Polyphenolic Profiling of Forestry Waste by UPLC-HDMSE

: Polyphenols constitute a diverse array of naturally occurring secondary metabolites found in plants which, when consumed, have been shown to promote human health. Greater consumption may therefore aid in the fight against diseases such as obesity, diabetes, heart disease, cancer, etc. Tree bark is polyphenol-rich and has potential to be used in food supplements. However, it is important to gain insight into the polyphenol profile of different barks to select the material with greatest concentration and diversity. Ultra-performance liquid chromatography (UPLC) was coupled with an ion mobility time-of-flight high-definition / high-resolution mass spectrometer (UPLC-HDMS E ) to profile ethanol extracts of three common tree barks ( Pinus contorta, Pinus sylvestris, Quercus robur ) alongside a commercial reference (Pycnogenol ® extracted from Pinus pinaster ). Through the use of Progenesis QI informatics software, 35 high scoring components with reported significance to health were tentatively identified across the three bark extracts following broadly the profile of Pycnogenol ® . Scots Pine had generally higher compound abundances than in the other two extracts. Oak bark extract showed the lowest abundances but exhibited higher amounts of naringenin and 3- O -methylrosmarinic acid. We conclude that forestry bark waste provides a rich source of extractable polyphenols suitable for use in food supplements and so can valorise this forestry waste stream.


Introduction
Polyphenols encompass a very broad range of compounds (e.g., flavonoids, phenolic acids, polyphenolic amides) which can be present in some foods in high concentrations [1]. Consumption of polyphenols can provide significant benefits to human health [2,3]. Most of these positive effects are attributed to their antioxidant and antimicrobial properties which may help prevent a range of diseases, such as cancer and bacterial infections [4,5]. There is also evidence that polyphenols can cross the blood-brain barrier and participate in the regulation of neuropeptides involved in mental wellbeing [6,7]. In relation to the current obesity crisis, polyphenols have also been shown to promote satiety and reduce food intake [8,9]. Specifically, dietary polyphenols have been shown to reduce the proliferation of adipocytes, suppress triglyceride accumulation, stimulate lipolysis, and reduce inflammation [10]. Polyphenols have also been shown to positively alter the gut microbiome [11]. The wide range of perceived benefits associated with polyphenols has led to calls from health agencies to both increase the consumption of polyphenol-rich foods, to breed crops with higher polyphenol contents and to potentially supplement food with polyphenols to promote human wellbeing [12]. A range of controlled trials have subsequently confirmed the benefits of these approaches [13].

Forestry Waste Samples
Representative bulk samples of bark waste were provided by a commercial forestry contractor (B.R. Warner Ltd., Amlwch, Anglesey, UK). The samples were from mature Lodgepole pine (LPP; Pinus contorta), Scots pine (SP; Pinus sylvestris), and Oak (O; Quercus robur). The commercial bark-derived dietary supplement, Pycnogenol ® extracted from Maritime pine (Pinus pinaster) was used as a reference material. Reviews on the use of Pycnogenol ® and its impacts on human health are presented elsewhere [37][38][39].

Sample Preparation
Changes in sample preparation methods will alter the polyphenol profile obtained in the subsequent extract. Ethanol was chosen here to provide the highest abundance of polyphenols [40] and also to comply with the EU directive for Good Manufacturing Practice (GMP) for food stuffs [41]. Briefly, each bark sample was ground to a fine powder and then 10 g placed in a glass beaker containing 100 mL of ethanol. After covering with Parafilm ® , the mixture was sonicated in an ultrasonic bath for 30 min. Ethanol was chosen as a solvent due to its legislative approval for use in the food industry [41].
The mixture was then left for 24 h at 4 • C before re-sonicating for 30 min. Once the solid material had settled, the liquid layer was placed in polypropylene tubes and centrifuged (10,000 rev min −1 , 30 min). The resultant supernatant was concentrated to 10 mL by gentle heating to 60 • C in a fume cupboard (i.e., 1 mL extract g −1 bark). The resultant extract was kept at −20 • C until required. The method described was repeated for each bark type in quadruplicate.

Analytical Instrumentation
An untargeted, discovery method was developed in negative ion mode using an I-class UPLC with a Synapt G2-Si in HDMS E mode (Waters UK Ltd., Wilmslow, UK). HD refers to ion mobility while MS E is a data-independent terminology for an acquisition that gathers mass spectrometer (MS) data, within a specified mass range, on all ions formed in the gas phase. These parent ions are subsequently fragmented to create product ions. The Synapt G2-Si is a quadrupole time-of-flight MS (Q-ToF) with incorporated ion mobility. A Z-Spray TM source was used in which chromatographically separated analytes arrive via one probe and the lock mass is infused via another. A metal baffle was set to switch periodically to allow either the analytes or the lock mass to enter the MS. Data were acquired and stored as continuum spectral data. Leucine enkephalin (Tyr-Gly-Gly-Phe-Leu) was the lock mass chosen for mass axis correction.

UPLC Conditions
The UPLC used was a Waters I-class instrument equipped with a Waters Cortecs UPLC C18+ 2.7 µm × 2.1 mm × 100 mm superficially porous column. This stationary phase has a positive charge present on the surface which provides better selectivity and peak shape for negatively charged analytes such as polyphenols. The use of 0.1% acetic acid as the mobile phase modifier was found to be the combination of modifier and concentration that gave the highest signal-to-noise ratio for polyphenolic compounds. Briefly, the mobile phase consisted of water modified with 0.1% acetic acid in A and MeOH modified with 0.1% acetic acid in B. The flow rate was 0.5 mL min −1 , the column temperature 40 • C and the injection volume 1.0 µL. The mobile phase composition was initially 90% A with 10% B, changing linearly to 1% A with 99% B over 4 min, and finally back to the initial conditions over 0.2 min.

Synapt G2-Si Conditions
Data were acquired and stored as continuum in a mass range of 50 to 1200 Da in negative ion resolution mode. The cone voltage was set to 40 V and the scan time was set to 0.2 s using an average of 3 scans and a mass window of ± 0.5 Da. The leucine enkephalin lock mass (554.2615 Da) was acquired every 30 s throughout the run but was not used for on-the-fly correction; this mass reference was acquired and stored for later use in data processing.

Data Processing
HDMS E data were processed using Progenesis QI software (NonLinear Dynamics Ltd., Newcastle upon Tyne, UK). Firstly, the acquired data were imported, then aligned to compensate for the small drifts in retention time between runs. A between-subjects experimental design was chosen creating 3 groups for the bark extracts, 1 for the reference (Pycnogenol ® ), and 1 group for blank extracts. Peaks were picked to locate the analytes in the samples and then ions were deconvoluted. These ions were then compared to the ChemSpider Polyphenols database [42] with a 5 ppm precursor tolerance. Theoretical fragmentation (in silico) was also performed with a fragment tolerance of 5 ppm.'Only isotope similarity scores above 90% were used and also a filter for elemental composition was set to take into account compounds with elemental composition of H, C, and O. As a result of collecting ion mobility data as well as MS E data (HDMS E mode), Progenesis QI was more able to distinguish co-eluting components due to differences in drift times of the parent ions.
Identifications made through comparison to external databases should always be treated as tentative. In Progenesis QI, data were filtered so that only the best quality data were selected for further investigation. Filters were setup to show only compounds that had ANOVA p values ≤ 0.01 and blanks were the lowest mean. All compound abundances lower than 100, with no fragmentation and showing blanks as the highest mean, were removed from the dataset. Of a maximum possible score of 60 for this experimental setup only scores above 40 were chosen for further evaluation. Metrics for retention time similarity and collision cross-section (CCS) similarity were zero.
After processing through Progenesis QI, data were then exported to EZInfo (Umetrics, Umeå, Sweden) which provided a multivariate analysis (MVA) approach to this discovery data. Multivariate data analysis was achieved by undertaking Principal Component Analysis (PCA) in EZInfo giving an overview of the sample data via a scores plot. This plot shows the observations that are likely to be most similar (close together) and also the ones that are most dissimilar (far away) allowing for the visualisation of atypical observations, trends and other patterns within the data. The heat map was produced in Matlab (MatWorks Inc., Natlick, MA, USA).

Bioactive Phenolic Compounds in Tree Bark Extracts
A summary of the results obtained from the analysis of the three bark extracts is presented in the detailed Excel file in Supplementary Materials (XL-SM). Tentative identifications for compounds are ordered with the highest scoring results from top to bottom. In summary, a total of 35 components were tentatively identified across the three bark extracts with scores of 40 or above. Accurate mass values are given for parent and main product ions plus retention time (min.), drift time (ms) and normalised abundances for all sample injections. All components showed very low ANOVA-p and q values, indicating a false discovery rate (FDR) approaching zero in many cases with a maximum FDR of 0.5%. Isotope similarity scores above 90 and mass errors of ≤ 5 ppm can also be seen. Pycnogenol ® , which was used as a reference material, generally exhibited higher abundances of polyphenols than the bark extracts, which was probably due to the extraction and concentration resulting from its proprietary production process. Additional references are given for each identified polyphenol component which have been shown to have bioactive significance in terms of human health benefits. Three components which could not be distinguished from their isobaric species have their various possibilities listed, which is felt to be better than exclusion. Total ion chromatograms and ion intensity graphs to illustrate the alignment and vector editing processes, are available in supplementary materials (Figures S1-S8). Molecular structures of the 35 identified components can be found in Figure S9.
As an aid to visualisation, a heat map has been created using averaged abundances for each sample type across the 35 identified polyphenols ( Figure 1). From this overview of the identifications it can be seen that the 2 Pinus sp. more closely represent the general trend of Pycnogenol ® (Pinus pinaster) than does oak.
It is also noted that compound abundances are generally higher in the Scots Pine extract than in the other two bark extracts with the Oak sample showing the lowest abundances. Oak, on the other hand, does exhibit higher amounts of naringenin and 3-O-methylrosmarinic acid than the other samples. Several biological activities have been ascribed to naringenin, including antioxidant, antitumor, antiviral, antibacterial, anti-inflammatory, antiadipogenic, and cardioprotective effects. The most promising activity being related to cardiovascular disease protection, in pure form and in complex polyphenolic mixtures, and also naringenin's ability to improve endothelial function [43]. The 3-O-methylrosmarinic acid is thought to contribute to the properties of the Cistus genus which is a plant used in traditional folk medicine for wound healing and its anti-inflammatory properties [44]. It is also noted that compound abundances are generally higher in the Scots Pine extract than in the other two bark extracts with the Oak sample showing the lowest abundances. Oak, on the other hand, does exhibit higher amounts of naringenin and 3-O-methylrosmarinic acid than the other samples. Several biological activities have been ascribed to naringenin, including antioxidant, antitumor, antiviral, antibacterial, anti-inflammatory, antiadipogenic, and cardioprotective effects. The most promising activity being related to cardiovascular disease protection, in pure form and in complex polyphenolic mixtures, and also naringenin's ability to improve endothelial function [43]. The 3-O-methylrosmarinic acid is thought to contribute to the properties of the Cistus genus which is a plant used in traditional folk medicine for wound healing and its anti-inflammatory properties [44].
As an overview of the data, a scores plot, using principal component analysis (PCA) with Pareto scaling, was created which showed tight clustering of the bark extracts sample replicates and a general separation between bark types ( Figure 2). This figure includes an expanded area view of the more closely situated clusters. This shows tight clustering within each group of replicates and clear differences between the sample types. Furthermore, also available in the supplementary material is a loadings bi-plot ( Figure S11) which illustrates how the components relate to the samples i.e., the closer an ion is to a sample cluster the more this describes the sample's composition and therefore also highlights the differences between the bark extracts. As an overview of the data, a scores plot, using principal component analysis (PCA) with Pareto scaling, was created which showed tight clustering of the bark extracts sample replicates and a general separation between bark types ( Figure 2). This figure includes an expanded area view of the more closely situated clusters. This shows tight clustering within each group of replicates and clear differences between the sample types. Furthermore, also available in the supplementary material is a loadings bi-plot ( Figure S11) which illustrates how the components relate to the samples i.e., the closer an ion is to a sample cluster the more this describes the sample's composition and therefore also highlights the differences between the bark extracts.  To continue to point out the highlights of this discovery work (additional references for each compound are available in supplementary information) it can be seen in XL-SM and Figure 1 that quercetin, eriodictyol and (−)-epicatechin are compounds with significant abundance. Dietary supplementation with quercetin or plant extracts containing quercetin has been shown to attenuate high fat diet induced obesity and insulin resistance [45] and also decreases inflammation [46]. Quercetin is seen to be abundant across all 4 extracts. Eriodictyol, which is particularly prominent in Scots Pine, has been shown to stimulate insulin secretion in mice islets, improving glucose tolerance and increasing plasma insulin in non-diabetic and diabetic rats [47]. Furthermore, research shows that neuro-inflammatory response to experimental stroke is inhibited by eriodictyol [48] as is inflammation in osteoarthritis [49]. Epicatechin's benefits have been discussed prolifically in literature with a primary focus on anti-oxidant, anti-microbial, anti-inflammatory, and anti-cancer effects [50] plus, more specifically, cardiovascular and neuropsychological health [51]. To continue to point out the highlights of this discovery work (additional references for each compound are available in supplementary information) it can be seen in XL-SM and Figure 1 that quercetin, eriodictyol and (−)-epicatechin are compounds with significant abundance. Dietary supplementation with quercetin or plant extracts containing quercetin has been shown to attenuate high fat diet induced obesity and insulin resistance [45] and also decreases inflammation [46]. Quercetin is seen to be abundant across all 4 extracts. Eriodictyol, which is particularly prominent in Scots Pine, has been shown to stimulate insulin secretion in mice islets, improving glucose tolerance and increasing plasma insulin in non-diabetic and diabetic rats [47]. Furthermore, research shows that neuro-inflammatory response to experimental stroke is inhibited by eriodictyol [48] as is inflammation in osteoarthritis [49]. Epicatechin's benefits have been discussed prolifically in literature with a primary focus on anti-oxidant, anti-microbial, anti-inflammatory, and anti-cancer effects [50] plus, more specifically, cardiovascular and neuropsychological health [51].
Quercetin, which is seen in all samples analysed here, and quercetin-3 -glucuronide, which is abundant in lodgepole pine bark, have both been shown to be active against human breast cancer [52]. Furthermore, also abundant in lodgepole pine is norathyriol which is noted for its potential towards suppression of skin cancers induced by UV radiation [53], as a new candidate for the treatment of hypouricaemic [54] and also its regulatory effect on lipid metabolism making it useful for protection against hepatic lipid metabolic disorders and the treatment of non-alcoholic fatty liver disease [55].
The diverse array of higher abundance components is most evident in Scots Pine; some of these components have already been mentioned. In addition, procyanidin (B1, B2, B3 or B4), is observed in Scots Pine bark in very high abundance and is also evident in the other extracts to a lesser extent. These molecules are the pigments often associated with apples, grapes, and berries and their related health benefits. Procyanidins have been reported to target diverse molecular switches in carcinogen metabolism including inflammation, cell proliferation, cell cycle, apoptosis, and the development of new blood vessels (angiogenesis) and consequently studies on Procyanidins have shown that they inhibit the proliferation of various cancer cells in vitro and in vivo [56].

Tentative Identification of Polyphenols
It is often the case that databases do not contain fragmentation data so in silico theoretical predictions are relied upon here. This can be quite effective in the context of polyphenolic compounds as their mechanisms of fragmentation are well documented. They are generally in the category referred to as Retro-Diels-Alder (RDA) reactions [57]. Furthermore, the sugar moiety, which is present in many polyphenols, may also fragment by RDA. Often, a water loss precedes the RDA fragmentation by remote hydrogen rearrangement, forming an unsaturated sugar moiety, which facilitates the RDA process. All mass spectra are available in the supplementary material ( Figure S10 +e+M] − was used for confirmation. From literature, naringenin was expected to be present in oak [63] and was found in greatest abundance in the oak extract data. Naringenin was identified by the [M-H] − adduct only as the fragmentation was not conclusive and would therefore require more work to make this satisfactory.

Potential Use of Bark Waste for Nutraceutical Production
The wide spectrum of polyphenols found in these bark samples illustrates the potential to produce a nutritional supplement from a range of tree species. It also raises the possibility to blend products from different tree species to obtain a more balanced bioactive phenolic profile. This may involve the use of other tree species not investigated here which may have vastly different physical structures and chemical compositions (e.g., Betula spp., Salix spp., and Eucalyptus spp.). In addition, an expansion of this study is necessary to provide more detail on the importance of factors which may affect the polyphenol profile, such as tree stand age, harvesting season, time, and conditions of storage prior to processing. Future work should also focus on the most efficient and cost-effective extraction techniques. One such route, would be to investigate the use of supercritical fluid extraction i.e., liquid CO 2 (and possibly modifiers such as ethanol) as a food-friendly alternative to the processes detailed here. The fast diffusion rate of liquid CO 2 may result in more rapid extraction. Furthermore, it is shown that different enzymatic pre-treatments can facilitate control of selectivity as to which components are extracted [64]. Extraction solvents may also be recycled from batch to batch. Future work should also focus on the use of the solid waste remaining after the extraction process e.g., conversion into fuel pellets [65].

Conclusions
Here we demonstrate the effective use of UPLC-HDMS E for the detailed analysis of forestry waste and its application in the development of novel food nutritional supplements. In our extracts we identified 35 components with bioactive properties which have the potential to benefit human and animal health by providing a preventative measure against many life threatening conditions. Furthermore, the phenolic components can be easily extracted from low value forestry waste making the low carbon process suitable for commercialisation.
Supplementary Materials: The following are available online at http://www.mdpi.com/2227-9717/8/11/1411/s1, Figure S1: Total ion chromatogram of Lodge Pole Pine, Figure S2: Ion intensity map of Lodge Pole Pine, Figure S3: Total ion chromatogram of Oak, Figure S4: Ion intensity map of Oak, Figure S5: Total ion chromatogram of Scots Pine, Figure S6: Ion intensity map of Scots Pine, Figure S7: Total ion chromatogram of Pycnogenol ® , Figure S8: Ion intensity map of Pycnogenol ® , Figure S9 Molecular structures of identifications ordered from high to low score value, Figure S10 Mass spectra of identifications ordered from high to low score value, Figure S11 Loadings bi-plot of the bark extract data, XL SM is a spreadsheet of identifications, abundances and additional references.
Author Contributions: The project was conceptualized by C.M.P. who was also responsible for methodology and formal analysis. The samples were prepared and analysed by C.M.P. The first draft of the manuscript was prepared by C.M.P. and reviewed and edited by D.L.J. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.