Bud-Derivatives, a Novel Source of Polyphenols and How Different Extraction Processes Affect Their Composition

The use of herbal food supplements, as a concentrate form of vegetable extracts, increased so much over the past years to count them among the relevant sources of dietetic polyphenols. Bud-derivatives are a category of botanicals perceived as a “new entry” in this sector since they are still poorly studied. Due to the lack of a manufacturing process specification, very different products can be found on the market in terms of their polyphenolic profile depending on the experimental conditions of manufacturing. In this research two different manufacturing processes, using two different protocols, and eight species (Carpinus betulus L., Cornus mas L., Ficus carica L., Fraxinus excelsior L., Larix decidua Mill., Pinus montana Mill., Quercus petraea (Matt.) Liebl., Tilia tomentosa Moench), commonly used to produce bud-derivatives, have been considered as a case study. An untargeted spectroscopic fingerprint of the extracts, coupled to chemometrics, provide to be a useful tool to identify these botanicals. The targeted phytochemical fingerprint by HPLC provided a screening of the main bud-derivatives polyphenolic classes highlighting a high variability depending on both method and protocol used. Nevertheless, ultrasonic extraction proved to be less sensitive to the different extraction protocols than conventional maceration regarding the extract polyphenolic profile.


Introduction
In recent decades, food supplements have an important impact on the consumers showing a significant expectation for their health and well-being [1]. They are concentrated sources of nutrients or bioactive compounds endowed with nutritional or physiological effects and, due to their presumed health benefits, they can supplement the common diet [2,3].
In particular, the interest in herbal food supplements (botanicals) is exponentially grown and consequently the relative market has increased in all the world [4]. Botanicals are become among the most popular into the food supplements category, due to the general belief which "natural" is better, healthier and safer than synthetic drugs, although this is not always true [4]. In Italy, more than 20% of the Italian population is considered "regular" consumer of these herbal products, as highlighted from the recent European PlantLibra (Plant Food Supplements: Levels of Intake, Benefit and Risk Assessment) consumer survey [5]. The wide range of herbal food supplements on the market and the

Raw Samples
Buds, belonging to eight different vegetable species (Carpinus betulus L., Cornus mas L., Ficus carica L., Fraxinus excelsior L., Larix decidua Mill., Pinus montana Mill., Quercus petraea (Matt.) Liebl., Tilia tomentosa Moench) were collected from plants spontaneously grown in the Turin Province (Italy) and were immediately authenticated by an agronomist. Sampling has been performed in two years (2018-2019), from February to April, during the bud break ("balsamic period"). Table 1 reports the geo-localization coordinates of the different collection sites and the scientific naturalistic illustrations (specifically achieved during the Finnover project) of all the eight vegetable species investigated.

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1). and were immediately authenticated by an agronomist. Sampling has been performed in two years (2018-2019), from February to April, during the bud break ("balsamic period"). Table 1 reports the geo-localization coordinates of the different collection sites and the scientific naturalistic illustrations (specifically achieved during the Finnover project) of all the eight vegetable species investigated. and were immediately authenticated by an agronomist. Sampling has been performed in two years (2018-2019), from February to April, during the bud break ("balsamic period"). Table 1 reports the geo-localization coordinates of the different collection sites and the scientific naturalistic illustrations (specifically achieved during the Finnover project) of all the eight vegetable species investigated.

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as possible. The manufacturing was performed both in an Italian company (Geal Pharma Turin, Italy)

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as possible. The manufacturing was performed both in an Italian company (Geal Pharma Turin, Italy)

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as possible. The manufacturing was performed both in an Italian company (Geal Pharma Turin, Italy) and by the Authors in the analytical laboratory of the University of Genoa (Department of Pharmacy).

Chemicals
MilliQ ultrapure water, obtained by means of a Millipore equipment (Bedford, MA, USA) was used throughout. All chemicals employed for the extract preparations and for the subsequent analysis were HPLC-grade. They were supplied by VWR International S.r.l (Milan, Italy) and Sigma-Aldrich (St. Louis, MO, USA). Purity of all the used standards for HPLC analysis of BDs has been reported in the Supplementary Materials (Table S1).

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as possible. The manufacturing was performed both in an Italian company (Geal Pharma Turin, Italy)

Bud-Derivatives Manufacturing Applying Two Different Methods
Fresh buds, after their collection, were immediately processed to prepare the corresponding BDs in order to minimize any degradation preserving the peculiar phytocomplex as much undamaged as The following two different preparation methods of BDs were investigated: the conventional cold Maceration (M) [9], and a more rapid and innovative procedure by Ultrasounds (US) recently described by the Authors [11]. Moreover, for both preparation methods two different extraction solvents and different sample/solvent ratios were investigated too ("Protocol A" and "Protocol B", see Figure 1), in order to evaluate both the proper identification of the botanical species and the traceability of the BDs independently from their manufacturing process (Table 2). Each extraction was performed in duplicate.

Conventional Cold Maceration (M) as Traditional Method
BDs were prepared using a cold maceration by an Italian Company of botanicals (Geal Pharma, Bricherasio, Turin) following two different experimental manufacturing protocols, reported in Table  2: (A) A 21 days maceration of buds in glycerol/ethanol 96% (1/1 w/w) with a 1:20 bud/solvent ratio (considering the dry weight) has been performed, according to the official method of glyceric macerates reported in the European Pharmacopoeia VIII edition [9] ("M_A"). (B) A 3 months maceration of buds in a mixture of water/glycerol/ethanol 96% (50/20/30 w/w/w) as extraction solvent with a bud/solvent ratio variable (considering the fresh weight) depending on the botanical species (see Table 2) has been used, according to the method optimized and used by the Company to produce glyceric macerates ("M_B").
In both methods, after the maceration step, the extracts, namely BDs, have been obtained by a preliminary filtration, a manual pressing and a second filtration after two days of decanting. The so obtained BDs were stored at 4 °C in the dark until their further analysis.

Conventional Cold Maceration (M) as Traditional Method
BDs were prepared using a cold maceration by an Italian Company of botanicals (Geal Pharma, Bricherasio, Turin) following two different experimental manufacturing protocols, reported in Table 2: (A) A 21 days maceration of buds in glycerol/ethanol 96% (1/1 w/w) with a 1:20 bud/solvent ratio (considering the dry weight) has been performed, according to the official method of glyceric macerates reported in the European Pharmacopoeia VIII edition [9] ("M_A"). (B) A 3 months maceration of buds in a mixture of water/glycerol/ethanol 96% (50/20/30 w/w/w) as extraction solvent with a bud/solvent ratio variable (considering the fresh weight) depending on the botanical species (see Table 2) has been used, according to the method optimized and used by the Company to produce glyceric macerates ("M_B").
In both methods, after the maceration step, the extracts, namely BDs, have been obtained by a preliminary filtration, a manual pressing and a second filtration after two days of decanting. The so obtained BDs were stored at 4 • C in the dark until their further analysis.

Green Extraction: Pulsed Ultrasound-Assisted Extraction (US) as Alternative Method
Fresh buds were finely ground by a Grindomix 200 M (Retsch, Haan, Germany) for 20 s at 5000 rpm, and then sieved by a 150 µm sieve, in order to improve the efficiency of the following extraction step [46]. PUAE was performed directly by an Hielscher UP200St sonicator (Teltow, Germany) equipped with an ultrasonic titanium sonotrode (7 mm of diameter), at a constant frequency of 26 kHz. The pulsed mode, referring to an alternation of "on" time and "off" time of the sonicator, guarantees a lowering increase in temperature, which better preserve the phytocomplex, and greater energy savings compared to continuous treatments [47]. The experimental sonication conditions (amplitude 30%, duty cycle 65%, extraction time 20 min) were previously optimized by the Authors on the same raw materials [11].
The same two experimental extraction conditions described in the paragraph 2.3.1 ("Protocol A" and "Protocol B", see Figure 1) were employed ("US_A" and "US_B", see Table 2). The extracts obtained were filtered for Buchner (Whatman n. 1 paper), centrifuged at 3000 rpm for 10 min and then stored at 4 • C in the dark until analysis.

Spectroscopic Analysis: UV-Visible Fingerprint
UV-Visible absorption spectra (200 nm-900 nm) were recorded by a spectrophotometer Agilent Cary 100 (Varian Co., Santa Clara, CO, USA) with 0.5 nm resolution, using rectangular quartz cuvettes with 1 cm path length. BDs, before the spectroscopic analysis, were suitably diluted in the corresponding extraction solvent (glycerol/ethanol 1/1 w/w or water/glycerol/ethanol 50/20/30 w/w/w) depending on the followed experimental protocol ("Protocol A" and "Protocol B", respectively). Dilution was necessary to avoid signal saturation but was subsequently considered in order to make a comparison between the different spectra achieved. BDs spectra were acquired in duplicate and then averaged. The collection was performed at room temperature (25 ± 1 • C), against a blank solution represented by the corresponding extraction solvent.
The external standard method was used for quantitative determination of bioactive compounds. Stock solutions of cinnamic acids and flavonols with a concentration of 1.0 mg·mL −1 were prepared in methanol: five calibration standards were prepared by dilution with methanol; stock solutions of benzoic acids and catechins with a concentration of 1.0 mg·mL −1 were prepared in 95% methanol and 5% water. In this case, five calibration standards were prepared by dilution with 50% methanol-water.
An Agilent 1200 High-Performance Liquid Chromatograph coupled to an Agilent UV-Vis diode array detector (Agilent Technologies, Santa Clara, CA, USA) was used for the chromatographic analysis. Four chromatographic methods were used to separate the bioactive molecules on a Kinetex C18 column (4.6 × 150 mm, 5 µm, Phenomenex, Torrance, CA, USA). Several mobile phases were used for bioactive compound identification and UV spectra were recorded at different wavelengths, based on HPLC methods, previously tested and validated [10,40] Biomarkers were selected for their demonstrated positive healthy properties and antioxidant capacity by literature in relation to the use of this plant-derived products. All single compounds were identified in samples by comparison and combination of their retention times and UV spectra with those of authentic standards in the same chromatographic conditions. Each sample was analyzed in triplicate and results were reported as mean value ± standard deviation to assess the repeatability of the employed methods. Multivariate data analysis has been performed by CAT (Chemometric Agile Tool) software, one advanced chemometric multivariate analysis tool based on R, developed by the Chemistry Group of the Italian Chemical Society [49].
PCA was applied as common multivariate statistical method of unsupervised pattern recognition. Its aim is extracting important information from the data and decreasing the high-dimensional dataset volume by maintaining the important information [50,51].

Data Matrices Organization
A data matrix A 32,601 consisting of 32 rows (corresponding to the BDs analyzed, 4 samples for each of the eight botanical species investigated) and 601 columns (the absorbance values in the range of 200-500 nm of the UV-Visible spectra, with 0.5 nm of resolution) was prepared and further analyzed by PCA. Standard normal variate (SNV) transform and column autoscaling were previously performed on the spectral data to remove multiplicative effects of scattering and to scale the data, respectively [52].
Available sample were divided in two different subsets: a calibration (or training) set and a test (or evaluation) set in order to build and validate the statistical model, respectively [53]. For a reliable validation strategy, it is important that data used as test set were not used to build the model in order to avoid the overestimations of the prediction ability [53]. 32 samples, previously reported in Table 2, were selected for the construction and identification of the model (Calibration set). The representative calibration data set consisted of 4 extracts (M_A, M_B, US_A, US_B) for each botanical species investigated (Carpinus betulus L., Cornus mas L., Ficus carica L., Fraxinus excelsior L., Larix decidua Mill., Pinus montana Mill., Quercus petraea (Matt.) Liebl., Tilia tomentosa Moench). Furthermore 16 BDs, obtained both by conventional maceration and ultrasound extraction respectively from the same eight vegetal species, were randomly selected and used as an independent set to test the model and assess its validity (Test set, Table 3). All the pre-treated UV-Visible absorption spectra, in the range 200-500 nm, are reported in Figure 2. For each species, the four averaged spectral profiles corresponding to the Calibration set (Table 2) (Table 2) are highlighted in grey while in red are reported the External Test set samples (Table 3).
Then, a data matrix B 32,620 consisting of 32 rows and 620 columns was prepared and analogously analyzed by PCA. B 32,620 rows correspond to the 32 BDs analyzed (Calibration set), and columns are the absorbance values of the UV-Visible spectra after SNV in the range 200-500 nm coupled to the chromatographic quantifications by HPLC (4 polyphenolic classes and 13 bioactive compounds). The data set was previously scaled by using a block scaling procedure [54], with the aim to give to the spectroscopic and chromatographic variables a comparable influence in the data analysis. In fact, this pretreatment allows to divide variables in different blocks whose values will be scaled to attain the same block-variance after pretreatment. Moreover, the variables belonging to the same block are equally weighted.

Results and Discussion
The quality control of vegetal material is critical both if the botanical product is to be used as a drug or as an herbal food supplement. For consumer safety and the protection of who operate in this industrial field, quality control should be applied throughout the different processing steps, from the raw material to the final product. Scientific-naturalistic illustrations of the most common buds used in BDs production (Table 1) have been realized within the Finnover project by an expert botanical graphic designer, in order to provide a useful first tool for the operators in the BDs manufacturing. In fact, this peculiar raw material is generally spontaneously collected and mistakes in the attribution of some botanical species may be possible. For this, bud illustrations could represent a preliminary control of these vegetable materials after their collection in the point of view of a controlled manufacturing chain of BDs.
Moreover, a strategy based on the untargeted UV-Visible fingerprinting coupled to chemometrics allows rapid screening of the polyphenolic BDs profile to obtain a preliminary control tool to identify the botanical species. Figure 2 show the UV-Visible spectral profiles, after SNV pretreatment of the data, recorded for the eight vegetable species investigated: Carpinus betulus L., Cornus mas L., Ficus carica L., Fraxinus excelsior L., Larix decidua Mill., Pinus montana Mill., Quercus petraea (Matt.) Liebl., Tilia tomentosa Moench. The extracts were obtained by the conventional maceration and the innovative green extraction (M or US) respectively, using the two experimental protocols (A or B) as described in detail in Table 2. Ultrasounds represent one of the innovative processing techniques of officinal plants [39]. In fact, several companies already exploit innovative applications of ultrasound to obtain liquid foods, beverages, and alcoholic drinks [55,56]. Previously, the Authors described PUAE as an alternative time-saving method to the conventional maceration for the extraction of the polyphenolic fraction from buds [11]. Particularly, PUAE on a lab pilot reactor demonstrated to be an excellent approach for a rapid (20 min vs. 21 days or 3 months of maceration, depending on the Protocol applied) and efficient extraction of phenolic compounds.

Bud-Derivatives Identification: UV-Visible Fingerprint
Looking at Figure 2, the spectra of the different vegetable species are quite different, highlighting as the pattern of absorbances in the UV-Visible region is strictly connected with the botanical origin of the plants. On the contrary, for each botanical species the spectral differences due to the extraction method (M or US) and to the extraction solvent (Protocol A or B), are minimal. The 501-900 nm interval has been preliminarily removed because there were none interesting absorptions in this spectral region at the assayed concentrations.
PCA, an unsupervised pattern recognition technique [50,51], was applied in order to explore and to analyze the data set using a multivariate approach since the analytical information contained in each spectrum was considered as a multivariate fingerprint. Particularly, the data matrix A 32,601 , whose rows are the extracts (Calibration set) and the columns are the absorbances recorded in the spectral range 200-500 nm, was considered. PCA was performed on the pretreated and autoscaled data matrix. The first two principal components (PCs) of the data set (A 32,601 ), which together explained the 77.9% of the total information of the data set since they visualize almost the 80% of the total variance, were firstly taken into account. Figure 3a,b shows the PCA score plots on the 1st-2nd principal components (PC1-PC2) obtained from the above-mentioned data matrix. In Figure 3a the extracts are categorized according to the vegetable species and each one is visualized with a different color (Carpinus betulus L.: black, Cornus mas L.: red, Ficus carica L: green., Fraxinus excelsior L.: blue, Larix decidua Mill.: brown, Pinus montana Mill.: light blue, Quercus petraea (Matt.) Liebl.: orange, Tilia tomentosa Moench: pink). In Figure 3b, for each vegetable class all the extracts belonging to the calibration set were indicated with their identification code (see Table 2). PC1, the direction of maximum variance which explains almost the 60% of the total information, allows good discrimination between the botanical class regardless of the extraction method (M or US) and the experimental preparation protocol (A or B). Particularly, the Fraxinus class (blue, lowest scores on PC1) separates from Ficus (green) and Pinus (light blue) which have higher scores on PC1. PC2, which explains the 21.1% of the remaining variance, allows to mainly separate Larix class (brown, highest scores on PC2) from Quercus (orange) and Carpinus (black, lowest scores on PC2). Figure 3c,d show the PCA score plots on the PC1-PC3, which explain together the 69.3% of the total variance of the data set. A good separation among the above cited botanical classes is also highlighted except for Larix and Carpinus ones. In fact, these latter separate on PC2 (Figure 3a,b) and since PCs are orthogonal, they are uncorrelated and no duplicate information are shown in their plots [50].
In Figure 3e,f, the projections of the external test set (red samples) were reported on the PC1-PC2 and PC1-PC3 score plots respectively, showing a good correspondence with the calibration set for each botanical species.
The spectral variables having greater importance (loading values) on the first three PCs are represented by spectral areas near the following absorbances (in ascending order): 200 nm, 212 nm, 240 nm, 275 nm, 310 nm, 360 nm, 420 nm, as highlighted in the Loading plot on PC1-PC2-PC3 (Figure 4).
Several of them could be related to some secondary metabolites largely distributed in plant material (even in buds) such as tannins, whose structural variability depends on the vegetal species and even among organs of the same plant species [57]. The chemotaxonomic values of tannins have been recognized in the literature for several botanical species [58,59] and, the distribution of hydrolysable tannins has been used as chemotaxonomic markers by several authors [60].
It is well known that the different classes of tannins present characteristic absorption bands in the UV spectral region. Particularly as far as hydrolysable tannins are concerned, gallotannins show two characteristic absorption maximums, λ max around 212 nm and λ max around 275 nm, with an inflection point (λ min) around 242 nm; ellagitannins present strong absorption near 200 nm and a shoulder around 277 nm and another absorption near 360 nm. Instead condensed tannins (or proanthocyanidins), chemically defined as flavonoid polymers in which the phenolic hydroxyls are partially or totally esterified with gallic acid, present an absorption around 200 nm, a λ min between 258-259 nm and λ max between 279-281 nm [57]. Nevertheless, also other polyphenols, such as hydroxycinnamic acids and flavonoids, could contribute to the UV-Visible fingerprints, even if some of them are more ubiquitarians and lesser species-specific [61,62]. Furthermore, as far as flavonoids are concerned, it is important to underline that their absorptions in the Visible are almost negligible at the measured concentrations, which are instead useful to avoid saturation of the UV region.
The fingerprint UV-Visible, at least in a preliminary screening step, seems to discriminate the peculiar polyphenols composition of BDs and could be a simple and quick method to confirm the proper identification of the botanical source after the botanic check by a professional botanist.
highlighted except for Larix and Carpinus ones. In fact, these latter separate on PC2 (Figure 3a,b) and since PCs are orthogonal, they are uncorrelated and no duplicate information are shown in their plots [50].
In Figure 3e,f, the projections of the external test set (red samples) were reported on the PC1-PC2 and PC1-PC3 score plots respectively, showing a good correspondence with the calibration set for each botanical species.  (Table 2); (c) PC1-PC3 score plot with BDs categorized according to the vegetable species; (d) PC1-PC3 score plot with BDs categorized according to their identification code ( Table 2); (e) PC1-PC2 score plot obtained projecting the external test set samples (highlighted in red); (f) PC1-PC3 score plot obtained projecting the external test set samples (highlighted in red).
PC3 score plot obtained projecting the external test set samples (highlighted in red).
The spectral variables having greater importance (loading values) on the first three PCs are represented by spectral areas near the following absorbances (in ascending order): 200 nm, 212 nm, 240 nm, 275 nm, 310 nm, 360 nm, 420 nm, as highlighted in the Loading plot on PC1-PC2-PC3 ( Figure  4). Several of them could be related to some secondary metabolites largely distributed in plant material (even in buds) such as tannins, whose structural variability depends on the vegetal species and even among organs of the same plant species [57]. The chemotaxonomic values of tannins have been recognized in the literature for several botanical species [58,59] and, the distribution of hydrolysable tannins has been used as chemotaxonomic markers by several authors [60].
It is well known that the different classes of tannins present characteristic absorption bands in the UV spectral region. Particularly as far as hydrolysable tannins are concerned, gallotannins show two characteristic absorption maximums, λ max around 212 nm and λ max around 275 nm, with an inflection point (λ min) around 242 nm; ellagitannins present strong absorption near 200 nm and a shoulder around 277 nm and another absorption near 360 nm. Instead condensed tannins (or proanthocyanidins), chemically defined as flavonoid polymers in which the phenolic hydroxyls are partially or totally esterified with gallic acid, present an absorption around 200 nm, a λ min between 258-259 nm and λ max between 279-281 nm [57]. Nevertheless, also other polyphenols, such as hydroxycinnamic acids and flavonoids, could contribute to the UV-Visible fingerprints, even if some of them are more ubiquitarians and lesser species-specific [61,62]. Furthermore, as far as flavonoids are concerned, it is important to underline that their absorptions in the Visible are almost negligible at the measured concentrations, which are instead useful to avoid saturation of the UV region.
The fingerprint UV-Visible, at least in a preliminary screening step, seems to discriminate the peculiar polyphenols composition of BDs and could be a simple and quick method to confirm the proper identification of the botanical source after the botanic check by a professional botanist.  Figure 5 shows the PCA plots of the data matrix B 32,620 on PC1-PC2, which together explained the 76.2% of the total variance. Foods 2020, 9, x FOR PEER REVIEW 13 of 23 Figure 5 shows the PCA plots of the data matrix B32,620 on PC1-PC2, which together explained the 76.2% of the total variance. PCA was performed on the pretreated and autoscaled data matrix, after the block scaling treatment in order to consider in the data analysis the same importance for the spectroscopic and chromatographic variables [63]. The PC1-PC2 score plot (Figure 5a) highlights a good separation between the vegetal species. Particularly PC1, which represents the direction of maximum variance explaining the 55.4% of the total information, allows good discrimination between Fraxinus class (blue, highest scores on PC1), Ficus (green) and Pinus (light blue) classes, which have lowest scores on this PC. As highlighted in the Biplot (Figure 5b) the variables having greater importance (loading value) on this separation are represented by total cinnamic acids, caffeic acid, coumaric acid and hyperoside content which are high in Fraxinus species and very low in Pinus one (as reported in Table  4). Instead PC2, which explains the 20.7% of the remaining information, allows mainly to separate Carpinus (black) and Cornus (red) classes from all the other ones. These species result particularly rich in tannins (catechins and benzoic acids). Table 4. Bioactive classes and total phenolics in the analyzed samples.

Cinnamic Acids
Flavonols Benzoic Acids Catechins Total Phenolics PCA was performed on the pretreated and autoscaled data matrix, after the block scaling treatment in order to consider in the data analysis the same importance for the spectroscopic and chromatographic variables [63]. The PC1-PC2 score plot (Figure 5a) highlights a good separation between the vegetal species. Particularly PC1, which represents the direction of maximum variance explaining the 55.4% of the total information, allows good discrimination between Fraxinus class (blue, highest scores on PC1), Ficus (green) and Pinus (light blue) classes, which have lowest scores on this PC. As highlighted in the Biplot (Figure 5b) the variables having greater importance (loading value) on this separation are represented by total cinnamic acids, caffeic acid, coumaric acid and hyperoside content which are high in Fraxinus species and very low in Pinus one (as reported in Table 4). Instead PC2, which explains the 20.7% of the remaining information, allows mainly to separate Carpinus (black) and Cornus (red) classes from all the other ones. These species result particularly rich in tannins (catechins and benzoic acids). In the Supplementary materials an example (Larix decidua) of chromatographic pattern was reported. As shown in Figure 5, the addition of chromatographic variables does not greatly improve the taxonomic separation previously obtained by the only UV-Visible fingerprint ( Figure 3). However, these results show that the main polyphenols evaluated could be useful markers for identifying the botanical species regardless of the extraction method and the experimental preparation protocol.

Phenolic Composition of BDs
In this study, the health-promoting compounds were grouped into four different polyphenolic classes in order to assess the contribution of each class to the phytocomplex composition of buds belonging to the eight different species: cinnamic acids (as sum of caffeic acid, chlorogenic acid, coumaric acid, ferulic acid), flavonols (as sum of hyperoside, isoquercitrin, quercetin, quercitrin and rutin), benzoic acids (ellagic and gallic acids) and catechins ((+)catechin and (-)epicatechin). The identification and quantification of each single bioactive compound, expressed in mg/100 g FW , is reported in the Supplementary Materials (Table S2). For a better data visualization, Figure 6 shows the radar plot, made considering for each botanical species the mean values obtained from the 4 different extracts (M_A, M_B, US_A, US_B) for each marker compound quantified. coumaric acid, ferulic acid), flavonols (as sum of hyperoside, isoquercitrin, quercetin, quercitrin and rutin), benzoic acids (ellagic and gallic acids) and catechins ((+)catechin and (-)epicatechin). The identification and quantification of each single bioactive compound, expressed in mg/100 g FW, is reported in the Supplementary Materials (Table S2). For a better data visualization, Figure 6 shows the radar plot, made considering for each botanical species the mean values obtained from the 4 different extracts (M_A, M_B, US_A, US_B) for each marker compound quantified. Several markers of cinnamic acids were considered but not detected in all the extracts. Fraxinus excelsior BDs showed the highest content in cinnamic acids (ranged from 113.53 ± 6.70 to 829.03 ± 2.26 mg/100 g FW), and as shown in Table S2, ferulic and chlorogenic acids were the most abundant. Cornus mas and Tilia tomentosa species showed very low amounts of ferulic acid (respectively 12.14 and 11.73 mg/100 g FW), while in the other species it was not detected. In recent years, several physiological functions of ferulic acid have been demonstrated [64]. Particularly, its free radical scavenging activity and its cholesterol-lowering activity, together with the low toxicity, suggested its chemo preventive effects on heart diseases [65]. Chlorogenic acid is also involved in beneficial effects on human health due to its anti-inflammatory, antioxidative, anti-aging and anticancer activities [66]. Chlorogenic acid was detected only in Fraxinus excelsior BDs (ranges from 43.88 to 489.94 mg/100 gFW), in all the other species it was not detectable. Li et al. 2013 reported that chlorogenic acid and flavonols may be considered the main phenolic compounds responsible for in vitro anti-cancer property (i.e., against breast, colon, liver and lung cancer) [66]. As regards the total flavonol content, it was highly variable among species. The highest content was quantified in Cornus mas species (mean value: 705.71 mg/100 g FW) while the lowest value in Pinus montana (mean value: 25.31 mg/100 g FW). As highlighted in Figure 6. The mean content of each phenolic marker (caffeic acid, chlorogenic acid, coumaric acid, ferulic acid, hyperoside, isoquercitrin, quercetin, quercitrin and rutin, ellagic acid, gallic acid, (+)catechin and (-)epicatechin) for the eight botanical species investigated. very abundant in almost all the described species (Table S2) while gallic acid was not detectable in Ficus carica, Fraxinus excelsior, Larix decidua and Pinus montana species. The highest content in ellagic acid was identified in Cornus mas extracts, followed by Larix decidua and Quercus petraea BDs (Figure 6, in red).
Catechins have important effects on human health thanks to its antioxidant, anti-inflammatory, antidiabetic, and antimicrobial properties [67]. The intake of foods and dietary supplements rich in catechins could have an important role in the prevention of various diseases (i.e., cardiovascular diseases), inhibition of lipid peroxidation, improvement of blood flow, elimination of several toxins and inhibition of human cancer cell line proliferation and cyclooxygenase enzymes [70]. All the vegetal species considered in this research were a good source of catechins (catechin and epicatechin) as shown in Table 3 and Table S2. Particularly, as highlighted in Figure 6, catechin represented the phenolic marker of Pinus montana BDs (in violet), while epicatechin was more abundant in Carpinu betulus and Quercus petraea extracts (in light green).
All BDs analyzed showed a good content of phenolics although there was a high variability both between the different vegetal species and between the extracts obtained by the different manufacturing method and experimental conditions starting from the same botanical species. Figure 7 showed the radar plots of each botanical species in order to better highlight the phenolic composition of the 4 different extracts (M_A, M_B, US_A, US_B).
As showed in Figure 7, the manufacturing methods (conventional maceration or sonication) and the experimental conditions used for the preparation of BDs (i.e., extraction solvent, extraction time, solid/ solvent ratio, extraction time) strongly influenced the phenolic extraction yield despite having removed the variability of the raw material (same batch of buds for each vegetal species). Generally US_A (green line) and US_B (red line) appears more similar in terms of phenolic composition respect to M extracts (M_A and M_B), except for some species, such as Pinus montana and Larix decidua, in which there is a greater homogeneity in the polyphenolic profile of the final products. In almost all species, the M_B extract (yellow line) is the most different from the others. In example, the M_B extract of Cornus mas was rich in catechin which was not detected in extracts obtained by different extraction conditions (M_A, US_A, US_B). Analogously, rutin represents a phenolic marker of the M_B extract of Fraxinus excelsior, while it was poorly detectable in the other extracts of the same species. Surely Protocol A, according to the European Pharmacopoeia, provided a higher alcoholic concentration of the extraction solvent than protocol B and it is known that a higher solvent polarity allows a higher phenolic extraction from plant materials [71]. Moreover, Protocol A used an higher solid/solvent ratio because it is evaluated on the dry weight of the raw material while following the industrial Protocol (B), the fresh weight of buds was taken into account. Regarding the effect of ultrasounds, the implosion of cavitation bubbles on the material surface results in micro-jetting which generates several effects such as surface peeling, detexturation, erosion and cell breakdown [40]. Probably, the destruction of vegetal cells allowed to increase the extraction yield making up for the lower alcohol content of protocol B.
Due to the lack of a single regulation and an unique preparation protocol for these botanicals, very different products can be found on the market in terms of their polyphenolic fraction depending on both the raw materials (i.e., taking into account their specific agro-environmental and biological traits) and on the experimental conditions of manufacturing (method of preparation, extraction solvent, solid/solvent ratio, extraction time).

Conclusions
Although BDs have been widely used in traditional medicine because of the peculiar content of buds in phenolic compounds, nowadays they are a category of botanicals still poorly studied. The lack of detailed scientific information and a clear and unique regulation, it makes these products high risk and vulnerable for accidental mistakes in the attribution of the botanical species, but also frauds and adulterations. Moreover, the polyphenols content of BDs is strongly influenced by the manufacturing processes whose parameters are often not strictly defined (e.g., solvent ratios in the extraction mixtures, raw material/extraction mixture ratios, extraction time) and thus they affect their final compositions.
This research, within the Finnover project, aims to answer to the growing demand for efficient quality control in the BDs field to guarantee the proper attribution of the botanical source and their content. Moreover, a manufacturing process specification should be advisable to monitor the bioactive contents.
UV-Visible spectroscopy and HPLC-DAD analysis have been employed to obtain an untargeted and a targeted phytochemical fingerprint of BDs, respectively. UV-Visible coupled with an appropriate chemometric data processing is a simple, rapid and low-cost technique proved to be very useful to identify the botanical source regardless the manufacturing method and the experimental conditions used. Moreover, the targeted phytochemical fingerprint by HPLC-DAD allowed to obtain a detailed screening of the BDs polyphenolic profile which highlighted an high variability due to the different vegetal species and to the manufacturing method and protocol. The ultrasonic extraction of buds compared to conventional maceration proved less sensitive to the different extraction protocols.
The proposed strategy offers to those operating in this industrial sector an untargeted method for the identification of the bud's botanical species and a green extraction strategy (PUAE) which is more robust with respect to the different extractive protocols that can be used. The same approach, described for BDs, could be analogously applied to other botanical productions.