Integrated Chemical Characterization, Network Pharmacology and Transcriptomics to Explore the Mechanism of Sesquiterpenoids Isolated from Gynura divaricata (L.) DC. against Chronic Myelogenous Leukemia

Chronic myelogenous leukemia (CML) is a serious threat to human health, while drugs for CML are limited. Herbal medicines with structural diversity, low toxicity and low drug resistance are always the most important source for drug discoveries. Gynura divaricata (L.) DC. is a well-known herbal medicine whose non-alkaline ingredients (GD-NAIs) were isolated. The GD-NAIs demonstrated potential anti-CML activity in our preliminary screening tests. However, the chemical components and underlying mechanism are still unknown. In this study, GD-NAIs were tentatively characterized using UHPLC-HRMS combined with molecular networking, which were composed of 75 sesquiterpenoids. Then, the anti-CML activities of GD-NAIs were evaluated and demonstrated significant suppression of proliferation and promotion of apoptosis in K562 cells. Furthermore, the mechanism of GD-NAIs against CML were elucidated using network pharmacology combined with RNA sequencing. Four sesquiterpenoids would be the main active ingredients of GD-NAIs against CML, which could regulate PD-L1 expression and the PD-1 checkpoint pathway in cancer, PI3K/AKT, JAK/STAT, TGF-β, estrogen, Notch and Wnt signaling pathways. In conclusion, our study reveals the composition of GD-NAIs, confirms its anti-CML activity and elucidates their underlying mechanism, which is a potential countermeasure for the treatment of CML.


Introduction
Chronic myelogenous leukemia (CML) is a rare malignant proliferative hematopoietic disease [1]. The molecular hallmark of CML is the Philadelphia chromosome, which involves the fusion of the v-abl Abelson murine leukemia viral oncogene homologue 1 (ABL1) gene on chromosome 9 with the breakpoint cluster region (BCR) gene on chromosome 22 [1]. The BCR-ABL1 oncoprotein induces the activation of tyrosine kinase (TK), stimulates multiple signaling pathways and alters the expression of genes/molecules, which ultimately induce CML [2,3]. These pathways include Janus kinase/signal transducer and activator of transcription (JAK/STAT), Hedgehog (Hh), transforming growth clustered network indicated that these nodes (compounds) may be analogs of rupestonic acid that share similar fragmentation patterns. The fragmentation pattern of rupestonic acid was carefully studied and summarized. Thereafter, the connected compounds were tentatively characterized as a cluster of sesquiterpenoids, and a total of 75 compounds were tentatively characterized, of which 61 sesquiterpenoids were reported for the first time (Table 1 and Figure 2). by analyzing the fragmentation patterns of adjacent nodes, which facilitates the simultaneous characterization of abundant and trace analogs [26].
Driven by advances in MN, UHPLC-HRMS combined with MN was used to tentatively characterize GD-NAIs. First, the MN of GD-NAIs was established ( Figure 1). Red nodes represent compounds that were identified by GNPS library search, while green nodes represent unknown compounds. The results showed that the only red node was identified as a sesquiterpenoid rupestonic acid (compound 1), yet others remain unknown. The clustered network indicated that these nodes (compounds) may be analogs of rupestonic acid that share similar fragmentation patterns. The fragmentation pattern of rupestonic acid was carefully studied and summarized. Thereafter, the connected compounds were tentatively characterized as a cluster of sesquiterpenoids, and a total of 75 compounds were tentatively characterized, of which 61 sesquiterpenoids were reported for the first time (Table 1 and Figure 2).  (GD-NAIs). The red node was identified as rupestonic acid (compound 1) by GNPS library search, and the green nodes remain unknown. Table 1. Detailed information on tentatively characterized sesquiterpenoids within Gynura divaricata (L.) DC.

Compound
Molecular Formula

The Effect of GD-NAIs on Suppressing CML Cell Proliferation
First, the cytotoxicity of GD-NAIs on L02 normal liver cells was evaluated by MTT assay. The results showed that the IC 50 value of GD-NAIs was more than 1 mg/mL. Then, the effect of GD-NAIs at safe concentrations on K562 cells was assessed. The results demonstrated that GD-NAIs (30, 60, 120, 250, 500 and 1000 µg/mL) significantly inhibited the viability of K562 cells ( Figure 3A). The inhibitory effect of GD-NAIs was also assessed in the acute erythroid leukemia (AEL) cell line HEL. The results demonstrated that GD-NAIs (15,30,60,120,250, 500 and 1000 µg/mL) were able to inhibit the viability of HEL cells ( Figure 3B). As a positive drug for CML, cytarabine (20 nM) showed an obvious inhibitory effect on the viability of K562 and HEL cells ( Figure 3A,B). Furthermore, the effect of GD-NAIs and GD-E on the proliferation of K562 cells was measured by CCK-8 assay.
The results indicated that GD-NAIs and GD-E (50, 100 and 200 µg/mL) conspicuously inhibited the proliferation of K562 cells in a concentration-dependent manner ( Figure 3C). The inhibitory ability of GD-NAIs on the proliferation of K562 cells at a concentration of 200 µg/mL was much stronger than that of cytarabine (20 nM) ( Figure 3C). Besides, the inhibitory ability of GD-NAIs (100 and 200 µg/mL) was stronger than that of GD-E ( Figure 3C). These results indicated that GD-NAIs are able to suppress the proliferation of K562 cells in a concentration-dependent manner.

The Effect of GD-NAIs on Inducing Cell Apoptosis
One of the hallmarks of CML is the deregulation of apoptosis [39,40]. Therefore, the effect of GD-NAIs on the apoptosis of K562 cells was verified. The cell morphology was observed and showed that the number of K562 cells in the GD-NAIs (50, 100 and 200 µg/mL)-treated and cytarabine (20 nM)-treated groups was markedly less than that of the control group ( Figure 3D). GD-NAIs (50, 100 and 200 µg/mL) and cytarabine (20 nM) caused bursts and the death of K562 cells ( Figure 3D). The Hoechst 33258 assay further demonstrated that many bright blue cells appeared in the GD-NAIs (50, 100 and 200 µg/mL)-treated and cytarabine (20 nM)-treated groups but seldom in the control group ( Figure 3E), indicating that GD-NAIs could induce apoptosis of K562 cells. In addition, Annexin V-FITC/PI staining showed that the percentage of apoptotic cells in the GD-NAIs (50, 100 and 200 µg/mL)-, GD-E (50, 100 and 200 µg/mL)-treated and cytarabine (20 nM)treated groups was much higher than that in the control group ( Figure 3F). Moreover, the proapoptotic ability of GD-NAIs was stronger than that of GD-E ( Figure 3F). Taken together, these results indicated that GD-NAIs can significantly promote apoptosis of K562 cells.

Target Prediction and Screening of GD-NAIs against CML
To predict the mechanism of GD-NAIs against CML, network pharmacology was carried out. First, a total of 68 components of GD-NAIs were screened out using absorption and drug-likeness parameters (Table S1 (Supplementary Materials)). Then, 583 gene targets of GD-NAIs were obtained from the SwissTargetPrediction network database, and 2871 gene targets of CML were collected through OMIM, GeneCards, PharmGKB, Drugbank and Therapeutic Target Database ( Figure 4A). A total of 294 overlapping genes were considered potential targets of GD-NAIs against CML ( Figure 4A; Table S2).

The Effect of GD-NAIs on Inducing Cell Apoptosis
One of the hallmarks of CML is the deregulation of apoptosis [39,40]. Therefore, the effect of GD-NAIs on the apoptosis of K562 cells was verified. The cell morphology was observed and showed that the number of K562 cells in the GD-NAIs (50, 100 and 200 μg/mL)-treated and cytarabine (20 nM)-treated groups was markedly less than that of the control group ( Figure 3D). GD-NAIs (50, 100 and 200 μg/mL) and cytarabine (20 nM) caused bursts and the death of K562 cells ( Figure 3D). The Hoechst 33258 assay further demonstrated that many bright blue cells appeared in the GD-NAIs (50, 100 and 200 μg/mL)-treated and cytarabine (20 nM)-treated groups but seldom in the control group ( Figure 3E), indicating that GD-NAIs could induce apoptosis of K562 cells. In addition, Annexin V-FITC/PI staining showed that the percentage of apoptotic cells in the GD-NAIs (50, 100 and 200 μg/mL)-, GD-E (50, 100 and 200 μg/mL)-treated and cytarabine (20 nM)treated groups was much higher than that in the control group ( Figure 3F). Moreover, the

PPI Network of Targets of GD-NAIs against CML
To investigate the interactions between the targets of GD-NAIs against CML, 294 genes were imported into STRING to construct a PPI network. The results showed that the PPI network contained 277 nodes and 1823 edges with an average degree of 12.4 ( Figure 4B). Among the PPI network, each node represented a gene target; the larger the node size was, the greater the degree value. The interaction strength was expressed by the number of node connections. The topological analysis of the PPI network based on the network parameter of degree was applied to screen out the key targets. The 57 targets' values were higher than the two-fold median value (18). The top 30 targets based on the three-fold median value were selected ( Figure 4B), which were tentatively regarded as the key targets of GD-NAIs against CML. A further screening parameter was set as the five-fold median value; thus, the top 10 targets including SRC, HSP90AA1, CTNNB1, AKT1, MAPK3, EGFR, EP300, PIK3R1, MAPK1 and PIK3CA were screened out ( Figure 4B). These targets might be the key targets of GD-NAIs against CML.

Target Prediction and Screening of GD-NAIs against CML
To predict the mechanism of GD-NAIs against CML, network pharmacology was carried out. First, a total of 68 components of GD-NAIs were screened out using absorption and drug-likeness parameters (Table S1 (Supplementary Materials)). Then, 583 gene targets of GD-NAIs were obtained from the SwissTargetPrediction network database, and 2871 gene targets of CML were collected through OMIM, GeneCards, PharmGKB, Drugbank and Therapeutic Target Database ( Figure 4A). A total of 294 overlapping genes were considered potential targets of GD-NAIs against CML ( Figure 4A; Table S2).

PPI Network of Targets of GD-NAIs against CML
To investigate the interactions between the targets of GD-NAIs against CML, 294 genes were imported into STRING to construct a PPI network. The results showed that the PPI network contained 277 nodes and 1823 edges with an average degree of 12.4 (Figure 4B). Among the PPI network, each node represented a gene target; the larger the node size was, the greater the degree value. The interaction strength was expressed by the number of node connections. The topological analysis of the PPI network based on the network parameter of degree was applied to screen out the key targets. The 57 targets' values were higher than the two-fold median value (18). The top 30 targets based on the three-fold median value were selected ( Figure 4B), which were tentatively regarded as the key targets of GD-NAIs against CML. A further screening parameter was set as the five-fold median value; thus, the top 10 targets including SRC, HSP90AA1, CTNNB1, AKT1, MAPK3, EGFR, EP300, PIK3R1, MAPK1 and PIK3CA were screened out ( Figure 4B). These targets might be the key targets of GD-NAIs against CML.

KEGG Pathway Enrichment Analysis of Targets of GD-NAIs against CML
KEGG pathway enrichment analysis of the 294 overlapping genes was applied to predict the signaling pathways regulated by GD-NAIs. The results revealed that the targets of GD-NAIs were mainly enriched in PI3K/AKT, MAPK, Endocrine resistance, Ras, FoxO, Th17 cell differentiation, JAK/STAT and TGF-β signaling pathways ( Figure 5; Table  S3), which are all potential therapeutic targets of CML.

KEGG Pathway Enrichment Analysis of Targets of GD-NAIs against CML
KEGG pathway enrichment analysis of the 294 overlapping genes was applied to predict the signaling pathways regulated by GD-NAIs. The results revealed that the targets of GD-NAIs were mainly enriched in PI3K/AKT, MAPK, Endocrine resistance, Ras, FoxO, Th17 cell differentiation, JAK/STAT and TGF-β signaling pathways ( Figure 5; Table S3), which are all potential therapeutic targets of CML. Due to the great number of targets, the top 30 targets of GD-NAIs against CML were used to construct a compound-target-pathway network. The results showed that 92 nodes and 488 edges made up the network, which included 1 herbal medicine, 34 com-

Compound-Target-Pathway Network and Analysis
Due to the great number of targets, the top 30 targets of GD-NAIs against CML were used to construct a compound-target-pathway network. The results showed that 92 nodes and 488 edges made up the network, which included 1 herbal medicine, 34 compounds, 30 targets, 26 pathways and 1 disease ( Figure 6). The greater the degree value, the more nodes were connected to this node, which proved that the node contributed significantly to the network. Thirty-four components were related to the top 30 targets (Figure 6). Compounds 68, 71, 69 and 75 had large degree values (Figure 6), indicating that these compounds might be the potential active components of GD-NAIs in the treatment of CML.

Compound-Target-Pathway Network and Analysis
Due to the great number of targets, the top 30 targets of GD-NAIs against CML w used to construct a compound-target-pathway network. The results showed that nodes and 488 edges made up the network, which included 1 herbal medicine, 34 co pounds, 30 targets, 26 pathways and 1 disease ( Figure 6). The greater the degree val the more nodes were connected to this node, which proved that the node contributed s nificantly to the network. Thirty-four components were related to the top 30 targets (F ure 6). Compounds 68, 71, 69 and 75 had large degree values (Figure 6), indicating t these compounds might be the potential active components of GD-NAIs in the treatm of CML.

Gene Expression Profile Regulated by GD-NAIs
To elucidate the underlying mechanism of GD-NAIs in inhibiting proliferation and inducing apoptosis of CML cells, RNA-seq was performed after K562 cells were treated with GD-NAIs (200 µg/mL) for 48 h. Hierarchical clustering analysis showed that there was a large gene expression difference between the control and GD-NAIs-treated groups ( Figure 7A). Volcano plot statistics and the graph of DEGs revealed that there were 141 DEGs upregulated by GD-NAIs and 318 DEGs downregulated by GD-NAIs ( Figure 7B), suggesting that GD-NAIs mainly played an inhibitory function on gene expression in K562 cells. DO enrichment analysis suggested that DEGs regulated by GD-NAIs were mainly involved in hematopoietic system disease, cancer, leukemia, hematologic cancer, cellular proliferation and lymphoid leukemia ( Figure 8A), which was in line with the therapeutic action of GD-NAIs on CML. GO enrichment analysis showed that DEGs were markedly enriched in positive regulation of hemopoiesis, hematopoietic or lymphoid organ development, negative regulation of signal transduction, and negative regulation of cell population proliferation ( Figure 8B). KEGG pathway enrichment analysis demonstrated that the upregulated DEGs were significantly associated with the renin-angiotensin system, valine, leucine and isoleucine biosynthesis, sphingolipid metabolism and soon ( Figure 8C). Conversely, the downregulated DEGs were prominently enriched in PD-L1 expression and the PD-1 checkpoint pathway in cancer, pathways in cancer, TGF-β, estrogen-signaling pathways and so on ( Figure 8D), which were all targets against CML. Reactome enrichment analysis revealed that the upregulated DEGs were mainly involved in metallothioneins-binding metals, the response to metal ions, metabolism of angiotensinogen to angiotensins, the biosynthesis of DPAn-3-derived 13-series resolvins and so on ( Figure 8E). However, the downregulated DEGs were mostly enriched in ESR-mediated signaling, signaling by Notch, signaling by Wnt, PI3K/AKT signaling in cancer, signaling by receptor tyrosine kinases and so on, which were all targets for therapeutic intervention ( Figure 8F). Combining the network pharmacology analysis, GO, KEGG and Reactome enrichment analysis, PD-L1 expression and the PD-1 checkpoint pathway in cancer, JAK/STAT, TGF-β, estrogen, Notch and Wnt were the main signaling pathways regulated by GD-NAIs for the treatment of CML.

Discussion
The development of CML is associated with multiple biological processes and signaling cascades [4]. Treatment of CML with herbal medicines has multitarget, multi-pathway and overall coordination characteristics. In the present study, we first tentatively characterized the chemical ingredients in GD-NAIs. The red node identified as rupestonic acid (compound 1) via library search ( Figure 9A

Discussion
The development of CML is associated with multiple biological processes and signaling cascades [4]. Treatment of CML with herbal medicines has multitarget, multi-pathway and overall coordination characteristics. In the present study, we first tentatively characterized the chemical ingredients in GD-NAIs. The red node identified as rupestonic acid (compound 1) via library search ( Figure 9A)   Compound 2 produced a precursor ion at m/z 251.1631 (−4.3 ppm, C15H22O3), wh was 2 Da higher in mass than rupestonic acid (compound 1) and yielded the same prod ions of methyl cycloheptane. In addition, the characteristic ions at m/z 233.1541 a 205.1591 corresponded to the successive loss of H2O and CO. Therefore, compound 2 w tentatively identified ( Figure 9B). Compound 3 exhibited a precursor ion at m/z 253.1 (−4.0, C15H24O3) and was then rapidly identified in a similar way ( Figure S1).  Figure 9C). Similar MS/MS fragmentations were also observed in these fo neighboring compounds (5, 6, 7 and 8) (Figures S2-5). Other compounds were sub quently tentatively characterized based on the above-mentioned fragmentation patter Not surprisingly, these tentatively characterized compounds were all sesquiterpenoid Sesquiterpenoids are one of the most attractive components in natural product che istry [9]. There are hundreds of natural sesquiterpenoids. Studies have confirmed that s quiterpenoids have inhibitory effects on leukemia [41][42][43]. Several examples are giv including that curcumol isolated from Curcuma longa L. can induce the differentiation a apoptosis of CML cells by blocking the BCR/ABL-JAK2/STAT3 and PI3K/AKT-JNK s naling pathways and activating the BH3-only gene [11]. Pseudolaric acid B derived fr Pseudolarix kaempferi was able to induce cell apoptosis in acute promyelocytic leukem HL-60 cells by inhibiting tubulin polymerization, preventing cell division and activat caspase-3 [12]. Dihydroartemisinin significantly induced apoptosis and inhibited vascu endothelial growth factor (VEGF) expression in K562 cells [13]. We tentatively charact  Figure 9C). Similar MS/MS fragmentations were also observed in these four neighboring compounds (5, 6, 7 and 8) (Figures S2-S5). Other compounds were subsequently tentatively characterized based on the above-mentioned fragmentation patterns. Not surprisingly, these tentatively characterized compounds were all sesquiterpenoids.
Sesquiterpenoids are one of the most attractive components in natural product chemistry [9]. There are hundreds of natural sesquiterpenoids. Studies have confirmed that sesquiterpenoids have inhibitory effects on leukemia [41][42][43]. Several examples are given, including that curcumol isolated from Curcuma longa L. can induce the differentiation and apoptosis of CML cells by blocking the BCR/ABL-JAK2/STAT3 and PI3K/AKT-JNK signaling pathways and activating the BH3-only gene [11]. Pseudolaric acid B derived from Pseudolarix kaempferi was able to induce cell apoptosis in acute promyelocytic leukemia HL-60 cells by inhibiting tubulin polymerization, preventing cell division and activating caspase-3 [12]. Dihydroartemisinin significantly induced apoptosis and inhibited vascular endothelial growth factor (VEGF) expression in K562 cells [13]. We tentatively characterized 61 new sesquiterpenoids and 14 known sesquiterpenoids from GD-NAIs, and no pharmacological activity against CML of these ingredients has been reported, which needs to be further explored.
CML is characterized by the loss of proliferation control [6]. We thus monitored the effect of GD-NAIs on the proliferation of K562 cells and observed that GD-NAIs conspicuously suppressed the proliferation of K562 cells in a concentration-dependent manner. Another hallmark of CML is a defect in apoptosis [39,40]. We found that GD-NAIs could induce apoptosis of K562 cells. These results indicated that GD-NAIs were able to suppress the proliferation and promote apoptosis of CML cells, which is consistent with previous studies showing that sesquiterpenoids have an inhibitory effect on CML [11,13]. Furthermore, we have experimentally proved that GD-NAIs demonstrate stronger proliferation inhibition and proapoptotic ability on K562 cells than that of GD-E. Meanwhile, we demonstrated that GD-NAIs had low toxicity to normal liver cells, indicating that GD-NAIs have high potential for clinical application.
Subsequently, the pharmacological mechanism of GD-NAIs against CML was investigated through a combination of network pharmacology and RNA-seq. We proved that GD-NAIs can exert an inhibitory effect on CML by regulating multiple signaling pathways, including PD-L1 expression and the PD-1 checkpoint pathway in cancer, the PI3K/AKT, JAK/STAT, TGF-β, estrogen, Notch and Wnt signaling pathways. PD-L1 expression and the PD-1 checkpoint pathway in the cancer signaling pathway plays an important role in tumor immunity. PD-1 binds to its ligand PD-L1 and inhibits T-cell activity. Abnormally high PD-LI expression on tumor cells mediates tumor immune escape [44]. An efficient therapeutic strategy for CML is to block this signaling pathway [3]. The PI3K/AKT signaling pathway regulates a large number of cellular processes, such as the cell cycle, proliferation, differentiation and apoptosis [5,7,45], whose dysregulation in leukemia stem cells (LSCs) increases ROS production and promotes the survival of LSCs and their drug resistance [6]. Furthermore, dysregulation of the JAK/STAT signaling pathway causes proliferation and resistance to apoptosis, which is a hallmark of BCR-ABL-transformed CML cells [46]. It has been reported that LSCs have a self-renewal capacity to generate leukemia progenitor cells by activating the TGF-β, Notch and Wnt signaling pathways [6]. The TGF-β signaling pathway participates in a wide range of cellular processes, such as growth, proliferation and apoptosis [47]. It was efficient to inhibit this pathway to decrease the maintenance of LSCs and eliminate CML leukemia-initiating cells [48]. According to the literature, inhibiting the Notch signaling pathway has anti-CML activity [49]. Moreover, inhibiting the estrogen signaling pathway induces apoptosis in a variety of leukemia cells [50]. As predicted by network pharmacology, compounds 68, 71, 69,

Cell Culture
The cell lines (K562, HEL and L02) were purchased from American Type Culture Collection (Bethesda, MD, USA). The K562 and HEL cells were cultured in RPMI 1640 medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) with 10% FBS (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) and 100 U/mL penicillin-streptomycin (Beyotime, Shanghai, China). L02 cells were cultured in DMEM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). Cells were maintained at 37°C with a 5% CO2 atmosphere. , and hydrochloric acid (XiLong Scientific, Chengdu, China) were used for isolation and enrichment of GD-NAIs. A Milli-Q water purification system (Billerica, MA, USA) was used to generate ultrapure water. Cytarabine was procured from Energy Chemical (Anqing, China).

Preparation of GD-NAIs
Samples were prepared according to a previously published method [24]. Whole fresh GD plants (1 kg) were prepared by the impregnation method with 20 L methanolwater (50:50, v v −1 ) with 0.1% hydrochloric acid for two weeks. After filtration, concentration and lyophilization, the extracts of GD (GD-E) were collected for further processing.

Preparation of GD-NAIs
Samples were prepared according to a previously published method [24]. Whole fresh GD plants (1 kg) were prepared by the impregnation method with 20 L methanol-water (50:50, v v −1 ) with 0.1% hydrochloric acid for two weeks. After filtration, concentration and lyophilization, the extracts of GD (GD-E) were collected for further processing.
Before uploading to the global natural products social (GNPS) molecular network web server (http://gnps.ucsd.edu (accessed on 1 January, 2021)), raw UHPLC-HRMS data files were prepared by ProteoWizard's msConvert [51]. The precursor ion mass tolerance and fragment ion mass tolerance were set as ± 0.02 Da on the GNPS platform, and the other parameters were set as default values. In addition, the apoptosis of K562 cells was determined by Hoechst 33258 assay. After treatment with GD-NAIs (50, 100 and 200 µg/mL) for 24 and 48 h, the cells were harvested and washed with cold PBS. The supernatants were discarded, and the cells were resuspended in cell stain buffer. Thereafter, 5 µL of Hoechst 33258 stain (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was added to the cells and incubated for 15 min at 37 • C. Cryogenic centrifugation was performed to remove the staining solution. Eventually, the samples were photographed on a fluorescence microscope (Olympus, Tokyo, Japan) at a magnification of ×200. In addition, the cells were photographed using light microscopy (Nikon, Tokyo, Japan). ployed to screen active components based on absorption and drug-likeness parameters. With the help of the SwissTargetPrediction network database (http://www.swisstargetprediction. ch/ (accessed on 15 January 2022)) [53], component-related gene targets were retrieved and duplicate genes were removed. The screening condition of genes was probability >0.1.

Construction of Protein-Protein Interaction (PPI) Network
The gene targets in Sections 4.8.1 and 4.8.2 were imported into Venny (http://www. liuxiaoyuyuan.cn/ (accessed on 8 May 2022)) [59]. The overlapping genes were regarded as targets of GD-NAIs against CML. STRING (https://cn.string-db.org/ (accessed on 8 May 2022)) [60] was used to construct the PPI network of targets of GD-NAIs against CML. The organism was set as "Homo sapiens", the minimum required interaction score was set as 0.700, and disconnected nodes in the network were hidden. Cytoscape 3.7.2 was used for topological analysis of nodes to select the key targets in the PPI network. The threshold value of the first screening was set as the two-fold median value and the further screening parameter was set as the five-fold median value.

Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway Enrichment Analysis
The Metascape database (https://metascape.org/gp/index.html#/main/step1 (accessed on 9 May 2022)) [61] was used for KEGG pathway enrichment analysis. A total of 294 overlapping genes of GD-NAIs and CML were uploaded to Metascape. Parameters were default values. The KEGG results were visualized using Cytoscape 3.7.2.

Construction of Compound-Target-Pathway Network
According to the relationship between overlapping genes with ingredients and KEGG enrichment pathways, the compound-target-pathway network was constructed and visualized using Cytoscape 3.7.2.

Statistical Analysis
Data are presented as the mean ± standard deviation (SD). Differences between groups were analyzed by one-way univariate analysis of variance (ANOVA). A p-value < 0.05 was considered to be statistically significant (marked as *). Higher significance levels were established at p-value < 0.01 (marked as **) and p-value < 0.001 (marked as ***).

Conclusions
In conclusion, we revealed the chemical components of GD-NAIs, confirmed the anti-CML activity, and elucidated the associated underlying mechanism of GD-NAIs against CML by combining UHPLC-HRMS-MN, network pharmacology and RNA-seq. Seventy-five sesquiterpenoids were tentatively characterized in GD-NAIs, in which four sesquiterpenoids would be the main active ingredients against CML, regulating PD-L1 expression and the PD-1 checkpoint pathway in cancer, PI3K/AKT, JAK/STAT, TGF-β, estrogen, Notch and Wnt signaling pathways. Our study is meaningful for understanding the pharmacological activity of GD-NAIs and provides an alternative agent against CML.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ph15111435/s1, Figure S1: Mass spectrum and proposed fragmentation pattern of compound 3; Figure S2: Mass spectrum and proposed fragmentation pattern of compound 5; Figure S3: Mass spectrum and proposed fragmentation pattern of compound 6; Figure  S4: Mass spectrum and proposed fragmentation pattern of compound 7; Figure S5: Mass spectrum and proposed fragmentation pattern of compound 8; Table S1: The absorption and drug-likeness parameters of 75 compounds in GD-NAIs; Table S2: 294 overlapping genes; Table S3: Detailed information of KEGG pathways which met the requirements of Gene count ≥10.

Conflicts of Interest:
The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper.