Identification of Rubiae Radix et Rhizoma and Its Adulterants Based on “Mass Spectrometry Matrix”

ABSTRACT

Galium elegans Wall. in Roxb. (GWR) and Rubia tibetica Hook. f. (RTH) are common adulterants of rubiae radix et rhizoma (RRR). To strengthen RRR adulteration regulation, this paper conducted the adulteration analysis of RRR according to the “mass spectrometry (MS) matrix”. Ultra‐high‐performance liquid chromatography‐quadrupole time‐of‐flight‐full information tandem MS expression analyzed the GWR, RTH, and RRR to obtain quantized data. The ionic intersections were extracted from different batches of the same herbs as their “ionic representation”, respectively. Then, the ionic matrix of each herb relative to each other was screened out, and the top‐n ions were regarded as the “MS matrix” of the three herbs. The “MS matrix” was used for matching test samples to obtain a credibility index. On the other hand, chemometrics was used to validate. RRR adulteration can be realized according to the “MS matrix”. Note that, 3% adulterants can still be identified and three market samples were identified as adulterants. Further research proved that the adulteration identification based on the “MS matrix” of RRR, RTH and GWR is reasonable and reliable. The RRR adulteration analysis can be realized according to the “MS matrix”. The “MS matrix” has certain proprietary and has an important reference for RRR adulteration regulation.

Abbreviations

CI

credibility index

DM

digital matrix

EMRT

exact mass retention time

ESI

electrospray ionisation

GWR

Galium elegans Wall. in Roxb.

LE

leucine enkephalin

m/z

mass‐to‐charge ratio

MS^E

full information tandem mass spectrometry expression

RMS

market samples of rubiae radix et rhizoma

RRR

rubiae radix et rhizoma

Rt

retention time

RTH

Rubia tibetica Hook. f.

1. Introduction

Rubiae radix et rhizoma (RRR), known as “Qian Cao”, is the dried root and rhizome of Rubia cordifolia L. [1]. It can cool blood, remove blood stasis, and stop bleeding [1, 2]. RTH is the dried root and rhizome of R. tibetica Hook. f. and RTH can clear heat, anti‐histamine, and cool blood [3]. In addition, GWR is the root of the Galium elegans Wall. in Roxb., and it can invigorate blood, antioxidants, and relieve tendons [4]. Although they are all Rubiaceae plants, their efficacy is different. On the other hand, the GWR and RTH are not included in the Chinese Pharmacopoeia. However, they were often confused in the medicinal materials market. Even unscrupulous merchants mixed GWR and RTH into RRR to seek exorbitant illegal benefits due to their similar features such as nodular and cylindrical roots, dark brown surface, fine longitudinal wrinkles, more xylem, and cork cell multi‐row [5]. Therefore, it is necessary to establish effective identification methods.

Based on traits and microscopic characteristics, Guo et al. used 3D polarized light techniques to identify RRR and its confused species. The results showed that RRR and its confused species could be distinguished by their surface roughness, the morphology of cork layer cells, and the density of needle crystal bundles [6]. Li et al. determined the contents of alizarin, lucid, xanthin, purpurin, 6‐hydroxyrubiadin, chrysophanol, and mulligan in RRR and its confused species. They achieved effective differentiation based on multi‐component chemometrics analyses [7]. Chen et al. established the DNA barcode identification system, and the results showed that the combined sequence can successfully identify RRR and its adulterants. However, the identification effect on the species belonging to the same genus is not ideal [8]. These studies favor RRR adulteration regulation. But there are some shortcomings: (1) Traits and microscopic characteristics need professional knowledge reserves; (2) In the analysis of DNA bar code, DNA is easily destroyed or even cannot be extracted, and the identification effect of adulterants belonging to the same species is not ideal; (3) In chemical analysis, only a single or a few components were used as quality control marker.

The proposal of a “mass spectrometry (MS) matrix” facilitated the adulteration identification of Chinese medicines to a certain extent [9]. In simple terms, an “MS matrix” is defined as “a collection of ions in a digital form that uniquely identifies a specific herb.” On the other hand, given the deficiency of current identification, successful applications of “digital identity”, and considering that chemical compositions tend to have better transferability and stability in sample pretreatment, this study established the “MS matrix” of RRR, GWR, and RTH for identifying RRR and its adulterants to enrich the identification methods of RR and strengthen its market regulation and quality control [10, 11].

2. Experimental Materials and Analytical Methods

2.1. Herbs and Chemical Agents

Ten RRR, eight RTH and six GWR were collected from different places in China. Fourteen positively adulterated samples including 0% RTH, 3% RTH, 5% RTH, 10% RTH, 20% RTH, 50% RTH, 100% RTH and 0% GWR, 3% GWR, 5% GWR, 10% GWR, 20% GWR, 50% GWR and 100% GWR [9, 12]. Twenty RRRs were purchased at the market. Table S1 shows the herbal information.

The Milli‐Q Advantage A10 ultrapure water system prepared deionized water. Thermo Honeywell Trading Co., Ltd provided the methanol (Lot: E1058‐US), formic acid (Lot: L1670) and Acetonitrile (Lot: Y5BA1H).

2.2. Analytical Conditions

MS adopted positive electrospray ionisation (ESI⁺), negative ESI (ESI⁻) and full information tandem MS expression (MS^E) mode. m/z: 50–1200 Da; Scan time: 0.1 s; Source and desolvation temperature were 120°C and 400°C; Desolvation and cone gas were 800 and 50 L/h; Capillary and Cracking: 3.0 kV and 10–40 V [12, 13]. The MS resolution was set to 3.5 × 10⁴. Sodium formate and leucine enkephalin (LE, 556.28, [M+H]⁺; LE, 554.26, [M‐H]⁻; quality control marker) were used for mass‐axis and lock spray corrections, respectively, prior to sample analysis, and real‐time mass‐axis corrections were performed during the sample analysis process using LE. On the other hand, we used the ultra‐high‐performance liquid chromatography (UHPLC) BEH‐C₁₈ column for chromatographic separations. Column temperature: 35°C; mobile phase velocity: 0.3 mL/min; sample load: 2 µL; Table 1 shows analytical mobile phase and gradient elution conditions.

TABLE 1.

The gradient elution conditions of mobile phases.

Time (min)	0.1% Formic acid water (%)	Acetonitrile (%)	Flow (L/min)
0.00	95	5	0.3
23.00	5	95	0.3
26.00	5	95	0.3
26.10	95	5	0.3
30.00	95	5	0.3

2.3. Sample Preparation

The herbs were crushed and passed through a number three sieve (50 mesh; pore size: 355 ± 13 µm). Then mixed RRR powder of each batch to make an RRR mixed sample. In contrast, mixed GWR powder and RTH powder of each batch to make the GWR and RTH mixed samples, respectively. The RRR mixed sample was mixed with different proportions of RTH or GWR mixed powder (0%, 3%, 5%, 10%, 20% and 100%) to prepare positive adulterants [11]. Accurately weighed 0.5 g of RRR, RTH, GWR, positively adulterated samples and market herbs in an Erlenmeyer flask. Then, 25 mL methanol was added and sonicated for 30 min (power: 450 W, frequency: 40 kHz) [12]. Finally, filtered with a 0.22 µm filter membrane to obtain samples [13].

2.4. Algorithms Principles

Version 2.4.69 of the QI software was used to process MS of test samples and blank solvent (methanol) [12]. At the same time, the parameters were set as follows: ESI+ or ESI‐; Peak limit: automatic; Rt limits: 3.00–23.00 min. Further, saved the “ionic matrix” in the following format [9, 11, 12]:

$$\mathrm{DM}=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_n& m_n& i_n\end{array}\right]$$

The algorithm formula includes the following steps [9, 10, 11, 12, 13]:

(1) Blank deduction: if the Rt and m/z of ions in samples and methanol satisfied ΔRt ≤ 0.10 min and Δm/z ≤ 0.01 Da, these ions will be removed from the sample's “ionic matrix”.

If the “ionic matrix” of RRR ₁, RRR ₂ and RRR ₃ (RRR samples from different batches) are recorded as r ₁, r ₂ and r ₃; The “ionic matrix” of GWR ₁, GWR ₂ and GWR ₃ (GWR samples from different batches) are recorded as g ₁, g ₂ and g ₃, respectively; The “ionic matrix” of RTH ₁, RTH ₂ and RTH ₃ (RTH samples from different batches) are recorded as t ₁, t ₂ and t ₃, respectively and the blank “ionic matrix” are recorded as b ₁, b ₂ and b ₃ [14, 15, 16].

$$\begin{array}{ccc}\hfill r_1& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r1}& m_{r1}& i_{r1}\end{array}\right]r_2=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r2}& m_{r2}& i_{r2}\end{array}\right]r_3=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r3}& m_{r3}& i_{r3}\end{array}\right]\hfill \end{array}$$

$$g_1=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g1}& m_{g1}& i_{g1}\end{array}\right]g_2=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g2}& m_{g2}& i_{g2}\end{array}\right]g_3=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g3}& m_{g3}& i_{g3}\end{array}\right]$$

$$b_1=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b1}& m_{b1}& i_{b1}\end{array}\right]b_2=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b2}& m_{b2}& i_{b2}\end{array}\right]b_3=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b3}& m_{b3}& i_{b3}\end{array}\right]$$

$$t_1=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{t1}& m_{t1}& i_{t1}\end{array}\right]t_2=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{t2}& m_{t2}& i_{t2}\end{array}\right]t_3=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{t3}& m_{t3}& i_{t3}\end{array}\right]$$

The new “ionic matrix” of R ₁, R ₂ and R ₃ are obtained by removing background ions from r ₁, r ₂ and r ₃ [16]:

$$\begin{array}{ccc}\hfill R_1=r_1-r_1\cap b_1& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r1}& m_{r1}& i_{r1}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r1}& m_{r1}& i_{r1}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b1}& m_{b1}& i_{b1}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R1}& m_{R1}& i_{R1}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill R_2=r_2-r_2\cap b_2& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r2}& m_{r2}& i_{r2}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r2}& m_{r2}& i_{r2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b2}& m_{b2}& i_{b2}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R2}& m_{R2}& i_{R2}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill R_3=r_3-r_3\cap b_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r3}& m_{r3}& i_{r3}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{r3}& m_{r3}& i_{r3}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b3}& m_{b3}& i_{b3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R3}& m_{R3}& i_{R3}\end{array}\right]\hfill \end{array}$$

For GWR and RTH, the new “ionic matrix” can be obtained by excluding the blank ions. It can be defined as G ₁, G ₂ and G ₃ for GWR and T ₁, T ₂ and T ₃ for RTH:

$$\begin{array}{ccc}\hfill G_1=g_1-g_1\cap b_1& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g1}& m_{g1}& i_{g1}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g1}& m_{g1}& i_{g1}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b1}& m_{b1}& i_{b1}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G1}& m_{G1}& i_{G1}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill G_2=g_2-g_2\cap b_2& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g2}& m_{g2}& i_{g2}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g2}& m_{g2}& i_{g2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b2}& m_{b2}& i_{b2}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G2}& m_{G2}& i_{G2}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill G_3=g_3-g_3\cap b_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g3}& m_{g3}& i_{g3}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g3}& m_{g3}& i_{g3}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b3}& m_{b3}& i_{b3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G3}& m_{G3}& i_{G3}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill T_1=t_1-t_1\cap b_1& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g1}& m_{g1}& i_{g1}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g1}& m_{g1}& i_{g1}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b1}& m_{b1}& i_{b1}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T1}& m_{T1}& i_{T1}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill T_2=t_2-t_2\cap b_2& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g2}& m_{g2}& i_{g2}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g2}& m_{g2}& i_{g2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b2}& m_{b2}& i_{b2}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T2}& m_{T2}& i_{T2}\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill T_3=t_3-t_3\cap b_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g3}& m_{g3}& i_{g3}\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{g3}& m_{g3}& i_{g3}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{b3}& m_{b3}& i_{b3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T3}& m_{T3}& i_{T3}\end{array}\right]\hfill \end{array}$$

(2) Acquisition of shared ions: If the Rt and m/z of ions satisfied ΔRt ≤ 0.10 min and Δm/z ≤ 0.01 Da, Using the above ions construct a new “ionic matrix” of shared ions.

Define R, G and T as the “ionic matrix” of RRR, RTH and GWR shared ions, respectively [16].

$$\begin{array}{ccc}\hfill G=G_1\cap G_2\cap G_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G1}& m_{G1}& i_{G1}\end{array}\right]\cap \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G2}& m_{G2}& i_{G2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G3}& m_{G3}& i_{G3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_G& m_G& i_G\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill R=R_1\cap R_2\cap R_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R1}& m_{R1}& i_{R1}\end{array}\right]\cap \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R2}& m_{R2}& i_{R2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R3}& m_{R3}& i_{R3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_R& m_R& i_R\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill T=T_1\cap T_2\cap T_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T1}& m_{T1}& i_{T1}\end{array}\right]\cap \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T2}& m_{T2}& i_{T2}\end{array}\right]\hfill \\ & \cap & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T3}& m_{T3}& i_{T3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_T& m_T& i_T\end{array}\right]\hfill \end{array}$$

(3) Acquisition of specific ions: the RRR‐specific ions can be obtained by comparing RRR's shared ions to the raw ions of RTH and GWR and removing similar ions. Similarly, the “ionic matrix” of RTH and GWR‐specific ions can be obtained in the same way.

x, y and z represent the “ionic matrix” of specific RRR, RTH and GWR ions. The union of R ₁, R ₂, and R ₃ is defined as x = (R ₁∪R ₂∪R ₃); The union of G ₁, G ₂ and G ₃ is defined as y = (G ₁∪G ₂∪G ₃) and the union of T ₁, T ₂ and T ₃ is defined as z = (T ₁∪T ₂∪T ₃).

$$\begin{array}{ccc}\hfill x=R_1\cup R_2\cup R_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R1}& m_{R1}& i_{R1}\end{array}\right]\cup \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R2}& m_{R2}& i_{R2}\end{array}\right]\hfill \\ & \cup & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{R3}& m_{R3}& i_{R3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_x& m_x& i_x\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill y=G_1\cup G_2\cup G_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G1}& m_{G1}& i_{G1}\end{array}\right]\cup \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G2}& m_{G2}& i_{G2}\end{array}\right]\hfill \\ & \cup & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{G3}& m_{G3}& i_{G3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_y& m_y& i_y\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill z=T_1\cup T_2\cup T_3& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T1}& m_{T1}& i_{T1}\end{array}\right]\cup \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T2}& m_{T2}& i_{T2}\end{array}\right]\hfill \\ & \cup & \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_{T3}& m_{T3}& i_{T3}\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_z& m_z& i_z\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill X=R-y-z& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_R& m_R& i_R\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_y& m_y& i_y\end{array}\right]\hfill \\ & & -\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_z& m_z& i_z\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_X& m_X& i_X\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill Y=G-x-z& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_G& m_G& i_G\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_x& m_x& i_x\end{array}\right]\hfill \\ & & -\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_z& m_z& i_z\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_G& m_G& i_G\end{array}\right]\hfill \end{array}$$

$$\begin{array}{ccc}\hfill Z=T-x-y& =& \left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_T& m_T& i_T\end{array}\right]-\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_x& m_x& i_x\end{array}\right]\hfill \\ & & -\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_y& m_y& i_y\end{array}\right]=\left[\begin{array}{ccc}Rt& m/z& I\\ •••& •••& •••\\ t_G& m_G& i_G\end{array}\right]\hfill \end{array}$$

(4) Acquisition of “MS matrix”: In the X, Y and Z “ionic matrix”, Top‐100 ions were used as the final “MS matrix” for RRR, RTH and GWR. Figure 1 shows the acquisition process.

FIGURE 1.

The acquisition process of "mass spectrometry matrix".

(5) Adulteration analysis: The “MS matrix” of RRR, RTH and GWR matched various test samples to obtain the credibility index (CI). If the Rt and m/z of ions in the “MS matrix” and test samples satisfied ΔRt ≤ 0.10 min and Δm/z ≤ 0.01 Da, These ions are considered a successful match. The CI arithmetic formula is as follows [11, 12, 16]:

$$\mathrm{CI}=\frac{\begin{array}{cccc}number& of& matched& ions\end{array}}{100}\times 100\%$$

3. Results and Discussions

3.1. MS Results

Figure 2 shows base‐peak chromatograms. Blank methanol does not interfere with experimental analysis. As a whole, the base‐peak chromatograms of RRR, RTH and GWR differed. But, it is difficult to identify adulterated samples. Therefore, the “MS matrix” was proposed to identify RRR adulterated samples.

FIGURE 2.

The base‐peak chromatograms of RRR, RTH, GWR, positive samples and market herbs (A) blank; (B) RTH; (C) GWR; (D) RRR; (E) 50% GWR; (F) 50% RTH; (G) marker herb 01; (H) marker herb 07. GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

Before experimental analysis, we optimized the analysis condition. The MS^E acquisition can get more MS data [12, 14, 15]. In MS^E analysis, we examined the different ionic modes. The ionic strength is more significant in the ESI⁺ mode. In terms of collision energy, there were more ions when it is 10–40 V. At the same time, compared to Acetonitrile, 50% methanol‐water, Methanol has the best extraction effect.

3.2. MS Matrix

Eight blank methanol were used to deduct interfering ions. they contain 9485–10,827 [Rt‐m/z‐I]. The average is 10,023 and the RSD is 5.81%. Table 2 shows a number of units in samples constructed “MS matrix”.

TABLE 2.

The number of [Rt‐m/z‐I] in three herbs used to extract "Mass Spectrometry Matrix".

Herbs	Batch	Number	Herbs	Batch	Number	Herbs	Batch	Number
RRR	RRR01	20229	RTH	RTH01	21754	GWR	GWR01	25567
	RRR03	19575		RTH02	22264		GWR02	25237
	RRR04	21288		RTH04	22061		GWR03	25859
	RRR05	20996		RTH05	22150		GWR05	25673
	RRR06	20721		RTH06	22096		GWR06	25230
	RRR07	20340		RTH08	21794		—	—
	RRR08	20603		—	—		—	—
	RRR10	19748		—	—		—	—

Abbreviations: GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

Based on the algorithm principle in 2.4 selection, the “MS matrix” of RRR, RTH and GWR can be obtained. Tables S2–S4 show the “MS matrix” of three herbs. In addition, we explored the deviation thresholds. The Rt was really skewed, but not up to 0.1 min. So, we set ΔRt ≤ 0.10 min. At the same time, we comprehensively compare the “MS matrix” with different Δm/z. The results showed that when Δm/z ≤ 0.01 Da, it ensures sufficient precision as well as allowing us to screen further based on ionic strength [9, 16].

3.3. Identification of Single Herb

The “MS matrix” of RRR, RTH and GWR matched the “ionic matrix” of test samples. Table 3 shows CI results.

TABLE 3.

The matching credibility index results of test samples.

Herbs	Batches	Match ions	Ions in mass spectrometry matrix of RRR, RTH or GWR	CI (%)
RRR	RRR02	96	100‐RRR	96.0
	RRR09	96	100‐RRR	96.0
	RRR02	2	100‐RTH	2.00
	RRR09	1	100‐RTH	1.00
	RRR02	0	100‐GWR	0.00
	RRR09	2	100‐GWR	2.00
RTH	RTH03	93	100‐RTH	93.0
	RTH07	86	100‐RTH	86.0
	RTH03	5	100‐RRR	5.00
	RTH07	7	100‐RRR	7.00
	RTH03	2	100‐GWR	2.00
	RT07	2	100‐GWR	2.00
GWR	GE04	98	100‐GWR	98.0
	GE04	2	100‐RRR	2.00
	GE04	0	100‐RTH	0.00

Abbreviations: CI, credibility index; GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

The CIs of herbs were grouped compared to their own “MS matrix” (Group I). The CIs of herbs compared to non‐self “MS matrix” were grouped (Group II). Since the data were not normally distributed (p = 0.025 < 0.05), non‐parametric tests were used for statistical analysis. Table 4 shows the non‐parametric test results. p = 0.002 < 0.01 indicates a significant difference between the two groups of CIs. It also shows that the “MS matrix” of RRR, RTH, and GWR are somewhat proprietary, enabling the identification of three herbs.

TABLE 4.

The results of non‐parametric test.

Class	Median (P25, P75)		Mann Whitney‐U	Mann Whitney‐Z	p
	Group I (n = 5)	Group II (n = 10)
CI	96.000 (89.5, 97.0)	2.000 (0.8, 2.8)	0	−3.124	0.002**

*p < 0.05; **p <0.01.

Abbreviation: CI, credibility index.

During the analysis, we explored the matching situation when the “MS matrix” contains different ions. Table 5 shows CI results. There was no significant difference in CIs for RRR and GWR when the “MS matrix” contained 100, 200 and 300 ions, respectively. However, as the number of ions increases, RTH's CI decreases. So, to address the RTH match situation, top‐100 ions an “MS matrix” uniformly.

TABLE 5.

The matching results when the "mass spectrometry matrix" contains 100, 200 and 300 ions.

Herbs	Batch	100 ions	200 ions	300 ions
RRR	RRR02	96%	92%	95%
RTH	RTH03	93%	88%	84%
GWR	GWR04	98%	99%	99%

Abbreviations: GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

3.4. Adulteration Analysis

3.4.1. Analysis of Positive Adulterants

The “MS matrix” of RRR, RTH and GWR were used to match adulterated samples. Table 6 shows the CIs of positively adulterated samples of RRR and RTH compared with the “MS matrix” of three herbs.

TABLE 6.

The credibility index (CI) results of positively adulterated samples of RRR and RTH.

Positively adulterated samples	"Mass spectrometry matrix" of RTH	"Mass spectrometry matrix" of RRR	"Mass spectrometry matrix" of GWR
0% RTH	0.00%	98.0%	0.00%
3% RTH	22.0%	60.0%	2.00%
5% RTH	24.0%	62.0%	2.00%
10% RTH	32.0%	54.0%	1.00%
20% RTH	45.0%	54.0%	2.00%
50% RTH	57.0%	50.0%	0.00%
100% RTH	86.0%	7.00%	2.00%

Abbreviations: GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

From 3% RTH to 100% RTH, there is a gradual upward trend in the CIs in matching positively adulterated samples to the “MS matrix” of RTH, and the CI of 3% RTH is the smallest and is not less than 22.0%. At the same time, the CI of 0% RTH is only 0.00%. On the other hand, from 0% RTH to 50% RTH, the CIs are not less than 50.0%. But the CI of 100% RTH is only 7.00%. In other words, the CI of 50% RTH is more than 7.00 times that of 100% RTH, suggesting that RTH adulteration analysis based on “MS matrix” and CI can be achieved without affecting RRR identification. In addition, the CIs of GWR are not higher than 2.00%.

Table 7 shows the CIs of positively adulterated samples of RRR and GWR compared with the “MS matrix” of three herbs. From 0% GWR to 100% GWR, there is also a gradual upward trend in the CIs in matching positively adulterated samples to the “MS matrix” of GWR, and the CI of 3% GWR is the smallest and is not less than 30.0%. At the same time, the CI of 0% GWR is only 0.00%. From 0% GWR to 50% GWR, the CIs of matching positively adulterated samples to the “MS matrix” of RRR are higher than 53.0%. However, the CI of 100% GWR is only 0.00%. In addition, the CIs in matching positively adulterated samples to the “MS matrix” of RTH are not higher than 16.0%.

TABLE 7.

The credibility index (CI) results of positively adulterated samples of RRR and GWR.

Positively adulterated samples	"Mass spectrometry matrix" of GWR	"Mass spectrometry matrix" of RRR	"Mass spectrometry matrix" of RTH
0% GWR	0.00%	100%	0.00%
3% GWR	30.0%	92.0%	8.00%
5% GWR	44.0%	92.0%	14.0%
10% GWR	57.0%	79.0%	16.0%
20% GWR	78.0%	65.0%	14.0%
50% GWR	79.0%	54.0%	13.0%
100% GWR	100%	0.00%	0.00%

Abbreviations: GWR, Galium elegans Wall. in Roxb.; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

According to the “MS matrix” and CI, it can realize adulteration analysis of positively adulterated samples. Further, combined with Tables 6 and 7, considering that herbs can have 3% impurities, data bias, and data fluctuations, we initially presume the limit of RTH's CI detection to 25.0%. If the CI > 25.0% when compared with RTH's “MS matrix”, the sample is adulterated with RTH. Similarly, we initially presume the limit of GWR's CI detection to 35.0%.

3.4.2. Analysis of RRR Market Samples

Twenty batches of RRR market samples (RMS) were analyzed according to the “MS matrix”. Table 8 shows CI results. The CIs matching RMS09 and RMS10 samples to RTH's “MS matrix” are 82.0% and 62.0%, respectively, which is much greater than RTH's detection threshold of 25.0%. It indicated that RTH is mixed into RMS09 and RMS10 samples. When compared with RTH's “MS matrix”, the CIs of the remaining samples are all less than 25.0%, and the CIs of matching all the RMS to RRR's “MS matrix” are all higher than 51.0%, which indicated that these market samples are RRR and are not doped with RTH. On the other hand, for market samples except for RMS17, their CIs are all less than GWR's detection threshold of 35.0% when compared with GWR's “MS matrix”, indicating that these market samples are not doped with GWR. However, the MCs matching RMS17 samples to GWR's “MS matrix” is 81.0%, much more significant than GWR's detection threshold of 35.0%. It indicated that GWR is mixed into RMS17 samples.

TABLE 8.

The credibility index (CI) results of RRR market samples.

RR market samples	"Mass spectrometry matrix" of RRR	"Mass spectrometry matrix" of RTH	"Mass spectrometry matrix" of GWR
RMS01	60.0%	18.0%	0.00%
RMS02	57.0%	17.0%	0.00%
RMS03	62.0%	12.0%	2.00%
RMS04	59.0%	12.0%	3.00%
RMS05	64.0%	13.0%	2.00%
RMS06	71.0%	15.0%	1.00%
RMS07	64.0%	14.0%	1.00%
RMS08	71.0%	12.0%	1.00%
RMS09	68.0%	82.0%	19.0%
RMS10	69.0%	62.0%	7.00%
RMS11	63.0%	13.0%	0.00%
RMS12	52.0%	15.0%	0.00%
RMS13	88.0%	14.0%	4.00%
RMS14	83.0%	15.0%	4.00%
RMS15	79.0%	19.0%	4.00%
RMS16	58.0%	17.0%	2.00%
RMS17	67.0%	18.0%	81.0%
RMS18	57.0%	17.0%	2.00%
RMS19	65.0%	14.0%	2.00%
RMS20	59.0%	12.0%	3.00%

Abbreviations: GWR, Galium elegans Wall. in Roxb.; RMS: market samples of rubiae radix et rhizoma; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

3.4.3. Validation of Adulterant Identification

Adulteration identification based on the “MS matrix” was verified by performing chemometric analysis to extract differential ions. The quantized data for 10 batches of RRRs and eight batches of RTHs have been imported into SCIMA 14.1 software. Rt‐m/z and I were used as the independent variable and the dependent variable to explore the RTH and GWR's differential ions [17, 18, 19]. Finally, the adulteration identification based on the “MS matrix” was verified by determining whether the RTH's proprietary ions can be extracted in market RRs. Figure 3 shows the chemometric results. Figure 3A shows that the RRR and RTH can be distinguished with R2X = 0.71 and Q2 = 0.67. Further, OPLS‐DA and S‐Plots analysis explored RTH's proprietary ions. Figure 3B shows that the RRR and RTH can be distinguished with Q2 = 0.99. Figure 3C results indicate that the OPLS‐DA model is not overfitting [20]. In the S‐Plots (Figure 3D), the data points at the two ends are the potential differential components. The red point is chemical component A (Rt 21.89 min_m/z 609.27). Similarly, as shown in Figure 4, we can get GWR's proprietary ion‐chemical component B (Rt 7.42 min_m/z 977.41).

FIGURE 3.

The results of chemometric analysis of RRR and RTH: (A) the results of PCA analysis; (B) the results of OPLS‐DA analysis; (C) the results of model validation; (D) the results of S‐Plots. OPLS‐DA, Orthogonal Partial Least Squares Discriminant Analysis; PCA, Principal component analysis; RRR, rubiae radix et rhizoma; RTH, Rubia tibetica Hook. f.

FIGURE 4.

The results of chemometric analysis of RRR and GWR: (A) the results of PCA analysis; (B) the results of OPLS‐DA analysis; (C) the results of model validation; (D) the results of S‐Plots. GWR, Galium elegans Wall. in Roxb.; OPLS‐DA, Orthogonal Partial Least Squares Discriminant Analysis; PCA, Principal component analysis; RRR, rubiae radix et rhizoma.

Compounds A and B are proprietary chemical markers of RTH and GWR. Figure S1 shows the extraction results of compound A. A can be extracted from the RTH samples, 3% positively adulterated samples of RTH and RRR, RMS09 and RMS10 samples, and their ionic strengths are higher than 1.5 × 10⁶. However, it can not be detected in the RRR, 0% RTH, and remaining RMS samples. Similarly, as shown in Figure S2, the B can be extracted from the GWR samples, 3% positively adulterated samples of GWR and RRR, and the RMS17 market sample and their ionic strengths are higher than 2.0 × 10⁶. However, it can not detected in the RRR samples, 0% GWR sample, and remaining RMS samples. The above results indicate that the RMS09 and RMS10 market samples are adulterated with RTH, and the RMS 17 market sample is adulterated with GWR. So the validation analysis based on chemometrics proves that the “MS matrix” is reasonable. Unfortunately, we could not determine the chemical structures of differential components. It does not matter and we use it as an auxiliary method to verify the adulterant identification.

3.5. Discussion of “MS Matrix” Authentication

The quality control of traditional Chinese medicine (TCM) is also developing towards digitalization [21, 22, 23]. This study established the “MS matrix” of RRR and its adulterants (RTH and GWR) to realize the adulteration identification analysis efficiently and rapidly. The “MS matrix” has broad applicability and can be dynamically adjusted by adjusting the thresholds of Rt, m/z and the number of ion outputs to adapt to different analytical scenarios. In addition, compared to previous studies [9, 16], previous analytical methods have only considered the shared ions of multiple batches of the same herbal medicine as the “MS matrix”. However, herbs in the same family and genus often contain many same chemical components, which makes identification difficult if only the shared ions are considered! Therefore, based on shared ions in this study, the different chemical compositions between herbs of the same family and genus were also considered to construct an “MS matrix”. The results show that the identification effect can be significantly improved. On the other hand, This study also adds the validation analysis of the reasonable reliability of “MS matrix” based identification by chemometric analysis. The RTH and GWR's proprietary ions were identified for mass spectrometry information extraction by chemometric analysis. At the same time, using mixed positive samples as quality control samples, even with only 3% adulteration, the respective proprietary ions can still be extracted, whereas the proprietary ions can not be extracted in the unadulterated RRR samples, which can prove that adulterant identification based on “MS matrix” is reasonable.

In addition, the two proprietary ions are contained in their respective “MS matrix”. For the adulteration analysis of RRR, exploring differential chemical compositions based on chemometrics analysis alone indeed enables adulteration analysis of RRR. However, as the number of herbal species increases, differential chemical composition will likely not be captured based on chemometrics. However, the “MS matrix” will remain proprietary, just as the differences between ABC, ACB and BCA are. “MS matrix” is a collection of ions in a digital form uniquely identifying a herb. It is not limited to a single chemical component but is based on an arrangement and combination of multiple chemical components. The arrangement and combination of multiple chemical components make Chinese medicine's “MS matrix” more proprietary. It allows us to establish the proprietary quantized characterization of different herbs. The “MS matrix” no longer focuses on identifying and analyzing the structure of compounds but instead realizes adulteration identification and analysis from the perspective of multi‐chemical components quantized characterization and realizes the effective use of unknown components. However, the “MS matrix” also has limitations, which are not conducive to elucidating the material basis and proprietary chemical composition analysis of Chinese medicine.

This study belongs to the category of non‐targeted identification, so it is necessary to control the experimental analysis conditions strictly. Under the uniform sample pretreatment and analytical conditions, the concern is with differences in results. As we all know, from the perspective of differences, the chemical compositions of TCM in different places of origin and periods have variability, and even the mass spectrometry information collected for the same TCM in different periods is also different [9, 16]. However, this will not interfere with the “MS matrix” significantly. First, even if the origin and harvesting period are different, the same chemical composition exists in the same Chinese medicine, and the “MS matrix” focuses on these shared chemical compositions at the beginning, followed by the different chemical compositions of different herbs. For the detection deviation caused by the influence of ionization and temperature during detection, we can correct the mass axis with leucine enkephalin (LE) in real‐time. In addition, we can control the bias of Rt and m/z. It is worth noting that the “MS matrix” is not static but can be obtained dynamically by adjusting the Rt and m/z deviation thresholds and the number of ion outputs, which guarantees a wide range of applicability and improves tolerance of deviation because of fluctuations in instrument parameters.

It is undeniable that this study has some shortcomings. Although the herbal materials come from different producing areas and harvesting periods and meet the requirements of the relevant varieties, The sample size is limited and needs to be increased for in‐depth exploration. In addition, we started with the matching bias to improve the analytical accuracy and fault tolerance. However, we still need to explore from the perspective of system applicability in the follow‐up study, such as the small‐scale fluctuation of instrumental parameters. On the other hand, matching identification based on “MS matrix” combined with chemometrics analysis can be developed into a scientific methodology.

4. Conclusion

Based on UHPLC‐quadrupole time‐of‐flight‐MS^E analysis, this study established the “MS matrix” of RRR and its adulterants (RTH and GWR), to realize the adulteration identification analysis with CI ≥86.00%, even if only 3% adulterants can still be identified. The market samples of RMS09, RMS10 and RMS17 were adulterated samples. In addition, the reliability of the “MS matrix” was verified through chemometric analysis. It can provide a reference for establishing the “digital identity” of Rubiae radix et rhizoma and adulteration analysis.

Authors Contributions

Xianrui Wang: writing‐original draft, methodology, validation and software. Minghua Li: writing‐review & editing, data curation, methodology and writing‐original draft. Yu Zhang: validation, visualization, data curation, methodology and writing‐original draft. Jiating Zhang: data curation, investigation, formal analysis, software and writing‐original draft. Wenguang Jing: investigation, validation, data curation and software. Xiaohan Guo: conceptualization, methodology, supervision and validation. Xianlong Cheng: funding acquisition, supervision, writing–review & editing and project administration. Feng Wei: data curation, funding acquisition, supervision and software.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information

Data Availability Statement

The data are available on request from the authors.