Characterization of Four
Commercial Grape Seed Extracts Using UHPLC-UV/CAD/HRMS, GC-FID, and GC-MS
Reviewed: Sica VP, Mahoney C, Baker TR. Multi-detector
characterization of grape seed extract to enable in silico safety assessment. Front Chemistry. 2018;6:334. doi:
10.3389/fchem.2018.00334.
Keywords: GC-FID, GC-MS, grape
seed extract, procyanidin, UHPLC-UV/CAD/HRMS, Vitis
vinifera
As part of an effort to
assess the composition of grape (Vitis vinifera,
Vitaceae) seed extracts (GSEs), commercially available bulk materials from four
suppliers were characterized using a combination of ultra high-performance
liquid chromatography with ultraviolet (UHPLC-UV), charged aerosol detection
(CAD) and high-resolution mass spectrometry (HRMS), and gas chromatography (GC)
with either a flame ionization (FID) or a MS detector.
In addition to the
qualitative data, compounds present in GSE above a threshold of toxicological
concern (TTC, i.e., higher than 400 μg/g extract, which was established based
on a daily dosage of 210 mg GSE) were quantified using UHPLC-CAD and
“identified” using high-resolution tandem MS for an in vitro
safety assessment. Whenever possible, a complete chemical structure was
assigned to a compound, although in many cases, only the compound class was
determined based on similarity of the MS fragmentation pattern with those of
known compounds. For obvious reasons, the stereochemistry could not be
assigned. Since the CAD provides inconsistent results for volatile compounds,
GC-FID and GC-MS was used to identify and quantify those molecules. In addition
to GSE, authentic extracts of peanut (Arachis hypogaea,
Fabaceae) and Maritime pine (Pinus pinaster,
Pinaceae) were analyzed and compared with GSE to ensure absence of potential
adulterants.
Using the UHPLC-CAD
approach, 91% of the GSE could be accounted for. The CAD chromatogram showed 39
peaks, reportedly consisting of at least 83 components, assigned by the authors
to one of three groups: polar, nonpolar, and polyphenols. For the reference
GSE, polar compounds (salts, amino acids, organic acids, and sugars) made up
16% of the extract, while nonpolar compounds (fatty acids and sterols) accounted
for 1%. The remainder of the extract (83%) consisted of polyphenolic compounds.
A hump seen in the chromatogram, which was formed by all grape seed
proanthocyanidins with a degree of polymerization (DP) ≥ 6, was regarded as
just one peak based on toxicological considerations, but this “peak” made up
75% of all GSE polyphenols. The authors then used molecular weight filters to
determine the size distribution, and concluded that a large portion of the high
molecular weight fraction were due to polyphenols with a molecular weight above
100,000 Da.
The UHPLC-CAD fingerprint
allowed an easy distinction of the four commercial GSEs from peanut skin and Maritime
pine bark extracts. The polyphenol contents in the four commercial GSEs ranged
from 82-93%, with 72-87% being hexamers or larger polymers. There were some
differences with regards to the gallic acid content among the four commercial
GSE samples, from non-detected up to 3% of the extract. Generally, the
differences were minor, and can be explained by variations in the source
material and/or manufacturing process.
For the in vitro
safety assessment, the daily intake of the main component classes (flavonoids,
tannins, and lignans) was compared to the intake of similar components from
common food sources. As an example, a daily intake of ca. 160 mg of “tannins”
(defined by the authors as proanthocyanidins with a degree of polymerization of
6 and higher) was calculated for the GSE, which was then compared to “tannin” intake
from foods such as cocoa (Theobroma cacao,
Malvaceae), black tea and green tea (Camellia sinensis,
Theaceae), or grape products. None of the compound classes in GSE was present
at concentrations that represent a safety risk.
Comment: One of the great advantages of using a CAD system is the
ability to obtain quantitative data across a large set of compounds in a
mixture without the need of standard compounds for each of these molecules.
This is ideal for an herbal extract, which often contains thousands of
different molecules. As such, this detector merits a more widespread use in the
herbal dietary supplement industry. The UHPLC-CAD fingerprint is suitable to
distinguish GSE from peanut skin and Maritime pine bark extracts, although it
would have been interesting to see if the approach also works for mixtures of
GSE with, e.g., peanut skin extracts.
While certainly providing a
lot of information about the composition of an herbal extract, the ability of
HRMS to unequivocally identify natural products is limited. For example, the
stereochemistry and connectivity of the flavan-3-ol monomers cannot be
established using this approach.
As a side note, the
molecular weight distribution of grape seed proanthocyanidins is a matter of
debate. Spranger et al.1 obtained a maximum DP of 34.5 determined
after thiolysis, which is in a similar range as the maximum DP of 31.5 obtained
by Sun et al.2 Based on the DP, Spranger et al. calculated a maximum
molecular weight for grape seed proanthocyanidins of 15,952.8 Da, well below
the >100,000 Da reported by Sica et al. in this paper.
References
- Sun
B, Leandro C, Ricardo da Silva JM, Spranger I. Separation of grape and wine
proanthocyanidins according to their degree of polymerization. J Agric Food Chem. 1998;46(4):1390-1396.
- Spranger I, Sun B,
Mateus AM, de Freitas V, Ricardo da Silva JM. Chemical characterization and
antioxidant activities of oligomeric and polymeric procyanidin fractions from
grape seeds. Food Chem. 2008;108(2):519-532.