Attached Nitrogen Test by ¹³C-¹⁴N Solid-State NMR Spectroscopy for the Structure Determination of Heterocyclic Isomers

Abstract

Differentiation of heterocyclic isomers by solution ¹H, ¹³C and ¹⁵N NMR spectroscopy is often challenging due to similarities in their spectroscopic signatures. Here, ¹³C{¹⁴N} solid-state NMR spectroscopy experiments are shown to operate as an “attached nitrogen test”, where heterocyclic isomers are easy to distinguish based on 1D nitrogen-filtered ¹³C solid-state NMR. We anticipate that these NMR experiments will facilitate the assignment of heterocycle isomers during synthesis and natural product discovery.

Graphical Abstract

The determination of molecular structure is a foundational pillar of synthetic chemistry and natural product discovery. Despite the suite of available techniques to probe molecular structure, we often still “see through a glass, darkly” when assigning spectroscopic data, where the vast possibilities of atomic connectivity may lead to errors in structural assignment. This ambiguity in structural assignment is particularly true in natural product discovery, where common spectroscopic techniques cannot often distinguish between isomeric products. For example, a 2011 review reported over 150 misassigned natural products between 2001–2010.¹ Multiple other reviews have also discussed the misassignment of natural products.^2–4 Several examples of natural products that were misassigned by solution NMR spectroscopy are shown in Figure 1.^5–7

Figure 1.

Selected natural products (1-4) with originally proposed structures (upper) and revised structures (lower, highlighted in green).

Total synthesis of a target molecule is a classical avenue to confirm atomic connectivity and identify errors in originally proposed structures.^2–4 A key drawback, however, is that total synthesis is a time and labor consuming endeavor. Further, even after total synthesis is completed, single crystal X-ray diffraction (SCXRD) is sometimes required to unambiguously determine molecular structure, but diffraction quality single crystals are not obtainable in all cases. Solution NMR spectroscopy is the workhorse method for probing molecular structure within organic molecules and natural products. Nearly all organic systems are suitable for NMR spectroscopy, and isotropic chemical shifts and scalar (J-) couplings reveal unique information on the local chemical environment of the probed nuclei. Two-dimensional (2D) homonuclear and heteronuclear correlation NMR experiments are powerful tools to determine molecular structure. However, heteronuclear correlation solution NMR experiments on organic systems are often limited to ¹H-¹³C, such that ¹³C NMR signal assignment is based solely on ¹³C chemical shifts and ¹H-¹³C scalar (J-) couplings. The assignment of ¹³C NMR signals to a single isomer in systems containing nitrogen heterocycles may become ambiguous when using 2D ¹H-¹H and ¹H-¹³C solution NMR spectroscopy techniques because changes in the nitrogen atom location within a heterocycle often does not alter the observed ¹H-¹H or ¹H-¹³C J-couplings. Indeed, the misassigned natural products shown in Figure 1 differ from their corrected structures by the location and connectivity of the nitrogen atoms.

Thus, information about direct connectivity of carbon and nitrogen atoms would be immensely valuable to discriminate between possible heterocyclic isomers. Here, we report the application of ¹³C{¹⁴N} solid-state NMR experiments that exploit ¹³C-¹⁴N dipolar couplings to identify C atoms directly bonded to N atoms.^8–15 This “attached nitrogen test” requires no isotopic labeling and the working organic chemist will find that such spectra are easily interpretable, akin to the interpretation of NOE difference spectra. We demonstrate the utility of ¹³C{¹⁴N} solid-state NMR spectroscopy for structure determination through model case studies that address the misassignments described in Figure 1. Lastly, we demonstrate the powerful utility of N-filtered ¹³C NMR spectra to aid in the accurate assignment of more complex molecular scaffolds relevant to natural products and pharmaceuticals.

Nitrogen has two NMR active isotopes, ¹⁴N and ¹⁵N, with ¹⁵N being the preferred nucleus to probe in NMR spectroscopy because it is a spin I = 1/2 nucleus, whereas ¹⁴N is spin I = 1 quadrupolar nucleus. Unfortunately, ¹³C-¹⁵N NMR experiments are challenging at natural isotopic abundance (0.004 % probability of having a ¹³C-¹⁵N spin pair) and are sometimes only feasible in concentrated systems and/or with sensitivity enhancement techniques, such as dynamic nuclear polarization (DNP).^{16–22 13}C-¹⁴N NMR experiments are attractive because ¹⁴N is 99.6 % abundant. However, the quadrupolar nature of ¹⁴N means that one-bond ¹³C-¹⁴N J-couplings (¹J ~ 10–15 Hz) often cannot be observed in solution NMR spectra due to the self-decoupling of ¹⁴N that occurs because of the continuous alternation of the ¹⁴N spin states by rapid longitudinal (T₁) relaxation.

Fortunately, ¹⁴N can be readily probed in solid-state NMR experiments. Here, we use the ¹³C{¹⁴N} Resonance Echo Saturation Pulse DOuble Resonance (RESPDOR) NMR experiment to obtain 1D N-filtered ¹³C NMR spectra (Figure 2A). We note that ¹³C{¹⁴N} solid-state NMR experiments have been used for over two decades to obtain structural information in organic and biomolecular systems.^8–15 However, the goal of our work is to demonstrate the value and simplicity of ¹³C{¹⁴N} solid-state NMR experiments to the practicing chemist to differentiate heterocyclic isomers.

Figure 2.

(A) ¹³C{¹⁴N} PM-RESPDOR pulse sequence. (B) ¹³C{¹⁴N} RESPDOR curves for the ¹³C NMR signals of Hist at (orange) 173 ppm and (green) 128 ppm. The experimental data points are shown as circles and numerical simulations are shown as solid lines. (C) ¹³C{¹⁴N} RESPDOR spectra of Hist recorded (red, dashed) with or (black, solid) without a ¹⁴N PM saturation pulse. The difference spectrum is shown below.

We first optimized the experimental conditions for the “attached nitrogen” ¹³C{¹⁴N} RESPDOR experiments using histidine hydrochloride monohydrate (Hist) as a model compound. In the ¹³C{¹⁴N} RESPDOR experiment, two ¹³C NMR spectra are recorded; one with and one without a ¹⁴N phase-modulated (PM) saturation pulse.^23–24 Taking the difference of the two NMR spectra yields an N-filtered ¹³C NMR spectrum because the ¹³C NMR signal will have reduced intensity when pulsing on ¹⁴N if it is covalently bonded to N (see SI for more discussion). Experiments on Hist showed that ca. 1.3 ms of ¹³C REDOR recoupling (τ_rec) was optimal for maximizing the difference ¹³C NMR signal for C atoms covalently bonded to N atoms and minimizing dephasing for C atoms not bonded to N atoms (Figure 2B and S35). For Hist, a total of 1 hour of spectrometer time was required to obtain the ¹⁴N-filtered ¹³C NMR spectrum that shows only ¹³C NMR signals from C atoms exhibiting C-N covalent bonds (Figure 2C). We note that similar “attached nitrogen” ¹³C{¹⁴N} NMR spectra can be recorded without ¹³C dipolar recoupling to cause signal dephasing by evolution of ¹³C-¹⁴N J-couplings and residual dipolar splittings (Figure S36).^25–26 However, the RESPDOR experiment will generally be more sensitive because the dipolar coupling is over one order of magnitude larger than the J-coupling and residual dipolar splitting (Figure S37, see SI for more discussion). For Hist, the RESPDOR experiment with dipolar recoupling was ca. two times more sensitive (SNR min^−1/2) than the analogous experiment without dipolar recoupling.

The core atoms of heterocycles typically do not display characteristic ¹H or ¹³C chemical shifts that would allow for their straightforward identification through 1D ¹H/¹³C NMR spectroscopy without prior knowledge of chemical shifts (Figure S38). Even with modern 2D ¹H homonuclear and/or ¹H-¹³C heteronuclear correlation solution NMR experiments, spectral interpretation is still often ambiguous due to similarities in the observed correlations for different isomers, making it challenging to differentiate isomers on unknown, highly substituted heterocyclic systems (see SI for solution NMR spectra).

For example, the structure of Aspernigrin A (1) was initially assigned via ¹H{¹³C} HMBC but was later corrected by SCXRD (Figure 1).^{5, 27–28} The ambiguity in assigning the 2- or 4-pyridone heterocyclic cores could have been easily addressed with an “attached nitrogen” ¹³C{¹⁴N} RESPDOR NMR experiment. To test this hypothesis, we recorded ¹³C{¹⁴N} RESPDOR NMR spectra of acridinone (5) and phenanthridinone (6) as model compounds for Aspernigrin A (Figure 3A–B).

Figure 3.

Comparison of ¹³C{¹⁴N} RESPDOR NMR spectra of (A) acridinone (5), (B) phenanthridinone (6), (C) pyrazole (7), (D) imidazole (8), (E) oxazole (10) and (F) isoxazole (11). ¹³C{¹⁴N} RESPDOR spectra were recorded with 1.28 ms of total dipolar recoupling and (red, dashed) with or (black, solid) without a ¹⁴N PM saturation pulse. The difference spectrum is shown below. NMR signals correspond to the highlighted C atoms on the structures.

The ¹⁴N-filtered ¹³C NMR spectra allow for clear differentiation of the 2- versus 4-pyridone core. Isomer 6 displays two ¹³C NMR signals exhibiting a C-N covalent bond (Figure 3B). Importantly, one of the ¹³C NMR signals attached to N in 6 clearly shows a diagnostic chemical shift associated with a carbonyl carbon (> 150 ppm), while that of 5 does not. 5 displays only one ¹³C NMR signal exhibiting a C-N covalent bond due to the C₂ symmetry of the compound (Figure 3A). In the more substituted Aspernigin A (1a and 1b), the corrected structure will show two carbonyl ¹³C NMR signals in the ¹⁴N-filtered ¹³C NMR spectrum as opposed to one in the misassigned structure (Figure 1).

We next examined the differentiation of azole-type heterocycles using the “attached nitrogen” ¹³C{¹⁴N} RESPDOR experiment. Model heterocycles pyrazole (7) and imidazole (8) form the core of numerous drug scaffolds (Figure 3C–D). Particularly for highly substituted rings, ¹H-¹³C correlations can be ambiguous, and without prior knowledge of ¹H and/or ¹³C chemical shifts, the assignment of the ¹H/¹³C NMR spectra to a single isomer can lead to error (Figure S38). Alternatively, comparison of the ¹⁴N-filtered ¹³C NMR spectra enable the easy assignment of the two heterocyclic isomers, where 7 displays two ¹³C NMR signals exhibiting C-N covalent bonds, while imidazole 8 displays three ¹³C NMR signals exhibiting C-N covalent bonds (Figure 3C–D).

Differentiation of isoxazole and oxazole heterocyclic isomers provide a compelling illustration of the utility of the “attached nitrogen” ¹³C{¹⁴N} RESPDOR NMR experiment. Even when assisted by 2D solution NMR experiments, conclusive assignment of the oxazole can be elusive without comparison to the isoxazole isomer (and vice versa), a form of structure determination by total synthesis (Figure S3–5 and S8–10). Alarmingly, however, such an approach is more complicated when one considers that both the oxazole and isoxazole can be prepared from the same starting ketoxime (9, Scheme 1).²⁹

Scheme 1.

Preparation of Oxazole and Isoxazole Heterocyclic Isomers from Ketoxime.

The ¹⁴N-filtered ¹³C NMR spectra of oxazole (10) and isoxazole (11) enable the straightforward differentiation of the heterocyclic isomers (Figure 3E–F). 10 displays two ¹³C NMR signals with C-N bonds, while 11 only exhibits one ¹³C NMR signal with a C-N bond.

Finally, to illustrate the ability of “attached nitrogen” ¹³C{¹⁴N} RESPDOR experiments to aid in the determination of more complex molecules relevant to natural products and pharmaceuticals, we performed experiments on a multi-component API, where the free-base molecule forms a co-crystal with phosphoric acid (12, Figure 4A).³⁰ The ¹³C{¹⁴N} RESPDOR experiments were performed with either conventional NMR at room temperature or with dynamic nuclear polarization (DNP) at ca. 100 K. In a DNP experiment, the NMR signal intensity is enhanced by 1–2 orders of magnitude by transferring the polarization of electron spins from a polarizing agent (e.g., TEKPol) to the nuclear spins. ¹H→¹³C CPMAS DNP enhancements were ≥ 10, meaning that a ¹H→¹³C CPMAS NMR spectrum with the same signal-to-noise ratio could be acquired ca. 100 times faster with DNP than conventional room temperature NMR spectroscopy (Figure S39).

Figure 4.

(A) Crystal structure of co-crystal 12. H, C, N, O, F, P and S atoms are white, grey, blue, red, green, orange and yellow, respectively. (B) DNP-enhanced ¹³C{¹⁴N} RESPDOR spectra recorded with 1.2 ms of dipolar recoupling and (red) with or (black) without a ¹⁴N saturation pulse. The difference spectrum is shown below. (C) DNP-enhanced ¹³C{¹⁴N} RESPDOR curves of co-crystal 12 (see Figure S42 for all RESPDOR curves). The circles correspond to the experimental data points and the solid lines correspond to numerical simulations.

1D ¹⁴N-filtered ¹³C NMR spectra of 12 were obtained in ca. 40 min with DNP and a 3.2 mm rotor or ca. 17 hours with conventional room temperature solid-state NMR spectroscopy and 2.5 mm rotor (Figure 4B and S40, respectively). The 1D ¹⁴N-filtered ¹³C NMR spectrum reveals all ¹³C NMR signals exhibiting C-N covalent bonds. We note that C14 has reduced intensity in the DNP spectrum due to ¹³C signal overlap with the DNP solvent (see SI for more discussion).

The large sensitivity gains provided by DNP also enabled the acquisition of ¹³C{¹⁴N} RESPDOR curves that provide detail as to the rate of signal build-up and the extent of signal dephasing (Figure 4C and S42). In turn, the shape of these curves are dependent on the type and number of nitrogen atoms within a ca. 4 Å radius (Table S2). Therefore, fitting of the experimental ¹³C{¹⁴N} RESPDOR curves with numerical simulations facilitates the assignment of all ¹³C NMR signals spatially proximate to nitrogen atoms (Figure 4C and S42). This capability permits complete ¹³C signal assignment by comparing the ¹³C{¹⁴N} RESPDOR curves with 1 ¹³H- C heteronuclear correlation NMR spectra and plane-wave DFT GIPAW³¹ calculated ¹³C chemical shifts (Figure S43–45).

In conclusion, the determination of molecular structure is a foundational pillar of organic synthesis and natural product discovery. However, typical ¹H-¹H and ¹H-¹³C scalar (J-) correlation solution NMR experiments reveal the same homonuclear/heteronuclear correlations for many heterocyclic isomers, meaning that spectroscopic assignment to a single isomer without prior knowledge of ¹H and/or ¹³C chemical shifts is often ambiguous. Here, ¹³C{¹⁴N} RESPDOR solid-state NMR spectroscopy experiments are shown to enable the easy acquisition of 1D ¹⁴N-filtered ¹³C solid-state NMR spectra, which effectively operates as an “attached nitrogen test”.

The practical utility of ¹³C{¹⁴N} RESPDOR solid-state NMR spectroscopy experiments to differentiate heterocyclic isomers was demonstrated for three different model systems. In all three examples, the heterocyclic isomers could be easily distinguished from the ¹⁴N-filtered ¹³C NMR spectra, where 1D and 2D ¹H and ¹³C correlation solution NMR spectroscopy were ambiguous, particularly if one did not have prior knowledge of the molecular structure. We also demonstrated how ¹⁴N-filtered ¹³C NMR spectra can aid in the structural characterization of more complex molecular scaffolds relevant to natural products and pharmaceuticals.

We anticipate that ¹³C{¹⁴N} RESPDOR solid-state NMR spectroscopy experiments will provide practicing chemists with a simple method to obtain 1D ¹⁴N-filtered ¹³C solid-state NMR spectra that can greatly aid in ¹³C NMR signal assignment and differentiation of heterocyclic isomers. 1D ¹³C{¹⁴N} NMR experiments on 12 were performed at room temperature with 2.5 mm rotors, ca. 20 mg of material and 17 hours of spectrometer time. Therefore, even in the absence of sensitivity enhancement by DNP, ¹³C{¹⁴N} NMR experiments can be feasibly applied to molecules with comparable size and complexity as natural products. Therefore, “attached nitrogen tests” could be especially useful in natural product discovery because they will reduce structural ambiguities and misassignments.