Breakdown of chlorophyll: A nonenzymatic reaction accounts for the formation of the colorless “nonfluorescent” chlorophyll catabolites

Materials. Reagents used were reagent-grade commercial, and solvents were distilled before use. Reagents and HPLC solvents were from Fluka and Merck. Sep-Pak-C18 cartridges were from Waters.

Chromatography. Flash chromatography was done on silica gel 60 from Fluka. TLC was done on plates (0.25 mm) with silica gel 60 F₂₅₄ from Merck.

HPLC. For analytical work, we used a Hypersil (Cheshire, U.K.) ODS (5 μm, 250 × 4.6 mm i.d.) as a column and a Gynkotek (Germering, Germany) M480G pump with vacuum on-line degasser. For preparative work, we used a Hypersil ODS (5 μm, 250 × 21.2 mm i.d.) and a Gynkotek M300 pump. Solvents were degassed by sonication. Detectors used were a Gynkotek diode array detector UVD 340 and a Hitachi SPD-6AV UV/visible detector. All analytical chromatograms were taken at 5°C, all preparative chromatograms were taken at room temperature, and data were processed by a Gynkotek HPLC data system and GYNKOSOFT 5.50.

Spectroscopy. To obtain UV/visible spectra we used a Hitachi-U3000 spectrophotometer at λ_max(nm)/(log ε). Concentrations were calculated following Lambert-Beer's law. To obtain CD spectra we used a Jasco-J715 spectra-polarimeter at λ_max(nm) and λ_min(nm)/(Δε). To obtain NMR spectra we used a Varian Unity_plus 500 [δ(C¹HD₂OD) = 3.31 ppm and δ(¹³CD₃OD) = 49.0 ppm]. For MS we used a Finnigan MAT 95-S in (+)-ion mode. For fast atom bombardment–MS we used a cesium gun at 20 keV and a glycerol matrix. Electrospray ionization–MS was done at 3.2 kV with water/MeOH mixtures as solvents.

Senescent Leaves from C. japonicum Sieb and Zucc. Yellow leaves from C. japonicum trees were collected in late September in the botanical institutes of the University of Zürich and the University of Innsbruck and were extracted as described by Curty and Engel (18).

Preparation of Ca-FCC-2 (2). A sample of ≈70 nmol of Ca-FCC-2 (2) in 10-ml solution (MeOH/water, 30:70, vol/vol) was prepared with extracts from ripe sweet pepper (Capsicum annuum) and isolated as described by Mühlecker et al. (16). The FCC 2 was stored frozen (at —80°C) for ≈20 days before being used for the experiments described here.

Isolation of Cj-NCC-2 (1). A crude solid extract (5.8 g) was obtained essentially as described (18) from 5.25 kg (fresh weight) of yellow (senescent) leaves of C. japonicum. A total of 4.8 g of this solid extract, which contained impure Cj-NCC-1 (3) (18) and Cj-NCC-2 (1) in a ratio of ≈23:1, was subjected to flash chromatography. A sample of ≈1 g of a yellow powder was obtained, which contained still impure Cj-NCC-1 (3) and Cj-NCC-2 (1) in a ratio of ≈7:1. Flash chromatography (silica gel; eluent/dichloromethane/MeOH/water 800:200:1.5, vol/vol) of this powder yielded a fraction containing 26.1 mg of a crude sample of Cj-NCC-2 (1), free of 3 (and other fractions containing both 1 and 3 in varying amounts). A homogeneous fraction containing 1 was collected (16.5 min) from 3.6 mg of the crude 1, subjected to preparative HPLC [10 ml/min (200 mM) ammonium acetate in MeOH/water 65:35 vol/vol (pH 6.8)], and desalted with a Sep-Pak cartridge. After the solvent was evaporated in vacuo at <0°C, 0.88 mg (1.4 μmol) of pure Cj-NCC-2 (1) was obtained (see Supporting Text, which is published as supporting information on the PNAS web site, www.pnas.org, for details).

Characterization of Cj-NCC-2 (1). The sample of 0.86 mg of authentic 1 was characterized as follows: MS (nano-electrospray ionization) m/z (%), 631.2 (18), 630.2 (57), 629.2 (100, [M+H]⁺), 597.2 (26, [M+H-32]⁺), 506.2 (12, [M+H-123]⁺); high-resolution MS, experimental 629.2990, calculated 629.2975. UV/visible, (MeOH) λ_max (log ε) 239 (sh, 4.34), 312 (4.25); CD (MeOH) λ_min/max (Δε) 225 (24.0), 251 (—8.1), 258 (—7.5), 283 (—17.9), 317 (6.5) (see Fig. 2); ¹H NMR (500 MHz, ≈2.3 mM in CD₃OD, 26°C, tetramethylsilane) δ = 0.99 (H₃C-8²), 1.93 (H₃C-18¹), 1.94 (H₃C-2¹), 2.08 (H₃C-12¹), 2.24 (H₃C-7¹), 2.30–2.36 (H₂C-17²), 2.42 (H₂C-8¹), 2.50 (H_AC-20), 2.60–2.75 (H₂C-17¹), 2.84 (H_BC-20), 3.75 (H₃C-13⁵), 3,89/3.93 (H₂C-10), 4.00 (HC-1), 5.34 (H_cisC-3²), 6.09 (H_transC-3²), 6.43 (HC-3¹), 9.33 (HC-5), 9.51 (NH); ¹³C-NMR, (125 MHz, CD₃OD, 26°C, tetramethylsilane) from heteronuclear single quantum correlation/heteronuclear multiple bond correlation spectra δ = 8.7 (7¹), 9.0 (12¹), 9.2 (18¹), 12.5 (2¹), 15.1 (8²), 17.5 (8¹), 21.7 (17¹), 23.5 (10), 30.2 (20), 37.1 (15), 39.2 (17²), 52.6 (13⁵), 61.8 (1), 67.7 (13²), 112.1 (12), 115.4 (18), 118.9 (3²), 120.5 (17), 124.2 (19), 124.3 (16), 126.0 (8), 125.8 (13), 126.9 (3¹), 128.7 (3), 129.1 (6), 133.9 (11), 134.4 (7), 137.7 (9), 156.6 (2), 161.4 (14), 171.6 (13³), 174.9 (4), 177.4 (5), 180.7 (17³), 197.7 (13¹) (Figs. 2 and 4; and see Figs. 6–8, which are published as supporting information on the PNAS web site).

Preparation of the NCC 5a by Tautomerization of Ca-FCC-2 (2). A sample of 6 ml of the solution of Ca-FCC-2 (2) (16) [containing ≈42 nmol (26 μg) of 2] was first degassed, then acidified to pH 4.9 by addition of 0.6 ml of degassed buffer (1 M KH₂PO₄/K₂HPO₄) under an Ar atmosphere (pH control by pH electrode). The oxygen-free reaction mixture was stored at ambient temperature with protection from light for 20.5 h. The reaction mixture was then desalted by filtration through a Sep-Pak cartridge, followed by elution with methanol. The solution of the crude product in methanol exhibited the spectrum of an NCC, as analyzed by UV/visible spectroscopy, and was indicated to contain ≈46 nmol [29 μg, as calculated with log ε_312nm (1) = 4.25] of 5a. The crude product obtained from this experiment, as well as the one of a second batch, in which 3 ml of the original solution of 2 was converted under the same conditions [giving 19 nmol (12 μg) of 5a], were purified with preparative HPLC [10 ml/min of 200 mM ammonium acetate in MeOH/water, 65:35 vol/vol (pH 6.8)]. The homogeneous fractions containing 5a (R_t = 21.9 min) were combined and desalted. The sample contained 38 nmol (24 μg) of the NCC 5a, according to its UV/visible absorbance spectrum.

Qualitative Kinetic Analysis of the Formation of the NCC 5a from the FCC 2. The preparative isomerization reaction was monitored by recording the UV/visible spectrum of the reaction mixture at 0, 15, 30, 90, 110, 270, and 1,230 min, and by analytical HPLC of samples of the reaction mixture collected at 0, 15, 110, 270, and 1,230 min (see Fig. 3). The observed time dependence of the relative amounts of 2, 5b, and 5a was analyzed based on a model of two consecutive first-order reactions (19), providing the estimated rates (at pH 4.9) of k₁ = 0.025 min^—1 for the isomerization of 2 to 5b and k₂ = 0.0055 min^—1 for the epimerization of 5b to 5a (for details see Fig. 9, which is published as supporting information on the PNAS web site).

Spectroscopic Characterization of the NCC 5a and Identification with Authentic Cj-NCC-2 (1). Spectroscopic data of 5a includes: MS (nano-electrospray ionization) m/z (%), 631.2 (12), 630.1 (52), 629.1 (100, [M+H]⁺), 597.1 (8, [M+H-32]⁺), 506.1 (5, [M+H-123]⁺); UV/visible, (MeOH) λ_max (relative ε) 238 (sh, 1.00), 313 (0.62); CD, (MeOH) λ_min/max (Δε) 226 (23.7), 252 (—8.6), 258 (—8.8), 281 (—18.4), 314 (5.1); ¹H NMR (500 MHz, 0.22 mM in CD₃OD, 26°C, tetramethylsilane), δ = 0.98 (H₃C-8²), 1.94 (H₃C-2¹ and H₃C-18¹), 2.07 (H₃C-12¹), 2.24 (H₃C-7¹), 2.28–2.36 (H₂C-17²), 2.42 (H₂C-8¹), 2.46 (H_AC-20), 2.61–2.67 (H_AC-17¹), 2.72–2.78 (H_BC-17¹), 2.84 (H_BC-20), 3.74 (H₃C-13⁵), 3.92 (H₂C-10), 3.94–3.99 (HC-1), 5.34 (H_cisC-3²), 6.08 (H_transC-3²), 6.43 (HC-3¹), 9.33 (HC-5). The UV/visible, CD, ¹H-NMR, and MS spectra of the NCC 5a showed no significant difference when compared with the spectra of authentic Cj-NCC-2 (1) (Figs. 2 and 4; and see Fig. 10, which is published as supporting information on the PNAS web site).

Identification of Cj-NCC-2 (1) and the NCC 5a by HPLC. Samples of the NCC 5a and authentic Cj-NCC-2 (1) from C. japonicum showed identical HPLC retention times and gave a single peak upon HPLC coinjection [flow: 500 μl/min; elution gradient at 0–40 min: from 200 mM ammonium acetate in MeOH/water 55:45 vol/vol (pH 6.8) to 100% 200 mM ammonium acetate in MeOH (pH 6.8); elution gradient at 40–60 min: 100% 200 mM ammonium acetate in MeOH (pH 6.8); see Fig. 11, which is published as supporting information on the PNAS web site].

Analysis of the raw extracts of degreened leaves of C. japonicum by HPLC indicated the presence of Cj-NCC-1 (3) (as described earlier) (18), in addition to small amounts of a second, rather unpolar NCC, named Cj-NCC-2 (1) here. The two NCCs, 3 and 1, were found in a ratio of ≈23:1. Using repeated flash chromatography and reversed-phase HPLC, a pure sample of Cj-NCC-2 (1) could be isolated. The structure of authentic Cj-NCC-2 (1) was elucidated by spectroscopic means. The molecular formula of Cj-NCC-2 (1) was determined with high-resolution MS as C₃₅H₄₀N₄O₇, based on its pseudomolecular ion at m/z = 629.2990. This result indicated Cj-NCC-2 (1) has a molecular weight smaller than the one of any known NCC (6, 18, 20–24) and relates to Cj-NCC-1 (3) (18) by its lack of one oxygen atom. The NCC 1 was thus identified as an isomer of a primary FCC, either the primary FCC (Bn-FCC-2) obtained from oil seed rape (Brassica napus) (13) or its C-1-epimer from sweet pepper (Capsicum annuum), named Ca-FCC-2 (2) (16). The UV/visible and CD spectra of a solution of 1 in methanol showed the absorption characteristics of the known NCCs (see Figs. 2 and 7; refs. 18 and 20–24). Homonuclear and heteronuclear NMR spectroscopy provided consistent signal assignment for all 35 C atoms of 1 and for 34 H atoms, i.e., for all carbon-bound H atoms, except for the exchange labile HC-13². The spectroscopic data allowed the constitution of 1 to be determined as that of a 3¹,3²-didehydro-1,4,5,10,15,20,22,24-octahydro-(13²-methoxycarbonyl)-4,5-dioxo-4,5-secophytoporphyrin. The NCC 1 is a natural NCC lacking a peripheral oxidative refunctionalization. So far, a hydroxyl group or a polar, oxygen-bearing functional group at the ethyl group extending from ring B was considered to be a characteristic of the natural NCCs (6, 18, 20–24). Such a function was assumed to be required as the basis for further peripheral refunctionalization, to assist the inter-organellar transport from the senescent chloroplast to the vacuole (17, 25, 26).

In earlier studies directed at determining the structures of FCCs, the FCC samples were noted to undergo uncharacterized chemical changes rather readily (13). Our interest has now turned to the question of whether FCCs would have a high tendency to engage in a specific tautomerization to NCCs, in analogy to the tautomerization reactions of hydroporphyrins, studied earlier by Eschenmoser (27). When a deoxygenated sample of ≈42 nmol (26 μg) of the chemically rather labile Ca-FCC-2 (2) (16) was stored at pH 4.9 in MeOH/water (30:70 vol/vol) at ambient temperature, a well-defined reaction was observed by UV/visible spectroscopy and HPLC. Conversion of the FCC 2 occurred within 120 min and a spectrum typical of an NCC appeared (see Fig. 3). The observed time dependence of the relative concentrations of 2, 5b, and 5a was analyzed by using a model of two consecutive reactions with first-order kinetics (see Fig. 9), and the isomerization of the FCC 2 to the NCC 5b could be estimated to occur with an effective rate of ≈0.025 min^—1 (half-life of 2 ≈27 min).

After overnight storage at ambient temperature, the crude product contained mainly the NCC 5a, according to HPLC analysis (Fig. 3). Monitoring by HPLC and UV/visible spectroscopy revealed the first product of the reaction not to be 5a, but a slightly less polar intermediate, the NCC 5b. Within ≈20 h the intermediate 5b disappeared nearly completely, with an estimated rate of ≈0.0055 min^—1 (i.e., half-life of 5b ≈130 min), to give the NCC 5a, which was isolated as the final isomerization product. The FCC 2 thus eventually isomerized in a rather clean (and nearly complete) reaction to the NCC 5a, which was isolated and subjected to spectroscopic and chromatographic analysis. The NCC 5a proved identical with Cj-NCC-2 (1). A sample of the tautomerization product 5a and a sample of authentic 1 were coinjected (at similar concentrations) for HPLC analysis, giving only one single HPLC peak (see Fig. 11). The electrospray ionization–MS, UV/visible, CD, and ¹H-NMR spectral data collected for the sample of 38 nmol (24 μg) of NCC 5a were all indistinguishable from those obtained for the natural NCC, Cj-NCC-2 (1). The ¹H-NMR spectra of 1 (in unbuffered solutions in methanol) showed an intriguing concentration dependence, resulting in slight chemical shift differences of the signals caused by H atoms of the propionic acid side chain mainly (see Figs. 4, 6, and 10). Only at comparable concentrations of 1 and 5a were the ¹H-NMR spectra of the two samples practically identical. The observed concentration effects have not been examined further so far. They might be caused by a varying degree of dissociation of the acidic propionic acid function at C-17³ or to (partial) intermolecular association (28).

In senescing higher plants, Chl a and Chl b both are degraded within several days to pheophorbide a (5–7), which is rapidly broken down further to polar, colorless, and nonfluorescent Chl catabolites (see Fig. 1), which accumulate in the vacuoles of senescent leaves (5, 6, 15). Under conditions where senescenceinduced Chl breakdown occurred with high rates, FCCs were observed temporarily and suggested to be catabolic intermediates preceding the NCCs (29, 30).

The structures of two isomeric FCCs from two plant species have been analyzed so far (13, 16). Both were identified as primary FCCs, differing from their catabolic precursor, pheophorbide a, only by the minimal structural changes required by the enzymatic oxygenolytic ring opening reaction and the subsequent regioselective enzymatic reduction, which generates the C-1 stereocenter (13, 16). The two primary FCCs were found to be C-1 epimers (16), pointing to the existence of two classes of reductases in higher plants (12, 13) that introduce the stereocenter at C-1 stereoselectively but with either configuration. The reducing enzyme, tentatively named red Chl catabolite (RCC) reductase, was suggested to be inactive in a mutant of Arabidopsis thaliana, the leaves of which developed necrotic behavior (31).

The two epimeric primary FCCs set the stage for one of the two possible stereochemical lineages in Chl breakdown, which are both observed to a similar extent in higher plants (32). The absolute configuration at C-1 appears not to be altered during the subsequent catabolic transformations and the NCCs are also indicated to fall into the two groups, depending on the stereochemistry of their precursor FCC (22, 32): e.g., the NCCs 1 and 3 from leaves of C. japonicum fall into the same stereochemical group as the NCCs from C. annuum and are therefore derived from a primary FCC identical to Ca-FCC-2 (2) (16, 22). Despite the indicated stereochemical divergence of the NCCs at C-1, the long wavelength parts of the CD spectra of the known natural NCCs are all similar (see, e.g., Fig. 2; refs. 6, 18, and 20–24). A common absolute configuration at the chiral centers at C-15 (and C-13²) of the natural NCCs is thought to be responsible for the latter. The corresponding UV absorbance at ≈315 nm is caused by ring B (4, 5, 24), which experiences its nearest chiral center at C-15 (see Fig. 1 and ref. 33 for the numbering used). Analysis of the nuclear Overhauser effect spectra of various NCCs allowed the relative configuration at C-15 and C-13² to be derived as being the same in all of them. As configurational equilibration at the exchange labile C-13² produces the more stable trans arrangement of the bulky groups at C-15 and C-13², the configurationally stable chiral center C-15 directs the stereochemical preference at C-13² (4, 5, 18, 20, 22–24). The absolute configuration of the stereocenters C-15 and C-13² of the NCCs is thus determined upon generation of the chiral center C-15.

We report here on an isomerization of the natural primary FCC 2 to the NCC 1, which occurs nonenzymatically and with high stereoselectivity at the crucial chiral center C-15. The NCC 1 (Cj-NCC-2) from extracts of senescent leaves of C. japonicum is a nonpolar natural NCC. The FCC 2 is identical to the unknown primary FCC from C. japonicum (22), the catabolic precursor of the NCC 1. The sample of 2 actually used here was obtained as Ca-FCC-2 from C. annuum (16).

In mildly acidic aqueous medium, at pH 4.9, the FCC 2 isomerizes to an NCC with a half-life of <30 min at ambient temperature. This behavior of the monoacid 2 contrasts with the much higher stability of the methyl esters of 2 and of isomeric FCCs (34). The remarkable effect of the type of carboxylic acid functionality provides direct evidence for a relevant role in the isomerization process of the propionic acid function extending from C-17, which is S-configurated in the FCC 2 but is prochiral in the NCCs. Indeed, as deduced from the mechanism inferred for the acid-catalyzed isomerization of the FCC 2 to the NCC 5a (= 1), the propionic acid function assists to accomplish the crucial step in the isomerization reaction (see Fig. 5) by stereoselective intramolecular protonation at C-15. Molecular models indicate such an intramolecular protonation to be feasible at C-15 from the re-side of the intermediate “enamine” 4 only, but not from the si-side, to result in R configuration at C-15 in 5a and 5b. Accordingly, the protonation product would not lead directly to 5a, an NCC with a stable relative trans arrangement of the bulky groups at C-15 and C-13². However, it would require the formation of an intermediate epi-NCC (5b) with the alternative cis arrangement. An unstable NCC is indeed observed as an intermediate in the isomerization reaction of the FCC 2 to the NCC 5a. This intermediary NCC is thus suggested to be (13²epi)-5a (= 5b), the C-13² epimer of 5a.

The deduced crucial role of the propionic acid function in the mechanism of the stereoselective isomerization of the FCC 2 to the NCC 1 (= 5a) provides a basis for a stereochemical assignment of C-15 and C-13² in the NCC 1. An intramolecular protonation involving a stereochemical transposition from C-17 to C-15 is invoked as the basis for the highly effective induction of the absolute configuration as 15-R (and 13²-S). The absolute configuration at C-15 in 1 is the same as the one in the other known NCCs: two of the three chiral centers at the skeleton of the natural NCCs can now (tentatively) be assigned as 15-R and 13²-S. The common stereochemistry of the NCCs at C-15 is a predictable outcome of the hypothetical nonenzymatic isomerization and is in striking contrast to the stereodivergence of the tetrapyrrolic Chl catabolites at C-1 (16), the chiral center introduced enzymatically during Chl breakdown. Such a surprising stereodivergence of an enzymatic step was rationalized as a feature of genuine catabolic processes (16), as it is likely to have no metabolic consequence. Considering this, the apparently common stereochemical build-up of the NCCs at their C-15 (and C13²) is all the more paradoxical. As shown here, it can be rationalized as the consequence of an inherent tendency of the short-lived FCCs to isomerize spontaneously and stereoselectively in weakly acidic medium.

The structure of Cj-NCC-1 (3) and other polar natural NCCs point to refunctionalization reactions at several side chains (6, 18, 20, 23), which are likely to occur in the chloroplast and at the stage of the FCCs (15). Such peripheral transformations are indicated to assist the transport of the FCCs from the senescent chloroplast to the vacuole (17, 25, 26). According to the current topographical model of Chl breakdown (6, 8), FCCs are imported into the vacuoles before their transformation to the NCCs. The vacuoles, therefore, not only function as the place of storage for the NCCs in senescent plants (8, 14), but also as the site of their formation (5, 6).

Direct in vivo measurements of the vacuolar pH during senescence have not been performed. It is commonly accepted that the acidity of the vacuoles changes only marginally, if at all, as plant cells age. Thus, the rather acidic environment in the vacuoles (pH ≈4–5; refs. 17 and 35) would induce FCCs to undergo a rapid stereoselective isomerization to the NCCs, as shown here. The deduced half-life of the FCCs of 30 min or less (depending on the effective acidity) in the vacuole is kinetically competent with the temporary appearance of minor amounts of FCCs during senescence-induced Chl breakdown. Indeed, from comparison of the apparent rate of import of FCCs into the vacuole (limited by the overall rate of disappearance of pheophorbide a) with the rate of the isomerization of the FCC 2 at pH 4.9, FCCs can be estimated to represent only ≈1% (or less) of the total amount of Chl catabolites in the vacuoles.

The delineated kinetic and stereochemical criteria provide evidence for the relevance of a nonenzymatic isomerization of FCCs to NCCs during Chl breakdown. The FCCs are shown to be programmed by nature to further isomerize to the NCCs in an acidic medium. This feature of Chl breakdown has a remarkable manmade parallel in Woodward's total synthesis of Chl: There a surprisingly effective, stereoselective build-up of the ring D periphery of the chlorin macrocycle was achieved by strategic exploitation of the principle “that two things cannot be at the same place at the same time” (36, 37).

According to our studies the senescent leaf does not depend on an enzyme for the formation of NCCs from FCCs, an important step in Chl breakdown. Instead, interorganellar transport and a compartmentalized and suitable reaction medium (the acidic vacuolar sap) are used in an economic way to achieve this step in a well-controlled catabolic pathway. Export of the FCCs from nonacidic metabolically active compartments to the vacuole is recognized as a critical catabolic step in Chl breakdown (6, 8). In case this export to the vacuoles was slow, FCCs would accumulate in senescent leaves, as may upstream catabolites, like RCC. Indeed, accumulation of RCC is assumed to be responsible for the lesion mimic phenotype observed in A. thaliana mutants defective in RCC reductase (31).

The formation of NCCs was considered as the “ultimate” step in the breakdown of Chl in senescent higher plants (6). However, in the presence of air the NCCs are oxidized to rust-colored products (3, 4, 6). In some senescent leaves NCCs are processed further oxidatively to other tetrapyrroles (38) and Chl appears to be broken down even to monopyrrolic fragments (39). In view of the recent identification of the tetrapyrrolic heme degradation product bilirubin as an antioxidant cytoprotectant (40), the pronounced sensitivity of the NCCs to oxidants may turn out to have physiologic relevance in plant senescence.

In conclusion, the fleetingly observable FCCs can now be considered to be the proper products of the enzyme-catalyzed part of Chl breakdown in senescent higher plants. Once deposited in the vacuoles, the FCCs are geared to isomerize to NCCs, whose further fate and possible physiological role deserve further attention. These findings also underline the crucial position of the ring opening monooxygenase in aerobic Chl catabolism (10). Aside from being the only known senescence-specific enzyme in higher plants, this latter oxygenase or variants of it are also known to be the key enzymes in Chl catabolism in senescent green algae (6, 41).

Abbreviations: Chl, chlorophyll; FCC, fluorescent Chl catabolite; NCC, nonfluorescent Chl catabolite; RCC, red Chl catabolite.

We thank Sonja Berger for technical assistance and Dr. Christian Eichmüller for measuring NMR spectra. This work was supported by Austrian National Science Foundation Projects 13503 and 16097, and Swiss National Science Foundation Grant 31.63628.