Tracer studies with crude U-13C-lipid mixtures. Biosynthesis of the lipase inhibitor lipstatin.

The biosynthesis of the pancreatic lipase inhibitor lipstatin was investigated by fermentation experiments using cultures of Streptomyces toxytricini, which were supplied with soybean oil and a crude mixture of U-13C-lipids obtained from algal biomass cultured with 13CO2. Lipstatin was analyzed by one- and two-dimensional NMR spectroscopy. 13C total correlation spectroscopy and INADEQUATE experiments show that two fatty acid fragments containing 14 and 8 carbon atoms, respectively, are incorporated en bloc into lipstatin. The 14-carbon fragment is preferentially derived from the unsaturated fatty acid fraction, as shown by an experiment with hydrogenated U-13C-lipid mixture, which is conducive to labeling of the 8-carbon moiety but not of the 14-carbon moiety. The data indicate that the lipstatin molecule can be assembled by Claisen condensation of octanoyl-CoA with 3-hydroxy-delta5,8-tetradecanoyl-CoA obtained by beta oxidation of linoleic acid. The formation of lipstatin from acetate units by a polyketide-type pathway is ruled out conclusively by these data. The data show that surprisingly clear labeling patterns can be obtained in studies with crude, universally 13C-labeled precursor mixtures that are proffered together with a large excess of unlabeled material. One- and two-dimensional 13C total correlation spectroscopy analyses are suggested as elegant methods for the delineation of contiguously 13C-labeled biosynthetic blocks.

Lipstatin is produced by Streptomyces toxytricini (1). The structure is characterized by a ␤ lacton ring carrying two aliphatic residues with chain lengths of 6 and 13 carbon atoms (Ref. 6 and Fig. 1). One of the side chains contains two isolated double bonds and a hydroxy group esterified to N-formylleucine. Structurally, lipstatin is closely related to the esterase inhibitor esterastin, which contains a N-acetylasparagine side chain instead of N-formylleucine (7). Tetrahydrolipstatin can be obtained by catalytic hydrogenation of lipstatin (6). Total synthesis of tetrahydrolipstatin has been reported (8).
Nothing is known about the biosynthesis of lipstatin. Initial biosynthetic studies with isotope-labeled acetate were not con-ducive to the incorporation of label into lipstatin. 1 The long alkyl side chains of lipstatin could be biosynthesized from acetate via a polyketide pathway or from preformed fatty acids. S. toxytricini can be grown on medium containing large amounts of lipids, and it appeared likely that these were actively metabolized by the microorganism. In that case, 13 Clabeled lipids could be incorporated into lipstatin by partial degradation of preformed fatty acids or by total degradation via acetyl-CoA.
Triglycerides with appropriate 13 C labeling were not commercially available. We therefore decided to use a mixture of universally 13 C-labeled lipids, which was obtained by acetone extraction of algal biomass grown with 13 CO 2 . This complex mixture was extensively diluted with soybean oil (natural 13 C abundance) for the preparation of culture media. In conjunction with advanced NMR techniques, this approach was conducive to the elucidation of the building blocks of lipstatin. This method appears to be generally useful for biosynthetic studies on compounds with a putative origin from fatty acid precursors.

MATERIALS AND METHODS
U- 13 C-Lipid-The crude U-13 C-lipid mixture used in this study was purchased from Dr. H. Oschkinat (European Molecular Biology Laboratory, Heidelberg, Germany). It was prepared by acetone extraction of algal biomass (Scenedesmus obliquus) grown on 13 CO 2 . The acetone extract was evaporated to dryness under reduced pressure. The black, oily material was used as a supplement to the culture medium without purification.
Hydrogenation of U- 13 C-Lipid-A solution of 13 C-lipid (crude algal extract, 277 mg) in 6 ml of ethanol was hydrogenated over 550 mg of Raney nickel (50°C, 10-bar hydrogen, 3 h). The product still contained approximately 8% of C18:1 fatty acid according to GC 2 analysis. The hydrogenation was therefore repeated using similar conditions, yielding 124 mg of fully hydrogenated material.
Fatty Acid Composition of 13 C-Algal Lipid-An aliquot of the 13 Clabeled lipid mixture (20 mg) and 1.25 mg of pentadecanoic acid were dissolved in 10 ml of 0.5 M sodium methylate in methanol. The mixture was heated to 60°C in a screw cap glass for 30 min and was then acidified with 4 ml of 3.6% hydrochloric acid in methanol. An aliquot of 5 ml was diluted with 3 ml of water, and the mixture was extracted with 5 ml of n-hexane. The resulting fatty acid methyl esters were identified by GC/MS on a capillary column (DBWAX; 20 m, 0.3-m film; temperature gradient, 120 -250°C, 4°C/min; splitless injection). Quantitative GC analysis of the methyl esters was performed on a capillary column (OV-225; 25 m, 0.25 m; temperature gradient, 140 -220°C, 3°C/min). Pentadecanoic acid methylester was used as internal standard for quantification, and dihomo-␥-linoleic acid ethyl ester (97.4% purity by weight) was used as a reference substance. The 13 C content of fatty acid residues was determined by GC/MS analysis (electron impact ioniza-tion) of the methyl esters.
Fermentation-For biosynthetic studies, the U-13 C-labeled precursors had to be diluted extensively with unlabeled material. Per 35 ml of fermentation medium, 1.8 g of soybean oil and 0.1 g of the 13 C-labeled lipid mixture or hydrogenated 13 C-labeled lipid mixture were used. The lipids were emulsified in water using 0.5 g of soybean lecithin as an emulsifier. The medium also contained soybean flour (1.4 g) and glycerol (0.7 g). The pH was adjusted to 7.4 before sterilization (121°C, 20 min). After seeding with S. toxytricini, the culture was incubated for 7 days at 27°C and 70% relative humidity on a rotary shaker at 220 rpm and 5-cm throw.
Isolation of Lipstatin-Fermentation broth (58 ml) was extracted with 150 ml of acetone and 100 ml of hexane. After separation of the organic layer, the aqueous layer was extracted three times with 100 ml of a 1:1 mixture of acetone and hexane. The combined organic extracts were dried with sodium sulfate and concentrated to yield a green oil (3.1 g) containing 173 mg of lipstatin. The material was dissolved in hexane (40 ml) and placed on a column of silica gel (Bond Elut, 10 g). The column was developed with hexane (40 ml) and hexane/ethyl acetate at dilutions of 20:1 (v/v, 300 ml), 10:1 (200 ml), 20:3 (200 ml), and 5:1 (200 ml), yielding 220 mg of crude material after evaporation of solvent. Second chromatography on silica gel using the same procedure afforded 163 mg of semipure material. Further purification was done by reversed phase chromatography. Lipstatin was dissolved in 50% aqueous isopropanol (25 ml) and placed on a column of Bond Elut C-18 (10 g) which was developed with 50% isopropanol (225 ml) and 60% isopropanol (150 ml). Fractions containing lipstatin were concentrated by evaporation under reduced pressure, and the residue was extracted with ethyl acetate. The extract was dried with sodium sulfate and was concentrated to dryness, yielding 112 mg of lipstatin with 88% purity (high performance liquid chromatography, percentage of area) containing 9% of lipstatin analogs.
Two-dimensional double quantum-filtered COSY, DEPT, and INAD-EQUATE experiments were performed according to standard Bruker software (DISR87). Phase-sensitive two-dimensional TOCSY spectroscopy was done according to the method of Bax and Davis (9). 1 H-Detected multiple quantum 1 H-13 C chemical shift correlation experiments (HMQC and HMBC) were performed according to the methods of Bax and Subramanian (10) and Bax and Summers (11). Samples were not rotated during two-dimensional experiments.
Data acquisition and processing parameters for two-dimensional experiments were: COSY, 32 scans/t 1 increment, 2.0-s relaxation delay, 480 ϫ 2048 raw data matrix size, zero filled to 2048 words in t 1 and processed with 2-Hz Gaussian in the f 1 dimension, and 90°-shifted sine bell filtering in the f 2 dimension; TOCSY, 32 scans/t 1 increment, 2.0-s relaxation delay, 62-ms MLEV-17 mixing period preceded and followed by 2.5-ms trim pulses, 90°pulse width, 44 s, 512 ϫ 2048 raw data matrix size, zero filled to 2048 words in f 1 and processed with 2-Hz Gaussian in the f 1 dimension, and 90°-shifted sine bell filtering in the f 2 dimension; HMQC, 256 scans/t 1 increment, start of coherence experiment 159 ms after a bilinear rotation decoupling pulse, 3.5-ms delay period for evolution of 1 J CH corresponding to a coupling of 143 Hz, 13 C decoupling during acquisition by globally optimized alternating phase rectangular pulse sequence 1, 500 ϫ 1024 raw data matrix size, zero filled to 1024 words in t 1 and processed with 10-Hz Gaussian in f 1 , and 90°-shifted squared sine filtering in f 2 ; HMBC, 64 scans/t 1 increment, 1.8-s relaxation delay, 3.5-ms delay for suppression of 1 J CH , 60-ms delay period for evolution of long-range couplings ( 2 J CH and 3 J CH ) corresponding to a coupling of 8 Hz, and 350 ϫ 2048 raw data matrix size, zero filled to 1024 words in t 1 and processed with 20-Hz Gaussian in f 1 and in f 2 ; and INADEQUATE, 128 scans/t 1 increment, 1.5-s relaxation delay, Ernst-type phase cycle, 5.0-ms delay for evolution of 1 J CC , and 350 ϫ 2048 raw data matrix size, zero filled to 2048 words in t 1 and processed with 60°-shifted sine bell filtering in f 1 and f 2 .
Phase-sensitive two-dimensional 13 C TOCSY experiments were performed with a MLEV-17-based mixing period (9). The 13 C excitation pulse was generated in the transmitter high power output level (90°p ulse, 5 s). 13 C pulses for mixing were generated in the transmitter low power output level amplified with a BFX5 unit (90°pulse, 30 s). The MLEV-17 mixing period was 45 ms and was preceded and followed by 2.5-ms trim pulses. The data were acquired in the phase-sensitive mode using time-proportional phase increments. Other data acquisition and processing parameters were: 48 scans per t 1 increment, 2.0-s relaxation delay, and 400 ϫ 2048 raw data matrix size, zero filled to 2048 in t 1 and processed with 90°-shifted, squared sine bell functions in f 1 and Gaussian broadening in f 2 .
One-dimensional 13 C TOCSY experiments were performed with selective excitation using a Gaussian-or half-Gaussian-shaped pulse of 5 ms in length generated by the transmitter output of a selective excitation unit (Bruker). The transfer of magnetization between 13 C atoms was achieved by a MLEV-17-based mixing period (45 ms) preceded and followed by trim pulses (2.5 ms). The pulses for mixing were generated in the low power output of the transmitter amplified with a BFX5 unit (90°pulse, 30 s).

NMR Signal Assignment of Lipstatin
A 1 H and 13 C NMR analysis of lipstatin providing assignments for some of the carbon atoms has been reported (6). However, since the biosynthetic study depended crucially on unequivocal assignments for all 13 C signals, a more detailed NMR analysis using double quantum-filtered COSY, 1 H TOCSY, and inverse carbon-proton correlation experiments such as HMQC and HMBC was in order. Additional assignment information was afforded by 13 C TOCSY and INADE-QUATE analysis of multiple 13 C-labeled lipstatin samples obtained in the labeling experiments described below. 1 H and 13 C NMR signal assignments of lipstatin are summarized in Table I.

Analysis of U-13 C-Lipid Mixture
An acetone extract of totally 13 C-labeled algal biomass obtained by growth of Scenedesmus obliquus on 99% enriched 13 CO 2 as a carbon source was obtained from Dr. H. Oschkinat. After evaporation of the solvent, the viscous oil appeared almost black. The dark color was in part due to the presence of chlorophylls in considerable amount.
For assessment of the fatty acid content, the crude algal lipid mixture was subjected to methanolysis, and the resulting fatty acid methyl esters were analyzed by coupled gas chromatography/mass spectrometry (G. Oesterhelt, Hoffmann-La Roche AG). The results are shown in Table II. The combined fatty acid residues account for about 40% (w/w) of the total material. Eleven fatty acids with chain lengths of 14 -18 carbon atoms were determined in widely different abundance. In the saturated fraction, palmitic acid was the dominant component. The combined unsaturated fatty acids accounted for 32% (w/w) of the crude material. The dominant components were linolenic acid (about 21%) and linoleic acid (about 8%). The 13 C abundance was approximately 97%.
An aliquot of the U-13 C-lipid mixture from algae was subjected to two consecutive cycles of catalytic hydrogenation over Raney nickel. Gas chromatographic/mass spectrometric analysis of fatty acid methyl esters obtained by methanolysis confirmed that virtually all double bonds had been removed by hydrogenation (Table II). The dominant fatty acid in the hydrogenated mixture was stearic acid (about 25%, w/w).

Biological Studies
Fermentation with U-13 C-Lipid-Fermentation experiments were performed with S. toxytricini and the culture medium reported earlier (1). To detect the joint transfer of 13 C-labeled atoms, it is important to dilute the U-13 C-labeled precursors extensively with unlabeled material. We used the 13 C-labeled lipid mixture and unlabeled soybean oil at a ratio of 1:17.4 (w/w) for the preparation of the fermentation medium. After cultivation for 7 days, lipstatin was isolated as described under "Materials and Methods." The analysis of the 13 C satellites in the 1 H NMR spectrum of lipstatin gave 13 C enrichments of about 4%. This indicated that the 13 C-labeled algal lipid had been metabolized by the organism at a similar rate as soybean oil and had served as a   precursor of the lipstatin molecule. The 13 C NMR spectrum of lipstatin revealed the presence of extensive 13 C-13 C coupling in the biolabeled molecule (Fig. 2). For example, the 13 C signal of C-3 was characterized by a central signal at 74.9 ppm and by a doublet ( 1 J CC , 39.2 Hz) resulting from coupling to one adjacent 13 C atom (C-2 or C-4). The 13 C satellites of C-5 appeared at a distance of 78.2 Hz. Coupling to either C-4 or C-6 should result in a coupling constant of about 40 Hz due to the aliphatic nature of C-4 and C-6. Therefore, the observed coupling signature must result from simultaneous coupling to two adjacent 13 C atoms (i.e. C-4 and C-6) yielding a pseudo-triplet ( 1 J CC , 39.1 Hz) where the central component overlaps with the uncoupled singlet (Fig. 2). It follows that the carbon atoms C-4 -C-6 were derived en bloc from the uniformly 13 C-labeled algal lipid. Similarly, the signal of C-2 was characterized by simultaneous 13 C coupling to two 13 C spins via one-bond coupling ( 1 J CC , 42.4 Hz and 38.0 Hz, respectively) and to one 13 C spin via two-bond coupling ( 2 J CC , 3.6 Hz) (Fig. 2).
The analysis of 13 C-13 C coupling in terms of multiplicities in the one-dimensional 13 C NMR spectrum is summarized in Table III. These data show already that biosynthetic modules containing more than two carbon atoms were incorporated into the lipstatin molecule. However, the size of these respective building blocks can be addressed much better by two-dimensional NMR analysis as described below.
A section of a two-dimensional 13 C INADEQUATE experiment is shown in Fig. 3. The double quantum-filtering technique monitors pairs of adjacent 13 C atoms but not isolated 13 C nuclei. Due to the low natural abundance of 13 C, INADE-QUATE is notorious for its low sensitivity. However, pairs of 13 C atoms that were contiguously incorporated from multiple 13 C-labeled precursors are diagnosed with high sensitivity. The data in Fig. 3 indicate 13 C-13 C coupling between the carbon atoms 3 and 4, 4 and 5, and 5 and 6, suggesting the presence of molecules with contiguous 13 C labeling between C-3 and C-6. A more detailed analysis (Table III) proves that 13 C-13 C coupling occurs between each individual pair of adjacent carbon atoms in both alkyl side chains, thus suggesting that these modules were incorporated en bloc from their respective precursors. However, 13 C-13 C coupling was not detected between C-3 and C-2, as shown by the absence of signals at the positions marked by open circles in Fig. 3. It follows that the bond between C-2 and C-3 was formed in the biosynthetic pathway and was not present in the totally labeled precursor molecule. Bond formation involving one labeled precursor molecule would most frequently involve an unlabeled reaction partner, and the frequency of 13 C-13 C coupling along this bond should be as low as about 4% (i.e. undetectable) compared with about 97% for jointly transferred carbon pairs. Contiguous 13 C labeling throughout both alkyl side chains is also obvious by 13 C TOCSY methods. The physical basis of the TOCSY experiment is the exchange of magnetization between directly coupled spins by a radiofrequency field (spin lock field). In the case of a two-dimensional experiment, the result of the mixing process is that a spin ( 1 H or 13 C) shows a correlation cross-peak to each of the nuclei in a contiguous spin system.
A transfer of magnetization under the influence of a spin lock field was introduced by Davis and Bax (12) in 1985. To improve the performance of the experiment, periodic phase alteration (MLEV-17) of the spin lock pulses was implemented (10). Traditionally, the TOCSY experiment is used to assign scalar coupled 1 H spin systems. More recently, TOCSY pulse trains were also applied to transfer magnetization between 13 C of labeled biopolymers (13). However, the 13 C TOCSY experiment is used infrequently in the evaluation of biosynthetic pathways and therefore requires some technical comments. The most important difference between 1 H and 13 C TOCSY methods is the larger spectral width of the 13 C chemical shift range. Efficient transfer of magnetization by the spin lock pulse requires a sufficiently high power of spin lock field B 1 . Typically, ␥B 1 should exceed the spectral width. However, the maximum strength of the B 1 field is limited by the thermal and electronic stability of the probe head during the relatively long mixing period (typically 10 -60 ms).
To optimize the experimental parameters, we performed a series of two-dimensional 13 C TOCSY experiments with [U-13 C]lysine. A spin lock field of 8 kHz applied for 45 ms enabled the transfer of magnetization over a relatively wide frequency range (15 kHz) between all of the six carbon atoms of lysine. However, it should be noted that the cross-peak intensity between the carboxylic carbon and the side chain carbon atoms decreased approximately by a factor 10 relative to the other cross-peaks, which had similar intensities.
To improve the limited digital resolution of this two-dimensional experiment (e.g. for extraction of coupling constants), we performed a selective excitation of a single 13 C frequency followed by a 13 C TOCSY mixing process. The selective excitation was achieved by a 5-ms Gauss or semi-Gauss pulse. In model experiments with [U-13 C]lysine, we were able to transfer magnetization from the ␣-carbon to C-3-C-6 with similar efficiency and to C-1 with decreased intensity. Obviously, the spin lock field (8 kHz) was too weak to achieve efficient magnetization transfer between C-1 and C-2 of lysine, which are separated by approximately 12 kHz from each other. Since the intensities of C-3-C-6 in the one-dimensional 13 C TOCSY spectrum were similar, the applied spin lock field was optimal for a frequency range of 3-4 kHz.
These results served as a basis for the 13 C TOCSY experiments with lipstatin. Fig. 4 shows a part of a phase-sensitive two-dimensional 13 C TOCSY experiment with lipstatin from the fermentation with U-13 C-lipid encompassing the aliphatic spectral region (80 -10 ppm). The transfer of magnetization is highlighted in Fig. 4 among C-3-C-6 and among C-2 and C-1Ј-C-6Ј. Again, no transfer of magnetization was observed between C-2 and C-3. Additionally, a series of selective one-dimensional 13 C-TOCSY experiments was performed (Table  III). Fig. 5 shows spectra obtained by selective excitation of C-3 and C-2, respectively, and subsequent isotropic mixing. Due to the physical basis of the experiment, only signals were observed that result from magnetization transfer from the excited carbon. Consequently, the highly crowded 13 C-coupled onedimensional 13 C spectrum can be edited by this spectroscopic technique (Fig. 5). It should be noted that the 13 C spin lock field was not strong enough to achieve magnetization transfer from the aliphatic C-6 to the alkene carbon atoms. However, the signal of C-6 in the one-dimensional 13 C TOCSY experiment is a pseudo-triplet (Fig. 5B), which clearly indicates contiguous 13 C coupling to the unsaturated C-7 of lipstatin. In combination with signal multiplicities in the one-dimensional 13 C NMR spectrum (i.e. double doublets of the inner chain carbon atoms, indicating contiguous coupling) and the INADEQUATE results, this proves the presence of isotopomers with contiguous 13 C labeling extending from C-16 to C-3, and from C-1 to C-6Ј (Fig. 6A). We conclude that the lipstatin molecule is assembled from a 14-carbon (C-16 -C-3) and an 8-carbon (C-1-C-6Ј) moiety, which can both be derived en bloc from the U-13 C-lipid mixture supplied as a precursor.
Two pairs of labeled carbon atoms (i.e. C-1Љ/C-2Љ, and C-4Љ/ C-5Љ) were incorporated into the leucine side chain from the 13 C-labeled lipid mixtures. This signifies the diversion of multiple 13 C-labeled components of the lipid mixture to the biosynthesis of the amino acid.
Fermentation with Hydrogenated U-13 C-Lipid-The presence of two isolated double bonds in the 14-carbon moiety suggested that it might be derived preferentially or exclusively from the unsaturated fatty acid fraction of the 13 C-labeled precursor mixture (i.e. more specifically, from linoleic acid). To check this hypothesis, we used a totally hydrogenated U-13 Clipid mixture for an incorporation experiment.
The hydrogenated U-13 C-lipid was mixed with soybean oil at a ratio of 1:17.4 (w/w), and the mixture was proffered to a growing culture of S. toxytricini. Lipstatin was isolated and analyzed by NMR spectroscopy as described above. The 13 C NMR signals of the 8-carbon moiety were again characterized by the presence of 13 C-coupled satellites, whereas the carbon atoms of the 14-carbon moiety showed no 13 C-13 C coupling in the one-dimensional NMR experiment (Table III). Analysis by two-dimensional INADEQUATE and 13 C TOCSY spectroscopy is summarized in Table III and confirmed that the 8-carbon moiety but not the 14-carbon moiety was consecutively labeled (Fig. 6B).
It follows that the unsaturated 14-carbon moiety was not biolabeled in the experiment with the hydrogenated U-13 C- FIG. 4. Part of a two-dimensional 13 C TOCSY spectrum of lipstatin from fermentation with U-13 C-lipid using a spin lock field of 8 kHz applied for 45 ms. The contiguous 13 C spin system comprising C-3-C-6 is indicated above the diagonal; the contiguous 13 C spin system comprising C-2 and C-1Ј-C-6Ј is indicated below the diagonal. lipid. Apparently, this part of lipstatin was exclusively derived from the unsaturated fraction of the unlabeled soybean oil supplement in this experiment. The labeling pattern of the leucine residue was the same as in the experiment described above. DISCUSSION We have demonstrated that the lipstatin backbone is assembled from two moieties consisting of 8 and 14 carbon atoms, respectively, which were both contiguously labeled with 13 C from a U-13 C-lipid mixture. Since the uniformly 13 C-labeled lipids were proffered together with a large excess of unlabeled lipid material, this result shows that both building blocks were derived from precursor lipids by partial catabolism of fatty acid residues. The de novo synthesis of the building blocks from smaller units such as acetate could not possibly yield lipstatin molecules with uninterrupted 13 C labeling of long alkyl chains, since labeled and unlabeled precursor molecules would be interspersed at random, thus conducing to noncontiguous 13 C labeling, as shown in the formylleucine moiety of lipstatin. In this case, the TOCSY transfer of magnetization along the alkyl chain would be interrupted.
Catalytic hydrogenation destroys the potential of the 13 Clabeled lipid mixture to serve as a precursor for the 14-carbon moiety. However, the hydrogenated lipid mixture is still used efficiently as a precursor of the 8-carbon moiety and of leucine. It follows that the 14-carbon moiety is specifically derived from an unsaturated component of the U-13 C-lipid mixture. Linoleic acid has the same pattern of double bonds as the unsaturated side chain of lipstatin and could be converted to an appropriate precursor by ␤ oxidation (Fig. 7). Alternative pathways for the partial degradation of linoleic acid, such as the pathway described recently in rat liver (14,15), cannot be ruled out on basis of the available data.
A possible scenario for the subsequent reaction steps is immediately obvious, but the details require further study. The hydroxy group of the 14-carbon module could be aminoacylated prior to or after the Claisen condensation. Reduction of the carbonyl group generated by the Claisen condensation should yield a hydroxy group, and the ␤-lactone ring of compound 7 could be formed with the CoA moiety as a leaving group.
It should be noted that this biosynthetic pathway was observed in a medium containing large amounts of saturated and unsaturated fatty acids. Under these conditions, the 14-carbon moiety is entirely derived from the unsaturated fatty acid pool, as shown by the experiments with the hydrogenated U-13 Clipid mixture. Apparently, desaturation of fatty acid does not play a significant role under these culture conditions.
The data also show that degradation products of the proffered lipid are diverted to the biosynthesis of leucine. The incorporation of two 13 C pairs from the precursor mixture is well in line with the biosynthetic pathway of the amino acid, as summarized in Fig. 9. The biosynthesis of leucine involves the condensation of pyruvate and acetyl-CoA molecules. A mixture of 13 C 3 -pyruvate and 13 C 2 -acetyl-CoA with the respective unlabeled precursors should yield the observed labeling pattern, as shown in Fig. 9. Whereas the details have not been investigated, 13 C 3 -pyruvate could be formed during the fermentation from the glycerol part of the U-13 C-lipid mixture. This should be conducive to the observed isotope distribution.
The use of a crude mixture of 13 C-labeled precursors for biosynthetic studies is an unusual approach. As shown with the present example, it can yield results of optimum clarity under appropriate conditions. The crucial part of the present experiments is the use of a mixture of totally 13 C-labeled lipids with a large excess of unlabeled lipids. In conjunction with modern one-and two-dimensional NMR technology, this approach can define the length of the biosynthetic building blocks with a minimum of experimental effort. This strategy is not limited to mixtures of lipids. Indeed, we have shown independently that it can also be used successfully using crude mixtures of 13 Clabeled carbohydrates or amino acids. 3