Assay for Chemical Purity and Isotope Enrichment

General Considerations

Two criteria relevant to Omicron product quality are chemical purity and the level of isotope enrichment. The former is established by various analytical methods, including HPLC, GLC, and NMR. Usually one or a combination of these methods, when used in conjunction with physical tests such as melting point and optical rotation, proves sufficient to establish the purity of our products.

A related issue is the hydration state of the product, which bears on the true cost for the consumer (i.e., Are you paying a high price for waters of hydration?). For crystalline products, hydration state is determined by X-ray crystallography or from Karl-Fischer titration in those cases where the information is unavailable from the literature. For Omicron products sold as syrups, we assume that H₂O comprises 20% of the syrup weight, the maximal percentage determined from empirical studies conducted in our laboratory.

The question of isotope enrichment level is less easily established with certainty. Two common analytical methods used for this assay, NMR and mass spectrometry, yield results whose quality depends highly on the care taken with data collection and analysis, and on the behavior of the specific compound under investigation. At Omicron, we rely on NMR-based assays for many, but not all, of these determinations. In the following, we briefly describe some of the details of the NMR-based assay and the criteria used to judge its reliability.

The NMR-Based Assay of Enrichment Level

Quantitative NMR requires careful adjustment of the acquisition parameters and careful preparation of the sample. We illustrate how isotope enrichment level is determined by ¹H NMR using D-[1-¹³C]glucose as the test sample.

D-[1-¹³C]Glucose exists mainly as α-glucopyranose 1 (38%) and β-glucopyranose 2 (62 %) in aqueous solution at 30° C (Zhu et al., J. Org. Chem. 2001, 66, 6244-6251). The anomeric H1 signals of both forms in an unlabeled sample are distinguished from those of the remaining ring protons by their characteristic chemical shifts and their multiplicities (the doublets are caused by the presence of ³J_H1,H2 whose magnitude depends on the H1-C1-C2-H2 torsion angle). When ¹³C is introduced at C1 at significant levels over natural abundance (1.1 %), four H1 signals appear for each form; the additional large ¹J_C1,H1 coupling (typically ~160-170 Hz in aldopyranosyl rings) is superimposed on the smaller ³J_H1,H2 coupling, yielding a doublet of doublets. Thus, in D-[1-¹³C]glucose, each anomer produces six H1 signals, two arising from the residual unlabeled molecules and four arising from the C1-labeled molecules.

This pattern is illustrated in Figure 1, which shows partially-relaxed ¹H NMR spectra of D-[1-¹³C]glucose (H1 region only) obtained from an inversion-recovery T₁ determination (see below). The inverted signals A-C are attributed to H1 of 1, where signal B arises from the unlabeled molecules, and signals A and C from the labeled molecules. An analogous set of signals (D, F and G) is observed for H1 of 2. Signal E arises from residual HOD in the solvent. It is evident that, in principle, the level of ¹³C-enrichment can be determined by appropriate integration of the H1 signals associated with each anomer (data for each anomer provides redundant information on the enrichment level at C1). For example, the integration of signals A+C yields the amount of labeled species and the integration of signal B yields the amount of unlabeled species.

While this approach appears straightforward, there are serious pitfalls to be avoided. Signal-to-noise ratio must be excellent, and signals must be adequately digitized to eliminate undesirable peak-area truncation. Care must be taken during signal integration to avoid user bias; the latter can be avoided by using a computerized lineshape analysis commonly available on many modern spectrometers. Finally, signal area distortions arising from differential spin-lattice relaxation times must be eliminated. The latter is addressed through the incorporation of adequate interpulse delay times during data acquisition to insure complete nuclear relaxation.

Choosing proper interpulse delay times hinges on knowing the spin-lattice relaxation properties of the molecule. This information can be obtained through direct measurement of spin-lattice relaxation times, T₁, in the sample under solution conditions similar to those used for routine determinations of isotope enrichment level. The T₁ values of H1 attached to ¹²C will differ from those of H1 attached to ¹³C. Although the gyromagnetic ratio of ¹³C is ~1/4 that of ¹H, the short internuclear distance between C1 and H1 (the C-H bond length is ~1.08 Å) renders the ¹³C at C1 a potent source of dipole-dipole relaxation for H1. For example, consider the T₁ values extracted from the partially-relaxed (inversion recovery) spectra shown in Figure 1. For H1 of 2, T₁ = 1.8 s when attached to ¹²C and 0.8 s when attached to ¹³C. For H1 of 1, T₁ = 3.3 s when attached to ¹²C and 0.9 s when attached to ¹³C. The ratio, T₁(¹²C)/T₁(¹³C) is not constant since this ratio depends highly on molecular geometry. In 1, H1 is equatorial and thus relaxation mediated by other carbon-bound hydrogens is less potent (because these protons are too far away from H1; 1/T₁ is proportional to r_CH^-6) than for H1 of 2, which is axial and thus relatively close to the axial H3 and H5. Consequently, providing a new relaxation pathway through the ¹³C nucleus at C1 affects the T₁ of 1 more than that of 2.

Importantly, the data in Figure 1 show that an interpulse delay of 3.3 s x 5 = 16.5 s is required to insure the attainment of thermal equilibrium between pulses. The value of knowing T₁ values is apparent; without this information, selection of the proper interpulse delay cannot be made. An improper choice can result in inaccurate determinations of enrichment levels.

Reliability of the NMR-Based Assay for Enrichment Level

In order to validate isotope percent determinations by NMR, a parallel study of five compounds was undertaken: four samples of D-[1-¹³C]glucose containing different levels of ¹³C-enrichment at C1, and D-[1,6-¹³C₂]glucose. ¹H NMR spectra were collected as described above, and the same samples were analyzed by an independent mass spectrometry facility (Metabolic Solutions, Inc.). The results are shown in Table 1. The computed enrichment levels determined from the NMR and MS data are in good agreement, suggesting that the NMR-based assay yields reliable isotope enrichment levels to within ± 0.2-0.4% of the actual value (we assume that the MS data yield the true enrichment level). Based on these results, we rely mainly on NMR-based methods for routine assays of enrichment level. Clearly, however, this approach is not applicable to all types of labeled products; for example, assays of uniformly ¹³C- and ²H-labeled compounds cannot be conducted readily by NMR. In these cases, mass spectrometry provides the required information.

All Omicron products are accompanied by one source of data indicating the level of isotope enrichment of the sample. Additional assays can be requested by the customer; charges for these additional assays are paid by the customer.

Table 1
A Comparison of Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) Determinations of Percent Isotopic Enrichment in Isotopically-Labeled Glucoses
	Percent 13C-Enrichment
Compound	NMR Assaya	MS Assayb
D-[1-13C]glucose 1	96.3	96.7 (0.22)
D-[1-13C]glucose 2	100.0	99.7 (0.03)
D-[1-13C]glucose 3	99.8	100.0 (0.22)
D-[1-13C]glucose 4	96.6	96.2 (0.1)
D-[1,6-13C2]glucose 5	99.9	99.1 (0.16)
aUsing an 8.8 s recycle time.
bStandard deviation in parentheses, n = 3
data collected by Metabolic Solutions Inc.		return to text

Figure 1