Content uploaded by Alice C L Churchill
Author content
All content in this area was uploaded by Alice C L Churchill on Oct 05, 2021
Content may be subject to copyright.
NMR Spectroscopy DOI: 10.1002/anie.200603821
Differential Analysis of 2D NMR Spectra: New Natural Products from
a Pilot-Scale Fungal Extract Library**
Frank C. Schroeder,* Donna M. Gibson, Alice C. L. Churchill, Punchapat Sojikul,
Eric J. Wursthorn, Stuart B. Krasnoff, and Jon Clardy
The efficient analysis of small-molecule mixtures underlies
many endeavors in chemical biology. The sensitivity of mass
spectrometry (MS) has resulted in its widespread adoption for
such analyses, and today rapid automated LC-MS analyses
are widely used. Several recent studies have demonstrated the
feasibility of NMR spectroscopic analyses of complex small-
molecule mixtures, including the use of diffusion-ordered
spectroscopy (DOSY)[1] or principal component analysis
(PCA) in metabolomics,[2] as well as the characterization of
crude unfractionated natural product extracts using routine
two-dimensional NMR spectra.[3] Compared to MS analyses,
2D NMR spectroscopic investigations of small-molecule
mixtures offer the benefit of more detailed structural
information, which is of particular relevance for the detection
of novel chemotypes. However, the complexity of 2D spectra
typically obtained for small-molecule mixtures has limited a
broader implementation of NMR spectroscopy for their
characterization. Herein, we describe a simple procedure
for the differential analysis of arrays of 2D NMR spectra and
demonstrate its utility for the detection of new natural
products from a small library of fungal extracts.
Fungi are prolific producers of natural products derived
from terpene,[4] polyketide,[5] and nonribosomal peptide path-
ways.[6] Several lines of evidence indicate that only a fraction
of the biosynthetic capabilities of fungi (and other cultured
organisms) are discovered in traditional screening operations,
as most secondary metabolite pathways are not expressed
under the culture conditions used. Various approaches are
being pursued to increase the accessible fraction of fungal
metabolomes, and anecdotal evidence suggests that fungi
respond to even small variations in their culturing protocol by
starting (or stopping) the biosynthesis of specific natural
products.[7–11] Clearly, a more systematic exploration of factors
modulating secondary metabolite biosynthesis in fungi (and,
by the same token, in bacteria) would be highly desirable. In a
pilot study, we used differential analyses of 2D NMR spectra
for the characterization of a small library of fungal extracts
derived from a Tolypocladium cylindrosporum strain, cul-
tured with a variety of protocols, which quickly revealed two
new terpenoid indole alkaloids.
T. cylindrosporum strain TC705 was selected from a group
of insect-pathogenic fungi[12] because it has a number of
nonribosomal peptide and polyketide biosynthetic genes that
suggest a high metabolic potential for the production of
secondary metabolites.[13] For our studies, TC705 cultures
were grown using seven different protocols, based on four
different media (YM, SDY, mEM, and diEM; see Supporting
Information for full details). Three protocols (YM-SDY, YM-
mEM, and YM-diEM) included growing cultures in a two-
step fermentation procedure, whereby each culture is initi-
ated using a nutrient-rich medium and then transferred to a
minimal or partially nutrient-deficient medium.[14] For sub-
sequent NMR spectroscopic analyses, ethyl acetate extracts of
the fungal broths were used.[14]
The initial NMR spectroscopic analysis of the unfractio-
nated extracts was based on double quantum filtered
correlation spectroscopy (DQF-COSY), as previous experi-
ence had shown that a single DQF-COSY spectrum often
provides sufficient information to recognize the presence of
significant quantities of any unusual small molecules.[3] DQF-
COSY spectra were acquired for 25 extracts derived from
three repetitions of the seven culturing protocols and four
media controls, using a set of acquisition parameters opti-
mized for high resolution in both frequency dimensions. As
expected, the resulting DQF-COSY spectra were extremely
complex, and a detailed cross-peak-by-cross-peak analysis of
all 25 spectra was not feasible. To address this challenge, we
developed a simple two-step protocol for a differential
analysis of the DQF-COSY spectra (Scheme 1).
The first step consisted of a graphical analysis based on
multiplicative stacking of bitmaps derived from magnitude
mode versions of the DQF-COSY spectra.[15] This technique
clearly distinguished signals present in only one spectrum
from signals common to several spectra. For example, overlay
[*] Dr. F. C. Schroeder, Prof. Dr. J. Clardy
Biological Chemistry and Molecular Pharmacology
Harvard Medical School
Boston, MA 02115 (USA)
Fax: (+1) 607-255-3407
E-mail: Frank_Schroeder@harvard.edu
Prof. Dr. D. M. Gibson
USDA-ARS-Plant Protection Research Unit
Ithaca, NY 14853 (USA)
Dr. A. C. L. Churchill, Dr. P. Sojikul
Boyce Thompson Institute for Plant Research
Ithaca, NY 14853 (USA)
Dr. S. B. Krasnoff
Department of Plant Pathology,
Cornell University, Ithaca, NY 14853 (USA)
E. J. Wursthorn
Department of Chemistry and Chemical Biology
Cornell University, Ithaca, NY 14853 (USA)
[**] This work was supported by a grant from the NIH (CA59021 to J.C.).
F.C.S. and E.J.W. thank Jerrold Meinwald (Cornell University) and
Matthew Gronquist (SUNY Fredonia) for their interest in the project
and helpful discussions.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angewandte
Chemie
901Angew. Chem. Int. Ed. 2007,46, 901–904 !2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
of the spectrum obtained for the YM-SDY protocol (Fig-
ure 1a) with the three spectra obtained for the SDY-only
protocol, the YM medium control, and the SDY medium
control showed that most signals present in the YM-SDY
spectrum correspond to compounds also present in the media
controls or the SDY-only extract, as indicated by partial
extinction and strong color shifts of these signals (Figure 1 b).
Only a small subset of signals in the YM-SDY spectrum
remained unaffected by superposition of the three other
spectra. These signals correspond to compound(s) present
only in the YM-SDY extract. In this manner, a simple
graphical manipulation of the COSY-derived bitmaps, which
can be accomplished using commonly available image editing
software, clearly distinguished signals corresponding to com-
pounds produced only under a specific culturing protocol
from signals of compounds produced under most conditions.
The graphical manipulation of the bitmap spectra described
here is significantly more efficient than subtraction of spectra,
because it results in obvious color shifts and partial extinction
of signals common to two spectra even in cases where the
concentration of the corresponding compound in the two
extracts being compared is vastly different. In fact, the
efficacy of this comparison method is limited primarily by the
dynamic range and sensitivity of the NMR spectrometer.[15]
The second step consisted of a more detailed analysis of
the signals representing unique or unusual metabolites in a
specific extract. The corresponding spin systems were char-
acterized based on the phase-sensitive originals of the DQF-
COSY spectra (Step 2 in Scheme 1; Figure 1c). For extracts
containing structurally intriguing components, additional
heteronuclear single quantum correlation (HSQC) and het-
eronuclear multiple bond correlation (HMBC) spectra were
acquired.
Subjecting the DQF-COSY spectra of the extracts derived
from the seven culturing protocols and media controls to this
evaluation protocol immediately revealed significant differ-
ences. Extracts derived from protocols using mEM or diEM
Scheme 1. Two-step differential analysis of DQF-COSY spectra
obtained for a library of fungal extracts.
Figure 1. a) Section of the magnitude-mode DQF-COSY spectrum
obtained for the YM-SDY extract. b) Same spectrum after multiplicative
stacking with spectra for the YM extract, SDY extract, and SDY
medium control, showing partial extinction and strong color shifts for
signals that are present in the YM-SDY spectrum and in at least one of
the YM, SDY, and SDY medium spectra. Cross-peaks unaffected by the
multiplication layers represent compounds present only in the YM-SDY
extract. Cross-peaks marked with black rectangles correspond to
compounds 6and 7, whereas those marked with red rectangles
represent fatty acid ethanolamides. c) Phase-sensitive representation
of the DQF-COSY spectrum used for detailed characterization of the
spin systems of 6and 7.
Communications
902 www.angewandte.org !2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2007,46, 901–904
media contained large amounts of 3-hydroxyisobutyric acid
(1) and its acetyl derivative 2, whereas the furan derivatives 3
and 4were produced only under the two protocols using
mEM medium (Scheme 2).[16] The analyses of compounds
specific to the three two-media combination protocols were
particularly interesting. Under the YM-mEM protocol, large
amounts of compounds represented by several spin systems in
the aromatic and aliphatic regions were produced. These
compounds were also detected in extracts derived from the
YM-SDY protocol, although in smaller amounts (Figure 1 b).
Analysis of the YM-mEM DQF-COSY spectra and addi-
tional HSQC, HMBC, and NOESY spectra of the unfrac-
tioned extracts suggested structures 6and 7, which represent
two previously unreported terpenoid indole alkaloids
(Scheme 2).[17–19] As these alkaloids constitute new natural
products, we subsequently isolated 6and 7from the best-
producing YM-mEM media combination by reversed-phase
HPLC and confirmed the structural assignments shown.[20]
Detailed analysis of the DQF-COSY spectra derived from
the seven fermentation protocols showed that 6and 7are
consistently produced in high yields only in the YM-mEM
protocol, while the amounts produced under the YM-SDY
protocol varied. Furthermore, analysis of the extracts by both
NMR spectroscopy and HPLC-UV confirmed that extracts
derived from single-medium mEM or SDY cultures contained
none or only trace amounts of 6and 7.[14]
Only one of the major components in these extracts was
produced under all seven protocols. This compound was
identified as the known fungal metabolite pyridoxatin (5),
which is produced as a major component under all but one of
seven protocols.[21] The exception was the YM-mEM media
combination, in which case only trace amounts of 5were
found.
To validate the results of our NMR-based analyses, all
media extracts were subjected to additional HPLC/electro-
spray ionization (ESI) MS analyses, which showed significant
differences between the various extracts as well. However,
the ESI mass spectra alone provided little structural informa-
tion compared to the NMR spectroscopic analyses. Further-
more, positive electrospray ionization efficiencies of secon-
dary metabolites identical to “secondary metabolites”, such
as pyridoxatin (5) or the terpenoids 6and 7, were orders of
magnitude lower than those for peptides and other amino acid
derivatives, which resulted in a strongly skewed representa-
tion of the actual compositions. Accordingly, the strongest
peaks in the HPLC/ESI-MS analyses represented amino acids
and several series of oligopeptides, whereas the NMR spectra
indicated that peptides account for only a small fraction of the
total extracts. The major peptide components as identified by
HPLC/ESI-MS were a series of efrapeptins,[22] variable
amounts of which were produced under all protocols. In
addition, two series of as-yet unidentified peptides were
produced under the YM-mEM protocol. In this regard,
HPLC/ESI-MS and NMR spectroscopic analyses comple-
ment each other.
These results show that, as predicted by earlier PCR
analysis,[13] the metabolism of TC705 is highly variable and
responds strongly to changes in culturing conditions. The two-
stage differential analysis of NMR spectra obtained for the
unfractionated extracts allowed the rapid detection of two
new natural products.[23] The scope of such NMR spectro-
scopic characterization of largely unfractionated extracts
from fungal, bacterial, and other sources could be easily
extended. The analyses described here are primarily limited
by the finite dynamic range of NMR spectroscopy, and as a
consequence, most components accounting for less than a few
percent of the total extracts cannot be reliably characterized
because signal-to-noise ratios for the corresponding signals
are too low. Compounds missed by the NMR spectroscopic
analysis included the various oligopeptides that were detected
by LC-MS. Detection limits could be lowered considerably by
including a coarse prefractionation of the extracts prior to
NMR spectroscopic analysis. As graphical comparison of the
DQF-COSY spectra is fast, the corresponding increase in the
number of spectra could be easily addressed. Acquisition of
larger numbers of spectra could be accomplished for example
by using recently introduced capillary NMR technology
(CapNMR).[24]
In comparing our analyses to other NMR-based
approaches for characterizing complex mixtures of small
molecules,[1,2] it should be noted that our primary goal was the
detection and characterization of novel metabolites. Our
approach thus focuses on extracting structural information
(connectivity information) instead of determining character-
istic quantitative differences in integrated signal intensity.
Differential analysis of NMR spectra provides a useful
tool for a non-discriminatory characterization of small-
molecule mixtures, with many potential applications in
metabolomics and natural products chemistry. Among these,
the possibility of complementing bacterial and fungal genetics
Scheme 2.
Angewandte
Chemie
903Angew. Chem. Int. Ed. 2007,46, 901–904 !2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org
with NMR-based differential analysis of corresponding
changes in secondary metabolite production is particularly
intriguing.
Received: September 18, 2006
Published online: December 20, 2006
.
Keywords: high-throughput screening · metabolism ·
natural products · NMR spectroscopy · structure elucidation
[1] a) J. C. Cobas, P. Groves, M. Martin-Pastor, A. De Capua, Curr.
Anal. Chem. 2005,1, 289 – 305; b) R. T. Williamson, E. L.
Chapin, A. W. Carr, J. R. Gilbert, P. R. Graupner, P. Lewer, P.
McKamey, J. R. Carney, W. H. Gerwick, Org. Lett. 2000,2, 289 –
292.
[2] S. Halouska, R. Powers, J. Magn. Reson. 2006,178, 88 – 95.
[3] a) A. E. Taggi, J. Meinwald, F. C. Schroeder, J. Am. Chem. Soc.
2004,126, 10364–10 369; b) A. T. Dossey, S. S. Walse, J. R.
Rocca, A. S. Edison, ACS Chem. Biol. 2006,1, 511.
[4] S. Inouye, S. Abe, H. Yamaguchi, Mycol. Ser. 2004,20, 379 – 399
(Handbook of Fungal Biotechnology (2nd ed.)).
[5] B. J. Rawlings, Nat. Prod. Rep. 1999,16, 425 – 484.
[6] a) T. B. Ng, Peptides 2004,25, 1055 – 1073; b) H. von D!hren,
Adv. Biochem. Eng./Biotechnol. 2004,88, 217 –264 ; c) R. J. Cole,
M.A. Schweikert, Handbook of Secondary Fungal Metabolites,
Academic Press, San Diego, USA, 2003.
[7] a) C. L. Preisg, J. A. Laakso, U. M. Mocek, P. T. Wang, J. Baez,
G. Byng, J. Nat. Prod. 2003,66, 350 – 356 ; b) Y. Okuno, M.
Miyazawa, J. Nat. Prod. 2004, 67, 1876–1878 ; c) Y. Tian, H. Guo,
J. Han, D. Guo, J. Nat. Prod. 2005,68, 678 – 680.
[8] H. Oikawa, Y. Murakami, A. Ichihara, J. Chem. Soc. Perkin
Trans. 1 1992, 2949 – 2953.
[9] O. E. Christian, J. Compton, K. R. Christian, S. L. Mooberry,
F. A. Valeriote, P. Crews, J. Nat. Prod. 2005,68, 1592 – 1597.
[10] A. M. Calvo, R. A. Wilson, J. W. Bok, N. P. Keller, Microbiol.
Mol. Biol. Rev. 2002,66(3), 447 – 459.
[11] H. B. Bode, B. Bethe, R. H!fs, A. Zeeck, ChemBioChem 2002,3,
619 – 627.
[12] T. Lee, S-H. Yun, K. T. Hodge, R. A. Humber, S. B. Krasnoff,
B. G. Turgeon, O. C. Yoder, D. M. Gibson, Appl. Microbiol.
Biotechnol. 2001,56, 181 – 187.
[13] C. M. Ireland, W. Aalbersberg, R. J. Andersen, S. Ayral-Kalous-
tian, R. Berlinck, V. S. Bernan, G. T. Carter, A. C. L. Churchill, J.
Clardy, G. P. Concepcion, E. D. De Silva, C. Discafani, T. Fojo, P.
Frost, D. Gibson, L. M. Greenberger, M. Greenstein, M. K.
Harper, R. Mallon, F. Loganzo, M. Nunes, M. S. Poruchynsky, A.
Zask, Pharm. Biol. 2003,41, Supplement: 15 –38.
[14] See Supporting Information for a detailed description of the
culturing conditions and analytical procedures.
[15] See Supporting Information for a detailed description of the
“multiplicative stacking” technique for differential analysis of
the DQF-COSY spectra.
[16] R. Jadulco, P. Proksch, V. Wray, A. B. Sudarsono, U. Gr"fe, J.
Nat. Prod. 2001,64, 527 – 530.
[17] C. Li, J. B. Gloer, D. T. Wicklow, P. F. Dowd, Org. Lett. 2002,4,
3095 – 3098.
[18] M. C. Gonzales, C. Lull, P. Moya, I. Ayala, J. Primo, E. P. Yufera,
J. Agric. Food Chem. 2003,51, 2156 – 2160.
[19] H. Tomoda, N. Tabata, D.-J. Yang, H. Takayanagi, S. Omura, J.
Antibiot. 1995,48, 793 – 804.
[20] The configuration at C36 in the isopentyl side chains of 6and 7
and at C31 in 7was not determined.
[21] A. Jegorov, V. Matha, M. Husak, B. Kratochvil, J. Stuchlik, J.
Chem. Soc. Dalton Trans. 1993, 1287 – 1294.
[22] S. Gupta, S. B. Krasnoff, D. W. Roberts, J. A. A. Renwick, L.
Brinen, J. Clardy, J. Org. Chem. 1992,57, 2306 – 2313.
[23] Terpenoid indole alkaloids structurally related to 6and 7show
potent anti-insectan properties,[17,18] which seems intriguing
given that TC705 was isolated as an entomopathogenic fungus.
[24] a) M. Gronquist, J. Meinwald, T. Eisner, F. C. Schroeder, J. Am.
Chem. Soc. 2005,127, 10810– 10 811; b) F. C. Schroeder, M.
Gronquist, Angew. Chem. 2006,118, 7280 – 7290 ; Angew. Chem.
Int. Ed. 2006,45, 7122 – 7131.
Communications
904 www.angewandte.org !2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2007,46, 901–904
Supporting Information
© Wiley-VCH 2006
69451 Weinheim, Germany
1
Differential Analysis of 2D NMR Spectra:
New Natural Products from a Pilot-Scale Fungal Extract Library
Frank C. Schroeder,†,*, Donna M. Gibson,‡ Alice C. L. Churchill,║ Punchapat Sojikul,║
Eric J. Wursthorn, § Stuart B. Krasnoff,††, and Jon Clardy†
†Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston,
Massachusetts 02115, ‡USDA-ARS-Plant Protection Research Unit, Ithaca, New York 14853, ║Boyce
Thompson Institute for Plant Research, Ithaca, New York 14853, §Department of Chemistry and
††Department of Plant Pathology, Cornell University, Ithaca, New York 14853,
1. Strain and culture conditions. Cryogenically preserved mycelium of Tolypocladium
cylindrosporum W. Gams (ARSEF 705) was used to initiate cultures grown on plates of
Sabouraud’s dextrose agar plus yeast extract (1%). Plugs (10; 4 mm diameter) from 3-4-
week-old solid cultures were used to start liquid cultures.
For comparison of media effects on secondary metabolite production, 250 ml Erlenmeyer
flasks, each containing 100 ml of medium, were incubated in triplicate on a rotary shaker
(160 rpm, 24 + 2oC) for 14 days and harvested. For two-stage fermentation cultures, 10
plugs (4 mm diameter) were transferred to yeast malt (YM) broth (4 g yeast extract, 20 g
malt extract in 1 L deionized water), which was incubated for 3 days as above. An aliquot
(10 ml) of the resulting culture was then used to inoculate 100 ml of a second stage
medium, which was incubated for an additional 14 days before harvesting. For larger
scale cultures, 100 ml of culture grown in YM for 3 days was used to inoculate 1L of
second stage medium in 2.4L Fernbach flasks. Second stage media included two standard
media, Sabouraud’s dextrose broth + yeast extract (1%) and YM broth, as well as two
new formulations intended to provide stress conditions and affect fungal morphogenesis.
The mEM medium consisted of the following: L-sorbose, 20 g; KH2PO4, 0.8 g;
(NH4)H2PO4, 10 g; MgSO4.7H2O, 0.5 g; CaCl2 2H2O, 0.1 g; L-valine, 6 g; trace elements
solution, 1 ml, per liter, with a final pH of 4.5. The diEM medium contained the same
ingredients as mEM except that the (NH4)H2PO4 was replaced with the same amount, 10
g, of (NH4)2HPO4, which resulted in a pH of 7.2 for this medium. The trace elements
solution contained ZnCl2, 40 mg; FeCl3.6H2O, 200 mg; CuCl2.2H2O, 10 mg;
MnCl2.4H2O, 10 mg; Na2B4O7.10H2O, 10 mg; (NH4)6Mo7O24.4H2O, 10 mg; in a total
volume of 500 ml.
2. Extraction. Mycelium was separated from broth by vacuum filtration, then the broth
was extracted with three 50 ml aliquots of ethyl acetate per 100 ml of broth. The organic
layers were combined, dried over anhydrous Na2SO4, and the solvent was removed in
2
vacuo to afford an oily light brown residue. For each of the seven media combinations, at
least three extracts from three separate cultures were prepared. The choice of ethyl
acetate for extraction likely excluded from analysis some of the more polar secondary
products produced in culture; however, we felt that for this model study, it would be
appropriate to limit the scope of the analysis to the characterization of compounds of
intermediate to low polarity. After evaporation, aliquots of 30-40 mg of the ethyl acetate
extracts were suspended in CD3OD, filtered over a plug of cotton, and subjected to NMR
spectroscopy. Using more than 30-40 mg of evaporated extract per NMR sample did not
improve results because many compounds’ chemical shift values are slightly
concentration dependent and thus increasing concentration would reduce efficacy of the
ensuing overlay-based differential analysis.
3. NMR spectroscopy. All analyses were carried out using a VARIAN INOVA 600
MHz NMR spectrometer, which was equipped with a 5 mm inverse-detection HCN
probe. Shigemi tubes were used as needed. NMR spectra were acquired at 25 °C, using
the standard pulse sequences provided by VARIAN. For each fungal extract, a standard
phase-cycled phase-sensitive dqfCOSY spectrum was acquired, using the following
parameters: 600 ms acquisition time, 400-600 complex increments in F1, and 4, 8, or 16
scans per increment. Sweep width and the numbers of increments (ni) were chosen so
that the digital resolution of spectra obtained for different extracts was identical or at least
very similar. For further characterization of selected fungal extracts, additional gradient
HMQC (phase-sensitive), gradient HSQC (phase-sensitive), gradient or phase-cycled
HMBC, phase-cycled phase-sensitive NOESY spectra (using a mixing time of 600 ms) or
phase-sensitive ROESY spectra (mixing time 300 ms) were acquired.
4. Differential analysis.
4.1. Processing of the dqfCOSY spectra. Spectra were processed using either Varian’s
VNMR software or MestReC (www.mestrec.com). For magnitude-mode display, dqf-
COSY spectra were processed in MestReC using 4096 complex data points in F2, 2048
complex data points in F1, and squared sine bell functions as window functions in both
F2 and F1. For phase-sensitive display, a 60 degree-shifted squared sine bell was used for
windowing in F2 and a 90 degree shifted sine bell for windowing in F1.
4.2. Stacking of dqfCOSY spectra. For graphical comparison of magnitude-mode dqf-
COSY spectra, bitmaps from the MestReC on-screen display of the spectra, using the
default color scheme1, were imported into Adobe Photoshop (version CS2). Bitmaps
from a set of spectra to be compared were placed into two sets of layers of a single
Photoshop document. Using the various blending modes available in Photoshop, spectra
can be overlaid and compared in several different ways. For the purpose of detecting
differences between spectra, the dqf-COSY spectra were arranged as follows. On top of
a white background layer, seven layers were placed containing magnitude-mode-dqf-
COSY spectra from extracts corresponding to the seven culturing protocols. For these
seven layers, the blending mode was set to “difference”. On top of this stack of spectra,
another eleven layers were placed containing the seven dqf-COSY spectra from the same
1 A constant-hue color scheme is needed for the method of graphical comparison described here. The
MestReC default color scheme uses a light yellow-to-brown color scheme with sufficient hue accuracy.
3
seven extracts and additionally four spectra derived from media control extracts. For
these eleven layers, “multiply” was used as the blending mode. Finally, a “curves”
adjustment layer was placed as the top-most layer. This curves layer was used to invert
the color scheme and adjust gamma of the display to increase or decrease contrast. In
order to compare one spectrum – for example the spectrum obtained for the YM-SDY
protocol – with those of other protocols or media controls, this stack of 20 layers was
used as follows. Initially, visibility of all layers containing spectra was turned off, except
for the layer containing the YM-SDY spectrum in “difference” mode. The resulting
display is shown in Figure SI-1.
4.3. Identification of media-specific components. To compare the YM-SDY extract
with a media control extract, visibility of the “multiply” SDY media control layer was
turned on. The resulting display is shown in Figure SI-2. Signals representing media-
derived compounds are partially extinct and appear in blue or magenta, while signals
representing fungal metabolites appear unaltered. Turning on the visibility of additional
“multiply”-mode layers can be used to include additional extracts in this comparison. For
example, turning on visibility of the layer containing the spectrum obtained for the YM
fungal extract results in partial extinction and color changes for additional sets of signals,
which, as analyses of additional spectra showed, represent the two rotamers of
pyridoxatin, 1 (Figure SI-3). The remaining unaltered crosspeaks represent small amounts
of the terpenoid indole alkaloids 6 and 7, in addition to small quantities of 3-
hydroxypropionic acid derivatives.
4.4. Alternative approaches to differential analysis of 2D-NMR spectra. The
graphical comparison of 2D-NMR spectra as described above can be regarded as a
“weighing” of one spectrum against a series of related spectra. Essentially, the “multiply”
function weighs the RGB values of a given data point against the RGB values of
corresponding data points in the overlaid spectra. The use of the “multiply” blending
mode as a simple out-of-the-box weighing mode is convenient; however, other blending
modes could be designed that would provide even better suppression of signals not
unique to the spectrum under scrutiny. For example, simply subtracting the overlaid
spectrum in Figure SI-3 from the original spectrum in Figure SI-1 would yield a mask
that could be used to completely block out any partially extinct or color shifted signals in
Figure SI-3. In fact, in our experience color shifts of signals were generally better to
work with than complete signal extinction, because the preserved multiplicity structures
of partially suppressed peaks allows to confirm that they indeed represent
superposition(s) of equivalent crosspeaks in the compared spectra, as opposed to mutual
extinction of unrelated crosspeaks that accidentally occur at similar chemical shift values
in the compared spectra.
Comparing bitmaps derived from the on-screen display of spectra avoids the difficulties
associated with manipulating stacks of actual 2D-NMR data files. Such analyses could be
achieved using specialized NMR software packages; however, from our experience, these
rather expensive programs do not offer the flexibility required for fast comparison of
stacks of many complete high-resolution 2D spectra.
4
It should be noted that complete characterization of any interesting compounds detected
by dqf-COSY differential analysis is not always possible, due to excessive overlap in
either the dqf-COSY spectra or in the supplemental HMBC and HMQC spectra. In our
example, the terpenoid indole alkaloids 6 and 7 were originally detected in the YM-SDY
extracts, where 6 and 7 represent minor components, and then fully characterized using
the spectra of the YM-mEM extracts, which contain these compounds in larger quantities.
Nonetheless, analysis of the YM-SDY spectra alone produced sufficient information to
determine the key structural features of 6 and 7, and provided the motivation to pursue
further characterization and isolation of these compounds.
Instead of dqfCOSY spectra, several other types of two-dimensional NMR spectra could
be used for the differential analysis of complex small molecule mixtures, for example
TOCSY spectra with short mixing times. Phase-sensitive TOCSY spectra exhibit good
peak shape and excellent sensitivity, and short mixing times could be used to prevent
over-crowding of the spectra. However, for the differential analysis of fungal extracts,
which usually are not mass-limited, the slightly higher sensitivity of the TOCSY
experiment does not confer any significant advantage. We chose dqfCOSY spectra over
TOCSY spectra because the fine structure of the dqfCOSY crosspeaks allows extracting
more structural information (coupling constant information) than would be possible to
derive from TOCSY crosspeaks. This difference becomes relevant in the second step of
the differential analysis, in which signals identified as representing compounds of interest
get analyzed in greater detail. For this second step, the dqfCOSY method thus provides a
clear advantage. Nonetheless, for mass-limited samples, differential analysis based on
TOCSY spectra could be useful.
5
Figure SI-1. Magnitude-mode YM-SDY spectrum.
6
Figure SI-2. Magnitude-mode YM-SDY spectrum after overlay with the spectrum for the
SDY media control.
7
Figure SI-3. Magnitude-mode YM-SDY spectrum after overlay with the spectra for the
SDY medium control and the YM fungal culture. Note extinction and color shifts for the
signals representing pyridoxatin (5); for example, the crosspeaks for two of the
pyridoxatin methyl groups at 0.95/2.5 and 1.15/2.5 ppm. Most of the remaining
crosspeaks represent spin systems that belong to the terpenoid indole alkaloids 6 and 7.
8
Figure SI-4. Small section of the magnitude-mode COSY spectrum obtained for the YM-
mEM extract. B: Same spectrum after multiplication with spectra for the YM extract,
mEM extract, and mEM medium control. Crosspeaks labeled with black rectangles
represent terpenoid alkaloids 6 and 7.
5. HPLC-MS characterization of extracts. HPLC-MS analyses were carried out using
a Micromass Quattro I tandem mass spectrometer operated in positive-ion electrospray
mode and an Agilent Series 1100 liquid chromatograph (10 x 250 mm Supelco 5 µ ODS
preparative column eluted at a flow rate of 3.4 mL/min) equipped with a diode array
detector. A gradient elution was used, which started with a solvent composition of 5%
methanol and 95% water for three minutes, and then progressed to 100% methanol by 40
minutes. After UV-detection and before ESI-MS detection, the HPLC effluent was split
40:1. Prior to injection onto the HPLC, fungal extracts were dissolved in methanol and
filtered over glass wool.
6. Media-derived components. These included small amounts of free amino acids,
which were found in all of the extracts analyzed and, in the case of the diEM-media-
derived extracts, large amounts of various hydroxyalkylpyrazines, such as 8 and 9, which
are typical for the Maillard browning reaction and whose presence was not surprising
given the alkaline nature of the diEM medium. Furthermore, the extracts derived from the
mEM-media contained small amounts of several pyrrole-derivatives, such as 10,
representing another kind of condensation product of amino acids and glucose.1 SDY-
medium extracts revealed large amounts of bile acids and diketopiperazines, such as 11-
13.
9
O
N
O
O
HN
N
O
O
OH HN
N
O
O
HN
N
O
O
N
NOH
N
NOH
8910
11 12 13
7. Isolation of indole alkaloids 6 and 7. Ethyl acetate extracts derived from 1 L of YM-
mEM culture were subjected to reversed-phase column chromatography using octadecyl-
functionalized silica gel (Aldrich) and methanol containing 80-0% water as solvent.
Fractions containing 6 and 7, as detected by NMR spectroscopic analysis, were pooled
and subjected to HPLC. Preparative HPLC was carried out using an Agilent Series 1100
liquid chromatograph (10 x 250 mm Supelco 5 µ ODS preparative column eluted at a
flow rate of 3.4 mL/min) equipped with a diode array detector set to monitor wavelengths
of 230 nm and 280 nm. Gradient elution was used starting with a solvent composition of
60% methanol and 40% water, which was maintained for three minutes, and then
progressed to 100% methanol by 30 minutes. Under these conditions, compound 6 eluted
at 23.0 min and compound 7 at 21.4 min.
8. HPLC screening for the presence of trace amounts of 6 and 7 in single-medium
extracts. Using the HPLC system (UV detection) and conditions described above, single-
medium extracts were analyzed for the presence of small amounts of 6 and 7. In those
mEM and SDY repeats that showed trace amounts of the two terpenoids, their presence
was determined using isolated samples of the compounds as reference standards, showing
that the terpenoid content of these extracts was lower than 0.5%.
10
9. Spectroscopic data for compound 6.
NH
OH
OH
OH
O
O
O
OH
OH
O
H
H
H
HH
6, TC-705A
23
4
5
6
79
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24 25
26
35
36
37
38
39
27
28 29
30
31
32
33
34
9.1. NMR-spectroscopic data of 6, using CD3OD as solvent.
Chemical shifts were referenced to δ(CD2HOD) = 3.31 ppm and δ(CD2HOD) = 49.05
ppm. 13C-1H coupling constants were obtained from HMQC spectra, and 1H-1H coupling
constants from dqf-COSY spectra. NOE crosspeaks were qualitatively classified as
strong (s), medium strong (m) or weak (w). Coupling constants are given in Hertz [Hz].
Position
δ 13C [ppm]
13C-1H
coupling
constants
δ 1H [ppm]
1H-1H
coupling
constants
Relevant HMBC
Correlations
Relevant NOESY
Correlations
1 - - -
2 153.0 - -
3 51.9 - -
4 43.5 -
5a
27.0 JC-H = 125.0 1.67 J5a, 5b = 12,
J5a,6a = 5.6,
J5a,6b < 1
C-3, C-4, C-6, C-7,
C-13
H-26 (s)
5b 2.61 J5b,6a = 14,
J5a,6b = 5.3
C-6, C-4, C-26, C-7 H-25 (s)
6a 29.4 JC-H = 131.7 1.82 J6a,6b = 13, J6a,7
= 9.6
C-5, C-7 H-26 (s)
6b 2.24 J6b,7 = 7.9 C-5, C-12
7 72.8 JC-H = 150.0 4.33 J7,11 = 1 C-6, C-11, C-12 H-9 (m), H-28 (w)
9 72.3 JC-H = 148.4 3.58 J9,10 = 9.6 C-10, C-11, C-27,
C-28, C-29
10 72.4 JC-H = 141.7 3.95 J10,11 = 1 C-7 C-9, C-27,
C-30
H-29 (s), H-30 (s)
11 60.9 JC-H = 183.4 3.62 C-9, C-10, C-12,
C-13
H-14b (s),
H-14a (w)
12 68.7 - -
13 78.4 - -
14a 30.3 JC-H = 131.7 1.68 J14a,14b = 13.5,
J14a,15a = J14a,15b
= 2.5
C-14,C-16
11
14b 1.52 J14b,15a = 11.3,
J14b,15b = 3.7
C-13, C-14,
C-15
H-26 (s)
15a 21.6 JC-H = 128.4 1.94 J15a,15b = 13.5,
J15a,16 = 13.5
C-14, C-16 H-25 (s), H-14 (w)
15b 1.58 J15b,16 = 1.5 H-14 (s)
16 51.3 JC-H = 120.0 2.82 J16,17a = 10.8,
J16,17b = 6.5
C-3,C-25 H-26 (s)
17a 27.9 JC-H = 131.7 2.33 J17a,17b = 14.5 C-15, C-16,
C-18, C-2
H-25 (s)
17b 2.62 C-2, C-18, C-3
18 116.7 - -
19 124.5 - -
20 118.2 JC-H = 156.7 7.20 J20,21 = 8.3 C-18, C-22, C-24
21 121.2 JC-H =156.7 6.87 J21,23 = 1.5 C-19, C-23, C-35
22 132.7 - -
23 113.1 JC-H = 158.4 7.21 C-19, C-21, C-35
24 142.0 - -
25 16.2 JC-H = 126.7 1.27 C-2, C-3, C-4
26 18.7 JC-H = 125.0 1.15 C-3, C-4, C-5, C-13
27 76.25 - -
28 28.4 JC-H = 126.7 1.26 C-9, C-27, C-29
29 16.7 JC-H = 126.7 1.28 C-9, C-27, C-28
30 96.90 JC-H = 165.0 4.98 J30,31 = 4.6 C-32, C-10 H-29 (s)
31 72.44
4
JC-H = 141.7 3.27 C-30, C-33, C-34
32 78.4 - -
33 26.4 JC-H = 125.0 1.20 C-31, C-32, C-34
34 25.3 JC-H = 125.0 1.21 C-31, C-32, C-33
35a 39.0 JC-H = 123.4 2.53 J35a,36 =10.3,
J35a,35b =14.5
C-36, C-37, C-23,
C-21, C-22
35b 3.06 J35b,36 = 1.5 C-23, C-22, C-21,
C-36
36 81.4 JC-H = 141.7 3.55 C-22, C-37, C-38,
C-39
37 73.6 - -
38 24.5 JC-H = 126.7 1.23 C-36, C-37, C-39
39 25.9 JC-H = 125.0 1.25 C-36, C-37, C-38
12
9.2. Additional NMR-spectroscopic data of 6, using DMSO-d6 as solvent.
For corroboration of the stereochemical assignments at positions 3, 4, 7, 9, 10, 11, 12, 13,
and 16 in compound 6 an additional set of spectra using DMSO-d6 as solvent was
acquired. Using DMSO-d6 as the solvent revealed proton signals corresponding to the
various OH-groups in 6. Chemical shifts were referenced to δ(CD2HCOCD3) = 2.48 ppm
and δ(CD3COCD3) = 39.51 ppm. ROESY-crosspeaks were qualitatively classified as
strong (s), medium strong (m) or weak (w).
Position
δ 13C [ppm]
δ 1H [ppm]
Relevant HMBC
Correlations
Relevant ROESY
Correlations
1 - -10.47 -
2 151.73 - -
3 50.13 - -
4 42.04 -
5a
25.37 1.67 C-4, C-6, C-7, C-13 H-26 (s)
5b 2.40 C-6, C-4, C-26, 13-OH H-25 (s)
6a 28.40 1.64 H-26 (s), H-11 (w), H-10 (w)
6b 2.13 C-4, C-7, C-12
7 70.48 4.27 C-6, C-11, C-12 H-9 (m), H-28 (w), 13-OH (m)
9 70.82 3.42 C-10, C-11, C-27, C-28,
C-29
H-28 (s)
10 70.06 3.99 C-9, C-27, C-30 H-29 (s), H-30 (s),
H-26 (m)
11 58.70 3.52 C-9, C-10, C-12, C-13 H-14b (s), H-26 (m)
H-14a (w)
12 66.88 - -
13 76.26 - -
13-OH 4.51 C-4, C-13, C-14 H-14a (s), H-15a (m), H-25 (m)
14a 28.23 1.54 C-4, C-13,C-16
14b 1.463 H-26 (s)
15a 20.31 1.78 H-25 (s), H-14 (w)
15b 1.461 H-14 (s)
16 49.27 2.67 C-17,C-25 H-26 (s)
17a 26.72 2.23 C-2, C-15, C-16, C-18 H-25 (s)
17b 2.52 C-2, C-3, C-18
18 114.23 - -
19 122.31 - -
20 116.78 7.112 C-18, C-22, C-24
21 119.98 6.76 C-19, C-23, C-35
22 132.04 - -
23 112.15 7.107 C-19, C-21, C-35
24 139.89 - -
25 15.79 1.152 C-2, C-3, C-4, C-16
13
26 17.78 1.025 C-3, C-4, C-5, C-13
27 74.27 - -
28 27.98 1.155 C-9, C-27, C-29
29 16.49 1.182 C-9, C-27, C-28
30 95.23 4.86 C-32, C-10, C-27 H-29 (s)
31 76.63 3.06 C-30, C-33, C-34
31-OH 4.69 C-30, C-31, C-32
32 70.63 - -
32-OH 3.96 C31, C-32, C-33, C-34
33 26.45 1.033 C-31, C-32, C-34
34 25.16 1.062 C-31, C-32, C-33
35a 37.66 2.34 C-36, C-37, C-23, C-21,
C-22
35b 2.93 C-23, C-22, C-21, C-36
36 79.58 3.27 C-22, C-37, C-38, C-39
36-OH 4.23 C-35, C-36, C-37
37 71.45 - -
37-OH 4.17 C-36, C-37, C-38, C-39
38 24.11 1.065 C-36, C-37, C-39
39 26.45 1.097 C-36, C-37, C-38
9.3. Elucidation of relative configuration of 6.
NH
OH
OH
OH
O
O
O
OH
OH
O
H
H
H
HH
6, TC-705A
23
4
5
6
79
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24 25
26
35
36
37
38
39
27
28 29
30
31
32
33
34
The ROESY crosspeaks 26-H/16-H in combination with the ROESY crosspeaks between
25-H/13-OH show that H-16 and C-26 are cis oriented, and that 13-OH and C-25 are cis
oriented as well. The absence of ROESY signals between 25-H and 26-H, between 25-H
and 16-H, or between 26-H and 13-OH confirms this assignment, along with several
ROESY crosspeaks of the protons in position 5, 14, and 15. This defines the relative
configuration at positions 3, 4, 13, and 16 as shown. The ROESY crosspeaks 26-H/6-Ha,
26-H/11-H, and 26-H/10-H establish spatial proximity between these protons, whereas
the ROESY crosspeak 7-H/9-H establishes cis-orientation of these two protons. The
absence of significant ROESY crosspeaks and the large coupling constants observed for
6-Ha/7-H and 9-H/10-H indicate 1,2-diaxial configuration for these two pairs of protons.
Therefore, protons 7-H and 9-H must be oriented opposite to protons 10-H and 6-Ha as
well as the methyl group in position 26. The configuration in positions 11 and 12 follows
from the ROESY crosspeaks 26-H/11-H and 6-Ha/11-H. Finally, the configuration at
position 30 and the assignments for the two methyl groups attached to carbon 27 follow
from the ROESY crosspeaks 10-H/30-H and 29-H/30-H as well as 9-H/28-H.
14
9.4. Plots of NMR spectra of 6 acquired using DMSO-d6 as the solvent.
Figure SI-5. 1H-NMR spectrum of 6 in DMSO-d6.
15
Figure SI-6. Partial plot of the dqfCOSY spectrum of 6,
16
Figure SI-7. ROESY spectrum (mixing time 300 ms) of 6 in DMSO-d6.
17
Figure SI-8. Phase sensitive couples HMQC spectrum of 6.
18
Figure SI-9. HMBC spectrum of 6.
9.5. Additional spectroscopic data of compound 6.
Positive-ion electrospray MS m/z 656.4 (M+H)+, 678.4 (M+Na)+;
UV (CH3OH) λmax 232 nm, 281 nm with shoulder at 296 nm;
[α]23D -59.4 (c 0.12, CH3OH).
19
10. Spectroscopic data for compound 7.
7, TC-705B
NH
OH
OH
OH
O
OH
OH
O
H
H
H
H
23
4
5
6
79
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24 25
26
35
36
37
38
39
27
28 29
NMR-spectroscopic data (CD3OD, 600 MHz). Chemical shifts were referenced to
δ(CD2HOD) = 3.31 ppm and δ(CD2HOD) = 49.05 ppm. 13C-1H coupling constants were
obtained from HMQC spectra, and 1H-1H coupling constants from dqf-COSY spectra.
NOE crosspeaks were qualitatively classified as strong (s), medium strong (m) or weak
(w). Coupling constants are given in Hertz [Hz].
Position
δ 13C [ppm]
13C-1H
coupling
constants
δ 1H [ppm]
1H-1H
coupling
constants
Relevant HMBC
Correlations
Relevant NOESY
Correlations
1 - - -
2 153.0 - -
3 51.9 - -
4 43.5 -
5a
26.9 JC-H = 125.0 1.66
J5a, 5b = 12,
J5a,6a = 5.6,
J5a,6b < 1
C-3, C-4, C-6, C-7,
C-13
H-26 (s)
5b 2.62 J5b,6a = 14,
J5a,6b = 5.3
C-6, C-4, C-26, C-7 H-25 (s)
6a 29.1 JC-H = 131.7 1.83 J6a,6b = 13, J6a,7
= 9.6
C-5, C-7 H-26 (s)
6b 2.27 J6b,7 = 7.9 C-5, C-12
7 72.6 JC-H = 150.0 4.19 J7,11 = 1 C-6, C-11, C-12 H-9 (m), H-28 (w)
9 77.3 JC-H = 148.4 3.35 J9,10 = 9.3 C-10, C-11, C-27,
C-28, C-29
10 68.4 JC-H = 141.7 3.93 J10,11 = 1 C-7, C-9, C-27 H-29 (s)
11 64.7 JC-H = 183.4 3.49 C-9, C-10, C-12,
C-13
H-14b (s),
H-14a (w)
12 70.5 - -
13 78.3 - -
14a 30.3 JC-H = 131.7 1.68 J14a,14b = 13.5,
J14a,15a = 2.5,
J14a,15b = 2.5
C-14, C-16
14b 1.53 J14b,15a = 11.3,
J14b,15b = 3.7
C-13, C-14,
C-15
H-26 (s)
20
15a 21.6 JC-H = 128.4 1.95 J15a,15b = 13.5,
J15a,16 = 13.5
C-14, C-16 H-25 (s), H-14 (w)
15b 1.58 J15b,16 = 1.5 H-14 (s)
16 51.3 JC-H = 120.0 2.82 J16,17a = 10.8,
J16,17b = 6.5
C-3,C-25 H-26 (s)
17a 27.9 JC-H = 131.8 2.34 J17a,17b = 14.5 C-15, C-16,
C-18, C-2
H-25 (s)
17b 2.62 C-2, C-18, C-3
18 116.7 - -
19 124.5 - -
20 118.2 JC-H = 156.7 7.20 J20,21 = 8.3 C-18, C-22, C-24
21 121.2 JC-H =156.5 6.87 J21,23 = 1.5 C-19, C-23, C-35
22 132.7 - -
23 113.1 JC-H = 158.4 7.21 C-19, C-21, C-35
24 142.0 - -
25 16.2 JC-H = 126.7 1.27 C-2, C-3, C-4
26 18.7 JC-H = 125.0 1.12 C-3, C-4, C-5, C-13
27 73.3 - -
28 24.9 JC-H = 126.6 1.21
2
C-9, C-27, C-29
29 27.2 JC-H = 126.5 1.21
9
C-9, C-27, C-28
35a 38.96 JC-H = 123.4 2.54 J35a,36 =10.3,
J35a,35b =14.5
C-36, C-37, C-23,
C-21, C-22
35b 3.06 J35b,36 = 1.5 C-23, C-22, C-21,
C-36
36 81.4 JC-H = 141.7 3.55 C-22, C-37,
C-38, C-39
37 73.6 - -
38 24.5 JC-H = 126.8 1.23 C-36, C-37, C-39
39 25.9 JC-H = 125.0 1.26 C-36, C-37, C-38
Positive-ion electrospray MS m/z 556.4 (M+H)+, 578.4 (M+Na)+.
1 Sannai, A.; Fujimori, T.; Kato, K. Agric. Biol. Chem. 1982, 46, 429-434.