Total Synthesis of Loroxanthin

The first total synthesis of loroxanthin (1) was accomplished by Horner-Wadsworth-Emmons reaction of C25-apocarotenal 8 having a silyl-protected 19-hydroxy moiety with C15-phosphonate 25 bearing a silyl-protected 3-hydroxy-ε-end group. Preparation of apocarotenal 8 was achieved via Stille coupling reaction of alkenyl iodide 10 with alkenyl stananne 9, whereas phosphonate 25 was prepared through treatment of ally alcohol 23 with triethyl phosphite and ZnI2. The ally alcohol 23 was derived from the known (3R,6R)-3-hydroxy C15-aldehyde 20, which was obtained by direct optical resolution of racemate 20 using a semi-preparative chiral HPLC column.


Introduction
Loroxanthin (1) (Figure 1), one of the in-chain hydroxylated carotenoids [1], has been isolated from various green algae and is sometimes found as ∆2-fatty acid esters of the C19-hydroxy moiety [2][3][4][5]. It is presumed to be biosynthesized by C19-hydroxylation of lutein (2) and to be a biosynthetic precursor of siphonaxanthin (4), which is a major photosynthetic pigment of green algae [1,6,7]. The 3 ,6 -trans-configuration of the ε-end group in 1 was determined [8] by comparison of its 1 H-NMR spectrum with that of lutein (2), while the absolute configurations at C3 and C6 in 1 have been assigned as 3R, 6 R [8] from the close resemblance of its circular dichroism (CD) spectrum with that of synthetic (3R,6 R)-loroxanthin model compound 3 [9]. However, compound 3 lacks a hydroxy group at C3 and thus its stereostructure has remained in dispute. Although some biological activities of 1 have been reported [10,11], its properties and functions are not yet well understood due to its limited availability from natural sources. Thus, interest in its function and structure prompted us to undertake the first total synthesis of 1.

Introduction
Loroxanthin (1) (Figure 1), one of the in-chain hydroxylated carotenoids [1], has been isolated from various green algae and is sometimes found as Δ2-fatty acid esters of the C19-hydroxy moiety [2][3][4][5]. It is presumed to be biosynthesized by C19-hydroxylation of lutein (2) and to be a biosynthetic precursor of siphonaxanthin (4), which is a major photosynthetic pigment of green algae [1,6,7]. The 3′,6′-trans-configuration of the ε-end group in 1 was determined [8] by comparison of its 1 H-NMR spectrum with that of lutein (2), while the absolute configurations at C3 and C6′ in 1 have been assigned as 3R,6′R [8] from the close resemblance of its circular dichroism (CD) spectrum with that of synthetic (3R,6′R)-loroxanthin model compound 3 [9]. However, compound 3 lacks a hydroxy group at C3′ and thus its stereostructure has remained in dispute. Although some biological activities of 1 have been reported [10,11], its properties and functions are not yet well understood due to its limited availability from natural sources. Thus, interest in its function and structure prompted us to undertake the first total synthesis of 1.

Results and Discussion
Loroxanthin model compound 3 was previously synthesized by utilizing the Shapiro reaction of C 13 -(arylsulfonyl)hydrazone 5 with C 27 -polyenal 6 and subsequent acid-promoted allylic rearrangement of the resulting adduct 7 and following Lindlar reduction [9] as shown in Scheme 1. However, this procedure cannot be applicable to the synthesis of loroxanthin itself, because the 3 -hydroxy-ε-end moiety of loroxanthin (1) is labile to the acidic conditions [12,13]. Moreover, many stereoisomers were produced during the conversion of compound 7 into compound 3.

Results and Discussion
Loroxanthin model compound 3 was previously synthesized by utilizing the Shapiro reaction of C13-(arylsulfonyl)hydrazone 5 with C27-polyenal 6 and subsequent acid-promoted allylic rearrangement of the resulting adduct 7 and following Lindlar reduction [9] as shown in Scheme 1. However, this procedure cannot be applicable to the synthesis of loroxanthin itself, because the 3′-hydroxy-ε-end moiety of loroxanthin (1) is labile to the acidic conditions [12,13]. Moreover, many stereoisomers were produced during the conversion of compound 7 into compound 3.

Scheme 1. Previous synthesis of loroxanthin model compound 3.
Hence, we planned to synthesize loroxanthin (1) by the condensation of C25-apocarotenal 8 with an appropriate C15-building block A having a 3-hydroxy-ε-end group at the C11′-C12′ double bond position on 1 as shown in Scheme 2. The apocarotenal 8 was expected to be prepared by a three-component connection, which involves the Stille coupling reaction of previously reported C11-alkenyl stananne 9 [14] with the C4-alkenyl iodide 10 and the Wittig reaction with the C10-phosphonium salt 12 [15]. The compound 10 would be converted from the alkenyl stannane 13, which was previously prepared from 1,4-butynediol [16,17]. Hence, we planned to synthesize loroxanthin (1) by the condensation of C 25 -apocarotenal 8 with an appropriate C 15 -building block A having a 3-hydroxy-ε-end group at the C11 -C12 double bond position on 1 as shown in Scheme 2. The apocarotenal 8 was expected to be prepared by a three-component connection, which involves the Stille coupling reaction of previously reported C 11 -alkenyl stananne 9 [14] with the C 4 -alkenyl iodide 10 and the Wittig reaction with the C 10 -phosphonium salt 12 [15]. The compound 10 would be converted from the alkenyl stannane 13, which was previously prepared from 1,4-butynediol [16,17]. First, the C25-apocarotenal 8 was prepared as shown in Scheme 3. According to the reported method [16,17], the O-tert-butyldimethylsilyl (TBS)-protected alkenyl stannane 13 was prepared by Pd-catalyzed stereoselective hydrostannylation of 1,4-butynediol and First, the C 25 -apocarotenal 8 was prepared as shown in Scheme 3. According to the reported method [16,17], the O-tert-butyldimethylsilyl (TBS)-protected alkenyl stannane 13 was prepared by Pd-catalyzed stereoselective hydrostannylation of 1,4-butynediol and subsequent regioselective silylation of the resulting dihydroxy alkenyl stannane. The silylation yield improved (63% to 74%) by changing the reaction solvent from the reported N,N-dimethylformamide (DMF) to CH 2 Cl 2 . The alkenyl stannane 13 was treated with I 2 to provide the labile alkenyl iodide 15, which was promptly reacted with ethyl vinyl ether in the presence of pyridinium p-toluenesulfonate (PPTS) to give the 1-ethoxyethyl (EE)protected alkenyl iodide 10 in a high yield. Stille coupling reaction of this alkenyl iodide 10 with the C 11 -alkenyl stananne 9 [14] under Baldwin's conditions [18] [Pd(PPh 3 ) 4 , CsF, CuI] gave the desired coupling product 16 in a good yield. Desilylation of compound 16 by treatment with tetrabutylammonium fluoride (TBAF) and subsequent MnO 2 oxidation of the resulting alcohol 17 provided the aldehyde 18. The Wittig condensation of C 15aldehyde 18 with C 10 -phosphonium salt 12 [15] in the presence of NaOMe as a base, followed by acidic treatment and subsequence protection of the hydroxy groups on the resulting condensed products 19 with triethylsilyl (TES) groups to yield 9Z (all-trans)-C 25apocarotenal 8 (30% from 18), accompanied by some stereoisomers (mainly contained 9Z,11Z-isomer: 42% from 18). Palladium-catalyzed isomerization [19] of the latter, with careful observation of reaction progress by HPLC, afforded the desired 9Z-isomer of 8 (21%) along with its all-E (9-cis)-isomer (42%). The preference for 9E (9-cis)-isomer indicates its higher thermodynamic stability over the 9Z (9-trans)-isomer (Supplementary Materials). Next, the preparation of C15-building block A (Scheme 2) having a 3-hydroxy-ε-end group was investigated. Mayer and Rüttimann reported [20] the preparation of the optically active phosphonium salt 22 (Scheme 4) for the total synthesis of lutein (2). They described that treatment of the tertiary allyl alcohol 21 with aqueous (aq.) hydrogen bromide afforded the unstable primary allyl bromide, which was reacted with triphenylphosphine, Next, the preparation of C 15 -building block A (Scheme 2) having a 3-hydroxy-ε-end group was investigated. Mayer and Rüttimann reported [20] the preparation of the optically active phosphonium salt 22 (Scheme 4) for the total synthesis of lutein (2). They described that treatment of the tertiary allyl alcohol 21 with aqueous (aq.) hydrogen bromide afforded the unstable primary allyl bromide, which was reacted with triphenylphosphine, followed by treatment of the resulting phosphonium bromide 22 (X = Br) with NaCl solution to yield the phosphonium chloride 22 (X = Cl).
Mar. Drugs 2022, 20, x FOR PEER REVIEW 5 of 12 time under reflux conditions was shortened from overnight (12 h) as described in the literature [21] to 30 min. TES-protected phosphonate 25 was obtained by alcoholysis of the acetyl group of 24 and subsequent silylation of the resulting hydroxy group.  [13,20] Finally, Horner-Wadsworth-Emmons reaction of apocarotenal 8 with phosphonate 25 in the presence of sodium bis(trimethylsilyl)amide [NaN(TMS)2] followed by desilylation with TBAF afforded loroxanthin (1) stereoselectively in 61% yield in two steps. Total yield of 1 from alkenyl stananne 9 including isomerization recovery of apocarotenal 8 was 15% over 9 steps and that from (3R,6R)-aldehyde was 39% over 7 steps. Its 1 H-NMR spectral data were in good agreement with the reported data [8]. While the CD spectrum of the natural product was reported to be non-conservative with a weak positive Cotton effect around 250 nm [8], that of the synthetic product showed a relatively clear curve as shown in Figure 3. Good similarity with the reported spectrum of (3R,3′R,6′R)-lutein (2) [13] having the same configuration was observed.  Several attempts to prepare the phosphonium salt 22 from the allyl alcohol 23 were disappointingly unsuccessful, whereas the phosphonate 25 could be prepared as an efficient building block A as shown in Scheme 4. Their precursor (3R,6R)-allyl alcohol 23 was derived from the known (3R,6R)-aldehyde 20 [13]. Khachik and Chang obtained (3R,6R)-20 by lipase-mediated kinetic acetylation of racemate 20 [13]. We found that both enantiomers of 20 can be separated using a chiral HPLC column (CHIRALPAK IF; Daicel, Tokyo, Japan) and EtOH-tertiary butyl methyl ether (TBME) (1:9) as an eluent with high efficiency [ Figure 2a]. Thus, direct optical resolution of racemate 20 was performed using a semipreparative chiral column as shown in Figure 2b. Approximately 1 g of the racemate 20 could be separated into each pure enantiomer in 5 h by repeated injection of 70 mg of sample at 20-min intervals. Next, after protecting of the hydroxy group of (3R,6R)-20 with an acetyl group, the resulting compound was reduced to the allyl alcohol 23. Referring to Wiemer's method [21], this was treated with triethyl phosphite and ZnI 2 under refluxing dry tetrahydrofuran (THF) to give phosphonate 24 in a good yield. The reaction time under reflux conditions was shortened from overnight (12 h) as described in the literature [21] to 30 min. TES-protected phosphonate 25 was obtained by alcoholysis of the acetyl group of 24 and subsequent silylation of the resulting hydroxy group. yield of 1 from alkenyl stananne 9 including isomerization recovery of apocarotenal 8 was 15% over 9 steps and that from (3R,6R)-aldehyde was 39% over 7 steps. Its 1 H-NMR spectral data were in good agreement with the reported data [8]. While the CD spectrum of the natural product was reported to be non-conservative with a weak positive Cotton effect around 250 nm [8], that of the synthetic product showed a relatively clear curve as shown in Figure 3. Good similarity with the reported spectrum of (3R,3′R,6′R)-lutein (2) [13] having the same configuration was observed.  Finally, Horner-Wadsworth-Emmons reaction of apocarotenal 8 with phosphonate 25 in the presence of sodium bis(trimethylsilyl)amide [NaN(TMS) 2 ] followed by desilylation with TBAF afforded loroxanthin (1) stereoselectively in 61% yield in two steps. Total yield of 1 from alkenyl stananne 9 including isomerization recovery of apocarotenal 8 was 15% over 9 steps and that from (3R,6R)-aldehyde was 39% over 7 steps. Its 1 H-NMR spectral data were in good agreement with the reported data [8]. While the CD spectrum of the natural product was reported to be non-conservative with a weak positive Cotton effect around 250 nm [8], that of the synthetic product showed a relatively clear curve as shown in Figure 3. Good similarity with the reported spectrum of (3R,3 R,6 R)-lutein (2) [13] having the same configuration was observed.  In summary, the first total synthesis of loroxanthin (1) was accomplished via stereoselective condensation of C25-apocarotenal 8 with C15-phosphonate 25. Preparation of apocarotenal 8 was achieved via Stille coupling reaction of alkenyl iodide 10 with alkenyl stananne 9, whereas phosphonate 25 was prepared through treatment of ally alcohol 23 with triethyl phosphite and ZnI2. The phosphonate 25 would be a versatile building block for carotenoids with 3-hydroxy-ε-end group such as lutein (3) and siphonaxanthin (4). Recent our achievement of the total synthesis of 19-deoxysiphonaxanthin, siphonaxanthin biosynthetic precursor, also proves usefulness of phosphonate 25 [22]. This method will provide a material needed for investigating the biological functions of 1.

General
UV-VIS spectra were recorded on a JASCO V-650 instrument (JASCO, Tokyo, Japan), with ethanol solutions. IR spectra were measured on a Perkin-Elmer spectrum 100 FT-IR spectrometer (Perkin-Elmer, Yokohama, Japan), with chloroform solutions. 1 H-and 13 C-NMR spectra were determined on a Varian Gemini-300 or a Varian NMR System AS 500 In summary, the first total synthesis of loroxanthin (1) was accomplished via stereoselective condensation of C 25 -apocarotenal 8 with C 15 -phosphonate 25. Preparation of apocarotenal 8 was achieved via Stille coupling reaction of alkenyl iodide 10 with alkenyl stananne 9, whereas phosphonate 25 was prepared through treatment of ally alcohol 23 with triethyl phosphite and ZnI 2 . The phosphonate 25 would be a versatile building block for carotenoids with 3-hydroxy-ε-end group such as lutein (3) and siphonaxanthin (4). Recent our achievement of the total synthesis of 19-deoxysiphonaxanthin, siphonaxanthin biosynthetic precursor, also proves usefulness of phosphonate 25 [22]. This method will provide a material needed for investigating the biological functions of 1.

General
UV-VIS spectra were recorded on a JASCO V-650 instrument (JASCO, Tokyo, Japan), with ethanol solutions. IR spectra were measured on a Perkin-Elmer spectrum 100 FT-IR spectrometer (Perkin-Elmer, Yokohama, Japan), with chloroform solutions. 1 H-and 13 C-NMR spectra were determined on a Varian Gemini-300 or a Varian NMR System AS 500 superconducting FT-NMR spectrometer (Varian Inc., Palo Alto, CA, USA), with CDCl 3 solutions. The chemical sifts are expressed in ppm relative to tetramethylsilane (TMS) (δ = 0) as internal standard for 1 H-NMR and CDCl 3 (δ = 77.0) for 13 C-NMR. J Values are given in Hz. Mass spectra were taken on a Thermo Fisher Scientific Exactive spectrometer (Thermo Fisher Scientific, Bremen, Germany).
Flash column chromatography (CC) was performed on using Kanto Silica Gel 60 N. Preparative HPLC was carried out on a Shimadzu LC-6A with a UV-VIS detector (Shimadzu, Kyoto, Japan). Preparative HPLC was carried out on a JASCO PU-2080 with a UV-VIS detector (JASCO, Tokyo, Japan). HPLC analyses were performed on Shimadzu-LC-20AT instrument (Shimadzu, Kyoto, Japan) with a photodiode arrey detetor (GL-Sciences, Tokyo, Japan).
All operations were carried out under nitrogen or argon. Evaporation of the extract or the filtrate was carried out under reduced pressure. In solvent extraction procedure, organic layer was dried over anhydrous Na 2 SO 4 . Ether refers to diethyl ether, and hexane to n-hexane. NMR assignments are given using the carotenoid numbering system.