Gracilosulfates A–G, Monosulfated Polyoxygenated Steroids from the Marine Sponge Haliclona gracilis

Seven new polyoxygenated steroids belonging to a new structural group of sponge steroids, gracilosulfates A–G (1–7), possessing 3β-O-sulfonato, 5β,6β epoxy (or 5(6)-dehydro), and 4β,23-dihydroxy substitution patterns as a common structural motif, were isolated from the marine sponge Haliclona gracilis. Their structures were determined by NMR and MS methods. The compounds 1, 2, 4, 6, and 7 inhibited the expression of prostate-specific antigen (PSA) in 22Rv1 tumor cells.

During the search for bioactive compounds from the Northwestern Pacific deep-water marine invertebrates [21,22], we collected the pale orange sponge Haliclona gracilis near Shikotan Island, Russia, whose extract exhibited hemolytic and antifungal activities.
The genus Haliclona (order Haplosclerida, family Halinidae) is represented by more than 600 species [23]. Marine sponges of Haliclona genus have been extensively examined, and more than 200 The genus Haliclona (order Haplosclerida, family Halinidae) is represented by more than 600 species [23]. Marine sponges of Haliclona genus have been extensively examined, and more than 200 various bioactive metabolites including steroids, alkaloids, macrolides, polyketides, cyclic peptides, long-chain sphingoid bases, merohexaprenoids, and cyclic bis-1,3-dialkylpyridinium salts have been isolated, and different activities, including cytotoxic and antitumor effects, have been reported [1].
Sponges of Haliclona genus have provided very few sulfated steroids [1]. Thus, only two trisulfated steroids have been isolated in one Indo-Pacific Haliclona sponge [17], while monosulfated polyoxygenated steroids have never been isolated from this genus. Moreover, thus far, the sponge H. gracilis has not been chemically investigated.
The 1 H NMR analysis of the fractions obtained after diverse chromatographic separations suggested the presence of polar metabolites, inspiring our extensive investigation. Here, we report the details of the isolation and structure determination of compounds 1−7, belonging to a new group of naturally occurring monosulfated polyoxygenated steroids with a 3β-O-sulfonato, 5β,6β-epoxy (or 5(6)-dehydro), or 4β,23-dihydroxy substitution pattern as a common structural motif. Additionally, anticancer activities of 1, 2, 4, 6, and 7 were evaluated.

Results and Discussion
The concentrated EtOH extract of the sponge was partitioned between n-BuOH and H2O. The organic extract was concentrated and the obtained residue was fractionated by flash chromatography on a YMC gel column. Further separation using reversed-phase HPLC resulted in the isolation of seven new steroids, gracilosulfates A-G (1-7, Figure 1). Compound 1 was isolated as a white, amorphous solid. The molecular formula of 1 was determined to be C28H47NaO7S from the [M − Na] -ion peak at m/z 527.3045 in the (−)HRESIMS spectrum. The fragment ion peak at m/z 97.9606 in the (−)HRESIMS/MS spectrum and absorption band at 1213 cm -1 in the IR spectrum revealed the presence of a sulfate group in 1.
The 1 H NMR spectrum of 1 (Table 1) showed signals attributable to six methyl groups at δH 1.16 (s), 0.94 (d), 0.91 (d), 0.82 (d), 0.74 (d), and 0.70 (s); four oxygen-bearing methine protons at δH 4.27 Compound 1 was isolated as a white, amorphous solid. The molecular formula of 1 was determined to be C 28 H 47 NaO 7 S from the [M − Na] − ion peak at m/z 527.3045 in the (−)HRESIMS spectrum. The fragment ion peak at m/z 97.9606 in the (−)HRESIMS/MS spectrum and absorption band at 1213 cm −1 in the IR spectrum revealed the presence of a sulfate group in 1.
The 1 H NMR spectrum of 1 (Table 1) 3.55 (br.d), 3.53 (br.d), and 3.17 (br.d); and a series of other methine and methylene multipletes. The 13 C NMR (Table 2) and DEPT spectra of 1 revealed the presence of 28 signals, corresponding to 6 methyls, 8 methylenes, 11 methines, and 3 nonprotonated carbons (one bearing oxygen atom). These data evidenced a C-28 steroidal skeleton. Structure determination of 1 began with HMBC correlations from CH 3 -19 to C-1, C-5, C-9, and C-10. The COSY correlations (Figure 2) delineated the spin system H 2 -1 to H-4, which included protons of oxygenated methines at C-3 and C-4 based on their characteristic chemical shifts. The sequences of protons from H-6 to H-8, H-8 to H 2 -12, H-8 to H 3 -27, and H-24 to H 3 -28 were also established from COSY correlations and indicated the third oxymethine group at C-23. The cross peaks H-4/OH and H-23/OH in the COSY spectrum recorded in DMSO−d 6 ( Figure S11) and the 13 C chemical shifts for C-4 (δ C 77.7) and C-23 (δ C 71.7) implied OH substitution, while the chemical shift for C-3 at δ C 80.1 was more consistent with a sulfate half-ester O(SO 3 )Na [14]. The 13 C NMR signals at δ C 64.2 (CH) and 66.4 (C) and 1 H NMR signal at δ H 3.17 indicated the presence of trisubstituted epoxy ring [24]. The epoxy group was placed at C-5 and C-6 on the basis of HMBC correlations from H 2 -1, H-4, and H-19 to C-5 and from H-6 to C-4, C-8, and COSY correlations H-6/H 2 -7.  (Table 2) and DEPT spectra of 1 revealed the presence of 28 signals, corresponding to 6 methyls, 8 methylenes, 11 methines, and 3 nonprotonated carbons (one bearing oxygen atom). These data evidenced a C-28 steroidal skeleton. Structure determination of 1 began with HMBC correlations from CH3-19 to C-1, C-5, C-9, and C-10. The COSY correlations ( Figure 2) delineated the spin system H2-1 to H-4, which included protons of oxygenated methines at C-3 and C-4 based on their characteristic chemical shifts. The sequences of protons from H-6 to H-8, H-8 to H2-12, H-8 to H3-27, and H-24 to H3-28 were also established from COSY correlations and indicated the third oxymethine group at C-23. The cross peaks H-4/OH and H-23/OH in the COSY spectrum recorded in DMSO−d6 ( Figure S11) and the 13 C chemical shifts for C-4 (δC 77.7) and C-23 (δC 71.7) implied OH substitution, while the chemical shift for C-3 at δC 80.1 was more consistent with a sulfate half-ester O(SO3)Na [14]. The 13 C NMR signals at δC 64.2 (CH) and 66.4 (C) and 1 H NMR signal at δH 3.17 indicated the presence of trisubstituted epoxy ring [24]. The epoxy group was placed at C-5 and C-6 on the basis of HMBC correlations from H2-1, H-4, and H-19 to C-5 and from H-6 to C-4, C-8, and COSY correlations H-6/H2-7. The 20R configuration was demonstrated by the NOESY cross-peak H3-18/H-20 and chemical shift value of CH3-21 at δH 0.94 [26].
The absolute configuration at C-23 was assigned by application of the Mosher's method. Esterification of 1 with (R)-and (S)-α-methoxy-α-(trifluoromethyl)-phenylacetyl chlorides (MTPACl) yielded the 23-MTPA adducts 1S and 1R, respectively, while C-4 hydroxy group was not modified. Interpretation of 1 H NMR chemical shift differences Δδ between 1S and 1R ( Figure 3) revealed that the absolute configuration of C-23 is R. The JH23/H24 coupling constant was 7.3 Hz, which indicated anti relationship of the H-23 and H-24 protons [27] (Figure 4). The NOESY cross peak for H2-22/H3-28 suggested the gauche relationship between the C-22 methylene and C-28 methyl groups, as shown in Figure 3. These data allowed us to determine the 24S absolute configuration.   [26].
The absolute configuration at C-23 was assigned by application of the Mosher's method. Esterification of 1 with (R)-and (S)-α-methoxy-α-(trifluoromethyl)-phenylacetyl chlorides (MTPACl) yielded the 23-MTPA adducts 1S and 1R, respectively, while C-4 hydroxy group was not modified. Interpretation of 1 H NMR chemical shift differences ∆δ between 1S and 1R ( Figure 3) revealed that the absolute configuration of C-23 is R. The J H23/H24 coupling constant was 7.3 Hz, which indicated anti relationship of the H-23 and H-24 protons [27] (Figure 4). The NOESY cross peak for H 2 -22/H 3 -28 suggested the gauche relationship between the C-22 methylene and C-28 methyl groups, as shown in Figure 3. These data allowed us to determine the 24S absolute configuration.
The molecular formula of the second isolated compound, gracilosulfate B (2), was determined as C28H47NaO8S (m/z 543.3002 [M − Na] − ) on the basis of the negative ion HRESIMS analysis. The 1 H NMR data of 2 resembled those of 1, except for the presence of an oxymethine group in 2 (δH 4.14) instead of a methylene group for 1. Analysis of 1D and 2D NMR (COSY, HSQC, and HMBC) spectra allowed us to assign all the observed 1 H and 13 C signals for 2 (Tables 1 and 2). The localization of the additional hydroxy group at C-11 followed from the HMBC correlations between H-11, C-8, and C-13 ( Figure S16) and COSY data. The equatorial disposition of H-11 was evident from the small 3 JHH vicinal coupling to H-9 and H2-12 (Table 1) and confirmed by the relatively low field shift of H-8 (Table 1) caused by the 1,3-diaxial relationship of this proton to the hydroxy group at C-11. The configurations of other stereogenic centers of the ring portion were assigned using similar principles used for 1. The similarity of the NMR data of the side chains of steroids 2 and 1 suggested the same (20R,23R,24S) configuration. Thus, gracilosulfate B was defined as (20R,23R,24S)-4β,11β,23trihydroxy-5β,6β-epoxy-24-methylcholest-3β yl sulfate.
The molecular formula of gracilosulfate C (3), determined as C27H45NaO7S from HRESIMS data (m/z 513.2891[M-Na] -), was one methylene unit less than that of 1. The spectroscopic properties of 3 were similar to those of 1 and differed only by the signals of steroid side chain (Tables 1 and 2). A combination of 2D NMR data showed the lack of a C-28 methyl group, while the remaining portion of the molecule was intact in 1. The configuration of the ring moiety of 3 was assumed to be the same as that of 1 on the basis of the complete overlapping of proton and carbon resonances in NMR spectra. The configuration of the stereogenic center at C-23 was determined by the MTPA method as 23S ( Figure 3). Thus, the gracilosulfate C was defined to be 24-demethyl derivative of gracilosulfate A (1), namely, (20R,23S)-4β,23-dihydroxy-5β,6β-epoxycholest-3β yl sulfate.
Gracilosulfate D (4) with a molecular formula C28H45NaO7S, confirmed by HRESIMS, was isolated as an optically active white amorphous solid. In addition to the signals relative to 3β-Osulfonato-4β,23-dihydroxy structure, the 1 H and 13 C NMR spectra of 4 (Tables 1 and 2 Figure S32), in addition to COSY data ( Figure S30), allowed placement of a hydroxy group at C-15 position, whereas the HMBC correlations between H2-28 and C-23, C-24, and C-25 confirmed the position of terminal methylene group at C-24. The coupling pattern associated with H-15 (ddd, J = 7.9, 5.8, 2.2 Hz) indicated that the hydroxy group at C-15 is βpositioned [28]. The configurations of other stereocenters of the steroid nucleus were assigned by NOESY ( Figure 2) and coupling constants data (Table 1). Thus, the structure of 1 was defined as (20R,23R,24S)-4β,23-dihydroxy-5β,6β-epoxy-24methylcholest-3β yl sulfate and was named gracilosulfate A.
The molecular formula of the second isolated compound, gracilosulfate B (2), was determined as C 28 H 47 NaO 8 S (m/z 543.3002 [M − Na] − ) on the basis of the negative ion HRESIMS analysis. The 1 H NMR data of 2 resembled those of 1, except for the presence of an oxymethine group in 2 (δ H 4.14) instead of a methylene group for 1. Analysis of 1D and 2D NMR (COSY, HSQC, and HMBC) spectra allowed us to assign all the observed 1 H and 13 C signals for 2 (Tables 1 and 2). The localization of the additional hydroxy group at C-11 followed from the HMBC correlations between H-11, C-8, and C-13 ( Figure S16) and COSY data. The equatorial disposition of H-11 was evident from the small 3 J HH vicinal coupling to H-9 and H 2 -12 (Table 1) and confirmed by the relatively low field shift of H-8 (Table 1) caused by the 1,3-diaxial relationship of this proton to the hydroxy group at C-11. The configurations of other stereogenic centers of the ring portion were assigned using similar principles used for 1. The similarity of the NMR data of the side chains of steroids 2 and 1 suggested the same (20R,23R,24S) configuration. Thus, gracilosulfate B was defined as (20R,23R,24S)-4β,11β,23-trihydroxy-5β,6β-epoxy-24-methylcholest-3β yl sulfate.
The molecular formula of gracilosulfate C (3), determined as C 27 H 45 NaO 7 S from HRESIMS data (m/z 513.2891[M-Na] − ), was one methylene unit less than that of 1. The spectroscopic properties of 3 were similar to those of 1 and differed only by the signals of steroid side chain (Tables 1 and 2). A combination of 2D NMR data showed the lack of a C-28 methyl group, while the remaining portion of the molecule was intact in 1. The configuration of the ring moiety of 3 was assumed to be the same as that of 1 on the basis of the complete overlapping of proton and carbon resonances in NMR spectra. The configuration of the stereogenic center at C-23 was determined by the MTPA method as 23S ( Figure 3). Thus, the gracilosulfate C was defined to be 24-demethyl derivative of gracilosulfate A (1), namely, (20R,23S)-4β,23-dihydroxy-5β,6β-epoxycholest-3β yl sulfate.
The molecular formula of C 28 H 45 NaO 8 S was assigned by HRESIMS to gracilosulfate E (5). The 1D (Tables 1 and 2) and 2D NMR analysis showed that gracilosulfate E (5) differs from 4 in the 5,6-epoxy group, replacing trisubstituted double bond. The configurations of the ring moiety were assigned on the basis of the analyses of proton-proton coupling constants (Table 1) and NOESY data. The absolute configuration of the side chain of 5 was determined to be the same as in 4 by comparison of 1 H and 13 C chemical shifts. Thus, gracilosulfate E (5) was determined to be (20R,23R)-4β,15β,23-trihydroxy-5β,6β-epoxy-24-methylenecholest-3β yl sulfate.
Gracilosulfate F (6) of molecular formula C 28 H 45 NaO 8 S was a close analogue of gracilosulfate D (4) showing only an additional oxygen atom. Inspection of 1D (Tables 1 and 2) and 2D NMR data allowed placement of an additional hydroxy group at C-11. The configuration at C-11 was deduced from NOESY correlation of H-11 to axial proton H-1 and small vicinal coupling constant of H-11 (Table 1), which is consistent with an equatorial disposition for this proton, thereby placing the hydroxy group in an axial position. The configurations of remaining stereogenic centers of the ring portion were the same as those of 4, as established on the basis of analyses of proton-proton coupling constants (Tables 1 and 2) and NOESY data. The absolute configuration of the side chain was determined to be the same as that of 4 by comparison of 1 H and 13 C chemical shifts, and finally the structure of 6 was established as (20R, 23R)-4β,11β,15β,23-tetrahydroxy-24-methylenecholest-5(6)-en-3βyl sulfate.
Gracilosulfate G (7) showed the molecular formula C 28 H 47 NaO 7 S as determined by HRESIMS. On the basis of the results of the 1D NMR spectra, we were able to assign a trisubstituted double bond and four oxygen-bearing methine groups. The same steroid core constitution and configurations as in gracilosulfate D (4) were inferred from 1D (Tables 1 and 2) and 2D NMR analysis. The proton and carbon resonances attributable to the side chain of 7 were coincident with those of 1 and 2 (Tables 1 and 2). Thus, gracilosulfate G was defined as (20R, 23R, 24S)-4β,15β,23-trihydroxy-24-methylcholest-5(6)-en-3βyl sulfate.
Next, antitumor activity of compounds 1, 2, 4, 6, and 7 were determined in human prostate cancer cells 22Rv1. Of note, this cell line reveals resistance to androgen receptor (AR)-targeted therapy due to the expression of AR-V7 (AR transcript variant V7), which lacks the androgen-binding site [29,30]. The compounds exhibited moderate cytotoxic activity in the cancer cells after 48 h of treatment. Thus, compound 7 exhibited IC 50 = 64.4 ± 14.9 µM, while the other tested compounds had IC 50 > 100 µM (docetaxel was used as a positive control and exhibited IC 50 = 17.3 ± 6.3 nM). However, all compounds were able to effectively inhibit the expression of PSA (prostate-specific antigen) in 22Rv1 cells ( Figure 5). Earlier, only two monosulfated polyoxygenated steroids have been shown to exert cytotoxic activity on human cancer cell lines [10]. On the other hand, non-sulfated polyoxygenated steroid aragusterol with potent antitumor activities was isolated from a sponge of the genus Xestospongia [31]. Interestingly, for compounds 6 and 7, this effect was already detected at a concentration of 10 µM. PSA is a well-known downstream target of AR signaling. Thus, suppression of PSA expression may indicate an inhibition of this pathway. AR signaling is essential for the growth and survival of prostate cancer cells, with its targeting playing a central role in the modern therapy of advanced prostate cancer. The ability of the isolated compounds to suppress AR signaling can be explained by the similarity of their structures to androgen ligands, which may result in a binding to androgen receptors and therefore blocking of AR-mediated signaling in prostate cancer cells. Gracilosulfate G (7) showed the molecular formula C28H47NaO7S as determined by HRESIMS. On the basis of the results of the 1D NMR spectra, we were able to assign a trisubstituted double bond and four oxygen-bearing methine groups. The same steroid core constitution and configurations as in gracilosulfate D (4) were inferred from 1D (Tables 1 and 2) and 2D NMR analysis. The proton and carbon resonances attributable to the side chain of 7 were coincident with those of 1 and 2 (Tables 1 and 2). Thus, gracilosulfate G was defined as (20R, 23R, 24S)-4β,15β,23-trihydroxy-24methylcholest-5(6)-en-3βyl sulfate.
Next, antitumor activity of compounds 1, 2, 4, 6, and 7 were determined in human prostate cancer cells 22Rv1. Of note, this cell line reveals resistance to androgen receptor (AR)-targeted therapy due to the expression of AR-V7 (AR transcript variant V7), which lacks the androgen-binding site [29,30]. The compounds exhibited moderate cytotoxic activity in the cancer cells after 48 h of treatment. Thus, compound 7 exhibited IC50 = 64.4 ± 14.9 µM, while the other tested compounds had IC50 > 100 µM (docetaxel was used as a positive control and exhibited IC50 = 17.3 ± 6.3 nM). However, all compounds were able to effectively inhibit the expression of PSA (prostate-specific antigen) in 22Rv1 cells ( Figure 5). Earlier, only two monosulfated polyoxygenated steroids have been shown to exert cytotoxic activity on human cancer cell lines [10]. On the other hand, non-sulfated polyoxygenated steroid aragusterol with potent antitumor activities was isolated from a sponge of the genus Xestospongia [31]. Interestingly, for compounds 6 and 7, this effect was already detected at a concentration of 10 µM. PSA is a well-known downstream target of AR signaling. Thus, suppression of PSA expression may indicate an inhibition of this pathway. AR signaling is essential for the growth and survival of prostate cancer cells, with its targeting playing a central role in the modern therapy of advanced prostate cancer. The ability of the isolated compounds to suppress AR signaling can be explained by the similarity of their structures to androgen ligands, which may result in a binding to androgen receptors and therefore blocking of AR-mediated signaling in prostate cancer cells. Figure 5. Effects of the compounds on PSA expression in 22Rv1 cells. The cells were treated with the compounds for 24 h, then the proteins were extracted and examined using Western blotting. β-actin was used as a loading control.

General Procedures
Optical rotations were measured using a PerkinElmer 343 polarimeter (Waltham, MA, USA). IR spectra were recorded using spectrophotometer Equinox 55 (Bruker, Ettlingen, Germany). The 1 H Figure 5. Effects of the compounds on PSA expression in 22Rv1 cells. The cells were treated with the compounds for 24 h, then the proteins were extracted and examined using Western blotting. β-actin was used as a loading control.

Cell Lines and Culture Conditions
The human prostate cancer cell line 22Rv1 was purchased from ATCC (Manassas, VA, USA). Cells were cultured according to the manufacturer's instructions in RPMI media containing 10% FBS (Invitrogen, Carlsbad, USA). Cells were continuously kept in culture for a maximum of 3 months, and were routinely examined for stable phenotype and mycoplasma contamination.

In Vitro MTT-Based Drug Sensitivity Assay
The in vitro cytotoxicity of individual substances was evaluated using a MTT-based assay, which was performed as previously described [33]. Treatment time was 48 h.

Western Blotting
Preparation of protein extracts and Western blotting were performed as described previously [34]. For the detection of PSA, expression the anti-PSA/KLK3 antibodies was used (Cell Signaling, #5365, 1:1000). Treatment time was of 24 h.

Conclusions
In summary, we isolated gracilosulfates A-G, new steroids from the marine sponge H. gracilis, possessing a rare 3β-O-sulfonato, 4β-hydroxy moiety [1]. To date, only one pregnane steroid [35] and two polyhydroxy steroids [36] with such a fragment have been isolated from the sponge Stylopus australis and the starfish Coscinasterias tenuispina, respectively. In addition, the 5β,6β epoxy fragment is unprecedented in sulfated steroids [1]. Finally, the combination of 3β-O-sulfonato, 5β,6β-epoxy (or 5(6)-dehydro), and 4β,23-dihydroxy moieties is unprecedented, taking into account structures of all previously known natural sulfated steroids. Interestingly, these compounds are able to inhibit PSA expression in human hormone-independent prostate cancer cells, suggesting inhibition of AR signaling, a central target for the treatment of advanced prostate cancer. Funding: Isolation and establishment of chemical structures were partially supported by the RSF grant #20-14-00040 (Russian Science Foundation). Search for bioactive compounds from the Northwestern Pacific deep-water marine invertebrates was partially supported by the Grant of the Ministry of Science and Higher Education of the Russian Federation, grant #2020-1902-01-006.