Abstract

   Background: Paeonia lactiflora Pall. (PLP) is a plant with excellent
   ornamental and therapeutic value that can be utilized in traditional
   Chinese medicine as Paeoniae Radix Alba (PRA) and Paeoniae Radix Rubra
   (PRR). PRA must undergo the “peeling” process, which involves removing
   the cork and a portion of the phloem. PLP’s biological function is
   strongly linked to its secondary metabolites, and the distribution of
   metabolites in different regions of the PLP rhizome causes changes in
   efficacy when PLP is processed into various therapeutic compounds.
   Methods: The metabolites of the cork (cor), phloem (phl), and xylem
   (xyl) were examined in the roots of PLP using a metabolomics approach
   based on UPLC-Q-Exactive-Orbitrap-MS/MS (UPLC-MS/MS), and the
   differential metabolites were evaluated using multivariate analysis.
   Results: Significant changes were observed among the cor, phl, and xyl
   samples. In both positive and negative ion modes, a total of 15,429
   peaks were detected and 7366 metabolites were identified. A total of
   525 cor-phl differential metabolites, 452 cor-xyl differential
   metabolites, and 328 phl-xyl differential metabolites were evaluated.
   Flavonoids, monoterpene glycosides, fatty acids, sugar derivatives, and
   carbohydrates were among the top 50 dissimilar chemicals. The key
   divergent metabolic pathways include linoleic acid metabolism,
   galactose metabolism, ABC transporters, arginine biosynthesis, and
   flavonoid biosynthesis. Conclusion: The cor, phl, and xyl of PLP roots
   exhibit significantly different metabolite types and metabolic
   pathways; therefore, “peeling” may impact the pharmaceutical effect of
   PLP. This study represents the first metabolomics analysis of the PLP
   rhizome, laying the groundwork for the isolation and identification of
   PLP pharmacological activity, as well as the quality evaluation and
   efficacy exploration of PLP.

   Keywords: Paeonia lactiflora Pall., UPLC-MS/MS, untargeted
   metabolomics, stoichiometry, KEGG pathway analysis

1. Introduction

   Paeonia lactiflora Pall., with a rich history of approximately 3000
   years, has been mentioned in ancient texts such as the Book of Songs
   and Shennong Bencao Jing, marking its significance in traditional
   Chinese medicine. Its first recorded medicinal use dates back to
   Fifty-two Prescriptions, and today, it is extensively employed in the
   pharmaceutical field [[38]1]. Pharmacological studies have demonstrated
   its efficacy in treating a variety of conditions including depression,
   atherosclerosis, rheumatoid arthritis, liver damage, ulcerative
   colitis, Parkinson’s disease, and Alzheimer’s disease [[39]2].

   Clinically, PLP is utilized in two primary forms: Paeoniae Radix Rubra
   (PRR) and Paeoniae Radix Alba (PRA). As per the 2020 edition of the
   Chinese Pharmacopoeia [[40]3], PRA is the dried medicinal material of
   the Ranunculaceae family’s Paeonia lactiflora Pall., boiled and peeled,
   known for its blood-nourishing and menstruation-regulating properties,
   as well as liver-soothing and pain-relieving effects. PRR, on the other
   hand, represents the direct use of the dried medicinal material of the
   plant, effective in clearing heat, cooling blood, breaking blood
   stasis, and alleviating pain. PLP is composed of three layers from the
   outside in: cor, phl, and xyl. The “skin” removed during the PRA
   production process includes the cork and a minimal amount of the
   phloem, whereas the dried PRA contains portions of both the phloem and
   xylem. Researchers believe that the pharmacological and pharmacodynamic
   differences between PRA and PRR stem from their distinct processing
   methods, which impact the concentration of chemical constituents and
   thus the efficacy of the medicinal material [[41]1]. Studies have
   identified significant amounts of benzoic acid and albiflorin in the
   outer skin of the PLP root rhizome [[42]4,[43]5], with paeoniflorin and
   albiflorin primarily located in the cork and cortex [[44]6].
   Paeoniflorin, serving as a quality marker for both PRA and PRR, is
   crucial to the quality of PLP medicinal materials [[45]3]. Hence,
   research into the various parts of the PLP root can deepen our
   understanding of its chemical composition differences and
   pharmacological properties, further improving quality control and
   providing more reliable support for clinical applications.

   Recent advancements in metabolomics and chemometrics have offered new
   perspectives in understanding PLP. Studies using modern instruments
   such as HPLC, ^1H-NMR, UPLC-Q/TOF-MS, and HPLC-DAD-ESI-MS have
   elucidated the chemical composition differences among various peony
   varieties [[46]7], cultivation regions [[47]8], and different parts of
   the plant such as the roots, stems, and flowers [[48]9]. Furthermore,
   research utilizing these methods has aided in analyzing the
   distribution of eleven chemical components including peony glycosides
   in the PLP root [[49]10]. However, limitations in the sensitivity,
   separation, and detection thresholds of instruments such as HPLC and
   UPLC have restricted the analysis to only a few chemical components,
   leaving a gap in understanding the full spectrum of chemical variations
   in the PLP root. Additionally, a detailed exploration of the chemical
   composition differences among the cork, phloem, and xylem of the PLP
   root remains unexplored. This research gap forms the basis of this
   study, which focuses on investigating the chemical component
   distribution in these specific parts of the PLP root to more
   comprehensively reveal the chemical characteristics and pharmacological
   mechanisms of PLP.

   The integration of untargeted metabolomics with chemometrics has become
   an important and valuable tool in various life science research areas
   in recent years, including biomarker discovery, disease diagnosis, and
   the quality assessment of food and herbal medicines [[50]11,[51]12].
   The strength of untargeted metabolomics lies in its ability to analyze
   and compare a multitude of complex components without the need for
   prior knowledge of the material composition or key metabolic products
   [[52]13]. Chemometric methods can be used to extract crucial
   information from raw data, revealing potential correlations among a
   vast array of variables. Tandem mass spectrometry coupled with
   UPLC-MS/MS is widely regarded as the preferred and effective method for
   the analysis of traditional Chinese medicine, providing structural
   fragment information of compounds at different retention times
   [[53]14]. UPLC-MS/MS not only enables the chemical analysis of
   traditional Chinese medicine samples but also facilitates the analysis
   of different parts and sources of plants. Additionally, it allows for a
   rough estimation of the relative content of components based on signal
   intensity [[54]15].

   In summary, this study, by combining untargeted metabolomics and
   chemometric methods with UPLC-MS/MS, aims to reveal the metabolic
   profile differences in various parts of the PLP root and further
   explore the impact of these differences on the chemical properties and
   pharmacological mechanisms of PLP. This research not only promises to
   provide new scientific evidence for the quality control of PLP but also
   deepens our understanding of this ancient herbal medicine, supporting
   its broader medical applications. It also offers new perspectives and
   possibilities for drug development and clinical treatment.

2. Results and Discussion

2.1. Multivariate Statistical Analysis

   In this study, a total of 15,429 substance peaks were detected, with
   7366 unique metabolites being identified. Among them, 5109 were
   identified under ESI+ mode and 2257 were identified under ESI− mode.
   Notably, no metabolites were detected simultaneously in both ESI+ and
   ESI− modes, indicating distinct metabolomic profiles captured by each
   ionization technique. The base peak chromatograms for both positive and
   negative ion modes are presented in [55]Supplementary Material Figure
   S1. Multivariate statistical analyses, including principal component
   analysis (PCA), partial least-squares–discriminant analysis (PLS-DA),
   and orthogonal partial least-squares–discriminant analysis (OPLS-DA),
   were applied to the detected substance peaks.

   PCA was utilized to understand the overall situation of the
   metabolites. PLS-DA was employed to distinguish metabolic differences
   between different groups, and OPLS-DA was used for model refinement,
   noise reduction, and enhancing analytical power and effectiveness. It
   emphasized the results provided by PCA and PLS-DA and highlighted the
   differences both between and within the sample groups. Additionally,
   seven-fold cross-validation and 200-times response permutation testing
   (RPT) were conducted to assess whether the models suffered from
   overfitting.

   The model parameters for multivariate statistical analysis among the
   three groups are shown in [56]Supplementary Material Table S1. The
   results of the PCA, PLS-DA, and OPLS-DA analyses are presented in
   [57]Figure 1. The PCA plots demonstrate clear separations between the
   three groups, with high values of R2X[1] indicating that the first
   principal component significantly contributes to the explanation of
   data variability. No obvious outliers were observed, indicating the
   good statistical consistency of the model ([58]Figure 1(A1–C1)). In the
   PLS-DA plots, the values of R2X[1] and R2X[2] confirm the
   differentiation capability of PLS-DA, showcasing its efficiency. The
   values of parameters R2Y and Q2 were 0.995 and 0.992; 0.998 and 0.993;
   and 0.998 and 0.944, respectively ([59]Supplementary Material Table
   S1), indicating the good predictive ability of the PLS-DA model
   ([60]Figure 1(A2–C2)). The OPLS-DA plots further revealed significant
   differences between cor, phl, and xyl. The values of R2X[1] and R2X0[1]
   demonstrated the efficacy of OPLS-DA in capturing key differences
   between groups ([61]Figure 1(A3–C3)). The 200-times permutation test
   for cross-validation showed that R2 was close to 1 and Q2 was less than
   0, indicating the high adaptability and reliability of the OPLS-DA
   model in handling complex metabolic data ([62]Figure 1(A4–C4)).

Figure 1.

   [63]Figure 1
   [64]Open in a new tab

   Multivariate analysis among the cor, phl, and xyl groups. (A): cor-phl;
   (B): cor-xyl; (C): phl-xyl; (1): PCA; (2): PLS-DA; (3): OPLS-DA; (4):
   permutation. R2X (cum): cumulative explained variance in the
   X-direction; R2Y (cum): cumulative explained variance in the
   Y-direction; Q2 (cum): cumulative predictive ability of the model; R2
   and Q2: parameters of the permutation test, used to evaluate model
   overfitting.

2.2. Analysis of Differential Metabolites

   Univariate analysis methods such as the Student’s t-test and fold
   change analysis were employed to compare the expression of differential
   metabolites between the different experimental groups. The fold change
   (FC) values of the detected substance peaks were converted to log2
   (FC), and the p-values from the Student’s t-test (p = 0.05) were
   converted to −log10 (p-value) for the construction of volcano plots
   ([65]Figure 2).

Figure 2.

   [66]Figure 2
   [67]Open in a new tab

   Comparison of differential metabolites among the cor, phl, and xyl
   groups. (A): cor-phl; (B): cor-xyl; (C): phl-xyl; (D): the number of
   differential metabolites among the groups. Differential metabolites
   were selected with a threshold of variable importance in projection
   (VIP) > 1.0, p-value < 0.05. Red dots indicate significant upregulation
   (log2(FC) > 0), blue dots indicate significant downregulation (log2(FC)
   < 0), and gray dots indicate no significant difference.

   As indicated in the volcano plots, significant metabolic changes exist
   between cor and xyl, with the highest number of significantly
   downregulated metabolite points, suggesting the potential inhibition of
   key metabolic pathways in cor ([68]Figure 2B). Between phl and xyl,
   there are fewer significantly downregulated points and more
   non-significant gray points ([69]Figure 2C), indicating smaller
   biological differences between these two tissues. Further analysis
   revealed that the metabolites between phl and xyl show higher
   significance, and the change magnitude between cor and phl is greater,
   suggesting more significant metabolite variations between phl and xyl
   and a larger magnitude of changes between cor and phl.

   A total of 525 differential metabolites were identified between the cor
   and phl groups in the PLP root, with 203 metabolites upregulated and
   322 metabolites downregulated. Between cor and xyl, 452 differential
   metabolites were identified, with 220 upregulated and 232
   downregulated. Between phl and xyl, 328 differential metabolites were
   identified, with 200 upregulated and 128 downregulated ([70]Figure 2D).

2.3. Analysis of the Top 50 Differential Metabolites

   To further analyze the expression differences in metabolites among
   different samples, hierarchical clustering was conducted on all
   significantly differential metabolites. Heatmaps and correlation
   analysis plots for the top 50 differential metabolites were created
   using Origin 2022 software based on VIP values and Pearson
   product–moment correlation coefficients ([71]Figure 3 and [72]Figure
   4). Details of the top 50 differential metabolites can be found in
   [73]Supplementary Material Table S2.

Figure 3.

   [74]Figure 3
   [75]Open in a new tab

   Heatmap of the top 50 differential metabolites among the cor, phl, and
   xyl groups. (A): cor-phl; (B): cor-xyl; (C): phl-xyl. The horizontal
   axis represents the sample names, and the vertical axis represents the
   top 50 differential metabolites. The color gradient from green to red
   indicates the expression abundance of metabolites from low to high.

Figure 4.

   [76]Figure 4
   [77]Open in a new tab

   Correlation analysis of top 50 differential metabolites among the cor,
   phl, and xyl groups. (A): cor-phl; (B): cor-xyl; (C): phl-xyl. The
   correlation coefficient ranges from −1 to +1, with close to +1 (red)
   indicating a strong positive correlation, close to −1 (blue) indicating
   a strong negative correlation, and close to 0 indicating no significant
   correlation.

   Through categorization and comparison of the top 50 differential
   metabolites, 678 pairs of metabolites between cor and phl showed a
   significant positive correlation, while 423 pairs exhibited a
   significant negative correlation. Between cor and xyl, there were 648
   pairs with a significant positive correlation and 600 pairs with a
   significant negative correlation. Between phl and xyl, 608 pairs were
   significantly positively correlated, and 586 pairs were negatively
   correlated.

   The study identified a diverse range of compound categories, including
   monoterpenoid glycosides, tannins, flavonoids and their glycosides,
   coumarins, organic acids, amino acids, nucleosides, sugars and their
   derivatives, and nitrogen-containing heterocyclic compounds.
   Monoterpenoid glycosides, particularly paeoniflorin and albiflorin,
   showed consistent positive correlations across cor, phl, and xyl,
   indicating their stable biological roles in the different tissues of
   peony roots, such as anti-inflammatory, anti-tumor, and immune
   regulatory functions [[78]16,[79]17,[80]18]. The high abundance of
   flavonoids and their glycosides in phl reflects the tissue’s activity
   in antioxidation and environmental stress resistance [[81]19,[82]20].
   The low abundance of coumarin compounds in cor suggests their potential
   specific roles in the regulation of cork layer growth. Furthermore, the
   higher abundance of fatty acids in phl and the highest abundance of
   Ubiquinol 8 in cor indicate cor’s potential function in maintaining
   cell membrane stability and energy transfer [[83]21,[84]22]. Overall,
   the levels of primary metabolites (such as amino acids, nucleosides,
   and sugars) and secondary metabolites (such as monoterpenoid
   glycosides, flavonoids, coumarins, etc.) were generally higher in phl
   than in cor, suggesting phl’s advantage in energy metabolism and the
   synthesis of secondary metabolites.

   However, the metabolic differences between phl and xyl showed less
   regularity, hinting at potential complex metabolic regulatory
   mechanisms. Further research into these metabolic differences and their
   roles in plant physiology and adaptability will provide a deeper
   understanding of the biological functions of peony roots and their
   mechanisms in responding to environmental changes.

2.4. Enrichment Analysis of Metabolic Pathways

   Enrichment analysis of metabolic pathways was performed using the Kyoto
   Encyclopedia of Genes and Genomes (KEGG) IDs of differential compounds
   to obtain the results of pathway enrichment. A p-value threshold of
   ≤0.05 was established for selecting the top 10 significantly enriched
   pathways among the differential metabolites ([85]Figure 5).
   Subsequently, a bubble chart was constructed to graphically represent
   these pathways ([86]Figure 6).

Figure 5.

   [87]Figure 5
   [88]Open in a new tab

   Enrichment graphs of the top 10 metabolic pathways among the cor, phl,
   and xyl groups. (A): cor-phl; (B): cor-xyl; (C): phl-xyl. The red line
   indicates a p-value of 0.01, and the blue line indicates a p-value of
   0.05. When the top of a bar exceeds the blue line, the corresponding
   pathway is significantly enriched.

Figure 6.

   [89]Figure 6
   [90]Open in a new tab

   Bubble charts of the top 20 among the cor, phl, and xyl groups. (A):
   cor-phl; (B): cor-xyl; (C): phl-xyl. The vertical axis represents the
   metabolic pathways, and the horizontal axis represents the enrichment
   factor. A larger Rich factor indicates greater enrichment. The color
   gradient from red to green indicates decreasing p-values; larger
   bubbles indicate a greater number of compounds enriched in the pathway.

   The results showed that linoleic acid metabolism, galactose metabolism,
   and ABC transporters were the primary metabolic pathways contributing
   to the differences between cor and phl in PLP. Arginine biosynthesis
   and flavonoid biosynthesis were the main metabolic pathways
   differentiating phl and xyl. It was noted that significant differential
   metabolites included L-Arginine, a product of the arginine biosynthesis
   pathway, and Quercetin 3-(2′′′-feruloylsophoroside), Kaempferol
   3-(2-(E)-p-coumarylrhamnoside), and 6-Methoxykaempferol
   3-rhamnoside-7-(4′′′-acetylrhamnoside), which are products of the
   flavonoid biosynthesis pathway. These metabolic pathways collectively
   contribute to the distinctiveness among different parts within PLP.

   Linoleic acid, an essential fatty acid in humans, primarily functions
   in cell membrane synthesis and inflammation regulation [[91]22,[92]23].
   Previous studies have reported that linoleic acid can protect the
   liver, improve liver fibrosis [[93]24,[94]25], and offer cardiac
   protection in rats with chronic heart failure by modulating its
   metabolism [[95]26]. It also alleviates lipopolysaccharide-induced
   acute inflammation and multi-organ damage [[96]23]. Galactose, a vital
   carbohydrate in cell metabolism, can be directly absorbed by target
   tissues, promoting energy production, glycosylation, and other
   metabolic functions [[97]27]. Due to its chemical properties that allow
   for transmembrane diffusion, galactose is clinically used to generate
   glycoconjugates for improved drug absorption, as successfully
   demonstrated in supplementing dopamine for Parkinson’s patients
   [[98]28]. Regulating galactose metabolism has been shown to alleviate
   acute liver damage and improve some complications of endotoxemia
   [[99]29,[100]30]. Earlier studies indicated that fatty acid and sugar
   metabolites are more abundant in phl, suggesting that PLP might
   regulate metabolic pathways through phl, thus exerting hepatoprotective
   and menstrual regulation effects.

   L-Arginine, a precursor to proteins and NO, is known as a “vascular
   scavenger”, acting as a vasodilator and an endogenous
   anti-atherosclerosis molecule in the cardiovascular system
   [[101]31,[102]32]. Its significant metabolic roles have been confirmed,
   as L-Arginine positively influences energy metabolism through various
   physiological and metabolic mechanisms and can orally delay or prevent
   progressive renal failure in mice [[103]33]. Flavonoids, as secondary
   metabolites widely present in plants, play crucial roles in plant
   growth and resistance to adverse stress [[104]20]. Research shows that
   stress conditions enhance the expression of flavonoid synthase genes,
   leading to flavonoid accumulation. These compounds are transported to
   organelles through various pathways, adjusting plant growth patterns to
   adapt to the environment [[105]34]. Additionally, the hydroxyl groups
   in flavonoid structures confer strong antioxidant properties [[106]35].
   Flavonoids released from roots can also act as signaling molecules,
   promoting the colonization and growth of root microorganisms, thereby
   strengthening plant–microbe symbiosis against environmental stress
   [[107]36,[108]37]. Cor, phl, and xyl may modulate metabolic pathways
   through changing metabolites, thereby enabling PLP to exert various
   pharmacological effects, such as the anticoagulant action of PRR in a
   rat model of acute blood stasis. Further study into the changes of
   metabolic biomarkers in PLP and their impact on metabolic pathways is
   anticipated to provide references for discovering new therapeutic