Organic Molecule Identification

The bonding arrangement of a small organic molecule for which the exact mass is known can be inferred from 1D ¹H and ¹³C and 2D ¹H/¹³C HSQC and HMBC spectra. Samples of 1 mg (FW < 500) of 95% pure material can be routinely identified with a standard suite of NMR spectra acquired in several hours; if a one-to-one correspondence of peaks and atoms is present, straightforward data analysis rapidly yields complete ¹H and ¹³C NMR resonance assignments. For some samples, however, the data may indicate that an atom in a particular bonding arrangement is present as more than one possible peak: such complications can arise from a single chemical species having multiple conformations that are in slow exchange due to bond rotation barriers (e.g. amides) or from different chemical species (isomers, if only a single mass is present). Such samples require extensive data analysis or even additional data acquisition: spectra may be acquired in other solvents, and at different temperatures or magnetic fields, to eliminate the possibility that any of the additional NMR signals might derive from some unexpected contaminating or major species.

Core personnel recently helped BCM Associate Professor Jin Wang and colleagues prove the chemical identity of RT-AM (Figure 1), the acetoxymethyl protected membrane-permeable version of the compound RT that they synthesized as a probe for quantitative real-time imaging of glutathione.

Exact mass spectra agree with the structure predicted from the synthetic scheme (Figure 1), but a mixture of E and Z geometric isomers at the critical double bond where the reversible Michael addition occurs (red oval in Figure 1) means the 1D ¹H (or ¹³C) spectra will show two sets of resonances (see Figure 2a), which complicates the NMR data analysis. The reversibility of the Michael addition, a key feature of RT that allows rapid kinetics of the probe response to glutathione, enables RT (and RT-AM) to convert between the E and Z isomers, so even pure samples of one isomer will revert to a mixture over time. 2D ¹H/¹³C multiple bond correlations (Figure 2b) nevertheless allow each of the signals in the 1D spectra to be assigned to one isomer or the other. In these 2D spectra, J coupling between hydrogens and carbons located two or three bonds away give rise to correlation peaks at the ¹H and ¹³C chemical shifts, allowing the chemical nature of the neighboring atoms to be identified and thus the bonding arrangement to be deduced.

A second complication in these NMR spectra is slow rotation about the amide bond (purple oval in Figure 1), which affects the line shapes of resonances for atoms on the nitrogen side of the amide bond. Hydrogens D and D’ (Figure 1) will have different chemical shifts because they are in different environments; each is also expected to appear as two peaks in the 1D ¹H spectrum because of the E/Z isomers (Figure 3a). Rotation about the amide bond allows the two amide substituents within a given isomer to exchange positions, as demonstrated by positive NOE cross peaks (Figure 3b); the lack of cross peaks between resonances in different isomers indicates that equilibration between the E and Z forms is not detected on a time scale of 400 milliseconds. Rotational exchange about the amide bond occurs at different rates for the E and Z isomers, giving rise to different line widths for the resonances of the two isomers as well as different relative intensities of NOE cross peaks and auto peaks.

The effect of conformational averaging on the 1D ¹H spectrum is even more dramatic for hydrogens E and E’ of Figure 1: the broad linewidths cause signals for the more rapidly exchanging isomer to collapse nearly to the baseline (Figure 4, top). Cooling the sample from 299.5 K to 269.5 K slows this exchange and reveals the expected four distinct peaks for hydrogens E and E’ (Figure 4, bottom). Conformational exchange can affect the line width and sensitivity of signals in any of the standard NMR experiments used to probe small molecule structure, but by changing the solvent or the temperature one can usually reveal the underlying species present (i.e., the broad peaks in Figure 4 top do not give detectable ¹H/¹³C HMBC correlations but all peaks in Figure 4, bottom do).

The effect of conformational averaging on the 1D ¹H spectrum is even more dramatic for hydrogens E and E’ of Figure 1: the broad linewidths cause signals for the more rapidly exchanging isomer to collapse nearly to the baseline (Figure 4, top). Cooling the sample from 299.5 K to 269.5 K slows this exchange and reveals the expected four distinct peaks for hydrogens E and E’ (Figure 4, bottom). Conformational exchange can affect the line width and sensitivity of signals in any of the standard NMR experiments used to probe small molecule structure, but by changing the solvent or the temperature one can usually reveal the underlying species present (i.e., the broad peaks in Figure 4 top do not give detectable ¹H/¹³C HMBC correlations but all peaks in Figure 4, bottom do).

Despite the presence of one configurational heterogeneity and one conformational heterogeneity on the millisecond to second time scale, a careful analysis of the 2D HSQC/HMBC one- and multiple-bond correlations identifies a complete set of resonances for the expected E and Z species in a 4:3 ratio (Figure 5), and with all NMR peaks accounted for, we can conclude that the sample comprises only the expected RT-AM E and Z isomers.

X Jiang, J Chen, A Bajic, C Zhang, X Song, SL Carroll, ZL Cai, M Tang, M Xue, N Cheng, CP Schaaf, F Li, KR MacKenzie, AM Ferreon, F Xia, MC Wang, M Maletic-Savatic, and J Wang, Quantitative real-time imaging of glutathione, Nat Commun. 2017, 8:16087.