|Home | About | Journals | Submit | Contact Us | Français|
We describe a simple and accurate method for quantitation by solvent 13C-satellites (QSCS), of very small amounts of natural products using microprobe NMR spectroscopy. The method takes advantage of integration of 13C satellite peaks of deuterated solvents, in particular CDCl3, that have favorable intensities for measurements of samples in NMR microcoils and microprobe tubes in the 1–200 nanomole range.
Recent developments in instrumentation have greatly increased the sensitivity of NMR measurements and have facilitated structural elucidation of natural products at the sub-milligram level. Microcoil probes1 and narrow diameter tube microprobes2 offer significant advantages for measurements of small samples of natural products.3 Reduction of the sample volume by a factor n at a constant sample mass will increase sensitivity of NMR signals in proportion to n, at the same absolute concentration-sensitivity or signal-to-noise (S/N). Microcoils typically have fill volumes of 2.5–5 µL while 1 mm and 1.7 mm tube probes offer fill volumes of about 7 and 30 µL respectively. In practice, microcoil and capillary probe designs optimized for small fill volumes sacrifice some S/N, yet the gain in mass-sensitivity more than compensates for this loss. For example, a recently developed commercial 1.7 mm 1H-13C-15N CPTCI cryoprobe operating at 600 MHz achieves a nominal S/N of 1000:1 for 1H (ASTM standard) when compared to 9000:1 for a 5 mm TXI probe at the same field strength with sample fill volumes of 30 µL and 750 µL, respectively. Thus, the increase in mass-sensitivity of a 1.7 mm cryoprobe compared to the 5 mm cryoprobe can be estimated as the product 1/9 × 750/30, or a factor of 2.7. Compared to a conventional "room temperature" 5 mm probe at the same field strength, this represents >10-fold improvement in mass-sensitivity. Higher gains in S/N have been realized with recent innovations in probe design; for example the 600 MHz 1 mm high-temperature superconducting (HTS) probe.4
Natural products structure elucidation at the sub-milligram level is most suited to microcoil and microtube NMR, however independent estimation of the amount of natural product undergoing measurement is not trivial. For example, 1 nanomole (nmol) of a compound of molecular mass 1000 is equivalent to 1 µg. Gravimetric assay of this amount on a standard analytical balance of nanomoles of material is impossible as the precision of the instrument is typically limited to 0.1–0.01 µg. Accuracy suffers when the net must be calculated by subtraction of two large numbers: the gross mass and the tare mass of the sample container (often 1500–13000 mg). Submilligram samples may be weighed on a microbalance down to ~0.1 µg, however, transfers, especially from solutions, may be cumbersome and these instruments are not commonly available in natural products laboratories. UV-visible spectroscopy can be used for estimating concentrations, but only for compounds that contain known chromophores, or where the extinction coefficient, ε, can be reasonably estimated. The NMR quantitation method known under the acronym ERETIC inserts an "electronic standard" into the FID in the form of a shaped rf signal of variable power that can be integrated along with the sample FID.5 The method has considerable advantages but it typically requires a 3-channel spectrometer and licensed software. Moreover, the accuracy of ERETIC may suffer from high dynamic range problems encountered in nanomole samples. The use of 13C-satellites for quantitative NMR (qNMR) has been exploited for validation of reference phytochemicals.6,7 Here we describe Quantitation by Solvent 13C Satellite (QSCS), a simple method suitable for calculating the absolute amounts of natural product typically measured by NMR in the extremely small volumes of 1–1.7 mm tubes or microcoils. QSCS requires measurement of nothing more than the signals already present in a sample in a solution of CDCl3 during normal NMR measurements. It can be used to obtain the mass for additional quantitative spectroscopic calculations (e.g. UV-vis molar absorptivity, ε, [α]D and CD), and other applications in the elucidation of natural products at the 'nanomole-scale'.8 The latter includes calculation of natural product yields in the nanomole range as demonstrated here for a new natural product, phorbaside F (1), isolated from the marine sponge Phorbas sp.
The advent of micrcoil- and microtube probes has allowed NMR measurements of extremely small amounts of compounds including natural products. During the course of investigations of sub-micromole samples of new natural products from Phorbas sp., we required a reliable way to quantitate the amount of sample in a solution of CDCl3 (99.8% deuterium) during NMR measurements a 1.7 mm Bruker 600 MHz micro-cryoprobe. Since our liquid handling regimen reduced the entire natural product sample to a solution in a constant volume of CDCl3 (35 µL) and the range of sample sizes in our studies was 1–150 nanomole, the dominant peaks observed in the 1H NMR spectra were from residual CHCl3 from the CDCl3 (99.8% D). Excellent peak line-shape, high signal and low artifact levels were observed near the base of the CHCl3 signal at 1% signal height due, in part, to efficient vibration insulation of the 14T cryomagnet employed in this study. Highly sensitive cryoprobes allow ready observation of signals for the 1H-13C couplet of 13CHCl3 (J = 209 Hz), the so-called "13C-satellites", even after only one scan. Although the main 12CHCl3 signal intensity is generally too large for accurate quantitation of nanomole samples due to dynamic range considerations, the 13CHCl3 solvent satellite integrals are comparable in magnitude to those of typical methine CH signals in the natural product sample being measured. Thus, we were motivated to exploit the constant presence of the 13C-satellite signals for nanomole-scale quantitation of the sample.
A series of volumetric solutions of cholesterol (2) in CDCl3 (99.8% D) were prepared containing 1, 10 and 50 nmol/35 µL and measured at 600 MHz in a 1.7 mm cryo-microprobe under standard conditions (see Experimental Section) that included two pre-acquisition dummy scans. The FID’s were Fourier-transformed and the resulting spectra phased, baseline-corrected, and integrated to give ratios of selected integrals ACH for H-3 and H-6 to Asat which is defined in Eqn 1 as the sum integrals of the left-hand (AL) and right hand (AR) components of the 1H-13C couplet of residual 13CHCl3. The number of moles of sample, m, is related to the ratio of signal integral ACH to satellite integrals Asat by the proportionality constant, c (Eqn. 2), which may be obtained from the slope of a standard curve (Figure 1) of the measured integrals of the 13C-satellites against the amount of standard in the fixed volume of sample solution.
From linear regressions of the standard curves, we obtained a proportionality constant of c = 0.097 nmol–1 (R = 0.99972) for the H-3 proton of 2 and similar value (c = 0.099 nmol−1, R = 0.9992) for H-6. Therefore, under our standard conditions, c is approximately 0.10 nmol−1. To put this another way, the presence of 10 nmol of sample in the tube will give 1H sample CH integrals that will be approximately equal to the sum of the two 13C satellite integrals for residual 13CHCl3 in NMR grade CDCl3 of 99.8%D atom isotopic purity.9
Having optimized a simple method for measuring mole amounts of samples in 1.7 mm NMR tubes, we turned our attention to quantitation of a new natural product that was isolated in nanomole amounts from the marine sponge, Phorbas sp. Previous results of our investigation of a single specimen of Phorbas sp. collected in Western Australia, yielded two sets of different cytotoxic macrolides: phorboxazoles A (3), B (4)10 in 1996 and more recently the glycosidic macrolides, phorbasides A-E11 [e.g., phorbaside A (5) and D (6)]. Repurification of minor fractions yielded the new natural product phorbaside F (1) that was shown by HRESIMS (m/z 649.2753 [M+Na]+, Δmmu = 0.40) to be a lower homolog of 5.
Figure 3 shows the downfield region of the 1H NMR spectrum of 1 (600 MHz, CDCl3, 99.8%D) and the associated 13C-satellite signals of residual solvent (CHCl3). The integral ratio ACH/Asat was measured for the downfield signals and averaged (ACH/ Asat=116±0.05) together with the proportionality constant c established above allowed determination of the amount of 1 present in the NMR tube to be 11.8±0.6 nmol (7.7 µg).12 Since the entire sample of 1 isolated from Phorbas sp. had been transferred to the 1.7 mm NMR tube, we calculated that the yield of phorbaside F from Phorbas sp. was 3 × 10–8 % dry weight. The complete structure of 1 was established from HRMS and 2D NMR experiments (COSY, HSQC, HMBC) performed on the entire sample and compared with 5. HSQC of 1 revealed a characteristic AB pattern attributed to an isolated C-2 methylene group (δ 2.42, J = 14.6 Hz, 1H; δ 2.51, J = 14.6 Hz, 1H; δc 44.8 ppm, CDCl3), consistent with absence of the C-2 Me group of 5 (Table 1). The 13C NMR chemical shift of C2 is similar to that of C-2 in callipeloside A (7) 13 (δ 2.50, d; 2.56, d, AB J = 12.7 Hz, 2H; δc 46, CD3OD), a related chlorocyclopropane macrolide with an almost identical macrolide core. The remaining 1H and 13C NMR chemical shifts of 1 were essentially the same as those of 5; therefore, we assigned the same relative configuration to both. The Cotton effect observed in the CD spectrum of 1 (λmax 232, Δε +13.5), was identical in sign and magnitude to that of 5. Since we had earlier assigned the absolute configurations at C-13, C-18 and C-19 of phorbasides A (5) and B by quantitative analysis of the Cotton effects arising from π-π* transitions of the hyperconjugated ene-yne-chlorocyclopropane chromophores (λmax 236 nm), 11,14 both 1 and 5 share the same absolute stereostructure.
The use of 13C satellite signals for quantitative NMR analysis (qNMR) has precedence. qNMR has been used for validation, certification and quality control of reference phytochemicals.6,7 A recent novel application by Claridge and coworkers15 described a solution to an outstanding problem – quantitative measurement of diastereomeric very high diastereomeric ratios (dr's) – by integrating the minor diastereomer against the 13C satellite of the major isomer, which enables measurements of dr's of up to 1000:1 (99.8% diastereomeric excess). In the present application, QSCS allows absolute quantitation of the mole amounts of sample contained within the NMR microtube.8 By providing quantitation independent of gravimetric methods, QSCS facilitates further characterization of natural products at the nanomole-scale by other important spectroscopic parameters that are mass- or mole-dependent, including specific rotation ([α]D), V-visible absorbtivity (ε) and CD (Δε).
Several comments on residual solvent NMR signal intensities are pertinent to QSCS. Commercial CDCl3 for NMR is manufactured to stringent specifications that meet quality control standards for purity and nominal atom % of deuterium content that also holds the residual CHCl3 concentration at reasonably constant levels. For example, quality control measurements of batches of commercial CDCl3 (nominally, 99.8% D) manufactured within a three-month period gave a range of deuterium content from 99.80 to 99.89% D (mean = 99.834%; standard deviation = ±0.021, n = 30).16 Assuming the isotopic natural abundance of 13C in CDCl3 is invariant, the slope of the curves in Figure 1 should only change slightly between different solvent batches if measured under the same conditions. Intramolecular spin-lattice relaxation dominates nuclear spin magnetization recovery in bulk 12CHCl3. Chloroform exhibits very long relaxation times (T1 and T2) as may be expected for inefficient spin-lattice relaxation of a small, rapidly tumbling molecule with only one H atom. Dinesh and Rogers measured longitudinal spin-lattice relaxation times in carefully purified, oxygen-free 12CHCl3 and obtained a value of T1 = 86±2s at 25.0 °C.17 Under normal usage, CDCl3 will show greater longitudinal spin relaxation, governed by an apparent T1* that differs from T1 due to the presence of additional relaxation pathways, in particular paramagnetic relaxation induced by the presence of dissolved oxygen. To our best knowledge, the T1’s of dilute solutions of 13CHCl3 in bulk 12CHCl3 at natural abundance have not been reported, however, it was expected to be similar to 12CHCl3 given the low efficiency of intramolecular dipolar relaxation from 1H to the low-γ nucleus 13C. In fact, measurements of the apparent T1* 's of residual 13CHCl3 and 12CHCl3 in atmosphere-equilibrated ‘CDCl3’ NMR solvent (600 MHz, 25 °C) by inversion recovery experiments gave almost equal values (T1* = 7.3 s and 6.9 s, respectively), both significantly shorter than that of pure oxygen-free CHCl3. The reported T1’s of CH and CH2 protons in 1-dehydrotestosterone (8), a steroid comparable in size and shape to 2, are much shorter (e.g., CH2’s, T1 = 1.0–1.2 s; CH’s 1.3–1.7 s)18 than those of residual CHCl3 molecules. Consequently, partial saturation of the solvent residue 13C-satellites upon repetitive rf pulse NMR experiments may be greater in CHCl3 than those of molecules of similar or greater size than 2, and give a disproportionately lower integrals.19
Changes in intensity of the 13C-satellite signals of CHCl3 may also be expected with variation of experimental parameters such as the relaxation delay, d1, the presence of paramagnetic materials (particularly, triplet molecular oxygen), and the temperature.17 Partial saturation of residual solvent is established largely by the first two dummy scans in the pulse sequence and changes in d1 may influence ACH/Asat.. In addition, differential recovery of magnetization of the sample signal during d1 and the acquisition time may further alter the value of ACH/Asat.. This should be more pronounced for molecules with T1's largely different from 2. On the other hand the relatively large number of scans used for 1D spectra may obviate these factors if the system is under dynamic saturation conditions. In order to address these issues experimentally, we varied the relaxation delay from d1 = 1 s to d1 = 1.5 s and found ACH/Asat only decreased by about 9% for ACH/Asat for H-3 and H-6 with 10 nmol of 2 and 20% when d1 = 2 s (see Table S1, Supporting Information), consistent with a larger relative increase of Asat compared to ACH. Nevertheless, the linearity of the standard curve (Figure 2) shows the spin-lattice relaxation pathways for sample and residual CHCl3 protons are constant over the range of concentrations of 2 that were tested (0.0285–1.42 mM). Consequently, the ratio ACH/Asat and the constant c can be reliably used for quantitation of samples if measurements are made under the same experimental conditions, particularly d1, used for calibration with 2. The total number of scans acquired for each sample of 2 was >400. For smaller molecules, with longer T1's or data sets acquired with fewer scans, it may be preferable to use more dummy scans (n= 8–16). Nevertheless, it should be stressed that application of solvent 13C satellite integration for natural products quantitation should always be preceded by calibration with a standard molecule displaying fast relaxation (short T1’s) and well-resolved integrals for each solvent batch.
Other experimental factors can also influence c. These include field strength, nature of the NMR solvent, probe fill factor and, of course, atom % deuterium content of the solvent that determines the levels of residual 1H isotopomers. To some extent, less-than optimal sample volumes in the NMR tube may also affect c. Microcoil probes should suffer less from this effect because coil edge-effects are eliminated by "infinitely" constant dielectric composition in the axial direction of the capillary.20 Calibration should only be needed once for particular combinations of spectrometer, probe, temperature and solvent batch, and should give confidence for nanomole-level quantitation of natural products when NMR spectra of samples are recorded under similar conditions. Solvent overlap with sample signals must be considered when appropriate. The NMR 13C satellite quantitation method can be applied to compounds with vinyl or aromatic signals present in the region δ 7–8 ppm provided that at least one of the two halves of the satellite 13C-1H doublet is clearly observed and can be integrated. If only one of AL or AR can be accurately integrated, a proportionality constant of 2c should be used for calculating the number of nmol. For natural products with limited solubility in CDCl3, the method may be extended to other NMR solvents after appropriate external calibration with a standard (e.g. Figure 1). The 1H-13C couplets of residual non-deuterated isotopomers in common NMR solvents (e.g. CD3OD, (CD3)2SO, (CD3)2CO, CD3CN, CD3COOH) have smaller 1JCH values than CHCl3 (e.g., d6-acetone CD2HCOCD3, JCH = 126 Hz; CHD2OD, JCH = 139 Hz). Observation of the 13C-satellites of multiply-deuterated solvents may be more difficult than with CDCl3 due to additional splitting by 13C-2H scalar coupling, resulting in lower peak heights and occurrence of the satellites signals in midfield regions of the 1H NMR spectrum.
Care should be taken with the handling and storage of enolizable NMR solvents (e.g. d6-acetone) that may undergo deuterium exchange with adventitious H2O and alter the residual protonated isotopomer content.21 This will be of special concern for NMR measurements of samples that are appreciably basic or acidic, although under neutral conditions exchange should be negligible. In the most prudent approach, the application of QSCS in other solvents will take into consideration the foregoing issues on a case-by-case basis. Finally, it should be noted that the precision of QSCS will be no better than that of electronic integration of signals in NMR spectroscopy, typically 5–10%, although this may be improved by averaging values of ACH/Asat over several integral measurements.
We have illustrated here a very simple method for absolute quantitative estimation of nanomole-amounts of natural products by comparative measurements of integrals of 13C satellites of NMR solvent in a microprobe. The method is illustrated for samples dissolved in CDCl3, including structure elucidation and the quantitation of a new natural product, phorbaside F (1) from Phorbas sp. Quantitation by solvent 13C-satellite (QSCS) should be applicable to samples in other NMR solvents that give rise to measurable 13C-satellite signals that are well separated from the major 12C isotopomer signal and the sample signals. QSCS provides expands the window of discovery and characterization of new natural products, especially those that can be procured only in vanishingly small amounts.8
1H NMR spectra were recorded on a Bruker DRX spectrometer equipped with an Avance III console and a CPTCI 1.7 mm probe operating at 599.82 MHz. Measurements of T1 were conducted using a standard inversion recovery pulse sequence (180°– τ −90°) using a 5 mm CPTCI probe, and the data processed using the TOPSPIN T1/T2 module. The same batch of NMR grade solvent CDCl3 (99.8% D, Cambridge Isotopes, Inc.) was used for all measurements. HPLC-grade solvents (CH3OH, CH3CN) were re-distilled prior to use in the final stages of HPLC purification of natural product and sample transfers. Other general procedures are reported elsewhere.22
Samples of 1 and other natural products for NMR measurements were prepared by co-evaporation with CDCl3 (99.8%D) and dried under a stream of N2 immediately prior to transfer to 1.7 mm NMR tubes. Samples were dissolved in CDCl3 (99.8%D) and the solutions centrifuged before transferring to NMR tubes using a Hamilton gas-tight syringe fitted with an extra long, narrow gauge needle (26G, 10.8 cm). Standard solutions of cholesterol (Aldrich, 99%) in CDCl3 were made by serial dilutions of a stock solution (2.85 mM) prepared by dissolving accurately weighed samples of 2 (±0.01 mg) in the calculated volumes of CDCl3. Final concentrations were 0.0285, 0.285 mM and 1.42 mM, corresponding to 1, 10, and 50 nmol in 35 µL. Aliquots (35 µL) of each standard solution were transferred to 1.7 mm NMR tubes (Bruker Biospin) and their 1H NMR spectra recorded as described below.
FID’s were acquired at room temperature (25 °C) under steady state conditions on non-spinning samples using a standard one-pulse acquisition experiment with a non-selective pulse corresponding to a pulse width of 3.3 µs, or 30° tip angle (15 dB attenuation of 100W pulse power) with the following parameters: two dummy pulses to achieve steady state, followed by acquisition of 240–480 FID’s of 32 K complex points and an acquisition time of 2.6564 s (12335.526 Hz spectral width). FID’s were processed under identical conditions using iNMR software (Mestre Labs, Spain) as follows: Fourier transformation with exponential apodization (line broadening = 0.50 Hz), followed by manual phase correction and auto-baseline correction (fifth order polynomial fit). The signals of the left and right components (AL and AR, respectively) of the 1H-13C couplet of the residual solvent signal 13CHCl3 were integrated and summed to give Asat, taking care to exclude the large central 12CHCl3 signal. The integral, ACH, for each of the 1H NMR signals of cholesterol (2) H-3 (δ 3.52, m, 1H), H-6 (δ 5.35, bs, 1H) and the methyl group H-18 (δ 0.67, 3H) were measured and the ACH/Asat ratios plotted against the number of nmol of sample (Figure 1 and Supporting Information).
The previously described CCl4-soluble fraction (93-054-B-1, 350 mg)10a,11,from a MeOH extract of Phorbas sp. (93–054) was separated by flash chromatography on a pre-packed cartridge (silica, Analogix RS-12, 12 g, 2 × 7.5 cm) using stepped gradient of hexane-ethyl acetate mixtures (0–100%) to yield seven fractions. Fraction 5 (93-054-B1-5, 12.5 mg) was further purified twice by reversed-phase HPLC (phenylhexyl column, 250 × 10 mm, 90:10 MeOH-H2O, then phenylhexyl column, 250 × 4.6 mm, 60:40 CH3CN-H2O, HPLC grade, re-distilled) to afford a pure UV-active fraction, 93-054-B1-12 (tR = 12.0 min) which was collected in a 1.8 mL vial and dried under the stream of N2. The residual solvent was removed by co-evaporation from CDCl3 and the sample of pure 1 (7.7 µg, 2 ×10–8 % dry weight) was re-dissolved in CDCl3 (99.8% D, 35 µL) and transferred to a 1.7 mm NMR tube using a gas-tight syringe as described under "Sample Preparation" UV (MeOH), λmax 237 nm, ε 12,000; CD (MeOH) Δε 232 (+13.5), 241 (+12.5); 1H NMR (600 MHz, CDCl3, 99.8% D), see Table 1.; HRESIMS m/z 649.2753 [M+Na]+, calcd 649.2749 for C32H47ClO10Na. 2D NMR spectra (COSY, HSQC, HMBC, 600 MHz, 1.7 mm CPTCI microprobe) were acquired using standard Bruker software. Total NMR time required for acquisition of 2D NMR data was about 36 hours, see Supporting Information for 2D NMR spectra.
We are grateful to E. Genetti (Cambridge Isotopes, Inc.) for discussions on deuterium content of commercial CDCl3, the Scripps Research Institute Mass Spectrometry facility for HRMS measurements, and T. Handel and A. Jansma for assistance with the inversion recovery experiments. This research was supported grants from the NIH (R01 AI039987, R01 CA122256).
Supporting Information Available.
Calculations of moles of 13CHCl3 in commercial CDCl3 (99.8%D), HSQC spectra for 1, and a table of dependence of ACH/Asat upon relaxation delay, d1, for 2 in CDCl3.