is likely that this reaction proceeds through an alkynylogous mechanism, which favors γ-addition up to 75% yields. The latter builds on previous work in the Lepore group towards methods for the diverse synthesis of allenyl carbonyl compounds,80 particularly allenoates.

55 The preparation of α-methyl phenyl allenoate was carried out as described in Chapter 1 of this dissertation (Scheme 17). For bis-aryl vitisinol D analogs, the installation of an aryl carbinol at the γ-position was carried out in one pot. To accomplish this, γ-deprotonation was accomplished using LiHMDS. Various aromatic aldehydes were then added followed by quenching with TESCl (Scheme 22). This resulted in the desired allenol compounds

95a - c in good yields, although a high degree of diastereoselectivity was not noted. To our knowledge, this direct approach to γ-carbinols starting with a γ-unsubstituted allenoate has never been previously reported.

Scheme 22. Direct and regioselective substitution at the γ-carbon 2.2.3 Convergent synthesis of bicyclo[3.2.1]oct-2-ene-4,8-diols

Synthesis of mono or bis-aryl analogues of vitisinol D is achieved in an analogous manner to the protocol described in Chapter 1. In the synthesis of bis-aryl analogues, γ-carbinol allenoates add to the two α-positions of the ketoester promoted by cesium carbonate or potassium tert-butoxide. Unlike the sulphone substrates (Chapter 1, Scheme 16), heat is not required to promote this addition with oxy-esters. Addition to the β-carbon of allenoates

52b or 95a-c (Scheme 23) leads to formation of lactones 96 and 97a-c with elimination of the triethylsilyl ether, leading to a highly conjugated system. It is unclear if this elimination precedes or follows the formation of a lactone through acyl substitution of the phenyl ester by the enolate oxygen. Side products, notably various addition products that failed to annulate, were present with all analogues. Yields of the bis-methoxyphenyl lactone were 56 unsatisfactory and ranged from 27-37%, necessitating time-consuming scale ups and repetitions of reactions. Yields of the chlorophenyl analogue lactones were unsatisfactory, but also preliminary. They might be improved with more optimization.

Scheme 23. Convergent synthesis of mono and bis aryl lactone intermediates 2.2.4 Diastereoselectivity and mechanistic hypothesis

In these vitisinol D analogs, the relative configuration of three of the four consecutive stereocenters of vitisinol D is controlled in this synthesis, with the final stereocenter lost due to not installing the third aryl substituent of the natural product. These analogues differ from vitisinol D in that bridgehead carbon (1) bears a carboxyl substituent not present on the natural product. This moiety must be removed in the total synthesis of vitisinol D, but for medicinal purposes, the presence of a carboxylic acid at that position may be favorable in order to enhance serum solubility.

Scheme 24. Diastereoselective mechanism of addition/annulation of allenoates to β-aryl cyclopentyl ketoester The construction of the final bond (C5-C4) leading to compounds 98-99 leads to well- defined stereochemistry for the C5 carbon since only syn addition of the 3-carbon bridge is possible (Scheme 24). However, the C4 carbon has also been produced with complete control of configuration via the reductive aldol mechanism. In this reaction, the facial selectivity of the aldol addition of the reduced aldehyde to the five-membered ring enolate is controlled by a diastereoselectivity cascade from C7 to C1 to C5. This results in the exo- orientation of the aryl group, and the cis relationship of the aryl group to the ester moiety.

Note that this is similar to the three center diastereoselectivity described by Rodriguez noted in Chapter 1 (Scheme 11) but a crucial difference is that the orientation of the prochiral aldehyde is not influenced by any nearby chiral center, so the observed exo orientation of the C4 hydroxyl must be explained through another interaction.

Scheme 25. Proposed diastereoselective mechanism in reductive aldol rearrangement Diastereoselectivity for a hydroxyl on a bicyclo[3.3.1]nonane skeleton was noted in passing by Nicolaou (see Scheme 4 and reference 9). However, the mechanism was not explored in this precedent, as subsequent oxidation of the hydroxyls rendered any diastereoselectivity at those positions inconsequential. In our reductive aldol, compounds

98-99 are isolated as single diastereomers, which raises the question of how the transition state for the reductive aldol rearrangement controls the relative stereochemistry of the

59 hydroxyls. Because our final tartget compounds are not oxidized at the hydroxyls, there is a compelling reason to investigate this diastereoselectivity. We propose a Zimmerman-

Traxler-like transition state as a plausible explanation of the data (Scheme 25).

Zimmerman-Traxler like transition states have been proposed in similar intramolecular systems.81 In the proposed transition state A, the reduced α-β unsaturated aldehyde is chelated to aluminum along with the enolate oxygen such that the former is in the s-trans configuration and an approximation of the 6-membered Zimmerman-Traxler chair conformation is achieved. This transition state leads to facial selectivity of the enolate attack on the aldehyde carbonyl in B. This results in exo configuration of the hydroxyl on the C4 (on the 3-carbon bridge) in C. This mechanism also provides a rationale for the observed preference for reduction of the carbonyl on C8 (on the 1-carbon bridge) of G on the face leading to the endo hydroxyl, as the other face would be less sterically accessible to DIBAL-H due to steric crowding by the exo hydroxyl on the 8-carbon. On the other hand, an open transition state as shown in D would allow intramolecular aldol addition on the opposite face of the aldehyde carbonyl in E, leading to endo orientation of the hydroxyl on C4 in F.

The relative stereochemistry of the diastereomer 98 obtained experimentally was determined by 1-D and 2-D NMR examinations of its spectra and by comparison to the spectra of closely related compounds such as the similar diketone compound (104 Figure

16). The exo orientation of the oxygen on C4 was determined due to a lack of coupling between H4 and H5. H5 is split into a broad triplet by coupling with H6A and H8, with the latter key interaction shown unambiguously on COSY. The exo orientation of H8 is corroborated by an NOE between H8 and the phenyl ring. There is an NOE between H4

60 and the acetyl group on C4, and no other NOE for H4, consistent with its orientation in the pseudoaxial/endo position.

Figure 16. Assignment of observed relative stereochemistry Ambiguity was removed further by comparing with the spectrum of similar compound x, with ketones at C4 and C8. In compound x, there was no difference in the multiplicity of

H7, a doublet of doublets split by the diastereotopic protons H6A and H6B. Instead of a doublet of doublets (which has the appearance of a triplet) however, H5 was split into a doublet in the absence of a second adjacent proton on C8 or C4. This removes any doubt that the doublet of doublets represents the proton on C7 in both compounds, and that the triplet is the bridgehead carbon in compound 98. The change in the multiplicity of H5 from triplet to doublet in comparing 98 to compound 104 respectively is only consistent with H4 being in the pseudoaxial/endo position (expected axial-bridgehead J = 0.1).82

2.2.5 End stage synthesis

Upon synthesis of 98-99, compounds were acyl protected at the hydroxyl positions in nearly quantitative yields using a standard procedure, leading to 100-101. This reduced the polarity of the compounds, making chromatographic separation of the final product easier.

The final product of the synthesis for all analogues was the carboxylic acid. Upon treatment with TBAF in THF at room temperature, the silyl ester was cleanly removed to reveal 102- 61 103 in acceptable yields. It is also possible to directly deprotect diols 98-99 to reveal carboxylic acids 105-106 (Scheme 27), although the higher polarity of these products makes purification more difficult.

Scheme 26. End stage synthesis

Scheme 27. Direct deprotection of diols

62 2.3 Biological testing

Figure 17. Initial screening of compounds against resveratrol for neuroprotection against oxidative stress in Drosophila melanogaster (vertical bars shown as mean  S.E.M, 2-4 replicates each). Resveratrol was found to enhance the survival rate of flies after anoxia was induced via nitrogen asphyxiation. Our compounds showed promising initial activity in this model, but it was ultimately abandoned due to engineering issues that negatively affected reproducibility. Instead we sought an experiment to measure any effect of these compounds on neuronal function under oxidative stress more directly in vivo.

63 The three initial compounds were instead subjected to the neuromuscular junction (NMJ) post synaptic transmission assay (see section 2.1.2). Resveratrol was included as a reference neuroprotectant. Using a similar to past procedures protocol33, wandering third instar larvae (approximately 110 hours old) from Drosophilia melanogaster were utilized.

Larvae were dissected, and the internal organs and central nervous system were removed.

Intercellular electrodes measured the post-synaptic excitatory junction potentials with a resistance of 60-90 M. Excitatory junction potentials were measured until synaptic failure occurred, as characterized by a recorded potential amplitude of < 1 mV.

Electrophysiological recordings from the larval preparations were conducted in hemolymph-like 3 (HL3) saline containing 2.25 mM H2O2 in order to quantify neurotransmission’s tolerance to acute oxidative stress. In each experiment, the potential neuroprotectant was added to the HL3 saline along with the H2O2 so that both reached the neuromuscular junction concomitantly.

When a wash of H2O2 (2.25 mM) alone was introduced, the time to synaptic failure was shortened from approximately 150 min to approximately 40 min (Figure 17 one-way

ANOVA, F(4,58)= 8.731, P<0.05). Introducing resveratrol at 25 nM along with the 2.25 mM wash of H2O2 restored the time to synaptic failure to approximately 150 min.

Compound 103 was tested at 25 nM and extended time to synaptic failure to approximately

100 minutes with 2.25 mM hydrogen peroxide.

Figure 18. Dose-response plot of most active compound against resveratrol (vertical bars shown as mean  S.E.M, 2-4 replicates each). Although this compound has significant activity, it is less active than resveratrol, which is a partial falsification of the hypothesis that vitisinol D analogues modeled on fisetin might be especially active in this model versus resveratrol. To our surprise, mono-aryl compound

102 was more potent than resveratrol at restoring time to synaptic failure at 0.5 and 1 nm versus 2.25 mM hydrogen peroxide (Figure 18, one-way ANOVA, F(4,58)= 8.731,

P<0.05). This led to a reappraisal of our hypothesis in which the similarities between the mono-aryl vitisinol D analogues and the structure of resveratrol were noted. The precursor

100 was the least active compound tested, which suggests that either the carboxylic moiety is important for solubility or substrate binding, or the bulky ester may interfere with binding. In all of these single diastereomer compounds, it is plausible that the orientation

65 of the hydroxyl groups may have direct bearing on the activity of the compounds in their interaction with their biological targets.

2.4 Conclusion

With the existing data it is impossible to definitively rule out hypotheses about the target of the active compounds, such as the SIRT activation hypothesis. This is because neuroprotective activity at the neuron could be due to the direct action of the compounds, or it could be the result of a cascade initiated by the direct action of the compounds at some upsteam target.

One conclusion, however, can be drawn. Our data conclusively falsifies the hypothesis that resveratrol and the other compounds tested were conveying neuroprotection from oxidative stress by the mechanism of free radical scavenging. The rationale for drawing that conclusion is this: the effective dose of the most active compound was found to be in the low single digit nanomolar range, and resveratrol had similar activity (Figure 8). At the same time, these compounds were introduced along with hydrogen peroxide at 2.25 mM.

The reactive oxygen species is present at six orders of magnitude higher concentration than the potential antioxidant.

This argument is supported by the observation that the isoresveromorph compounds, compared to resveratrol, have far less conjugation. Resveratrol’s fourteen carbons are all sp2 hybridized and in complete conjugation with one another. Out of the isoresveromorphs’ corresponding fourteen carbons, six are sp3 hybridized, including two of the hydroxyl groups, and of the remaining eight sp2 hybridized carbons, only six are on conjugation with hydroxyl moieties, namely those oxygens on the phenyl ring. Lastly, those oxygens are protected as methoxy groups in all of the isoresveromorph compounds tested. The point of

66 this analysis is that if the mechanism of neuroprotection for resveratrol were free radical scavenging, we should not expect any of our isoresveromorph compounds to match resveratrol’s neuroprotectant activity, let alone exceed it.

Therefore, we can tentatively conclude that the isoresveromorph compounds convey neuroprotection through a mechanism other than free radical scavenging. Further experimentation is warranted to identify the biological target.

2.5 Experimental Section

ESI-TOF MS spectrometer (DART ion source). All reagents were purchased from commercially available sources and were used without further puriﬁcation. All solvents were dried over activated 3 Å molecular sieves.

Synthesis of compound 88

A flame-dried flask was affixed with an addition funnel and charged with THF (200ml).

Under Ar, phenyl bromide (2eq, 40mmol) was added. The vessel was cooled to -78°C and n-BuLi (2 eq, 25ml, 1.6M) was added dropwise. After stirring for 1h at temperature, CuCN

(1eq, 20mmol) was added through the addition funnel with 2x15ml THF. This was stirred at temperature, then the dry ice bath was removed to ensure complete dissolution of the

CuCN with 10 min additional stirring. The dry ice bath was returned and the reaction was stirred at -78°C for 10 min longer. Cyclopentene-2-one (1eq, 20mmol) was added diluted in 4 ml THF through the addition funnel dropwise over 20 min. The solution turned yellow in color, and was left stirring at temperature for 2 h. Ethyl cyanoformate (2 eq, 40 mmol) was added, and the reaction was stirred at temp for an additional 5 h. The reaction was quenched with saturated ammonium chloride and the organic extracts were stripped of solvent and purified via silica column using a gradient of ethyl acetate in hexanes, giving pure 88 in 45% yield.

Synthesis of compound 89

To a round bottom flask was affixed a Dean Starke trap and a condenser. Compound (1 eq,

8,89mmol) was taken 88 in 90 ml toluene with dibutyltin oxide (0.1 eq, 0.89 mmol) and

TMS-ethanol (10 eq, 26.7 mmol) and the solution was refluxed for 3 days. After work-up, 68 the organic extracts were concentrated and purified via silica column using 4% EtOAc in hexanes to give pure 89 in 74% yield.

General Procedure for the Preparation of Compounds 96-97. To a solution of 89 (0.37−2.0 mmol, 1 equiv) in acetone (5 mL mmol−1) was added Cs2CO3 (0.37−2.0 mmol, 1 equiv) and 52b or 95 (0.44−2.4 mmol, 1.2equiv). The reaction mixture was stirred at room temperature for 1-5 days. Upon reaction completion (as monitored by TLC), the crude reaction mixture was filtered and the filtrate was evaporated and purified using silica gel column chromatography with ethyl acetate/ hexanes (3:7) to give pure 96-97 in 56−80% yields.

Compound 107 was taken in a flame-dried flask and charged with THF and cooled to -

78°C under Ar. To it was added DIBAL-H (1.1 equ, 1M in hexanes) The reaction was stirred at temperature for 4 h, after which it was worked up with HCl solution and the crude was concentrated and taken to the next step. The presumed intermediate ketol was dissolved in DCM, and DMP (1.1eq, 1.4M in DCM) was added. The reaction was stirred for 30 min and the crude was purified via silica column using 20% EtOAc in hexanes to give compound 108.

69 HNMR: 6.32-6.29 (t, 1H, J = 1.96 Hz), 6.22-6.16 (m, 3H), 4.14-3.99 (m, 4H), 3.74 (s, 6H),

3.65-3.61 (d, 1H, J = 7.83), 3.42-3.37 (dd, 1H, J = 3.91 Hz, 9.78 Hz), 2.39-2.26 (m, 2H J

= 3.91 Hz, 10.17 Hz), 2.18 (s, 3H), 0.76-0.72 (m, 2H) -0.03 (s, 9H)

CNMR: 198.0, 197.8, 166.2, 161.7, 161.2, 145.4, 128.1, 98.8, 68.1, 64.0, 62.3, 55.5, 47.6,

32.4, 31.8, 29.3, 25.5, 22.9 21.6 17.3 14.4, -1.44

General Procedure for the Preparation of Compounds 98-99.

To a solution of lactone (96-97) (1 equiv) in DCM (0.1M) at -20°C was added 1 M DIBAL-

H solution (1 equiv), and the reaction mixture was stirred at 0 °C for 4 h. The reaction was worked up with 3N HCl solution. Organic compounds were extracted with DCM three times, and the pooled organic extracts were stripped of solvent and purified by silica gel column to give the pure products 98-99 as single diastereomers. Yields range from approx.

20-60%

Compound 98

1H NMR: 6.27 (s, 3H), 4.87-4.85 (d, 1H) 3.96-3.83 (m, 2H) 3.75 (s, 6H) 3.55-3.52 (d, 1H)

3.17-3.13 (dd, 1H) 2.70-2.66 (td, 1H) 1.86 (s, 3H) 1.73 (s, 3H) -0.05 (s, 9H)

13C NMR: 174.74, 160.46, 147.20, 129.68, 127.25, 106.95, 97.48, 76.38, 74.76, 62.93,

61.67, 55.22, 51.33, 40.37, 32.88, 16.84, 16.70, 15.54, 1.74

(M+Na) expected 471.2173 found 471.2180, (M+2Na) found 919.4463

70 Compound 99

1H NMR: 7.43-7.39 (d, 1H), 7.33-7.29 (dt, 1H) 6.88-6.83 (dt, 1H) 6.81-6.76 (d, 1H) 6.49-

6.42 (m, 1H) 6.41-6.34 (m, 1H), 6.33-6.26 (m, 4H) 4.99-4.90 (dd, 1H) 3.86-3.71 (M, 15H),

3.67-3.58 (m, 1H), 3.41-3.33 (dd, 1H), 2.79-2.71 (dd 1H), 2.17 (s, 1H), 2.04 (s, 1H), 2.01

(s, 2H), 1.28-1.18 (m, 3H), 1.01-0.95 (m, 1H), 0.98-0.83 (m, 2H), 0.54-0.47 (m, 2H), 0.01-

-0.18 (m, 9H)

13C NMR: 160.44, 132.99, 132.21, 131.02, 129.96, 129.75, 127.27, 123.33, 114.00, 113.56,

107.04, 105.29, 97.63, 97.54, 74.60, 62.99, 62.91, 61.52, 55.32, 55.25, 55.17, 51.39, 40.29,

32.73, 31.57, 22.62, 21.03, 18.32, 17.03, 16.80, 14.20, -1.78, -1.89

(M+Na) expected 589.2592 found 589.2593

General Procedure for the Preparation of Compounds 100-101

Compound 98 or 99 was dissolved in DCM (0.1M) and acetic anhydride (5 eq), DMAP

(0.1 eq), and pyridine (5 eq) were added under argon with stirring at room temperature.

Reaction was stirred from several hours to overnight, when TLC spotting showed complete conversion of the starting material to a less polar product. Solvent was removed under vacuum and the crude product was purified via silica column chromatography with ethyl acetate/hexanes (3:7) to afford pure 100-101 in near-quantitative yields.

Compound 100 ‘Fly 3’

1H NMR: 6.30-6.26 (m, 3H), 5.52-5.50 (d, 1H), 5.00-4.98 (s, 1H), 3.83-3.77 (m, 1H), 3.75-

3.72 (s, 6H), 3.18-3.12 (dd, 1H), 2.95-2.91 (dd, 1H) 2.05 (s, 3H), 2.02 (s, 3H), 1.90 (s, 3H)

1.69 (s, 3H)-0.05-0.22 (s, 9H) 71 13C NMR: 171.12, 170.60, 160.53, 145.94, 133.72, 124.43, 109.99, 106.70, 98.20, 76.33,

74.30, 62.58, 60.75, 55.25, 52.68, 37.40, 33.19, 21.25, 21.10, 16.75, 16.50, 16.35, -1.76

+ + + HRMS: Expected MW: 532.69 Found [M+H] 533.2528 [M+NH4] 550.2813 [M+Na]

555.2404

Compound 101

1H NMR: 7.37-7.33 (d, 2H); 6.88-6.85 (d, 2H), 6.68-6.62 (d, 1H), 6.36-6.26 (m, 4H), 5.57-

5.54 (d, 1H), 5.12-5.10 (s, 1H), 3.83-3.77 (s, 3H), 3.77-3.72 (s, 6H), 3.72-3.66 (m, 1H),

3.62-3.53 (m, 1H), 3.36-3.30 (dd, 1H), 3.02-2.97 (dd, 1H), 2.11-2.08 (s, 3H), 2.07-2.05 (s,

3H), 1.01—0.18 (m, 9H)

13C NMR: 171.16, 170.50, 170.20, 160.51, 159.28, 145.60, 137.06, 132.53, 129.98, 127.43,

126.98, 124.23, 114.00, 106.83, 98.16, 76.33, 74.16, 62.54, 60.94, 55.32, 55.26, 52.81,

37.48, 33.13, 21.28, 21.11, 17.95, 16.86, -1.83

General procedure for the synthesis of compounds 102-103

To a stirred solution of compound 100-101 (0.047−0.065 mmol, 1 equiv) dissolved in THF

(2 mL mmol−1) was added TBAF·3H20 (10 eq), and the reaction was stirred at room temperature for 1h. Upon reaction completion (as monitored by TLC), solvent was

72 evaporated and the residue was puriﬁed using silica gel column chromatography with ethyl acetate/hexanes (3:7) to afford pure 102-103.

Compound 102 ‘Fly 2’ 66% yield.

1H NMR: 7.37-7.34 (d, 2H) 7.00-6.94 (d, 1H) 6.83-6.79 (d, 2H) 6.42-6.39 (d, 2H) 6.29-

6.24 (d, 1H) 6.20-6.18 (t, 1H) 5.54-5.52 (d, 1H) 5.12-5.09 (b, 1H) 3.81-3.76 (s, 3H) 3.76-

3.69 (s, 6H) 3.28-3.23 (dd, 1H) 3.12-3.06 (m, 1H) 2.99-2.94 (t(b) 1H) 2.09-2.08 (s, 3H)

2.05-2.04 (s, 3H) 1.85-1.83 (s, 3H)

13C NMR: 174.25, 171.56, 170.70, 160.01, 158.71, 148.88, 140.01, 131.03, 130.45, 127.28,

127.00, 125.22, 113.72, 106.44, 97.89, 77.51, 77.23, 75.50, 61.91, 58.78, 55.28, 55.12,

52.88, 52.26, 37.91, 33.89, 24.22 21.66, 21.22, 20.45, 19.72, 18.11

HRMS: Expected MW: 432.46 Expected (M+Na) 455.1676 Found (M+Na) 455.1685

Found (M+NH) 450.2098 Found (M+H) 433.1826

Compound 103 ‘Fly 1’

1H NMR: 7.37-7.34 (d, 2H) 7.00-6.94 (d, 1H) 6.83-6.79 (d, 2H) 6.42-6.39 (d, 2H) 6.29-

6.24 (d, 1H) 6.20-6.18 (t, 1H) 5.54-5.52 (d, 1H) 5.12-5.09 (b, 1H) 3.81-3.76 (s, 3H) 3.76-

3.69 (s, 6H) 3.28-3.23 (dd, 1H) 3.12-3.06 (m, 1H) 2.99-2.94 (dd 1H) 2.09-2.08 (s, 3H)

2.05-2.04 (s, 3H) 1.85-1.83 (s, 3H)

13C NMR: 174.25, 171.56, 170.70, 160.01, 158.71, 148.88, 140.01, 131.03, 130.45, 127.28,

127.00, 125.22, 113.72, 106.44, 97.89, 77.51, 77.23, 75.50, 61.91, 58.78, 55.28, 55.12,

52.88, 52.26, 37.91, 33.89, 24.22 21.66, 21.22, 20.45, 19.72, 18.11

General procedure for the synthesis of compounds 105-106

To a stirred solution of compound 98-99 (1 equiv) dissolved in THF (2 mL mmol−1) was added TBAF·3H20 (10 eq), and the reaction was stirred at room temperature for 1h. Upon reaction completion (as monitored by TLC), solvent was evaporated and the residue was puriﬁed using silica gel column chromatography with ethyl acetate/hexanes (3:7) to afford pure 105-106.

Compound 106

1H NMR: 7.47-7.45 (d, 1H), 7.35-7.33 (d, 1H), 6.82-6.79 (d, J =8.81Hz, 1H), 6.79-6.75

(m, 1H), 6.50-6.46 (m, 1H), 6.43-6.39 (dd, 8.32Hz 2H) 6.32-6.25 (m, 1H) 6.22-6.20 (q,

1H) 4.91-4.89 (d, J = 5.79, 0.5H), 4.83-4.81 (d, J = 5.64Hz, 0.5H), 3.79-3.78 (d, 3H) 3.77

(s, 2H), 3.74-3.73 (s, 6H) 3.62-3.59 (b, 1H) 3.56-3.53 (B, 1H), 3.35-3.28 (m, 1H), 2.74-

2.60 (M, 8H), 2.04 (S, 3H)

HRMS: expected 466.53 found [M + H]+ 467.2056 [M + Na]+ 489.1866

General synthesis of γ-carbinol 95:

LiHMDS (1 eq) was prepared freshly in THF in an oven dried flask with a stir bar at -78°C.

Gamma silyl-protected phenyl allenoate (1 eq) was added dropwise over several minutes.

After 45 minutes stirring at temperature, aldehyde (1 eq) was added. After an additional hour of stirring at temperature, TESCl (1.5eq) was added and the reaction was left stirring overnight, coming to room temperature on its own. After workup with saturated

74 ammonium chloride, the organic extracts were purified via silica gel column using a gradient from 0-20% ethyl acetate in hexanes to afford pure 95a in 75% yield. NMR spectra was identical to published spectra.

75 Chapter 3

Ammonium catalyzed cycloadditions and evidence for a cation-π

interaction with alkynes

3.1 Introduction

3.1.1 Cation-π Interaction

Over the past two decades, cation–π interactions have been characterized and identified in numerous biological and host–guest systems.83 More recently, this relatively new intermolecular force has found application in organic synthesis playing a key role in a variety of stereoselective reactions.84

Scheme 28. Selected reactions stabilized by cation-π interactions A few conspicuous examples include the development by Dougherty of cyclophane host- guest complex 109 (Scheme 28) that catalyzes an alkylation reaction leading to ammonium cations,85 face-selective additions by Yamada to pyridinium/arene complex 110 generated from ester addition to 111 leading to the stereoselective synthesis of chiral 1,4- dihydropyridines 112,86 and intramolecular Schmidt reactions of 2-substituted ketone 113 with regiocontrol via diazonium/phenyl interactions in intermediate 114 leading to bridged bicyclic lactam 115 by Aube.87 While these and other examples demonstrate a cation binding force in aromatic systems, analogous cation interactions with other π-systems have only rarely been reported. In particular, reports of associations between alkynes and cations that are best explained through an electrostatic quadrupole-monopole interaction are scarce.

Figure 19. Cation-π Interaction of Arenes, and Proposed Ammonium-Alkyne Interaction The cation-π interaction described subsequently in this chapter between an ammonium cation and a terminal alkyne is proposed as the best explanation of the observed association between the two moieties, as well as the best explanation of spectral effects one would expect from such an interaction. The ammonium-alkyne interaction is proposed to operate through the same electrostatic quadrupole-monopole dynamic as the more standard cation-

π interaction involving an arene as the quadrupole and some cationic monopole (Figure

19). One rare precedent for a cation-π interaction involving a carbon-carbon triple bond as the electron-rich quadrupole is the interaction of alkynyl sidearms with complexed alkali metals in diaza-18-crown-6 ethers (116 Figure 20).88

Figure 20. Complexation of cation in crown ether involving alkynyl sidearms 3.1.2 Cyclization reactions

Previously, Lepore reported a cyclization reaction of β-alkynyl hydrazines (116, Scheme

29) to give azaproline derivatives 117 in high enantiomeric excesses under kinetic resolution conditions.89 Although this cyclization was unexpected, the new route to azaproline derivatives was worth exploration both due to the mechanistic questions it posed 78 and because azaprolines have become more relevant recently with applications in bioorganic90 and medicinal chemistry.91 Notably, these reactions were catalyzed by chiral ammonium and phosphonium salts to give nonracemic azaproline products, whereas other methods utilized chiral transition metal catalysts such as magnesium bisoxazole,92 and titanium BINOL-ate.93

Scheme 29. Previous discovery from the Lepore Group While heteroatom additions to unactivated alkynes are well known in the presence of transition metal catalysts such as palladium94 and gold,95 there are only a few reports of such reactions involving non-metal catalysts. Hammond and co-workers demonstrated that gem-difluoro propargyl amides 118 undergo 5-endodig cyclization to give the γ-lactam

119 in the presence of stoichiometric tetrabutylammonium fluoride (TBAF).96 This reagent was also shown by Jacobi97 and later by Sakamoto and coworkers98 to eﬃciently give isoindole products 120 via a 5-exo-dig cyclization of 121 (Scheme 30).

Scheme 30. Precedents for TBAF catalyzed cycloadditions These previous studies pointed to TBAF as a soluble source of fluoride thought to be an ideal mild base for this transformation. However, the previous success by Lepore in the synthesis of non-racemic azaprolines from propargyl hydrazines using chiral ammonium

79 catalysts suggested a more discriminating interaction with the substrate. We sought to develop a mechanistic hypothesis to account for this intramolecular cyclitive attack of carbamate nitrogen in substrates such as compound 116 to an unactivated alkyne in which an ammonium cation plays some catalytic role. We hypothesized that the ammonium cation may be stabilizing a transition state associate with the rate-limiting step and proposed a possible interaction between ammonium and the alkyne moiety to explain the proximity and coordination, at the very least, of the ammonium during the transformation. This proximity hypothesis was based largely on the observation during Lepore’s kinetic resolution studies that a chiral ammonium cation is sensitive to the configuration of the quaternary carbon directly bonded to the carbon-carbon triple bond moiety.

3.1.3 Synthesis of propargyl hydrazines

The γ-silyl allenoate (123 Scheme 31) was conveniently prepared by dehydration of β- ketoester 122 also developed by Lepore and Maity.99 Propargyl hydrazine 126 was synthesized from γ-silyl allenoate 124 via DBU catalyzed addition of diisopropyl azodicarboxylate 125. Subsequent treatment with TBAF in methanol solution removes the silyl group from the terminal alkyne without promoting the cyclization that was discovered serendipitously by Maity while trying to deprotect the alkyne with TBAF in THF. Finally, it was observed that treatment of the deprotected propargyl hydrazine 116a (and similar propargyl hydrazines) with TBAB and KF led quickly to the cycloaddition product 117a

(and similar dehydroazaproline derivatives). However, KF used alone did not lead to cyclization, even with prolonged reaction times. This result led us to speculate that the ammonium cation may be playing some novel catalytic role.

Scheme 31. Synthesis of β-alkynyl hydrazines and azaprolines 3.2 Mechanistic Investigation

3.2.1 Initial Transition State Hypotheses

Our initial hypothesis for the role of ammonium salts in our reaction was that these act as phase transfer agents serving to solubilize weaker bases such as fluoride.100 We noted that stronger, soluble bases such as potassium tert-butoxide led to rapid cyclization in organic solvents. With a strong enough base, no tetraalkylammonium salt was needed, but we were not convinced that phase transfer catalysis alone explained Maity’s previous observation that catalytic TBAB enhanced reaction rate with weaker bases such as carbonates and fluorides. From the conjecture that ammonium might be playing a more active role in the mechanism of base-promoted cycloaddition, we hypothesized an early cycloaddition transition state (A, Figure 21) with deprotonation of the carbamate nitrogen being the rate- limiting step, followed by intramolecular 5-endo-dig nucleophilic addition. An alternative hypothesis, a cycloaddition with a late transition state B would be plausibly stabilized by an ionic interaction of the resulting vinyl anion intermediate. In this alternative hypothesis, the ammonium’s role is not to activate the alkyne towards nucleophilic addition but rather to stabilize the formation of this probably high-energy vinyl anion. In a third hypothesis, 81 the role of ammonium in transition state C is to assist the deprotonation of the carbamate nitrogen through ion pairing with the developing negative charge. The cation-π interaction in this hypothesis explains the initial proximity of the ammonium cation to the carbamate moiety.

Figure 21. Transition state analysis to account for the role of ammonium. From each hypothesis can be derived a number of testable predictions that could falsify it if contravened by experimental evidence. In the case of the early cyclization TS hypothesis, one would expect to see a rather large observed change in the alkynyl bond order upon infrared or Raman spectroscopy when the propargyl hydrazine is in the presence of TBAB

(TBAB alone does not lead to cyclization due to the non-basicity of bromine). The late cyclization hypothesis predicts a vinyl anionic intermediate rather than some concerted transition state leading directly to 117.

One key piece of circumstantial evidence supporting the tetraalkylammonium’s involvement in the mechanism that is compatible with all three TS hypotheses is the kinetic resolution reaction developed by Maity and Lepore in their precedent work.

Figure 22. Kinetic resolution catalyzed by a chiral ammonium catalyst We reproduced a key kinetic resolution experiment from Maity’s work (Figure 22) where

Maruoka’s catalyst A was used to kinetically resolve a racemic mixture of 116b and yield highly enantioenriched 117b. This result led us to reason that the ammonium cation must be sensitive to the steric crowding at the quaternary carbon of 116b, and its proximity to that region of the molecule would be explained by an interaction with the alkyne. However, one limitation of these data is that the kinetic resolutions were performed as heterogeneous biphasic systems which introduces the confounding factor of phase transfer catalysis into any explanation of the catalytic mechanism involving the tetraalkyl ammonium salt.

3.2.2 Rate experiments

Attempts were made to carry out a homogeneous version of the kinetic resolution cyclization described in the previous section. We argued that a reaction that proceeds under homogeneous conditions would point to another role for the added ammonium salts other than phase transfer catalysis. An initial experiment was carried out using pyrollidine in

THF. Analysis by TLC revealed that the reaction proceeded slowly in the presence of catalytic tetrabutylammonium bromide (TBAB). By contrast, almost no reaction was 83 evident by TLC after several days in the absence of added TBAB. To qualitatively analyze this hypothesis, we performed further homogeneous reactions employing soluble weak bases with basicities within several log units of fluoride. For this study (Table 4), we chose

1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and sodium phenoxide, which were entirely soluble in several common organic reaction solvents (The pKa in acetonitrile of conjugate acids are HF (25.2), DBU-H+ (24.3), and phenol (29.1).101 We observed a substantial rate enhancement with these bases in the presence of TBAB whose bromide counter anion is essentially non-basic. For example, the reaction of substrate 116a in MeCN in the presence of stoichiometric DBU gave azaproline product 117a in 50% yield in 722 min (t1/2).The same reaction gave a t1/2 of 196 min in the presence of added TBAB (1 eq) for a relative rate of 3.7 (Table 4, entry1)

a t1/2 (min) Entry Base Solvent Salt No Salt Salt Rel Rateb c 1 DBU MeCN Bu4NBr 722 196 3.7

2 NaOPh MeCN Bu4NBr 1140 52 22

3 NaOPh MeCN Et4NBr 1140 54 21 n 4 NaOPh MeCN Pr4NBr 1140 52 22

5 NaOPh MeCN Oct4NBr 1140 46 25

6 NaOPh THF Bu4NBr 7515 33 228

7 NaOPh pyridine Bu4NBr 360 125 2.9

8 NaOPh DMF Bu4NBr 70 1 70 a t1/2 represents time required for 50% conversion of the starting material into product determined using NMR. Reactions performed at 0.1 M. bRelative rate c = t1/2(no R4NBr) /t1/2(R4NBr). Reaction performed at 0.025 M to ensure solubility. Table 4. Role of base in TBAB-catalyzed cyclization of 50a under homogeneous conditions.

84 Interestingly, the ionic base NaOPh led to much higher relative rates in MeCN and other solvents. The higher ratio observed with this charged base may be due to its ionic attraction to the tetrabutylammonium cation, which we propose to be associated with the triple bond of the substrate. Several other ammonium bromide salts also catalyzed the cyclization under homogeneous conditions (entries 3–5). The most significant rate enhancement was measured in tetrahydrofuran (THF) (entry 6). The reaction was very slow in the absence of

TBAB and product decomposition was observed after long reaction times. It is possible that the differences in relative rates are a function of the degree of dissociation of the TBAB ion pair in the solvents examined (tetraalkyl ammonium bromides (including TBAB) are usually well dissociated in dipolar aprotic solvents).102 Nevertheless, considering the complete solubility of the bases used in this study, it appears that the ammonium salt enhanced these cyclization reactions by something other than a phase transfer mechanism.

3.2.3 Raman evidence

Based on these homogeneous reaction data, we surmised that an alternative pathway for ammonium salt involvement in the aforementioned cyclization reactions (and those of

Hammond and Sakamoto) could be through a cation–π association with the carbon–carbon triple bond in the substrate. To our knowledge, ammonium–π interactions have not been demonstrated with alkynes. Such an interaction should be theoretically possible here since they are thought to be largely electrostatic (possibly with some π to σ* involvement),103 which is not exclusive to aromatic systems. To directly observe this interaction, we turned to Raman spectroscopy (RS) since the polarizable carbon–carbon triple bond appears as an intense band (more so than IR) that is easily recognizable in an uncrowded region of the spectrum. Moreover, RS has been used to elucidate metal–tryptophan interactions in a

85 periplasmic protein (CusF)104 and, more recently, to understand conformational changes in small molecules induced by ammonium/indole associations.105

Figure 23. Raman titration of alkynyl hydrazine with TBAB Thus, a series of spectra of 116a were measured using RS in the absence or presence of varying amounts of TBAB noting especially the alkyne stretching (ν C≡C), carbonyl stretching (ν C=O), and CH2/CH3 deformation (δ CHn) regions (Figure 23). In these experiments, TBAB was mixed in a dichloromethane solution of compound 116a and then drops of this homogeneous mixture were deposited onto a slide for analysis. Spectral intensities were normalized so that all spectra have the same carbonyl stretching band intensity (Figure 12). The inset provides a detailed view of the alkyne stretching band of

116a, showing a peak shift to lower wavenumber and intensity decrease as increasing amounts of TBAB were added. Critically, other vibrational modes showed no shift in wavenumber in the presence of TBAB. Similar trends were also observed in different solvents using the RS method described above. Thus when compound 116a was mixed with TBAB (5 eq), a shift of -11 cm-1 was observed in THF and -8 cm-1 in DMF.

86 Other alkynes, including one bearing no other functional group, also exhibit a down-shift of the C–C triple bond wavenumber in the presence of TBAB. Using the same TBAB titration experiments as described above for compound 116a, five other terminal alkynes exhibited wavenumber shifts ranging from 4 to 12 cm-1 (Table 5). There appears to be no correlation of the C–C triple bond shift with other functional groups present in the molecule. Raman measurements of alkyne 126 in the presence of KBr gave no significant shift (Table 5, entry 7). This data point is an important control because one worry was that the rate enhancing effect of TBAB might have been merely due to an increase in the ionic strength of the solution from the added salt. If this were the case, then we would expect to see any effect due to added salts on Raman spectroscopy equally with KBr. The lack of any significant red shift upon saturation of the sample with KBr suggests that it is the tetrabutylammonium cation that is causing this shift to occur.

TBAB (0 eq) TBAB (5 eq) Δ Entry Alkyne (cm-1) (cm-1) (cm-1)a 1 116a 2124 2113 11 2 126 2125 2114 11 3 127 2124 2117 7 4 128 2121 2109 12 5 129 2114 2110 4 6 130 2120 2108 12 7b 126 with 2125 2125 <1 KBr (5 eq) aRepresents the change in wavenumbers in the sample measured with TBAB (5 eq) relative to the sample containing no TBAB. bNo TBAB was added. Table 5. Raman measurements of alkynes with added TBAB An interaction between the tetrabutylammonium cation and the carbon–carbon triple bond of alkynes seems to be the most parsimonious explanation of these data, and the best explanation of the nature of this proposed interaction is that it is a cation–π interaction. The direction of the shift towards lower wavenumber suggests that electron density from the triple bond is drawn towards the cation, slightly decreasing the bond order of the alkyne though the maximum limit of this shift is well below the change in bond order seen in transition metal alkyne interactions (The p-coordination of an alkynyl unit by Cu(I) within a tetra-nuclear mixed-metal platinum(II)–copper(I) complex led to a shift of 82 cm-1).106

Propargyl hydrazine 116a was observed through RS using the same drop deposition technique employed above both alone and with five equivalents of TBAB. The (νN-H) carbamate hydrogen stretch normally has a frequency of approx. 3300cm-1 as a broad, weak absorbance. In the presence of TBAB, the peak is shifted to approx. 3420cm-1. This shift to higher frequency, or ‘blue shift’, is consistent with the loss of a stabilizing interaction of

88 the approximate magnitude of a hydrogen bond (there is a report of a shift from as much as 3410 cm-1 to 3310 cm-1 from a non-hydrogen bonded carbamate N-H stretch to the hydrogen bonded counterpart).107 One plausible explanation is that the preferred conformation of 116a and related propargyl hydrazines maybe be the hydrogen-bonded rotomer A (Figure 19). The association of TBAB with the alkynyl moiety may induce a conformational shift, as in rotomer B where that hydrogen bond is lost to the carbamate proton.

Figure 24. Raman spectra showing blue shift of N-H stretch with added TBAB 3.2.4 DFT calculations

Computational modeling was used to further elucidate mechanistic possibilities in the cycloaddition of propargyl hydrazine 116a promoted by TBAF. These studies were also undertaken to reveal any potential interaction between the C-C triple bond and the ammonium ion. These DFT calculations were performed by Prof. Andrew Terentis employing B3LYP functional and 6-311+G(2d, p) basis set and implicit CPCM solvent

(DCM) model. These computational studies were initially intended to distinguish between two hypothesis that have emerged from our previous studies of this cycloaddition reaction. 89 Hypothesis 1: The tetraalkylammonium cation activates the C-C triple bond towards nucleophilic cyclitive attack.

Hypothesis 2: The tetraalkylammonium increases the acidity of the carbamate nitrogen by stabilizing its conjugate base.

To investigate Hypothesis 1, structural optimization and natural bond orbital (NBO) calculations were performed on the propargyl hydrazine substrate alone and in the presence of TBAB. Natural bond order analysis has long been considered useful for deducing subtle intermolecular interactions.108 With these calculations, we sought to observe perturbations to the system in the presence of ammonium. The result of these calculations was that the

NBO charge on the alkynyl carbon (C1) in the absence of TBAB was -0.186, which was not significantly different from the charge in the presence of TBAB (-0.179). With so little difference in the electron density of the C1 carbon (0.007), it is unlikely that the carbon is more reactive to cyclitive attack (which is Hypothesis 1). This computational finding is consistent with our Raman data (see previous section), in which only a small red shift of the C-C alkynyl stretching frequency (~ 10 cm-1) was observed in a dose-dependent manner with added TBAB. A direct activation of the C-C triple bond by metals such as Cu(I) usually results in a much larger shift (100 cm-1), signifying a change in the bond order from that of an alkyne to that of an alkene.106

These calculations also point to a conformational change of the propargyl hydrazine 116a in the presence of TBAB. The distance between carbamate nitrogen N5 and C1 is shortened from 4.76 Å to 3.47 Å (Figure 25). This finding suggests a third hypothesis in which the ammonium catalyzes the reaction by orienting C1 and N5 in an “attack-ready” position that is more favorable for cyclization. In addition, the N5 proton is 2.67 Å away from the

90 ethyl ester carbonyl, making the explanation of the (ν N-H) carbamate hydrogen stretch undergoing a blue shift in the presence of TBAB due to a disrupted hydrogen bond more plausible.

Figure 25. Computational evidence for a conformational change in the presence of TBAB Modelling experiments were performed to evaluate Hypothesis 2 by calculating the energies of the propargyl hydrazine conjugate base (deprotonated at carbamate nitrogen

N5), both in the absence and in the presence of TBAB. In the experimental system, these reactive species cannot be isolated since the quickly undergo cycloaddition. However, for our computational study, these conjugate bases were not allowed to react further so that we could study their energetics and molecular geometries. Sodium and bromide were included as counter ions (to the alkynyl hydrazine anion and tetrabutylammonium, respectively) in these calculations to maintain overall charge neutrality of the systems in nonpolar solvent

(DCM).

The results of these calculations showed that the geometry of the optimized structure of the propargyl hydrazine anion (sodium salt) gave a distance of 4.66 Å between C1 and N5 (A,

Figure 26). The NBO charge in this system was -0.641 on N5 (sodium salt). In the presence

91 of TBAB, the optimized system gave a C1 - N5 distance of 3.50 Å and an NBO charge on

N5 of -0.601 (B). TBAB did not greatly change the charges on the C-C triple bond in this calculation. The N-H bond of the propargyl hydrazine 116a is slightly less polarized in the presence of TBAB (see relative NBO differences, Table 6). The calculations appear to support Hypothesis 2 in that they show the TBAB plausibly stabilizing the propargyl hydrazine anion (N5) relative to the sodium salt alone, although more calculations are needed to quantify this in terms of a pKa difference. This would suggest a more acidic

N(5)-H bond in the presence of TBAB due to its stabilizing effect on the conjugate base.

In addition, TBAB may be further expediting the subsequent cyclization reaction by promoting an “attack-ready” position as in the previous calculation described above.

Figure 26. Computational results in deprotonated system with and without TBAB

92 Atom NBO without TBAB NBO with TBAB

0.245 0.259

-0.472 -0.463

0.413 0.406

-0.641 -0.601

Table 6. Selected NBO changes due to presence of TBAB In addition to evaluating the two hypotheses discussed above, our computations also shed light on the reactivity differences of propargyl hydrazine anion A (Figure 26) in the presence of TBAB. In these studies, a ring-closed version of anion A was computationally constructed to give structure C; however, it optimized into another ring-opened structure

E. The ring-closed intermediate C is 6.8 kJ/mol higher in energy than A, which is less of an endergonic difference than might have been supposed for the formation of a vinyl anion.The allene E that C immediately optimized into is 103 kJ/mol lower in energy than anion A, suggesting that the allene E is highly thermodynamically favored, and that the experimentally observed formation of ring-closed products may indicate a kinetic trap.

When the open anion was cyclized in the presence of TBAB (B to D) the ring-closed vinyl anion was stabilized and did not optimize into the allene.

Figure 27. Computational results in deprotonated system with and without TBAB

Figure 28. Relative thermodynamic stability of reactive intermediates We note that this computational finding is in general agreement with the predictions of

Alabugin (see following section) as it suggests that a [2,3] Wittig rearrangement leading to the allene product is intrinsically preferred over cyclization. Indeed, one of the largest theoretical objections to any proposed mechanism that results in the vinyl anionic intermediate is that it seems implausible that such a presumably strong base would be produced from a mechanism triggered by the reaction of a modestly strong base with the

94 substrate. However, the vinyl anion intermediate in very similar tandem cycloaddition- conjugate addition of 131 to aldehyde catalyzed by TBAF has recently been trapped by

Nagy (Scheme 32).109 An explanation of why intermediate 132 is formed, leading after electrophilic quenching to the 5-endo-dig cycloaddition product 133 rather than a [2,3] sigmatropic shift is needed.

Scheme 32. Tandem cycloaddition/aldehyde addition trapping vinyl anion 3.2.5. Role of the tetraalkylammonium salt

The DFT and Raman spectroscopic evidence suggest that the addition of TBAB induces a conformational change and that this change, perhaps as much if not more so than an ion- pairing stabilization of the N5 anion, might best explain the rate acceleration of the cycloaddition when TBAB is used with modest bases, or by extension when TBAF is used to promote the cyclative addition. These data call for a subtler examination of the possible mechanisms of interaction between tetraalkylammonium salts (TBAB will continue to be examined in particular) and propargyl hydrazine 116a.

Catalytic systems, including organocatalysts, that are used to induce a certain conformation in the substrates that is optimal for acheiving the desired transition state are known as template catalysts.110 A change in the substrate’s conformation is achieved by the template molecule usually through weak electrostatic intermolecular interactions such as hydrogen bonds or dipole-dipole interactions. Template catalysis is a common strategy for supramolecular self-assembly and a concept applied to host-guest design in supramolecular chemistry.111 The concept is exemplified in the use of potassium cation to preorganize

95 ethylene polyethers such as 134 (Scheme 33) into a conformation conducive to the

Williamson ether synthesis of 18-crown-6 ether 135.112

Scheme 33. Cation templated synthesis of potassium 18-crown-6 ether Recent work by Hasegawa and Ema have highlighted the intriguing possibilities for tetraalkylammonium species to catalyze reactions through subtle electrostatic interactions that stabilize the reactants in part through the induction of conformational changes conducive to the transition states as well as possibly stabilizing transitional states and reactive intermediates.113 Hasegawa and Ema developed a series of catalysts in which an

MgII porphyrin is tethered to tetraalkylammonium auxiliaries, employing a synergy of organocatalysis and organometallic catalysis to the solvent-free synthesis of cyclic carbonates from epoxides and carbon dioxide (136, Figure 29). DFT calculations were performed to investigate the role of the tetraalkylammonium in the catalytic cycle. It is well understood that in tetraalkylammonium cations, the positive charge is located not on the more electronegative nitrogen atom but is spread out among the hydrogen atoms of the alkyl substituents. They found that the bromide interacted with these hydrogens

Figure 29. Tetralkyl bromide linker used to preorganize and stabilize reaction complex together as an anion-binding site, and this stabilized the bromide as well as the entire reactant complex. Interestingly, bromide was found to be more nucleophilic in these models than chloride or iodide (nucleophilicity increases from I-

However, just as salient is the observation made by these authors that the quaternary ammonium cation stabilized anionic species that were produced during the catalytic cycle in an “induced fit manner” and were accompanied by conformational changes. This corroborates the hypothesis that tetraalkylammonium cations can act as template catalysts where the mechanism of catalysis, in part, is in the preorganization of the substrate into an optimal conformation leading to the transition state.

97 In light of the subtler role tetraalkylammonium salts can play in preorganizing and stabilizing reactants by coordinating both through the alkyl substituents of the ammonium cation and through the counter anion, a refinement to the cation-π interaction between tetrabutylammonium bromide and propargyl hydrazine 116a is proposed. Based on the relative positioning of the quaternary ammonium and bromide to 116a in its protonated form (complex B, Figure 21), the induced conformational change in 116a may be due to some combination of cation-π and anion-π interactions. An anion-π interaction between an alkyne, particularly a terminal alkyne, is predicted by the same theoretical quadrupole- monopole electrostatic interaction that predicts the cation-π interaction between the alkyne and ammonium. Examples of quadrupole-monopole interactions corresponding to alkyne- cation interactions and anlyne-anion interactions are depicted in Figure 30. It is compatible with both theory and the evidence that an induced conformational change is effected in

116a through a complimentary combination of subtle attractive interactions between the alkyne moiety and the coordinated complex of tetrabutylammonium and bromide.

Figure 30. Representation of cation-π and anion-π interaction with an alkyne’s quadrupole moment 3.3 Discussion

Alabugin found that nucleophilic intramolecular cyclization of alkynes into five-membered rings, designated 5-endo-dig cyclizations under Baldwin’s rules, displayed theoretical features anomalous compared to smaller and larger analogues. Alabugin found that,

98 contrary to Baldwin’s rules114 for nucleophilic cyclizations, exo-dig attack is favored over endo-dig attack by stabilizing orbital interactions via the obtuse, rather than the acute, trajectory of the nucleophile towards the alkyne (Figure 31).115 The calculated transition state energy of 5-endo-dig cyclization is lower than expected, while the incipient bond length is longer than expected, compared to other endo-dig reactions. The proposed explanation is that 5-endo-dig transition states are stabilized through in-plane aromaticity.

Figure 31. Baldwin’s rules modification for 5-endo-dig cycloadditions and aromatic TS Because of the aromatic transition state, Alabugin has proposed calling this kind of nonpericyclic ring closing reaction an aborted anionic [2, 3] sigmatropic shift. The reasoning is that the cyclic anion formed in transition between the hypothetical true [2, 3] sigmatropic shift is unusually stabilized by this in-plane aromaticity so that ring closure represents an energy well between the acyclic anion precursor and the [2, 3] sigmatropic shift product. What would have been a transition state has been ‘flattened down’, energetically speaking, into a ‘caldera’ as Doering put it.116 This theoretical concept helps to explain one aspect out our cycloaddition reaction that has always seemed implausible:

99 that a vinyl anion could be generated as an intermediate from an initial acid-base reaction between a carbamate and a relatively weak base.

The question of why we do not observe the [2, 3] sigmatropic shift product, allene, under any reaction conditions so far, is still far from obvious. Alabugin predicts that exocyclic hyperconjugation with a sigma acceptor will stabilize the vinyl anionic intermediate leading to the 5-endo-dig product. Supporting evidence for the latter prediction is

Hammond’s gem 3, 3-difluoro nitrogen anions that cyclize, purportedly through the heterocyclic anion being stabilized by negative hyperconjugation between the incipient anion and the σ*C-F bond. However, Alabugin’s analysis also predicts that having an endocyclic sigma acceptor such as N or O facilitates the [2, 3] Wittig rearrangement product, the allene, rather than the 5-endo-dig cycloaddition product. This would lead us to expect the allene as the sole or major product, when in fact the allene is not observed to form at all in any of our cycloaddition reactions. On the contrary, it has been shown that anionic cycloaddition catalyzed by TBAF will produce the heterocyclic products with two endocyclic heteroatoms (N-N or O-N) without any endocyclic heteroatom or electron withdrawing group, albeit under gentle reflux as opposed to the quick room temperature cyclization as in 116a where there is an ester at the propargyl position.

100

Figure 32. Comparison of theoretical predictions with experimental results One important point is that the above prediction of a [2, 3] Wittig rearrangement given the endocyclic heteroatoms is supported by Terentis’ calculation (Figure xyz) that showed the anionic propargyl hydrazine 116a finding an energy minimum by rearranging into an allene after being ring-closed. This has not been observed experimentally with any base strong enough to promote reaction by deprotonating 116a and similar compounds (Figure 32).

Heterocycle 117a or related heterocycles are obtained exclusively. Intriguingly, Terentis showed that TBAB stabilizes the ring-closed, vinyl anionic intermediate, leading to the hypothesis that the tetraalkylammonium cation is stabilizing this intermediate, as these ring closure are almost always efficiently catalyzed by TBAF or TBAB and a suitable base.

The congruence of empirical and calculational data seem to support several catalytic roles for a tetraalkylammonium salt through the process of base-promoted 5-endo-dig cycloaddition of propargyl hydrazines and closely related compounds.

101

Figure 33. Mechanistic hypothesis for tetraalkylammonium assisted 5-endo-dig cycloaddition The following three-step mechanism is proposed (Figure 33). First, association of tetrabutylammonium bromide with propargyl hydrazine 112a causes a conformational change in the latter from A to B, plausibly explained by quadrupole-monopole electrostatic interactions between the alkyne functional group and the quaternary ammonium salt. This preorganizes 116a into an ‘attack ready’ conformation for cycliation and may weakly stabilize the anionic intermediate C through ion-pairing, with the ammonium salt still tethered by a weak association with the alkyne. The formation of C, rather than some concerted step leading from B directly to D is indirectly supported by the 26 kJmol-1 calculated stabilization of B in the presence of TBAB, and by the observation that there is a fine sensitivity of the rate to the strength and character of the base (pyrazoline and attenuated phenoxides such as p-fluorophenoxide result in slower TBAB catalyzed cyclization compared to sodium phenoxide with TBAB, see also Table 5, entries 1 & 2) which suggests that deprotonation of a less acidic species such as B to form a reactive

102 intermediate such as C is the rate limiting step. If, on the other hand, concerted cyclization occurs directly from B to an N5-H+ version of D, then the role of the base would simply to shuttle a proton from N5 to the vinyl anion on C2 and this would be unlikely to require stronger bases in order to accelerate the reaction. Mechanistic involvement in the cycloaddition from C to D by electrostatic activation of the alkyne towards nucleophilic attack by the quaternary ammonium salt is ruled out by RS evidence and by DFT calculation. However, the quaternary ammonium salt plausibly stabilizes the known vinyl anionic intermediate D as suggested by DFT results that suggest, in agreement with

Alabugin’s analysis, that there is at least a theoretical preference for a [2, 3] Wittig rearrangement to allene anionic intermediate E that is being overcome by some combination of effects that stabilize D, leading to the experimentally observed cycloaddition.

3.4 Conclusion

The mechanism of intramolecular cyclative addition of nitrogen to unactivated terminal alkyne resulting in 5-endo-dig products in a range of structurally related compounds has been investigated by closely examining the case of a representative propargyl hydrazine, compound 116a, and the role of tetrabutylammonium salts, particularly tetraammonium bromide, in catalyzing the base-mediated cycloaddition. Although 116a and closely related analogues will cyclize with base alone, the reaction rate is greatly accelerated when catalytic TBAB is combined with the base and is very rapid when catalytic TBAF is used alone. The hypothesis that the mechanism of tetrabutylammonium is something other than phase transfer catalysis is supported by the homogeneous reaction rate study, where added

TBAB was unambiguously shown to significantly accelerate the rate of cycloaddition

103 promoted by soluble bases in a variety of solvents. The hypothesis that the mechanism of catalysis might involve an electrostatic interaction between the carbon-carbon triple bond and the quaternary ammonium salt is supported by Raman experiments that established a dose-dependent shift of the alkynyl carbon-carbon stretch frequency of propargyl hydrazine 116a to lower wavenumber in the presence of TBAB that is consistent with a weak intermolecular interaction. DFT calculations also predicted this effect. This effect was found to be general for terminal alkynes in the presence of TBAB. It has gained acceptance as evidence of a cation-π interaction between the permanent quadrupole moment of the alkyne moiety and the monopole charge of the ammonium cation. However, two potentially important subtleties that must be considered are: 1) the tetrabutylammonium cation is a multipoint-charge species that has been shown to coordinate substrates and counterions into preorganized reaction complexes and to stabilize developing intermediary charges throughout the reaction in much the same manner as an enzyme, and 2) it is consistent with the theoretical principles underwriting cation-π hypothesis and with DFT calculations that an anion-π interaction between the bromide anion and the positive region of the alkyne’s permanent quadrupole moment may be also be operant, allowing for a tandem electrostatic interaction between the quaternary ammonium cation and the alkyne. It is worth emphasizing that the cation-π and anion-π effects are not either-or propositions in the overall hypothesis; they may both be complementing one another and resulting in a more complex interaction between the alkyne and the ammonium salt. Raman evidence suggests that a hydrogen bond may be lost to the carbamate proton on N5 of propargyl hydrazine 116a upon addition of TBAB.

Taken together with the DFT calculations that show TBAB to induce a conformational

104 change from a conformation where the N-H is closer to an ester carbonyl oxygen to a more attack ready conformation, this supports hypothesis that TBAB acts as a template catalysts to preorganize 116a for cycloaddition. The ion-pairing hypothesis is not supported by DFT calculations, which indicated that the anionic charge of deprotonated N5 is not greatly reduced in the presence of TBAB, but we cannot yet rule out this effect as a subtle stabilizing factor in the mechanism. DFT calculations do support a stabilizing effect by

TBAB on the ring-closed vinyl anionic intermediate, supporting the calculational analysis of Alabugin, although the preference for the cycloaddition product rather than the [2, 3]

Wittig rearrangement product in 116a is counter to Alabugin’s predictions and this cannot be entirely attributed to stabilization by tetraalkylammonium salts.

3.5 Experimental Section

General procedure for Raman study:

Solutions of each alkyne with TBAB at various concentration ratios were prepared in

DCM unless otherwise stated. For the solutions with TBAB:alkyne molar ratios of 0:1,

0.25:1, 0.5:1, 0.75:1, and 1:1 the concentration of the alkyne was initially 100 μM. For the

5:1 solution, the concentration of the alkyne was initially 20 μM. Each solution was vortexed vigorously before 10 μL of solution was pipetted onto a glass microscope slide.

Raman spectra were collected immediately within the solution droplet before it had dried using an XploRA ONE Raman microscope (Horiba Scientific, Edison, NJ) with 12 mW,

532 nm laser excitation, 10x microscope objective, 1800 g/mm grating, 200 μm slit width,

500 μm confocal hole diameter, and 1-5 min accumulation time per spectrum. All spectra were manually baseline corrected and calibrated against the standard Raman band of silicon (520.7 cm-1).

105 Chapter 4

Teaching Lab Module to Elucidate a Possible Cation-π Interaction in an Intramolecular

Ammonium-Alkyne Cyclization

4.1 Introduction

4.1.1 The cation-π interaction in chemical education

While cation-π interactions have long been known and described in the literature,117 they are included in relatively few introductory or even advanced textbooks118 and laboratory experiments119 of organic chemistry. This is an unfortunate educational omission since cation-π interactions are being used more frequently to explain exciting new phenomena in biology,120 materials science,121 and chemistry.122

An example of cation-π interactions explaining the coordination of cations and arenes is potassium complex 127 (Figure 34) characterized by Mulvey and Sherrington123. The cation-π interaction involving an alkyne as the π-system is encountered much less frequently in the literature, but evidential support is increasing, as demonstrated in the structurally authenticated complexes such as cesium complex 128 characterized by Long124

(see also 3.1.1. Figure 20 for another example). In complex 127, the cation-π interaction explains the coordination between potassium and the arenes. In complex 128, the cation-π

106 interaction explains the coordination between cesium and the terminal alkyne moieties of the N,N',N''-trimethyl-1,4,7-triazacyclononane-chromium complexes.

Figure 34. Cation-π interactions involving arenes and alkynes as π-systems A review of papers published in the Journal of Chemical Education reveals that examples of articles dealing with the teaching of the scientific method were far more numerous in past decades.125 Among the more recent teaching innovations involving the teaching of the scientific method, Giunta has advocated a historical approach.126 While the history of science is replete with excellent teachable examples of the scientific method in practice, there seems to be a lack of emphasis on employing more recent research developments to a foundational teaching of the scientific method. Some approaches to the teaching of the scientific method employ very general observable phenomena, such as Tannenbaum’s simple yet methodical demonstration of boiling water.127 Overway has developed a simple yet interesting module in which students are led to realize the advantage of evidenced based decision making versus intuition through the typically counterintuitive statistical results of a shell game, which is presented as a vehicle for teaching the decision-making aspects of

107 the scientific method.128 There are also practical modules that use chemical phenomena related to everyday products to teach hypothesis formation and experiment to nonmajors, such as Eichler’s fountain reaction of mint candy and diet cola.129 What seem to be lacking in the recent literature are more intermediate-level chemical models that teach or reinforce the basic logical structure of the scientific method to science and pre-health majors.

The process of discovery that accompanied the mechanistic investigation detailed in

Chapter 3 led us to interact with the scientific method and the process of hypothesis formation and experimental design in ways that seemed to lend themselves to the development of an educational project. The rate experiments outlined in Chapter 3 could plausibly be recreated by undergraduates in a simplified form, and it also happened that the organic teaching laboratory in our institution utilized IR spectroscopy as a mainstay in the teaching lab modules. We would therefore have a chance to have students also try to recreate our Raman spectroscopy evidence for the cation-π interaction between propargyl hydrazine 116a and TBAB with IR spectroscopy. The general conception was to introduce students to the scientific method by having them recapitulate our process of hypothesis formation and experimental design, and once having confirmed our results ruling out the phase-transfer catalysis hypothesis of TBAB’s mechanistic role in the reaction, find corroborating evidence for the ammonium-alkyne interaction on and IR spectroscopy version of our Raman experiment. Three benefits of this project are that it allows students to interact with contemporary chemical research, it teaches the scientific method using a more intermediate-level approach than is typically seen, and it allowed students to aquire novel data in support of a hypothetical phenomenon in an area of science that is not yet considered settled.

108 4.1.2 Evidence supporting a hypothesis

Evidence from both spectroscopic techniques and from chemical rate experiments are commonly used to support mechanistic hypotheses. Although many chemists hold to the position that a mechanism can never be proven, there has been some recent disagreement on this point.130 Many scientists, following the philosopher of science Karl Popper, hold the view that science does not advance by proving hypotheses, but rather by designing experiments that would falsify an incorrect hypothesis.131 A hypothesis that survives this falsification process then gains support. It should be noted that the question of whether a mechanism can be proven does not refer to proof in the strict mathematical sense, but whether a mechanistic hypothesis can gain the same degree of support as other common chemical hypotheses such as those, for example, relating to a compound’s structure.

Traditionally, reaction mechanisms were elucidated mainly through reaction rate experiments, and the curved arrow mechanism and the structure of the transition state were considered conjectural despite being compatible with the evidence. More recently, computational methods allow analysis of transition states132 and reaction mechanism pathways133 using high levels of theory. There have also been improvements in spectroscopic technology such as femtosecond spectroscopy134 and other techniques have led some to reevaluate whether mechanistic hypotheses may no longer qualify for this necessarily conjectural status. Rigorous quantitative analysis of reaction mechanisms through rate experiments to deduce the rate law of reactions has been developed as an instructional module by Miller.135

109 4.1.3 The role of spectroscopy

Questions about transition states, intermolecular interactions and conformational changes during the course of a reaction are still investigated using deductive reasoning from the analysis of reaction rate under various changes to the system, but the use of spectroscopic techniques allows the discovery of more direct evidence for mechanistic hypotheses.

Basu,136 citing a precedent,137 argued that a C-H-π interaction could be inferred from NMR spectroscopy showing the perturbations that would be expected as a consequence of such an interaction. As referenced in the discussion about whether a mechanism can ever be proven, spectroscopic techniques are often taken as a more direct observation of a chemical phenomenon than classic falsification experiments involving reaction rates in which the inferred phenomena are either ruled out or not. The question of whether or not this is the optimal attitude, or whether, on the contrary, falsification experiments that deduce mechanisms through hypothesizing about the outcomes of reactions should be seen as on a par with experiments in which mechanisms are deduced by spectroscopic observance of some evidence of an interaction, is beyond the scope of this work. But one of the goals of this educational project is to draw attention to this topic of debate about the practice of science.

110 4.2 Experimental design and results

This experiment is divided into two parts. It begins by introducing the student to a hypothetical argument based on indirect evidence for an ammonium-alkyne interaction. It was observed that propargyl hydrazine 116a (Figure 15) is converted to cyclic product 117a in the presence of KF only when tetrabutylammonium bromide (TBAB) is present. In the introductory lecture, concepts were explained pertaining to two possible competing hypotheses that might account for the catalytic effect of the added ammonium salt. One hypothesis is that the ammonium functions as a phase transfer agent for the insoluble fluoride (Figure 36).

Figure 35. Evidence of ammonium salt having a catalytic role in the cyclization

111

Figure 36. Phase transfer mechanism as a hypothesis of TBAF catalysis The alternative hypothesis is that the ammonium cation interacts with the alkyne, subtly influencing the cyclization mechanism. Students are prompted to think of an experiment that could distinguish between these two hypotheses and falsify one or the other. What experiment can be designed to demonstrate phase transfer catalysis? The answer is to use a soluble base and perform the reaction as a homogeneous solution, with and without added

TBAB, to see whether TBAB enhances the rate of reaction compared to the soluble base alone (Figure 37). If the addition of TBAB to a homogeneous reaction mixture increases its rate, then the phase-transfer catalysis hypothesis would be revealed as an insufficient explanation.

112

Figure 37. Presentation of competing hypotheses and falsifiable predictions The second part of the lab is based on the recently published spectroscopic evidence for this ammonium-alkyne interaction. In order to obtain direct evidence of the ammonium- alkyne interaction, we recently reported the use of Raman spectroscopy to observe the effect of TBAB on various terminal alkynes, including the propargyl hydrazine. The observed shift in frequency of the carbon-carbon triple bond stretching vibrational mode toward lower wavenumber was considered sufficient direct evidential support for a cation-

π effect between the ammonium cation and the alkyne. The mechanistic analysis discussed in Chapter 3 presents a case for the involvement of the ammonium salt in the mechanism that is supported by empirical evidence and DFT calculations. In order to allow students the opportunity to gather novel data in the teaching lab, we adapted this TBAB titration experiment for IR spectroscopy, using the spectroscopic equipment available in the teaching labs. Students performed their own TBAB titration by preparing samples of propargyl hydrazine in solution with increasing amounts of TBAB and drop depositing them on salt crystals. They then analyzed the samples using IR spectroscopy to semi- quantitatively observe the effect of the ammonium ion on the alkyne stretching frequency.

Students analyzed the data to determine whether evidence supports a cation-π interaction.

As discussed earlier, our group reported that an unusual TBAF-mediated intramolecular cycloaddition of a carbamate nitrogen to an unconjugated alkyne formed substituted

113 azaproline derivatives. Subsequent studies led to the hypothesis that the mechanism involves a cation-π interaction between the ammonium ion (with the positive charge spread between the methylene groups directly attached to the electronegative nitrogen) and the C-

C triple bond of the substrate. The ammonium cation itself was shown to have a catalytic effect, and, critically, experiments revealed that this was due to something other than phase- transfer catalysis. Raman spectroscopy showed that in the presence of ammonium cation, a dose-dependent shift occurred in the wavelength of the carbon-carbon triple bond stretching mode consistent with a cation-π interaction. Critically, other vibrational modes showed no shift in wavenumber in the presence of TBAB.

The propargyl hydrazine 116a was chosen as the substrate for this experiment because it was the compound used to extensively study this mode of organocatalysis in our published work. The product, dehydroazaproline derivative 117a, is representative of a class of compounds that may have interesting applications in peptide design, specifically as beta- turn inducers in protein folding.138 Therefore, the reaction is of interest beyond its usefulness in studying the organocatalysis and its possible role in this unusual reaction. The goals of this lab were to guide students through the process of forming plausible hypotheses and designing experiments to choose between them, while allowing them to interact with contemporary research. Unlike most undergraduate organic chemistry experiments, which are drawn from long-established research and classic named reactions, this lab recapitulates the key experiments used in recent research. Students are given the chance to work in small groups in order to recreate the team dynamic of a research group, and they are encouraged to collaborate on hypothetical reasoning and then to come up with a team strategy to carry out the experiments. The two most plausible hypotheses for the organocatalysis reaction

114 under investigation are presented to the students, along with the scientific method rationale for the experiments that will decide between the rival hypotheses using the criterion of falsification. However, students are reminded that these may not exhaust the possibilities.

Students participating in this experiment had completed Organic Chemistry 1 and were enrolled in this lab course with a co-requisite of Organic Chemistry 2.

4.3 Discussion

4.3.1 Outcome of teaching module

This experiment was carried out by 232 second or third-year undergraduates over two semesters as the capstone learning experience of the organic chemistry lab course. The experiment, including the introductory lecture and the write-up, was completed over four sessions of two-and-a-half hours each. The class was divided into research teams to more fully replicate the experience of peer interactions in graduate research. Each team of four students was instructed to prepare a division of labor plan such that all members had a role in every stage of the experiments while working together in a coordinated manner. Each team member kept an individual lab notebook for the experiment, and write-ups were completed individually.

In the first part of the experiment, students tested the phase-transfer hypothesis for the role of TBAB in the catalyzed reaction. Each team ran the control reaction combining sodium phenoxide with propargyl hydrazine 116a in ethyl acetate, as well as the same reaction with the addition of TBAB. Each reaction was monitored individually with a sample taken at regular 5-minute intervals for TLC analysis. The TLC plates were developed using

KMnO4, and the progression of the spot corresponding to 116a to the less polar spot known

(from reliable research data shared with the students) to be the azaproline derivative

115 product 116a, was evaluated. By visually comparing the approximate rate at which the spot for compound 116a faded away and the spot for compound 117a appeared and grew stronger, the students were asked to make a qualitative judgement about whether the reaction with added TBAB proceeded to completion faster than the control reaction.

In the second part of the experiment, students carried out a reaction using the same reactants plus TBAB with acetonitrile as a solvent. The reaction was timed and TLC data were collected and evaluated in the same way as before. Students were reminded of the two hypotheses being considered, and were encouraged to discuss the results and their implications with group members throughout the lab period.

Students were then prompted to think about how IR spectroscopy can be used to more directly observe an ammonium interaction with the reactant. Perhaps there will be a change on the vibrational frequency of the C-C triple bond? Such changes should be detectable by

IR spectroscopy, taking advantage of the fact that the signal appears in a relatively uncluttered region of the spectrum. To obtain this direct observation, students prepared of a series of samples of propargyl hydrazine 116a in ethyl acetate and dichloromethane

(DCM) (to improve TBAB solubility) with increasing amounts of TBAB. The technique of drop deposition was used: each sample was deposited onto a NaCl crystal disc, dried, and measured using transmission IR spectroscopy. The alkynyl carbon- carbon stretch was identified at 2120 cm-1, and the spectra were expanded and overlaid to highlight that peak.

For each research group’s titration, all spectra were plotted on a single graph with each labeled according to the molar equivalent of TBAB added (see Figure 38 for a representative student example). After obtaining and overlaying the spectra of their titration samples, students printed a copy for each group member and annotated and

116 examined the peaks for evidence of a dose-dependent shift in wavenumber with added

TBAB. Discussion of the implications for the competing hypotheses, this time emphasizing possible confirmatory evidence for the cation-π hypothesis, was encouraged between group members.

Qualitative Rate Experiment Additive Solvent Ranking 1 TBAB EtOAc 2nd No 2 EtOAc 3rd TBAB 3 TBAB MeCN 1st Table 7. Student Rate Experiments

A rate increase was generally qualitatively inferred by TLC from the increasing product intensity and decreasing starting material intensity when TBAB was added to the reaction in ethyl acetate, compared to the uncatalyzed reaction in ethyl acetate. The TBAB catalyzed reaction was fastest in acetonitrile in acetonitrile, showing that the rate is solvent dependent. The order of reaction rates, from fastest to slowest, is TBAB catalyzed in acetonitrile, followed by TBAB catalyzed in ethyl acetate, and finally uncatalyzed in ethyl acetate. (Table 6). The observed relative reaction rates are in agreement with the published data (obtained by more precise NMR experiments) used to establish the catalytic role of the ammonium cation in homogeneous solutions and the importance of the solvent in determining the reaction rate. However, it is important to note that due to unfamiliarity with the technique of TLC spotting and TLC plate development, not all students were able to produce sufficiently high quality TLC plates to see the difference in rate. The results from the IR titration were uniform across the student groups and showed unambiguously that, in the presence of increasing TBAB, the carbon-carbon triple bond stretching frequency changed noticeably.

117

Figure 38. Representative Student IR Titration of 116a with TBAB Students were asked to consider the original competing hypotheses: catalysis by phase- transfer versus a cation-π interaction. The phase-transfer catalysis hypothesis was falsified by the rate experiments. However, students were guided towards realizing that the falsification of one possibility does not automatically validate the alternative, since there may be other possibilities that were not considered. This is why the IR titration was needed to provide supporting evidence for the cation-π hypothesis. The IR titration study (Figure

38) showed an unambiguous dose dependent shift of the alkynyl carbon-carbon stretching frequency with added TBAB, an observation best explained by the cation-π interaction between the ammonium cation and the alkyne. This result was reproduced in all students’ data. Furthermore, the direction of the shift was uniformly towards lower wavenumber with increasing TBAB, just as one would expect given the nature of the hypothetical electrostatic interaction slightly decreasing the bond order of the alkyne. This evidence was taken to confirm the predictions of the ammonium-alkyne cation-pi interaction hypothesis.

118 For the final portion of the research experience, each student was asked to compose a conclusion in essay format. In this conclusion, they recounted the background information and initial hypotheses given in the introductory lecture. They were expected to take into consideration the potential falsification of the phase-transfer hypothesis by the reaction rate experiments, as well as the potential confirmation of the predictions of the cation-π hypothesis by the IR titration. They were asked to draw some provisional conclusions, keeping in mind that the two hypotheses that were examined may not exhaust the possible explanations. Finally, they were asked to think critically and to come up with a further hypothesis about the reaction and an experiment to test their hypothesis. A qualitative review of student write-ups suggested that students generally comprehended the concepts introduced in this experiment.

4.3.2 Evaluation of pedagogical goals

The main pedagogical goal of this project was the teaching of hypothesis-based experimental design. Other goals of the project were to introduce undergraduates to contemporary research problems in organic chemistry, to introduce students to the experimental investigation of reaction mechanisms and molecular interactions, and to demonstrate the usefulness of spectroscopic techniques not just as methods of characterizing compounds, but as means of collecting decisive evidence in mechanistic investigations. Secondary goals were to train students to work in small groups and learn what it is like to apply a team-based approach to research.

Students were introduced to the concept of hypothesis-based experimental design by the example problem introduced in this lab module. An emphasis was placed on the importance of generating two or more incompatible hypothetical explanations for the phenomenon in

119 question. In this case, the two hypotheses offered as possible explanations for the role of

TBAB in catalyzing the reaction are both plausible explanations of the phenomenon (the increased reaction rate, i.e. catalysis). They are both compatible with the initial data, and it is not immediately clear how one or the other of them (or both) could be ruled out by observation. Strictly speaking, they are not incompatible explanations since each hypothesis posits a type of mechanism of catalysis, and, in the experimental reaction, it is always possible that more than one mechanism is at work. In this project, that possibility poses no great difficulty, because the two competing hypotheses may be reformulated into one hypothesis: the catalytic effect of TBAB is merely due to a phase transfer mechanism.

Since this investigation is concerned with seeing what evidence is compatible with a novel intermolecular interaction posited as an alternative mechanism of organocatalysis, it suffices to find conditions in which this modified hypothesis could be falsified (which are the same as the conditions used in this module). Once this hypothesis has been falsified (in other words, once a catalytic effect of TBAB has been observed in a system in which phase transfer catalysis cannot operate) this is enough to answer the objection that the evidence for our hypothesized interaction may be fully explained by phase transfer catalysis, leaving our proposed explanation redundant and thus our hypothesis lacking corroboration. We rationally expect that in heterogeneous systems, TBAB is acting both as a phase transfer catalyst and as an organocatalyst along the lines of our proposed explanation. In such a system, it is possible that both mechanisms could have a combined effect on the rate. But it is necessary to design experiments that might tease this possibility out. We simply need to establish evidence of the non-phase transfer mechanism. On the other hand, if it had been discovered instead that all catalysis ceases in the homogeneous system, then this

120 would be fatal to our proposed alternative mechanism. It would not falsify our proposed intermolecular interaction outright, since the interaction might be present but mechanistically impotent, but the interaction would be relegated to mere speculation unless and until some other observable consequence could be identified that would be best explained as an effect of the interaction.

The goal of exposing students to contemporary research was achieved by the design of the module, but it remains to be shown that this approach leads to better learning outcomes.

Further study is needed to evaluate this pedagogical hypothesis. Quantitative longer-term student outcomes might be evaluated between curricula that incorporate contemporary synthetic organic chemistry research and those that include only teaching modules based only on older, well-understood research problems.

Evaluation of the effectiveness of teaching mechanistic investigation to undergraduates was assessed informally through qualitative appraisal of the student experimental write- ups. The general impression was that, although the logic of the rate experiments chosen to decide between the two hypotheses was adequately recreated in the student write-ups, the understanding of this logic as demonstrated by student proposals for further experiments to elucidate the mechanism were highly derivative and exhibited a shallow or complete lack of understanding of the logic of experimentation in most cases. It seems reasonable to conclude that a basic understanding of the logic of hypothesis-based experimental design will only be instilled at the undergraduate level through repeated exposure to the subject across multiple laboratory and lecture modules.

The secondary goal of conveying to students the utility of spectroscopic techniques for mechanistic elucidation as well as for molecular characterization was aided by serendipity

121 in this project since we had a unique opportunity to have students gather novel infrared spectroscopic data that might either corroborate or challenge the published Raman spectroscopic evidence in support of the hypothesis of the cation-π interaction (see chapter

3). IR spectroscopic evidence had been found to corroborate the Raman findings prior to the development of this teaching model, but those data were not published along with the

Raman data. With the publication of this lab module, novel IR data that corroborate the cation-π hypothesis were gathered by undergraduates while taking this lab module. Such an opportunity to have students learn about the research process by directly participating in it is usually reserved for just a handful of advanced undergraduates engaging in directed independent research with primary investigators as an extracurricular project. While this aspect of the lab module enhanced the experience for the students involved, it was, again, the result of a serendipitous opportunity and cannot be recreated generally in other lab modules, or even in future repetitions of this module, now that the data have been published.

Finally, the team building aspect that was another secondary pedagogical goal worked quite successfully in the execution of this lab module. It was made clear to students in the introductory lecture that they were to think of themselves as members of research teams working under the supervision of a primary investigator. Once they had been formed into small teams, they were asked to come up with their own division of labor to carry out all of the experimental steps in a coordinated fashion. Without exception, all student teams responded well to this challenge, and all teams were able to accomplish their tasks in the time allotted.

122 4.4. Conclusion

This lab module adapted a recently published investigation of a novel interaction into an exercise in applying the scientific method to contemporary research. The experiment showed how hypothetical reasoning leads to experimental design. Students found indirect and direct evidence of an intermolecular cation-pi interaction. By interacting with contemporary research and thinking critically about hypothesis formulation and the various ways of obtaining scientific evidence, students had an opportunity to interactively explore the process of scientific discovery.

123 Appendixes

124 Appendix A: Selected Spectra

1 H NMR (CDCl3, 400 MHz)

TMS

52a

125

13 C NMR (CDCl3, 100 MHz)

52a

CDCl3

126

1 H NMR (CDCl3, 400 MHz)

52c TMS

127

13 C NMR (CDCl3, 100 MHz)

52c

CDCl3

128

1 H NMR (CDCl3, 400 MHz)

129

13 C NMR (CDCl3, 100 MHz)

53a

CDCl3

130

1 H NMR (CDCl3, 400 MHz)

TMS

53b

CDCl3

131

13 C NMR (CDCl3, 100 MHz)

53b

CDCl3

132

1 H NMR (CDCl3, 400 MHz)

TMS

54a

10a

CDCl3

133

13 C NMR (CDCl3, 100 MHz)

54a

CDCl3

134

1 H NMR (CDCl3, 400 MHz)

TMS

54b

10b

CDCl3

135

13 C NMR (CDCl3, 100 MHz)

54b

CDCl3

136

1 H NMR (CDCl3, 400 MHz)

TMS

CDCl3

137

13 C NMR (CDCl3, 100 MHz)

CDCl3

138

1 H NMR (CDCl3, 400 MHz)

TMS

CDCl3

139

13 C NMR (CDCl3, 100 MHz)

CDCl3

140

1 2

H NMR (CDCl3, 400 MHz) -

1 -

TMS 0

6.10 1.001.00 3.47 3.46 2.97 2.97 2.97 3.46 3.47 1.001.00 6.10

0.97 0.98 0.97

2.95 Chemical Shift (ppm) Shift Chemical

CDCl3

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

141

- 13 C NMR (CDCl3, 100 MHz)

80 CDCl3

100

120 Chemical Shift (ppm) Shift Chemical

140

160

180

200

220

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

142

2 -

1 H NMR (CDCl3, 400 MHz)

1 -

7.73 1.78 3.35 3.52 4.32 3.14 0.64 0.99 0.94 1.88 5.91 0.86

0.99

1.03

2.86

Chemical Shift (ppm) Shift Chemical

CDCl3

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

143

20 -

13 C NMR (CDCl3, 100 MHz)

CDCl3 80

100

120

(ppm) Shift Chemical

140

160

180

200

220

0.05 0.10 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

- -

Intensity Normalized

144

2 -

1 H NMR (CDCl3, 400 MHz)

1 -

0 8.16

1.87

1.00 2.09 6.53 1.17 1.01 1.01 1.33 4.15 4.28 4.15 1.33 1.01 1.01 1.17 6.53 2.09 1.00 5

3.13 Chemical Shift (ppm) Shift Chemical

7 CDCl3

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

145

20 -

13 C NMR (CDCl3, 100 MHz)

CDCl3 80

100

120 Chemical Shift (ppm) Shift Chemical

140

160

180

200

220

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Normalized Intensity Normalized

146

2 -

1 1 - H NMR (CDCl3, 400 MHz)

TMS 0

4 3.94 7.23 1.00 0.68 1.20 3.02 3.87 3.05 14.61 3.05 3.87 3.02 1.20 0.68 1.00 7.23 3.94

1.59

6 Chemical Shift (ppm) Shift Chemical

7 CDCl3

2.00 1.25 2.592.80 2.561.38 2.592.80 1.25 2.00

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

147

20 -

0 13 C NMR (CDCl3, 100 MHz)

80 CDCl3

100

120 (ppm) Shift Chemical

140

160

180

200

220

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

148

2 -

- 1 H NMR (CDCl3, 400 MHz)

0 9.56

2.16

2.99

3.20

2 3.42

2.22

3 1.04

0.95

1.15 4

7.56

3.92

0.90

Chemical Shift (ppm) Shift Chemical

7 CDCl3 4.60 0.95 2.10 2.00

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

149

20 -

13 C NMR (CDCl3, 100 MHz)

80 CDCl3

100

120 Chemical Shift (ppm) Shift Chemical

140

160

180

200

220

0.05 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 -

Intensity Normalized

150

2 -

1 - H NMR (CDCl3, 400 MHz)

0 9.37

1.93

2 3.28 2.53 3.28

1.00

1.12

2.68

4 15.53

0.821.35

Chemical Shift (ppm) Shift Chemical

CDCl3 6.23 1.88 1.88

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

151

20 -

13 C NMR (CDCl3, 100 MHz)

CDCl3

100

120 (ppm) Shift Chemical

140

160

180

200

220

0.05 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

152

2 -

1 - TMS 1 H NMR (CDCl3, 400 MHz)

0.37 0.28 4.35 5.82 0.54 0.65 1.03 6.67 2.25 6.67 1.03 0.65 0.54 5.82 4.35 0.28 0.37

CDCl3 Chemical Shift (ppm) Shift Chemical

0.95 1.00 2.13 2.63 1.97 2.63 2.13 1.00 0.95

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

153

2 -

1 1 - H NMR (CDCl3, 400 MHz)

0 9.19

3.40 5.78 1.17 1.00 2.74 3.27 2.05 3.27 2.74 1.00 1.17 5.78 3.40

1.03 2.79 1.03 Chemical Shift (ppm) Shift Chemical

7 CDCl3

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

154

20 -

13 0 C NMR (CDCl3, 100 MHz)

CDCl3 80

100

120 Chemical Shift (ppm) Shift Chemical

140

160

180

200

220

0.05 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Intensity Normalized

155

gCOSY (CDCl3, 500 MHz)

-0.5

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5 F1 Chemical Shift Shift (ppm) Chemical F1

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8 7 6 5 4 3 2 1 0 F2 Chemical Shift (ppm)

156

gHMBC (CDCl3, 500 MHz)

8 7 6 5 4 3 2 1 0 F2 Chemical Shift (ppm)

157

gHSQC (CDCl3, 500 MHz)

100

110 Shift (ppm) Chemical F1

120

130

140

150

160

170

180

8 7 6 5 4 3 2 1 0 F2 Chemical Shift (ppm)

158

gCOSY (CDCl3, 500 MHz)

-1.0

-0.5

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5 Shift (ppm) Chemical F1

5.0

5.5

6.0

6.5

7.0

7.5

8.0 8 7 6 5 4 3 2 1 0 -1 F2 Chemical Shift (ppm)

159

gHMBC (CDCl3, 500 MHz)

-10

100

110

120

F1 Chemical Shift Shift (ppm) Chemical F1 130

140

150

160

170

180

190

200

210

220

8 7 6 5 4 3 2 1 0 -1 F2 Chemical Shift (ppm)

160

gHSQC (CDCl3, 500 MHz)

-8

88 Shift (ppm) Chemical F1 96

104

112

120

128

136

144

152

8 7 6 5 4 3 2 1 0 -1 F2 Chemical Shift (ppm)

161 Appendix B: Raman data (Table 5)

Entry 1. Raman titration of 116a with added TBAB. The inset peaks correspond to the alkyne CC stretch and the arrow indicates the trend in peak change accompanying increasing TBAB:116a molar ratios: 0:1 (pure 116a), 0.25:1, 0.5:1, 1:1, and 5:1 (red curve).

162 2125 cm-1 pure 3

2114 cm-1

3 + 5 eq TBAB Raman Intensity Raman