Cross-Coupling as a Key Step in the Synthesis and Structure Revision of the Natural Products Selagibenzophenones A and B

: Selagibenzophenone A ( 1 ) and its isomer selagibenzophenone B ( 2 ) were recently described as natural products from Selaginella genus plants with PDE4 inhibitory activity. Herein, we report the ﬁrst total syntheses of both compounds. By comparing spectroscopic data of the synthetic compounds with reported data for the isolated material, we demonstrate that the structure of one of the two natural products was incorrectly assigned, and that in fact isolated selagibenzophenone A and selagibenzophenone B are identical compounds. The synthetic strategy for both 1 and 2 is based on a cross-coupling reaction and on the addition of organometallic species to assemble the framework of the molecules. Identifying a suitable starting material with the correct substitution pattern is crucial because its pattern is reﬂected in that of the targeted compounds. These syntheses are ﬁnalized via global deprotection. Protecting the phenols as methoxy groups provides the possibility for partial control over the selectivity in the demethylation thanks to differences in the reactivity of the various methoxy groups. Our ﬁndings may help in future syntheses of derivatives of the biologically active natural product and in understanding the structure–activity relationship.

Selagibenzphenone A (1) is a naturally occurring benzophenone derivative, which was recently isolated from Selaginella pulvinata [2]. The determination of the structure revealed that the natural product contains three 4 -hydroxyphenyl rings in positions 2, 4, and 6 of one of the benzophenone rings, the aromatic ring B (Figure 1, 1). The compound has demonstrated inhibitory activity against PDE4, with a promising EC 50 value of 1.04 µM. A closer analysis of the reported 1 H and 13 C NMR spectra of isolated selagibenzophenone A (1) and selagibenzophenone B (2) showed their striking similarity. We put forth two explanations for this similarity: (a) their origin is coincidental, and both compounds actually display similar spectral characteristics; or more likely, (b) only one natural product exists in nature and the structure of the other one was assigned incorrectly. It is not uncommon for the structure of a natural product to be incorrectly solved, as shown by the numerous examples of synthetic work published in the literature that have resulted in subsequent corrections to previously proposed structures of isolated compounds [20,21]. Therefore, we decided to synthesize both molecules. Based on the comparison of spectroscopic data of synthetic and isolated materials, we shed light on this discrepancy in this study. The synthesis and our findings are summarized in the following discussion.

Results and Discussion
Our synthetic strategy relied on the formation of a benzophenone moiety in both cases via an addition of organometallic species to an aldehyde, followed by re-oxidation to a ketone and a cross-coupling reaction with a suitably substituted starting material. The substitution pattern of the starting aromatic synthon is crucial because it will be reflected in the final substitution pattern of compounds 1 and 2. We identified commercially available 2,4,6-tribromobenzaldehyde (6) and methyl gallate (8) as suitable starting materials for compounds 1 and 2, respectively (Scheme 1). Liang and Wang [4] in 2018, and later Xu and Tan [3] in 2020, reported the isolation of a novel benzophenone analogue, selagibenzophenone B (2) (referred to as selaphenin A by Liang and Wang), with potential anticancer activity. The authors proposed that compound 2 differed from selagibenzophenone A (1) in the position of the substitution of the benzophenone core. Benzophenone 1 contains three 4 -hydroxyphenyl rings in positions 2, 4, and 6 of aromatic ring B, whereas selagibenzophenone B (2) has 4 -hydroxyphenyl rings in positions 3, 4, and 5 ( Figure 1, 2).
A closer analysis of the reported 1 H and 13 C NMR spectra of isolated selagibenzophenone A (1) and selagibenzophenone B (2) showed their striking similarity. We put forth two explanations for this similarity: (a) their origin is coincidental, and both compounds actually display similar spectral characteristics; or more likely, (b) only one natural product exists in nature and the structure of the other one was assigned incorrectly. It is not uncommon for the structure of a natural product to be incorrectly solved, as shown by the numerous examples of synthetic work published in the literature that have resulted in subsequent corrections to previously proposed structures of isolated compounds [20,21]. Therefore, we decided to synthesize both molecules. Based on the comparison of spectroscopic data of synthetic and isolated materials, we shed light on this discrepancy in this study. The synthesis and our findings are summarized in the following discussion.

Results and Discussion
Our synthetic strategy relied on the formation of a benzophenone moiety in both cases via an addition of organometallic species to an aldehyde, followed by re-oxidation to a ketone and a cross-coupling reaction with a suitably substituted starting material. The substitution pattern of the starting aromatic synthon is crucial because it will be reflected in the final substitution pattern of compounds 1 and 2. We identified commercially available 2,4,6-tribromobenzaldehyde (6) and methyl gallate (8) as suitable starting materials for compounds 1 and 2, respectively (Scheme 1). The synthesis of selagibenzophenone A (1) commenced with the Suzuki crosscoupling reaction of 2,4,6-tribromobenzaldehyde (6) with a three-fold excess of boronic acid 9 (Scheme 2). In the presence of tetrakis(triphenylphosphine)palladium (0) and Scheme 1. Retrosynthetic analysis and identification of suitable starting materials (6 and 8). The synthesis of selagibenzophenone A (1) commenced with the Suzuki cross-coupling reaction of 2,4,6-tribromobenzaldehyde (6) with a three-fold excess of boronic acid 9 (Scheme 2). In the presence of tetrakis(triphenylphosphine)palladium (0) and potassium carbonate, this reaction provided benzaldehyde 10 in 79% yield. In the next step, aldehyde 10 was subjected to the Grignard reaction with 4-methoxyphenylmagnesium bromide (11) to furnish secondary alcohol 12 in 91% yield. Further oxidation of the alcohol led to the formation of ketone 13 in 74% yield.

Scheme 1. Retrosynthetic analysis and identification of suitable starting materials (6 and 8).
The synthesis of selagibenzophenone A (1) commenced with the Suzuki crosscoupling reaction of 2,4,6-tribromobenzaldehyde (6) with a three-fold excess of boronic acid 9 (Scheme 2). In the presence of tetrakis(triphenylphosphine)palladium (0) and potassium carbonate, this reaction provided benzaldehyde 10 in 79% yield. In the next step, aldehyde 10 was subjected to the Grignard reaction with 4methoxyphenylmagnesium bromide (11) to furnish secondary alcohol 12 in 91% yield. Further oxidation of the alcohol led to the formation of ketone 13 in 74% yield.

Scheme 2. Synthesis of selagibenzophenone A (1).
Our attempts to demethylate anisole moieties led to an unexpected outcome. Using conditions commonly applied for demethylation of methylphenyl ethers and employing boron tribromide [22,23] in CH2Cl2 at 0 °C, a new product was formed in 53% yield and with a significantly higher polarity, indicating the formation of free phenols. However, NMR analysis revealed the presence of one remaining methoxy group at the aromatic ring A, as depicted in the structure of compound 14 (Scheme 2). The resistance of this methoxy group to demethylation can be explained by the decrease in the Lewis basicity of this particular methoxy group, which was caused by the electron-withdrawing effect of the carbonyl moiety in the para position, thus decreasing the reactivity towards boron tribromide. Such a reactivity has already been described in the literature for similar systems [24]. Despite the fact that such a selectivity in the deprotection step can be beneficial in the synthesis of derivatives of this natural product for medicinal chemistry Our attempts to demethylate anisole moieties led to an unexpected outcome. Using conditions commonly applied for demethylation of methylphenyl ethers and employing boron tribromide [22,23] in CH 2 Cl 2 at 0 • C, a new product was formed in 53% yield and with a significantly higher polarity, indicating the formation of free phenols. However, NMR analysis revealed the presence of one remaining methoxy group at the aromatic ring A, as depicted in the structure of compound 14 (Scheme 2). The resistance of this methoxy group to demethylation can be explained by the decrease in the Lewis basicity of this particular methoxy group, which was caused by the electron-withdrawing effect of the carbonyl moiety in the para position, thus decreasing the reactivity towards boron tribromide. Such a reactivity has already been described in the literature for similar systems [24]. Despite the fact that such a selectivity in the deprotection step can be beneficial in the synthesis of derivatives of this natural product for medicinal chemistry purposes and for understanding the structure-activity relationship, this approach is not applicable for the synthesis of the natural product. Increasing the reaction temperature to 25 • C or to a refluxing temperature did not change the outcome of the reaction either. In addition, applying harsh conditions, as described in the synthesis of related selaginpulvilins C and D, namely using neat MeMgI at 160 • C [12,15], led to the decomposition of the material and to the formation of a complex mixture of products.
Considering the above, we hypothesized that the remaining methoxy group could also be cleaved using nucleophilic instead of electrophilic conditions. Indeed, when applying sodium ethanethiolate in DMF at 100 • C [24], the remaining methoxy group was cleaved and tetraphenol 1 was formed in 42% yield. Moreover, subjecting the fully protected compound 13 to the same reaction conditions resulted in the cleavage of all methoxy groups and in the formation of the desired product 1 in 44% yield (Scheme 2). However, we found that this demethylation was not reproducible despite extensive research. The reasons for the lack of reproducibility of this protocol remain elusive.
These unsatisfactory results, combined with the unpractical use of a large excess of sodium ethylthiolate, which has an unpleasant odor, prompted us to develop a more reliable route to compound 1, employing an alternative and easily removable tert-butyldimethyl silyl (TBS) protective group. Therefore, the second-generation synthesis began with the synthesis of boronic acid 17 from 4-bromophenol (15), which was achieved in two steps, namely protection of the phenol moiety, yielding 91% of bromide 16; and introduction of boronic acid via lithium-halogen exchange, reaction with isopropyl borate, and in situ hydrolysis. Suzuki coupling of aldehyde 6 and boronic acid 17 under similar conditions to those applied in the previous synthesis provided aldehyde 18 in 79% yield. In the next step, aryl bromide 16 was treated with t-butyl lithium and the resulting organolithium species reacted with aldehyde 18. The immediate oxidation of the crude reaction mixture provided ketone 19 in 57% yield over two steps. Global deprotection of TBS groups employing HF-pyridine resulted in the formation of the natural product (1) in 82% yield (Scheme 3).
selaginpulvilins C and D, namely using neat MeMgI at 160 °C [12,15], led to the decomposition of the material and to the formation of a complex mixture of products.
Considering the above, we hypothesized that the remaining methoxy group could also be cleaved using nucleophilic instead of electrophilic conditions. Indeed, when applying sodium ethanethiolate in DMF at 100 °C [24], the remaining methoxy group was cleaved and tetraphenol 1 was formed in 42% yield. Moreover, subjecting the fully protected compound 13 to the same reaction conditions resulted in the cleavage of all methoxy groups and in the formation of the desired product 1 in 44% yield (Scheme 2). However, we found that this demethylation was not reproducible despite extensive research. The reasons for the lack of reproducibility of this protocol remain elusive.
These unsatisfactory results, combined with the unpractical use of a large excess of sodium ethylthiolate, which has an unpleasant odor, prompted us to develop a more reliable route to compound 1, employing an alternative and easily removable tertbutyldimethyl silyl (TBS) protective group. Therefore, the second-generation synthesis began with the synthesis of boronic acid 17 from 4-bromophenol (15), which was achieved in two steps, namely protection of the phenol moiety, yielding 91% of bromide 16; and introduction of boronic acid via lithium-halogen exchange, reaction with isopropyl borate, and in situ hydrolysis. Suzuki coupling of aldehyde 6 and boronic acid 17 under similar conditions to those applied in the previous synthesis provided aldehyde 18 in 79% yield. In the next step, aryl bromide 16 was treated with t-butyl lithium and the resulting organolithium species reacted with aldehyde 18. The immediate oxidation of the crude reaction mixture provided ketone 19 in 57% yield over two steps. Global deprotection of TBS groups employing HF-pyridine resulted in the formation of the natural product (1) in 82% yield (Scheme 3).

Scheme 3. Improved synthesis of selagibenzophenone A (1).
In the synthesis of selagibenzophenone B (2), the hydroxy groups of gallate 8 were converted into triflates in a reaction with triflic anhydride in the presence of triethylamine (Scheme 4). This reaction provided the desired triflate 20 in 96% yield. Suzuki crosscoupling of compound 20 and 3.15 equivalents of 4-methoxyphenyl boronic acid (9) proceeded smoothly and furnished the trisarylated aromatic ester 21 in 71% yield. Reduction of the ester moiety in compound 21 was pursued next. When using DIBAL-H Scheme 3. Improved synthesis of selagibenzophenone A (1).
In the synthesis of selagibenzophenone B (2), the hydroxy groups of gallate 8 were converted into triflates in a reaction with triflic anhydride in the presence of triethylamine (Scheme 4). This reaction provided the desired triflate 20 in 96% yield. Suzuki crosscoupling of compound 20 and 3.15 equivalents of 4-methoxyphenyl boronic acid (9) proceeded smoothly and furnished the trisarylated aromatic ester 21 in 71% yield. Reduction of the ester moiety in compound 21 was pursued next. When using DIBAL-H at −78 • C, this reaction resulted in the formation of the desired aldehyde, albeit with partial overreduction to alcohol. Therefore, the crude reaction mixture was subjected to re-oxidation with pyridinium chlorochromate (PCC) to provide aldehyde 22 in 83% yield. Alternatively, LiAlH 4 can be used for a complete reduction of the ester to primary alcohol, and after re-oxidation with PCC, aldehyde 22 was obtained in 72% overall yield. The aromatic ring D was introduced into the structure via Grignard reaction with 4-methoxyphenylmagnesium bromide. The resulting alcohol was subjected to the PCC-mediated oxidation without further purification and yielded the desired ketone 23 in 61% yield (over two steps). To our delight, subjecting compound 23 to BBr 3 in CH 2 Cl 2 at room temperature resulted in the formation of the desired polyphenol 2 in 36% yield along with monomethoxy derivative 24 in 48% yield (Scheme 4). and after re-oxidation with PCC, aldehyde 22 was obtained in 72% overall yield. The aromatic ring D was introduced into the structure via Grignard reaction with 4methoxyphenylmagnesium bromide. The resulting alcohol was subjected to the PCCmediated oxidation without further purification and yielded the desired ketone 23 in 61% yield (over two steps). To our delight, subjecting compound 23 to BBr3 in CH2Cl2 at room temperature resulted in the formation of the desired polyphenol 2 in 36% yield along with monomethoxy derivative 24 in 48% yield (Scheme 4).

Scheme 4. Synthesis of selagibenzophenone B (2).
Having both desired compounds 1 and 2 in hand, we compared their analytical data. The spectra of the synthetic compounds 1 and 2 are clearly different ( Figure 2). In fact, the change in the chemical shifts of corresponding protons of these molecules in the 1 H NMR spectra are more significant than it appears at first glance. The chemical shift in the signals of aromatic rings A and D in both compounds are strongly influenced by the anisotropic effect of aromatic rings C and E. In compound 1, protons H A and H B (ring A) are shielded by aromatic rings C and E and shifted upfield relative to protons H A and H B in compound 2 (1: H A = 7.42 ppm and H B = 6.60 ppm; 2 H A = 7.78 ppm and H B = 6.76 ppm). Conversely, in compound 2, the shielding zone of rings C and E affects the protons of the aromatic ring D, meaning the signals H C and H D are shifted upfield in compound 2 relative to those in compound 1 (1: H C = 7.58 ppm and H D = 6.90 ppm; 2 H C = 6.66 ppm and H D = 6.48 ppm). The effect is most significant on proton H C , for which the difference in chemical shift is nearly one ppm. Similarly, in both compounds, protons H E and H F from rings C and E are affected by the anisotropic shielding of either conjugated system of the carbonyl group together with ring A in compound 1 or by aromatic ring D in compound 2. The chemical shifts of protons H E (1: 7.11 ppm; 2 6.90 ppm) clearly show that the shielding of the aromatic ring D in 2 is stronger than that of the conjugated carbonyl-aromatic ring A system in 1.

Scheme 4. Synthesis of selagibenzophenone B (2).
Having both desired compounds 1 and 2 in hand, we compared their analytical data. The spectra of the synthetic compounds 1 and 2 are clearly different ( Figure 2). In fact, the change in the chemical shifts of corresponding protons of these molecules in the 1 H NMR spectra are more significant than it appears at first glance. The chemical shift in the signals of aromatic rings A and D in both compounds are strongly influenced by the anisotropic effect of aromatic rings C and E. In compound 1, protons H A and H B (ring A) are shielded by aromatic rings C and E and shifted upfield relative to protons H A and H B in compound 2 (1: H A = 7.42 ppm and H B = 6.60 ppm; 2 H A = 7.78 ppm and H B = 6.76 ppm). Conversely, in compound 2, the shielding zone of rings C and E affects the protons of the aromatic ring D, meaning the signals H C and H D are shifted upfield in compound 2 relative to those in compound 1 (1: H C = 7.58 ppm and H D = 6.90 ppm; 2 H C = 6.66 ppm and H D = 6.48 ppm). The effect is most significant on proton H C , for which the difference in chemical shift is nearly one ppm. Similarly, in both compounds, protons H E and H F from rings C and E are affected by the anisotropic shielding of either conjugated system of the carbonyl group together with ring A in compound 1 or by aromatic ring D in compound 2. The chemical shifts of protons H E (1: 7.11 ppm; 2 6.90 ppm) clearly show that the shielding of the aromatic ring D in 2 is stronger than that of the conjugated carbonyl-aromatic ring A system in 1.
Based on the findings described above, the coincidental similarity for the spectra of isolated selagibenzophenones A and B was ruled out. Consequently, the structure of one of the isolated compounds was incorrectly assigned. For this reason, we compared the spectra of both synthetic compounds 1 and 2 with the spectra of the isolated selagibenzophenones reported in the literature. The chemical shift in the signals observed in 1 H and 13 C NMR spectra are summarized in Table S1 (see supplementary information). The reported spectra of both isolated compounds correspond to the spectra of synthetic compound 1. As such, the structure of the isolated selagibenzophenone B was assigned incorrectly and the compound previously reported as selagibenzophenone B was in fact selagibenzophenone A. Based on the findings described above, the coincidental similarity for the spectra of isolated selagibenzophenones A and B was ruled out. Consequently, the structure of one of the isolated compounds was incorrectly assigned. For this reason, we compared the spectra of both synthetic compounds 1 and 2 with the spectra of the isolated selagibenzophenones reported in the literature. The chemical shift in the signals observed in 1 H and 13 C NMR spectra are summarized in Table S1 (see supplementary information). The reported spectra of both isolated compounds correspond to the spectra of synthetic compound 1. As such, the structure of the isolated selagibenzophenone B was assigned incorrectly and the compound previously reported as selagibenzophenone B was in fact selagibenzophenone A.

General
All of the chemicals were purchased from the common sources, namely Merck KGaA (Darmstadt, Germany), Acros Organics (part of Thermo Fisher, Geel, Belgium), Alfa Aesar (part of Thermo Fisher, Kandel, Germany), Strem Chemicals (Kehl, Germany), PENTA Chemicals (Prague, Czech Republic), Fluorochem (Headfield, United Kingdom), and Cambridge Isotope Laboratories (Tewksbury, MA, United States), Inc. All of the reagents were used without further purification unless otherwise noted. Solvents used in the reactions were distilled and dried prior the use. The reactions were monitored by TLC using Merck TLC (Merck KGaA, Darmstadt, Germany) silica gel 60 F254 plates, using UV lamp (254 nm) detection and Hanessian's stain (CAM). NMR spectra were recorded on a Bruker Avance III spectrometer (Bruker, Billerica, MA, United States, 400 MHz and 600 MHz for 1 H NMR and 100 MHz and 150 MHz for 13 C NMR, respectively) and Varian NMR Solutions 300 (Varian, Inc., Palo Alto, CA, United States, 300 MHz for 1 H NMR and 75 MHz for 13 C NMR). All chemical shifts δ are reported in ppm with a reference to a residual solvent. Mass spectrometry was performed on a VG-Analytical ZAB SEQ (VG Analytical, Manchester, United Kingdom). Infrared spectrum were measured in KBr with a Thermo

Conclusions
In conclusion, we accomplished the total synthesis of the natural product selagibenzophenone A (1), comprising a Suzuki coupling and an addition of an organometallic aromatic compound to a carbonyl moiety to assemble the backbone of the natural product. Further adjustment of the oxidation state and liberation of the phenols led to the synthesis of selagibenzophenone A (1) and to confirmation of the proposed structure. The synthesis was performed using two different protecting group strategies. In the first, the phenols were protected as methoxy groups and partial control over the selectivity of the deprotection was gained, depending on the deprotection method used. This will be useful in the future synthesis of derivatives of the natural product and determination of the structureactivity relationship. However, the protocol leading to the formation of the desired natural product lacked reproducibility, which prompted us to develop a second-generation synthesis procedure, using easily removable TBS protecting groups. This approach allowed us to achieve the first reliable total synthesis of selagibenzophenone A (1).
In addition, we achieved the total synthesis of compound 2, which had been described as a natural product known as selagibenzophenone B by Xu and Tan [3] and by Liang and Wang [4]. Our synthetic studies and comparison of our data with previously reported data revealed that the structure of the isolated material, described as selagibenzophenone B, is in fact misassigned and that the isolated compound is selagibenzophenone A. This finding is important not only for natural product chemists but also for the medicinal chemistry community because several biological activities have been reported for selagibenzophenone B (2) when they should be instead ascribed to selagibenzophenone A (1).
Currently, follow-up studies are being conducted in our laboratory and in the laboratories of our collaborators, where natural and unnatural selagibenzophenones are being prepared and assessed for their biological effects.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/catal11060708/s1: Figure S1: Numbering of selagibenzophenones A and B and key HMBC correlations for selagibenzophenone B; Table S1: Comparison of NMR shifts of synthetic and isolated selagibenzophenones A and B; Figure S2: 1 H NMR spectra of compound 10 in CDCl 3 (400 MHz); Figure S3: 13