Oxygenated Theonellastrols: Interpretation of Unusual Chemical Behaviors Using Quantum Mechanical Calculations and Stereochemical Reassignment of 7α-Hydroxytheonellasterol

A total of eight new oxygenated 4-exo-methylene sterols, 1–8, together with one artifact 9 and six known sterols 11–16, were isolated from the marine sponge Theonella swinhoei collected from the Bohol province in Philippines. Structures of sterols 1–8 were determined from 1D and 2D NMR data. Among the sterols, 8α-hydroxytheonellasterol (4) spontaneously underwent an allylic 1,3-hydroxyl shift to produce 15α-hydroxytheonellasterol (9) as an artifact; this was rationalized by quantum mechanical calculations of the transition state. In addition, the 1,2-epoxy alcohol subunit of 8α-hydroxy-14,15-β-epoxytheonellasterol (5) was assigned using the Gauge-Independent Atomic Orbital (GIAO) NMR chemical shift calculations and subsequent DP4+ analysis. Finally, comparison of the 13C chemical shifts of isolated 7α-hydroxytheonellasterol (6) with the reported values revealed significant discrepancies at C-6, C-7, C-8, and C-14, leading to reassignment of the C-7 stereochemistry in the known structure.

Generally, structures of the natural sterols can be analyzed from the key NMR (HMBC, NOESY) correlations arising from the methyl groups at C-18 and C- 19. However, precise analyses utilizing conventional NMR techniques are problematic in some cases because of the complex overlapping of non-functionalized sp 3 methylene peaks in the 1 H NMR spectrum and the absence of 1 H signals from oxygenated tertiary carbon atoms. Current advances in the prediction of NMR shielding constants employing quantum mechanical calculations have provided alternative tools to clarify the ambiguities in the course of structure determination [31,32]. For instance, the structure of conicasterol F, bearing a tetra-substituted epoxide at C-8 and C-14, was deduced from GIAO calculations of 13 C NMR chemical shifts and DFT-calculated ROE-distances [26].
As a part of our ongoing research to isolate bioactive and structurally interesting natural products, we investigated the metabolites of T. swinhoei, collected from the Bohol province in Philippines and identified eight novel theonellasterol analogs 1-8 (Figure 1), one artifact 9, and six known sterols 11-16 ( Figure S2, Supporting Information). Herein, we report the structural assignments highlighted with DFT calculations to provide a rationale for the unusual chemical behaviors of oxygenated 4-exo-methylene sterols. The structure of 8α-hydroxy-14,15-βepoxytheonellasterol (5) was deduced using GIAO chemical shift calculations. In addition, the structure of 7α-hydroxytheonellasterol (6), determined by Faulkner and Qureshi in 2000 [28], was reevaluated due to significant discrepancies between the reported 13 C NMR chemical shifts and the spectroscopic data obtained in this study.

Results and Discussion
A total of eight new oxygenated theonellasterols 1-8 were obtained from the hexane extract of T. Swinhoei. Theonellasterol-5,8-oxide (1) was isolated as a colorless oil. Its molecular formula was determined to be C 30  Additionally, an sp 2 methine (δ C 116.8, δ H 5.80) and two oxygenated tertiary carbons (δ C 91.9, 86.8) were detected as characteristics of compound 1, suggesting identification of a new analog. While the exocyclic 4,4-di-substituted 4,30 -olefin and endocyclic tetra-substituted olefin at 8,14 or 8,9 were known to be structural features of theonellasterols, the sp 2 methine in 1 indicated the presence of an endocyclic tri-substituted olefin that may be generated by an isomerization or rearrangement of the tetra-substituted olefin. Moreover, the additional oxygenated tertiary carbons and a higher degree of hydrogen deficiency (DBE = 7) compared to theonellasterol A (DBE = 6) indicated the existence of an oxygenated theonellasterol framework bearing an additional ring [33,34].
15β-Hydroxytheonellasterol (3) was isolated as a colorless oil. Its molecular formula was determined to be C 30  and 2D NMR spectra of 3 were almost identical to those of theonellasterol A. However, an additional oxymethine (δ C 70.3, δ H 4.64) was detected, of which the location was assigned as C-15 by HMBC correlations from δ H 4.64 to C-13 (δ C 43.8)/C-17 (δ C 54.2) and from H 2 -16 (δ H 1.95, 1.60) to δ C 70.3. The β-orientation of the hydroxyl group at C-15 was established from the NOESY correlation between H-15 and H-17 (δ H 1.60).
(a) Due to its labile nature, sterol 4 was entirely decomposed into a complex mixture of unidentifiable compounds within several days. However, extended storage in benzene afforded an artifact as a single compound (Figure 4). Its molecular formula was determined to be C 30 1.49) and HMBC correlations with C-13 (δ C 43.5)/C-14 (δ C 141.1)/C-17 (δ C 54.1) to be assigned at C-15. Based on the stereochemistry of 3, the artifact was determined to be 15α-hydroxytheonellasterol (9), which was produced through an allylic 1,3-hydroxyl migration of 4.
HRFABMS (m/z [M − H2O + H] + 425.3781, calcd 425.3783). The 1D and 2D NMR data were almost identical to those of compound 3, except for the deshielded oxymethine (δC 84.4, δH 4.97). The oxymethine exhibited a 1 H-1 H COSY cross peak with H2-16 (δH 1.49) and HMBC correlations with C-13 (δC 43.5)/C-14 (δC 141.1)/C-17 (δC 54.1) to be assigned at C-15. Based on the stereochemistry of 3, the artifact was determined to be 15α-hydroxytheonellasterol (9), which was produced through an allylic 1,3-hydroxyl migration of 4.  Calculation of transition state energy using the Linear Synchronous Transit (LST) method revealed that the energy barrier for the transformation of 4 to 9 was only 0.6 kcal/mol, which can explain the instability of 4 ( Figure 5). The structure of the transition state (TS) was turned out to be almost identical to that of 9. In the transition state, the B-ring was flipped to a chair-like conformation, bringing O-8 and sp 2 C-15 (1.43 Å ) in close proximity. The atomic distance between C-8 and O-8 in TS was measured to be 3.09 Å , suggesting that the C-8-O-8 bond was actually broken before the TS to form a new C-O bond at C-15. In addition, the C-8-C-14 bond length was estimated to be 1.34 Å , indicating olefin migration from △ 14,15 to △ 8,14 ( Figure S4, Supporting Information). Although 9 was slightly more stable than 4 at rt (G° = −0.3 kcal/mol), the low activation energy and the formation of a more rigid tetra-substituted olefin perhaps shifted the chemical equilibrium toward 9. Figure 5. Free energy diagram for 1,3-hydroxyl migration in 8α-hydroxy-theonellasterol (4) to generate 15α-hydroxytheonellasterol (9). Geometry optimizations of compounds 4, 9, and the transition state (TS) were performed at the mPW1PW91/6-31G* level of theory. Calculation of transition state energy using the Linear Synchronous Transit (LST) method revealed that the energy barrier for the transformation of 4 to 9 was only 0.6 kcal/mol, which can explain the instability of 4 ( Figure 5). The structure of the transition state (TS) was turned out to be almost identical to that of 9. In the transition state, the B-ring was flipped to a chair-like conformation, bringing O-8 and sp 2 C-15 (1.43 Å) in close proximity. The atomic distance between C-8 and O-8 in TS was measured to be 3.09 Å, suggesting that the C-8-O-8 bond was actually broken before the TS to form a new C-O bond at C-15. In addition, the C-8-C-14 bond length was estimated to be 1.34 Å, indicating olefin migration from 14,15 to 8,14 ( Figure S4, Supporting Information). Although 9 was slightly more stable than 4 at rt (∆G • = −0.3 kcal/mol), the low activation energy and the formation of a more rigid tetra-substituted olefin perhaps shifted the chemical equilibrium toward 9.
8α-Hydroxy-14,15-β-epoxy-theonellasterol (5) was isolated as an amorphous powder. Its molecular formula was determined to be C 30  bringing O-8 and sp 2 C-15 (1.43 Å ) in close proximity. The atomic distance between C-8 and O-8 in TS was measured to be 3.09 Å , suggesting that the C-8-O-8 bond was actually broken before the TS to form a new C-O bond at C-15. In addition, the C-8-C-14 bond length was estimated to be 1.34 Å , indicating olefin migration from △ 14,15 to △ 8,14 ( Figure S4, Supporting Information). Although 9 was slightly more stable than 4 at rt (G° = −0.3 kcal/mol), the low activation energy and the formation of a more rigid tetra-substituted olefin perhaps shifted the chemical equilibrium toward 9.  Due to the limited spectroscopic data for the 1,2-epoxy alcohol subunit of 5, GIAO NMR chemical shift calculations were employed to support the assignments (Table 1). Although our observations suggested the maximum possibility of the 14,15-epoxide isomers 5-I and 5-II, the formation of 8,14-epoxide isomers 5-III and 5-IV could not be ruled out. The 13 C NMR chemical shift calculations of the four sets of 8,14,15-isomers using the mPW1PW91/6-31G** level of theory and subsequent DP4+ analysis indicated 100% probability of 8α-hydroxy-14,15-β-epoxy-isomer 5-I [26]. The correlation coefficient (R 2 ) in the regression analysis of the experimental versus calculated 13 C chemical shifts of 5-I was calculated to be 0.9908, indicating that the structure assignment was highly reliable. As anticipated from the downfield shift of H-5, the B-ring in the optimized structure adopted a boat conformation to rationalize the downfield shift of H-5 by 1,4-flagpole interactions (Table S4, Supporting Information).
Compound 6 was isolated as a colorless needle-shaped solid. The molecular formula was determined to be C 30 H 50 O 2 by HRESIMS (m/z [M + Na] + 465.3705, calcd 465.3709), indicating six degrees of unsaturation. Comparison of the NMR spectra of 6 with the previously reported data revealed an oxygenated theonellasterol-type framework bearing an additional oxymethine (δ C 66.7, δ H 4.64). The HMBC correlation from H 2 -6 (δ C 1.77, 1.59) to δ C 66.7, as well as the 1 H-1 H COSY cross peak between the protons at δ H 4.64 and H 2 -6, suggested that the oxymethine was positioned at C-7.
Because the isolation of 7α-hydroxytheonellasterol was reported in 2000 by Faulkner and Qureshi [28], compound 6 was initially considered to be 7β-hydroxytheonellasterol, as deduced from the comparison of 1 H and 13 C chemical shifts (Figure 6a). However, a lack of NOESY signals corresponding to H-7 led us to synthesize 3,7-dimethyl ether 10 from 6. Surprisingly, the NOESY data of 10 indicated a correlation between OMe-7 (δ H 3.14) and H-9 (δ H 2.29), supporting the α-orientation of the C-7 hydroxyl group. In addition, methylation of swinhosterol C (11), known as 7α-OMe-theonellasterol, afforded a compound that was spectroscopically identical to 10 (Figure 6b,c). Single crystal X-ray diffraction of 6 further confirmed a 7α-hydroxytheonellasterol structure (Figure 7). Considering the large differences in the 13 C chemical shifts at C-6, C-7, C-8, and C-14, we speculate that the previously reported compound is the 7β-epimer of 6. The correlation coefficient (R 2 ) in the regression analysis of the experimental versus calculated 13 C chemical shifts of 5-I was calculated to be 0.9908, indicating that the structure assignment was highly reliable. As anticipated from the downfield shift of H-5, the B-ring in the optimized structure adopted a boat conformation to rationalize the downfield shift of H-5 by 1,4-flagpole interactions (Table S4, Supporting Information). Because the isolation of 7α-hydroxytheonellasterol was reported in 2000 by Faulkner and Qureshi [28], compound 6 was initially considered to be 7β-hydroxytheonellasterol, as deduced from the comparison of 1 H and 13 C chemical shifts (Figure 6a). However, a lack of NOESY signals corresponding to H-7 led us to synthesize 3,7-dimethyl ether 10 from 6. Surprisingly, the NOESY data of 10 indicated a correlation between OMe-7 (δH 3.14) and H-9 (δH 2.29), supporting the α-orientation of the C-7 hydroxyl group. In addition, methylation of swinhosterol C (11), known as 7α-OMetheonellasterol, afforded a compound that was spectroscopically identical to 10 (Figure 6b,c). Single crystal X-ray diffraction of 6 further confirmed a 7α-hydroxytheonellasterol structure (Figure 7). Considering the large differences in the 13 C chemical shifts at C-6, C-7, C-8, and C-14, we speculate that the previously reported compound is the 7β-epimer of 6.  8β-Hydroxy-7α-formyl-B-northeonellasterol (7) was isolated as a colorless oil. Its molecular formula was determined to be C30H50O3 by HRESIMS (m/z [M + Na] + 481.3660, calcd 481.3658). Analysis of the 1D and 2D NMR data of 7 revealed an oxygenated theonellasterol-like skeleton bearing an aldehyde (δC 204.2, δH 9.75) and an additional oxygenated tertiary carbon (δC 87.2). Since the aldehyde moiety is known as a unique feature of 8β-hydroxy-B-norconicasta-6α-aldehyde among the sterols isolated from T. swinhoei [37], the 6/5/6/5-fused cyclic backbone of 7 was assigned by comparing the NMR data. Compound 7 could be differentiated from 8β-hydroxy-B-norconicasta-6αaldehyde in the ethyl substituent at C-24, which was assigned based on the HMBC correlations from a triplet methyl group (δH 0.95) to C-24 (δC 46.9)/C-28 (δC 23.8).
28-Homoswinhoeisterol (8) was isolated as a yellow oil. The molecular formula was determined to be C30H48O2 by HRESIMS (m/z [M + Na] + 465.3549, calcd 463.3552). The IR spectrum of 8 clearly indicated the presence of a hydroxyl group (3343 cm −1 ) and a ketone group (1593 cm −1 ). The features of the IR and NMR spectra of this compound were almost identical to those of swinhoeisterol A. The only difference was found in the triplet methyl group (δH 0.84), which was involved in a spin system for H-24-H2-28-CH3-29, as evident from the 1 H-1 H COSY spectrum. This suggested that the ethyl group was located at C-24. In addition, the plausible biogenetic pathway reported by Zhang et al. [21] indicated that compound 8 could be originated from swinhosterol A through an intramolecular aldol-type reaction, which strongly suggested that the absolute configuration of 8 is (3S,5R,7R,10S,13R,17R,20R,24S). To date, 28-homoswinhoiesterol (8) is the only 6/6/5/7-fused cyclic sterol derived from C30 sterols such as theonellasterol A and swinhosterol A. 8β-Hydroxy-7α-formyl-B-northeonellasterol (7) was isolated as a colorless oil. Its molecular formula was determined to be C 30 H 50 O 3 by HRESIMS (m/z [M + Na] + 481.3660, calcd 481.3658). Analysis of the 1D and 2D NMR data of 7 revealed an oxygenated theonellasterol-like skeleton bearing an aldehyde (δ C 204.2, δ H 9.75) and an additional oxygenated tertiary carbon (δ C 87.2). Since the aldehyde moiety is known as a unique feature of 8β-hydroxy-B-norconicasta-6α-aldehyde among the sterols isolated from T. swinhoei [37], the 6/5/6/5-fused cyclic backbone of 7 was assigned by comparing the NMR data. Compound 7 could be differentiated from 8β-hydroxy-B-norconicasta-6α-aldehyde in the ethyl substituent at C-24, which was assigned based on the HMBC correlations from a triplet methyl group (δ H 0.95) to C-24 (δ C 46.9)/C-28 (δ C 23.8).
28-Homoswinhoeisterol (8) was isolated as a yellow oil. The molecular formula was determined to be C 30 H 48 O 2 by HRESIMS (m/z [M + Na] + 465.3549, calcd 463.3552). The IR spectrum of 8 clearly indicated the presence of a hydroxyl group (3343 cm −1 ) and a ketone group (1593 cm −1 ). The features of the IR and NMR spectra of this compound were almost identical to those of swinhoeisterol A. The only difference was found in the triplet methyl group (δ H 0.84), which was involved in a spin system for H-24-H 2 -28-CH 3 -29, as evident from the 1 H-1 H COSY spectrum. This suggested that the ethyl group was located at C-24. In addition, the plausible biogenetic pathway reported by Zhang et al. [21] indicated that compound 8 could be originated from swinhosterol A through an intramolecular aldol-type reaction, which strongly suggested that the absolute configuration of 8 is (3S,5R,7R,10S,13R,17R,20R,24S). To date, 28-homoswinhoiesterol (8) is the only 6/6/5/7-fused cyclic sterol derived from C 30 sterols such as theonellasterol A and swinhosterol A.
Generally, the 24S configurations of 24-ethyl-sterol analogs are deduced from the 13 C chemical shift differences between CH 3 -26 and CH 3 -27 [38]. However, the differences for compounds 2 (0.6 ppm), 5 (0.7 ppm), and 7 (0.8 ppm) were not significant enough for determining the configuration of C-24, and hence, a complementary method was required for the assignment. To establish a universal database using 1 H NMR data, the absolute values of δ H-26 -δ H-27 were obtained from the sets of sterol-type compounds bearing a (21R,24R) or (21R,24S)-21,26-dimethyl-24-ethylhexane side chain (Table S11, Supporting Information). Values calculated for the (21R,24R)-set were higher than 0.04 ppm, whereas those for the (21R,24S)-set were smaller than 0.04 ppm. The validity of our database was evaluated using six known 24S-ethyl-sterols isolated in this study: theonellasterol A (12), E (13), G (14), and K (15); swinhosterol A (16) and C (11). In all the cases, the differences were smaller than 0.03 ppm to prove the reliability of the database. Further, this method was extended to the new compounds 1-8. The differences for all of them were in the desirable range (<0.03 ppm), confirming their 24S configuration ( Figure S8, Supporting Information).

General Experimental Procedures
Specific optical rotations were obtained on a Rudolph Research Analytical (Autopol III) polarimeter (Rudolph Research Analytical, Hackettstown, NJ, USA). IR spectra were recorded on a JASCO FT/IR-4100 spectrophotometer (JASCO Corporation, Tokyo, Japan). The 1D ( 1 H and 13 C) and 2D (COSY, HSQC, HMBC, and NOESY) NMR spectra were taken in C 6 D 6 or CDCl 3 using Bruker 600 MHz spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) at 297.1 K. 1 H NMR spectra were collected after 64 scans, and 13 C NMR spectra were collected at a range of 10,000~15,000 scans depending on the sample concentrations. The mixing time for NOESY experiments was set as 0.3 s. Chemical shifts are reported in parts per million relative to C 6 D 6 (δ H 7.16, δ C 128.4) and CDCl 3 (δ H 7.26, δ C 77.1). High resolution mass-spectra were obtained on a Waters Q-TOF spectrometer (Waters

Biological Material
The biological material was collected in March 2016 from the Bohol province in Philippines (9 • 43 31.86" N, 124 • 32 35.57" E) at a depth of 15 m using scuba diving. The sponge was kept frozen at −20 • C until identified as Theonella swinhoei and chemically analyzed. A voucher sample (163PIL-102) has been stored at the marine biotechnology center, Korea Institute of Ocean Science & Technology (KIOST).

13 C Chemical Shift Calculations
The conformational searches were performed using the Macromodel software (Maestro Materials Science 3.7.013 based on Maestro Core 12.3.013, MMshare Version 4.9.013, Release 2020-1, Platform Windows-x64; New York, NY, USA). The conformers within an energy threshold of 5 kJ/mol were optimized employing DFT calculations at the mPW91PW1/6-3lG* level of theory to estimate gas phase energies and Gibbs free energies. All of the optimizations were performed at "fine" grid density and "ultrafine" accuracy level. The structure that has the lowest gas phase energy was selected, and NMR shielding constants were calculated with the mPW91PW1/6-3lG**/CPCM benzene basis set. The calculated 13 C chemical shifts of compounds 5-I-IV were referenced to the 13 C chemical shift of tetramethylsilane (TMS), computed with the same level of theory (for the details, see Supporting Information).

Conclusions
A total of eight new oxygenated 4-exo-methylene sterols 1-8 and six known sterols (11)(12)(13)(14)(15)(16) were isolated from T. swinhoei. The C-7 stereochemistry of the reported 7α-hydroxytheonellasterol has been revised based on the outcome of a series of chemical modifications and the X-ray crystallography data of 6. The stereo and regiochemistry of the 1,2-epoxyalcohol moiety in 8α-hydroxy-14,15-β-epoxy-theonellasterol (5) was determined by GIAO chemical shift calculations. The reaction pathway for the 1,3-hydroxyl migration of 4 was calculated using quantum mechanical calculations to explain the observed reaction spontaneity. In addition, the unusual downfield shifts observed for H-5 in compounds 4 and 5 were rationalized through geometry optimizations, which indicated the presence of an 8α-hydroxyl group in 6/6/6/5-fused cyclic sterols.