Use of Long-Range C-H Heteronuclear Multiple Bond Connectivity in the Assignment of the 13 C NMR Spectra of Complex Organic Molecules

Abstract

The structural elucidation of complex organic molecules relies heavily on the application of proton detected heteronuclear NMR. Among these techniques, the HMBC NMR experiment is probably the most useful 2D NMR method The HMBC (C-H) experiment allows the assignment of structural fragments through correlations between protons and carbons separated by more than one bond, usually two or three bonds (JCH and JCH) via 1H,13C-coupling constants. It is also possible to obtain valuable information through longer correlations, JCH n>3, performing several HMBC experiments with different long-range delays and using a deeper threshold in the contour plot. There have been several attempts to improve the results of the HMBC experiment, mainly focused on the question of optimization of the longrange delay, ∆2. The D-HMBC, 3D-HMBC, CT-HMBC, ACCORD-HMBC, IMPEACH-MBC and CIGARHMBC experiments which provide much better experimental access to sample long-range couplings are briefly discussed. These long-range correlations have proven to be crucial in the structure elucidation of molecules with proton deficient skeleton. INTRODUCTION concert with Heteronuclear Multiple Quantum Coherence, HMQC [7], has proven to be extremely useful for the total structure elucidation and NMR spectral assignments of numerous natural products and complex organic molecules. At the beginning, the sensitivity of HMBC was low as compared with HMQC; however, this characteristic has improved significantly with the introduction of pulsed field gradients [8-10] into these experiments in the early 1990 s. It allowed the receiver gain of the spectrometer to be significantly raised, since the unwanted coherences were already filtered off in the probe head. Moreover, the addition of pulsed field gradients into NMR pulse sequences yields spectra with fewer artifacts and decreases the data collection time, because the selection of the desired coherence pathways occurs without extensive phase cycling. Today, the gradient modification of the HMBC sequence [11] has become a routine standard, accessible to most operators of NMR instruments capable to generate pulsed gradients. In this review we will focus our attention only on Heteronuclear Multiple Bond Correlation (HMBC) [1-3] and modifications of it. Other types of heteronuclear correlation experiments will not be treated here, since HMBC is the most widely used experiment to observe 13C-1H long range couplings. The structural elucidation of complex organic molecules relies heavily on the application of proton detected heteronuclear NMR. Among these techniques, the HMBC NMR experiment is probably the most useful 2D NMR method [4], since it detects 13C-1H long range couplings using inverse detection of the 1H signal, the most sensitive NMR nucleus. Inverse detection techniques also present a considerably higher sensitivity when compared to older 2D experiments [5]. The sensitivity is particularly good when the 1H signals to be observed appear as sharp lines. The HMBC experiment gives a wealth of structural and assignment information through long-range correlation signals for C,H spin pairs, that can span quaternary carbons or heteronuclei, providing a way to link structural fragments together. Therefore, it can be efficiently used to elucidate the molecular skeleton. Two reviews on this topic were published about ten years ago [5,6]. The use of HMBC in Generally, the HMBC (C-H) experiment is described as a technique that allows the assignment of structural fragments through correlations between protons and carbons that are separated by more than one bond, usually through two or three bonds (JCH and JCH) [11,12] via 1H,13C-coupling constants, despite the early observation of a crucial four bond C-H correlation in the HMBC spectrum of antibiotic distamycin A [5], The valuable information that can be obtained through these correlations (JCH n>3 ), is generally discarded because the relative intensities of the resonances are directly related to the magnitude of the coupling constants. Therefore, for JCH n>3 the cross peaks show a low intensities. This characteristic has been used as a criterion to *Address correspondence to this author at the Departamento de Quimica Organica y Fisicoquimica, Facultad de Ciencias Quimicas y Farmaceuticas, Universidad de Chile, Casilla 233, Santiago 1, Chile; Tel.: +56-2-6782865; Fax: +56-2-6782868; e-mail: raraya@ciq.uchile.cl 254 Araya-Maturana et al. discard signals in a computational method of analysis of 2D NMR spectra [13]. have values from 1 to 25 Hz. In practice, a delay shorter than the theoretical value, 65 to 100 ms, is employed to avoid the decay of the 1H magnetization during this delay [15]. Usually, the experimental settings of these parameters in the HMBC experiment are the result of a compromise: when the spin coupling constant is about 8 Hz, the ∆2 delay in the sequence is about 60 ms, allowing an optimum transfer for correlations. EXPERIMENTAL DETAILS When the signals appear as broad lines due to complex splittings, HMBC suffers from a considerable lack of sensitivity. As a consequence, the detection of cross peaks becomes difficult. This problem arises when the power mode data processing causes the cancellation of antiphase signal components. The situation is made worse when the separation of these components is small and when their signals are broad. The efficiency of the HMBC technique is affected also by spectrometer instabilities resulting in t1noises ridges, which also originate from protons bound to 12C. This problem is solved by the application of pulsed field gradients [8-10] leading to much better results in a fraction of the time, since the unwanted coherences are already filtered off in the probe head. The pulse sequence of the HMBC experiment is shown in Fig. (1) [11]. Running the experiment under these conditions, several important couplings could give rise to only very small correlation signals, or they may be completely lost. Furthermore, the direct translation of the connectivities observed in the HMBC spectrum into the bonding network, may be hampered by the fact that JCH and 3J correlations cannot be distinguished [13]. Different methods have been developed to solve both problems mentioned before. In particular, a method to distinguish between JCH and JCH correlations obtained in HMBC experiments has been described some years ago. The experiment is known as 1,1-ADEQUATE and yields only two bond 1H-13C connectivities in H-C-C moieties, allowing differentiation of HMBC from two and three 1H-13C bond connectivities [12, 15]. An additional advantage of this method is that it permits to observe correlations that are missing in the HMBC experiment. A second problem associated with inverse protondetected heteronuclear shift correlation experiments is the lack of resolution in the indirectly detected dimension (F1). For a given spectral width, an increase in F1 resolution requires an increase in the number of t1 increments. Generally, not all spectral regions are crowded enough to need such a treatment. F1 restricted 2D maps can be a great help to ensure a proper spectral analysis [14]. Theoretically, in the HMBC pulse sequence the optimum choice for the first delay is calculated from the expression ∆1 = 1/(2 1J(C,H)). Generally, the 1J(C,H) coupling constants span a range of values from 130 to 160 Hz. The 1J(C,H) filter delay in the pulse sequence currently implemented, is calculated by entering an average value of JCH, usually 140 or 125 Hz, depending on whether there are aromatic or alkene groups, respectively, giving a value of ∆1 = 3.6 or 4.0 ms. In the same manner, the optimum value of the second delay is calculated from ∆2 = 1/(2 nJ(C,H)), where nJ(C,H) is the long range coupling constant. However a drawback of the experiment is due to the range of values of 2J (C,H) and 3J (C,H) spin coupling constants, which can In general, the observed correlations in different HMBC experiments depend on the value the long-range delay, obtained from the individual long-range C-H coupling constant responsible for creating the heteronuclear multiple quantum coherence. Usually, the long-range delay is optimized for a value between 5 and 10 Hz for 1H-13C longrange correlation experiments. The choice is generally made on an arbitrary basis rather than from a knowledge of the actual value of the couplings. On this basis, a first approximation to observe more long-range correlations is to perform several HMBC experiments with different long-range delays and using a deeper threshold in the contour plot. Each one of the spectra obtained in this series of experiments, will show different long-range correlations, according to the value of nJ(C,H). Actually, different delay times will enhance the proper signals and the rest may not be detected. This technique, optimized for small couplings, was employed to observe two and four-bond 1H-13C correlations and unambiguously assign the 13C NMR signals of several ∆ 1 ∆ 2 p1 t1/2 t1/2 aq p2 1H