Structure elucidation of uniformly ¹³C labeled small molecule natural products

Abstract

Utilization of ²H, ¹³C, and ¹⁵N isotopically labeled proteins and peptides is now routine in biomolecular NMR investigations. The wide-spread availability of inexpensive, uniformly ¹³C enriched glucose now makes it possible to isolate uniformly ¹³C labeled natural products from microbial fermentation. We now wish to describe an approach for the rapid structural characterization of uniformly ¹³C labeled natural products that avoids the pitfalls of relying on parameters typically employed in biomolecular NMR studies.

Introduction

Natural products from microbial and fungal fermentations are a rich source of bioactive molecules that can provide an effective entry point into the development of therapeutic agents for a wide variety of human and animal health issues.^[1] Although the field of natural products isolation and structure elucidation is, in many respects, mature, it is still a generally time-consuming process that sometimes may lead to structural ambiguity. In the past decade, natural product-based drug discovery has fallen into disfavor with most major pharmaceutical research companies.^[2] In this report, we present a streamlined approach to rapid structure characterization of natural products utilizing uniform (U) ¹³C labeling. This strategy has become feasible because of the increased availability and affordability of labeled substrates such as U-¹³C labeled glucose. Using labeled substrates in fermentation media produces metabolites highly enriched with ¹³C. With the ¹³C label in place, the backbone of virtually any natural product can be quickly established by standard and long-range optimized ¹H-decoupled ¹³C–¹³C COSY experiments.^[3–5] The ¹H NMR chemical shifts can be subsequently assigned through the use of an F₁-decoupled HSQC experiment. The constant time CT-HSQC described by Bax and coworkers^[6] provides a method to achieve F₁ decoupling in the HSQC experiments of uniformly ¹³C-labeled peptides and proteins. Herein, we expand the theoretical framework of the CT-HSQC approach to make it universally applicable to practically any U-¹³C labeled organic small molecule. The resulting structure elucidation strategy is demonstrated both theoretically and experimentally using the natural product enterocin,^[7] 1, as an example. Practical limitations of the experimental approach will also be described in this report.

¹H–¹³C HSQC: a cornerstone of structure elucidation

Natural abundance ¹H–¹³C HSQC experiments^[8,9] have represented a cornerstone of small molecule structure elucidation protocols for more than 20 years. The experiment offers high sensitivity, straight-forward and robust setup, while providing a ¹H/¹³C fingerprint of the molecule. Multiplicity-editing adds additional information content relating to the number of protons covalently attached to each carbon atom. Recently developed pure shift analogs of the HSQC experiment have served to further increase resolution and sensitivity.^[10–12] In addition to ¹H and ¹³C chemical shifts, HSQC spectra provide a wealth of structurally relevant information. In fact, there are successful examples of using just ¹H and multiplicityedited ¹H–¹³C HSQC data for automated structure verification purposes.^[13]

Uniform ¹³C labeling of small molecules offers opportunities to access quaternary ¹³C chemical shifts, as well as ¹H–¹³C and ¹³C–¹³C coupling constant information, the latter of which are not easily accessible at ¹³C natural abundance.^[14] At the same time, U-¹³C labeling introduces additional J_CC and J_CH pathways during t₁ evolution in the HSQC experiment, which are negligible at natural abundance. Evolution of J_CC and J_CH couplings affects sensitivity, lineshape, and the phase of HSQC cross-peaks and complicates data interpretation, and therefore is undesirable and to be avoided. The constant time (CT) approach was first introduced in 1992 to avoid J_CH evolution and to minimize J_CC evolution in ¹H–¹³C HSQC of uniformly ¹³C-labeled proteins and peptides.^[6] The success of CT-HSQC experiment is heavily reliant on the structural uniformity of amino acid residues, which equates to uniformity of the corresponding J_CC coupling constants.

Applying the same approach to a much broader range of ¹³C–¹³C coupling constants and chemical shifts is non-trivial. Recently, a modified version of ¹H–¹³C HSQC for U-¹³C labeled small molecules was published.^[15] However, the modified experiment still focuses on the same narrow range of J_CC couplings as the original CT publication and hence has limited general applicability. We believe that the issue can benefit from a thorough theoretical analysis of potential modulations introduced by J_CC couplings, similar to the analysis performed for the ADEQUATE experiments.^[16]

CT-HSQC and its limitations

The constant time (CT) pulse sequence element and an example of an HSQC pulse sequence are shown in Fig. 1. Constant time pulse sequence elements simplify, but do not remove, J_CC evolution. Evolution caused by J_CC introduces a modulation to the magnetization that is a cosine function of the J_CC coupling constant and CT value:

$$\begin{array}{cc}\hfill \mathrm{M}\left(J\right)=\text{cos}(\pi \times {}^{\mathrm{A}}{J}_{\text{CC}}\times \text{CT})& \times \text{cos}(\pi \times {}^{\mathrm{B}}{J}_{\text{CC}}\times \text{CT})\hfill \\ \hfill & \times \text{cos}(\pi \times {}^{\mathrm{C}}{J}_{\text{CC}}\times \text{CT}).\hfill \end{array}$$ (1)

Figure 1.

Pulse sequence elements of the CT-HSQC experiment and the principle of CT evolution.

Here, the terms ^AJ_CC, ^BJ_CC, and ^CJ_CC denote one-bond ¹³C–¹³C homonuclear coupling constants between the carbon of interest and up to three neighboring carbons, A, B, and C. Long-range ⁿJ_CC couplings can also contribute to the modulation significantly in some cases because some ⁿJ_CC couplings can be rather large, ranging up to >16 Hz.^[17]

In the case of amino acids and, consequently, peptides and proteins, the mathematical description of the CT-HSQC experiment can be significantly simplified because of structural uniformity, which translates into uniformity of the corresponding J_CC coupling constants. All aliphatic ¹J_CC values will be within a 32–40 Hz range, and long-range ⁿJ_CC couplings are small enough in amino acids to consider them negligible. One-bond and long-range couplings to the carbonyls of the peptide backbone are selectively suppressed, which is possible because of narrow and predictable range of their ¹³C chemical shifts.^[6] The simplified formula uses average aliphatic ¹J_CC value and a number of covalently bound carbon neighbors, N:

$$\mathrm{M}\left(J\right)\approx {\text{cos}}^{\mathrm{N}}(\pi \times J_{\text{av}}\times \text{CT}).$$ (2)

Choosing CT to match 1/J_av (26.6 ms) allows multiplicity-editing based on the number of covalent carbon neighbors. CT matched to 2/J_av (53.2 ms) leads to correlation responses with the same phase.^[6] Long-range ⁿJ_CC are again considered negligible with a corresponding cosine term close to 1.

One can easily see that while this approach is powerful, provided that the assumption regarding J_CC is valid, the method quickly fails, becoming unusable in the more general case of an organic small molecule. Small molecule ¹J_CC values typically are in 25–70 Hz range,^[18,19] with known outliers as low as 6–15 Hz for cyclopropyl and 15–20 Hz for cyclobutyl carbons and as high as 85–95 Hz for protonated acetylenic carbons.^[19,20] Similarly, long-range ⁿJ_CC constants are also more highly variable and can range to as large as 16 Hz,^[17] thus contributing significantly to the modulation of magnetization. As an example, at an optimization of 53.2 ms for the constant time period as in the original publication,^[6] a long-range coupling of 9 Hz will essentially null the modulation function, thereby eliminating an HSQC correlation.

Analysis of the limits of CT approach

In the general of case of an organic molecule there are four possible situations: 0, 1, 2, and 3 carbon neighbors. We will provide a mathematical treatise for every case. The effect of ⁿJ_CC couplings will be considered later.

N= 0: No J_CC couplings.

$${\mathrm{M}}_0\left(J\right)\equiv 1.$$ (3)

N= 1: One J_CC coupling.

$${\mathrm{M}}_1\left(J\right)=\text{cos}(\pi \times J_{\text{CC}}\times \text{CT}).$$ (4)

N= 2: Two J_CC couplings.

$${\mathrm{M}}_2\left(J\right)=\text{cos}(\pi \times {}^{\mathrm{A}}{J}_{\text{CC}}\times \text{CT})\times \text{cos}(\pi \times {}^{\mathrm{B}}{J}_{\text{CC}}\times \text{CT}).$$ (5a)

Applying a trigonometric transformation cos(x) × cos(y) = 0.5 × [cos(x + y) + cos(x − y)] transforms (5a) from the multiplicative to the additive function:

$$\begin{array}{cc}\hfill {\mathrm{M}}_2\left(J\right)=0.5& \times [\text{cos}(\pi \times \text{CT}\times {(}^{\mathrm{A}}{J}_{\text{CC}}{+}^{\mathrm{B}}{J}_{\text{CC}}))\hfill \\ \hfill & +\text{cos}(\pi \times \text{CT}\times {(}^{\mathrm{A}}{J}_{\text{CC}}{-}^{\mathrm{B}}{J}_{\text{CC}}))].\hfill \end{array}$$ (5b)

The resulting arguments are now (^AJ_CC + ^BJ_CC) and (^AJ_CC − ^BJ_CC) instead of ^AJ_CC and ^BJ_CC.

It is now easy to see that within the range of one-bond carbon– carbon couplings typically encountered with small organic molecules the function M₂(J) can adopt any value between −1 and +1, including null. Moreover, even small changes in the ¹J_CC value may dramatically impact the result and thereby response intensity. For example, consider a CT value of 53.2 ms used in the original publication^[6] (matched to the average aliphatic ¹J_CC coupling for proteins and peptides):

$$\text{If}{}^{\mathrm{A}}{J}_{\text{CC}}=50\mathbf{Hz}\text{and}{}^{\mathrm{B}}{J}_{\text{CC}}=38\text{Hz}\to {\mathbf{M}}_2\left(\mathit{J}\right)=-0.5,$$ (6)

$$\text{If}{}^{\mathrm{A}}{J}_{\text{CC}}=47\mathbf{Hz}\text{and}{}^{\mathrm{B}}{J}_{\text{CC}}=38\text{Hz}\to {\mathbf{M}}_2\left(\mathit{J}\right)=-0.0,$$ (7)

$$\text{If}{}^{\mathrm{A}}{J}_{\text{CC}}=44\mathbf{Hz}\text{and}{}^{\mathrm{B}}{J}_{\text{CC}}=38\text{Hz}\to {\mathbf{M}}_2\left(\mathit{J}\right)=+0.5.$$ (8)

Considering that 38, 44, 47, and 50 Hz are all within the typical range of ¹J_CC coupling constants for organic molecules, one can easily see how an otherwise minor 3 Hz difference can cause dramatic variations in both sign and relative intensity of cross-peaks in a CT-HSQC spectrum. This hypothetical example clearly illustrates challenges of using CT-HSQC for a uniformly ¹³C-labeled organic molecule such as a natural product.

N= 4: Three J_CC couplings.

$$\begin{array}{cc}\hfill {\mathrm{M}}_3\left(J\right)=\text{cos}(\pi \times {}^{\mathrm{A}}{J}_{\text{CC}}\times \text{CT})& \times \text{cos}(\pi \times {}^{\mathrm{B}}{J}_{\text{CC}}\times \text{CT})\hfill \\ \hfill & \times \text{cos}(\pi \times {}^{\mathrm{C}}{J}_{\text{CC}}\times \text{CT}).\hfill \end{array}$$ (9a)

Equation (9a) can be simplified considering that the only case when a protonated carbon can have three other covalently bound carbons if it is an sp³ carbon. In this case, direct carbon–carbon ¹J_CC couplings will be confined to a relatively narrow range, and original CT assumption can still be used here:

$${\mathrm{M}}_3\left(J\right)={\text{cos}}^3(\pi \times J_{\text{CC}}\times \text{CT}).$$ (9b)

Graphical representation of M₁(J) and M₃(J) are shown in Fig. 2, and M₂(J) behavior is shown in Fig. 3. For all cases CT is set to 53.2 ms as in the original publication.^[6] All functions have intensity and sign variations that are dependent on ¹J_CC coupling constants.

Figure 2.

Graphical representation of response intensity M₁(J) (green) and M₃(J) (red) at CT of 53.2 ms.

Figure 3.

Graphical representation of response intensity M₂(J) at CT of 53.2 ms. Ten M₂(J) functions are shown with J_CC incremented from 30 Hz to 75 Hz with 5 Hz step and with variable J represented by ^BJ_CC.

Zero crossings are especially undesirable because it means that certain coupling constants will result in null magnetization, effectively erasing the corresponding HSQC cross-peak.

Optimization of CT for a general case

It is clear from the analysis of Eqns (3–9) that there is no ‘ideal’ setting of CT that will adequately cover the typical range of one-bond carbon–carbon couplings in organic molecules without modulation of the function crossing zero (often more than once) within this range. Figure 4 provides a visual aid showing the M₂(J) function at varied CT values.

Figure 4.

Graphical representation of response intensity M₂(J) at different CT values. For each CT value, ten M₂(J) functions are shown with ^AJ_CC incremented from 30 Hz to 75 Hz with 5 Hz step and with variable J represented by ^BJ_CC.

As a first approach to identifying a ‘universal’ CT setting, one could try to find a compromise CT value that would be sufficient for most but not all possible one-bond couplings, similar to the approach utilized for the inverted ¹J_CC 1, n-ADEQUATE experiment.^[21] Using similar methodology, we calculated that the CT value of 20 ms would represent such compromise. With this optimization, M₂(J) will cross zero at 25 Hz and 75 Hz (see Fig. 5), which may be considered acceptable because 25 and 75 Hz are on the edges of the typical range of one-bond couplings encountered for most natural products.^[18–20] However, further analysis of M₃(J) behavior shows that choosing a 20-ms optimization for CT is a rather poor choice in terms of a universal optimization. With 20-ms CT optimization, ¹J_CC coupling constants in the range from 19 to 31 Hz (as well as from 69 to 81 Hz) will result in response intensity below 5% of maximum possible value, which for practical purposes can be considered null (see Fig. 5). Having ¹J_CC coupling constants ranges of 19–31 Hz and 69–81 Hz as null regions is very undesirable because they overlap with the typical range of one-bond ¹³C–¹³C couplings in organic small molecules and natural products. Large long-range ¹³C–¹³C couplings would also attenuate down the signal intensity, although less drastically. With the optimization of CT for 20 ms, the intensity loss because of long-range carbon– carbon couplings, ⁿJ_CC, of 16, 10, and 7 Hz is calculated to be 40%, 20%, and 10%, respectively. Of course an investigator could still use a 20-ms optimization for the CT parameter if no ¹J_CC couplings are smaller than 32 Hz or larger than 68 Hz are expected in the molecule, but it is definitely not universal for organic small molecules and natural products.

Figure 5.

Graphical representation of M₁(J) (grey), M₂(J), and M₃(J) (black) response intensities at 20 ms CT value. For visualization purposes, ten M₂(J) functions are shown with ^AJ_CC incremented from 30 Hz to 75 Hz in 5 Hz step and with the variable J represented by ^BJ_CC.

A closer look at the formula (5b) makes it clear that in order to decouple J_CC evolution in a truly uniform fashion, one has to satisfy all possible ranges of (^AJ_CC + ^BJ_CC) and (^AJ_CC − ^BJ_CC). The only way to accomplish this goal is by creating a very slow cosine wave, which is achieved with a 5 ms optimization of the CT interval (see Fig. 6). This setting represents truly universal optimization: in a coupling range of 0–96 Hz it yields both constant phase and reasonable signal intensity (zero crossing is at 100Hz). Additionally, this optimization is very insensitive to smaller ⁿJ_CC couplings: even the largest ⁿJ_CC of 16 Hz would cause an intensity loss of less than 3%.

Figure 6.

Graphical representation of M₁(J) (grey) and M₂(J) response intensities at 5 ms optimization of the constant time interval. For visualization purposes, ten M₂(J) functions are shown with ^AJ_CC incremented from 30 Hz to 75 Hz with 5 Hz step and with variable J represented by ^BJ_CC.

Optimization of the CT interval for a universal value of 5 ms has one major disadvantage: a short constant time interval limits t1, which in turn limits the resolution of the indirectly determined frequency domain F₁ to 200 Hz. Linear prediction during processing can somewhat improve F₁ digital resolution. However, this disadvantage does not affect the structure elucidation protocol for ¹³C-labeled molecules, because high resolution ¹³C data are available via 1D ¹³C and 2D ¹³C–¹³C COSY spectra. These points will be further considered below in the results and discussion section.

Results and discussion

¹H decoupled ¹³C–¹³C COSY

The first step in the carbon backbone determination of a uniformly ¹³C labeled natural product is to acquire a ¹H decoupled ¹³C–¹³C COSY spectrum.^[4] These data quickly provide information on the carbon–carbon connectivity of all neighboring carbon atoms. Because of the high incorporation of ¹³C label, these experiments can be acquired in a short experiment time as shown below using a 7.8-mg sample of enterocin (1). One-bond carbon–carbon correlations afford partial structures, which can be successively linked together through heteroatoms utilizing long-range (ⁿJ_CC, n = 2,3) ¹³C–¹³C correlations via a ¹³C–¹³C COSYLR^[5] experiment. In the case of ¹³C–¹³C COSY, the choice of the t₁ evolution time governs the size of the ¹³C–¹³C coupling constant leading to the observation of a cross-peak. As the t₁ interval becomes longer, progressively smaller ¹³C–¹³C coupling constants will afford correlations. For a given spectral width, the t₁ evolution time can be increased by either increasing the number of F₁ increments or by using COSYLR with an explicitly specified t₁ evolution delay. Both of these approaches provide comparable data, but COSYLR proved to be more practical because it allows one to set any t₁ evolution interval while keeping the number of F₁ increments small and governed only by needed resolution in F₁ indirectly acquired dimension.

Figure 7 shows the ¹H decoupled ¹³C–¹³C COSY spectra optimized for one-bond and for long-range ¹³C–¹³C couplings. The spectra were acquired in 35 min and in 4 h and 45 min, respectively using a Bruker 600-MHz AVANCE III triple resonance spectrometer equipped with 1.7-mm MicroCryoProbe.^™

Figure 7.

¹H decoupled ¹³C–¹³C COSY spectra of the enterocin. A.) ¹H decoupled ¹³C–¹³C COSY spectrum optimized for one-bond correlations. The experiment was acquired with 2048 points in the direct F₂ dimension and 256 points in the indirect F₁ dimension for a maximal t1 interval of 8.5 ms. Pulses affording a 45° flip angle were used. Each FID was acquired with four scans giving a total acquisition time of 35 min. Every possible one-bond ¹³C–¹³C correlation has been observed (see Fig. 9) with an average signal-to-noise ratio of 82:1. B.) ¹H decoupled ¹³C–¹³C COSYLR spectrum optimized for 2- and 3-bond correlations. The experiment was acquired with 2048 points in the direct F₂ dimension and 256 points in the indirect F₁ dimension, the t₁ delay was set 50 ms. Again, pulses affording a 45° flip angle were used. Each FID was acquired with 32 scans giving a total acquisition time of 4 h 45 m.

Universally optimized ¹H–¹³C CT-HSQC

Structure elucidation paradigms for U-¹³C labeled natural products generally utilize a ¹H-decoupled ¹³C–¹³C COSY as the primary structural tool, while CT-HSQC data are employed in a supplementary role to establish ¹H–¹³C direct correlations. Figure 8 shows CT-HSQC spectra acquired with CT intervals of 5 and 26.6 ms acquired on a 7.8 mg sample of U-¹³C labeled enterocin. The acquisition times were 16 and 85 min, respectively, using a Bruker 600-MHz spectrometer equipped with a 1.7-mm MicroCryoProbe.^™

Figure 8.

CT-HSQC spectra of U-¹³C labeled enterocin shown with normalized vertical scales. Positive contours are black and negative contours are red. Enterocin structure with highlighted groups, corresponding to the highlighted HSQC correlations, is shown. A.) CT-HSQC spectrum with CT interval of 5 ms. The data were acquired with 2048 points in the direct F₂ dimension and 128 points in the indirect F₁ dimension. Four transients were accumulated per t₁ increment giving a total acquisition time 16 m. B.) CT-HSQC spectrum acquired with the CT interval optimized for 26.6 ms. The data were acquired with 2048 points in the direct F₂ dimension and 682 points in the indirect F₁ dimension. Four transients were accumulated per t₁ increment giving a total acquisition time of 85 m. Two missing correlations and two correlations with the inverted phase clearly illustrate the importance of employing the optimized CT interval.

One can easily appreciate the robustness of the universally optimized CT-HSQC spectrum acquired with CT optimized to 5 ms: no correlations are missing, phase characteristics are good, and there are only small variations in relative peak intensities. Other CT settings are noticeably less reliable. For example, a CT-HSQC spectrum acquired with 7.5 ms optimization has some peaks missing and other peaks exhibit severe phase distortions (see Figure S1 in the Supplementary Information). The CT optimization of 26.6 ms (default value for proteins and peptides^[6]) results in a spectrum with multiple missing correlations and with inverted phase for other responses. Experimental results in these three spectra confirm the theoretical considerations developed above. All missing or inverted responses are in quantitative agreement with nulls and inversions of the response intensity governed by the Eqn (1) (see Figures S1–S4 in the Supplementary Information).

Structure elucidation of enterocin

A ‘rough’ carbon skeleton of the enterocin molecule was assembled using ¹H decoupled ¹³C–¹³C COSY optimized for large couplings (see Figs. 7 and 9). This skeleton consisted of several substructures that are formed whenever a contiguous carbon– carbon chain is interrupted by a heteroatom (Fig. 9). The resultant substructures were connected together using long-range ¹H decoupled ¹³C–¹³C COSYLR data and ¹H–¹³C connectivities from the optimized CT-HSQC spectrum (see Figs. 8 and 9). The total acquisition time for all necessary experiments (short-range ¹{H} ¹³C–¹³C COSY, long-range ¹{H} ¹³C–¹³C COSYLR, CT-HSQC, 1D ¹H, and 1D ¹³C) was 5 h 35 m for a 7.8 mg sample of U-¹³C labeled enterocin, with ¹³C–¹³C COSYLR requiring 4 h 45 m, which represents 85% of total acquisition time.

Figure 9.

Structure of enterocin derived directly from the ¹³C–¹³C COSY data. Bonds shown in red represent one-bond carbon–carbon correlations observed in ¹H decoupled ¹³C–¹³C COSY experiment. Arrows represent multiple-bond correlations observed in ¹H decoupled ¹³C–¹³C COSYLR experiment acquired with a 50 ms delay. Strong ⁿJ_CC correlations are shown using solid arrows, and weak correlations are denoted by dashed arrows.

Enterocin fermentation and isolation

The ascidian, Didemnum psammathode, was collected in October, 2010, in the Florida Keys (24° 37.487′, 81° 27.443′). For cultivation, a sample of ascidian (1 cm³) was rinsed with sterile seawater, macerated using a sterile pestle in a micro-centrifuge tube, and dilutions were made in sterile seawater, with vortexing between steps to separate bacteria from heavier tissues. Dilutions were separately plated on three media: ISP2, R2A, and M4. Each medium was supplemented with 50 μg/ml cycloheximide and 25 μg/ml nalidixic acid. Plates were incubated at 28 °C for at least 28 days. Strain WMMB285 was identified as a Streptomyces sp. based on the 16S rDNA sequence (GenBank JX960758.1).

A 10 mL seed culture (25 × 150 mm tubes) in medium ASW-A (20-g soluble starch, 10 g glucose, 5 g peptone, 5 g yeast extract, 5 g CaCO₃ per liter of artificial seawater) was inoculated with strain WMMB285 and shaken (200 RPM, 28 °C) for seven days. The seed culture was used to inoculate 100 mL of ¹³C-enriched ASW-A media (10 g U-¹³C-glucose/L) containing Diaion HP20 (4% by weight). After 7 days, filtered HP20 and cells were washed with water and extracted with acetone. The acetone extract was subjected to liquid–liquid partitioning using 30% aqueous methanol and chloroform (1:1). The chloroform partition was fractionated by size exclusion chromatography applying Sephadex® LH-20 as stationary phase and a solvent mixture of chloroform:methanol 50:50 as mobile phase. Five fractions were obtained, and the fraction containing enterocin was purified by HPLC.

A Shimadzu LC system was used, equipped with a LC-20AT pump, SPD-M20A diode array detector, a SIL-20AC HT auto sampler, and a FRC-10A fraction collector. The LC time program consisted on a gradient of methanol 15%/water 85% to methanol 60%/water 40% in 8 min, then increased from methanol 60%/water 40% to methanol 100% in 1 min, and held for 3 min. The column used was a Phenomenex Onyx monolithic column, 100 × 4.6 mm. The flow rate used was 3 mL/min. Deoxyenterocin eluted at 2.45 min and was collected in fractions 3 to 7, while enterocin, eluted at 3.70 min and was collected in fractions 9 to 11.

Conclusions

We have demonstrated a rapid and efficient structure elucidation protocol for U-¹³C labeled natural products. The protocol we propose consists of ¹H decoupled ¹³C–¹³C COSY optimized for shortand long-range couplings and CT-HSQC optimized for use with natural products and organic small molecules. We developed a theoretical framework and have demonstrated the experimental verification of the techniques for expanding the scope of CT-HSQC experiment from narrow amino acid applications^[6] to a protocol that is universally applicable for labeled natural products and organic small molecules. The only trade-off of the universally optimized CT-HSQC is lower F₁ resolution, which, should not be a major impediment to the use of the protocol for structure elucidation because of the availability of high resolution ¹³C information from 1D ¹³C and 2D ¹³C–¹³C COSY data.