Formation of Tetrahydrofurano-, Aryltetralin, and Butyrolactone Norlignans through the Epoxidation of 9-Norlignans.

Epoxidation of the C=C double bond in unsaturated norlignans derived from hydroxymatairesinol was studied. The intermediate epoxides were formed in up to quantitative conversions and were readily further transformed into tetrahydrofuran, aryltetralin, and butyrolactone products-in diastereomeric mixtures-through ring-closing reactions and intramolecular couplings. For epoxidation, the classical Prilezhaev reaction, using stoichiometric amounts of meta-chloroperbenzoic acid (mCPBA), was used. As an alternative method, a catalytic system using dimeric molybdenum-complexes [MoO2L]2 with ONO- or ONS-tridentate Schiff base ligands and aqueous tert-butyl hydroperoxide (TBHP) as oxidant was used on the same substrates. Although the epoxidation was quantitative when using the Mo-catalysts, the higher temperatures led to more side-products and lower yields. Kinetic studies were also performed on the Mo-catalyzed reactions.


Introduction
In 1936, Haworth introduced lignans as a group of plant-based compounds consisting of β-β -linked dimeric phenylpropane units ( Figure 1) [1]. Similar to other natural phenolics such as stilbenes and flavonoids, lignans are formed through the shikimic acid pathway [2]. Lignans have attracted great interest in research as they have a number of health benefits, such as being strong antioxidants, lowering the risk of coronary heart disease, and having neuroprotective effects and anticancer properties [3][4][5][6]. Lignans are also found in foods, for example, flaxseed, sesame, chickpeas and cereals (rye, oat, barley), and red wine contains high levels of lignans. [7]

Introduction
In 1936, Haworth introduced lignans as a group of plant-based compounds consisting of β-β´-linked dimeric phenylpropane units ( Figure 1) [1]. Similar to other natural phenolics such as stilbenes and flavonoids, lignans are formed through the shikimic acid pathway [2]. Lignans have attracted great interest in research as they have a number of health benefits, such as being strong antioxidants, lowering the risk of coronary heart disease, and having neuroprotective effects and anticancer properties [3][4][5][6]. Lignans are also found in foods, for example, flaxseed, sesame, chickpeas and cereals (rye, oat, barley), and red wine contains high levels of lignans. [7]   A subclass of lignans called norlignans are dimers composed of a phenylpropane unit and a phenylethane unit coupled through a β-β linkage ( Figure 1). The prefix "nor" indicates the lack of one or more carbon in the parent skeleton, and the term 9-norlignan is used when the lignan structure lacks C-9 (or γ carbon). As lignans, norlignans are also found in plants, having a range of biological activity [8,9].
The lignan 7-hydroxymatairesinol (HMR) can be isolated in large-scale from the knotwood of Norway spruce (Picea abies) as it constitutes up to 24% of the dry weight [10]. HMR can be used as raw material for the semi-synthesis of a range of different lignans, among others are a family of norlignans belonging to the imperanene family [9]. Some of these semi-synthetic and unsaturated 9-norlignans were used as substrates in this study ( Figure 2, substrates 1-4). The substrates were chosen for two reasons. Firstly, they were chosen to study the selectivity in the oxidation reactions (epoxidation) with substrates containing free phenolics (1 and 3) compared to protected (methoxylated) derivatives (2 and 4). Secondly, we wanted to study the outcome of the epoxidation reaction in the presence of a primary hydroxyl (1)(2), carboxylic acid (3) and methyl ester (4). A subclass of lignans called norlignans are dimers composed of a phenylpropane unit and a phenylethane unit coupled through a β-β´ linkage ( Figure 1). The prefix "nor" indicates the lack of one or more carbon in the parent skeleton, and the term 9-norlignan is used when the lignan structure lacks C-9 (or γ carbon). As lignans, norlignans are also found in plants, having a range of biological activity [8,9]. The lignan 7-hydroxymatairesinol (HMR) can be isolated in large-scale from the knotwood of Norway spruce (Picea abies) as it constitutes up to 24% of the dry weight [10]. HMR can be used as raw material for the semi-synthesis of a range of different lignans, among others are a family of norlignans belonging to the imperanene family [9]. Some of these semi-synthetic and unsaturated 9-norlignans were used as substrates in this study ( Figure 2, substrates 1-4). The substrates were chosen for two reasons. Firstly, they were chosen to study the selectivity in the oxidation reactions (epoxidation) with substrates containing free phenolics (1 and 3) compared to protected (methoxylated) derivatives (2 and 4). Secondly, we wanted to study the outcome of the epoxidation reaction in the presence of a primary hydroxyl (1-2), carboxylic acid (3) and methyl ester (4). Epoxidation reactions are undoubtedly important in organic chemistry and several methods have been developed to perform these reactions [11][12][13][14]. However, to our knowledge, no epoxidation reactions have previously been reported for substrates 1-4. As part of our work on oxidative transformation of lignans and norlignans, we wanted to investigate how these substrates react upon epoxidation. We chose two different methods for this study. First, the stoichiometric mCPBA-mediated epoxidation method known as the Prilezhaev reaction was studied. Previously, mCPBA has been used for Baeyer-Villiger oxidation of lignans containing ketone or aldehyde groups, for epoxidation of lignan structures containing double bonds, and for epoxidation of dihydro-naphthalene model compounds having a norlignan-type skeleton [15,16]. Because of the simplicity of the Prilezhaev reaction and the average high product yields for the reaction, mCPBA was a good epoxidation agent for the purpose of our study.
As we wanted to study more environmentally friendly alternatives for epoxidation, our attention turned to catalytic reactions using H2O2, tert-butyl hydroperoxide (TBHP), or even O2 as oxidants. Since the 1960s, numerous complexes of molybdenum have been used as catalysts for efficient epoxidation of olefins together with hydroperoxides [17]. More recently, scientists have further developed molybdenum complexes for this purpose [18]. Some complexes have attracted our interest, i.e., [MoO2L] complexes (L being a tridentate Schiff base ligand) since their synthesis is quite easy [19,20]. These complexes have a long shelf life and have been shown to exhibit interesting reactivity under green conditions. For example, epoxidations of cyclooctene and cyclohexene [19][20][21][22] and the natural substrate limonene has been demonstrated [23]. It has also been shown that epoxidation of sesquiterpenes could be achieved using TBHP as oxidant, and the reactivity was compared to mCPBA [24]. As the substrates previously studied for [MoO2L-TBHP] epoxidations were fairly simple and without functional groups, it was interesting to investigate how this Mo-catalyzed epoxidation could work on the more complex substrates 1-4. Epoxidation reactions are undoubtedly important in organic chemistry and several methods have been developed to perform these reactions [11][12][13][14]. However, to our knowledge, no epoxidation reactions have previously been reported for substrates 1-4. As part of our work on oxidative transformation of lignans and norlignans, we wanted to investigate how these substrates react upon epoxidation. We chose two different methods for this study. First, the stoichiometric mCPBA-mediated epoxidation method known as the Prilezhaev reaction was studied. Previously, mCPBA has been used for Baeyer-Villiger oxidation of lignans containing ketone or aldehyde groups, for epoxidation of lignan structures containing double bonds, and for epoxidation of dihydro-naphthalene model compounds having a norlignan-type skeleton [15,16]. Because of the simplicity of the Prilezhaev reaction and the average high product yields for the reaction, mCPBA was a good epoxidation agent for the purpose of our study.
As we wanted to study more environmentally friendly alternatives for epoxidation, our attention turned to catalytic reactions using H 2 O 2 , tert-butyl hydroperoxide (TBHP), or even O 2 as oxidants. Since the 1960s, numerous complexes of molybdenum have been used as catalysts for efficient epoxidation of olefins together with hydroperoxides [17]. More recently, scientists have further developed molybdenum complexes for this purpose [18]. Some complexes have attracted our interest, i.e., [MoO 2 L] complexes (L being a tridentate Schiff base ligand) since their synthesis is quite easy [19,20]. These complexes have a long shelf life and have been shown to exhibit interesting reactivity under green conditions. For example, epoxidations of cyclooctene and cyclohexene [19][20][21][22] and the natural substrate limonene has been demonstrated [23]. It has also been shown that epoxidation of sesquiterpenes could be achieved using TBHP as oxidant, and the reactivity was compared to mCPBA [24]. As the substrates previously studied for [MoO 2 L-TBHP] epoxidations were fairly simple and without functional groups, it was interesting to investigate how this Mo-catalyzed epoxidation could work on the more complex substrates 1-4.
The mechanism for the [MoO 2 L-TBHP] epoxidations can be explained through a Bartlett-like mechanism, implying the formation of a pentacoordinated species [MoO 2 L] followed by the coordination of TBHP (Scheme 1). The [MoO 2 L-TBHP] adduct was responsible of the oxygen transfer from the oxidant to the olefin. [19] Molecules 2019, 24, x FOR PEER REVIEW

of 14
The mechanism for the [MoO2L-TBHP] epoxidations can be explained through a Bartlett-like mechanism, implying the formation of a pentacoordinated species [MoO2L] followed by the coordination of TBHP (Scheme 1). The [MoO2L-TBHP] adduct was responsible of the oxygen transfer from the oxidant to the olefin. [19]  species is coordinated to TBHP followed by Bartlett-type epoxidation of olefins [19].
The aim of this study was to evaluate the traditional mCPBA-reaction and compare it to the novel Mo-catalytic version for epoxidation of the presented substrates. Furthermore, the subsequent reactions of the epoxy-intermediates to form different norlignan structures were the target of investigation.

Results and Discussion
All norlignans reacted at the double bond forming the epoxide-intermediate as two diastereomers. The formed epoxides were subject to rapid intramolecular nucleophilic attack to give the final products (Scheme 2, products A and B). Since both stereoisomers of the epoxides were formed, and the subsequent nucleophilic attacks were not stereospecific, four possible stereoisomers were formed. However, some diastereomers were formed only in trace amounts.
The aim of this study was to evaluate the traditional mCPBA-reaction and compare it to the novel Mo-catalytic version for epoxidation of the presented substrates. Furthermore, the subsequent reactions of the epoxy-intermediates to form different norlignan structures were the target of investigation.

Results and Discussion
All norlignans reacted at the double bond forming the epoxide-intermediate as two diastereomers. The formed epoxides were subject to rapid intramolecular nucleophilic attack to give the final products (Scheme 2, products A and B). Since both stereoisomers of the epoxides were formed, and the subsequent nucleophilic attacks were not stereospecific, four possible stereoisomers were formed. However, some diastereomers were formed only in trace amounts.
In addition, during the transformations product A was probably partially re-opened to the cationic intermediate and eventually transformed into the thermodynamic product B by the nonreversible Friedel-Craft type ring closure. In fact, we expected product A to be quantitatively converted into product B under acidic conditions, as similar pH dependent C-O bond openings and product transformations have been reported for other lignans [25]. During the reactions we could not clearly detect the transformation of A to B but upon treatment with excess acid the conversion was quantitative. In addition, during the transformations product A was probably partially re-opened to the cationic intermediate and eventually transformed into the thermodynamic product B by the nonreversible Friedel-Craft type ring closure. In fact, we expected product A to be quantitatively converted into product B under acidic conditions, as similar pH dependent C-O bond openings and product transformations have been reported for other lignans [25]. During the reactions we could not clearly detect the transformation of A to B but upon treatment with excess acid the conversion was quantitative.
In Table 1, the yields for products and side products are shown for each reaction.

m-CPBA-Mediated Epoxidation
The Prilezhaev reaction at room temperature gave quantitative conversions within 30 minutes for all substrates. No epoxides were detected on GC-MS nor NMR, as they rapidly reacted to product A and B (and a mixture of minor products).
The reaction worked well with free phenolic groups as 1 had a quantitative conversion to products 1A (isolated yield of 60% of four isomers) and 1B (isolated yield of 12% of two isomers) within 30 minutes at room temperature. Substrate 2 with protected phenolics (as methoxyl groups) also gave a quantitative conversion within 20 minutes at room temperature to products 2A (42% isolated yield as four isomers) and 2B (46% isolated yield as four isomers).
The Prilezhaev reaction gave 96% conversion of 3 after 30 minutes at room temperature to a reaction mixture with a product distribution, according to NMR, of 71% 3A (total isolated yield for the four isomers was 51%) and 24% 3B (total isolated yield for two isomers was 19%). Furthermore, the isolated product 3A could quantitatively be converted to 3B by stirring in trifluoroacetic acid (TFA) at room temperature for 20 minutes. Thus, a final isolated yield of 70% of 3B could be reached.
The Prilezhaev reaction of 4 using one equivalent mCPBA for 30 minutes resulted in an 83% conversion. Two products were observed, product 4B and 4D, (a 7-mCBA-coupled adduct, Figure 3 In Table 1, the yields for products and side products are shown for each reaction.

m-CPBA-Mediated Epoxidation
The Prilezhaev reaction at room temperature gave quantitative conversions within 30 min for all substrates. No epoxides were detected on GC-MS nor NMR, as they rapidly reacted to product A and B (and a mixture of minor products).
The reaction worked well with free phenolic groups as 1 had a quantitative conversion to products 1A (isolated yield of 60% of four isomers) and 1B (isolated yield of 12% of two isomers) within 30 min at room temperature. Substrate 2 with protected phenolics (as methoxyl groups) also gave a quantitative conversion within 20 min at room temperature to products 2A (42% isolated yield as four isomers) and 2B (46% isolated yield as four isomers).
The Prilezhaev reaction gave 96% conversion of 3 after 30 min at room temperature to a reaction mixture with a product distribution, according to NMR, of 71% 3A (total isolated yield for the four isomers was 51%) and 24% 3B (total isolated yield for two isomers was 19%). Furthermore, the isolated product 3A could quantitatively be converted to 3B by stirring in trifluoroacetic acid (TFA) at room temperature for 20 min. Thus, a final isolated yield of 70% of 3B could be reached.
The Prilezhaev reaction of 4 using one equivalent mCPBA for 30 min resulted in an 83% conversion. Two products were observed, product 4B and 4D, (a 7-mCBA-coupled adduct, Figure 3) in a 4:6 ratio. Substrate 4 could not react by pathway A, and the nucleophilic attack by mCBA was competing with the slower pathway B. This type of reaction has been reported previously [26]. The byproduct 4D could, however, be quantitatively hydrolyzed to 4B by stirring in TFA at room temperature for 20 min. in a 4:6 ratio. Substrate 4 could not react by pathway A, and the nucleophilic attack by mCBA was competing with the slower pathway B. This type of reaction has been reported previously [26]. The byproduct 4D could, however, be quantitatively hydrolyzed to 4B by stirring in TFA at room temperature for 20 minutes.

Mo-Catalyzed Epoxidation with TBHP
Most substrates were quantitatively converted by the Mo-catalysts. However, the higher temperatures resulted in partial water elimination of product B to a new unsaturated structure as seen for product C in Scheme 2. This side product was identified and quantified using GC-MS and NMR but was not further isolated and characterized as only small amounts were formed (≤7%). Additionally, some unidentified oligomeric side products were observed. The oligomers were likely

Mo-Catalyzed Epoxidation with TBHP
Most substrates were quantitatively converted by the Mo-catalysts. However, the higher temperatures resulted in partial water elimination of product B to a new unsaturated structure as seen for product C in Scheme 2. This side product was identified and quantified using GC-MS and NMR but was not further isolated and characterized as only small amounts were formed (≤7%). Additionally, some unidentified oligomeric side products were observed. The oligomers were likely formed through intermolecular reactions of the epoxide intermediates, as the double bond seemed to be reacting to products containing alcohol and ether functions. Analyses using high pressure size exclusion chromatography (HPSEC) revealed that the oligomers consisted mostly of di-and trimers. At 80 • C, traces of benzaldehydes were also formed through oxidative cleavage of double bond. Due to these various side reactions, the yield of products A and B were moderate, although the conversions to epoxides were quantitative (Table 1).
For the Mo-catalyzed reactions, no or very slow reactions were observed with less than 5 equivalents of TBHP or with reaction temperatures lower than 80 • C, which restricted the optimization of this reaction by lowering the reaction temperature. As a consequence, lower yields of the final products were obtained using MoO 2 L complexes performed at 80 • C (20%-40% formation of products A and B according to NMR of the reaction mixtures) compared to reactions with mCPBA operated at room temperature (>90% formation of products A and B).
The resulting reaction mixtures for each substrate were similar in the processes using both catalysts [MoO 2 (SAP)] 2 and [MoO 2 (SATP)] 2 . However, the reaction rates were higher with [MoO 2 (SATP)] 2 . This observation was in agreement with previous kinetic studies with both catalysts for epoxidation of cyclooctene, under solvent-free conditions with TBHP at 80 • C [20]. Substrate 1 reacted to identical reaction mixtures with both Mo-catalysts (Entries 1 and 2). As a side note, analyses of these reactions were done at one and three hours, and quantitative conversion for the [MoO 2 (SATP)] 2 -catalyzed reaction (Entry 3) was reached in a shorter time compared to the [MoO 2 (SAP)] 2 -catalyzed reaction (Entry 2).
In the absence of Mo-catalyst, even a large excess (10 equivalents) of TBHP did not react with substrate 1, exhibiting the role of the MoO 2 L complexes as catalyst.
To further investigate this, reaction kinetic studies were performed with substrate 2 (14.5 mM) in toluene, using 0.5 mol% dimeric Mo(VI)-precatalyst (i.e. 1% mol Mo) and 5 equivalents of TBHP (70 wt% in water) at 80 • C. The reaction was followed by GC-MS using dimethyl-dihydroimperanene as internal standard (Figure 4). Samples were taken at specific time intervals and dried in a vacuum. Derivatization of the samples was done using tetramethyldisilazane and chlorotrimethylsilane in pyridine prior to GC-MS analysis [27]. The results showed that the reaction catalyzed by [MoO 2 (SATP)] 2 was faster (as mentioned previously), with a quantitative conversion reached after 3 h 40 min ( Figure 5). For comparison, at 3 h 40 min, the [MoO 2 (SAP)] 2 -catalyzed reaction showed a conversion of less than 50% and, calculated from the trend of the reaction rate, a quantitative conversion was reached after approximately 9 h 30 min. The rate was 2.6 times slower than the [MoO 2 (SATP)] 2 -catalyzed reaction. For both catalysts, the final reaction mixture gave products 2A, 2B and 2C in roughly 3:5:1 ratio. Quantification using GC analyses showed a total conversion to products 2A, 2B and 2C at around 95%, for both reactions. Quantification using NMR analysis, however, indicated that the total yield (2A, 2B and 2C) was around 70% (calculated from the ratio of integrals in the 1 H spectrum). This difference may be explained by the fact that signal intensity in the total ion chromatogram is not fully quantitative, due to a different degree of ionization for the products. As each product was not quantitatively calibrated on GC, quantification using NMR seemed to be more reliable.
to a different degree of ionization for the products. As each product was not quantitatively calibrated on GC, quantification using NMR seemed to be more reliable.
Furthermore, the NMR spectra showed that the ratio of products 2A, 2B and 2C was roughly 6:2:1. The variation in product distribution between GC-MS and NMR may be partially explained by transformation of 2A to 2B during the derivatization step or upon injection to the GC-MS. Those types of transformations have been reported for other lignan structures as well [25].   to a different degree of ionization for the products. As each product was not quantitatively calibrated on GC, quantification using NMR seemed to be more reliable.
Furthermore, the NMR spectra showed that the ratio of products 2A, 2B and 2C was roughly 6:2:1. The variation in product distribution between GC-MS and NMR may be partially explained by transformation of 2A to 2B during the derivatization step or upon injection to the GC-MS. Those types of transformations have been reported for other lignan structures as well [25].   Furthermore, the NMR spectra showed that the ratio of products 2A, 2B and 2C was roughly 6:2:1. The variation in product distribution between GC-MS and NMR may be partially explained by transformation of 2A to 2B during the derivatization step or upon injection to the GC-MS. Those types of transformations have been reported for other lignan structures as well [25].
For substrate 3 (containing OH functions), the Mo-catalyzed reactions gave a mixture of monoand oligomeric products. Only traces of the expected ring closed products (3A and 3B) were detected using NMR and GC-MS. NMR analyses of the reaction mixtures showed that the double bond had reacted. Here, again, the high temperature in these reactions may result in unselective reactions of the epoxide intermediate. With lower equivalents of TBHP, no conversion occurred even with higher equivalents of pre-catalysts (Entries 8 and 10).
Substrate 4 reacted slower compared to the other norlignans. NMR showed the conversion to one major product (4B) in 14% or 19% for [MoO 2 (SAP)] 2 or [MoO 2 (SATP)] 2 respectively, as well as other mono-(4C, 4D and other unknown products) and oligomers. Analysis using HPSEC gave a rough estimation of 40% monomeric structures, 35% di-and trimers, and 25% small molecular structures including molybdenum complexes ( Figure 6). GC-MS revealed that the main degradation product was 3,4-dimethoxybenzaldehyde issued from the C=C oxidative cleavage. Substrate 4 reacted slower compared to the other norlignans. NMR showed the conversion to one major product (4B) in 14% or 19% for [MoO2(SAP)]2 or [MoO2(SATP)]2 respectively, as well as other mono-(4C, 4D and other unknown products) and oligomers. Analysis using HPSEC gave a rough estimation of 40% monomeric structures, 35% di-and trimers, and 25% small molecular structures including molybdenum complexes ( Figure 6). GC-MS revealed that the main degradation product was 3,4-dimethoxybenzaldehyde issued from the C=C oxidative cleavage.

Conclusion
The results suggest that all studied norlignans reacted in up to quantitative conversion to epoxide intermediates using mCPBA or by Mo-catalyzed reactions with TBHP as oxidant. However, the intermediate products rapidly reacted further through nucleophilic attack at the epoxides forming tetrahydrofuran or lactone structures (A) or aryltetralin (B) structures. The main products were formed by route A (Product A in Scheme 2), but the arylic ring closure at C-6´-position (Product B in Scheme 2) was a competing reaction. Product A could also be further transformed to B by treatment in acidic conditions. As a result of high temperatures, longer reaction times, and higher equivalents of peroxide, the Mo-catalyzed reactions also gave other mono-and oligomeric (di-and trimeric) side products resulting in lower yields of defined products (A and B). The results, however, showed that these Mo-catalysts can successfully be used for epoxidation of more complex substrates than previously reported, as long as the products are stable at those reaction conditions. Once optimized, the advantage of the MoO2L/TBHP procedures could lie in the fact that less post-reaction procedures are needed, interesting in terms of mass efficiency, and that TBHP is lighter than m-CPBA, interesting in terms of atom economy.

Conclusions
The results suggest that all studied norlignans reacted in up to quantitative conversion to epoxide intermediates using mCPBA or by Mo-catalyzed reactions with TBHP as oxidant. However, the intermediate products rapidly reacted further through nucleophilic attack at the epoxides forming tetrahydrofuran or lactone structures (A) or aryltetralin (B) structures. The main products were formed by route A (Product A in Scheme 2), but the arylic ring closure at C-6 -position (Product B in Scheme 2) was a competing reaction. Product A could also be further transformed to B by treatment in acidic conditions. As a result of high temperatures, longer reaction times, and higher equivalents of peroxide, the Mo-catalyzed reactions also gave other mono-and oligomeric (di-and trimeric) side products resulting in lower yields of defined products (A and B). The results, however, showed that these Mo-catalysts can successfully be used for epoxidation of more complex substrates than previously reported, as long as the products are stable at those reaction conditions. Once optimized, the advantage of the MoO 2 L/TBHP procedures could lie in the fact that less post-reaction procedures are needed, interesting in terms of mass efficiency, and that TBHP is lighter than m-CPBA, interesting in terms of atom economy.

Prilezhaev Reaction (mCPBA)
In a typical reaction, the substrate (1.0 equivalent) was dissolved in DCM (50 mM lignan concentration). Under stirring, mCPBA (1.0 equivalent) was added and the reaction continued at room temperature for 20-30 min. Aqueous saturated NaHCO 3 was added and the organic phase was separated. The aqueous phase was extracted with EtOAc. The combined organic phases were dried using Na 2 SO 4 , filtered, and concentrated in a vacuum. The isolated products were retrieved through purification using column chromatography with silica gel 60 as stationary phase and chloroform and methanol as eluents.

Mo-Catalyzed Epoxidation by TBHP
In a typical reaction, the substrate (1.0 equivalent) and molybdenum-catalyst (0.5 mol% of dimer) was dissolved in toluene (50 mM substrate concentration). The solution was heated to 80 • C, followed by dropwise addition of aqueous TBHP (70 wt%, 5 equivalents). The reaction mixture was stirred at 80 • C in an open air atmosphere. After an appropriate time (1-20 h) the reaction mixture was quenched by addition of excess solid MnO 2 , cooled to room temperature, and a dark precipitate (Mo-complex + MnO 2 ) was filtered off. The crude mixture was concentrated in a vacuum. The isolated products were retrieved through purification using column chromatography with silica gel 60 as stationary phase and chloroform and methanol as eluents.

Analytical Methods
HRMS of the negatively charged deprotonated products (positively charged Na-adducts of 4B) were recorded on a Bruker Daltonics micrOTOF-Q instrument using an electrospray ionization/quadrupole/time-of-flight systems. 1 H and 13 C-NMR spectra were recorded on a Bruker AV 500 spectrometer at 500 and 125 MHz, respectively. 2D-experiments were recorded using standard pulse sequences, and the chemical shifts are reported downfield from tetramethylsilane. GC-EIMS analyses were performed on an Agilent Technologies 7890A GC-system equipped with a 5975C EIMS-detector and an Agilent J&W HP-5ms GC Column (30 m × 0.25 mm, 0.25 µm film) (Agilent Technologies Inc., Santa Clara, CA, USA).
HP-SEC analyses were performed on an Agilent 1100 Series HPLC instrument equipped with a G1315B DAD-detector, 2 × Jordi Gel DVB 500A (300 mm × 7.8 mm) columns (Columnex LLC, New York, NY, USA; 40 • C), and a 50 mm × 7.8 mm guard column. One percent of AcOH in THF served as eluent at a flow rate of 0.8 mL/min.

Isolation of Products
The products were isolated from the reaction mixtures using a Teledyne ISCO combiFlash ® EZ Prep UV/ELSD-unit (Teledyne ISCO, Lincoln, NE, USA), with RediSep Rf gold silica flash chromatography columns (20-40 µm). A chloroform/methanol eluent system was used, starting from 0 vol% methanol and gradually over 45 min increasing the methanol ratio to 15 vol%.

Experimental Data
The experimental data for the major isomers of all products are listed here. The various diastereomers of each product were not separated through column chromatography, and only the major diastereomers could be characterized using NMR (see Supplementary Materials). Furthermore, the methoxyl-signals in 1 H-NMR spectra were overlapped with the signals for H-8 in 1A and 2A and 7-H and 8-H in 1B, 2B, 3B, and 4B, and the stereochemistry could therefore not be solved for these products. On the other hand, the stereochemistry for all isomers of 3A was solved.