Research article Received: 12 November 2014
Revised: 9 December 2014
Accepted: 12 December 2014
Published online in Wiley Online Library: 16 January 2015
(wileyonlinelibrary.com) DOI 10.1002/mrc.4205
Exploiting natural abundance 13C–15N coupling as a method for identification of nitrogen heterocycles: practical use of the HCNMBC sequence Steve Cheatham,a* Mike Klinea and Eriks Kupčeb In this paper, we detail the results of 1H–15N correlation data obtained via 13C–15N coupling at natural abundance on a number of classes of azoles including pyrazoles, imidazoles and triazoles. The experiment produces data that is highly complementary to direct 1H–15N HMBC type correlations in that it can provide 15N chemical shift data for nitrogen that may not show up in the HMBC. This is particularly advantageous in the triazoles where 15N chemical shift can be diagnostic of regiochemistry. Because of the consistency in JCN values among the azoles, the experiment produces distinctive correlation patterns that can be used for identification of regiochemistry. The experiment can also be used to directly measure 13C–15N coupling constants. Copyright © 2015 John Wiley & Sons, Ltd. Keywords: HCNMBC; HMBC; 2D NMR; 15N NMR; 13C–15N coupling
Introduction
Magn. Reson. Chem. 2015, 53, 363–368
Results and discussion We have investigated examples of four common heterocyclic classes including imidazoles (1), pyrazoles (2) and triazoles (3, 4) using the HCNMBC pulse sequence[8] (Scheme 1). Instead of direct measurement of the coupling constant by synthesis of 15N–13C labeled samples, the HCNMBC sequence allows examination of 1H–15N correlations derived from 13C–15N coupling at natural abundance in a simple, routine fashion. Assuming that the 1JCH inept transfer is roughly equivalent in the heterocyclic ring, then the cross peak intensity is related directly to the setting of the 13C–15N evolution delay δJCN. The values of 1 JCN and 2JCN have been measured for 1-methyl imidazole (5) and 1-phenyl pyrazole (6), and these compounds were used to evaluate the basic parameters affecting performance of the sequence[15,16] (Scheme 2).
* Correspondence to: Steve Cheatham, DuPont Crop Protection, Stine Haskell Research Center, Newark, DE 19714, USA. E-mail: [email protected] a DuPont, Crop Protection, Newark, DE, 19714, USA b Bruker UK Limited, Banner Lane, Coventry, CV4 9GH, UK
Copyright © 2015 John Wiley & Sons, Ltd.
363
Nitrogen containing heterocyclic systems constitute a significant percentage of agricultural and pharmaceutical products. Direct 1 H–15N long-range experiments are an invaluable tool for identifying the regioisomers that are often produced by synthesis, but the experiments are not without limitations. In addition to the implicit problem of differentiating two-bond from three-bond correlations, 1H–15N coupling can often be very small and difficult to observe.[1–3] There have been numerous experiments developed to address deficiencies in the standard HMBC including sequences using IMPEACH, CIGAR and other methodologies.[4–6] Williamson et. al. have addressed examination of very small nJNH couplings using a long-range 1H–15N HSQMBC approach to visualize correlations from 1H–15N couplings of less than 1 Hz.[7] Recently, we have introduced a new experiment, HCNMBC[8], that permits 1H–15N correlation via the one-bond 1H–13C and 13 C–15N couplings at natural abundance (Fig. 1). The HCNMBC experiment produces data that is complimentary to the direct 1H–15N correlation methods via a unique coherence pathway that provides an independent method of observation. Sample quantities required are reasonable and are on the same order as the amount necessary for 1,1-ADEQUATE.[9–11] One challenge in utilizing the new technique is that information on 13C–15N coupling constants of heterocycles is somewhat limited because of the difficultly of measuring the couplings at natural abundance and consequent requirement for synthesis of 15N or 15N–13C labeled materials. From the data available, it is clear that the values of nJCN (n = 1–3) vary widely depending on lone-pair orientation and other effects.[12,13] Despite the difficulty, several studies have demonstrated the value of 13C–15N coupling measurements for structure determination, especially in heterocyclic systems.[13,14] This is particularly true in a recent study of azide–tetrazole equilibria
in which analysis of JCN coupling patterns permitted unambiguous determination of structure.[14] The data outlined in this report demonstrates that this divergence of JCN values is common among various heterocyclic classes and holds significant promise as an independent method for assignment of structure.
S. Cheatham, M. Kline and E. Kupče
Figure 1. HCNMBC pulse sequence.[8] The two J optimization delays (δ and 1 13 13 15 δCN) for H– C and C– N respectively are noted in the figure. Pulse sequence code and phase cycle details have been previously published and may be obtained from the supplemental material of reference [8].
Scheme 1. Basic azole structures and numbering used in Tables 1–4.
Scheme 2. 1-Methylimidazole (5) and 1-phenylpyrazole (6) shown with numbering scheme referred to in text.
Imidazoles
364
In 1-methyl imidazole (5), reported coupling constants vary from approximately 14 Hz (1JC5N1) to less than 1 Hz (1JC4N3) and are slightly dependent on solvent.[15] Both the reported coupling constants for 1JC2N1 and 1JC5N1 are relatively large (13.4 and 11.6 Hz respectively in CDCl3).[15] The 2JC4N1 coupling constant is reported to be 5.8 Hz with the other coupling constants being less than 3 Hz.[15] This pattern of coupling constants is reflected in the HCNMBC data shown in Fig. 2. The upper value of the 13C–15N delay (δJCN) in the HCNMBC was chosen to match the expected upper range of coupling constants while the lower value was set to permit observation of coupling from smaller 1JCN and 2JCN values. The results clearly indicate that an excellent degree of tune ability is present in the experiment. The different results from the experiment are more regiospecific than those normally obtained from optimizing 1H–15N HMBC data. The corresponding 1H–15N HMBC (Fig. 2A) shows an extensive correlation network providing little differentiation between proton signals that in an unknown material can be problematic for regiochemical assignment. This result is obtained even though the setting of the JHN delay was optimized for a large coupling of 14 Hz. This contrasts with the behavior of the HCNMBC experiment. When the δJCN delay is set to a delay equivalent to 14 Hz, then only three cross peaks are evident in the aromatic region. Two strong cross peaks are present from the large 1JC2N1 and 1JC5N1 coupling.
wileyonlinelibrary.com/journal/mrc
1
15
Figure 2. (A) Standard H– N HMBC with J delay optimized at 14 Hz. HCNMBC with JCN delay optimized to 14 Hz (B) and 5 Hz (C).
A very weak cross peak is present from the 2JC4N1 coupling. A very different result is obtained when the δJCN delay is set to a delay equivalent to 5 Hz. In this case, the 1JC2N1 correlation is very weak, but, most importantly, one can now clearly see the 1JC4N3 coupling but not the 1JC2N3 correlation. Examination of other examples of imidazoles (1) indicated that the pattern of correlations was similar (Table 1). Tables 1–4 list correlations observed for each compound examined. The tables are organized to show whether a correlation exists between a proton and nitrogen and the 13C–15N bond through which coupling was transferred. Correlations are listed as either strong (S, blue), weak (W, yellow) or not observed (N, red). The differentiation between the strong and weak correlations is defined in this case as a greater than 50% reduction in cross peak intensity.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
Practical use of the HCNMBC sequence Table 1. Imidazoles. Comparison data for imidazoles examined in this study
Samples are listed along with the δJCN optimization and with the relative intensity of the correlation. S, strong; W, weak; N, not observed.
Table 2. Pyrazole correlations
Additional cross peaks not listed in the tables were observed in several samples and are listed along with the δJCN optimization and with the relative intensity, S, strong; W, weak. Again, strong and weak refer to a difference of at least 2× in relative cross peak intensity. a 1-Phenylpyrazole at 5 Hz – C2′N1 = W; C2′N2 = S; C3′N1 = S. b 1-[Pyridyl(2)]-5-aminopyrazole at 5 Hz – C6′N1 = S; C6′N2 = S; C3′N1 = S.
When the data in Table 1 are examined, certain themes develop. 1-Phenylimidazole and 1-acetylimidazole produce very similar correlations to those observed in 1-methylimidazole despite having electron withdrawing groups attached to N1 indicating that substitution of N1 is not critical. This result is consistent with studies indicating that 13C–15N coupling is heavily influenced by hybridization and orientation effects of the nitrogen lone pair.[11] The intensity of this correlation (effectively a two-bond 1H–15N) allows the data to be used in a similar fashion to 1,1-ADEQUATE.[9–11] Care must be taken, however, as the approximately 5 Hz C4N1 coupling also produces a correlation between H4 and N1. However, comparison with the data run at 5 Hz allows unambiguous assignment. Pyrazoles
Magn. Reson. Chem. 2015, 53, 363–368
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
365
The situation is very similar for 1-phenylpyrazole (6).[16] Table 2 lists the results of HCNMBC experiments run on a series of pyrazoles.
Direct correlation between the cross peaks observed and the coupling constants can again be achieved by examining the data for 1-phenylpyrazole. Measured 13C–15N couplings have been reported and are similar to those observed in imidazoles, i.e. 1JC5N1 (12.1 Hz), 1 JC4N2 (2.0 Hz), 1JC3N2 (1.2 Hz) and an unassigned coupling of 6.2 Hz that we were able to assign to 2JC4N1.[16] The correspondence between the measured couplings and the observed cross peaks provides qualitative information on the upper and lower limits of detectable couplings at the settings used. The correlation between H5 and N1 is easily differentiated at 14 Hz. A setting of 5 Hz allows the correlation from H4 to N2 to appear that is easily differentiated from H4 to N1. Importantly, the results are consistent for a variety of compounds. The H5–N1 (1JC5N1) correlation is diagnostic in all cases regardless of the setting of δJCN. Likewise, no correlation is observed between H5 and N2 consistent with a small value for 2 JC5N2 The correlation between H4 and N1 can be consistently identified by setting δJCΝ to 5 Hz. Finally, other long-range correlations
S. Cheatham, M. Kline and E. Kupče Table 3. 1,2,4-Triazole correlations
Additional cross peaks not listed in the tables observed in 1,2,4-triazoles. c 1-Vinyl-1,2,4-triazole at 14 Hz – C1′N1 = S; peaks observed at 5 Hz – C1′N1 = S; C2′N2 = S. d 1-(4-Hydroxyphenyl)-1,2,4-triazole at 5 Hz – C2′N1 = S; C2′N2 = S; C3′N1 = S.
Table 4. 1,2,3-Triazole correlations
Additional cross peaks observed in 1,2,3-triazoles not shown in the table. e 1-Phenyl-1,2,3-triazole at 5 Hz – C2′N1 = W; C2′N2 = W; C3′N1 = S.
are apparent upon the attachment of a phenyl or pyridyl group (Table 2 footnotes). We had demonstrated earlier that the HCNMBC experiment run on 4-bromopyrazole produced results that were two-bond specific when data were acquired with δJCN optimization of 5 Hz.[8] This correlation is not observed for 1JC3N2 for 1-phenyl pyrazole (1JC3N2 = 1.3 Hz) at the same setting. This correlation is identifiable for the both the 4-bromopyrazoles and 4iodopyrazoles. The data indicate that substitutions on the ring system can slightly affect the values of JCN. Triazoles
366
The HCNMBC experiment has proven particularly valuable in the identification of triazoles. Primary identification of triazole isomers can often depend on determination of the 15N chemical shift. Observing correlations to all three nitrogen of the various triazole isomers is often problematic. Results from examination of 1,2,4triazoles and 1,2,3-triazoles using the HCNMBC sequence are presented in Tables 3 and 4 respectively and show that the HCNMBC experiment generally permits determination of all nitrogen chemical shifts. The only exceptions noted so far are 4-substituted
wileyonlinelibrary.com/journal/mrc
1,2,3-triazoles (4) where correlations to N2 and N3 are not observed (Table 4). The data for 1,2,4-triazoles (3) examined so far is highly consistent with a pattern of correlating and non-correlating nuclei. This pattern among members of the class produces a defined ‘fingerprint’ for 1,2,4-triazoles. At a δJCN setting of 14 Hz, only the H5N1 and H4N3 correlations are present. When the optimization is changed to 5 Hz, the chemical shift of all nitrogen can be determined (Table 3). 1,2,3-Triazoles represent an important class of heterocycles in medicinal and agricultural chemistry, but because of the synthetic methodologies employed, both 4-substituted and 5-substituted isomers are commonly produced.[17] Identification of 4-substituted versus 5-substitued 1,2,3-triazoles can provide a challenge depending upon substituents. The data in Table 4 show that the HCNMBC experiment can provide confirmatory data to determine the regiochemistry. The table outlines correlations observed for 1-phenyl 4-carboxy and 5-carboxy triazole that are significantly different. In the case of the 4-carboxy analog, the only cross peak viewable is the strong H5N1 correlation. Obviously, in the 5-substituted analog, this position is blocked, and at a δJCN setting of 5 Hz, distinctive correlations between H4 to N2 and N3 are
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
Practical use of the HCNMBC sequence apparent. The data in Table 4 show that at a δJCN setting of 14 Hz, the strong H5N1 correlation is present in all examples. The correlation between H4 and N3 is only consistently observable at a δJCN setting of 5 Hz. Correlations between H5 and N2 or N3 are not observed at either setting. The trend is observed in all 1,2,3-triazoles so far examined. This data along with HSQMBC analysis of the 1 H–15N coupling constants where possible provides a definitive assignment of regiochemistry in this series.[18] Observation of long-range correlations Recently, Williamson et al. have utilized LRHSQMBC to observe very long-range 1H–15N correlations.[7] Several long-range couplings are apparent at a setting of 5 Hz in the compounds examined in
Tables 1–4 and are detailed in the footnotes. A significant number of four-bond 1H–15N correlations (three-bond 13C–15N correlations) are observed between attached aromatic substituents and the nitrogens of pyrazoles and triazoles. The HCNMBC experiment is also very well suited to optimization for studies of very small couplings as shown in Fig. 3. Optimization at 2 Hz emphasizes the four-bond 1H–15N correlations with excellent retention of signal to noise. Long 13C and 15N relaxation times make the experiment perform particularly well for optimization of small J values such as 2 Hz. This ability to emphasize long-range correlations will prove useful when trying to observe long-range 1H–15N correlations that can be critical to structure determination. Measurement of 13C–15N coupling constants The consistency of correlation patterns among the azoles was further examined to understand the dependence of cross peak intensity on the setting of the JCN delay. One observation in Table 3 is noteworthy in this regard. While the correlation between H5 to N1 is easily observed at a δJCN setting of 14 Hz, the correlation is not observed at 5 Hz in any of the examples shown in Table 3. This is surprising given that the 13C–15N coupling constant should be relatively large (>10 Hz) and is observed at both settings of δJCN in all other classes of heterocycles examined (Tables 1, 2 and 4). There are two major ways of determining the JCN coupling – (1) direct measurement in F1 and (2) analyzing signal intensities as a function of the JCN delay. The first technique relies on direct measurement of 13C–15N splitting in F1 by omitting the 180° carbon-13
Magn. Reson. Chem. 2015, 53, 363–368
13
15
Figure 4. C– N coupled HCNMBC spectra of 1-methyl imidazole. Data were acquired with a δJCN optimization of 14 Hz, spectral window 4.0 × 0.25 kHz, 1024 × 256 data points (F2 × F1), 2 s recycling delay; recorded on Bruker Avance III at 700 MHz in 12 h. A slight offset of 13/12 C isotope effects via one or two bonds. doublets is because of the
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
367
Figure 3. HCNMBC spectra of 1-phenyl-1,2,3,triazole with a δJCN optimization of 14 Hz (A), 5 Hz (B) and 2 Hz (C). Optimization at 14 Hz only shows the correlations via larger JCN couplings. Several new long-range correlations are visible at 5 Hz (red arrows) although some are still comparatively weak (denoted by blue arrows).
S. Cheatham, M. Kline and E. Kupče patterns emerge that are diagnostic both within individual classes and more globally among all the azoles we have examined so far. The potential for this new method of examining heterocyclic systems is significant. It provides an ‘orthogonal’ method to use in conjunction with direct 1H–15N correlation experiments to identify type and regiochemistry of nitrogen containing heterocycles. Optimization for observation of long-range correlations with J values of less than 2 Hz is straightforward. The experiment permits direct measurement of 13C–15N coupling constants on natural abundance samples.
Experimental 1
1
Figure 5. Normalized integral of the JH5N1 and JH3N4 cross peaks of 1methyl-1,2,4-triazole as a function of the setting of the JCN delay expressed in hertz.
pulse in the middle of the t1-evolution period (refer to Fig. 1). Figure 4 shows the results of the measurement performed on a sample of 1-methyl imidazole (4). The measured values are in line with those previously reported.[16] Minor differences are within the expected ranges as a result of different sample concentration, temperature and digitization levels. The two major drawbacks of this (‘classical’) technique are (a) the signal intensity is reduced by a factor of two increasing the measurement time by a factor of 4 and (b) generally, the JCN delay is unknown and the signal may be lost altogether. The second technique monitors the intensity of the signal as a function of the JCN delay.[19] The advantage here is that several measurements are carried out, and therefore, the chances of observing the signal in at least some of them are greater than with the classical approach. Furthermore, the zero crossing is sharp increasing the accuracy of the coupling measurement. A series of experiments with different values of δJCN optimization were run on 1-methyl-1,2,4-triazole to explore the response of the cross peaks. Examination of the behavior of the coupling in detail reveals that approximately 5 Hz results in a null in the H5N1 cross peak response in this series as shown in Fig. 5. Similarly, the response of the H3N4 cross peak follows a similar pattern although in this case, only a minimum value was detected. This null is because of the oscillatory behavior of the coupling response and is partly responsible for the defined fingerprint noted for the series. In both cases, the cross peak response passes through a broad maximum at the estimated value of the coupling constant, i.e. approximately 11 and 5 Hz for the H5N1 and H3N4 correlations, respectively and a minimum at 1/2 JCN. The most accurate estimation of the coupling constant can be determined by fitting the modulation, and we are currently exploring this technique in more detail. These results demonstrate that the experiment can not only be used for the identification of heterocyclic systems but also for the direct estimation of 13C–15N coupling constants at natural abundance in the same manner as previously demonstrated to measure 1JCN dipolar couplings in proteins.[19] Further work in this area is ongoing and will be detailed in a separate report.
Conclusions The pattern of couplings within the heterocycles examined is sufficiently consistent for use in structure elucidation. Correlation
Compounds were obtained from a variety of commercial sources. Data were acquired on a Bruker 600 MHz Avance and 700 MHz Avance III spectrometers using TCI cryoprobes. Samples of 13–16 mg/0.6 ml in DMSO-d6 were prepared to permit rapid data acquisition. Each dataset was acquired using 16 scans, a relaxation delay of 3 s and a matrix of 128 × 512 K for a total time of approximately 2 h per experiment. A separate sample of 40 mg/0.6 ml of 1-methyl-1,2,4-triazole in DMSO-d6 was used to acquire the series of data used to measure coupling constants. Acknowledgement We would like to thank Professor Tim Claridge of Oxford University Chemistry Department for kindly providing access to the 700 MHz Avance III system and 5 mm TCI cryoprobe.
References [1] R. M. Claramunt, D. Sanz, C. Lopez, J. A. Jimenez. Magn. Reson. Chem. 1997, 35, 35–75. [2] G. L. Martin, M. L. Martin, in NMR Basic Principles and Progress (Eds: P. Diehl, E. Fluck, R. Kosfeld), Springer, Berlin, 1981, pp. 269–299. [3] M. Witanowski, L. Stefaniak, G. A. Webb. Ann. Rep. NMR Spectrosc. 1986, 18, 194–199. [4] C. E. Hadden, G. E. Martin, V. V. Krishnamurthy. Magn. Reson. Chem. 2000, 38, 143–147. [5] G. E. Martin, C. E. Hadden. J. Nat. Prod. 2000, 63, 543–585. [6] M. Kline, S. Cheatham. Magn. Reson. Chem. 2003, 41, 307–314. [7] R. T. Williamson, A. V. Buevich, G. E. Martin. Tetrahedron Lett. 2014, 55, 3365–3366. [8] S. Cheatham, P. Gierth, W. Bermel, E. Kupce. J. Magn. Reson. 2014, 247, 38–41. [9] M. Köck, R. Kerssebaum, W. Bermel. Magn. Reson. Chem. 2003, 41, 65–69. [10] S. Cheatham, M. Kline, R. Sasaki, K. Blinov, M. Elyashberg, S. Molodtsov. Magn. Reson. Chem. 2010, 48, 571–574. [11] G. Martin, in Ann. Rep. NMR Spectrosc. (Ed: G. A. Webb), Elsevier, London, 2011, 74, pp. 215–291. [12] V. M. S. Gil, W. von Phillipsborn. Magn. Reson. Chem. 1989, 27, 409–430. [13] C. A. Montanari, J. P. B. Sandall, Y. Miyata, J. Miller. J. Chem. Soc., Perkin Trans. 2 1994, 2571–2575. [14] S. L. Deev, Z. O. Shenkarev, T. S. Shestakova, O. N. Chupakhin, V. L. Rusinov, A. S. J. Arseniev. Org. Chem. 2010, 75, 8487–8497. [15] M. Alei Jr., L. O. Morgan, W. E. Wageman, T. W. Whaley. J. Am. Chem. Soc. 1980, 102, 2881–2887. [16] G. E. Hawkes, W. Randall, J. Elguero, C. J. Marzin. J. Chem. Soc., Perkin Trans. 2 1977, 1024–1027. [17] A. Lauria, D. Riccard, F. Mingoia, A. Terenzi, A. Martorana, G. Barone, A. M. Almerico. Eur. J. Org. Chem. 2014, 3289–3306. [18] R. T. Williamson, B. L. Marquez, W. H. Gerwick, K. E. Kover. Magn. Reson. Chem. 2000, 38, 265–273. [19] J. J. Chou, F. Delaglio, A. Bax. J. Biomol. NMR 2000, 18, 101–105.
368 wileyonlinelibrary.com/journal/mrc
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
Revised: 9 December 2014
Accepted: 12 December 2014
Published online in Wiley Online Library: 16 January 2015
(wileyonlinelibrary.com) DOI 10.1002/mrc.4205
Exploiting natural abundance 13C–15N coupling as a method for identification of nitrogen heterocycles: practical use of the HCNMBC sequence Steve Cheatham,a* Mike Klinea and Eriks Kupčeb In this paper, we detail the results of 1H–15N correlation data obtained via 13C–15N coupling at natural abundance on a number of classes of azoles including pyrazoles, imidazoles and triazoles. The experiment produces data that is highly complementary to direct 1H–15N HMBC type correlations in that it can provide 15N chemical shift data for nitrogen that may not show up in the HMBC. This is particularly advantageous in the triazoles where 15N chemical shift can be diagnostic of regiochemistry. Because of the consistency in JCN values among the azoles, the experiment produces distinctive correlation patterns that can be used for identification of regiochemistry. The experiment can also be used to directly measure 13C–15N coupling constants. Copyright © 2015 John Wiley & Sons, Ltd. Keywords: HCNMBC; HMBC; 2D NMR; 15N NMR; 13C–15N coupling
Introduction
Magn. Reson. Chem. 2015, 53, 363–368
Results and discussion We have investigated examples of four common heterocyclic classes including imidazoles (1), pyrazoles (2) and triazoles (3, 4) using the HCNMBC pulse sequence[8] (Scheme 1). Instead of direct measurement of the coupling constant by synthesis of 15N–13C labeled samples, the HCNMBC sequence allows examination of 1H–15N correlations derived from 13C–15N coupling at natural abundance in a simple, routine fashion. Assuming that the 1JCH inept transfer is roughly equivalent in the heterocyclic ring, then the cross peak intensity is related directly to the setting of the 13C–15N evolution delay δJCN. The values of 1 JCN and 2JCN have been measured for 1-methyl imidazole (5) and 1-phenyl pyrazole (6), and these compounds were used to evaluate the basic parameters affecting performance of the sequence[15,16] (Scheme 2).
* Correspondence to: Steve Cheatham, DuPont Crop Protection, Stine Haskell Research Center, Newark, DE 19714, USA. E-mail: [email protected] a DuPont, Crop Protection, Newark, DE, 19714, USA b Bruker UK Limited, Banner Lane, Coventry, CV4 9GH, UK
Copyright © 2015 John Wiley & Sons, Ltd.
363
Nitrogen containing heterocyclic systems constitute a significant percentage of agricultural and pharmaceutical products. Direct 1 H–15N long-range experiments are an invaluable tool for identifying the regioisomers that are often produced by synthesis, but the experiments are not without limitations. In addition to the implicit problem of differentiating two-bond from three-bond correlations, 1H–15N coupling can often be very small and difficult to observe.[1–3] There have been numerous experiments developed to address deficiencies in the standard HMBC including sequences using IMPEACH, CIGAR and other methodologies.[4–6] Williamson et. al. have addressed examination of very small nJNH couplings using a long-range 1H–15N HSQMBC approach to visualize correlations from 1H–15N couplings of less than 1 Hz.[7] Recently, we have introduced a new experiment, HCNMBC[8], that permits 1H–15N correlation via the one-bond 1H–13C and 13 C–15N couplings at natural abundance (Fig. 1). The HCNMBC experiment produces data that is complimentary to the direct 1H–15N correlation methods via a unique coherence pathway that provides an independent method of observation. Sample quantities required are reasonable and are on the same order as the amount necessary for 1,1-ADEQUATE.[9–11] One challenge in utilizing the new technique is that information on 13C–15N coupling constants of heterocycles is somewhat limited because of the difficultly of measuring the couplings at natural abundance and consequent requirement for synthesis of 15N or 15N–13C labeled materials. From the data available, it is clear that the values of nJCN (n = 1–3) vary widely depending on lone-pair orientation and other effects.[12,13] Despite the difficulty, several studies have demonstrated the value of 13C–15N coupling measurements for structure determination, especially in heterocyclic systems.[13,14] This is particularly true in a recent study of azide–tetrazole equilibria
in which analysis of JCN coupling patterns permitted unambiguous determination of structure.[14] The data outlined in this report demonstrates that this divergence of JCN values is common among various heterocyclic classes and holds significant promise as an independent method for assignment of structure.
S. Cheatham, M. Kline and E. Kupče
Figure 1. HCNMBC pulse sequence.[8] The two J optimization delays (δ and 1 13 13 15 δCN) for H– C and C– N respectively are noted in the figure. Pulse sequence code and phase cycle details have been previously published and may be obtained from the supplemental material of reference [8].
Scheme 1. Basic azole structures and numbering used in Tables 1–4.
Scheme 2. 1-Methylimidazole (5) and 1-phenylpyrazole (6) shown with numbering scheme referred to in text.
Imidazoles
364
In 1-methyl imidazole (5), reported coupling constants vary from approximately 14 Hz (1JC5N1) to less than 1 Hz (1JC4N3) and are slightly dependent on solvent.[15] Both the reported coupling constants for 1JC2N1 and 1JC5N1 are relatively large (13.4 and 11.6 Hz respectively in CDCl3).[15] The 2JC4N1 coupling constant is reported to be 5.8 Hz with the other coupling constants being less than 3 Hz.[15] This pattern of coupling constants is reflected in the HCNMBC data shown in Fig. 2. The upper value of the 13C–15N delay (δJCN) in the HCNMBC was chosen to match the expected upper range of coupling constants while the lower value was set to permit observation of coupling from smaller 1JCN and 2JCN values. The results clearly indicate that an excellent degree of tune ability is present in the experiment. The different results from the experiment are more regiospecific than those normally obtained from optimizing 1H–15N HMBC data. The corresponding 1H–15N HMBC (Fig. 2A) shows an extensive correlation network providing little differentiation between proton signals that in an unknown material can be problematic for regiochemical assignment. This result is obtained even though the setting of the JHN delay was optimized for a large coupling of 14 Hz. This contrasts with the behavior of the HCNMBC experiment. When the δJCN delay is set to a delay equivalent to 14 Hz, then only three cross peaks are evident in the aromatic region. Two strong cross peaks are present from the large 1JC2N1 and 1JC5N1 coupling.
wileyonlinelibrary.com/journal/mrc
1
15
Figure 2. (A) Standard H– N HMBC with J delay optimized at 14 Hz. HCNMBC with JCN delay optimized to 14 Hz (B) and 5 Hz (C).
A very weak cross peak is present from the 2JC4N1 coupling. A very different result is obtained when the δJCN delay is set to a delay equivalent to 5 Hz. In this case, the 1JC2N1 correlation is very weak, but, most importantly, one can now clearly see the 1JC4N3 coupling but not the 1JC2N3 correlation. Examination of other examples of imidazoles (1) indicated that the pattern of correlations was similar (Table 1). Tables 1–4 list correlations observed for each compound examined. The tables are organized to show whether a correlation exists between a proton and nitrogen and the 13C–15N bond through which coupling was transferred. Correlations are listed as either strong (S, blue), weak (W, yellow) or not observed (N, red). The differentiation between the strong and weak correlations is defined in this case as a greater than 50% reduction in cross peak intensity.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
Practical use of the HCNMBC sequence Table 1. Imidazoles. Comparison data for imidazoles examined in this study
Samples are listed along with the δJCN optimization and with the relative intensity of the correlation. S, strong; W, weak; N, not observed.
Table 2. Pyrazole correlations
Additional cross peaks not listed in the tables were observed in several samples and are listed along with the δJCN optimization and with the relative intensity, S, strong; W, weak. Again, strong and weak refer to a difference of at least 2× in relative cross peak intensity. a 1-Phenylpyrazole at 5 Hz – C2′N1 = W; C2′N2 = S; C3′N1 = S. b 1-[Pyridyl(2)]-5-aminopyrazole at 5 Hz – C6′N1 = S; C6′N2 = S; C3′N1 = S.
When the data in Table 1 are examined, certain themes develop. 1-Phenylimidazole and 1-acetylimidazole produce very similar correlations to those observed in 1-methylimidazole despite having electron withdrawing groups attached to N1 indicating that substitution of N1 is not critical. This result is consistent with studies indicating that 13C–15N coupling is heavily influenced by hybridization and orientation effects of the nitrogen lone pair.[11] The intensity of this correlation (effectively a two-bond 1H–15N) allows the data to be used in a similar fashion to 1,1-ADEQUATE.[9–11] Care must be taken, however, as the approximately 5 Hz C4N1 coupling also produces a correlation between H4 and N1. However, comparison with the data run at 5 Hz allows unambiguous assignment. Pyrazoles
Magn. Reson. Chem. 2015, 53, 363–368
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
365
The situation is very similar for 1-phenylpyrazole (6).[16] Table 2 lists the results of HCNMBC experiments run on a series of pyrazoles.
Direct correlation between the cross peaks observed and the coupling constants can again be achieved by examining the data for 1-phenylpyrazole. Measured 13C–15N couplings have been reported and are similar to those observed in imidazoles, i.e. 1JC5N1 (12.1 Hz), 1 JC4N2 (2.0 Hz), 1JC3N2 (1.2 Hz) and an unassigned coupling of 6.2 Hz that we were able to assign to 2JC4N1.[16] The correspondence between the measured couplings and the observed cross peaks provides qualitative information on the upper and lower limits of detectable couplings at the settings used. The correlation between H5 and N1 is easily differentiated at 14 Hz. A setting of 5 Hz allows the correlation from H4 to N2 to appear that is easily differentiated from H4 to N1. Importantly, the results are consistent for a variety of compounds. The H5–N1 (1JC5N1) correlation is diagnostic in all cases regardless of the setting of δJCN. Likewise, no correlation is observed between H5 and N2 consistent with a small value for 2 JC5N2 The correlation between H4 and N1 can be consistently identified by setting δJCΝ to 5 Hz. Finally, other long-range correlations
S. Cheatham, M. Kline and E. Kupče Table 3. 1,2,4-Triazole correlations
Additional cross peaks not listed in the tables observed in 1,2,4-triazoles. c 1-Vinyl-1,2,4-triazole at 14 Hz – C1′N1 = S; peaks observed at 5 Hz – C1′N1 = S; C2′N2 = S. d 1-(4-Hydroxyphenyl)-1,2,4-triazole at 5 Hz – C2′N1 = S; C2′N2 = S; C3′N1 = S.
Table 4. 1,2,3-Triazole correlations
Additional cross peaks observed in 1,2,3-triazoles not shown in the table. e 1-Phenyl-1,2,3-triazole at 5 Hz – C2′N1 = W; C2′N2 = W; C3′N1 = S.
are apparent upon the attachment of a phenyl or pyridyl group (Table 2 footnotes). We had demonstrated earlier that the HCNMBC experiment run on 4-bromopyrazole produced results that were two-bond specific when data were acquired with δJCN optimization of 5 Hz.[8] This correlation is not observed for 1JC3N2 for 1-phenyl pyrazole (1JC3N2 = 1.3 Hz) at the same setting. This correlation is identifiable for the both the 4-bromopyrazoles and 4iodopyrazoles. The data indicate that substitutions on the ring system can slightly affect the values of JCN. Triazoles
366
The HCNMBC experiment has proven particularly valuable in the identification of triazoles. Primary identification of triazole isomers can often depend on determination of the 15N chemical shift. Observing correlations to all three nitrogen of the various triazole isomers is often problematic. Results from examination of 1,2,4triazoles and 1,2,3-triazoles using the HCNMBC sequence are presented in Tables 3 and 4 respectively and show that the HCNMBC experiment generally permits determination of all nitrogen chemical shifts. The only exceptions noted so far are 4-substituted
wileyonlinelibrary.com/journal/mrc
1,2,3-triazoles (4) where correlations to N2 and N3 are not observed (Table 4). The data for 1,2,4-triazoles (3) examined so far is highly consistent with a pattern of correlating and non-correlating nuclei. This pattern among members of the class produces a defined ‘fingerprint’ for 1,2,4-triazoles. At a δJCN setting of 14 Hz, only the H5N1 and H4N3 correlations are present. When the optimization is changed to 5 Hz, the chemical shift of all nitrogen can be determined (Table 3). 1,2,3-Triazoles represent an important class of heterocycles in medicinal and agricultural chemistry, but because of the synthetic methodologies employed, both 4-substituted and 5-substituted isomers are commonly produced.[17] Identification of 4-substituted versus 5-substitued 1,2,3-triazoles can provide a challenge depending upon substituents. The data in Table 4 show that the HCNMBC experiment can provide confirmatory data to determine the regiochemistry. The table outlines correlations observed for 1-phenyl 4-carboxy and 5-carboxy triazole that are significantly different. In the case of the 4-carboxy analog, the only cross peak viewable is the strong H5N1 correlation. Obviously, in the 5-substituted analog, this position is blocked, and at a δJCN setting of 5 Hz, distinctive correlations between H4 to N2 and N3 are
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
Practical use of the HCNMBC sequence apparent. The data in Table 4 show that at a δJCN setting of 14 Hz, the strong H5N1 correlation is present in all examples. The correlation between H4 and N3 is only consistently observable at a δJCN setting of 5 Hz. Correlations between H5 and N2 or N3 are not observed at either setting. The trend is observed in all 1,2,3-triazoles so far examined. This data along with HSQMBC analysis of the 1 H–15N coupling constants where possible provides a definitive assignment of regiochemistry in this series.[18] Observation of long-range correlations Recently, Williamson et al. have utilized LRHSQMBC to observe very long-range 1H–15N correlations.[7] Several long-range couplings are apparent at a setting of 5 Hz in the compounds examined in
Tables 1–4 and are detailed in the footnotes. A significant number of four-bond 1H–15N correlations (three-bond 13C–15N correlations) are observed between attached aromatic substituents and the nitrogens of pyrazoles and triazoles. The HCNMBC experiment is also very well suited to optimization for studies of very small couplings as shown in Fig. 3. Optimization at 2 Hz emphasizes the four-bond 1H–15N correlations with excellent retention of signal to noise. Long 13C and 15N relaxation times make the experiment perform particularly well for optimization of small J values such as 2 Hz. This ability to emphasize long-range correlations will prove useful when trying to observe long-range 1H–15N correlations that can be critical to structure determination. Measurement of 13C–15N coupling constants The consistency of correlation patterns among the azoles was further examined to understand the dependence of cross peak intensity on the setting of the JCN delay. One observation in Table 3 is noteworthy in this regard. While the correlation between H5 to N1 is easily observed at a δJCN setting of 14 Hz, the correlation is not observed at 5 Hz in any of the examples shown in Table 3. This is surprising given that the 13C–15N coupling constant should be relatively large (>10 Hz) and is observed at both settings of δJCN in all other classes of heterocycles examined (Tables 1, 2 and 4). There are two major ways of determining the JCN coupling – (1) direct measurement in F1 and (2) analyzing signal intensities as a function of the JCN delay. The first technique relies on direct measurement of 13C–15N splitting in F1 by omitting the 180° carbon-13
Magn. Reson. Chem. 2015, 53, 363–368
13
15
Figure 4. C– N coupled HCNMBC spectra of 1-methyl imidazole. Data were acquired with a δJCN optimization of 14 Hz, spectral window 4.0 × 0.25 kHz, 1024 × 256 data points (F2 × F1), 2 s recycling delay; recorded on Bruker Avance III at 700 MHz in 12 h. A slight offset of 13/12 C isotope effects via one or two bonds. doublets is because of the
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
367
Figure 3. HCNMBC spectra of 1-phenyl-1,2,3,triazole with a δJCN optimization of 14 Hz (A), 5 Hz (B) and 2 Hz (C). Optimization at 14 Hz only shows the correlations via larger JCN couplings. Several new long-range correlations are visible at 5 Hz (red arrows) although some are still comparatively weak (denoted by blue arrows).
S. Cheatham, M. Kline and E. Kupče patterns emerge that are diagnostic both within individual classes and more globally among all the azoles we have examined so far. The potential for this new method of examining heterocyclic systems is significant. It provides an ‘orthogonal’ method to use in conjunction with direct 1H–15N correlation experiments to identify type and regiochemistry of nitrogen containing heterocycles. Optimization for observation of long-range correlations with J values of less than 2 Hz is straightforward. The experiment permits direct measurement of 13C–15N coupling constants on natural abundance samples.
Experimental 1
1
Figure 5. Normalized integral of the JH5N1 and JH3N4 cross peaks of 1methyl-1,2,4-triazole as a function of the setting of the JCN delay expressed in hertz.
pulse in the middle of the t1-evolution period (refer to Fig. 1). Figure 4 shows the results of the measurement performed on a sample of 1-methyl imidazole (4). The measured values are in line with those previously reported.[16] Minor differences are within the expected ranges as a result of different sample concentration, temperature and digitization levels. The two major drawbacks of this (‘classical’) technique are (a) the signal intensity is reduced by a factor of two increasing the measurement time by a factor of 4 and (b) generally, the JCN delay is unknown and the signal may be lost altogether. The second technique monitors the intensity of the signal as a function of the JCN delay.[19] The advantage here is that several measurements are carried out, and therefore, the chances of observing the signal in at least some of them are greater than with the classical approach. Furthermore, the zero crossing is sharp increasing the accuracy of the coupling measurement. A series of experiments with different values of δJCN optimization were run on 1-methyl-1,2,4-triazole to explore the response of the cross peaks. Examination of the behavior of the coupling in detail reveals that approximately 5 Hz results in a null in the H5N1 cross peak response in this series as shown in Fig. 5. Similarly, the response of the H3N4 cross peak follows a similar pattern although in this case, only a minimum value was detected. This null is because of the oscillatory behavior of the coupling response and is partly responsible for the defined fingerprint noted for the series. In both cases, the cross peak response passes through a broad maximum at the estimated value of the coupling constant, i.e. approximately 11 and 5 Hz for the H5N1 and H3N4 correlations, respectively and a minimum at 1/2 JCN. The most accurate estimation of the coupling constant can be determined by fitting the modulation, and we are currently exploring this technique in more detail. These results demonstrate that the experiment can not only be used for the identification of heterocyclic systems but also for the direct estimation of 13C–15N coupling constants at natural abundance in the same manner as previously demonstrated to measure 1JCN dipolar couplings in proteins.[19] Further work in this area is ongoing and will be detailed in a separate report.
Conclusions The pattern of couplings within the heterocycles examined is sufficiently consistent for use in structure elucidation. Correlation
Compounds were obtained from a variety of commercial sources. Data were acquired on a Bruker 600 MHz Avance and 700 MHz Avance III spectrometers using TCI cryoprobes. Samples of 13–16 mg/0.6 ml in DMSO-d6 were prepared to permit rapid data acquisition. Each dataset was acquired using 16 scans, a relaxation delay of 3 s and a matrix of 128 × 512 K for a total time of approximately 2 h per experiment. A separate sample of 40 mg/0.6 ml of 1-methyl-1,2,4-triazole in DMSO-d6 was used to acquire the series of data used to measure coupling constants. Acknowledgement We would like to thank Professor Tim Claridge of Oxford University Chemistry Department for kindly providing access to the 700 MHz Avance III system and 5 mm TCI cryoprobe.
References [1] R. M. Claramunt, D. Sanz, C. Lopez, J. A. Jimenez. Magn. Reson. Chem. 1997, 35, 35–75. [2] G. L. Martin, M. L. Martin, in NMR Basic Principles and Progress (Eds: P. Diehl, E. Fluck, R. Kosfeld), Springer, Berlin, 1981, pp. 269–299. [3] M. Witanowski, L. Stefaniak, G. A. Webb. Ann. Rep. NMR Spectrosc. 1986, 18, 194–199. [4] C. E. Hadden, G. E. Martin, V. V. Krishnamurthy. Magn. Reson. Chem. 2000, 38, 143–147. [5] G. E. Martin, C. E. Hadden. J. Nat. Prod. 2000, 63, 543–585. [6] M. Kline, S. Cheatham. Magn. Reson. Chem. 2003, 41, 307–314. [7] R. T. Williamson, A. V. Buevich, G. E. Martin. Tetrahedron Lett. 2014, 55, 3365–3366. [8] S. Cheatham, P. Gierth, W. Bermel, E. Kupce. J. Magn. Reson. 2014, 247, 38–41. [9] M. Köck, R. Kerssebaum, W. Bermel. Magn. Reson. Chem. 2003, 41, 65–69. [10] S. Cheatham, M. Kline, R. Sasaki, K. Blinov, M. Elyashberg, S. Molodtsov. Magn. Reson. Chem. 2010, 48, 571–574. [11] G. Martin, in Ann. Rep. NMR Spectrosc. (Ed: G. A. Webb), Elsevier, London, 2011, 74, pp. 215–291. [12] V. M. S. Gil, W. von Phillipsborn. Magn. Reson. Chem. 1989, 27, 409–430. [13] C. A. Montanari, J. P. B. Sandall, Y. Miyata, J. Miller. J. Chem. Soc., Perkin Trans. 2 1994, 2571–2575. [14] S. L. Deev, Z. O. Shenkarev, T. S. Shestakova, O. N. Chupakhin, V. L. Rusinov, A. S. J. Arseniev. Org. Chem. 2010, 75, 8487–8497. [15] M. Alei Jr., L. O. Morgan, W. E. Wageman, T. W. Whaley. J. Am. Chem. Soc. 1980, 102, 2881–2887. [16] G. E. Hawkes, W. Randall, J. Elguero, C. J. Marzin. J. Chem. Soc., Perkin Trans. 2 1977, 1024–1027. [17] A. Lauria, D. Riccard, F. Mingoia, A. Terenzi, A. Martorana, G. Barone, A. M. Almerico. Eur. J. Org. Chem. 2014, 3289–3306. [18] R. T. Williamson, B. L. Marquez, W. H. Gerwick, K. E. Kover. Magn. Reson. Chem. 2000, 38, 265–273. [19] J. J. Chou, F. Delaglio, A. Bax. J. Biomol. NMR 2000, 18, 101–105.
368 wileyonlinelibrary.com/journal/mrc
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 363–368
