C15

-C14-C8:

-44° (2.5 Hz)

79° (-0.6 Hz)

No HMBC

J values in parentheses are predicted from the dihedral angles using the Karplus relation, modified in the case of C-H couplings for the smaller 3 JCH range. The "fat" HMBC crosspeak observed in Figure 11.30 between H-15 and C-17 is more consistent with j hydroxylation (predicted3 JCH = 2.5 Hz) than with a hydroxylation (0.8 Hz), while in neither case would we expect to see the corresponding H-17 to C-15 crosspeak (3 JCH = -0.4 or -0.5 Hz). The observation of an H-15 to C-13 crosspeak and the absence of an H-15 to C-8 crosspeak both support the assignment of j stereochemistry for the hydroxyl group. Even the observed 3Jhh coupling of 5.8 Hz is consistent with the predicted coupling (5.1 Hz) from the j isomer and too small for the predicted coupling (8.3 Hz) for the a isomer. NOE intensities would not be useful in this case: The H-17 to H-15 distance in the models is 3.91 A for the a isomer (trans relationship) and 3.48 A for the j isomer (cis relationship). Both are within the range (< 5 A) to give a measurable NOE—the simple 1/r6 relationship predicts an NOE twice as large for the cis relationship. The simple notion that cis-related protons will have large NOEs and trans-related protons will not show NOEs at all is clearly false. Only by isolating both isomers and carefully comparing the NOE intensities could we attempt to assign the stereochemistry based on NOE data alone.

11.6.2 Oxidation of ^-Carotene

To mimic the metabolism of j-carotene (Fig. 11.31), a sample was oxidized with m-chloroperbenzoic acid (mcpba) and a product with biological activity was partially purified. Due to higher molecular weight impurities, the 1D 1H spectrum was overlapped and difficult to interpret, but a small molecule component was easily identified in the HMQC spectrum due to its narrow linewidths. Figure 11.32 shows four crosspeaks from the downfield (lower left) region of the 500 MHz HMQC spectrum of the oxidation product

Figure 11.31

(insets) with an F2 slice through each crosspeak at the indicated 13C chemical shift in F1. The HMQC was acquired without13 C decoupling over a 24-h period, and the component that gives rise to these crosspeaks was estimated to be about 0.7 mg. In addition to the large 154-160 Hz) one-bond JCH couplings, the larger homonuclear (JHH) couplings can also be measured from the F2 slices. From these alone, a spin system can be identified in which one proton is coupled to two others with coupling constants of 14 Hz and 18 Hz: CH-CH-CH. Because the 1H and 13C shifts (5.5-7.4 ppm and 115-139 ppm, respectively) are in the olefinic region these large couplings suggest a trans olefin or an anti relationship between protons on two sp2-hybridized carbons connected by a single bond. Because the spin system is isolated (no other homonuclear J couplings), we can "cap" the spin system on either end with a quaternary carbon: Cq=CH-CH=CH-Cq. Another spin system can be deduced from the olefinic CH with a 4 Hz coupling to another proton (Fig. 11.32, top). An HMQC crosspeak at F2 = 4.67 ppm, F1 = 82.1 ppm also shows a single 4 Hz JHH coupling (not shown). These chemical shifts are in the region of an alcohol (CH-OH) group, shifted downfield by its proximity to the olefinic CH. Again we can "cap" the spin system at both ends: Cq=CH-CH(OH)-Cq. These fragments are shown in Figure 11.33, with the corresponding parts of the ^-carotene structure from which they could be derived indicated by circles (C6-C9 and C9-C13). This would indicate that oxygen was introduced at C8, with the double bond moving from C5-C6 to C6-C7.

Figure 11.33

An HMBC spectrum with phase-sensitive data presentation was acquired on a purer sample over a 60-h period to tie these fragments together and complete the structure. At an F\ chemical shift of 198.4 ppm three crosspeaks were observed (Fig. 11.34: insets and F2 slice). The F1 chemical shift indicates a ketone carbonyl carbon, and one of the crosspeaks, an antiphase doublet (J = 5 Hz) at F2 = 2.27 ppm, is at the right 1H shift for a methyl ketone: CH3-CO. The lack of any homonuclear couplings in addition to the 5 Hz active (2JCH) coupling is consistent because the methyl ketone would be a singlet in the 1H spectrum. The crosspeak at F2 = 6.17 ppm (Fig. 11.34, left of center) has an active (antiphase) coupling of 3.4 Hz and a passive (in-phase) coupling of 18 Hz. This is the doublet CH proton found in the HMQC spectrum at F2 = 6.17, F1 = 130.4 (Fig. 11.32) corresponding to the right-hand CH of the five-carbon fragment (Fig. 11.33). The crosspeak at F2 = 7.44 (Fig. 11.34, left) has a more complex coupling indicating more than one passive (JHH) coupling. This is the CH proton found in the HMQC spectrum at F2 = 7.44, F1 = 139.1, corresponding to the middle CH of the five-carbon fragment. So the ketone carbonyl of the CH3-CO fragment must be the quaternary carbon at the right side of the five-carbon fragment (Fig. 11.33), with two-bond correlations from the CO carbon to the CH3 proton and the right side CH proton (6.17 ppm) and a three-bond correlation to the middle CH proton (7.44 ppm). The downfield shift of this center CH (139.1 ppm 13C, 7.44 ppm 1H) relative to typical olefin values (120-130 ppm 13C, 5-6 ppm 1H) is explained by its ß position in an a,ß-unsaturated ketone (-Cß+-Ca = C-O- resonance structure).

HMBC F2 slice Fx = 198.4 ppm

Proton (ppm)

HMBC F2 slice Fx = 198.4 ppm

Proton (ppm)

15.14 27.37

30.47 27.52

15.14 27.37

30.47 27.52

Figure 11.35

Thus it appears that in addition to oxidation at C8 and shifting of the C6-C7 double bond, we have cleaved the ^-carotene structure at C13-C14, with oxidation of C13 to a ketone. Another quaternary carbon was identified in the HMBC spectrum at F1 = 79.8 ppm, with correlations to the proton at 5.56 ppm next to the CHOH (Fig. 11.33) and to a methyl group at 1.59 ppm. This corresponds to C5 of ^-carotene, which must be oxygenated and sp3-hybridized because its 13 C chemical shift (near 80 ppm) is characteristic of Cq-OH rather than an olefinic carbon. Analysis of many more crosspeaks in the HMBC spectrum, as well as a mass spectrum with molecular ion (M+) of nominal mass 290 (C18H26O3, 6 unsaturations) led to the cyclic peroxide structure shown in Figure 11.35. The mass spectral data was inconsistent with a diol structure (predicted m/z 292) so the cyclic peroxide was proposed. In all, 30 HMBC correlations were identified (arrows), making the structural assignment quite solid, and all 1H and 13C positions were assigned chemical shifts. All of this was accomplished on an impure sample without enough material for a simple 13 C spectrum.

0 0

Post a comment