15N-Labelling and structure determination of adamantylated azolo-azines in solution

Determining the accurate chemical structures of synthesized compounds is essential for biomedical studies and computer-assisted drug design. The unequivocal determination of N-adamantylation or N-arylation site(s) in nitrogen-rich heterocycles, characterized by a low density of hydrogen atoms, using NMR methods at natural isotopic abundance is difficult. In these compounds, the heterocyclic moiety is covalently attached to the carbon atom of the substituent group that has no bound hydrogen atoms, and the connection between the two moieties of the compound cannot always be established via conventional 1H-1H and 1H-13C NMR correlation experiments (COSY and HMBC, respectively) or nuclear Overhauser effect spectroscopy (NOESY or ROESY). The selective incorporation of 15N-labelled atoms in different positions of the heterocyclic core allowed for the use of 1H-15N (JHN) and 13C-15N (JCN) coupling constants for the structure determinations of N-alkylated nitrogen-containing heterocycles in solution. This method was tested on the N-adamantylated products in a series of azolo-1,2,4-triazines and 1,2,4-triazolo[1,5-a]pyrimidine. The syntheses of adamantylated azolo-azines were based on the interactions of azolo-azines and 1-adamatanol in TFA solution. For azolo-1,2,4-triazinones, the formation of mixtures of N-adamantyl derivatives was observed. The JHN and JCN values were measured using amplitude-modulated 1D 1H spin-echo experiments with the selective inversion of the 15N nuclei and line-shape analysis in the 1D 13С spectra acquired with selective 15N decoupling, respectively. Additional spin–spin interactions were detected in the 15N-HMBC spectra. NMR data and DFT (density functional theory) calculations permitted to suggest a possible mechanism of isomerization for the adamantylated products of the azolo-1,2,4-triazines. The combined analysis of the JHN and JCN couplings in 15N-labelled compounds provides an efficient method for the structure determination of N-alkylated azolo-azines even in the case of isomer formation. The isomerization of adamantylated tetrazolo[1,5-b][1,2,4]triazin-7-ones in acidic conditions occurs through the formation of the adamantyl cation.


Introduction
The incorporation of an adamantyl moiety in bioactive molecules and analogues of natural compounds is a widely used approach in medicinal chemistry [1]. The increased lipophilicity of adamantane-containing compounds compared with nonadamantylated derivatives [2] leads to considerably higher solubility of these compounds in blood plasma and their easier penetration through cell membranes. The conjugation of adamantane with heterocyclic compounds also provides a method to modify the pharmacological profile and frequently leads to a new type of bioactivity. For example, N-adamantyl tetrazoles 1 and 2 ( Figure 1A) demonstrate lower toxicity and, simultaneously, more potent activity against influenza A virus compared with the currently used antiviral drug rimantadine (1-(1-adamantyl)ethanamine) [3]. More recently, Roberge et al. described new inhibitors of the influenza A virus M2 proton channel. Among the studied compounds, adamantyl imidazole 3 showed good activity [4]. An azolo-azine core with a bridgehead nitrogen atom is found in many natural products [5,6] and biologically active synthetic compounds [7,8]. The purine-like scaffold of these nitrogencontaining heterocycles is frequently used in medicinal chemistry and drug design. For example, 6-nitro-1,2,4-triazolo- [5,1-c] [1,2,4]triazine 4 ( Figure 1A, Triazavirin ® ) was approved in Russia for the treatment of influenza [9]. This drug targets the viral protein haemagglutinin. The incorporation of an adamantyl moiety in azolo-azine structures could lead to the development of new multifunctional antiviral drugs.
Previously, we synthesized N-adamatylated derivatives of 1,2,4-triazolo [5,1-c] [1,2,4]triazines 5 and 6 by reaction with the adamantyl cation generated from 1-adamantanol in acidic medium [10]. The azolo-azine scaffold of these compounds has several nitrogen atoms that can react with alkylation reagents [11,12] ( Figure 1B). For this reason, the adamantylation of compounds 5 and 6 led to mixtures of N3-and N4-adamantylated isomers, which reisomerized into each other likely via the formation of an adamantyl cation and starting NH-heterocycle. The unambiguous determination of N-adamantylation site(s) in heterocycles 5 and 6 using well-established 1 H and 13 C NMR methods (such as 1D, 2D COSY, HMQC, HMBC, and INADE-QUATE spectra) was difficult because the heterocyclic moiety was covalently attached to the adamantane tertiary carbon that had no bound hydrogen atoms. Nuclear Overhauser effect spectroscopy (NOESY or ROESY) also did not provide unequivocal structures of the N-adamantylated derivatives [13,14]. For example, the attachment of an adamantyl group to the N1 or N3 atom in the azole ring of compounds 5 and 6 could not be distinguished by NOE data. Similar problems with the unambiguous determination of the product structure were also found for N-arylation or N-alkylation with tert-butyl fragments in the series of 1,2,3-triazole [15,16], tetrazole [17][18][19][20], and purine [21] derivatives. Meanwhile, knowledge of the accurate chemical structures of N-substituted heterocycles is essential for biomedical studies and computer-assisted drug design, e.g., molecular docking techniques. Thus, the development of effective methods for the unambiguous determination of N-alkylation site(s) in the azolo-azine series is important.
The data that are required to solve this problem could be provided by 15 N NMR spectroscopy. For monocyclic derivatives of azoles, the structures of N-alkylated regioisomers can be determined using 2D H-(C)-N multiple bond correlation (HCNMBC) experiments [22,23] using natural isotopic abun-dance. These experiments rely on the magnetization transfer through 13 C-15 N J-coupling constants (J CN ). However, the fusion of the azine ring to an azolo fragment increases the number of possible alkylation sites and considerably complicates the analysis of the J CN patterns. This issue, together with the inherently low sensitivity of natural abundance 15 N NMR spectroscopy, does not always permit the unambiguous positioning of alkyl (N-adamantyl, N-tert-butyl or N-aryl) fragments in azoloazines.
The incorporation of 15 N labels in nitrogen-containing heterocycles greatly facilitates the use of NMR spectroscopy for studies of molecular structures and mechanisms of chemical transformations [10,[24][25][26][27][28][29]. The labelling enhances the sensitivity of detection and permits the quantitative measurements of J CN and 1 H-15 N J-coupling constants (J HN ) even in a mixture of tautomeric forms [24,25]. Additionally, a method based on amplitude-modulated spin-echo experiments was found to be the most efficient way to measure J HN couplings [24]. Previously, the incorporation of a single 15 N label in position 1 of the 1,2,4-triazole fragment of compounds 5 and 6 and analysis of the J CN couplings permitted the unambiguous identification of the structures of the N3-adamantylated derivatives ( Figure 1B), while the structures of the N4-adamantylated products were determined by 13 C NMR spectroscopy via comparison with model compounds, N-methylated azolo-azines [10]. However, this preliminarily study did not evaluate the potential of the incorporation of several 15 N-labels and simultaneous analysis of the J CN and J HN coupling constants for the determination of the N-adamantylation site(s) in heterocycles.
To further study the mechanism of isomerization between compounds 15a and 15b, equimolar quantities of unlabelled N2-isomer 15a and its double-labelled non-adamantylated pre-Scheme 2: Synthesis and adamantylation of 15 N-labelled 20-15 N 2 and J HN and J CN data confirming the structures of adamantylated derivatives 21a,b-15 N 2 .The J HN couplings measured either in the 1D 1 H spectra or by amplitude-modulated 1D 1 H spin-echo experiments and detected in the 2D 15 N-HMBC spectra are shown by grey, magenta, and red arrows (see the legend in the figure). The measured J HN values (gray and magenta) with magnitudes J ≥ 14 Hz and J < 0.1 Hz are indicated by the bold solid and dashed arrows, respectively. The 1 H-15 N cross-peaks observed in the 2D HMBC spectrum for the unlabelled nitrogen atoms are classified into three categories: strong, medium and weak (bold solid, thin solid, and dashed red arrows, respectively). The J CN couplings with adamantane carbons measured in the 1D 13 С spectra are classified into three categories: J ≥ 2 Hz, 2 > J ≥ 1 Hz, and J < 1 Hz (bold solid, thin solid, and dashed green arrows, respectively).   Table 1).
The assignments of the 13 C and 15 N signals in the synthesized compounds were obtained by analysing the 2D 13 C-HMQC, 13 C-HMBC and 15 N-HMBC spectra and observing the 13 C-15 N and 1 H-15 N spin-spin interactions (see below). The 13 C assignment procedure for 19-15 N 2 , 20-15 N 2 , and 21a,b-15 N 2 was aided by the data from a previous study of unlabelled derivatives of compound 19 [12]. The 13 C-19 F J-coupling constants ( n J CF , Table 2) observed in the 1D 13 C spectra facilitated the assignment of the 13 C nuclei for the heterocyclic moieties of compounds 23-15 N 2 and 24-15 N 2 . The observations of the 3 J H2-C3a coupling constants (9.2 Hz) in the 1D 13 C spectra of 19-15 N 2 and 20-15 N 2 measured without proton decoupling confirmed the assignment of C3a to the signals at 160.23 ppm and 152.32 ppm, respectively. The obtained NMR assignments are collected in Table 1 ( 1 H, 15 N) and Table 2 ( 13 C).  J HN couplings, respectively). The J CN couplings became evident from the additional splitting of the corresponding signals in the 1D 13 C NMR spectra and were measured by nonlinear fits of the 13 С line shapes in the 1D spectra acquired with band-selective decoupling from 15 N nuclei [25] (Figure 3). This method allowed for the measurement of the 13 C-15 N spin-spin   Figure 3). The observation of the direct 1 J C1'-N2 (6.5 Hz) and other 13 C-15 N interactions for the C1' ( 2 J C-N3 3.8 Hz), C2' ( 2 J C-N2 0.4 Hz and 3 J C-N3     Table 1). The intensities of the HMBC cross-peaks for the 15 N-labelled nuclei demonstrated an approximate correlation with the measured J HN values (see Table 1). This provides a way to qualitatively estimate the J HN magnitudes for unla-belled and 15 N-labelled nuclei using the relative intensities of the HMBC cross-peaks, corrected for the degree of the isotope enrichment. The measured J HN couplings and HMBC 1 H-15 N spin-spin interactions are shown in Schemes 1-3.  (Table 1, Scheme 2). Notably, the weak cross-peak corresponding to the 4 J H2'-N5 coupling was also detected in the 15 N-HMBC spectrum ( Figure S22D in Supporting Information File 1).
For the adamantylated heterocycles 21a-15 N 2 and 24-15 N 2 , the J HN interactions between the adamantane protons and the labelled N1, N5 or N8 atoms were not detected by amplitude- NMR and Х-ray diffraction data revealing several rotameric configurations of adamantane substituents. The 13 C signals of the N-adamantyl substituents in compounds 15a,b-15 N 2 and 21a,b-15 N 2 measured at 45 °C demonstrated additional splitting, which was not connected to the 1 H-13 C and 13 C-15 N J-couplings. The C1' and C2' signals of adamantane (and C3' for 21a-15 N 2 ) were split into two components with a relative intensity ratio of ≈10:7 and a frequency difference of 0.5-1.2 Hz (Figure 3, Table 2). This revealed the presence of the two structural forms of the N-adamantylated heterocycles in solution with a slow (characteristic time ≥ 1 s) exchange between them. The rotation of the N-adamantyl substituents around the N-C1' bond in the bulky bicyclic heterocycles is likely hindered, and the observed conformational heterogeneity corresponds to the different rotameric configurations of the substituent. To test this hypothesis, additional NMR measurements at elevated temperature were carried out for compound 21a-15 N 2 . The 13 C 1D NMR spectrum measured at 70 °C with 1 H and 15 N decoupling did not demonstrate additional splitting ( Figure S25 in Supporting Information File 1). This confirmed that the studied NMR samples contained unique and chemically pure compounds, while the heterogeneity observed at 45 °C was connected to the presence of different rotameric states.
To confirm the determined positions of the N-adamantane substitutions, compounds 15a and 15b were studied by X-ray crystallography. Suitable crystals of 15a and 15b were obtained by slow evaporation from ethyl acetate solutions. The solved X-ray structures were in a full agreement with the results of the J CN and J HN analysis and confirmed the N2-substituted mesoionic form for compound 15a as well as the attachment of adamantane to the N1 atom in compound 15b. In accordance with expectations, the adamantane groups in the crystals of 15a and 15b were found disordered between two conformations with different rotameric configurations around the N-C1' bond ( Figure 5 and Supporting Information Files 2 and 3). These forms differ by the rotation around the N-C1' bond by 40-60°; thus, in each of them, one of the C2' atoms of the adamantane substituent is located approximately in plane with the heterocyclic moiety of the compound. The populations of the two conformational forms in the single crystals of 15a and 15b (4:1 and 17:3, respectively) differ from the populations of the conformers observed by NMR spectroscopy in DMSO solution (≈10:7). Interestingly, for 15a, the major conformer corresponds to a rotameric state with a screened N1 atom, but in the major conformer of 15b, the N2 atom of the heterocycle is screened. Notably, similar structural disorder was previously observed in the crystals of adamantylated tetrazolylpyrazole derivatives [37,41].

Discussion
Comparison of different NMR approaches for the determination of N-alkylation sites in fused heterocycles. The obtained data permit a comparison of the abilities of different NMR parameters ( 13 C and 15 N chemical shifts, J HN and J CN ) to provide structural information about the N-adamantylation sites in bicyclic heterocycles. The previous studies of azolo [5,1-c] [1,2,4]triazin-7-ones, 1,2,4-triazolo[1,5-a]pyrimidin-7-ones and tetrazolo-azines revealed that the 13 C chemical shifts of the nearest carbon atoms to N-alkyl fragments could be used as indicators for the formation of N-alkylated azolo-azines [12,42]. For the presently studied compounds, we can expect considerable changes in the chemical shifts of the bridgehead C3a and C8a atoms. The shifts of the other carbon atoms from the heterocyclic parts moieties of the compounds (C2, C5, C6, and C7) may also provide useful structural information.
In the studied tetrazolo-triazines and tetrazolo-pyrimidines (compounds 13-15 N 2 and 15a,b-15 N 2 and compounds from work [25]), the resonances of the bridgehead C8a atom were observed over a relatively narrow spectral range (144-155 ppm). In the triazolo-triazines and triazolo-pyrimidines 19-15 N 2 , 20-15 N 2 , 21a,b-15 N 2 , 23-15 N 2 , and 24-15 N 2 , the similar bridgehead C3a atoms are shifted slightly downfield (149-160 ppm). Here, we observed that N-adamantylation of the nitrogen atom directly attached to the C3a, C8a or C2 atoms induced up-field shifts of the corresponding 13 C signal. The majority of these shifts had relatively small magnitudes (Δδ −0.9 to −2.8 ppm), and a large shift was only observed for the C2 resonance of 21a-15 N 2 (Δδ −11.3 ppm ( Table 2)). However, these up-field shifts could not be used to determine the N-adamantylation site. The attachment of the adamantane moiety to other nitrogen atoms could lead to similar 13  A similar situation was observed for the carbon atoms that are separated from the N-adamantylation site by two covalent bonds ( Table 2). The attachment of adamantane to the N2 atom in compound 15a-15 N 2 induced a large down-field shift (Δδ +8.6 ppm) of the C8a resonance, while modification of the N4 atom in compound 21b-15 N 2 induced an up-field shift (Δδ −3.0 ppm) of the C6 resonance. Thus, the obtained data did not reveal an easily interpreted correlation between the 13 C chemical shifts and the position of N-adamantane substitu- ents. The same issue was previously noted in the study of N-alkylated tetrazolo[1,5-a]pyridine derivatives [43].
Similar to the situation observed for the 13 C nuclei, a comparison of the 15 N chemical shifts in the starting heterocycles 13-15 N 2 , 20-15 N 2 , and 23-15 N 2 and their N-adamantylated derivatives 15a,b-15 N 2 , 21a,b-15 N 2 , and 24-15 N 2 did not reveal a simple correlation with the position of the substituent group ( Figure 2, Table 1). Large changes in the 15 N resonance position were observed for the N2 atom in compounds 15a- 15  . For clarity, we should mention that the information that could be obtained from the 15 N chemical shifts is restricted by the pattern of the 15 N-label incorporation. In some cases, the isotopic labels were located far from the position of the attached adamantane group. This fact could partially explain the lack of correlation between chemical shifts and structure.
The obtained data indicated that changes in the 13 C and 15 N chemical shifts could not reliably determine the adamantylation sites in azolo-azines. Therefore, we focused our study on the analysis of 1 H-15 N and 13 C-15 N spin-spin interactions. Despite the relatively 'sparse' placement of 15 N labels, in all the synthesized compounds, the bridgehead C1' atom of the adamantyl fragment demonstrated detectable J CN couplings (Schemes 1-3, Figure 2). The observed J CN values greatly varied in magnitude. The direct and vicinal couplings ( 1,2 J CN ) were relatively large (6.5-2.7 Hz), while the geminal and longrange interactions ( 3,4 J CN ) were small (0.6-0.2 Hz). The fact that the 1 J CN and 2 J CN as well as the 3 J CN and 4 J CN couplings for the C1' atom had similar magnitudes indicated that additional data are required for the unambiguous determination of the adamantylation sites. For this purpose, we measured and analysed the 13 C-15 N and 1 H-15 N spin-spin interactions for the other atoms of the adamantane groups (Schemes 1-3). These additional sets of 2-4 J CN and 2-5 J HN data reliably identified the N-adamantylation sites in the all studied compounds. The proposed structures of 15a,b-15 N 2 were independently confirmed by Х-ray diffraction data.
One of the advantages of J CN and J HN data compared with chemical shift data is the usefulness of 'negative' information.
In the majority of the cases, the absence of a detectable 13 C-15 N or 1 H-15 N spin-spin interaction indicates the remote localization of the adamantane substituent and labelled nitrogen of the heterocycle. The obtained results showed that the structural information provided by the 1 H-15 N spin-spin interactions (measured by 1D 1 H spin-echo experiments or detected in 2D 15 N-HMBC experiments) is similar to the information obtained from the J CN couplings. However, these approaches are not equivalent. On one hand, the acquisition of J HN data requires less measurement time and less sophisticated equipment compared with that of J CN data (conventional broadband probe and two-channel NMR spectrometer versus triple-resonance probe and three-channel spectrometer, respectively). On the other hand, the structural characterization of the N-adamantylation site(s) in heterocycles based on the J HN data requires the preliminary assignment of the 15 N resonances. Therefore, the combination of these approaches based on the analysis of J CN and J HN couplings represents the most effective NMR tool for the determination of adamantylation sites in azolo-azines.
Possible mechanisms of the isomerization of N-adamantylated derivatives 15a and 15b. The isomerization of unlabelled 15a in the presence of 13a-15 N 2 (Scheme 4) elucidated the possible mechanism of the isomerization of 15a-15 N 2 into 15b-15 N 2 . This experiment confirmed that this rearrangement occurs via the formation of adamantyl cation 25 and heterocyclic base 13-15 N 2 (Scheme 5). Moreover, the equilibration of the isotope composition over the reaction products (15a*-15 N 2 , 15b*-15 N 2 and 13*-15 N 2 ) indicated that the transformation of 15a into 15b is reversible. Note that the protonation of compound 13 and its adamantylated derivatives probably plays an important role in the 15a 15b conversion in TFA solution.
The precise positions of the attached protons are unknown, and this determination requires additional investigation, but the analysis of calculated Mulliken charges in compounds 15a and 15b (see Supporting Information File 1, Scheme S2) suggests that the most negatively charged atom N8 undergoes the initial protonation. Similar mechanisms can be proposed to describe the isomerization of compounds 21a and 21b.

Conclusion
We reported the selective incorporation of two 15 N atoms at different positions of 1,2,4-triazolo[1,5-a]pyrimidine, azolo-1,2,4triazines and their N-adamantylated derivatives. The selective incorporation of the 15 N-labels into the azolo and azine rings of the heterocyclic structures led to the appearance of 1 H-15 N and 13 C-15 N J-coupling constants. The combined analysis of the J HN and J CN couplings allowed for the effective determination of the adamantylation sites in the azolo-azine series. To the best of our knowledge, the applicability of this approach for the structural determination of N-substituted heterocycles has not been previously considered. We suggest that the proposed method is generally applicable for the studies of N-alkylated heterocyclic compounds with a high abundance of nitrogen nuclei, where 13 C chemical shifts and 1 H-1 H NOE data cannot provide reliable structural information. The incorporation of the 15 N-labels also permitted the study of the mechanism of isomerization of N-adamantylated tetrazolo[1,5-b][1,2,4]triazin-7-one in TFA solution. The formation of an adamantyl cation and NH-tetrazolo-triazine during the isomerization reaction was confirmed.

Supporting Information
Supporting Information File 1