Characterization of polyphenols from plant materials through their silylation and 29Si NMR spectroscopy - Line assignment through 29Si, 13C spin-spin couplings

Article abstract of DOI:10.1002/mrc.1638

The lines in ²⁹Si NMR spectra of silylated polyphenols and some other compounds are difficult to assign owing to the absence of couplings with protons outside the silyl group. The assignment can be derived through small ⁿJ(²⁹

Full text of DOI:10.1002/mrc.1638

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2005; 43: 829–834
Published online 22 July 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1638
Characterization of polyphenols from plant
materials through their silylation and 29Si NMR
spectroscopy – line assignment through 29Si, 13C
spin–spin couplings
1
1
1
1
Jan Schraml,1∗ Vratislav Blechta, Jan Sy´kora, Ludmila Soukupova, Petra Curınova,
´
´
ˇ
´
2
2
´
ˇ
David Pronek and Jaromır Lachman
1
Institute of Chemical Process Fundamentals, Academy of Sciences of the Czech Republic, Rozvojova´ 135, 16502, Prague 6, Czech Republic
Department of Chemistry, Faculty of Agrobiology, Food and Natural Resources, Czech University of Agriculture, Kamy´cka´ 129,
2
16521 Prague 6, Czech Republic
Received 8 February 2005; Revised 4 June 2005; Accepted 8 June 2005
The lines in 29Si NMR spectra of silylated polyphenols and some other compounds are difficult to assign
owing to the absence of couplings with protons outside the silyl group. The assignment can be derived
n
through small J.29Si, 13C) couplings (n > 1). Using a previously described method for measurements of
these couplings, the assignment procedure is demonstrated here on three examples of trimethylsilylated
phenols: 7-hydroxyflavone, ferulic acid, and quercetin. In some cases the procedure can be used to identify
carbon atoms to which the siloxy groups are attached. Copyright  2005 John Wiley & Sons, Ltd.
KEYWORDS: NMR 29Si NMR; 13C NMR; nJꢀ29Si, 13C); line assignment; long-range couplings; polyphenols; flavonoids
This is of advantage in analyses of complicated mixtures
as it increases the reliability of components identification.
However, assignment of these lines to different silicon nuclei
in the molecule is a difficult task.
INTRODUCTION
Polyphenols are important constituents of plant materials.1
They are important not only for plant biochemistry but also as
nutrition components of herbivores and omnivores. Consid-
erable research effort has been made in understanding their
antibacterial action, their role as antioxidants, and in con-
nection with occurrence of cancer or cardiovascular diseases
in human beings.2–4 Obviously, separation of polyphenols
from plant materials, analysis of their mixtures, and identi-
fication of individual polyphenols are prerequisites for such
research.
Several years ago, we have shown that trimethylsilylation
followed by measurements of 29Si NMR spectra (‘29Si
NMR tagging’) could be a useful tool for analysis of
lignins, which are mixtures of compounds similar to the
polyphenols encountered in nutritious plants (for a review
see Ref. 5). Since then, advancement of NMR technology
(especially higher magnetic fields, more sensitive probes, and
development of LC-NMR) has lowered 29Si NMR detection
limits,6 making application of 29Si NMR tagging worthwhile
to apply to polyphenols produced by plants. In a preparatory
study, we have trimethylsilylated some model compounds
containing several phenolic hydroxyl groups, yielding 29Si
NMR spectra consisting of several lines for each compound.
The difficulty stems from the fact that the ‘phenolic’
trimethylsiloxy groups are attached to aromatic carbons
and there are no silicon spin–spin couplings to protons
outside the trimethylsilyl group (nJꢀ29Si,1H), n > 2). A
similar situation is encountered in trimethylsilyl esters of
polycarboxylic acids. The only possibility left for exact
experimental assignment in such cases is based on silicon
coupling to carbon-13 nuclei (nJꢀ29Si,13C), n > 1), provided
the 13C NMR spectra are assigned at least partially. Two
methods were developed specifically for this purpose: one
termed heteronuclear INADEQUATE7 and the other utilizing
selective 29Si decoupling.8 Recently, we have reported two
sensitive methods for measurements of small nJꢀ29Si,13C)
couplings9, which can be employed also for this purpose.
Both methods utilize INEPT-type polarization transfer for
signal enhancement and pulsed magnetic field gradients
for coherence pathway selection; the first one, (Si,C)gCOSY,
detects 13C signals and the other, (Si,C,Si)gHMQC, detects
29Si signals.
Feasibility of this approach to 29Si line assignment is
demonstrated here on three model compounds of increasing
complexity. The models were derived by trimethylsilylation
(R D SiꢀCH3ꢁ3ꢁ of the parent compounds (R D H): 7-
hydroxyflavone (1-H), ferulic acid (2-H), quercetin (3-H)
(Scheme 1). Of the two mentioned methods of measurements,
the latter has certain advantages (in general, narrow 29Si lines
ŁCorrespondence to: Jan Schraml, Institute of Chemical Process
Fundamentals, Academy of Sciences, Czech Republic, Rozvojova
135, Prague 6, 16502, Czech Republic. E-mail: schraml@icpf.cas.cz
Contract/grant sponsor: Grant Agency of the Czech Republic;
Contract/grant number: 525/02/0301; 203/03/1566; 203/03/D176.
Copyright  2005 John Wiley & Sons, Ltd.

830
J. Schraml et al.
Scheme 1. Carbon numbering in the studied compounds.
provide better resolution along F2 axis than 13C lines do and,
specifically, our probe has roughly 10 times higher sensitivity
in 29Si channel than in 13C channel) and so it was employed
here.
1% (v/v) of hexamethyldisilane (Aldrich, HMDSS, 29Si
secondary reference). The 13C NMR spectra were measured
using a spectral width of 30 000 Hz. WALTZ decoupling
was applied both during acquisition (1 s) and relaxation
delay (10 s). Zero filling to 128 K and a mild line-broadening
were used in data processing. The spectra were referenced
to the line of the solvent (CDCl3, υ D 76.99 ppm). The 29Si
NMR spectra were measured by an INEPT pulse sequence
as modified for Si(CH3ꢁ3 groups10 (75 ms polarization
and 16 ms refocusing delays optimized for polarization
transfer by 2Jꢀ29Si–C–1Hꢁ D 6.7 Hz from a nine-spin proton
system) using relaxation delay 10 s, spectral width 8000 Hz,
acquisition time 2 s, FID (free induction decay) data zero
filled to 128K. The spectra were referenced to the line of
HMDSS (υ D ꢀ19.79 ppm).
EXPERIMENTAL
Synthesis
The parent compounds were all of commercial
origin: 7-hydroxyflavone (7-hydroxy-2-phenyl-4H-chromen-
4-one, Aldrich 98%), ferulic acid ((E)-3-(4-hydroxy-3-
methoxyphenyl)acrylic acid, Fluka), and quercetin (2-(3,4-
dihydroxyphenyl)-3,5,7-trihydroxy-4H-chromen-4-one,
Lachema, p.a.). They were all trimethylsilylated by the
same procedure yielding 7-trimethylsiloxy-2-phenyl-4H-
chromen-4-one (1-TMS), trimethylsilyl ester of (E)-3(4-
trimethylsiloxy-3-methoxyphenyl)acrylic acid (2-TMS), and
2-(3,4-bis(trimethylsilyl)phenyl)-3,5,7-tris(trimethylsiloxy)-
4H-chromen-4-one (3-TMS).
The (Si,C,Si)gHMQC correlation experiment employed
the pulse sequence described earlier9with the INEPT part
as described above for the INEPT experiment; each sample
was measured with two values of polarization transfer delay
(Si > C) in the gHMQC part of the sequence, 160 and 500 ms.
The phenols were trimethylsilylated by bis(trimethyl-
silyl)acetamide (BSA) in 100% stoichiometric excess in a small
volume of acetonitrile as a solvent. The reaction mixture was
Intensity of the first gradient pulse was 9.93 gauss cmꢀ1
,
intensity of the second one alternated between 2.74 and -22.5
stirred at 65 C for 2 h. The solvent and the excess of silylation
gauss cmꢀ1 for the two FIDs needed for phase-sensitive 2D
°
1
detection. The 90 pulses of H, 13C, and 29Si were 18, 20,
°
°
reagent were distilled off in vacuum 1 kPa at 65 C.
and 8 µs long, respectively. For the selected pulse (i.e. the
11
°
first 90 carbon pulse) qSNEEZE shape 540 or 404 µs was
NMR spectra
13C and 29Si NMR spectral measurements in solution were
performed on a Varian UNITY-500 spectrometer (operating
at 499.9 MHz for 1H, at 125.7 MHz for 13C, and at 99.3 MHz
for 29Si NMR measurements) using a 5-mm SifC,Hg Nalorac
probe. The spectrometer is equipped with a X,Y,Z-Performa
gradient module, has four r.f. channels, two of them
fitted with waveform generators. The standard vnmr 6.1C
software was used for all the experiments except for
(Si,C,Si)gHMQC pulse sequence. All the spectra of the
used to cover 9000 or 11 000 Hz excitation bandwidth. Other
parameters were: acquisition time 2 s, spectral width in F2
(29Si) dimension 2000 Hz, with zero filling to 16 K data points
and Gaussian broadening (5 s), relaxation delay 8 s, gradient
duration 1 ms, spectrometer recovery time 1 ms. Spectral
width in F1 dimension was 8000 Hz or 10 000 Hz, 2 ð 64t1
increments, 24 scans per increment (total experiment time
8 h), zero filling to 256 and Gaussian broadening (0.005 s).
The resolution (especially along F2 axis) was chosen for the
assignment purposes and for more precise determination of
the coupling constants, and the resolution could be increased
substantially. The reported coupling constants were read off
°
pertrimethylsilylated compounds were recorded at 25 C
in 0.3–0.7 M solutions in dry CDCl3 (Aldrich, 99.8 atom %
D, stored over activated molecular sieve 3A) containing
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 829–834

Assignment through small 29Si, 13C spin–spin couplings 831
suggested simplistic approach – the nJꢀ29Si,13C) couplings
for n D 2 and n D 3 have comparable absolute values.16 The
recently proposed (Si,C,Si)gHMQC method holds promise
to overcome this obstacle by providing a means to measure
even smaller couplings expected for n > 3.
from the separations of lines in antiphase doublets; when the
separation within the doublet becomes comparable to the
linewidth, the apparent couplings are larger than the true
couplings. Simulations indicate that for Lorentzian lines, the
smallest separation observed here (0.4 Hz) corresponds to
true coupling 0.08 Hz smaller. No correction for the line
overlap was used.
The published assignments of the lines in the 13C NMR
spectra of the parent compounds (1-H and 3-H) were
derived from dimethylsulfoxide solutions12–15. Owing to
small differences between some carbon chemical shifts,
change of solvent, and substituent effects of TMS group, the
13C assignments in the trimethylsilylated products had to
be verified by a combination of usual procedures including
HETCOR, gHMBC, and INADEQUATE (in the case of 1-
TMS and 3-TMS). The assigned 13C chemical shifts are
summarized in Table 1, the detected nJꢀ29Si,13C) couplings in
Table 2.
The simplest compound, 1-TMS, containing only one
line in the 29Si NMR spectrum did not represent the 29Si
assignment problem but served as a test of this approach.
As it is apparent from Fig. 1a, the coupling constants can be
read from these experiments with a relatively high accuracy
and precision. However, even the high precision (estimated
at š0.1 Hz) of the measured couplings does not allow the
differentiation of two- and three-bond couplings based on
their values. In the case of an unknown compound, such a
result would not permit identification of the carbon atom
to which the trimethylsiloxy group is attached. However,
when it is a matter of 29Si NMR line assignment on a
known carbon skeleton, as is the case of 1-TMS, almost
the same couplings with three carbon nuclei in a row (C-
6, 7, 8) are possible only if the coupling to the middle
carbon (C-7) is through two bonds while the coupling with
its neighbors is through three-bond. At the same time, this
spectrum solves the assignment problem encountered in
the 13C NMR spectrum of the parent 1-H compound. For
this compound, contradictory assignments of the lines at
υ D 161.91 and 162.86 to C-2 and C-7 have been published.14,15
In the trimethylsilyl derivative, 1-TMS, the difference in
chemical shifts between these two lines is larger (υ D 160.15
RESULTS AND DISCUSSION
The three studied compounds (1-TMS, 2-TMS, 3-TMS) were
chosen to test the proposed approach on compounds with
increasing complexity and, at the same time, to investi-
gate derivatives of compounds found in natural material.
Previous determination of nJꢀ29Si,13C) couplings in the stud-
ied fragments (Si–O–C6H4 or Si–O–C(O)–C) indicated a
possible obstacle to the straightforward application of the
Table 1. 29Si chemical shifts and assigned 13C chemical shifts of compounds 1–3
Compound
2-TMS
Carbon
1-Ha
1-TMS
3-Hb
3-TMS
C-1
C-2
C-3
C-4
C-5
C-6
C-7
C-8
–
–
128.23
122.02
150.82
146.76
120.80
110.59
–
–
–
–
147.75
137.36
173.26
156.29
109.67
159.21
101.44
157.60
111.52
125.42
120.88
146.27
148.01
120.42
122.32
161.91c
106.62
176.37
126.69
115.13
162.86c
102.54
157.51
116.10
131.30
126.17
129.08
131.54
129.08
126.17
–
162.89
107.21
177.62
126.88
118.69
160.15
107.50
157.50
118.36
131.64
125.97
128.78
131.22
128.78
125.97
0.07
146.9
135.9
176.0
160.9
98.3
164.0
93.5
C-9
–
–
156.3
103.2
122.2
115.3
145.2
147.8
115.8
120.1
–
C-10
C- 10
C- 20
C- 30
C- 40
C- 50
C- 60
CH3 –Si
166.91
117.46
145.05
–
–
–
0.12, ꢀ0.38d
1.95, 0.36,
0.31, 0.21, 0.19
21.85, 21.62,
21.31, 21.02, 19.45
29Si
–
22.29
23.48, 21.50
–
a Data from DMSO-d6 solution, Ref. 14.
b Data from DMSO-d6 solution, Ref. 12, except for small numerical differences the data
agree with those of Ref. 13.
c Assignment of the two lines is interchanged in Refs 13 and 14.
d CH3 –O at 55.12 ppm.
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 829–834

832
J. Schraml et al.
Table 2. Long-range nJꢀ29Si,13C) coupling constants (absolute values) in Hza
Compound
1-TMS
2-TMS
3-TMS
Silicon υ
Carbon
C-1
C-2
C-3
C-4
C-5
C-6
C-7
22.29
23.48
21.50
21.85
21.62
21.31
21.02
19.45
0.5
0.5
1.3
2.4
1.6
0.5
0.5
2.9
2.6
0.5
0.2
2.1
1.3
0.6
0.5
0.5
1.2
1.8
2.0
1.8
0.4
0.4
1.8
2.0
2.0
C-8
C-9
C-10
C-10
2.4
2.2
0.4
0.5
0.5
1.3
2.3
1.5
0.5
C-20
C-30
1.8
2.2
1.3
0.5
40
C-40
C-50
C-60
0.5
30
Si–O–C-b
7
10
4
7
5
3
a No long-range 29Si, 13C couplings other than shown here were detected, estimated error š 0.1 Hz. The
values are line separations within the antiphase doublets, no correction for line overlap was used. The
values J < 0.6 Hz reported here might overestimate the true couplings by as much as 0.1 Hz.
b Silicon line assignment indicated as the carbon number to which the trimethylsiloxy group is attached,
see Scheme 1.
Figure 1. (a) A contour plot of the (Si,C,Si)gHMQC 2D spectrum of 1-TMS measured with polarization transfer 160 ms (on the
left – the relevant part of 13C NMR spectrum; on the right – traces through the cross-peaks corresponding to the 13C chemical
shifts of C-8, C-6, and C-7. The separations of the lines in the antiphase doublets are absolute values of the coupling constants in
Hz; down from the top these 2Jꢀ29Si,13C) and 3Jꢀ29Si,13C) couplings are 1.8, 1.8, and 2.0 Hz). (b) A contour plot of (Si,C,Si)gHMQC
2D spectrum of 1-TMS measured with polarization transfer of 500 ms (on the right – traces through the cross-peaks corresponding
to the 13C chemical shifts of C-10 and C-9. The 5Jꢀ29Si,13C) and 4Jꢀ29Si,13C) couplings are 0.4 Hz each.
and 162.89) and, most significantly, the line at 162.89 ppm
has no cross peak in the (Si,C,Si)gHMQC shown in Fig. 1,
and so it cannot arise from C-7 but rather belongs to C-
2 (in agreement with our assignment by other means). In
an attempt to see smaller couplings, the experiment was
repeated with a longer polarization delay of 500 ms (Fig. 1b)
which makes couplings of 0.4 Hz to C-9 and C-10 visible
5
(4Jꢀ29Si,13C) and Jꢀ29Si,13C), respectively). With the current
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 829–834

Assignment through small 29Si, 13C spin–spin couplings 833
lack of understanding of these couplings, it is not clear
why the 4Jꢀ29Si,13C) coupling to C-5 (υ D 126.88) is not
detected.
couplings. They occur at the chemical shifts of C-6, C-7,
and C-8 leaving no alternate assignment possible. This is
entirely reminiscent of the above-described example 1-TMS;
also no smaller couplings are revealed for this silicon in the
experiments with the short polarization delay. Analogously,
the 29Si line at υ D 21.62 is assigned to the silicon atom on
C-5 as this silicon couples with C-5, C-6, and C-10 either
The second model, 2-TMS, is closer to real examples;
it illustrates well what has been seen in the spectra of
the first model. The 29Si line at υ D 23.49 exhibits in the
(Si,C,Si)gHMQC spectrum intense cross peaks with C-20 and
C-10 and weak ones with C-30 (Fig. 2). Without going to
details of the coupling-constant values (Table 2), this silicon
signal obviously originates from the trimethylsilyl group
on C-1’. Similarly, as in the previous case, the couplings
over two and three bonds are of similar magnitude while
the coupling across four bonds is much smaller (about one-
third). The horizontal separation of the cross peaks follows
the trend one can expect: the largest separation is found
in the cross peaks of C-4 (2Jꢀ29Si,13C)) followed by C-5 and
C-3 (3Jꢀ29Si,13C)) and finally the smallest, of C-6, C-2 and
C-1 (4Jꢀ29Si,13C) and 5Jꢀ29Si,13C)). Among weak couplings,
the most intense cross peaks are due to 5Jꢀ29Si,13C) coupling
of C-1. It is also noteworthy that in this particular case all
these long-range couplings are revealed in one experiment
employing a polarization transfer delay of 160 ms.
through Jꢀ29Si,13C) or Jꢀ29Si,13C) couplings (the couplings
to C-10 and C-6 are somewhat smaller). Cross peaks due
to 4Jꢀ29Si,13C) or 5Jꢀ29Si,13C) are observed for C-7 and C-9.
The 29Si lines at υ D 21.31 and 21.02 can be assigned to
the trimethylsiloxy groups on C-40 and C-30, respectively,
in the same manner. Also, in these two cases the 2Jꢀ29Si,13C)
couplings are somewhat larger than the 3Jꢀ29Si,13C) couplings
experienced by the ortho positioned carbons. Finally, the
remaining 29Si line at υ D 19.950, which must be due to
siloxy group attached to C-3, manifests its position in line
with other above-described observations by its 2Jꢀ29Si,13C)
2
3
3
and Jꢀ29Si,13C) couplings to C-2 and C-3. Surprisingly, the
3Jꢀ29Si,13C) coupling to C-4 has anomalous magnitude; it is as
small as 4Jꢀ29Si,13C) or 5Jꢀ29Si,13C), comparable to the coupling
with C-1’.
Thethirdexample, 3-TMS, representstherealassignment
problem. Five 29Si NMR lines can be assigned on the basis of
the experience gained on the previous two models as follows.
The 29Si line at υ D 21.85 is due to the silicon of the siloxy
group on C-7 as there are three pairs of cross peaks in the
(Si,C,Si)gHMQC contour plot (Fig. 3), with the separations
within the pair corresponding to 2Jꢀ29Si,13C) or 3Jꢀ29Si,13C)
Besides assisting 29Si line assignments, the couplings
can also help in solving some 13C assignment problems.
For example, 2D (13C–13C) INADEQUATE of 3-TMS would
easily proceed from the characteristic chemical shift of C-4
to that of C-10. Further on there are, however, two coupling
pathways from C-10 to C-7: (i) C-10 ! C-5 ! C-6 ! C-7
Figure 2. Relevant parts of the contour plot of the (Si,C,Si)gHMQC 2D spectrum of 2-TMS measured with 160 ms polarization delay.
(On the left – 13C NMR spectrum).
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 829–834

834
J. Schraml et al.
Figure 3. Relevant parts of the contour plot of the (Si,C,Si)gHMQC 2D spectrum of 3-TMS measured with 160 ms polarization delay.
(On the left is the 13C NMR spectrum).
and (ii) C-10 ! C-9 ! C-8 ! C-7 with only 5 ppm chemical-
shift differences between the corresponding carbons. The
two coupling pathways are indistinguishable according to
INADEQUATE experiment. The Jꢀ29Si,13C) and Jꢀ29Si,13C)
couplings unambiguously identify the C-5 carbon line and
thus identify the carbon chemical shifts on each of the two
coupling pathways.
REFERENCES
1. Slanina J, Ta´borska´ E. Chem. Listy 2004; 98: 239.
2. Lachman J, Hamouz K, Orsa´k M, Pivec V. Rostl. Vy´r. 2000; 46:
231.
3. Lachman J, Orsa´k M, Pivec V. Zahradnictvi – Hort. Sci. (Prague)
2000; 27: 65.
2
3
4. Holasova´ M, Fiedlerova´ V, Smrcˇinova´ H, Orsa´k M, Lachman J,
Vavreinova´ S. Food Res. Intern. 2002; 35: 207.
5. Schraml J. Progr. Nucl. Magn. Reson. Spectrosc. 1990; 22: 289.
6. Knight CTG, Kinrade SD. Anal. Chem. 1999; 71: 265.
7. Past J, Puskar J, Alla M, Lippmaa E, Schraml J. Magn. Reson.
Chem. 1985; 23: 1076.
8. Past J, Puskar J, Schraml J, Lippmaa E. Collect. Czech. Chem.
Commun. 1985; 50: 2060.
9. Blechta V, Sy´kora J, Schraml J. Magn. Reson. Chem. 2004; 42: 968.
10. Schraml J. Collect. Czech. Chem. Commun. 1983; 48: 3402.
11. Kupcˇe E, Freeman R. J. Magn. Reson., A 1995; 112: 134.
12. Wawer I, Zielinska A. Magn. Reson. Chem. 2001; 39: 374.
13. Agrawal PK, Carbon-13 NMR of Flavonoids, Series: Studies in
Organic Chemistry, 59. Elsevier: Amsterdam, 1989.
14. Aksnes DW, Standnes A, Andersen OM. Magn. Reson. Chem.
1996; 34: 820.
CONCLUSION
The measurements of long-range nJꢀ29Si,13C) coupling
constants in compounds containing several trimethylsilyl
groups yield full and unambiguous assignments of all 29Si
lines provided the 13C NMR signals are assigned. The method
is especially valuable if siloxy groups are attached to quater-
nary carbons (especially aromatic or carboxylic) where other
methods of assignment fail. In some cases, the couplings can
be also used to identify the 13C NMR line of the carbon to
which the siloxy group is attached. When the trends in these
coupling constants are better known (including their depen-
dences on the number of intervening bonds, on substitution
both at silicon and carbon skeleton, etc.), the method could
be used even more efficiently.
15. Wenkert E, Gottlieb HE. Phytochemistry 1973; 16: 1811.
16. Schraml J, Past J, Puskar J, Pehk T, Lippmaa E, Brezˇny´ R. Collect.
Czech. Chem. Commun. 1987; 52: 1985.
Acknowledgments
The authors acknowledge the financial support provided by the
Grant Agency of the Czech Republic (Grant nos 525/02/0301;
203/03/1566; 203/03/D176).
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 829–834