Chemistry
Chiral glycoside probes were prepared in two versions, the first one classic, with benzyl protection of the hydroxyl groups of 2-deoxyglucose and the second one using less typical benzoyl protecting groups (Fig.1). Probes with benzyl protection were prepared in the usual manner. In the first step methyl 2-deoxyglucoside was prepared, which was then alkylated with benzyl bromide. Anomeric methyl group was removed in mild acidic condition to obtain compound 5a, ready for the final coupling with chiral secondary alcohol. In the case of benzoylated sugars we applied a two-step approach, with benzoylation of all hydroxyl groups in the first step following by selective deprotection of anomeric position. In this step, we tried to apply the deprotection with benzyl amine or ethanolamine, however this proved to be unsuccessful. This turned out to be the method of choice for two step but one-pot deprotection with transformation to glycosyl bromide and subsequent hydrolysis assisted with silver cations leading to formation of 5b. In the last step we applied Schmidt glycosylation with trichloroacetonitrile. Due to significant steric hindrance of used alcohols we experienced low yield of prepared probes. Additionally, in one case we isolated dehydrated glycoside 2B which also turned out to be useful for spectroscopic studies. As a result of chemical syntheses, 1A, 1B, 1C, 2A, 2B and 2C were obtained (Fig.2).
Nuclear Magnetic Resonance
Compound 1A, 1B, 1C, 2A, 2B and 2C were subjected to NMR studies in acetone-d6 or benzene-d6. Using standard DQF-COSY, TOCSY, HSQC, nd-HSQC, HMBC, and ROESY experiments, all of the connections in each isolated spin system of the resulting glycosides were traced in an uncomplicated manner The detailed procedure of the absolute configuration assignment by NMR will be presented on the instance of the molecule 1A.
Discussion of NMR results for 1A and 1B
In the nd-HSQC spectrum the coupling constant 1J C-H of the 1A molecule is 171.1 Hz (Fig. S6), which is synonymous with obtaining the alpha anomer. For the beta anomer 1B, a 1J C-H value of 158.1 Hz is observed (Fig. S15).17 An additional factor which confirms the glycoside conformation is the high chemical shift of the C1' (Table 2) signal in one-dimensional spectra. The 1'H proton shift (Table 1) is due to the effect of shielding the ester group -COPh.
The analysis of the 1A molecule confirmed that the substituents of the cyclohexane ring (aglycone) are in equatorial positions. Interpretation of vicinal constants of the coupling and dipolar couplings allowed to define (C1-(S*), C3-(R*), C5-(S*)) absolute configuration.
ROE diagnostic signals
The probe - aglycone ROEs proved the first diagnostic signal – H’1/H1. This dipolar coupling between the anomeric proton and the proton of oximetine must occur because it indicates the partial inhibition around the O-glycosidic bond. In addition, for 1A, two more ROEs between the protons of the aglycone unit and the protons of the chiral probe were observed, i.e., 5’H/7H and 5’H/8H - diagnostic signal (Table 3). ROESY NMR experiment confirmed the observed correlations in space simulated with the use of MD techniques (Fig. 4). These mentioned ROEs enabled the determination of the absolute configuration of 1A (Fig. 3a). All of the dipolar couplings for the 1A molecule can be observed in the ROESY spectrum (Fig. 3b).
In the case of beta 1B anomer, the ROESY spectrum revealed that a 1'H/1H signal indicative of partial inhibition of O-glycosidic bonding is present (Table S6). However, the beta anomer is not suitable as a chiral probe for determining the absolute configuration for molecule 1B because other diagnostic correlations between probe and aglycone are unobservable (Fig. 3A and Fig. S17). For molecule 1B, computer simulation results were also confirmed by spectroscopic studies (Fig. S51).
Discussion of NMR results for 1C, 2A, 2B and 2C molecules
According to the nd-HSQC spectrum, the obtained compounds 1C, 2A, 2B and 2C are alpha anomers. The values of coupling constants 1J C-H are presented sequentially: 171.2 Hz, 168.4 Hz, 165.6 Hz, 168.3 Hz (Fig. S23), (Fig. S31), (Fig. S39) and (Fig. S47). The NMR spectrum of benzoyl derivatives is less complex than that of benzyl derivatives because it does not contain the -CH2- signals (derived from -OBn) which are superimposed on the chemical protons or carbons’ shifts of the probe (Fig. S34), (Fig. S35). On the other hand, these signals are distinguishable because they focus within the certain range on the ppm scale.
Table 1. Chemical proton shifts and coupling constants for molecule 1A second
1 H NMR
Data for 1A
|
position
|
1H δ (ppm)
|
JH,H (Hz)
|
Aglycone Unit
|
1
|
3.61
|
10.7 (2), 10.7 (6ax), 4.1 (6eq)
|
2
|
1.39
|
10.7 (1), 11.1 (3ax), (3eq)*, 2.6 (7)
|
3ax
|
1.06
|
11.1 (2), 13,1 (3eq), (4ax)*, (4eq)*
|
3eq
|
1.72
|
(2)*, 13.1 (3ax), (4ax)*, (4eq)*
|
4ax
|
0.92
|
(3ax)*, (3eq)*, 12.3 (4eq), 12.1 (5)
|
4eq
|
1.71
|
(3ax)*, (3eq)*, 12.3 (4ax), 5.7 (5)
|
5
|
1.39
|
12.1 (4ax), 5.7 (4eq), (6ax)*, (6eq)*, 7.0 (10)
|
6ax
|
0.89
|
10.7 (1), (5)*, 12.4 (6eq)
|
6eq
|
2.25
|
4.1 (1), (5)*, 12.4 (6ax)
|
7
|
2.49
|
2.6 (2), 7.0 (8), 7.0 (9)
|
8
|
0.89
|
7.0 (7)
|
9
|
1.01
|
7.0 (7)
|
10
|
0.95
|
7.0 (5)
|
Probe Moiety
|
1'
|
5.43
|
0.8 (2’eq), 4.2 (2’ax)
|
2’ax
|
2.18
|
4.2 (1’), 12.9 (2’eq), 12.0 (3’)
|
2’eq
|
2.46
|
0.8 (1’), 12.9 (2’ax), 5.6 (3’)
|
3’
|
5.76
|
12.0 (2’ax), 5.6 (2’eq), 9.9 (4’)
|
4'
|
5.63
|
9.9 (3’), 9.9 (5’)
|
5'
|
4.49
|
9.9 (4’), 4.6 (6’a), 2.4 (6’b)
|
6'a
|
4.54
|
4.6 (5’), 12.1 (6’b)
|
6’b
|
4.62
|
2.4 (5’), 12.1 (6’a)
|
O-Bz (15H)
|
7.43 – 8.09
|
-
|
*These coupling constants could not be measured. Signal pattern remains partially unclear due to serve signal overlap and higher order effects
|
Table 2. Chemical carbon shifts for molecule 1A
13 C NMR
Data for 1A
|
position
|
|
13C δ (ppm)
|
Aglycone Unit
|
1
|
CH
|
75.0
|
2
|
CH
|
48.1
|
3
|
CH2
|
22.7
|
4
|
CH2
|
34.3
|
5
|
CH
|
31.2
|
6
|
CH2
|
39.4
|
7
|
CH
|
25.3
|
8
|
CH3
|
15.2
|
9
|
CH3
|
20.8
|
10
|
CH3
|
21.8
|
Probe Moiety
|
1'
|
anomeric carbon
|
92.9
|
2'
|
|
35.8
|
3’
|
|
70.2
|
4'
|
|
70.6
|
5'
|
|
68.9
|
6'
|
|
63.3
|
O-Bz
|
18 aromatic carbons
|
129.0 - 133.4
|
C=O
|
3 carbonyl carbons
|
165.2
165.3
165.6
|
Table. 3. Diagnostic ROE to proton for 1A compared to 1B
2D ROESY NMR
Data for 1A
|
2D ROESY NMR
Data for 1B
|
position
|
diagnostic ROE to proton
|
position
|
diagnostic ROE to proton
|
Probe Moiety
|
1’
|
1, 6eq, 8
|
1’
|
1, 8, 3’, 5’, 6eq, 7
|
3’
|
2’eq
|
3’
|
2’eq, 5’, 1’
|
4’
|
2’ax
|
4’
|
2’ax
|
5’
|
7, 8
|
5’
|
1’, 3’
|
ROE diagnostic signals
For all molecules 1C, 2A, 2B and 2C the ROESY spectrum confirmed the presence of the first diagnostic signal between the probe's 1'H proton and the aglycone's 1H proton (Table S9), (Table S12), (Table S15), (Table S18). Additionally, a second diagnostic signal was observed for each case. For 1C, the next dipolar coupling between the probe and the aglycone is 5'H/5exH. Both interactions allow the determination of the position of the aglycone in space, thus assigning the absolute configuration of 1C-(R*) borneol (Fig. 3a and Fig. S25). For 2A, three Overhauser effects were observed in the probe - aglycone relationship – 5’H/7H, 5’H/8H and 6’aH/8H. These interdependencies point to the C1-(S*) of menthol oxymethine carbon (Fig. 3a and Fig. S33). Similarly, in the case of the 2B molecule, the ROESY spectrum proves the occurrence of another dipolar coupling, which is important from the point of view of the present examinations – 5’H/7H which allows to recognize C1- (S *) aglycone (Fig. S41). An interesting case is the 2C molecule, as it is the only obtained glycoside with the smallest chiral secondary alcohol as an aglycone – 2-butanol (Fi. 3a and Fig. S48). In this instance, the second relatively weak diagnostic signal ROE - 5'H/4H was also recorded and confirmed the absolute configuration of 2-(S)-butanol.
Molecular Dynamics
In order to assess, whether a glycosidic bond formed by 2-deoxy-D-glucose derivatives in fact exhibits restricted conformational freedom, a set of molecular models of 1A, 1B, 1C, 2A, 2B and 2C were subjected to molecular dynamics studies. Moreover, models of the opposite enantiomers of the studied secondary alcohols with the 2-deoxy-D-glucose-based probes attached, namely 1A’, 1B’, 1C’, 2A’, 2B’ and 2C’ were also examined by the same computational methods.
All 12 studied systems displayed pronounced lack of conformational freedom of the glycosidic linkage, which was evidenced by the Ramachandran plots (Fig. 5 and Fig. S49-S60). The studied glycosides almost immediately assumed the geometry, in which the anomeric 1’H proton of a monosaccharide probe and the oxymethine 1H proton of an aglycone were in syn-type conformation (Fig. S61). This type of geometry was maintained through the remaining simulation time, as it was associated with the global energetic minima of all studied molecules. Therefore, the MD studies have suggested that the 1’H/1H ROEs could – and in fact, were – observed in the ROESY spectra of all examined systems. Nevertheless, since – on the basis of molecular modeling calculations - the 1’H/1H dipolar couplings were expected in case of all 12 studied glycosides, geometric requirements of the molecules to display all the observed diagnostic dipolar couplings at once, i.e. those involving 1’H, 5’H and 6’H protons, were met only by 1A, 1C, 2A, 2B and 2C. For instance, in case of 1A’, MD simulations have shown that if 1’H/1H ROE was recorded, ROEs 5’H/7H and 5’H/8H could not have been recorded since average distances between respective protons were too high (Fig. S50). On the other hand, the average interatomic distances in pairs 1’H/1H, 5’H/7H and 5’H/8H extracted from the MD simulation of 1A were a perfect match for the observed ROESY correlations (Fig. S49). These observations were identical for the rest of the pairs based on aglycone enantiomers (1C-1C’, 2A-2A’ and so on, see Fig. S51-S60) and strongly supported the applicability.