Figure 11.52

our experiment; the minimum t1 value is twice the gradient time. This can be remedied by introducing two 13 C spin echoes in the center of the evolution delay to "make room" for the two gradients (Fig. 11.52). Note that all chemical-shift evolution during the four A delays cancels, for 1H and for 13C. This is just one example of how you can get creative with pulse sequences: there are many ways to design an HMBC sequence and this is a very active area of research (e.g., "CIGAR," "ACCORDION," "ADEQUATE," "HSQMBC," etc.). Some sequences are designed for accurate measurement of long-range (2'3JCH) heteronuclear coupling constants, and there are many different strategies for getting rid of the one-bond artifacts ("low-pass filters"). Two big problems remain to be solved: removing the ambiguity of two-bond and three-bond correlations, and eliminating homonuclear (JHH) /-coupling evolution during the long (~50 ms) 1/(2/) delay. The complex logic of puzzle solving in interpreting HMBC data would be vastly simplified if we knew for each crosspeak whether it is a two-bond or a three-bond relationship. The homonuclear J evolution distorts the peak shapes by introducing antiphase terms, and this makes it difficult to accurately measure the long-range JCH coupling.


A case study of covalent structure deterimation by 2D NMR will help to illustrate the central importance of HSQC and HMBC, used in conjunction with the homonuclear COSY and ROESY experiments. Oxidation of the natural product Pristimerin with DDQ in diox-ane gave four products which were separated and purified (Fig. 11.53). The third fraction (LGJC3,4.1 mg) was dissolved in 0.5 mL of deuterated chloroform (CDCl3) and analyzed by one-dimensional (*H and 13C) and two-dimensional (HSQC, HMBC, TOCSY, COSY and ROESY) NMR using a Bruker DRX-600. In this case, we have a great deal of information about the sample since we know its origin and we can imagine that most of the carbon skeleton of Pristimerin is conserved in the product. In the following discussion, however, the problem will be approached initially as a "white powder" unknown to show how much we can conclude from NMR alone.

Carbon resonances in the 13C spectrum were numbered 1-32 in order of chemical shift from the farthest upfield (c1) to the farthest downfield (c32). Using the HSQC spectrum, the protons were named according to the carbon to which they are correlated, using "a" and

Figure 11.53

"b" to indicate the downfield and upfield resonances of a geminal pair (CH2 group): hi, h6a, h6b, and so on. To avoid confusion between this arbitrary numbering system and the structure-based numbering system of Pristimerin (Fig. 11.53), lower case "c" and "h" will be used for the arbitrary, chemical-shift based numbering (e.g., h23, c16, etc.) and upper case "C" and "H" will be used for structure-based references (C-5, H-8, etc.). The advantage of arbitrary numbering is that it does not bias us toward any particular structure and it does not have to be revised as our structural model evolves.

There are clearly three one-proton olefinic peaks (6.1-6.6 ppm), two coupled to each other with a 6.7 Hz coupling and a third coupled to one of the first two with a long-range coupling

of 1.7 Hz. These might correspond to the original H-1, H-6, and H-7 of Pristimerin, with the cis vicinal H-6 to H-7 coupling being the large coupling (6.7 Hz) and the long-range coupling (1.7 Hz) being between H-1 and H-6, a 5-bond coupling through the extended n system of the transoid diene (C1-C10-C5-C6). There are two OCH3 singlets, one at 3.30ppm, typical of a methyl ether, and one at 3.58 ppm, closer to the typical chemical shift for a methyl ester. A methyl ether is unexpected, and in fact for the first 1H spectrum, run on a less pure sample at 200 MHz, this peak was ignored. In the cleaner sample at 600 MHz it is clear that both peaks integrate to around three protons. In the upfield region (0.6-1.6 ppm) there are six methyl singlets, corresponding to the six methyl groups in the triterpene skeleton of Pristimerin, attached at carbons 4, 9, 13, 14, 17, and 20. Most of the other peaks are complex and overlapped.

Ignoring the TMS peak and the three CDCl3 peaks, there are 32 peaks in the 1H-decoupled 13C spectrum. One of these peaks, c22 at 77.23 ppm, was eventually assigned to the CHCl3 resonance (residual CHCl3 in the deuterated solvent and possible CHCl3 solvent left over from extraction and purification), leaving 31 carbons in the compound of interest. Of these 31 peaks, nine are in the downfield region typical for olefinic, aromatic or carbonyl carbons. Four of these nine peaks are in the typical olefinic region (116-131 ppm). Six are "short" peaks, indicating that they are quaternary carbons, including one of those in the olefinic region. Pristimerin would be expected to have ten downfield 13C peaks (sp2 carbons), nine from the A-B ring system (C-1-C-8 and C-10) and one from the methyl ester carbonyl

(C29), so it appears that one is missing. Perhaps one of these olefinic/aromatic/carbonyl carbons was converted by DDQ oxidation to an sp3-hybridized carbon that appears in the upfield (< 100 ppm) region (Fig. 11.55). LGJC3 (C3i) has one more carbon than Pristimerin (C30H40O4), consistent with the observation of an extra methyl group (methyl ether) in the proton spectrum. Combined with the observation of one fewer sp2-hybridized carbon in LGJC3, this suggests the addition of methanol to one of the sp2-hybridized carbons in the extended n system (C-1-C-8 and C-10) of Pristimerin. A carbon peak observed at 86.6 ppm in LGJC3 is typical for an oxygenated sp3-hybridized carbon, and this could be the carbon in the triterpene backbone where the methoxy group is attached. If so, the chemical shift puts it in the range of quaternary, singly oxygenated carbons as opposed to CH-O (70-80 ppm) or CH2-O (60-70 ppm).

11.10.3 2D HSQC Spectrum (Upfield Region: Fig. 11.56; Center and Downfield Regions: Fig. 11.57)

This is an edited, 13C-decoupled HSQC spectrum with negative intensities (CH2/methylene groups) shown in gray and positive intensities (CH/methine and CH3/methyl groups) in black. The carbons with no hydrogens attached (quaternary carbons) do not show up in this spectrum. A broad CH2 crosspeak was observed at SH = 1.26, SC = 29.7 due to long-chain hydrocarbon grease ((CH2)„), a common contaminant in solvents used for extraction and chromatography. The methyl carbon crosspeaks (c1, 2, 3, 9, 10, and 13) can be easily distinguished from the methine carbon crosspeaks because they are much more intense. Thus 8 CH3 peaks (6 upfield and 2 methoxy), 7 CH2 peaks (one degenerate and 6 in "pairs" at the same 13 C shift), and 4 CH peaks (three olefinic and one aliphatic) were observed in the HSQC spectrum, leaving 31 -8 -7 -4 = 12 quaternary carbons. By comparison with the 1D 13C spectrum, the 12 quaternary carbons can be identified and fall into three categories: five sp3-hybridized carbons without oxygen (30-45 ppm); one sp3-hybridized carbon with one oxygen (86.6 ppm); and six sp2-hybridized carbons (130-196 ppm: the "short" ones in the downfield region of the 13C spectrum). Primisterin has 7 CH3, 7 CH2,4



Edited, C decoupled


0 0

Post a comment