A new tool for NMR analysis of complex systems: selective pure shift TOCSY

Abstract

A new NMR experiment is proposed to aid the identification of components in complex systems, including mixtures. It greatly enhances spectral resolution, by generating a pure shift spectrum, with multiplet structure suppressed, only for those spins coupled to a chosen chemical shift.

---

Systems of many spins, whether in the form of large molecules or of mixtures of small ones, pose particular challenges in structure elucidation. The narrow range of proton chemical shifts makes spectral resolution critically important, and has provided the impetus for the development of a wide family of “pure shift” experiments in which the homonuclear multiplet structure that normally limits resolution is suppressed.^1–3 However in complex systems such as naturally-occurring mixtures, or in supramolecular or systems chemistry, even pure shift spectra are poorly resolved and show extensive overlap. Here a new experiment is described that uses 1D selective TOCSY⁴ to restrict the signals observed to those of spins with couplings to a chosen chemical shift, and PSYCHE⁵ to record those signals with broadband suppression of homonuclear couplings. The result of editing the spectral data in this way is an improvement of over two orders of magnitude in resolution, enabling simple spectra of single spin systems to be extracted from highly complex mixture spectra.

The new experiment is demonstrated here (Fig. 1) on peppermint (Mentha piperita) oil, a complex natural mixture. Typically in such mixtures a family of chemically-related species is present at a wide range of concentrations, giving rise to severe problems with both resolution and dynamic range in NMR analysis. The main components here (Scheme 1a–c) are menthol (ca. 30% w/w) and menthone (20%), but the majority of signals come from minor components (<10%).⁶ The aliphatic part of the conventional ¹H spectrum (Fig. 1a) illustrates this problem well, with almost wall-to-wall peaks of very different intensities. The PSYCHE pure shift spectrum (Fig. 1e) achieves an impressive improvement in resolution, but spectral overlap still frustrates efficient analysis. From the TOCSY spectrum it is clear that detailed analysis, even using 2D methods, is all but hopeless (see Fig. S2 in the ESI‡).

Fig. 1 500 MHz (a) conventional, (b–d) 1D selective TOCSY (4 min total acquisition time), (e) PSYCHE (2 h 50 min), (f–h) 1D selective TOCSY-PSYCHE ¹H spectra (1 h 35 min for menthol and 3 h 10 min for menthone and neomenthol) from peppermint oil 25% (v/v) in DMSO-d₆. (b and f) H₂ of menthone (selective pulse at 2.10 ppm) (c and g) H₁ of neomenthol (selective pulse at 3.91 ppm), and (d and h) H₁ of menthol (selective pulse at 3.17 ppm) were selected using RSNOB pulses (bandwidth 33 Hz for menthol and neomenthol and bandwidth of 21 Hz for menthone). Full spectra and all experimental parameters can be found in the ESI.‡ The small artefacts between 4_ax and 3_ax in (f) arise from strong coupling, and the signal marked with an “x” (b and f) arises from spillover excitation of another species with a resonance close to 2.10 ppm.

Scheme 1 Structures of the main compounds involved: (a) menthol, (b) neomenthol, (c) menthone, (d) provitamin D₃ and (e) vitamin D₃.

One step in the right direction is to use frequency-selective excitation⁷ of an isolated resonance followed by isotropic mixing (TOCSY)⁸ to excite the spectrum of one spin system only (Fig. 1b–d). This efficiently extracts the relevant resonances, but the complex individual spectra with overlapping multiplets still do not allow all the chemical shifts to be identified. Combining the two methods, selective TOCSY excitation and pure shift data acquisition, in a 1D selective pure shift TOCSY experiment solves the problem (Fig. 1f–h). Here well-resolved spectra of two the major components, menthol and menthone, and of the trace component neomenthol (Scheme 1a–c) are obtained (for full assignments see Section B1.3 in the ESI‡). In principle the same information should be available from a full 2D TOCSY-PSYCHE⁹ spectrum, but in practice the strong signals of major components give rise to weak sideband artefacts and overlapping peak tails that are of similar magnitude to the signals of minor components, greatly complicating analysis (see Fig. S3–S6 in the ESI‡). Using selective excitation elegantly bypasses this problem, as well as allowing the acquisition of the relevant data in a much shorter time and giving better resolution. It is also possible to speed up 2D acquisition if only a small part of the spectrum is of interest.¹⁰ In principle non-uniform sampling could be used to give very marked speed improvements,^11,12 but there are potential complications in complex mixtures.

The pulse sequence used (Fig. 2) combines 1D selective TOCSY, using DIPSI-2 for isotropic mixing¹³ flanked with zero quantum filters,¹⁴ with the PSYCHE method of pure shift acquisition.⁵ The required coherence transfer pathway is enforced with a combination of pulsed field gradients and phase cycling (see ESI‡ for a detailed description). In principle other pure shift methods could be used,^15,16 as demonstrated recently for a band-selective analogue.¹⁷ Here we demonstrate a fully broadband method, for which PSYCHE generally gives the best sensitivity and spectral purity.

Fig. 2 Pulse sequence for the 1D selective TOCSY-PSYCHE experiment. Narrow and wide filled rectangles represent hard 90° and 180° pulses respectively. Trapezoids with cross-diagonal arrows are low-power chirp pulses of small flip angle β. Trapezoids on either side of the DIPSI-2 isotropic mixing element are low power 180° chirp pulses used to suppress zero quantum coherences. The first selective 180° pulse is applied to an isolated resonance; typically RSNOB or REBURP shapes are used. A detailed description of the pulse sequence can be found in the ESI.‡

A second example is given for a mixture of vitamin D₃ and provitamin D₃ (Scheme 1d and e). This is particularly illustrative for the situation commonly encountered in step-by-step synthesis where a precursor has a chemical structure very similar to that of the end product. The path from precursor to vitamin D₃ here involves an electrocyclic ring opening followed by a [1,7]-hydride shift,¹⁸ which has a more pronounced effect on the chemical shifts of the cyclohexanol ring hydrogens than on the remainder of the spectrum.

The conventional ¹H spectrum (Fig. 3a) shows many complex overlapping multiplets, essentially preventing analysis. Selectively refocusing only the respective carbinolic hydrogens (H_3α) of the different spin systems and sharing coherence across the spin system with TOCSY, in conventional 1D selective TOCSY (see Fig. S8 in the ESI‡), here allows the signals of each species to be extracted cleanly. However in more complex spin systems multiplets are likely to overlap, complicating and in some cases preventing assignment. The full pure shift spectrum (Fig. 3b), on the other hand, gives better resolution of individual peaks, but again overlap complicates interpretation. The cleanest information by far comes from combining pure shift acquisition with 1D selective TOCSY editing (Fig. 3c and d). Here the spectra are well resolved and straightforward to analyse. The importance of this sort of separation of the spectra of chemically similar fragments is illustrated in Fig. 4, which shows the H_4β signals. In the pure shift spectrum (Fig. 4b) a clean singlet is observed, that would normally be assigned to a single species. The two selective experiments reveal that there are in fact two separate peaks from different spin systems (Fig. 4c and d) that differ in chemical shift by only a few parts per billion. It is important to note that, just as in conventional 1D and 2D TOCSY, signal integrals are not directly quantitative because the efficiency of magnetization transfer depends on the spin system parameters.

Fig. 3 500 MHz (a) conventional, (b) PSYCHE (2 h 25 min total acquisition time) and (c and d) 1D selective TOCSY-PSYCHE ¹H spectra (2 h 35 min) of a mixture of provitamin D₃ and vitamin D₃ in CDCl₃. The H₃ signals of vitamin D₃ (c) and of provitamin D₃ (d) were selected using 46 ms RSNOB pulses (bandwidth 51 Hz). An isotropic mixing time of 80 ms, a flip angle (β) of 20°, 32 transients and 100 data chunks of 12.5 ms duration were used in (b), (c) and (d).

Fig. 4 Expansions of the spectra in Fig. 3: 500 MHz (a) conventional, (b) PSYCHE and (c and d) 1D selective TOCSY-PSYCHE spectra of a mixture of provitamin D₃ and vitamin D₃ in CDCl₃, showing the region between 2.27 and 2.36 ppm. Using the selective experiments it is possible to differentiate between hydrogens H_4β of both vitamin D₃ (c) and provitamin D₃ (d), even though they are unresolved in the full PSYCHE spectrum (b).

In conclusion, the analysis of complex mixtures by NMR is of wide importance, but is often very frustrating. The new 1D selective TOCSY-PSYCHE experiment allows much more efficient analysis by selecting individual spin systems and recording the spectrum with pure shift resolution. This should allow an order of magnitude more complicated problems to be efficiently investigated.

Acknowledgements

This work was supported by Science Without Borders – Brazil (CNPq reference number 233163/2014-0) and by the Engineering and Physical Sciences Research Council (grant number EP/L018500/1). The authors gratefully acknowledge the assistance of Dr Mohammadali Foroozandeh and Dr Ralph Adams.

Notes and references

1. R. W. Adams, eMagRes, 2014, 3, 295.
2. K. Zangger, Prog. Nucl. Magn. Reson. Spectrosc., 2015, 86–87, 1.
3. L. Castañar and T. Parella, Magn. Reson. Chem., 2015, 53, 399.
4. D. G. Davis and A. Bax, J. Am. Chem. Soc., 1985, 107, 7197.
5. M. Foroozandeh, R. W. Adams, N. J. Meharry, D. Jeannerat, M. Nilsson and G. A. Morris, Angew. Chem., Int. Ed., 2014, 53, 6990.
6. M. Tsai, C. Wu, T. Lin, W. Lin, Y. Huang and C. Yang, Trop. J. Pharm. Res., 2013, 4, 577.
7. H. Kessler, H. Oschkinat, C. Griesinger and W. Bermel, J. Magn. Reson., 1986, 70, 106.
8. L. Braunschweiler and R. R. Ernst, J. Magn. Reson., 1983, 53, 521.
9. M. Foroozandeh, R. W. Adams, M. Nilsson and G. A. Morris, J. Am. Chem. Soc., 2014, 136, 11867.
10. V. M. R. Kakita, S. P. B. Vemulapalli and J. Bharatam, Magn. Reson. Chem., 2016, 54, 308.
11. A. Le Guennec, J. N. Dumez, P. Giraudeau and S. Caldarelli, Magn. Reson. Chem., 2015, 53, 913.
12. V. M. R. Kakita and R. V. Hosur, Magn. Reson. Chem., 2016, 17, 2304.
13. S. P. Rucker and A. J. Shaka, Mol. Phys., 1989, 68, 509.
14. M. J. Thrippleton and J. Keeler, Angew. Chem., Int. Ed., 2003, 42, 3938.
15. J. P. Garbow, D. P. Weitekamp and A. Pines, Chem. Phys. Lett., 1982, 93, 504.
16. N. H. Meyer and K. Zangger, Angew. Chem., Int. Ed., 2013, 52, 7143.
17. L. Castañar, M. Pérez-Trujillo, P. Nolis, E. Monteagudo, A. Virgili and T. Parella, ChemPhysChem, 2014, 15, 854.
18. E. Havinga and J. L. M. A. Schlatmann, Tetrahedron, 1961, 16, 146.

---

Footnotes

† Related pulse sequence code and macros are available from http://nmr.chemistry.manchester.ac.uk/. Full experimental data, pulse sequence code and macros can be downloaded from DOI: 10.15127/1.302716.

‡ Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra22807k

---