c ch

Figure 11.26

two correlations), a methine (maximum of three correlations), or a quaternary carbon (four correlations). The latter case is the most interesting, the case of a singlet methyl group. From the 1H shift in F2 of a singlet methyl group, we look along the vertical line in the HMBC spectrum and identify the four carbons. From the DEPT or edited HSQC, we can put them into categories: CH3, CH2, CH, or Cq. If only one of the four is quaternary, we know it has to be the one attached to our methyl group, and we know that the HMBC crosspeak is due to a 2 JCH relationship. This means that the other three are attached to the Cq, giving 3 JCH crosspeaks to our methyl protons. The beauty of this special case (singlet methyl group with only one correlation to a quaternary carbon) is that the three-bond versus two-bond ambiguity is removed: We can construct a molecular fragment consisting of five carbons from one vertical line in the HMBC spectrum (Fig. 11.26, lower right).

In Figure 11.25 the HSQC spectrum is aligned with the HMBC spectrum for comparison. In practice, this is done with NMR software in which two or more 2D spectra can be displayed simultaneously with crosshair cursors in each spectrum, controlled by the mouse and coordinated to be at the same chemical-shift position in F2 and F1 in all of the frames displayed. Another way to analyze the HMBC spectrum is with a printed list of chemical shifts taken from the HSQC spectrum. In this case, two lists are used: One in order of 1H chemical shift and one in order of 13C chemical shift (including the quaternary carbons). Both lists use the same arbitrary numbering (or lettering) system for the carbons. In this way the F1 (13C) and F2 (1H) shifts of an HMBC crosspeak can be searched in both lists, and the "nearest neighbor" shifts can be checked to see if there is any ambiguity in the assignment. If another resonance is very close, the assignment is made as a choice of possibilities, for example, H-5a/H-11b. These ambiguities can be resolved later as the structure becomes clear. The juxtaposition of the HSQC and HMBC spectra in Figure 11.25 will give you a feel for using correlated cursors in the NMR software. In this case, we use the HSQC spectrum itself as a "lookup table" for chemical shifts.

Because they are so intense, the methyl signals give strong one-bond artifacts in the HMBC spectrum. These are indicated by rectangles and have the same pattern as the methyl crosspeaks (a, b, c, e, and f) in the HSQC spectrum (left side) but with the additional 1 Jch coupling in F2. They are identical to the nondecoupled HMQC pattern for methyl groups (Fig. 11.15). Starting with methyl group a (upper right), the vertical line extending downward from the center of the one-bond artifact goes through four intense, antiphase doublet crosspeaks. Using the precise alignment with the HSQC spectrum, we can assign them as s (CH2), t/u (CH2/Cq), w (CH), and x (CH). Carbons t and u are too close in chemical shift to be distinguished in the F1 dimension (0.05 ppm difference), although

Figure 11.27

we could look for other HMBC crosspeaks that are very slightly higher or lower than this one to make the assignment. In this case we can use logic: Because methyl group a is a singlet in the 1H spectrum, it must be bonded to a quaternary carbon. The other three HMBC crosspeaks are not quaternary, so this crosspeak has to be to carbon u, the quaternary carbon. This gives us an unambiguous five-carbon fragment: CaH3-Cuq-(CsH2, CwH, CxH). In the cholesterol structure, this has to be C-18, the angular methyl group at the C-D ring juncture (Fig. 11.27). Focus on the top of the figure and pencil in the resonances (a-aa) as we go through the assignment process. So we can assign a = 18, s = 12, and w,x = 14,17. Carbons w and x are ambiguous because both C14 and C17 are CH carbons. Note that this confirms our assessment of w and x as two of the "crowded" CH carbons 9, 14, and 17. Methyl group a cannot be C19, the other singlet methyl group, because it has no HMBC correlation to a quaternary olefinic carbon (C5).

Starting again with the other singlet methyl group, c, we have HMBC correlations to carbons q (CH2), v (CH), and aa (olefinic Cq) (Fig. 11.25). The crosspeak at F1 = 36.47 ppm (just below carbon o), however, does not line up exactly with any of the HSQC crosspeaks. Careful examination of the DEPT and 13C spectra (Fig. 11.18) identifies this correlation as the quaternary carbon p. Because there are two correlations to quaternary carbons, there might be an ambiguity in how to arrange them, but one is olefinic (140.74 ppm) and cannot be directly connected to the methyl group (CH3-C(=C)-C) because then there would be only three HMBC correlations. The fragment is thus CcH3-Cpq-(Caaq, CvH, CqH2), and we can assign c = 19, p = 10, aa = 5, v = 9, and q = 1 using the structure-based numbering system (Fig. 11.27). This confirms the observation from both 1D selective NOE and 2D NOESY and ROESY spectra of a strong NOE from H4ax to the singlet methyl peak at 1.01 ppm (peak c in the HSQC).

Now we move to the doublet methyl groups. These have to be connected to a CH carbon because they appear as doublets in the 1H spectrum, and all three are on the side chain: C-21, C-26, and C-27. Moving down from the methyl b one-bond artifact in the HMBC spectrum (Fig. 11.25), we encounter a "stretched" HMBC crosspeak at 36 ppm that is much "fatter" in the vertical (F1) direction than the others. This is a correlation to two nearby carbons that is not resolved in F1. These correspond to peaks n (CH) and o (CH2) in the HSQC spectrum. A third and final crosspeak is observed to peak w (CH) at the bottom (a doublet methyl group can have only three correlations). Because there are correlations to two CH groups, it is not immediately clear which is directly bonded to the methyl group (of course, the COSY spectrum would tell us). But we have already assigned w and x to positions 14 and 17 (or 17 and 14) from carbon a (C-18). So it is clear that the fragment is CbH3-CnH-(C°H2, CwH), and we can assign b = 21, n = 20, o = 22, and w = 17 (Fig. 11.27). This also clears up the w/x ambiguity: x = 14.

The remaining two doublet methyl groups must be C-26 and C-27 of the isopropyl group. If you look closely at the center of the one-bond artifact for peaks e and f in the HMBC (Fig. 11.25), you will see a strong crosspeak that has a diagonal shape running from upper left to lower right, opposite to the direction (lower left to upper right) of the overlapping one-bond artifacts (and the overlapped HSQC crosspeaks). These are the HMBC correlations from C-26 to H-27 and from C-27 to H-26. Even when two carbons are exactly equivalent in an isopropyl group, there will be an HMBC correlation at the center of the one-bond artifact. Similarly, the pair of equivalent ortho or meta positions in a monosubstituted or p-disubstituted benzene ring will show three-bond HMBC crosspeaks at exactly the position of the 13C-decoupled HSQC crosspeak. We know this is not a one-bond artifact because it would be widely separated by the 1JCH coupling. Moving down along the F2 = He and F2 = Hf vertical lines, there are two "double-wide" crosspeaks corresponding to carbons i (CH) and r (degenerate CH2). These can be assigned to C-25 and C-24, respectively. The suspicion that degenerate CH2 groups might correspond to the flexible side chain is confirmed at least in the case of C-24.

Figure 11.28 shows the HMBC correlations from the olefin and alcohol portions of the molecule: H-6 (z) in F2 and C-3, C-6, and C-5 (y, z, and aa, respectively) in Fi, aligned with a portion of the HSQC spectrum. The olefinic proton H-6 (left side) correlates to carbons l/m, p (Cq), and t/u. Carbons p = C10 and u = C13 are already assigned from the methyl groups, so these three correlations can be assigned to p = C10 (Cq) and t = C4 (CH2), with l and m remaining ambiguous (l = C8 and/or m = C7, Fig. 11.27). From the olefinic carbons (Fig. 11.28, bottom), correlations can be seen from C-3 (y), C-5 (aa), and C-6 (z) to the proton of HSQC peak t (H-4ab). C-5 = aa (but not C-6 = z) shows a correlation to a proton position that intersects peaks q (C-1), k, and j (all at the downfield or equatorial proton of the geminal pair). This must be q: H-1eq, which bears an anti relationship to C-5 and is too far away (four bonds) from C-6 to give an HMBC correlation. Although four- and even five-bond correlations are common for JHH when a double bond is in the path, more than three-bond correlations are very rare in HMBC spectra. Both C-5 (aa) and C-6 (z) show another correlation to a proton position that intersects the i, k' and m' crosspeaks in the HSQC (k and m at the upfield or axial positions). Peak i has already been assigned (C-25, too far away), and the alignment is best with peak m' (see enlargement, Fig. 11.28, upper left), so the correlation can be assigned to the axial proton on Cm: H-7ax. Finally, the alcohol carbon y (C-3) shows correlations to proton peaks t and q (already assigned to H-4eq, H-4ax, and H-1eq) and to an F2 position a bit upfield of the one already assigned to m' (H-7ax). This aligns perfectly with peak k', which is offset to the right side and above peak m' (see enlargement, Fig. 11.28, upper left side). Because C-1 and C-4 are already assigned, the only remaining CH2 proton "within reach" of C-3 is H-2ax (k'). This example serves to illustrate how precise alignment and consideration of a number of overlapping possibilities are essential to the interpretation of HMBC spectra.

0 0

Post a comment