Complete 1 H NMR Assignment of Diosgenin Benzoate

Abstract

Diosgenin (1a) is used as a raw starting material for the synthesis of steroid hormones and contraceptives and is probably one of the most commercially important steroidal sapogenins. The literature on 1a is extensive, and topics range from its isolation from a variety of natural sources to its chemical modifications and therapeutic properties as an anti-cancer, anti-diabetic, anti-thrombosis, anti-inflammatory, and anti-viral agent, among others. ¹ The spectroscopic characteristics of 1a have been analyzed for a long time, but the assignment of the complete ¹H NMR profile remains incomplete due to the overlapping signals obtained with the most common spectrometers. In the current study, the 750-MHz ¹H NMR spectrum of diosgenin benzoate (1b) has been completely assigned by means of QM-based ¹H iterative full spin analysis (HiFSA), which is used for the complete ¹H NMR description of intricate natural products and the quality control of pharmaceuticals. ² The complete assignment was facilitated by taking advantage of some chemical shifts described for 1a and the coupling constants of the cholestane skeleton of cholesteryl benzoate, which were also recently calculated by HiFSA. ³

Introduction

In the wide world of natural compounds, there are classic molecules that are discussed often due to their impacts in pharmaceuticals. (25R)-spirost-5-en-3β-ol, namely diosgenin (1a) (Figure 1), is one of these molecules and is a fundamental raw starting material for the synthesis of steroid hormones and steroidal contraceptives. ⁴ 1a has also gained importance due to its utility as a broad therapeutic target or chemopreventive drug in the treatment of chronic diseases such as cardiovascular disease, cancer, and diabetes, among others. ¹

Figure 1.

Molecular structures of 1a and 1b.

In 1935, Fujii and Matsukawa isolated 1a for the first time from the rhizomes of Dioscorea tokoro. ⁵ Now, it is well known that 1a can also be isolated from ∼137 species from the genus Dioscorea and other botanicals, including some species of Trigonella, Costus, Smilax, and Paris. ⁶ The first application of 1a in the synthesis of hormones is accredited to Russell Marker, a researcher from Pennsylvania State University, who also discovered an excellent source for 1a in a Mexican Dioscorea species called “cabeza de negro” by natives. ⁷ With “Marker degradation”, 1a was used to produce progesterone. Using subsequent reactions, it can also be used to produce estradiol and testosterone. Thus, in the 1940s, it entered the forefront of the hormone industry in Mexico. ⁸

The production and purification of 1a have been optimized, ⁹ and it is now known that endophytes are a good alternative for the isolation of this sapogenin. ¹⁰ Several techniques have been employed for its characterization. Standard analytical methods include spectrophotometry, gravimetry, or thin layer chromatography (TLC), and more advanced methods include immunoenzymatic assays (ELISAs), gas chromatography, liquid chromatography, and ultra-high performance liquid chromatography. ¹¹ Nuclear Magnetic Resonance (NMR) is an alternative analytical method that can be very useful in the characterization of sapogenins and a wide range of natural products. Essentially, there are two reports dedicated to NMR analysis of 1a, but they are limited to only the description of the ¹H and ¹³C chemical shifts^12,13 and do not contain the coupling constants, which remain unknown until now.

In addition to these two classical papers, the NMR spectral assignment of 1a analogs is contained in several papers^14,15 as the excellent review by Simonet et al, ¹⁶ which provides the ¹³C and ¹H NMR spectroscopic data of aglycones from Agave saponins, however, the spectroscopic data of this compounds described chemical shifts without coupling constants.

Therefore, the detailed analysis of the coupling constants for economically important compounds such as 1a has application for quality control techniques as well as for inclusion and availability in repositories of natural products. ¹⁷

In this work, we describe for the first time the ¹H NMR fingerprint of diosgenin benzoate (1b), including the complete set of coupling constants. This was achieved by spectral calculation using HiFSA,^18‐20 which has been used to successfully describe the spectroscopic data of important natural products.^3,21,22

Materials and Methods

General

The benzoylation of 1a ²³ was carried out by its treatment with benzoyl chloride in pyridine to give 1b in quantitative yield, as white solid, which, after recrystallization from ethyl acetate-hexane, gave colorless needles: m.p. 237-238°. ²⁴ Approximately 10 mg of 1b was placed in a 5 mm NMR tube and dissolved in 0.9 mL of CDCl₃ with a small amount of TMS. The sample was degassed under an argon stream for 3 min and a final volume of 0.5 mL was obtained.

NMR Spectra

The spectra of 1b were acquired on a Bruker Ascend 750 spectrometer equipped with a cryoprobe. After a shimming process, a signal-line broadening of 0.8 Hz was obtained. The ¹H NMR spectrum was recorded with the following parameters: Acquisition time (AQ) = 2.18 s, 90^o pulse width (P1) = 7.19 ms, relaxation delay (RD) = 1.0 s, number of scans (NS) = 16, spectrum width (SW) = 15,000 Hz, and FT size = 262,144 data. The ¹³C NMR spectrum was acquired with the following parameters: 1024 scans, AQ = 1 s, RD = 2.0 s, 90^o P1 = 12.25 ms, SW = 45,454.5 Hz, and FT size = 32,768 data. The parameters for the HSQC spectrum of 1b were AQ = 0.08 s, NS = 8, and RD = 1.5 s; SW ¹H = 6334.5 Hz with 512 increments; and ¹³C = 33,786.8 Hz with 256 increments.

1H NMR Full Spin-Spin Analysis (HiFSA)

HiFSA was done using PERCH NMR software (v. 2011.1, PERCH Solutions Ltd, Kuopio, Finland). ²⁵ Briefly, the 750-MHz ¹H NMR experimental spectrum of 1b was imported in JCAMP format into the preparation module (PAC) and subjected to phase and baseline correction. A molecular model of 1b was constructed in the molecular modeling module (MMS), and a Monte Carlo conformational search was performed to obtain the minimum energy conformer to obtain the initial ¹H NMR spectrum. As a first approximation, all known chemical shifts for 1a^12,13 were incorporated in the spectral parameter editor (PMS). For the coupling constants in the obscured regions, the cholestane skeleton values of cholesterol ³ were used as a reference.

HiFSA was also replicated using the prerelease version of Cosmic Truth (CT), ²⁶ a web-based software for spectral analysis (NMR Solutions Ltd, Kuopio, Finland) since PERCH NMR software is commercially discontinued. Like PERCH, the ¹H NMR spectrum was imported in JCAMP format for CT, and the structure of 1b was imported in MOL format. The first spectrum calculated by CT was fed with values obtained from PERCH calculation. However, the iteration process was much faster and more efficient since CT takes advantage of current hardware and programming developments for optimized calculations. The HiFSA profiles obtained with PERCH and CT were compared (Tables S1 to S3, supporting information), which showed that the biggest differences in the chemical shifts and coupling constants between these platforms were below 0.4 Hz.

Results and Discussion

1H NMR Iterative Full Spin Analysis (HiFSA)

Although 1D and 2D NMR techniques now offer excellent options for the characterization of structurally intricate natural compounds, the complete NMR characterization is still pending for some industrial and pharmacologically relevant molecules. As an example, a description of the complete set of coupling constants is required for the NMR profile of 1a, which is the starting material for the synthesis of medical steroids such as progesterone. Thus, in this work, we detailed the assignment of 1b using HiFSA integrated in PERCH v.2011.1 software and reproduced it using the prerelease version of CT software. ²⁶

First, the 750-MHz free induction decay of 1b was imported into the PAC unit of the PERCH software and subjected to base line correction. Next, the 3D molecular structure of 1b was built in the MMS module, optimized, and subjected to conformational analysis by a Metropolis Monte Carlo search. The most stable conformer was used to get the initial predicted spectra containing the chemical shifts (δH), coupling constant (ⁿJ_H,H), and line width (Δν1/2) that are appropriate as initial points for the iteration process. The calculated δH values were then manually adjusted using the chemical shifts previously described for 1a.^12,13 For ⁿJ_H,H, some values for cholesterol were used as a reference. ³

After optimization using quantum-mechanical total line shape (QMTLS) iteration, the refinement of all the calculated ¹H NMR parameters was almost achieved. However, comparison of the experimental and calculated spectra showed a root-mean-square deviation (RMSD) greater than 0.1 Hz. Since PERCH is no longer available, the HiFSA was also reproduced with Cosmic Truth (CT) software which is currently in beta stage. ²⁶ Thus, the PMS file from PERCH calculation was imported into the more advanced CT platform, with which the iteration process was faster and more efficient, reaching an excellent RMSD of 0.089%. Finally, the values calculated with the CT software were transferred to the PERCH software, which also achieved an excellent result (RMS = 0.086%).

A complete numerical comparison of the HiFSA profiles obtained with PERCH and CT (Tables S1, S2, and S3, supporting information) showed that the largest difference in the chemical shifts was 0.38 Hz, which is below the digital point resolution of the measurement of 0.46 Hz. Chemical shift standard deviation was 0.29 Hz, which is also below the point resolution, while for the coupling constants it was 0.13 Hz, which is about 1/8 of the average line width of 1.4 Hz.

Interpretation of the ¹H NMR Spectra of 1b

Complete ¹H and ¹³C NMR spin data for 1b are described in Table 1, and graphical comparisons between experimental and calculated profiles are shown in Figure 2. The full chemical shifts of 1a were reported in 1993 by means of 2D NMR techniques. ¹² Generally, all the assignments obtained by HiFSA coincide with those described, except for diasterotopic hydrogens at C-26. Previously, H-26α (ax) (δ 3.38 ppm) had been assigned to 0.02 ppm downfield with respect to H-26β (eq) (δ 3.36 ppm). ¹² However, it is evident from Figure 1 that there is a tenfold difference (δ 0.2 ppm). Thus, the assignments of H-26α (ax) and H-26β (eq) were determined by HiFSA as 3.383 ppm and δ 3.477 ppm, respectively.

Figure 2.

CT-calculated (blue) and experimental (green) ¹H NMR spectra of diosgenin benzoate (1b) in CDCl₃ at 750 MHz.

Table 1.

¹ H Chemical Shifts (ppm), Coupling Orders, and Coupling Constants (Hz) of 1b.

H	δ <συπ>1</sup>Η	J
1α	1.2130	2J1 α ,1 β = −12.54, 3J1 α ,2 α = 3.85, 3J1 α ,2 β = 14.10, 4J1 α ,19 = −0.35
1β	1.9156	2J1 α ,1 β = −12.54, 3J1 β ,2 α = 3.37, 3J1 β ,2 β = 3.77
2α	2.0049	3J1 α ,2 α = 3.85, 3J1 α ,2 β = 14.10, 2J2 α ,2 β = −12.66, 3J2 α ,3 = 4.52, 4J2 α ,4 α = 2.24
2β	1.7386	3J1 α ,2 β = 14.10, 3J1 β ,2 β = 3.77, 2J2 α ,2 β = −12.66, 3J2 β, 3 = 11.42
3	4.8587	3J2 α ,3 = 4.52, 3J2 β ,3 = 11.42, 3J3,4α = 5.06, 3J3,4β = 11.63
4α	2.4782	4J2 α ,4 α = 2.24, 3J3,4α = 5.06, 2J4α,4 β = −13.25, 4J4 α ,6 = −0.25, 5J4 α ,7 α = 0.55, 5J4 α ,7 β = 0.30
4β	2.4658	3J3,4 β = 11.63, 2J4 α ,4 β = −13.25, 4J4 β ,6 = −2.11, 5J4 β ,7 α = 2.57, 5J4 β ,7 β = 2.67
6	5.4229	4J4 α ,6 = −0.25, 4J4 β ,6 = −2.11, 3J6,7α = 2.03, 3J6 , 7 β = 5.19, 4J6,8 = 0.77
7α	1.5892	5J4 α ,7 α = 0.55, 5J4 β ,7 α = 2.57, 3J6,7 α = 2.06, 2J7 α, 7 β = −17.71, 3J7 α, 8 = 10.45
7β	2.0241	5J4 α ,7 β = 0.30, 5J4 β ,7 β = 2.67, 3J6,7 β = 5.19, 2J7 α, 7 β = −17.71, 4J7β,8 = 5.29
8	1.6649	4J6,8 = 0.77, 3J7 α ,8 = 10.45, 3J7 β ,8 = 5.29, 3J8,9 = 11.04, 3J8,14= 10.95
9	1.0241	3J8,9 = 11.04, 3J9,11α = 4.68, 3J9,11 β = 12.54, 4J9,19= −0.6
11α	1.5511	3J9,11 α = 4.68, 2J11 α ,11 β = −13.79, 3J11 α ,12 α = 4.15, 3J11 α 12 β = 2.71
11β	1.4886	3J9,11 β = 12.54, 2J11 α ,11 β = −13.79, 3J11 β ,12 α = 13.40, 3J11 α 12 β = 2.71
12α	1.2045	3J11 α ,12 α = 4.15, 3J11 β ,12 α = 13.40, 2J12α,12 β = −12.56, 3J12 α ,18 = −0.58
12β	1.7527	3J11 α 12 β = 2.71, 3J11 α 12 β = 2.71, 2J12 α ,12 β = −12.56
14	1.1361	3J8,14= 10.95, 3J14,15 α = 5.62, 3J14,15 β = 13.75, 4J14,18 = −3.84
15α	2.0001	3J14,15 α = 5.62, 2J15 α ,15β = −12.05, 3J15α,16 = 7.50
15β	1.2977	3J14,15 β = 13.75, 2J15 α ,15β = −12.05, 3J15 β ,16 = 6.48
16	4.4218	3J15α,16 = 7.50, 3J15 β ,16 = 6.48, 3J16,17 = 8.70
17	1.7917	3J16,17 = 8.70, 3J17,20 = 6.7
18	0.8035	4J14,18 = −0.38,3J12 α ,18 = −0.58
19	1.0898	4J1 α ,19 = −0.35, 4J9,19= −0.6
20	1.8776	3J17,20 = 6.7, 3J20,21= 7.02
21	0.9805	3J20,21= 7.02
23α	1.6005	2J23 α ,23 β = −13.46, 3J23 α ,24 α = 4.22, 3J23 α ,24 β = 2.49
23β	1.6774	2J23 α ,23 β = −13.54, 3J23 β ,24 α =13.62, 3J23 β ,24 β = 4.66
24α	1.4574	3J23 α ,24 α = 4.22, 3J23 β, 24 α = 13.62, 2J24 α ,24 β = −13.19, 2J24 α ,25 = 12.31
24β	1.6247	3J23 α ,24 β = 13.62, 3J23 β ,24 β = 4.66, 2J24 α ,24 β = −13.19, 2J24 β ,25 = 3.66, 4J24 α ,26 α = 2.18
25	1.6273	3J24 α ,25 = 12.31, 3J24 β ,25 = 3.66, 3J25,26 α = 4.32, 3J25,26 β = 11.28, 3J25,27 = 6.57
26α	3.3830	4J24 α ,26 α = 2.18,3J25,26 α = 4.32, 2J26 α ,26 β = −10.95
26β	3.4772	3J25,26 β = 11.28, 2J26 α ,26 β = −10.95
27	0.7930	2J25,27 = 6.57
2’6’	8.0417	3J2’,3’= 7.84, 4J2’,4’= 1.31, 4J2’,6’= 1.76, 5J2’,5’= 0.62
3’5’	7.4324	3J2’,3’= 7.84, 4J3’,4’= 7.73, 4J3’,5’= 1.23, 5J2’,5’= 0.62
4’	7.5439	3J3’,4’= 7.43, 3J4’,5’= 7.43, 4J2’,4’= 1.31, 4J4’,6’= 1.31

Complementarily, these two signals also showed a direct HSQC correlation with the carbon signal at 66.86 ppm corresponding to C-26 (Table S4 and Figure S1, supporting information). This assignment is also in line with that of diosgenin acetate ²⁷ and two compounds structurally close to diosgenin, such as smilagenin ²⁸ and hecogenin. ²⁹ Thus, we can propose a correction for the chemical shifts of H-26α and H-26β for 1a. Additionally, with HiFSA, it was plausible to distinguish shifts of diasterotopic hydrogen atoms at C-11, which had not been resolved thus far.

One of the major advantages of HiFSA is its ability to separate overlapping and higher order signals that are usually annotated as multiplets without any coupling information. HiFSA provides a full spectral analysis also for such signals and defines all relevant coupling constant values for a complete ¹H NMR profile. Therefore, this methodology was exploited to determine the coupling patterns for 1b. Inspection of the geminal coupling constant (²J_HH) of methylenes at C-1, C-2, C-4, C-11, C-12, C-23, C-24, and C-26 of 1b showed values between −10.95 and −13.79 Hz, in accordance with A-, C-, and F-ring chair conformations.^27,30 For the B-ring, the ²J_HH of the methylene at C-7 (−17.71 Hz) was 3 Hz larger than the common vicinal coupling according to the torsion angle (˂ 20^o) adopted for the proximity of a double bond at C-5 and C-6. ³¹ For CH₂-15 of the five-membered D-ring, ²J_HH was −12.05 Hz, which is in line with values for some steroids measured by selective indirect J spectroscopy. ³²

Concerning the vicinal coupling constants (³J_HH) for the A-, C-, and F-rings of 1b, typical values for chair conformations were recognized. The axial-axial values for J_{1 α ,2 β} , J_{2 β, 3}, J_{3,4 β} , J_7α,8, J_8,14, J_8,9, J_{9,11 β} , J_{11 β ,12 α,} J_14,15β, J_{23 α, 24 β} , J_{24 β, 25}, and J_{25,26 β} were in the range of 10.95 to 14.10 Hz, and the axial-equatorial values for J_{1 α ,2 α} , J_{1 β ,2 β} , J_{2 α ,3}, J_{3,4 α ,}, J_{7 β ,8}, J_{9,11 α} , J_{11 α ,12α,} J_11β,_{12 β ,} J₁₄,_15α, J_{23 α ,24 α} , J_{23 β ,24 β,} J_{25,24 α} , and J_{25,26 α ,} were in the range of 3.76 to 5.61 Hz. The equatorial-equatorial values for J_{1 β ,2 α} , J_{11α,12 β} , and J_{23 β ,24 α} were in the range of 2.49 to 3.37 Hz. In a similar way for the B-ring, the ³J_HH values of the C-7 to C-9 portion were in line with a half-chair conformation originating from the distortion of the double bond at C5-C6. Thus, ³J_HH between H-6 and pseudoaxial H-7 _α and pseudoequatorial H-7 _β show values of 2.03 and 5.19 Hz, respectively. ³J_HH between H-8 and the same hydrogens were 10.45 and 5.29 Hz, respectively. The remaining vicinal coupling constants J_{14,15 α ,} J_{14,15 β ,} J_{15 α ,16}, J_{15 α ,16} J_16,17, and J_17,20 in the cyclopentane D-ring and tetrahydrofuran E-ring followed a Karplus-type relationship and are in accordance with a tendency toward twist and envelope conformations, respectively.^27,30 It is worth mentioning that the vicinal constants of the sterane system are coincident with those found for cholesterol determined by HiFSA. ³

Coupling constant values through four or more bonds in spirostane sapogenins have not been described to date since they are usually very small and are difficult to estimate. Calculated spectra of 1b by HiFSA showed three types of ⁴J_HH coupling. The first group included the well-known W arrangement ³³ for J_{1 α ,19}, J_9,19, J_{12 α ,18}, and J_14,18 with values in the range of −0.32 to −0.58 Hz for the coupling involving angular methyl groups at C-18 and C-19, as well as J_{2 α,4a} and J_24,26, which involve the coplanar equatorial hydrogens at C-2/C-4 and C-24/C-26 of 2.27 and 2.18 Hz, respectively. All the values correlate with the typical dihedral angles defined by the fragment H-C-C-C, as reported in pioneering works by Williamson ³⁴ and Barfield. ³⁵

The second group of long-range coupling constants covered allylic coupling constants for vicinal hydrogens on either side of the C5-C6 double bond with J_{4 α ,6} and J_{4 β ,6} values of −0.25z and −2.10 Hz, respectively. The difference between them is in agreement, with the magnitude of the cisoid (J_{4 α ,6}) coupling generally being greater than the transoid (J_{4 β ,6}) coupling in stereodefined systems such as steroids. ³⁶ Finally, the third group of long-range interproton coupling constants for 1b was two pairs of homoallylic couplings, J_{4 α ,7 α} and J_{4 α ,7 β} , which had values below 0.5 Hz, and J_{4 β ,7 α} and J_{4 β ,7 β} , with values close to 3 Hz. This in line with the angle pairs: ϕ = 352° (H4 _α -C4-C5-C6), ϕ’= 250° (C4-C5-C6-H_{7 α} ) and ϕ = 352° (H4 _α -C4-C5-C6), ϕ’ = 135° (C4-C5-C6-H7 _β ) corresponding to the spatial arrangement of the H4 _α - H7 _α and H4 _α - H7 _β resulting in an expected homoallylic coupling constant below 0.5 Hz. In the same way the angle pairs: ϕ = 108° (H4 _β -C4-C5-C6), ϕ’ = 250° (C4-C5-C6-H7 _α ), ϕ = 108° (H4 _β -C4-C5-C6) and ϕ’ = 135° (C4-C5-C6-H7 _β ) correspond for the arrangement of the H4 _β - H7 _α and H4 _β - H7 _β and the homoallylic coupling constants ranging between 2-4 Hz. These relationships are in accordance with Barfield and Sternhell's studies. ³⁷

Finally, with the values of the coupling constants well defined for 1b, the multiplets for the sterane skeleton could be correlated with the “signal templates” proposed by Kirk et al, in 1990. ³⁸ In this work, based on the accumulation of ¹H NMR data of several steroid hormones, signal templates conserving the appearances of proton signals of rings A, B, C, and D were proposed for easy recognition.

Thus, the splitting patterns of H-1α and H-1β for ring A, of H-8 and H-9 for ring B, of H-11α, H-11β, H-12α, H-12β and H-14 for ring C, and of H-15α, H-15β and H-17 for ring D are in accordance with proposed patterns for the steroid skeleton (Figure S2). In a dissimilar manner, the signals of H-2α, H-2β, H-7α and H-7β were not like those of the predicted templates. This was expected since the variability of the hydrogens at C-2 has been described due to the influence of the substituent effects and for the hydrogens at C-7 by the proximity of the double bond. ³⁸

Conclusions

The description of the complete 750-MHz ¹H NMR spectral profile of 1b was accomplished by using QM-based HiFSA. The RMS deviation between the experimental and calculated spectra was below 0.1%. The description of all chemical shifts and coupling constants was made possible by reordering some ¹H NMR chemical shifts previously reported for 1a and the coupling constant of the cholestane skeleton of cholesterol benzoate (1b) that we had described previously. The availability of the full ¹H NMR spectral data for commercially important 1a should be useful for the characterization of new derivatives with applications in the pharmaceutical industry.

Supplemental Material

sj-docx-1-npx-10.1177_1934578X231188214 - Supplemental material for Complete ¹H NMR Assignment of Diosgenin Benzoate

Supplemental material, sj-docx-1-npx-10.1177_1934578X231188214 for Complete ¹H NMR Assignment of Diosgenin Benzoate by Elvia Becerra-Martínez, Angel Ernesto Bañuelos-Hernandez and Nury Pérez-Hernández, Pedro Joseph-Nathan in Natural Product Communications