HPLC–NMR Revisited: Using Time-Slice High-Performance Liquid

Feb 24, 2013 - Time-based trapping of chromatographically separated compounds onto solid-phase extraction (SPE) cartridges and subsequent elution to N...
1 downloads 11 Views 2MB Size
Article pubs.acs.org/ac

HPLC−NMR Revisited: Using Time-Slice High-Performance Liquid Chromatography−Solid-Phase Extraction−Nuclear Magnetic Resonance with Database-Assisted Dereplication Kenneth T. Johansen, Sileshi G. Wubshet, and Nils T. Nyberg* Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark S Supporting Information *

ABSTRACT: Time-based trapping of chromatographically separated compounds onto solid-phase extraction (SPE) cartridges and subsequent elution to NMR tubes was done to emulate the function of HPLC−NMR for dereplication purposes. Sufficient mass sensitivity was obtained by use of a state-of-the-art HPLC−SPE−NMR system with a cryogenically cooled probe head, designed for 1.7 mm NMR tubes. The resulting 1H NMR spectra (600 MHz) were evaluated against a database of previously acquired and prepared spectra. The inhouse-developed matching algorithm, based on partitioning of the spectra and allowing for changes in the chemical shifts, is described. Two mixtures of natural products were used to test the approach: an extract of Carthamus oxyacantha (wild safflower), containing an array of spiro compounds, and an extract of the endophytic fungus Penicillum namyslowski, containing griseofulvin and analogues. The database matching of the resulting spectra positively identified expected compounds, while the number of false positives was few and easily recognized.

I

eluting compounds, can be used for dereplication purposes (Figure 1). Spectral databases, often in combination with alternative methods such as HPLC−mass spectrometry (MS) and HPLC− diode array detection (DAD), have been shown to provide efficient dereplication of natural product extracts.8−11,13,19 One-

t has been well established that the hyphenated technique high-performance liquid chromatography−solid-phase extraction−nuclear magnetic resonance (HPLC−SPE−NMR) can provide comprehensive information about plant metabolites present in crude extracts.1−7 The analysis of the results can be used to make early stop-or-go decisions in an isolation procedure, thus avoiding spending time and resources on already known compounds (dereplication).1−13 In HPLC− SPE−NMR, analytes are concentrated postcolumn on solidphase extraction (SPE) cartridges prior to elution and NMR analysis. The technique was developed from the shortcomings of coupling HPLC separations directly to the NMR instrument by means of a flow probe. Incompatibility of the amounts of analytes, solvent, and solvent modifiers between the separation system and the NMR analysis could thus be circumvented.6,14 Recent developments in probe head technology, leading to an increase in sensitivity and reduction of the sample size, have made it more efficient to elute the analytes from the SPE cartridges directly into NMR tubes.15−17 The direct hyphenation of the flow to the NMR instrument is hence interrupted, though this has a minor impact on the usefulness of the method. With the current level of automation, manual handling of the tubes is limited to transferring racks from a solvent handling robot to an automatic sample changer serving the NMR instrument. From this, it is apparent that the original idea about using NMR as a universal detector for compounds separated by HPLC18 could be reinvestigated. The aim of this work is to demonstrate that 1H NMR spectra, obtained through HPLC separation of extracts with time-sliced trapping of © XXXX American Chemical Society

Figure 1. Conceptual design of the experiments: HPLC separation with time-sliced trapping of eluate on SPE cartridges. The cartridges are eluted to capillary tubes and 1H NMR spectra are acquired. The spectral profiles are matched against a database of spectra and the suggested structures are presented. Spectra with good signal-to-noise ratio that did not give high match factors are marked with question marks. The data for this figure are the same as those in Figure 3, with data in the retention time range 11−20 min and chemical shifts in the range 4−6 ppm. Axes and labels are removed for clarity. Received: November 28, 2012 Accepted: February 23, 2013

A

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

dimensional 1 H NMR spectra provide complementary structural information, notably about stereochemistry. 1H chemical shifts of easily accessible signals have been used to search databases,12,20,21 but spectral profiles have not extensively been used as direct input in database searches. Sensitivity to changes in the solvent composition,22 concentration,23 and the prevalent magnetic field on the displayed spectral pattern makes the matching challenging. To take full advantage of the HPLC−SPE−NMR scheme presented here, we developed a robust matching algorithm that takes changes of chemical shifts into account. Each spectrum from the timesliced trapping is matched against an in-house created database, and the suggested identity and reference spectrum is presented for visual inspection. Two extracts from different sources were used to test the quality of the spectra, fidelity of the database, and the matching scheme in this context. The first was a prefractionated extract of Carthamus oxyacantha (wild safflower) expected to contain the lignan tracheloside, several isomers of glycosylated spiranes, and unsaturated fatty acids (compounds 1−11, Figure 2). From a previous study, reference spectra of pure constituents have been obtained.1 A reanalysis of the extract using the presented scheme could possibly indicate overlooked compounds. The second extract was obtained from a culture of an endophytic fungus, Penicillum namyslowskii, which was recovered from Rhododendron tomentosum.24 The extract contained the wellknown chlorinated fungistatic griseofulvin (14) together with three analogues (12, 13, and 15)25 and would thus constitute a good example that efficient dereplication saves time and increases the information yield.



EXPERIMENTAL SECTION Chemicals. Methanol-d4, acetonitrile-d3 (99.8 atom % D), HPLC-grade ethyl acetate, and griseofulvin were obtained from Sigma−Aldrich (Stenheim, Germany). HPLC-grade methanol and acetonitrile were from VWR (Fontenay-sous-Bois, France) and formic acid, p.a., was from Merck (Darmstadt, Germany). Deionized and filtered (0.22 μm) water was prepared in-house by use of a Milli-Q plus system (Millipore, Billerica, MA). Samples and Sample Preparation. Sampling and preparation of the extract of Carthamus oxyacantha M. Bieb. (Asteraceae) has previously been described as VLC fraction B.1 The endophytic fungus Penicillium namyslowskii K.M. Zalessky, isolated from Rhododendron tomentosum Harmaja and grown in malt extract broth, was prepared as previously described.24,25 Prior to the analysis, the two extracts were dissolved in methanol at 24 and 5 mg/mL, respectively. Instrument. The HPLC−SPE−NMR system consisted of an Agilent 1100 chromatograph composed of a quaternary pump, a photodiode array detector (DAD), an autosampler, a Bruker micrOTOF-Q II mass spectrometer (MS) equipped with an electrospray ionization (ESI) interface (operated via a 1:99 flow splitter), a Knauer Smartline 120 pump for postcolumn flow dilution, a Spark Holland Prospekt 2 SPE unit, a Gilson 215 liquid handler for automated filling of 1.7 mm NMR tubes, and a Bruker Avance III 600 MHz NMR spectrometer equipped with a SampleJet sample changer. Spectra were acquired by use of either a 5 mm TXI probe or a cryogenically cooled 1.7 mm TCI probe, both equipped with zgradients. Chromatographic separations were performed on a Phenomenex Luna C18(2) column (4.6 mm × 150 mm, 3 μm) operated at 40 °C.

Figure 2. Compounds identified in extract of Carthamus oxyacantha (1−11) and in fermentation broth of Penicillium namyslowskii (12− 15). Compounds (number in the database is given in parentheses): 1, tracheloside (1); 2, azaelic acid (11), 3−8, glycosylated spiro compounds (2−7); 9−11, fatty acids (8−10); 14, griseofulvin (11).

HPLC−SPE−NMR Analysis. Samples were separated by HPLC with a mobile phase composed of acetonitrile−water (5:95) + 0.1% formic acid (eluent A) and acetonitrile−water (95:5) + 0.1% formic acid (eluent B), with a linear gradient profile from 20% to 100% B in 40 min at 0.5 mL/min. Eluate in the retention time range 8.3−31.3 min was trapped on 47 GPphase SPE cartridges (30 s/trapping) by postcolumn addition B

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 3. HPLC-SPE-NMR analysis of an extract from Carthamus oxyacantha. (A) Contour plot of 47 1H NMR spectra (600 MHz, 300 K) acquired on compounds eluted from SPE cartridges with acetonitrile-d3. (B) Normalized HPLC chromatograms from UV detection (blue and green traces for 210 and 254 nm, respectively) and MS detection (red trace for the base peak chromatogram). The retention time axis is aligned with the corresponding spectra. The dashed box indicates the range of data presented in Figure 1. Chromatographic peaks identified from the database are indicated by compound numbers, and asterisks indicate compounds that have features in common with the identified glycosylated spiranes.



of water at 1.5 mL/min. Cartridges were dried overnight under a blanket of dry nitrogen and with pressurized nitrogen for 5 min before elution with 30 μL of acetonitrile-d3 into 1.7 mm NMR tubes. The injection volumes were 25 μL (0.6 mg) and 10 μL (0.05 mg), respectively. 1H NMR spectra were recorded at 300 K by use of a 30° flip angle with presaturation of the residual water signal by excitation sculpting and presaturation during the relaxation delay (4.0 s). One hundred twenty-eight transients were recorded with 64k data points covering a spectral width of 12 kHz (20 ppm). Data were multiplied with an exponential function with a line broadening factor of 0.3 Hz, Fourier-transformed to 128k data points, phase-corrected, and referenced to the residual solvent signal (CHD2CN at δ 1.94). Matching of Spectra. The comparison of reference and analyte spectra was done using in-house-developed functions running under Matlab (v. R2011a, Mathworks Inc.) with the Statistics Toolbox (v. 7.4) and Parallel Computing Toolbox (v. 5.0). The function is included as Supporting Information (Table S-1) and was executed with the default parameters specified in the file. The most important parameter values were a maximum shift of ±0.02 ppm and a threshold area of 105 (arbitrarily set to filter out spurious hits in noise regions). Prior to the matching, spectra were resampled by cubic interpolation to a common axis in the range 0−12 ppm by use of 16k data points. Reference Spectra for the Database. Fifty 1H NMR spectra of known compounds were included in the database. Compound names and a selection of acquisition parameters (solvent, sweep width, relaxation delay, number of data points, and probe head) are included as Supporting Information (Table S-2). They were all acquired on the same instrument (600 MHz) and processed with a line broadening factor of 0.3 Hz before Fourier transformation to 128k data points. Ranges with solvent signals, impurity signals, artifacts, and noise were manually set to 0 by in-house-developed functions running in Matlab (v. R2011a, Mathworks Inc.).

RESULTS AND DISCUSSION

The two extracts in this study, originating from a plant and liquid fermentation broth of a fungus, were separated by the same reversed-phase HPLC method with gradient elution. The chromatography was followed by postcolumn dilution with water, and the analytes were trapped on resin-based SPE cartridges. The trapping was based on 30 s time slices and included most of the compounds eluted during the analysis as determined from UV and MS detection. The cartridges were dried overnight under a blanket of dry nitrogen gas and eluted with acetonitrile-d3 directly to capillary NMR tubes in 96position racks. The racks were manually transferred to an autosampler and analyzed by NMR spectroscopy by use of a 1.7 mm cryogenically cooled probe (Figure 1). Each of the two mixtures was analyzed by single injection of approximately 600 and 50 μg, respectively. This is significantly higher than the amounts usually injected in traditional HPLC analyses and was dictated by the objective to balance the separation efficiency with the sensitivity of the NMR spectrometer. It would be possible to increase the signal-tonoise ratios in the NMR spectra by multiple injections with accumulative trappings.5,6,15 For this study we wanted to avoid the risk that retention time shifts would lead to trapping of different suites of compounds. The analysis of the extract of C. oxyacantha, presented as a pseudo-2D spectrum with chromatography traces as projection along the ordinate, showed the separation of a complex mixture and spectra with good signal-to-noise ratios (Figures 1 and 3). Since the composition of the solvent is practically constant for all samples, there are no excessive fluctuations on the position of residual solvent signals (acetonitrile at δ 1.94 and water at δ 2.15−2.23). The deuterium signal of acetonitrile-d3 was used as frequency lock, but the chemical shift of the water signal is usually sensitive toward concentration.26 The limited range of the water signal position is an indication that the drying of the cartridges was consistent throughout the procedure. Other signals present throughout the samples were commonly C

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 4. HPLC-SPE-NMR analysis of an extract of liquid fermentation broth of Penicillium namyslowski. (A) Contour plot of 47 1H NMR spectra (600 MHz, 300 K) acquired on compounds eluted from SPE cartridges with acetonitrile-d3. The box indicates the expanded range in the foreground, with dashed lines to indicate the different spectra along the retention time axis with a subset of assigned signals. (B) Normalized HPLC chromatograms from UV detection (blue and green traces for 210 and 254 nm, respectively) and MS detection (red trace for the base peak chromatogram). The retention time axis is aligned with the corresponding spectra. Numbers indicate the structures assigned to compounds in the respective chromatographic peaks.

value of all peak groups’ regression coefficients was returned as a match factor. A value of 100% would therefore indicate a perfect match of the spectrum to a specific reference spectrum, while the lowest possible match factor would be 0. Negative correlation coefficients were avoided since the phase parameters were set to give positive signals in both the spectra to be matched and for the reference spectra in the database. This approach has three key advantages. First, it is based on relative intensity values. It is therefore not dependent on standardized acquisition parameters or elaborate baseline corrections. However, this also means that even the smallest signal in the analyte spectra will be matched, and an intensity or area threshold should be applied to filter out randomly high matching factors in noise regions. Second, there is no need to assign or deconvolute signals in the reference spectra. Hence, it is easy to add previously acquired spectra in a continuous process during a study. The requirement for the operator is only to recognize and exclude solvent and impurity signals and to split actual analyte signals into groups by setting a threshold value. Third, practical use of the scheme shows that the match factors are fairly insensitive toward fine splitting of signals. Less resolved small couplings in the spectral database or in the analyte spectra still give a high match factor as long as the overall shapes of the signals are the same. This adds to the robustness of the matching scheme, as spectra with less than optimal line shapes can successfully be used. One obvious drawback is that the scheme cannot easily be used to identify components in a complex mixture with many overlapping signals, but for the current application where most of the components are well separated by HPLC−SPE, this does not pose a problem. The spectral database contained 50 1H NMR spectra that had been acquired earlier with the same instrument but not all with the same probe head (Table S-2, Supporting Information). Eleven of the spectra were obtained during a previous study of C. oxyacantha.1 The spectrum of an authentic sample of

observed impurity signals from methylene and methyl resonances at around δ 1.3 and 0.9. These were, however, of considerably smaller intensity than most of the signals from trapped and eluted analytes. Analysis of the endophyte extract (Figure 4) showed similar results with regard to residual solvent signals and quality of analyte signals. The chromatogram showed two main components and several smaller peaks. The first large chromatographic peak was a mixture (approximately 10:1) of one major component (12) and a minor component (13) as shown in the expanded section. The temporal resolution of the pseudo-2D spectrum is also indicated in the expanded section. Different compounds, indicated by different spectra, are clearly separated along the retention time axis with minor contributions from mixed components. In practical terms, the analyses were considerably longer than HPLC analyses with real-time detection, due to the cartridge drying and elution processes. Most of the preparation, however, was done in full automation, meaning that once the chromatography was finished, the drying and elution processes were started and the rack with filled tubes was ready for NMR analysis the following day. For this study we primarily analyzed all samples with standard 1D 1H pulse sequences with solvent suppression to reduce the signal from water and residual solvent signals. Each acquired analyte spectrum was matched against spectra in the database by an algorithm based on partitioning and shifting the spectra followed by linear regression. The program, written for Matlab, is included as Supporting Information (Table S-1), together with Figures S-1 and S-2 to illustrate the procedure. The reference spectra in the database were prepared by manually setting irrelevant data points (i.e., solvent signals, impurity signals, artifacts, and noise) to 0. The reference spectra were hence made up of well-defined peak groups flanked by 0s. Each peak group was compared to the corresponding points in the analyte spectrum, which were stepwise shifted to allow changes in the absolute position of the signals. The average D

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 5. Match factors for the comparison of database spectra (1−50, with compound categories shown in the middle) with experimental spectra, acquired in the HPLC−SPE−NMR mode, of (A) the extract of Carthamus oxyacantha and (B) the endophyte extract. The entries in the database are presented in Table S-2 (Supporting Information).

also illustrated in Figure S-3 (Supporting Information), where the distributions of all peak groups’ regression coefficients are indicated by box-and-whisker plots. Good hits are characterized by consistently high regression coefficients for peak groups as defined in the database spectra. Other spectra of unrelated compounds in the database (entries 12−50) did not give any significantly high match factors as can be seen in the mostly dark blue lower area of the matrix (Figure 5A, 95th percentile at 30%). The fatty acid (10E,12Z)-9-hydroxyoctadeca-10,12-dienoic acid (α-dimorphecolic acid, 9) was confirmed in fraction 45 (at retention time 30.3 min) by the database with a match factor of 85%. Even though the intensity of this compound was below the lowest contour plot level of Figure 3, the signal-to-noise was sufficient for the visual comparison guided by the high match factor and database spectrum (Figure 6). Two more similar

griseofulvin (14) was added to the database as the compound had been structurally characterized as one of the main components in the endophyte extract.25 The remaining 38 spectra were arbitrarily selected spectra of pure natural products acquired during a little more than a year. A complete query of one of the HPLC−SPE−NMR analyses with 47 spectra took about 5 s to compute once the data were read into memory. The results are presented as colored matrices with the database and tested spectra along the two different axes (Figure 5). Depending on the focus of the study, these matrices can be used either to find a specific compound among a series of spectra or to get an idea of the identity of the compound(s) represented in a particular spectrum. A red patch indicates a high match factor, and the suggested identity can be read on the vertical axis. Analysis of the plant extract showed that one particular compound in the database, azaelic acid (2), gave high match factors (interquartile range 62−81%) for many of the HPLC− SPE−NMR spectra in the set. The consistently high match factors were an effect of the ubiquitous fatty substances that appeared in all spectra and the fact that the reference spectrum of azaelic acid contained only three peak groups, and two of these appeared at δ 1.30 and 1.55. However, the largest match factor was 97% for the spectrum of the fraction collected at retention time 10.3 min (analyte spectrum 5), which also was confirmed by high-resolution mass spectrometry (HRMS). The other compounds previously found in a C. oxyacantha extract (1, 3−9), representing most of the major chromatographic peaks, could also be positively identified with the help of the database match factors (78−93%) and HRMS data. When the match factors indicated that several compounds could be possible for a particular fraction, it was in all cases the correct structures that had the highest rank, as verified from manual comparison of the 1H NMR spectra and HRMS data. As an example, fraction 17 contained a mixture of the partly coeluting compounds 4 and 5 with match factors of 78% and 79%, respectively. Compound 3 gave in the same spectrum a match factor of 74% but could be excluded as some of the database peak groups were not represented in the analyte spectrum. This value was also considerably lower than the highest match factor for the row of compound 3 (90%) that was found in fraction 11. Corresponding rows in Figure 5A are

Figure 6. Details from HPLC−SPE−NMR analysis of C. oxyacantha extract: (A) spectrum of fraction 45 at retention time 30.3 min and (B) prepared database spectrum of α-dimorphecolic acid.

compounds, (9Z,11E)- and (9E,11E)-13-oxooctadeca-9,11dienoic acid (10 and 11, respectively), were included in the database but did not appear among the captured fractions. This was either because they were eluted outside the analyzed retention time range or because the compounds were excluded in the VLC preparation of the extract. Interestingly, there were also two fractions in the analysis of the plant extract (marked with asterisks in Figure 3) that appeared similar to the six glycosylated spiranes (3−8), but they did not produce high match factors to any of the database spectra. HRMS data combined with information obtained from the 1H NMR experiment suggested C-7 epimers of the spiro E

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

integral verification of the whole system. Incomplete drying or any other malfunctioning during the elution process is revealed by comparison with other spectra in the series and the chromatograms recorded with real-time detection (DAD and MS). A related issue is the general applicability of the HPLC− SPE−NMR method. The characteristics of the chromatography step are different from those of the trapping step in terms of mobile and stationary phases. This suggests that there is no guarantee that separated compounds are trapped and transferred to the NMR tubes in good yields. This issue has been investigated for alkaloids in a recent article by Johansen et al.27 Using the current mode of HPLC−SPE−NMR gives an indication about the success of the trapping and elution processes. As NMR is a near-universal detector for organic compounds, chromatographic peaks detected by the DAD or MS should also appear in the corresponding 1H NMR spectra. Lack of NMR signals, or spectra with lower than expected signal-to-noise ratios, would indicate that the trapping and elution methods should be adapted for these compounds. A preliminary analysis of a mixture by the described method would hence reveal if the instrument and automation is in good order, if the compounds are suitable for the method, and if the acquired 1H NMR spectra match previously acquired spectra. From this information it is possible to guide further research activities such as focusing on structural characterization of unknown (not in the database) compounds, optimizing the absorption and desorption processes, or improving the robustness of the instrumentation and automation. Further possibilities for the presented method involve multiple trapping to increase the sensitivity, including NMR-based absolute quantification of known compounds, and improving the spectral databases in terms of number of spectra, metadata, and accessibility over the net. We expect that spectral databases will be able to be used at different laboratories as long as the solvent and magnetic field strengths of the instruments are the same.

alcohol (4) and the spiro hydroxyperoxide (7), respectively. The fractions containing compound 4, fractions 16 and 17, had a match factor to the corresponding database spectrum of 84% and 78%, respectively. The corresponding C-7 epimer in fraction 14 had a match factor of 41% to compound 4 (Figure S-3, Supporting Information). The match factor between the spectrum of fraction 20 to the database spectrum of compound 7 was 93%, while the fraction containing the C-7 epimer returned the corresponding match factor 60%. These constitute good examples that the 1D 1H NMR spectra contains stereochemical information that is not possible to obtain from MS data alone. Previously assigned signals for the isomers (4 and 7)1 and several two-dimensional NMR spectra (HSQC, HMBC, and COSY) were used to make complete assignments (Tables S-3 and S-4, Supporting Information). Even though the compounds were in small amounts, it was possible to acquire these spectra directly on the obtained fractions. The proposed structures were confirmed by NOESY spectra obtained in a separate experiment in which the mixture was chromatographically separated and the four isomers were accumulatively trapped (n = 3). The database matching of the spectra from analysis of the fermentation broth of P. namyslowskii gave similar results (Figure 5B). One of the major components, griseofulvin (14), gave high match factors in three consecutive fractions (86%, 97%, and 89%, respectively) with retention times in the range 19.3−20.3 min. Neither the analogue compounds (12, 13, and 15) nor any other entries of the database gave high match factors. The major components of this extract had in a previous study been structurally characterized by use of selectively captured chromatographic peaks.25 A preliminary HPLC− NMR analysis with expected chemotaxonomic components in the database would hence be an efficient dereplication step. A positive result (high match factors) means that the database spectrum should be visually compared with the experimental spectra. Conversely, negative results can be interpreted that the compound giving rise to the spectral profile is not represented in the database and thus might be interesting for further studies.





ASSOCIATED CONTENT

S Supporting Information *

CONCLUSIONS In this work we have demonstrated that mixtures of natural products can be chromatographically separated and detected by NMR spectroscopy as in the originally envisioned HPLC− NMR technique. It was accomplished by time-based trapping of compounds by SPE, followed by an exchange of mobile phase to fully deuterated solvent and elution into capillary NMR tubes (HPLC−SPE−NMR). As the obtained NMR spectra of the fractions were acquired in a fixed solvent, it was also possible to directly compare them with a database of reference spectra by use of an in-house-developed algorithm. Positively identified spectra could subsequently be confirmed by visual inspection, while further analyses could be limited to fractions with unknown spectra. With the relatively small database we tested, the number of false positives was low. The method to match analyte with reference spectra, based on partitioning of the reference spectra into peak groups and allowing for small differences of the chemical shifts, proved to be robust and simple to use. The high mass sensitivity of the instrumentation allowed analysis on single injection of the mixtures in amounts suitable for separation on standard 4.6 mm i.d. analytical-scale reverse-phase HPLC columns. The described approach to trap a high number of consecutive fractions, with subsequent elution into NMR tubes, features an

Four tables and three figures with additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Dr. Mysore V. Tejesvi (Department of Biology, University of Oulu, Finland) is gratefully acknowledged for the sample of endophytic fungus. Financial support from Drug Research Academy, University of Copenhagen (K.T.J. and S.G.W.) and Bruker BioSpin GmbH (K.T.J.) is gratefully acknowledged. Part of the work was funded by the project FungalFight, The Danish Council for Strategic Research (Grant 10-093473). The grant for upgrade of the hyphenation equipment used in this work was provided from the Danish Agency for Science, Technology and Innovation via National Research Infrastructure funds. F

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry



Article

REFERENCES

(1) Johansen, K. T.; Wubshet, S. G.; Nyberg, N. T.; Jaroszewski, J. W. J. Nat. Prod. 2011, 74 (11), 2454−2461. (2) Brkljača, R.; Urban, S. J. Liq. Chromatogr. Relat. Technol. 2011, 34 (13), 1063−1076. (3) Wolfender, J.-L.; Marti, G.; Queiroz, E. F. Curr. Org. Chem. 2010, 14 (16), 1808−1832. (4) Staerk, D.; Kesting, J. R.; Sairafianpour, M.; Witt, M.; Asili, J.; Emami, S. A.; Jaroszewski, J. W. Phytochemistry 2009, 70 (8), 1055− 1061. (5) Lambert, M.; Staerk, D.; Hansen, S. H.; Sairafianpour, M.; Jaroszewski, J. W. J. Nat. Prod. 2005, 68 (10), 1500−1509. (6) Jaroszewski, J. W. Planta Med. 2005, 71 (9), 795−802. (7) Lang, G.; Mayhudin, N. A.; Mitova, M. I.; Sun, L.; van der Sar, S.; Blunt, J. W.; Cole, A. L. J.; Ellis, G.; Laatsch, H.; Munro, M. H. G. J. Nat. Prod. 2008, 71 (9), 1595−1599. (8) Hansen, M. E.; Smedsgaard, J.; Larsen, T. O. Anal. Chem. 2005, 77 (21), 6805−6817. (9) Konishi, Y.; Kiyota, T.; Draghici, C.; Gao, J. M.; Yeboah, F.; Acoca, S.; Jarussophon, S.; Purisima, E. Anal. Chem. 2007, 79 (3), 1187−1197. (10) Fredenhagen, A.; Derrien, C.; Gassmann, E. J. Nat. Prod. 2005, 68 (3), 385−391. (11) Nielsen, K. F.; Månsson, M.; Rank, C.; Frisvad, J. C.; Larsen, T. O. J. Nat. Prod. 2011, 74 (11), 2338−2348. (12) Qiu, F.; Imai, A.; McAlpine, J. B.; Lankin, D. C.; Burton, I.; Karakach, T.; Farnsworth, N. R.; Chen, S. N.; Pauli, G. F. J. Nat. Prod. 2012, 75 (3), 432−443. (13) Sashidhara, K. V.; Rosaiah, J. N. Nat. Prod. Commun. 2007, 2 (2), 193−202. (14) Nyberg, N. T.; Baumann, H.; Kenne, L. Magn. Reson. Chem. 2001, 39 (5), 236−240. (15) Wubshet, S. G.; Johansen, K. T.; Nyberg, N. T.; Jaroszewski, J. W. J. Nat. Prod. 2012, 75 (5), 876−882. (16) Hilton, B. D.; Martin, G. E. J. Nat. Prod. 2010, 73 (9), 1465− 1469. (17) Kovacs, H.; Moskau, D.; Spraul, M. Prog. Nucl. Magn. Reson. Spectrosc. 2005, 46 (2−3), 131−155. (18) Bayer, E.; Albert, K.; Nieder, M.; Grom, E.; Keller, T. J. Chromatogr. 1979, 186, 497−507. (19) Schmidt, B.; Jaroszewski, J. W.; Bro, R.; Witt, M.; Staerk, D. Anal. Chem. 2008, 80 (6), 1978−1987. (20) Vliegenthart, J. F. G.; Dorland, L.; Vanhalbeek, H. Adv. Carbohydr. Chem. Biochem. 1983, 41, 209−374. (21) Jansson, P. E.; Kenne, L.; Widmalm, G. Carbohydr. Chem. 1987, 168 (1), 67−77. (22) Mitra, A.; Seaton, P. J.; Assarpour, R. A.; Williamson, T. Tetrahedron 1998, 54 (51), 15489−15498. (23) Pauli, G. F.; Kuczkowiak, U.; Nahrstedt, A. Magn. Reson. Chem. 1999, 37 (11), 827−836. (24) Tejesvi, M. V.; Kajula, M.; Mattila, S.; Pirttilä, A. M. Fungal Diversity 2011, 47 (1), 97−107. (25) Wubshet, S. G.; Nyberg, N. T.; Tejesvi, M. V.; Pirttilä, A. M.; Kajula, M.; Mattila, S.; Staerk, D. J. Chromatogr. A 2013, (submitted). (26) Keifer, P. A. J. Magn. Reson. 2009, 199 (1), 75−87. (27) Johansen, K. T.; Ebild, S. J.; Christensen, S. B.; Godejohann, M.; Jaroszewski, J. W. J. Chromatogr. A 2012, 1270, 171−177.

G

dx.doi.org/10.1021/ac303455j | Anal. Chem. XXXX, XXX, XXX−XXX