Occasionally we have the good fortune of enjoying a paradigm shift in how a discipline performs its research. NMR spectroscopy has had a number of such paradigm shifts over the years, 2D NMR methods being probably the most significant in the past several decades. Two-dimensional NMR began with very simple pulse sequences that delivered relatively high sensitivity. As 2D NMR pulse sequences became more sophisticated, leading eventually to hyphenated 2D NMR techniques such as GHSQC-TOCSY and GHSQC-NOESY, sensitivity losses associated with the correspondingly more complex pulse sequences led to information rich spectra but with the penalty of correspondingly lower sensitivity. Consequently, despite the interpretational advantages of some of the hyphenated 2D NMR experiments, they tend to be less frequently used because of their lower sensitivity and consequently longer acquisition times. Unsymmetrical indirect covariance processing, an extension of recent development in covariance NMR methods, offers a potential high sensitivity alternative to access what are normally low sensitivity hyphenated 2D NMR correlation data. Given two coherence transfer experiments, A → B and A → C, it is possible to indirectly determine B → C coherence transfer data. The application of unsymmetrical indirect covariance processing methods to generate GHSQC-COSY and GHSQC-NOESY spectra from the more readily acquired GCOSY, NOESY, and GHSQC 2D NMR spectra and related examples will be described.

1. Unsymmetrical Indirect Covariance NMR or NMR Outside of the Box! G. E. Martin, K. A. Blinov, and A. J. Williams Schering-Plough Research Institute Summit, NJ 07901 ACD Laboratories, Moscow Russian Federation ACD Laboratories, Toronto, Ontario, Canada

10. Indirect Covariance Processing Schematic representation of indirect covariance processing

11. Indirect Covariance Processing 2 mg sample in 180 µL d 6 -DMSO in 3 mm NMR tube. 18 ms conventional HSQC-TOCSY spectrum. Data acquisition time 16 h.

12. Indirect Covariance Processing Result from the IDC processing of the 18 ms HSQC-TOCSY spectrum.

13. Indirect Covariance Processing of IDR-GHSQC-TOCSY spectra Phase information is retained in the processed result, but not only in the usual sense of inverted direct responses!

14. Indirect Covariance Processing of IDR-GHSQC-TOCSY spectra As noted in the original work by Zhang and Bruschweiler, proton resonance overlap can give rise to artifact responses. Type I artifacts are inverted (red). Type II artifacts are indistinguishable on the basis of response phase. K. A. Blinov, N. I. Larin, M. P. Kvasha, A. Moser, A. J. Williams, and G. E. Martin, Magn. Reson. Chem. , 43 , 999 (2005).

15. Indirect Covariance Processing of IDR-GHSQC-TOCSY spectra Consider the more complex example of a polynuclear aromatic system with considerable overlap in the proton spectrum even at 600 MHz. H6 & H15 are completely overlapped and would be expected to give rise to artifact responses: Type I – red solid line Type II – dashed black line

16. Indirect Covariance Processing of IDR-GHSQC-TOCSY spectra The indirect covariance processed result of the IDR-HSQC-TOCSY spectrum shown on the previous slide. Type I artifacts are observed with negative phase (red); Type II artifacts are denoted by dashed black lines. These also differ in integrated peak volume. Projections of the 13 C spectrum are shown flanking the F 1 axis while a 13 C spectrum is plotted along F 2 .

17. Indirect Covariance Processing of IDR-GHSQC-TOCSY spectra Complete analysis of the Type I (red) and Type II (dashed black) artifact responses observed in the indirect covariance processed result from the IDR-HSQC- TOCSY spectrum of a complex polynuclear aromatic. Obviously the very long-range correlations are artifact responses. K.A. Blinov, N.I. Larin, M.P. Kvasha, A. Moser, A.J. Williams, and G.E. Martin, Magn. Reson. Chem. , 43 , 999 (2005).

18. Unsymmetrical Indirect Covariance Processing Unsymmetrical indirect covariance processing works on a pair of data matrices. In the case of IDR-HSQC-TOCSY spectra the data matrix is “decomposed” into a positive (relayed) and negative (direct) response matrix as shown schematically. K. A. Blinov, N. I. Larin, A. J. Williams, M. Zell, and G. E. Martin, Magn. Reson. Chem. , 44, 107 (2006).

19. Unsymmetrical Indirect Covariance Processing m,n -ADEQUATE Long-range carbon-carbon connectivities are shown for the C12 methine resonance. The sole artifact response observed involves C20. Red arrows denote mutually coupled resonant pairs; black arrows denote unidirectional correlations.

20. Unsymmetrical Indirect Covariance Processing GHSQC-COSY A more interesting possibility is found in the unsymmetrical indirect covariance co-processing of an HSQC spectrum and a COSY or TOCSY spectrum to produce the equivalent of an HSQC-COSY or HSQC-TOCSY spectrum. Data matrices were acquired with identical F 2 spectral windows and digitization using the simple sesquiterpene lactone autumnolide as a model compound for the study. K.A. Blinov, N.I. Larin, A.J. Williams, K.A. Mills, and G.E. Martin, J. Heterocyclic Chem., 43 , 163-166 (2005).

21. Unsymmetrical Indirect Covariance Processing GHSQC-COSY Standard GCOSY and multiplicity-edited GHSQC spectra of a 2 mg sample of autumnolide that might be acquired to elucidate a structure. Acquisition times were 10 and 60 m, respectively.

22. Unsymmetrical Indirect Covariance Processing Extending the Boundaries 18 msec IDR-HSQC-TOCSY spectrum of autumnolide acquired in 16 h using a 600 MHz spectrometer equipped with a 3 mm gradient indirect-detection NMR probe.

23. Unsymmetrical Indirect Covariance Processing GHSQC-COSY Unsymmetrical indirect covariance processing of COSY and HSQC spectra affords a data matrix, equivalent to an HSQC-COSY spectrum. Subjecting the calculated HSQC-COSY spectrum to indirect covariance processing, reduces the data to a presentation of 13 C- 13 C direct connectivity information identical to what one would obtain by subjecting an HSQC-TOCSY spectrum to this manipulation as described previously by Zhang and Bruschweiler.

24. Unsymmetrical Indirect Covariance Processing GHSQC-COSY HSQC-TOCSY, 18 ms mixing time 16 h acquisition at 600 MHz using a 2 mg sample and 3 mm probe. Unsymmetrical indirect covariance calculated HSQC-COSY spectrum. Total instrument time ~70 m; 4 s calculation from the processed spectra.

25. Unsymmetrical Indirect Covariance Processing – GHSQC-COSY Top trace – projection through F 1 unsymmetrical indirect covariance processed HSQC-COSY spectrum. Instrument time ~70 m. Signal-to-noise = 77:1. Bottom trace – projection through F 1 of the 18 msec HSQC-TOCSY spectrum. Instrument time 16 h. Signal-to-noise = 8:1. Time to equivalent s/n… a week?

26. Unsymmetrical Indirect Covariance Processing GHSQC-COSY Calculation of a GHSQC-COSY spectrum of strychnine from a multiplicity-edited GHSQC and a conven- tional GCOSY spectrum. Total data acquisition time was <<1 hr. Total post processing time was ~5 sec.

28. Unsymmetrical Indirect Covariance Processing GHSQC-COSY Comparison F 1 projections of strychnine GHSQC-COSY and GHSQC-TOCSY spectra. A) GHSQC-COSY spectrum calculated from a conven- tional GCOSY spectrum and a multiplicity-edited GHSQC spectrum. Data acquisition <<1 h. Post processing ~ 5 s. s/n = 144:1 B) 24 ms GHSQC-TOCSY spectrum of strychnine. Approx. 8 h. data acquisition. s/n = 40:1 Both spectra were subjected to magnitude calculation prior to F 1 projection.

31. Unsymmetrical Indirect Covariance Processing GHSQC-NOESY 44 h GHSQC-NOESY with Unsymmetrical indirect 450 ms mixing time. covariance processed GHSQC and NOESY data. 4.25 h of spectrometer time.

32. Unsymmetrical Indirect Covariance Processing GHSQC-NOESY Slices taken from the 44 h GHSQC-NOESY (left) and unsym- metrical indirect covariance calculated HSQC-NOESY spectra of ibuprofen (right). Slices were taken at the 13 C shift of the sec –butyl methine resonance at ~22.5 ppm.

33. Unsymmetrical Indirect Covariance Processing 13 C- 15 N Heteronuclear Shift Correlation The 13 C- 13 C INADEQUATE experiment depends on the statistical probability of two 13 C atoms being in the same molecule, a 1:10,000 probability based on the ~1% relative natural abundance of 13 C. The probability of adjacent 13 C- 13 C is correspondingly lower. Now consider the statistical probability of a 13 C and a 15 N anywhere in the molecular structure. Roughly a 1:27,000 probability based on 1.1% 13 C and 0.37% 15 N. The net result of these probabilities is that we have the very low sensitivity 13 C- 13 C INADEQUATE experiment but no 13 C- 15 N analog, at least not at natural abundance.

34. Unsymmetrical Indirect Covariance Processing 13 C- 15 N Heteronuclear Shift Correlation -Strychnine B ↔ C

35. Unsymmetrical Indirect Covariance Processing 13 C- 15 N Heteronuclear Shift Correlation - Strychnine G.E. Martin, P.A. Irish, B.D. Hilton, K.A. Blinov, and A.J. Williams, Magn. Reson. Chem. , 45 , in press (2007).

36. Unsymmetrical Indirect Covariance Processing 13 C- 15 N Heteronuclear Shift Correlation - Strychnine Although a 13 C- 15 N heteronuclear shift correlation experiment is infeasible experimentally, we can still calculate this correlation matrix using unsym- metrical indirect covariance processing methods. 1 J CN correlations arise via 2 J NH correlations in the 1 H- 15 N GHMBC data. 2 J CN and 3 J CN correlations arise via 3 J NH and 4 J NH correlations, respectively. The multiplicity arises via the phase of the multiplicity-edited 1 H- 13 C GHSQCAD spectrum direct response.

37. Unsymmetrical Indirect Covariance Processing 13 C- 15 N Heteronuclear Shift Correlation - Eburnamonine

41. Acknowledgements The authors would like to acknowledge Sr. Management of Schering-Plough Research Institute, particularly Drs. R. Imwinkelreid and J. B. Landis for their support. The authors would also like to acknowledge the contributions of B.D. Hilton and P.A. Irish to this work.