Lighting up spin systems: enhancing characteristic ¹H signal patterns of fluorinated molecules

Abstract

Fluorine is becoming increasingly prevalent in medicinal chemistry, both in drug molecules and in molecular probes. The presence of fluorine allows convenient monitoring of such molecules in complex environments by NMR spectroscopy. However, sensitivity is a persistent limitation of NMR, especially when molecules are present at low concentrations. Here, sensitivity issues with ¹H NMR are mitigated by sharing ¹⁹F photochemically-induced dynamic nuclear polarisation with ¹H nuclei. Unlike direct ¹H enhancement, this method enhances ¹H signals without significantly distorting multiplet intensities, and has the potential to enable the use of suitable molecules as low-concentration probes.

---

Fluorinated molecules account for over 20% of pharmaceuticals currently available, and are commonly found in compound libraries used for drug screening.^1,2 Fluorinated species are also increasingly used as probes for studying biological processes, due to the high sensitivity of ¹⁹F chemical shifts to the local environment, the absence of endogenous background signals, and 100% natural abundance.^3–7 However, structural information obtained by direct observation of sparse fluorine atoms is limited.

¹H NMR can provide much more information, by reporting multiple signals from the same molecule, and can be used to characterise the behaviour of a probe molecule in a complex system.^8,9 Unfortunately, ¹H spectra often suffer from signal overlap, especially in mixtures of protonated molecules, due to the limited chemical shift range and prominent signal multiplicity, making disentangling the spectra of individual species difficult. One strategy for mitigating the limitations of both nuclei is fluorine-edited selective TOCSY acquisition (FESTA), which exploits the exceptional chemical shift dispersion of ¹⁹F spectra to acquire ¹H spectra containing only ¹H signals that are within the same spin system as a selected ¹⁹F nucleus.¹⁰ FESTA has been previously demonstrated to be particularly powerful in mixture analysis, allowing characteristic ¹H subspectra of fluorinated components to be obtained.^10–12 Unfortunately, the beneficial ability to observe the characteristic ¹H fingerprint patterns of particular fluorinated molecules comes with a sensitivity penalty, which may be critical if molecules are present at low concentrations. Here we introduce a method to address this.

Strategies to improve the sensitivity of NMR include using very high magnetic fields, cryoprobes,¹³ parahydrogen-induced polarisation (PHIP),^14,15 and dynamic nuclear polarisation (DNP)^16–18 techniques. Unfortunately, these methods generally come at substantial cost and require additional complex hardware, such as gyrotrons and dissolution apparatus or parahydrogen generators. A potentially more convenient and relatively low-cost alternative to these approaches is photochemically-induced dynamic nuclear polarization (photo-CIDNP)^19–21 which, upon simple sample illumination, even using cheap LEDs, can lead to large signal enhancements in target molecules with low ionisation potentials, such as aromatic molecules. The relative enhancements obtained are larger for lower magnetic fields, so photo-CIDNP lends itself well to less expensive hardware, where the increase in sensitivity is at its most welcome.²²

Illumination (historically with lasers, more recently with inexpensive LEDs)^23–25 of a photosensitiser, such as fluorescein or flavin, present in the sample leads to a non-Boltzmann distribution of nuclear spins in a target molecule via a radical pair mechanism.²⁶ A hyperpolarised signal arises because there is an overpopulation of nuclear spin states which experience faster intersystem crossing and less efficient paramagnetic relaxation.²⁷ However, hyperpolarisation only occurs for nuclei that have a significant hyperfine coupling to an unpaired electron. Other signals from the molecule are not enhanced and, disappointingly, signal intensities may even be reduced, as the polarisation produced by photo-CIDNP can be absorptive or emissive.²⁸ The net result is to distort the characteristic signal pattern of a probe molecule, potentially making the spectrum unrecognisable. These signal distortions and losses hinder the application of direct ¹H photo-CIDNP as a signal enhancement approach for molecular detection and monitoring in mixtures. It has been noted, however, that there are significant benefits to using heavier heteroatoms such as ¹⁹F, ¹³C or ¹⁵N, which may result in greater hyperpolarization.^29–33 Given the ability to transfer magnetization from heteronuclei to protons, sharing such hyperpolarisation enables more sensitive heteronuclear correlation experiments, or can highlight through-space interactions between heteronuclei and neighbouring protons.^17,31,34

In this study, we demonstrate how the benefits of FESTA¹⁰ can be combined with those of photo-CIDNP to provide significant ¹H signal enhancements across a spin system that includes fluorine. This enables observation of the characteristic ¹H signals of a molecule with enhanced sensitivity. Using 6-fluoroindole (6FI), a common reagent in the preparation of fluorinated amino acids and antifungal and antibacterial agents,^35–37 as a model we show that the direct photo-CIDNP effect is here much greater for ¹⁹F than for ¹H (Fig. 1d and c). Transferring the large ¹⁹F hyperpolarisation to ¹H using FESTA demonstrates that multiple ¹H signals in the same spin system as fluorine can be greatly enhanced, with much less signal distortion than in direct ¹H photo-CIDNP (Fig. 1g and c). This approach also would allow subspectra for specific fluorinated molecules in complex mixtures to be obtained, using a frequency-selective shaped ¹⁹F pulse for the initial excitation. The recently proposed NMRtorch approach³⁸ was used to illuminate a sample containing 1 mM 6FI and a flavin mononucleotide as a photosensitizer prior to data acquisition. Matching control spectra were recorded without illumination.

Fig. 1 (a) 500 MHz ¹H and (b) 470 MHz ¹⁹F NMR spectra of 1 mM 6FI with 0.2 mM riboflavin 5′-monophosphate sodium salt in D₂O. (c) ¹H and (d) ¹⁹F NMR spectra with 2 s illumination using a single LED with nominal 460 nm peak emission. Example ¹H FESTA spectra obtained (e) in the dark and (f) with 2 s illumination using the pulse sequence of Fig. 2 selectively exciting H7 (7.18 ppm). (g) Summation of three ¹H FESTA spectra where H4, H5 and H7 were selected (further information can be found in the ESI†). Spectra (d), (f) and (g) were emissive, and are shown inverted with respect to the dark spectra to facilitate comparison. All spectra are scaled consistently for a given nucleus.

As 6FI is only present at 1 mM concentration, ¹H and FESTA NMR spectra recorded in the dark (Fig. 1a and e) suffer from limited signal-to-noise ratio. The ¹H NMR spectrum acquired with illumination (i.e., direct ¹H photo-CIDNP) showed a 32-fold absorptive signal enhancement of the integral of H2 (Fig. 1c). However, other ¹H signals in the molecule showed only modest signal enhancements (<5-fold) and, importantly, the multiplet patterns of H4, H5 and H7 were all significantly distorted (Fig. S2, ESI†), complicating the recognition of this spectrum as belonging to 6FI. ¹⁹F photo-CIDNP (Fig. 1d), on the other hand, resulted in 83-fold emissive signal enhancement compared to the corresponding dark reference spectrum (Fig. 1b).

As only certain 6FI ¹H signals are enhanced by direct ¹H photo-CIDNP, methods of sharing the benefits of hyperpolarisation amongst multiple ¹H signals were explored. Initially, ¹H selective TOCSY^39,40 (as previously proposed with photo-CIDNP by Goez et al.)⁴¹ was trialled to share the hyperpolarisation of H2 around the coupled ¹H network. However, the signal enhancement was only efficiently shared between H2 and H1, and gave only ∼12-fold enhancement of each with the optimum mixing time (Fig. S3, ESI†). The distant aromatic protons of 6FI (H4, H5 and H7) showed no significant enhancement (Fig. S3, ESI†), because the small inter-ring couplings J_HH (Table S2, ESI†) severely limit the effectiveness of TOCSY transfer.

To obtain more uniform enhancement across multiple ¹H nuclei, the use of a new approach, Light-FESTA (Fig. 2), was investigated. This exploits the strong ¹⁹F photo-CIDNP enhancement and the ability of FESTA to transfer that enhancement across a ¹H spin system with minimal multiplet distortion.¹⁰ In Light-FESTA the selection of different ¹⁹F-¹H coupling partners, i.e., transferring F6 magnetisation to H4, H5 or H7, leads to preferential enhancement of different signals (Table 1). However, summing the spectra from these three separate Light-FESTA experiments into a single Σ-Light-FESTA spectrum (Fig. 1g) leads to a more even distribution of signal intensities for H4, H5 and H7. This makes the spectrum easily recognisable as that of 6FI and allows straightforward signal comparison with the reference¹H NMR spectrum (Fig. 1a). This contrasts with the direct ¹H photo-CIDNP spectrum (Fig. 1c), which is barely recognisable, with the signal intensities for H4, H5 and H7 being lower and their multiplet structures severely distorted.

Fig. 2 Light-FESTA pulse sequence. The narrow black and wide grey rectangles represent hard 90° and 180° radiofrequency pulses, respectively. The shaped pulses represent selective 180° radiofrequency pulse applied to a chosen ¹H resonance coupled to ¹⁹F. The trapezoids on either side of the isotropic mixing element (DIPSI-2) represent zero-quantum coherence suppression elements.⁴⁰ Field gradient pulses are employed to enforce the coherence transfer pathway. Illumination (hν) is applied during the recovery delay, d1. If more than one fluorine resonance is present, the first ¹⁹F pulse should be frequency-selective. Further details are provided in the ESI.†

Table 1 Enhancements of signal integrals of 6-fluoroindole (1 mM) after 2 s illumination in the presence of riboflavin 5′-monophosphate sodium salt. Enhancement factors are provided for direct ¹H photo-CIDNP; three Light-FESTA experiments, selecting each in turn of the ¹H signals coupled to ¹⁹F (H4, H5 and H7); and the result of summing all three Light-FESTA experiments (Σ-Light-FESTA). Enhancement factors were calculated using absolute integrals relative to the dark ¹H NMR spectrum

	H4	H1	H7	H5	H2
Direct ^1H photo-CIDNP	4.0	3.3	2.1	1.4	32.4
Light-FESTA H5 (6.86 ppm)	5.0	0.3	7.2	16.6	0.8
Light-FESTA H7 (7.16 ppm)	14.9	2.9	4.1	7.2	0.5
Light-FESTA H4 (7.54 ppm)	6.1	0.4	10.9	3.7	0.7
Σ-Light-FESTA	8.7	1.2	7.4	9.2	0.7

---

Light-FESTA provides a significant signal enhancement (Fig. 1f) compared to standard FESTA (Fig. 1e), with minimal multiplet distortion. The background noise level was unaffected by illumination with NMRtorch.³⁸ The average sensitivity enhancement (defined as enhancement in the individual signal-to-noise ratios) was 34-fold for the H4–H5–H7 spin system (Table S3, ESI†), translating into a potential 1000-fold experimental time saving. Overall, Σ-Light-FESTA gives rise to >7 times more intense signals for the H4–H5–H7 spin system compared with a standard ¹H NMR spectrum recorded without illumination (Table 1) and is up to 49 times more sensitive than a standard FESTA experiment (Table S3, ESI†). Extra benefits include the ability to select a molecule of interest in complex mixtures, by applying selective ¹⁹F and ¹H pulses at defined frequencies. As the transfer of magnetisation to other ¹H nuclei within a spin system is dependent on the isotropic mixing time (Fig. S4, ESI†), running several experiments with judicious choices of mixing times and summing the resulting spectra can lead to more even distribution of hyperpolarisation.

Finally, although 1D selective TOCSY did not yield satisfactory results for 6FI, in cases where there is an extensive scalar coupling network with appreciable J_HH values, a selective TOCSY experiment may be advantageous. Another interesting possibility, currently being explored, is to design a pulse sequence that pools the ¹H and the ¹⁹F hyperpolarisation, sharing both among their neighbours.

In summary, we have shown that ¹⁹F and direct ¹H photo-CIDNP enhancements can be shared amongst multiple ¹H nuclei within the same spin system, using FESTA or selective TOCSY, with the former being much more effective. Transferring the greater hyperpolarisation of ¹⁹F to ¹H led to the enhancement of more signals, a more even distribution of signal intensities, and, in contrast to direct ¹H photo-CIDNP, preservation of normal multiplet structure. This makes the molecular spectral fingerprint easily recognisable. Light-FESTA not only eliminated the sensitivity penalty of the parent FESTA experiment, with a 34-fold enhancement, but actually increased the sensitivity of proton detection sevenfold compared to a standard ¹H NMR spectrum without illumination. Light-FESTA should therefore allow the analysis of molecules at low concentrations while retaining the selectivity advantages of FESTA. This approach will be particularly powerful when investigating fluorinated molecules known to show a photo-CIDNP effect. Such molecules could be used as probes to investigate complex system behaviour, or as constituents of compound libraries for drug screening.^42,43 Sample illumination here used the LED-based NMRtorch, a recent general tool for photo-NMR.³⁸ The approach presented in this paper can readily be extended to other spectral editing techniques, for example, to observe enhanced through-space interactions, determine molecular proximity, or quantify ligand–receptor interactions, as required in drug screening.^34,44 Although demonstrated using ¹⁹F nuclei, Light-FESTA could be extended to achieve selective photo-CIDNP enhancement of ¹H using other heteronuclei such as ¹³C or ¹⁵N.^31,33

The NMRtorch hardware was manufactured by the last author (APG). All authors contributed to the design of the experiments and pulse sequences, the analysis of the results, and the writing of the manuscript.

The authors gratefully acknowledge the Engineering and Physical Sciences Research Council (grant numbers EP/R018790/1, EP/V04835X/1, and EP/R513131/1 Project Reference 2297284) and the University of Manchester (Dame Kathleen Ollerenshaw Fellowship to L. C.) and the Comunidad de Madrid (grant number 2022-T1/BMD-24030 to L. C.) for supporting this work. We acknowledge the use of the Manchester Biomolecular NMR Facility and are grateful to Matthew Cliff for help with NMR equipment. For the purpose of open access, the authors have applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising. All NMR experimental data, pulse sequence codes for Bruker spectrometers, and processing macro are freely available at https://doi.org/10.48420/23056013.

Conflicts of interest

APG is the named author in patent application PCT/GB2021/051254, International Publication Number WO 2021/250372 A1. All other authors declare no competing interests.

References

1. M. Inoue, Y. Sumii and N. Shibata, ACS Omega, 2020, 5, 10633–10640.
2. I. Hyohdoh, N. Furuichi, T. Aoki, Y. Itezono, H. Shirai, S. Ozawa, F. Watanabe, M. Matsushita, M. Sakaitani, P.-S. Ho, K. Takanashi, N. Harada, Y. Tomii, K. Yoshinari, K. Ori, M. Tabo, Y. Aoki, N. Shimma and H. Iikura, ACS Med. Chem. Lett., 2013, 4, 1059–1063.
3. H. Chen, S. Viel, F. Ziarelli and L. Peng, Chem. Soc. Rev., 2013, 42, 7971.
4. J.-X. Yu, R. R. Hallac, S. Chiguru and R. P. Mason, Prog. Nucl. Magn. Reson. Spectrosc., 2013, 70, 25–49.
5. J. M. Edwards, J. P. Derrick, C. F. Van Der Walle and A. P. Golovanov, Mol. Pharmaceutics, 2018, 15, 2785–2796.
6. H.-L. Bao and Y. Xu, Chem. Commun., 2020, 56, 6547–6550.
7. A. M. Gronenborn, Structure, 2022, 30, 6–14.
8. Z. Xu, C. Liu, S. Zhao, S. Chen and Y. Zhao, Chem. Rev., 2019, 119, 195–230.
9. Q. Liu, Q.-T. He, X. Lyu, F. Yang, Z.-L. Zhu, P. Xiao, Z. Yang, F. Zhang, Z.-Y. Yang, X.-Y. Wang, P. Sun, Q.-W. Wang, C.-X. Qu, Z. Gong, J.-Y. Lin, Z. Xu, S.-L. Song, S.-M. Huang, S.-C. Guo, M.-J. Han, K.-K. Zhu, X. Chen, A. W. Kahsai, K.-H. Xiao, W. Kong, F.-H. Li, K. Ruan, Z.-J. Li, X. Yu, X.-G. Niu, C.-W. Jin, J. Wang and J.-P. Sun, Nat. Commun., 2020, 11, 4857.
10. L. Castañar, P. Moutzouri, T. M. Barbosa, C. F. Tormena, R. Rittner, A. R. Phillips, S. R. Coombes, M. Nilsson and G. A. Morris, Anal. Chem., 2018, 90, 5445–5450.
11. T. M. Barbosa, L. Castañar, P. Moutzouri, M. Nilsson, G. A. Morris, R. Rittner and C. F. Tormena, Anal. Chem., 2020, 92, 2224–2228.
12. G. Dal Poggetto, J. V. Soares and C. F. Tormena, Anal. Chem., 2020, 92, 14047–14053.
13. H. Kovacs, D. Moskau and M. Spraul, Prog. Nucl. Magn. Reson. Spectrosc., 2005, 46, 131–155.
14. S. B. Duckett and R. E. Mewis, Acc. Chem. Res., 2012, 45, 1247–1257.
15. W. Iali, P. J. Rayner and S. B. Duckett, Sci. Adv., 2018, 4, eaao6250.
16. A. Abragam and M. Goldman, Rep. Prog. Phys., 1978, 41, 395–467.
17. J. H. Lee, Y. Okuno and S. Cavagnero, J. Magn. Reson., 2014, 241, 18–31.
18. B. Plainchont, P. Berruyer, J.-N. Dumez, S. Jannin and P. Giraudeau, Anal. Chem., 2018, 90, 3639–3650.
19. R. Kaptein, K. Dijkstra, F. Müller, C. G. Van Schagen and A. J. W. G. Visser, J. Magn. Reson. (1969-1992), 1978, 31, 171–176.
20. M. Mompeán, R. M. Sánchez-Donoso, A. De La Hoz, V. Saggiomo, A. H. Velders and M. V. Gomez, Nat. Commun., 2018, 9, 108.
21. Y. Okuno, M. F. Mecha, H. Yang, L. Zhu, C. G. Fry and S. Cavagnero, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 11602–11611.
22. J. Bernarding, C. Bruns, I. Prediger and M. Plaumann, Appl. Magn. Reson., 2022, 53, 1375–1398.
23. P. Nitschke, N. Lokesh and R. M. Gschwind, Prog. Nucl. Magn. Reson. Spectrosc., 2019, 114–115, 86–134.
24. J. Bernarding, F. Euchner, C. Bruns, R. Ringleb, D. Müller, T. Trantzschel, J. Bargon, U. Bommerich and M. Plaumann, Chem. Phys. Chem., 2018, 19, 2453–2456.
25. H. Yang, H. Hofstetter and S. Cavagnero, J. Chem. Phys., 2019, 151, 245102.
26. G. L. Closs, J. Am. Chem. Soc., 1969, 91, 4552–4554.
27. J. Hore and R. W. Broadhurst, Prog. Nucl. Magn. Reson. Spectrosc., 1993, 25, 345–402.
28. R. Kaptein, J. Chem. Soc. D, 1971, 732–733.
29. C. E. Lyon, J. A. Jones, C. Redfield, C. M. Dobson and P. J. Hore, J. Am. Chem. Soc., 1999, 121, 6505–6506.
30. F. Khan, I. Kuprov, T. D. Craggs, P. J. Hore and S. E. Jackson, J. Am. Chem. Soc., 2006, 128, 10729–10737.
31. J. H. Lee, A. Sekhar and S. Cavagnero, J. Am. Chem. Soc., 2011, 133, 8062–8065.
32. Y. Okuno and S. Cavagnero, J. Magn. Reson., 2018, 286, 172–187.
33. A. Sekhar and S. Cavagnero, J. Phys. Chem. B, 2009, 113, 8310–8318.
34. I. Kuprov and P. J. Hore, J. Magn. Reson., 2004, 168, 1–7.
35. R. Curtis-Marof, D. Doko, M. L. Rowe, K. L. Richards, R. A. Williamson and M. J. Howard, Org. Biomol. Chem., 2014, 12, 3808–3812.
36. O. N. Burchak, E. L. Pihive, L. Maigre, X. Guinchard, P. Bouhours, C. Jolivalt, D. Schneider, M. Maurin, C. Giglione, T. Meinnel, J.-M. Paris and J.-N. Denis, Biorg. Med. Chem., 2011, 19, 3204–3215.
37. Y.-M. Na, Bull. Korean Chem. Soc., 2010, 31, 3467–3470.
38. J. E. Bramham and A. P. Golovanov, Commun. Chem., 2022, 5, 90.
39. H. Kessler, H. Oschkinat, C. Griesinger and W. Bermel, J. Magn. Reson. (1969-1992), 1986, 70, 106–133.
40. M. J. Thrippleton and J. Keeler, Angew. Chem., Int. Ed., 2003, 42, 3938–3941.
41. M. Goez, K. H. Mok and P. J. Hore, J. Magn. Reson., 2005, 177, 236–246.
42. F. Torres, A. Renn and R. Riek, Magn. Reson., 2021, 2, 321–329.
43. F. Torres, M. Bütikofer, G. R. Stadler, A. Renn, H. Kadavath, R. Bobrovs, K. Jaudzems and R. Riek, J. Am. Chem. Soc., 2023, 145, 12066–12080.
44. P. Lameiras, S. Patis, J. Jakhlal, S. Castex, P. Clivio and J.-M. Nuzillard, Chem. – Eur. J., 2017, 23, 4923–4928.

---

Footnote

† Electronic supplementary information (ESI) available: Containing further experimental data, guidance on acquiring experiments and pulse sequence code for Bruker spectrometers. See DOI: https://doi.org/10.1039/d3cc03557c

---