The 1H NMR Spectroscopic Effect of Steric Compression Is Found in [3.3.1]Oxa- and Azabicycles and Their Analogues

The through-space 1H NMR effect of steric compression by the lone-pair electrons of O- and N-atoms is shown in synthetic [3.3.1]oxa- and azabicycles. The electrons of the compressed proton bond are pushed away by the repulsive force generated by the lone-pair electrons of the heteroatom. There is a corresponding significant increase in the chemical shift of the compressed proton. The intensity of this deshielding effect is related to the proximity and overlap of the lone-pair or compressing atom. The steric compression decreases when the lone-pair electrons of the heteroatom and the compressed proton are not directly overlapped, for example, in [4.3.1]- and [3.2.1]azabicycles. Steric compression is also caused by a proton, deuterium, or an ethyl group close in space to the compressed proton. The protonated [3.3.1]azabicycle adopts a true-boat/true-chair conformation in its crystal lattice, but in solution the conformation is true-chair/true-chair.


INTRODUCTION
The intramolecular through-space interaction that causes steric compression is detectable in 1 H NMR studies. The chemical shifts of the compressed proton and its neighbor on the methylene group shift significantly. 1 Some inflexible C-skeleta (Figure 1), for example, half-cage cyclopentyl (1), norbornenes (2 and 3), and imino [14]annulene (4), have been designed and synthesized to demonstrate elegantly that the compressed proton must be close in space to the source of the repulsive force, for example, H, OH, 1,2 ether oxygen, 3 alkene π-cloud, 4−6 and NH. 7 For two protons attached to the same carbon of a cyclohexane ring in a chair conformation, it is known that the chemical shift of the equatorial proton (∼1.6 ppm) is larger than that of its geminal axial proton (1.1 ppm) by Δδ ∼ 0.5 ppm (at −103°C in CS 2 , 60 MHz), 8 and this is the result of the magnetic anisotropic effect. 9 For 4-H a and 4-H e of cyclohexanone, the difference between their chemical shifts (0.24 ppm, at <−185°C in a 5:1 CHClF 2 /CHCl 2 F mixture, 251 MHz) is relatively small. 10 C 19 -Norditerpenoid alkaloid methyllycaconitine (5, MLA) is one of the most potent competitive antagonists of α7-nicotinic acetylcholine receptors (α7-nAChRs) with a highly selective targeting of the snake venom toxin α-bungarotoxin (α-BgTx) binding sites. 11 In terms of structure, norditerpenoid alkaloids are hexacyclic with bridged structures leading to well-defined conformations. The synthesis of cyclic analogues mimicking MLA (5) 12−16 has been reported where the C2 axial and equatorial methylene group protons show a significant difference (∼1 ppm) in their chemical shifts. 17−22 In this study, the large 1 H NMR separations of the two signals on the methylene groups of different synthetic bridged [3.3.1]oxa-and azabicycles are reported. In order to explain this observation, comprehensive 1D/2D NMR spectroscopy and single-crystal X-ray analyses have been undertaken, which demonstrate that the effect of steric compression is found in these [3.3.1]bicycles and compared with the effect in analogues of various ring sizes.  13 C, heteronuclear single-quantum coherence (HSQC), heteronuclear multiple bond correlation (HMBC), correlation spectroscopy (COSY), and nuclear Overhauser enhancement spectroscopy (NOESY) were used to assign the product (8). 12−22 One of the protons attached to C7 is significantly deshielded and resonates at 2.86 ppm; the geminal methylene proton resonates at 1.53 ppm. The chemical shifts of these 7-H are separated by 1.33 ppm. This deshielding must be through space, as there are no significantly electronegative functional groups nearby.
The tertiary amine nitrogen is assumed to be responsible for the observed shift as it should be close to 7-H a , if the [3.3.1]azabicycle (8) adopts a chair/chair conformation. The deshielding was observed for 7-H a . This is an example of steric compression, where the lone-pair electrons of the N-atom generate a repulsive force pushing the electron cloud surrounding 7-H a away and thus decreasing its electron density, leading to 7-H a resonating at a low field. This explanation holds if the [3.3.1]azabicycle (8) adopts a chair/ chair conformation with the N-ethyl group in the equatorial position and therefore the N-atom lone-pair electrons are close to 7-H a . Due to the intramolecular hindrance in the "half-cage" structure, it is certainly possible that the AE-[3.3.1]bicycle (8) can adopt a boat/chair or a chair/boat conformation rather than an "obvious" chair/chair conformation. NOESY data ( Figure 2) of the [3.3.1]azabicycle (8) were obtained. 2-H a is correlated with 4-H a . 2-H e and 4-H e are correlated with 8-H e and 6-H e , respectively. Therefore, the piperidine ring of [3.3.1]azabicycle (8) has adopted a chair conformation.
The coupling pattern of 5-H of the compound (8) is shown in Figure 2, which is displayed as a dq peak ( The coupling patterns of 7-H a and 7-H e are different ( Figure  2), which were used to assign their orientations. 7-H a couples with 7-H e , 6-H a , 8-H a ( 2 J gem = 3 J aa = 12.4 Hz) and 6-H e and 8-H e ( 3 J ae = 6.1 Hz); therefore, it shows as a qt. 7-H e couples with 7-H a ( 2 J gem = 12.4 Hz); 6-H a , 8-H a ( 3 J ae = 6.1 Hz); and 6-  H e and 8-H e ( 3 J ae = 3.0 Hz); therefore, it resonates as a dtt. These assignments confirmed that the 7-H a is deshielded rather than shielded by the lone-pair electrons of the N-atom.
The coupling constants of 6-H a and 8-H a that are caused by 7-H a are equal (12.4 Hz); therefore, the dihedral angles ∠(6-H a )−C6−C7−(7-H a ) and ∠(8-H a )−C8−C7−(7-H a ) are the same or highly similar, indicating this chair conformation of the cyclohexane ring is true (not twisted). 23,24 NMR spectra of azabicycle (8) are also obtained in CDCl 3 , CD 3 OD, and d 6 -DMSO, and the chemical shifts for 7-H a and 7-H e are given in Table S8. In all three solvents, the 7-H a of azabicycle (8) is significantly deshielded; therefore, this effect cannot be attributed to solvent effects. In addition, variable temperature 1 H NMR experiments of azabicycle (8) in d 6 -DMSO were used to investigate the stability of the chair/chair conformation ( Figure S1). It is clear that the chemical shifts of both 7-H a and 7-H e barely change (∼0.2 ppm) when the azabicycle (8) solution was heated, so 7-H a still experiences steric compression; therefore, the molecule adopts a nearly inflexible chair/chair conformation at 25−125°C.
As two methyl groups are introduced at C7, the cyclohexane or piperidine ring flips into a boat conformation. The piperidine ring is in a chair conformation as shown by the NOESY correlations of 2-H a /4-H a , 2-H e /8-H e , and 4-H e /6-H e ( Figure  4). In contrast to the coupling pattern of 5-H in [3.3.1]azabicycle (8, Figure 2), 5-H of 7,7-dimethyl[3.3.1]azabicycle (10) displayed a dq with a large coupling constant ( 3 J x−x = 10.5 Hz) that originates from eclipsed 6-H e (Figure 4), confirming that the cyclohexane ring of 7,7-dimethyl[3.3.1]azabicycle (10) adopts a boat conformation.
To support the 1 H NMR assignment of 7-H a and 7-H e of the azabicycle (8), 7-alkyl substituted [3.3.1]azabicycle (12), (14), and (17) were synthesized (Scheme 3). In these [3.3.1]azabicycles (12), (14), and (17), the bulky 7-alkyl groups will preferentially adopt the equatorial positions and therefore the δ (7-H a ) can be demonstrated unequivocally. The δ (7-H a ) of products (12), (14), and (17)  2,4-Dinitrophenylhydrazine (2,4-DNPH) was used to derivatize these oily ketones (8), (12), and (17) in order to obtain crystalline derivatives (18−20, respectively Scheme 4) for single-crystal X-ray diffraction (SXRD). NMR data of dinitrophenylhydrazone (DNP) derivatives (18−20) also show that each 7-H a resonates at a low field (Table S9), suggesting that both the ketone starting materials (8), (12), and (17) and their DNP derivatives (18−20) adopt the same solution conformations. SXRD data (18−20) were obtained ( Figure 5). The [3.3.1]azabicyclic DNP derivatives (18), (19), (20a), and (20b) adopt true-chair/true-chair conformations with the N-ethyl groups in the equatorial positions and 9-imine hydrazinyl groups adopt E-configurations. 7-iPr and 7-Me groups of DNP derivatives (19), (20a), and (20b) are equatorial. These conclusions, based on the SXRD data, are consistent with the NMR studies of the synthetic [3.3.1]azabicycles (8), (12), and (17−20). The single crystals of 7-Me derivatives (20a) and (20b) are packed in the same unit cell; they are enantiomers.  (Figure 6), all the   Table S10), and the cycloheptane ring adopts a chair conformation. 24 Axial and equatorial are suitable for describing the orientations of protons attached to a sixmembered ring, for example, cyclohexane in a true-chair conformation. For cyclopentane and cycloheptane rings, protons attached to the ring are better described as pseudoaxial (a′) and pseudo-axial (e′). 25 In this study, a (a′) and a (e′) are preferred rather than exo and endo for retaining consistency with labeling used in the monoring system. Δδ 6 (27) were converted into 2,4-DNP derivatives in order to produce crystalline products (24) and (28), and crystals of [4.3.1]azabicyclic derivative (24) suitable for X-ray analysis were obtained. Twin-packed crystal structures in one unit cell in this crystal are stereoisomers (24a) and (24b), and these stereoisomers are extracted from the original SXRD data and shown separately, as shown in Figure 7. Chair conformations of piperidine rings with equatorial N-ethyls are displayed in both isomers with hydrazone imines in the E-configuration, and these two stereoisomers (24a) and (24b) can be distinguished as 1R-and 1S-esters. Unlike the twin-packed crystal structures of enantiomers 7-Me [3.3.1]azabicyclic DNP derivative (20a) and (20b), the two crystal stereoisomers (24a) and (24b) are not mirror images on the basis of comparison between them in different view angles ( Figure 8). A cycloheptane ring is more flexible and its conformations are more various than that of a cyclohexane ring. 26 The NMR spectroscopic study of the [4.3.1]azabicycles (23 and 24, Table S10) showed that the Δδ 3-H is slightly larger than Δδ 4-H , suggesting that 3-H a′ is experiencing more steric compression than 4-H a′ , and thus, it may, on average, sit closer to the N-atom. This theoretically preferred conformation of    Computer projections of the cycloheptane boat and chair conformations were reported by Bocian et al. 27 In their paper, they especially drew attention to the eclipsed hydrogens at the "stern", the left side of the projection (Figure 9) of the boat, and the chair conformations. In the twist-chair conformation, these previously eclipsed hydrogen atoms are now shown on the basis of SXRD data and NMR analysis to be staggered.

No Steric Compression in a Mono-Mannich Product.
To prove that 7-H a of the synthetic [3.3.1]azabicycles is being sterically compressed by the lone-pair electrons of the N-atoms, a mono-Mannich reaction was designed and carried out to give the monocyclic product (29). β-Keto ester (6) was treated with 0.9 equiv. formaldehyde and 0.9 equiv. ethyl amine, and the reaction was heated at 40°C, rather than under reflux, for 3 h giving the target product (29) (Scheme 6). In the mono-Mannich product (29), there is no piperidine ring; thus, the lone-pair electrons of the N-atom are away from 5-H that correlates to 7-H of the double-Mannich products (8).
The 13 C signal assignments of the mono-Mannich product (29) are assigned (Table S11) compared with those of β-keto ester (6, keto tautomer). Compared with Δ 7-H (1.33 ppm) of the double-Mannich product (8), Δ 5-H (0.09 ppm) of the mono-Mannich product (29) is significantly smaller, demonstrating that a significant 1 H NMR steric compression of 7-H a of [3.3.1]azabicycles requires the N-atom to be close in space to 7-H a .
2.4. Reducing 9-Ketone. An alternative explanation for the observed chemical shifts may possibly be attributed to the 9-ketone group of [3.3.1]azabicycle (8) displaying a long-range anisotropic effect on axial or equatorial protons. To investigate this, the ketone (8) was treated with lithium aluminum hydride to reduce both the 1-ester and the 9-ketone functional groups affording a diol (30, Scheme 7).
In CDCl 3 , CD 3 OD, and d 6 -DMSO, 9-ketone of azabicycle (8) has no obvious magnetic effect on 7-H a or 7-H e as chemical shifts of 7-H a and 7-H e of the diol (30) remain at similar values compared to those of 7-H a and 7-H e of ketone (8) when 9-ketone is reduced (Table S12). This is also    (Table S13), which means the NH(D) is able to provide steric compression acting on 7-H a . Therefore, the conformation of this salt (31) is determined to be true-chair/true-chair as significant compression is displayed. The compression is caused by the electrons in the new bond of NH/D, as there are no lone-pair electrons available. SXRD data of the chloride salt (32) were determined, which shows that this salt (32) adopts a trueboat/true-chair crystal conformation with the N-ethyl in the equatorial position ( Figure 12).
NMR spectroscopic analyses of the crystalline chloride salt (32) in d 4 -acetic acid gave similar results to those of the acetate salt (31) in d 4 -acetic acid. NMR data of the chloride salt (32) show a significant Δδ 7-H = 0.87 ppm (Table S13), suggesting that the solution conformation of this chloride salt (32) in d 4 -acetic acid is true-chair/true-chair. Therefore, the crystal conformation of the synthetic [3.3.1]azabicyclic chloride salt (32) is different from its solution conformation.
When the crystalline ketone chloride salt (32) was dissolved in D 2 O (or wet solvents, e.g., CD 3 OD and d 6 -DMSO), the ketal (hydrate) salt (33) was obtained. To determine the conformation of this salt (33), the 1 H signal of 5-H was employed ( Figure 12). On the basis of the shape (even though broad) of this 5-H, it does not contain a large coupling constant such as 3 J x−x (10.5 Hz) of the 5-H of 7,7-dimethyl   The key NOESY data of this methylated product (34) are given in Figure 13. The 2-H a of the compound (34) has a NOESY correlation with 4-H a , and 2-H a is also NOESY correlated with 8-H e , so the piperidine ring adopts a boat conformation. The 7-H a is close to the 2-H a in the space determined by NOESY correlation 2-H a /7-H a , thus the cyclohexane ring is in a chair conformation.
The N-2 has NOESY correlations with both 2-H a and 2-H e , but the N−Me only show NOESY correlation with the 2-H e ; therefore, the N−Me is in the flagpole position and the N−Et is determined to be in the bowsprit position.
The 7-H a experiences the effect of steric compression presented by δ (7-H a ) = 2.84 ppm > δ (7-H e ) = 1.87 ppm, Δδ 7-H = 0.97 ppm. It is notable that the steric compression acting on the 7-H a is caused by the methyl group of the N−Et rather than the lone-pair electrons of the N-atom, as the lonepair electrons are no longer available.
If the piperidine ring adopts a true-boat conformation, the methyl group of the N−Et is far away from 7-H a in space; therefore, the boat-like piperidine ring has to be monoflattened (only the N-atom is flattened rather than both the Natom and the C9 are flattened) allowing the methyl group of the N−Et to be close to the 7-H a showing a significant steric effect on the 7-H a .
Both 6-H a and 8-H a of methylated [3.3.1]azabicycle (34) ( Figure 13) contribute equal coupling constants of 3 J aa = 13. 8 Hz to the 7-H a , which suggests that the cyclohexane ring adopts a true-chair conformation. The value of the 2 J gem between 7-H a and 7-H e of the methylated derivative (34) is 16.1 Hz, which is significantly larger than that of the 2 J gem (12.4 Hz) of the 7-H a of azabicycle (8). This suggests that the ∠(7-H a )−C7−(7-H e ) of the methylated derivative (34) becomes smaller than that of the azabicycle (8), as 7-H a experiences a strong compression through space changing the geminal bond angle of ∠(7-H a )−C7−(7-H e ).

Synthesis and Analysis of [3.3.1]Oxabicycle.
To add further data about the effect of steric compression on 1 H NMR signals, a [3.3.1]oxabicyclic tetrahydropyranyl ether (36) was designed and then synthesized by intramolecular dehydration (Scheme 10). 28 The product was recrystallized from EtOAc (∼14 h). SXRD data of this

CONCLUSIONS
A through-space 1 H NMR effect of steric compression displayed in [3.3.1]azabicycles is demonstrated and fully discussed. It is caused by the lone-pair electrons of the Natom generating an intramolecular repulsive force acting on 7-H a leading to this proton being significantly deshielded. By comprehensive conformational analysis on these bicyclic compounds and their analogues via NOESY, coupling pattern analysis of key 1 H NMR signals, and SXRD the conformation of the bicycles and the configuration of protons of different methylene groups that experience steric compression are unambiguously assigned. The intensity of this compression is also proven to be related to the distance between the N-atom,   Table 1 and shown compared to the literature data in Figure  15.
Half-cage cyclopentyl (1), norbornenes (2 and 3), and imino [14]annulene (4) elegantly demonstrate that the compressed proton must be close in space to the source of the repulsive force, for example, H, OH, ether oxygen, alkene πcloud, and the only literature example of a secondary amine. Winstein and colleagues reported (in 1965) a new kind of steric compression on one proton of a methylene pair when the other proton is strongly compressed, for example, by an oxygen functional group. This was found together with unusually large deshielding effects in their half-cage or endo,endo-fused skeleta. 1,2 Such rigid geometries and enormous H−H or H−O steric oppositions are ideally suited for the study of effects of steric compression on chemical shifts, where the inside protons are strongly deshielded. Cava and Scheel (in 1967) concluded that the ether oxygen bridge exerts considerable shielding and deshielding effects on the methylene bridge protons, which are separated from each other by Δδ = 1.76 ppm, where such a large difference in the chemical shifts of the methylene bridge protons of a norbornene or norbornane was then unprecedented. 3 Marchand and Rose (in 1968) reported that of particular interest is the effect of the alkene electron π-cloud causing the unusually large value of Δδ = 1.49 ppm between the bridge norbornene protons. 4 A comparable value had been noted only once before in the literature by Cava and Scheel.
The expected aromatic character was found in syn-1,6-imino-8,13-methano [14]annulene (4), first synthesized by Vogel and colleagues. 7 This is the first, and outside of the [3.3.1]azabicycle of MLA (5) and related natural products and their analogues, and essentially the only (secondary) amine to show such a strong steric compression effect. The NH proton is exoorientated, the bridge methylene protons are so magnetically different; they are an AX system, −1.52 (d, H exo ) and 2.08 (d, H endo ) (J AX gem = 10.2 Hz). The chemical shifts of the two CH 2 protons therefore differ by 3.6 ppm! The chemical shifts of the exo-CH 2 -and the NH-bridge protons are observed at a relatively high field, the endo-CH 2 -bridge proton is strongly deshielded. Such a large Δδ might be due in a considerable part to its steric interaction with the spatially very close (H)Ngroup, possibly due to the van der Waals effect. The NHbridge proton is assigned to the exo-position with a high degree of certainty on the basis of its resonance at a relatively high field, NH exo [CCl 4 , tetramethylsilane (TMS)] −2.07 (br s) ppm. If this proton were to be in the endo-position, then it would not only have to be markedly deshielded as a result of the H−H (steric) compression but also show a not-present nuclear Overhauser effect. 7 This effect of steric compression can not only be caused by the lone-pair electrons of the N-or another heteroatom (e.g., O-atom) ( Figure 16) but also by a proton (deuterium) or alkyl, for example, ethyl group that is close in space to the compressed proton. This conclusion can help in understanding the conformations of molecules related to [3.3.1]bicycles as a true-chair/true-chair conformation allows a significant steric compression to be demonstrated.     3-(trimethylsilyl)-propionic-2,2,3,3-d 4 acid sodium salt (TMSP) as internal or external standards, and residual (protio) solvent peaks were also used as internal standards if required. Chemical shifts (δ H ) are reported as the position (accurate δ H of overlapping signals were extracted from 2D NMR spectra, e.g., HSQC, COSY, and NOESY), relative integral, multiplicity, and assignment. Multiplicity is abbreviated: s = singlet, d = doublet, t = triplet, q = quartet, quin = quintet, m = multiplet; br = broad. Coupling constants (J) are line separations (absolute values expressed in hertz) rounded and rationalized to 0.1 Hz. 13 C NMR spectra were recorded with complete proton decoupling on Bruker Avance III spectrometers ( 13 C Larmor precession frequency 100 and 125 MHz) at 25°C as well as 2D NMR experiments including HSQC and HMBC. Chemical shifts are expressed in a ppm downfield shift from TMS or TMSP as internal or external standards, and solvent peaks were also used as internal standards if required. Data are reported as the position (δ C ), number of attached protons (CH 3 , CH 2 , CH, quat = quaternary), and assignment.

EXPERIMENTAL
Positive-ion [M + H] + mode-mass spectrometry was performed on samples dissolved in methanol, using Bruker micrOTOF and Agilent Q-TOF mass spectrometers equipped with electrospray ionization (ESI) sources. Negative-ion [M − H] − mode-mass spectrometry was performed on samples dissolved in methanol, on an Agilent ESI-Q-TOF mass spectrometer. High-resolution mass spectra were within 5 ppm error unless otherwise stated.
Intensity SXRD data were collected at 150 ± 2 K on a Rigaku SuperNova Dual, EosS2 system using monochromated Cu Kα radiation (λ = 1.54184 Å). Unit cell determination, data collection, and data reduction were performed using CrysAlisPro software CrysAlisPro 1.171.39.46 (Rigaku Oxford Diffraction, 2018). An empirical absorption correction using spherical harmonics was employed. The structures were solved with SHELXT and refined by a full-matrix least-squares procedure based on F 2 (SHELXL-2018/3). 29 All nonhydrogen atoms were refined anisotropically. Hydrogen atoms were placed onto calculated positions and refined using a riding model.  7 g), acetic acid (20 mL), water (80 mL), and 50% w/v solution of potassium iodide in water (100 mL) were mixed and stored as a stock solution. The stock solution (10 mL) and acetic acid (20 mL) were mixed and made up to 100 mL with water to give Dragendorff's reagent.