Chemophenetic Approach to Selected Senecioneae Species, Combining Morphometric and UHPLC-HRMS Analyses

Herein, a chemophenetic significance, based on the phenolic metabolite profiling of three Senecio (S. hercynicus, S. ovatus, and S. rupestris) and two Jacobaea species (J. pancicii and J. maritima), coupled to morphometric data, is presented. A set of twelve morphometric characters were recorded from each plant species and used as predictor variables in a linear discriminant analysis (LDA) model. From a total 75 observations (15 from each of the five species), the model correctly assumed their species’ membership, except for 2 observations. Among the studied species, S. hercynicus and S. ovatus presented the greatest morphological similarity. A phytochemical profiling of phenolic specialized metabolites by UHPLC-Orbitrap-MS revealed 46 hydroxybenzoic, hydroxycinnamic, and acylquinic acids and their derivatives, 1 coumarin and 21 flavonoids. Hierarchical and PCA clustering applied to the phytochemical data corroborated the similarity of S. hercynicus and S. ovatus, observed in the morphometric analysis. This study contributes to the phylogenetic relationships between the tribe Senecioneae taxa and highlights the chemophenetic similarity/dissimilarity of the studied species belonging to Senecio and Jacobaea genera.


Introduction
The tribe Senecioneae (Asteraceae) encompasses more than 150 genera and 3000 species; approximately half of its species belong to the genus Senecio L., considering it one of the largest genera of flowering plants [1]. Senecio species have a wide distribution and they occur in various habitats-from low altitudes to high mountain communities, and from Arctic regions to hot tropical areas [1]. Although phylogenetic studies have been carried out classifying the taxa, the intergeneric relations are still vague [1,2]; some Senecio species have been recently transferred to a separate genus, Jacobaea Mill. [3]. Within the genus Senecio, hybridization was observed, e.g., S. hercynicus × S. ovatus [4]. Most taxa in the tribe can be identified by the existence of capitula (flower heads) with a typically uniseriate involucre. However, some species are poorly differentiated morphologically, and there is still uncertainty about recognition of their taxa [2,[5][6][7]. Senecio species are reported to accumulate sesquiterpenoids (eremophilanes, furanoeremophilanes, cacalols, eudesmanes, oplopanes, germacranes, etc.), pyrrolizidine alkaloids (PAs) [1,8,9], phenolic compounds [10][11][12][13][14][15], and various other secondary metabolites [4,9,16]. Senecio species have been described to possess analgesic [17] and hypoglycemic [18] activity, related to the typical for the genus sesquiterpene lactones, and insecticidal properties [19] related to the presence of PAs. Additionally, the taxa are reported to express strong antioxidant, cytotoxic, and antimicrobial activity attributed to the presence of phenolic compounds [11][12][13]15].
In the Bulgarian flora, Senecio hercynicus Herborg., S. ovatus (G. Gaertn. and Al.) Willd., S. rupestris Waldst. and Kit., and Jacobaea pancicii (Degen) Vladimirov and Raab-Straube are perennial plants distributed in the mountain regions, up to 1500 (2200) m a.s.l., while J. maritima number of ray flowers (X8) was positively associated to the involucr and number of disc flowers (X10). Additionally, the variables X6, X8, a positive correlation between each other. Next, a linear discriminant analysis (LDA) was performed o [29,30]. Based on several calculated parameters ( Figure S1), includin squares (RSS), adjusted R2, Mallow's Cp, and Bayesian information variable model was selected, including the variables X1, X4, X7, X8, X Then, 80% of the data was used as a training set (n = 60) and 20%-a a random principle. The training set was used to derive a linear mo species of a plant, based on the selected set of six parameters. The l to correctly predict the species on the test set (n = 15), but one (T dimensional plot derived from the linear model (Figure 2A), S. ov were not well-distinguished, while on the two-dimensional pl separated, except S. ovatus and S. hercynicus with a partial overlap (F Next, a linear discriminant analysis (LDA) was performed on the X 1-12 variables [29,30]. Based on several calculated parameters ( Figure S1), including the residual sum of squares (RSS), adjusted R2, Mallow's Cp, and Bayesian information criterion (BIC), a six variable model was selected, including the variables X 1 , X 4 , X 7 , X 8 , X 9 , and X 11 (Table S3). Then, 80% of the data was used as a training set (n = 60) and 20%-as a test set (n = 15), on a random principle. The training set was used to derive a linear model for predicting the species of a plant, based on the selected set of six parameters. The linear model was able to correctly predict the species on the test set (n = 15), but one (Table S4). On the one-dimensional plot derived from the linear model (Figure 2A), S. ovatus and S. hercynicus were not well-distinguished, while on the two-dimensional plot, all species were separated, except S. ovatus and S. hercynicus with a partial overlap ( Figure 2B).
Given these results, the morphological variability of the Senecio and Jacobaea species is not random and is long-established for the tribe Senecioneae taxa as prominent [31]. Although the Jacobaea is distinguished from Senecio sensu stricto, a clear morphological synapomorphies for Jacobaea have not yet been recognized [1]. The received data by the morphometrical study unequivocally confirm the taxonomical relationship of S. hercynicus and S. ovatus belonging to S. nemorensis group and the transfer of the last-mentioned species to the genus Senecio. Moreover, the results favor the delimitation of J. maritima and J. pancicii from the genus Senecio and the distinguishing of the other studied taxa [20]. a random principle. The training set was used to derive a linear model for predicting the species of a plant, based on the selected set of six parameters. The linear model was able to correctly predict the species on the test set (n = 15), but one (Table S4). On the onedimensional plot derived from the linear model (Figure 2A), S. ovatus and S. hercynicus were not well-distinguished, while on the two-dimensional plot, all species were separated, except S. ovatus and S. hercynicus with a partial overlap ( Figure 2B). Given these results, the morphological variability of the Senecio and Jacobaea species is not random and is long-established for the tribe Senecioneae taxa as prominent [31]. Although the Jacobaea is distinguished from Senecio sensu stricto, a clear morphological synapomorphies for Jacobaea have not yet been recognized [1]. The received data by the morphometrical study unequivocally confirm the taxonomical relationship of S. hercynicus and S. ovatus belonging to S. nemorensis group and the transfer of the last-mentioned species to the genus Senecio. Moreover, the results favor the delimitation of J. maritima and J. pancicii from the genus Senecio and the distinguishing of the other studied taxa [20].

UHPLC-HRMS Identification and Tentative Annotation of Specialized Natural Products
In order to establish a phenolic metabolite profiling, combined hydromethanolic plant extracts were prepared, i.e., LC-MS measurements on phenolic content were performed on the homogenized samples from the 15 collected plants from each species, as described in Section 3.4. Based on chromatographic retention times, MS and MS/MS accurate measurements, fragmentation patterns, and the comparison with reference standards and literature data, a total of 46 hydroxybenzoic, hydroxycinnamic, and acylquinic acids and their derivatives, 1 coumarin and 21 flavonoids, were annotated in the tested extracts. The LC-MS and MS/MS data of all 68 identified phenolic compounds are presented in

UHPLC-HRMS Identification and Tentative Annotation of Specialized Natural Products
In order to establish a phenolic metabolite profiling, combined hydromethanolic plant extracts were prepared, i.e., LC-MS measurements on phenolic content were performed on the homogenized samples from the 15 collected plants from each species, as described in Section 3.4. Based on chromatographic retention times, MS and MS/MS accurate measurements, fragmentation patterns, and the comparison with reference standards and literature data, a total of 46 hydroxybenzoic, hydroxycinnamic, and acylquinic acids and their derivatives, 1 coumarin and 21 flavonoids, were annotated in the tested extracts. The LC-MS and MS/MS data of all 68 identified phenolic compounds are presented in Table 1 along with their distribution in the studied extracts.

Chemophenetic Significance
The chemophenetic significance of phenolic metabolite profiling coupled to morphometric data of the studied Senecioneae species is presented. The raw LC-MS data of annotated specialized compounds were converted and further manipulated with the R programming language, as detailed in Section 3.6. Integration of the Full-MS intensity signals corresponding to the identified compounds allowed the determination of their AUC values. These AUC values were used as a relative quantitative measure for a particular compound, between the studied species. In order to do so, the AUC values were normalized from 0 to 100. Thence, a similarity/dissimilarity clustering analysis of the species was conducted for those compounds found in at least two, out of all five, species (Table 2, Figures 3 and 4). Figure 3 depicts a heatmap of the AUC values from Table 2. The dendrograms separated the compounds (columns) into five clusters, and the species (rows) into two clusters (Figure 3).
The clustering, by rows, did not differentiate the two genera. The greatest resemblance was generated between S. ovatus and S. hercynicus; J. pancicii showed greater similarity to the last-mentioned two species compared to J. maritima; S. rupestris was cast as a separate node. Table S5 presents the grouped compounds from Figure 3, where it is notable which compounds were characteristic for a given species. Hence, the contribution of the annotated phenolic compounds to the phenetic description of the selected taxa was determined. For example, hydroxybenzoic (1, 4, 6, 11, 16, 17, and 42) and hydroxycinnamic (10, 15, 19, 22,  27, and 28), derivatives as well as the flavonol glucosides (57 and 58) were dominant in S. rupestris, while diAQAs (37, 38, 39, 40, 43, 44, and 46), triAQA (47) and the flavonols (53, 59, 63, and 65) were in the highest amount in J. maritima. The coumarin 12, acylquinic acids (35 and 45) and flavonoid hexuronides (54, 55, and 60) were characteristic for S. ovatus. J. pancicii, on the other hand, presented the highest amount of AQAs (9, 23, and 30), hydroxycinnamic (5 and 8), and flavonol (48, 51, and 56) derivatives. A heatmap of the Euclidean distance and a PCA plot (of the data in Table 2) are shown in Figure 4, where similar clustering is observed, compared to that in Figure 3. Table 2. Normalized (by rows) AUC values of the identified specialized natural compounds found in at least two, out of all five, studied species. The cells show the normalized AUC, from 0 (in case the compound was not detected in the extract) to 100, by rows.  As phenolic content varies between different plant parts, the %CV of the morphometric characteristics, recorded for each plant species, were typically below 20%CV, except for the number of capitula per plant (X 12 ) reaching above 40%CV (Table S2). Noteworthy, for each of the plant species, the LC-MS measurements on the phenolic content were performed on homogenized samples from all 15 aerial plant samples, providing a representative phenolic profile. Overall, the morphometric data ( Figure 2) corroborates the taxonomical relationship of S. hercynicus and S. ovatus to the S. nemorensis group. Moreover, a delimitation was observed between the two Jacoboea species (J. maritima and J. pancicii) from the genus Senecio and distinguishing of the other studied taxa [20], and similar findings were detected by to the unsupervised clustering methods applied on the phytochemical data (Figures 3 and 4). In both morphometric characteristics and phenolics content, S. hercynicus and S. ovatus showed the highest similarity.  Vladimirov, 2012 [20]. A voucher specimen of S. hercynicus was deposited at Herbarium Academiae Scientiarum Bulgariae (SOM 177012). S. ovatus, S. rupestris, J. pancicii, and J. maritima specimens were given at Herbarium Facultatis Pharmaceuticae Sophiensis, Medical University-Sofia, Bulgaria (Voucher specimen № 11 631-11 634).

Morphometric Measurements
Morphometric measurements on the studied Senecio and Jacobaea species were performed on 15 randomly chosen plants, from each species, during the full flowering stage. The morphometric variability was determined using 12 quantitative characters (parameters) as follows: X 1 -root diameter [cm]; X 2 -stem height [cm]; X 3 -leaf length [cm]; X 4 -leaf width [cm]; X 5 -involucral bract length [cm]; X 6 -involucral bracts number per capitula; X 7 -ray flower length [cm]; X 8 -number of ray flowers per flower head; X 9 -disc flower length [cm]; X 10 -number of disc flowers per flower head; X 11 -flower head diameter [cm]; and X 12 -number of capitula per plant. The morphometric measurements are presented in Table S1. Descriptive statistics of the 12 characteristics was performed in the R programming language and presented in Table S2.

Chemicals and Reagents
Acetonitrile and formic acid for LC-MS, and methanol of analytical grade, were purchased from Merck (Merck, Bulgaria). The reference standards used for compound identification were bought from Phytolab (Vestenbergsgreuth, Germany).

Ultra-High-Performance Liquid Chromatography-High Resolution Mass Spectrometry (UHPLC-HRMS)
Elution was achieved on a reversed phase column Kromasil EternityXT C18 (1.8 µm, 2.1 × 100 mm, AkzoNobel, Sweden) column maintained at 40 • C. The binary mobile phase consisted of A: 0.1% formic acid in water and B: 0.1% formic acid in acetonitrile. The run time was 24.5 min. Prior to injection, the mobile phase was held at 50% B for 4.5 min, and then gradually turned at 5% B in 0.5 min. After injection, the % B was gradually turned to 60% B over 15 min, and then held at 60% B for 3 min, increased gradually to 95% B over 3 min, held at 95% B over 2 min, then turned to 50% B in 0.5 min. The retention time of the identified compounds ranged between 1.74 and 9.60 min. The flow rate and the injection volume were set to 300 µL/min and 1 µL, respectively. The effluents were connected on-line with a Q Exactive Plus Orbitrap mass spectrometer (ThermoFisher Scientific) where the compounds were detected. Data were processed with Xcalibur software 4.2 (ThermoFisher Scientific, Waltham, MO, USA).
Mass spectrometric analyses were carried out on a Q Exactive Plus Mass Spectrometer (ThermoFisher Scientific) equipped with a heated electrospray ionization (HESI-II) probe (ThermoFisher Scientific). The tune parameters were as follows: spray voltage 3.5 kV; sheath gas flow rate 38; auxiliary gas flow rate 12; spare gas flow rate 0; capillary temperature 320 • C; probe heater temperature 320 • C, and S-lens RF level 50. Acquisition was acquired at Full-scan MS and Data Dependent-MS 2 modes. Full-scan spectra over the m/z range 100 to 1000 were acquired in the negative ionization mode at a resolution of 70,000. Other instrument parameters for the Full MS mode were set as follows: AGC target 1e6, maximum ion time 80 ms, number of scan ranges 1. For DD-MS 2 mode, instrument parameters were as follows: microscans 1, resolution 17,500, AGC target 1e5, maximum ion time 50ms, MSX count 1, isolation window 1.0 m/z, stepped collision energy (NCE) 10, 30, and 60. Data acquisition and processing were carried out with Xcalibur 4.2 software (ThermoFisher Scientific).

File Conversions and Data Analysis
After the .raw (ThermoFisher Scientific) mass spectrometric files were obtained, they were converted to .ms1 (MS1 data) and .mgf (MS2 data) files using MSConvertGUI 3.1 (ProteoWizard). Then, the .ms1 and .mgf files were imported to RStudio (2021, Build 382) and further manipulated under the R programming language (version 4.2.1, 23 June 2022, "Funny-Looking Kid"). The MS2 spectra were screened for the presence of the available target (hydroxybenzoic acid derivatives and flavonoids) standard compounds. The screening was achieved by selecting spectra based on the following criteria: m/z error of the molecular ion < 15 ppm (minimum 0.0010 Da), retention time error < 2% (minimum 0.05 min, maximum 0.15 min), number of fragment ions match > 2/3, and error of the percentage intensity of matched fragment ion < 15. Similar MS2 scans found in the same chromatographic peak were grouped, i.e., the spectra were summed, the m/z were adjusted by weight averaging: where (m/z) avg is the recalculated m/z value, (m/z) i and int i are the m/z and the intensity of the ith fragment ion, respectively. The areas under the curve (AUC) of the identified compounds were calculated and normalized from 0 to 100. Data analysis was performed in the R programming language (R version 4.2.1., 23 June 2022, Funny-Looking Kid), operated under the RStudio environment (2022.07.2 Build 576). R packages used include: "MASS" [36], "klaR" [37], "caret" [38], "leaps" [39], "factoextra" [40], "cluster" [41], "lpSolve" [42], "DescTools" [43], "pheatmap" [44], and "arsenal" [45]. Distance matrices were generated using the "Euclidean" method, and hierarchical clustering was performed using the "ward.D2" method. The complete R code used for morphometric analysis is presented in the supplementary material.

Conclusions
Herein, a chemophenetic study of three Senecio (S. hercynicus, S. ovatus, and S. rupestris) and two Jacobaea species (J. pancicii and J. maritima) is presented. From the collected morphometric data, describing 12 parameters, a distinguishment of species by genera was performed using linear discriminant analysis (LDA). Among the studied species, S. her-cynicus and S. ovatus presented the greatest similarity, and hence, their formed clusters were the closest. Even though no overlap in the LDA analysis was observed between the Jacoboea and Senecio species, J. pancicii and J. maritima did not demonstrate likeness. A phytochemical analysis by UHPLC-Orbitrap-HRMS revealed a total of 46 hydroxybenzoic, hydroxycinnamic, and acylquinic acids and their derivatives, 1 coumarin and 21 flavonoids. Hierarchical and PCA clustering was then applied to the phytochemical data on combined plant material from each species. The data corroborated the similarity of S. hercynicus and S. ovatus, established in the morphometric analysis. The study highlights the similarity/dissimilarity, both morphometric, and in a manner of specialized metabolites, of the selected species belonging to Senecio and Jacobaea genera (Senecioneae).

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants12020390/s1, Table S1: Scaled and unscaled raw morphometric data; Table S2: Descriptive statistics on the morphometric data; Table S3: Selection of a model with n variables; Table S4: Prediction of membership of the test set (n = 15); Table S5: The compounds presented in Figure 3 grouped by the hierarchical clustering; Figure S1: Residual sum of squares (RSS), adjusted R2, Mallow's Cp, and Bayesian information criterion (BIC) for the standardized morphological data. R code: contains the code written in the R programming language for the analysis of the morphological data.

Data Availability Statement:
The data presented in this study are available in the article or supplementary material. The raw MS files are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.