Artemisinin derivatives maintain fibroblast normalization by acting on tumor-stroma interactions in oral tongue squamous cell carcinoma

Abstract

Fibroblasts can transform into cancer-associated fibroblasts (CAFs) when continuously stimulated by cancer cells, thereby playing a crucial role in cancer progression. Growing evidence indicates that targeted therapy for CAF can influence tumor progression. Dihydroartemisinin (DHA) and artemether (ARM), semisynthetic derivatives of the natural compound artemisinin, have exhibited anticancer effects in various tumors. In this study, we found that tumor cells secreted platelet-derived growth factor-BB (PDGF-BB), which stimulated fibroblasts to transition into the CAF phenotype (cell phenotype and secretory phenotype). CAFs promote Cal-27 cell proliferation by secreting lactate. We focused on the mechanisms by which DHA and ARM affect the tumor-stroma interactions. These findings demonstrated that DHA and ARM effectively suppressed the secretion of PDGF-BB from Cal-27 cells, maintaining the normal state of hOMF and preventing the proliferative effect on Cal-27 cells. These findings were confirmed in xenograft models. Our study showed that artemisinin derivatives prevent the progression of oral tongue squamous cell carcinoma (OTSCC) by inhibiting the production of PDGF-BB in cancer cells to maintain the normal state of fibroblasts, thus providing a potential avenue for targeted OTSCC treatment.

Introduction

Oral squamous cell carcinoma (OSCC) is a common malignant tumor of the head and neck region, originating from the oral cavity’s mucosal lining [1]. The cancer accounts for approximately 170,000 deaths annually worldwide [2]. Despite therapeutic advances, no significant improvement has been observed in long-term patient survival, primarily because the role of the tumor microenvironment in cancer development has been overlooked [3,4]. The association of the TME with cancer growth implicates non-cancerous stromal components in tumor progress, making them viable targets for innovative cancer therapies. Insight into these interactions offers promising therapeutic avenues for improved patient survival [5].

Cancer-associated fibroblasts (CAFs) are mesenchymal cells that constitute the tumor microenvironment (TME) and contribute to tumor progression in several ways [6-8]. These fibroblasts, primarily derived from host cells, transform irreversibly into CAFs under the persistent influence of cancer cells [9]. Various factors in tumors regulate the conversion of fibroblasts to CAFs, with existing studies showing that cytokines and tumor metabolites can induce fibroblast activation [10,11]. Cancer cells also secrete numerous factors that facilitate this conversion [12]. Moreover, metabolic changes in fibroblasts, particularly the shift towards aerobic glycolysis, are essential for fibroblasts’ phenotypic conversion [13]. Normal fibroblasts adopt a CAF phenotype under the influence of tumor cells and key components of the tumor microenvironment. CAFs promote tumor progression by secreting pro-oncogenic factors.

Platelet-derived growth factor (PDGF) consists of four polypeptides - A, B, C, and D - that form various dimeric combinations, which activate the PDGFα and PDGFβ receptors, thereby initiating receptor tyrosine kinase signaling [14,15]. This activation process facilitates multiple cellular and biological processes. PDGF-BB is mainly associated with tumor initiation and progression in the PDGF family [16]. PDGF-BB plays various roles in tumors, promoting tumor progression by enhancing tumor cell proliferation, inhibiting apoptosis, and mediating the transformation of endothelial cells into CAFs [17-19]. PDGF-BB induces aerobic glycolysis through the HIF-1α pathway to promote fibroblast transformation to CAFs [20]. PDGF-BB can promote this process through several mechanisms during the transformation of fibroblasts to cancer-associated fibroblasts (CAFs). Therefore, targeted inhibition of PDGF-BB signaling in fibroblasts may offer a potential therapeutic approach for tumor treatment.

Artemisinin is the first-line treatment for malaria [21]. Recently, a series of derivatives, such as dihydroartemisinin (DHA) and artemether (ARM), have been synthesized by modifying the parent structure of artemisinin. These derivatives exhibit good bioactivity and solubility and have emerged as potential therapeutic agents with broad-spectrum anticancer properties [22]. However, owing to the broad anticancer effects of these compounds, their molecular mechanisms are highly complex, and the mechanisms underlying their ability to inhibit tumor progression and metastasis remain unclear [23]. Research has shown that artemisinin derivatives target cyclin-dependent kinases (CDKs) to inhibit cell proliferation [24]. In addition, artemisinin derivatives can induce apoptosis in cancer cells [25]. Artesunate can reduce the expression of HSP70 and Bcl-2 in breast cancer cells and induce caspase-dependent apoptosis [26]. DHA induces apoptosis in melanoma cells through the NOXA-dependent pathway [27]. Artemisinin derivatives can also induce ferroptosis, selectively eliminating cancer cells while protecting normal cells [28,29]. Interestingly, the tumor microenvironment (TME) is an essential target for cancer treatment, and artemisinin and its derivatives also play a significant role in regulating TME. It has been shown that DHA can induce the polarization of tumor-associated macrophages (TAMs) towards the M1 phenotype in a dose-dependent manner [30]. DHA can also inhibit the polarization and infiltration of regulatory T cells (Tregs) in the tumor microenvironment by reducing IL-10 and IL-6 levels in melanoma [31]. Additionally, DHA promotes anti-tumor immunity by decreasing Tregs and increasing cytotoxic T cells’ infiltration (CD8+) [32]. Artemisinin derivatives also induce leukemia cell apoptosis by inhibiting the JAK2/STAT3 signaling pathway, thereby reducing the number of M2-type macrophages [33]. These compounds interfere with cancer cell growth by targeting specific proteins and cells.

Artemisinin derivatives inhibit tumor growth through various mechanisms, and numerous studies have shown that these derivatives can regulate the tumor immune microenvironment. However, there is limited understanding of how they modulate stromal cells within the tumor microenvironment. Whether inactivating, reducing, or normalizing CAFs can decrease cancer progression and metastasis remains unclear. In this study, we investigate how artemisinin derivatives contribute to the maintenance of normalized NFs in TME. Our findings indicate that DHA and ART can significantly hinder the transformation of NFs to CAFs by blocking cancer cell-mediated PDGF-BB secretion, thereby curtailing cancer advancement. This suggests that artemisinin derivatives have the capability to disrupt the intricate interactions between tumor cells and TME.

Material and methods

Cell culture and treatment

Neutrophils were sourced from BFBCOMPETENTCE LL.com. (Shanghai, China). Normal human oral mucosal keratinocytes (hOMK) and normal human oral mucosal fibroblasts (hOMF) were obtained from CellResearch Corp. (Singapore). OTSCC cells (Cal-27) were obtained from BioVector (Beijing, China). The cells were cultured at 37°C and 5% CO₂ in an incubator. The cells were cultured in DMEM containing 10% FBS (Gibco), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco). Adherent cells were subjected to experimentation when they reached 70%-80% confluence, whereas suspension cells were analyzed based on the observed formation of distinct cell clusters under a microscope. CAFs (Cal27-CM) were derived by treating hOMF with a conditioned medium from Cal-27 (Cal27-CM) at varying ratios for 48 h. CAFs (PDGF-BB) were obtained by treating the hOMF cells with PDGF-BB for 48 h.

Reagents

Antibodies against α-SMA (ab5694), PDGF-BB (ab23914), and β-tubulin (ab179513) were purchased from Abcam (Cambridge, UK). PI3K antibody (ServiceBio GB11525-50), AKT antibody (ServiceBio GB15689-50), and p-AKT antibody (ServiceBio GB150002-50) were purchased from ServiceBio. The p-PI3K antibody (Affinity AF3242) was purchased from Affinity. Recombinant human PDGF-BB was obtained from PeproTech (Beijing, China). Ki67 antibody (ServiceBio GB111499) and hematoxylin and eosin (HE) staining dye (ServiceBio G1005) were sourced from ServiceBio. Sodium oxamate (LDH-A lactate dehydrogenase inhibitor, MCE HY-W013032A) and lactate (MCE HY-B2227) were acquired from MedChemExpress (MCE). Sodium carboxymethyl cellulose (CMC-Na, MCE HY-Y1889A) and platelet-derived growth factor receptor (PDGFR) inhibitor (MCE CP73451) were also purchased from MCE. Dihydroartemisinin (Aladdin A140839-1) and artemether (Aladdin A107447-5) were obtained from Aladdin. Dimethyl sulfoxide (Solarbio D8371) was purchased from Solarbio. The MCT1/2 inhibitor was purchased from MCE (MCE AR-C155858). Pentobarbital sodium was purchased from Sigma Aldrich (Sigma Aldrich 11715-100 mg). The pan-AKT kinase inhibitor (AZD5363) was purchased from MCE (MCE HY-15431). The Akt activator (SC79) was purchased from MCE (MCE HY-18749).

Cell proliferation assay

Treatment of different cells (HOMF, Cal-27, and CAF cells) with DHA and ARM for 48 h. Afterward, CCK-8 reagent was added, and the solution was incubated for two h. The absorbance was measured using a spectrophotometer, and the cell proliferation rate was calculated from the OD value. The proliferation of Cal-27 cells in the presence of CAFs and lactic acid was assessed using the same method.

Cell apoptosis assay

Treatment of different cells (HOMF, Cal-27, and CAF cells) with DHA and ARM for 48 h. The culture supernatant was collected, and the detached cells were centrifuged at 1000 rpm for 5 min, after which the supernatant was discarded. The cells were then incubated for 20 min using the Annexin V-FITC Apoptosis Detection Kit (Beyotime Inst Biotech, China). Following incubation, apoptosis was detected using a flow cytometer (Sony ID 7000, Japan), and the apoptosis data were analyzed using FlowJo 10.9 software.

IC₅₀ determination of artemisinin derivatives and cytotoxicity-free dose screening

Cal-27 and hOMF cells were treated with DHA (0, 2, 4, 8, 16, 32, 64, 128, 256, and 300 μM) and ARM (0, 2, 4, 8, 16, 32, 64, 128, 256, and 512 μM) for 48 h. Following this, CCK-8 reagent was added, and the solution was incubated for 2 h. The absorbance was measured using a spectrophotometer. The absorbance measurements determined the rate of cell proliferation, and dose proliferation curves for DHA and ARM in Cal-27 and hOMF cells were plotted to calculate the IC₅₀ values. Based on the maximum concentration of DHA and ARM without apparent cytotoxicity, further experiments were performed to investigate their effects on Cal-27, hOMF, CAFs (Cal27-CM), and CAFs (PDGF-BB).

Cellular PDGF-BB secretion capacity assay

In a controlled assay to evaluate PDGF-BB secretion, Cal-27, hOMF, hOMK, and neutrophils were cultured at consistent densities in 6-well plates and incubated for 48 h. Clarified medium was obtained by centrifugation of the supernatant at 2000 × g for 5 min, and PDGF-BB levels were determined by ELISA (RD Systems, DBB00). After DHA and ARM treatment of Cal-27 cells for 48 h, PDGF-BB levels were assayed via the method described above.

Cell morphology and spindle index

Under an OLYMPUS inverted microscope, hOMF cells subjected to diverse conditioned media were observed for morphological analysis and captured through photography. Using ImageJ software for precise measurement, the cell dimensions were recorded systematically across 3-4 randomized assessments per sample, fulfilling a count of at least 40 cells to ensure data accuracy and validity. The spindle index formula: Spindle Index = length/width (cell) [34,35].

Analysis of cell lactate secretion capability

hOMF cells treated under various conditions were seeded in equal numbers in 6-well plates before further incubation for 48 h. Clarified medium was obtained by centrifugation of the supernatant at 2000 × g for 5 min, and lactate levels were determined using the Lactate Assay Kit (BioVision Technologies).

Western blot analysis

Proteins were extracted by lysing the cells and quantified using SDS-PAGE, followed by the transfer of the proteins to membranes (Millipore, MA, USA). The membranes were blocked with a blocking solution (Serviebio G2052) and then incubated with primary antibodies (α-SMA, PDGF-BB, Tubulin-β, 1:1000). The membranes were subsequently incubated with secondary antibodies for detection.

Transcriptome sequencing (RNA-seq)

The transcriptome sequencing (RNA-seq) in this study was performed using Cal-27 tumor cells (human origin) treated with DHA/DMSO. Total RNA was extracted from the cells using TRIzol reagent (Servicebio) after treatment. The sequencing data were analyzed by aligning to known reference genomes and annotation files. We used sequence similarity alignment to identify the expression abundance of protein-coding genes in different samples. The alignment results were obtained in counts. To eliminate the impact of differences in gene length and sequencing depth on expression quantification, we used the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) method to calculate gene expression levels. Subsequently, the gene counts for each sample were normalized using DESeq2 software. Normalization was performed using the BaseMean value to estimate expression levels, and fold changes were calculated for each gene. Differential significance testing was conducted using the Negative Binomial (NB) test, and protein-coding genes with significant differential expression were identified based on fold change and significance test results. All sequencing data have been uploaded to the GEO public database, and the data can be accessed via accession number GSE294841.

Animal source and breeding environment

The mice (BALB/c-nude) used in this study were all obtained from SPF (Beijing) Biotechnology Co., Ltd. The company’s animal production license number is (SCXK-2024-0001). The mice have passed 21 parasitic and infectious disease tests, including tests for hepatitis virus, parasites, and mousepox virus. The SPF mice were housed in a standardized animal breeding environment at the Kunming Medical University Animal Laboratory. The temperature in the laboratory was controlled between 22-24°C, with humidity maintained at 40%-60%. The mice were kept on a 12 h light/12 h dark cycle, with a light intensity of approximately 200-300 lux. The mice were fed sterilized feed, and their drinking water was treated distilled water, both of which were replaced regularly. All facilities in the mouse breeding environment were thoroughly cleaned and disinfected weekly to ensure a sterile environment.

Drug safety dose screening

Animal studies were sanctioned by the Ethics Review Committee for Animal Experiments at Kunming Medical University (kmmu20230673). Four-week-old male BALB/c nude mice were selected for this experiment, acquired, and reared in a specific pathogen-free (SPF) facility. DHA and ARM were formulated into gavage-friendly emulsions using CMC-Na before administration. The study involved 21 mice divided into seven distinct groups, encompassing three mice per grouping. Over 21 days, each group was orally administered either a control solution (CMC-Na) or different dosages of DHA (50, 25, and 10 mg/kg) and ARM (75, 35, and 20 mg/kg). Body weight assessments were conducted at three-day intervals throughout the study. After completing the 21 days of treatment, mice underwent a fasting period of 12 h before being euthanized. The weights of the liver and kidneys were then recorded, and serum analyses were performed to ascertain biochemical markers of liver and kidney functionality. By evaluating growth patterns, organ weights, and liver and kidney functionality, safe drug dosages were established.

Tumor xenograft model

Animals were obtained and housed, as previously described. A subcutaneous Cal-27 cell xenograft mouse model was established by injecting 5 × 10⁶ Cal-27 cells into the axillary region of mice. After six days, mice were divided into control and intervention groups. These groups underwent a daily gavage procedure for 21 uninterrupted days. Measurements were taken every third day to track changes in body weight and tumor size. On day 21, the mice were euthanized, and the final tumor weight was recorded. Moreover, a subcutaneous co-transplantation model was developed using hOMF/CAFs and Cal-27 cells. A combination of 5 × 10⁶ hOMF/CAFs cells and 5 × 10⁶ Cal-27 cells was injected into the axillary regions. Starting from day 6 after injection, measurements of tumor size and body mass began, followed by a regimen of daily gavage for 21 days. Every third day, we recorded the body mass and tumor dimensions. On the final day, the mice were euthanized, and tumor weights were recorded. The tumor volume was calculated using the formula: V = 1/2 × length × width².

Animal euthanasia

The animal experiments in this study comply with the ARRIVE guidelines. We performed euthanasia of the mice using a combination of sodium pentobarbital anesthesia and cervical dislocation. First, mice were anesthetized with an intraperitoneal injection of sodium pentobarbital (60 mg/kg), and their muscle tone, response to external stimuli, and respiratory rate were monitored to assess the anesthetic state until the mice were fully anesthetized [36]. Subsequently, the mouse’s tail was grasped and swiftly pulled upward to induce cervical dislocation. Throughout the procedure, we minimized noise and ensured the operation was quick and precise to reduce the mice’s stress response and minimize potential suffering.

Detection of PDGF-BB and lactate in serum from tumor-bearing mice

At the end of the animal experiments involving tumor-bearing mice, whole blood was collected from the mice. After allowing the blood to clot, the serum was separated by high-speed centrifugation (12,000 × g for 15-30 min). Serum PDGF-BB levels and lactate concentrations were measured using a PDGF-BB ELISA kit and a lactate assay kit.

Immunohistochemistry

Tumor tissues were fixed in 4% paraformaldehyde, followed by paraffin embedding and sectioning. The section thickness was 5 μm, and the slides underwent deparaffinization and rehydration. Subsequently, tissue sections were stained using anti-Ki67 and anti-α-SMA antibodies. Hematoxylin and eosin (HE) staining was performed to observe the morphological structure of the tissue. TUNEL staining was applied to the sections to detect apoptotic cells further. Images of the stained sections were captured using an Olympus VS120 imaging system. The percentage of Ki67-positive cells was used to quantify proliferating tumor cells, while the percentage of α-SMA-positive areas was used to quantify the expression levels of α-SMA protein in the tumor tissue. The percentage of TUNEL-negative cells was used to quantify non-apoptotic cells.

Statistical analysis

Data were analyzed using GraphPad Prism 9.5. Data are presented as the mean ± standard deviation (mean ± SD). Data processing involved t-tests and variance analysis (ANOVA), with Bonferroni correction applied for multiple comparisons. Unless otherwise stated in the figure legend, each data point represents the average of three biological replicates. Statistical significance is interpreted as follows: P < 0.05 (*).

Results

PDGF-BB was upregulated and associated with poor outcomes in head and neck squamous cell carcinoma

Analysis using the online database GEPIA-2 showed that the expression of PDGF-B was significantly higher in cancer tissues (head and neck squamous cell carcinoma, HNSC) than in normal tissues. Moreover, survival was lower in patients with high expression than those with low expression (Figure 1A, 1B). ELISA assays of the conditioned media from different cell types showed that the PDGF-BB levels in the conditioned media of oral tongue squamous cell carcinoma (OTSCC) cells (Cal-27) were significantly higher than those in oral keratinocytes (hOMK), fibroblasts (hOMF), and neutrophils (Figure 1C). In addition, in xenograft tumor models, we observed a significant increase in the serum levels of PDGF-BB in tumor-bearing mice and a significant decrease in the body weight of the mice (Figure 1D-H). Based on these findings, we concluded that PDGF-BB plays a significant role in the progression of OTSCC.

Figure 1.

PDGF-B expression and Kaplan-Meier analysis of HNSC. PDGF-B expression in normal tissues and HNSC (A). Kaplan-Meier analysis of HNSC patients with high or low PDGF-B expression (n = 467) (B). ELISA detection of PDGF-BB expression in different cell culture media (n = 3) (C). Schematic diagram of experimental design of mice xenograft tumor model (D). Comparison between the tumor-bearing and tumor-free mice (n = 3) (E). Representative H&E staining of tumor sections from tumor-bearing mice (n = 3) (F). ELISA measurement of PDGF-BB expression in the sera of tumor-bearing and tumor-free mice (n = 3) (G). Weight changes in the tumor-bearing and tumor-free mice (n = 3) (H). The data were expressed as the mean ± standard deviation (C, G, H). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

PDGF-BB induces fibroblast-to-CAF conversion to promote tumor cell proliferation

Based on our observations, OTSCC cells (Cal-27) secrete high levels of PDGF-BB (Figure 1). To simulate the activation of fibroblasts by tumor-secreted products, we co-cultured hOMF cells indirectly with conditioned media (Cal27-CM) derived from Cal-27 cell supernatants at ratios of 1/4, 2/4, and 3/4 (Figure 2A-F). As the proportion of conditioned media increased, hOMF cells exhibited morphological changes, an increased spindle index, upregulation of α-SMA activation protein, and significantly enhanced lactate secretion (Figure 2A-E). These observations implied that products secreted by tumors can convert hOMF cells into CAFs (Cal-CM). The CAFs (Cal-CM) significantly enhanced the proliferation rate of Cal-27 cells (Figure 2F). To verify the role of PDGF-BB in the activation of fibroblasts (Figure 2G-L), hOMF cells were exposed to varying PDGF-BB concentrations. The results showed that increasing PDGF-BB concentrations led to morphological changes in hOMF cells, an increased spindle index, upregulation of α-SMA activation protein, and a significant enhancement in lactate secretion (Figure 2G-K). The PDGF-BB-activated hOMF cells also significantly promoted the proliferation of Cal-27 cells (Figure 2L). Furthermore, we used the PDGF-R receptor inhibitor (CP73451). When Cal-27 cell supernatants treated with the inhibitor were used to culture hOMF cells, the activation of hOMF cells was significantly reduced (Figure 2M-Q). Moreover, the capacity of hOMF cells to enhance the proliferation of Cal-27 cells was considerably reduced (Figure 2R). The above results suggest that oral tongue squamous carcinoma cells induce fibroblast transformation to CAF by secreting PDGF-BB.

Figure 2.

PDGF-BB induces fibroblast-to-CAF conversion to promote tumor cell proliferation. Evaluation of fibroblast-to-CAF conversion markers in hOMF cells treated with Cal-27 conditioned medium (Cal27-CM) at 1/4, 2/4, and 3/4 ratios for 48 h (n = 3) (A-F). Morphology of hOMF cells (Scale bar: 100 μm; n = 3) (A). Western blot analysis of α-SMA protein in cells (n = 3) (B). Spindle index (n = 120) (C). Quantification of activated α-SMA protein in cells (n = 3) (D). Lactate secretion levels in CAFs (Cal27-CM) (n = 3) (E). The Effect of CAFs (Cal27-CM) on the proliferation of Cal-27 cells (n = 3) (F). Evaluation of fibroblast-to-CAF conversion markers in hOMF cells treated with 20, 40, and 60 ng/ml PDGF-BB for 48 h (n = 3) (G-L). Morphology of the hOMF cells (scale bar: 100 μm; n = 3) (G). Western blot analysis of α-SMA protein in cells (n = 3) (H). Spindle index (n = 120) (I). Quantification of α-SMA protein in cells (n = 3) (J). Lactate secretion levels in CAFs (PDGF-BB) (n = 3) (K). Effect of CAFs (PDGF-BB) on the proliferation of Cal-27 cells (n = 3) (L). Detection of fibroblast activation markers in hOMF cells treated with a 3/4 ratio of Cal-27 conditioned medium (Cal27-CM) and PDGF-R receptor blocker (CP73451) for 48 h (n = 3) (M-R). Cell morphology (scale bar, 100 μm; n = 3) (M). Western blot analysis of activated α-SMA in hOMF cells (n = 3) (N). Spindle index (n = 120) (O). Quantification of α-SMA protein in cells (n = 3) (P). Lactate secretion levels in the cells (n = 3) (Q). Effect of hOMF on the proliferation of Cal-27 cells (n = 3) (R). The data were expressed as the mean ± standard deviation (C-F, I-L, O-R). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

CAF-derived lactate promotes the proliferation of OTSCC cells

In our research, we found that both the conditioned medium from Cal-27 (Cal27-CM) and PDGF-BB can trigger the conversion of hOMF cells into a CAF-like phenotype. These CAFs exhibited a marked increase in lactate production (Figure 2). By supplementing with lactate externally, we observed that lactate supported the proliferation of Cal-27 cells (Figure 3E). In the study of the role of lactate in the CAF-mediated biological behavior of tumor cells, sodium oxamate, which inhibits LDH-A, was employed to prevent CAFs from secreting lactate (Figure 3A, 3C). The data revealed that blocking lactate release from CAFs considerably diminished their ability to enhance the proliferation of Cal-27 cells (Figure 3B, 3D). To further demonstrate that the lactate derived from CAFs is a key factor in stimulating tumor cell proliferation, we treated Cal-27 cells with CAF-conditioned medium supplemented with monocarboxylate transporter inhibitors 1 and 2 (MTCi1/2) (MTCi refers to a class of compounds that can inhibit the function of monocarboxylate transporter. These compounds work by reducing the activity of MCT, thereby blocking the transport of lactate [37-39]). The results showed that, compared to the CAF-conditioned medium alone, the CAFs-sup + MTCi group significantly inhibited Cal-27 cell proliferation (Figure 3F, 3G). These observations imply that lactate originating from CAFs is vital for encouraging the growth of cancer cells.

Figure 3.

Lactate from CAFs promotes the proliferation of Cal-27 cells and the inhibitory concentration of artemisinin derivatives on hOMF and Cal-27 cells. Lactate secretion levels in CAFs (Cal27-CM) treated with NaOx for 48 h (n = 3) (A). The effect of sodium oxalate-treated CAFs (Cal27-CM) on the proliferation of Cal-27 cells (n = 3) (B). Lactate secretion levels in CAFs (PDGF-BB) treated with NaOx for 48 h (n = 3) (C). Effect of sodium oxalate-treated CAFs (PDGF-BB) on the proliferation of Cal-27 cells (n = 3) (D). Proliferation of Cal-27 cells treated with 5, 10, and 15 mmol/L lactate (n = 3) (E). Proliferation of Cal-27 cells after adding monocarboxylate transporter (MCT) inhibitors 1 and 2 to CAFs (Cal27-CM)-conditioned medium (n = 3) (F). Proliferation of Cal-27 cells after adding monocarboxylate transporter (MCT) inhibitors 1 and 2 to CAFs (PDGF-BB)-conditioned medium (n = 3) (G). Effect of DHA on the proliferation of Cal-27 (H) and hOMF cells (I) at concentrations of 0, 2, 4, 8, 16, 32, 64, 128, 256, and 300 μM (n = 3) (H, I). Effect of ARM on the proliferation of Cal-27 (K) and hOMF cells (L) at concentrations of 0, 2, 4, 8, 16, 32, 64, 128, 256, 512, and 600 μM (n = 3) (K, L). Chemical structures and purity of DHA and ARM, with purity of DHA and ARM greater than or equal to 98% (J, M). The data were expressed as the mean ± standard deviation (A-I, K, L). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Artemisinin derivatives regulate the tumor microenvironment by targeting cancer cells in OTSCC

Based on the effects of DHA and ARM on the proliferation of Cal-27 and hOMF cells, we determined that the maximum non-inhibitory concentrations were 8 μM for DHA and 30 μM for ARM (Figures 3H-M, 4C). Furthermore, we found that at these concentrations (DHA: 8 μM and ARM: 30 μM), neither DHA nor ARM significantly induced apoptosis in Cal-27, hOMF, or CAF cells (Figure 4A, 4B) [40]. Cal-27 cells exposed to DHA (8 μM) and ARM (30 μM) for 48 h showed a notable reduction in PDGF-BB protein levels and secretion (Figure 4D-F). When we indirectly co-cultured the supernatants from DHA- and ARM-treated Cal-27 cells with hOMF cells, there was a considerable reduction in the spindle-shaped index of hOMF cells, a decline in the activation marker α-SMA, as well as decreased lactate secretion and reduced growth impact on Cal-27 cells (Figure 4G-L). We also investigated the implications of DHA and ARM on CAFs, revealing that these compounds did not substantially alter CAF activation markers (Figure 5A-E). Administering DHA or ARM to CAFs showed no notable change in their influence on the proliferation of Cal-27 cells (Figure 5F). Additionally, while exploring DHA and ARM as potential inhibitors of PDGF-BB-induced hOMF activation, we found that despite DHA (8 μM) and ARM (30 μM) reducing the transformation markers associated with PDGF-BB-induced conversion of hOMF to CAFs (Figure 5G-K), these CAFs did not significantly reduce Cal-27 cell proliferation (Figure 5L). These outcomes imply that artemisinin derivatives help normalize hOMF by primarily targeting oral tongue squamous carcinoma (OTSCC) cells to reduce PDGF-BB secretion.

Figure 4.

Artemisinin derivatives inhibit the secretion of PDGF-BB by Cal-27 cells, reducing the transformation of hOMF into CAF. Apoptosis results of flow cytometry (n = 3) (A). Apoptosis statistics (n = 3) (B). Effect of DHA (8 μM) and ARM (30 μM) on the proliferation of Cal-27, hOMF, CAFs (Cal27-CM), and CAFs (PDGF-BB) (n = 3) (C). Intracellular expression and quantification of PDGF-BB in Cal-27 cells treated with DHA or ARM (n = 3) (D, E). ELISA analysis of PDGF-BB expression levels in Cal-27 cells treated with DHA and ARM (n = 3) (F). Analysis of hOMF cell activation markers following 48-h treatment with Cal-27 (DHA/ARM)-conditioned medium (G-L). Morphology of the hOMF cells (scale bar: 100 μm; n = 3) (G). Western blotting analysis of α-SMA protein expression (n = 3) (H). Spindle index (n = 120) (I). Quantification of α-SMA protein expression (n = 3) (J). Lactate secretion in hOMF cells (n = 3) (K). Effect of hOMF on the proliferation of Cal-27 cells (n = 3) (L). The data were expressed as the mean ± standard deviation (B, C, E, F, I-L). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Figure 5.

Artemisinin derivatives have a limited effect on reversing CAF function and blocking PDGF-BB activation pathways. After 48 h treatment of CAFs (Cal27-CM) with DHA (8 μM) and ARM (30 μM), related CAFS markers were assessed (n = 3) (A-F). Morphology of the CAF cells (scale bar: 100 μm; n = 3) (A). Western blot analysis of α-SMA protein expression in CAF (n = 3) (B). Spindle cell index of CAF (n = 120) (C). Quantifying activated α-SMA protein expression in CAF cells (n = 3) (D). Lactate secretion in CAFS cells (n = 3) (E). Effect of CAF on proliferation of Cal-27 cells (n = 3) (F). After co-treatment of hOMF cells with PDGF-BB and DHA or ARM, the related activation markers in hOMF cells were assessed (n = 3) (G-L). Morphology of the hOMF cells (scale bar: 100 μm; n = 3) (G). Western blotting analysis of α-SMA protein expression in hOMF cells (n = 3) (H). Spindle cell index of hOMF cells (n = 120) (I). Quantification of α-SMA expression in hOMF cells (n = 3) (J). Lactate secretion in hOMF cells (n = 3) (K). Effect of hOMF on the proliferation of Cal-27 cells (n = 3) (L). The data were expressed as the mean ± standard deviation (C-F, I-L). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Artemisinin derivatives target the PI3K-AKT signaling pathway to inhibit tumor cell PDGF-BB expression

This study found that artemisinin derivatives can effectively inhibit the secretion of PDGF-BB and suppress its expression in tumor cells (Figure 4), providing further insight into their mechanism of action. We selected tumor cells treated with DHA for transcriptome sequencing, and the results showed that 242 genes were downregulated (P < 0.05, Log10 FC < -1) (Figure 6A). Further pathway enrichment analysis using WIKI and KEGG showed that both pathways were enriched in the PI3K-AKT pathway (Figure 6B, 6C). Western blotting revealed that after DHA and ARM treatment, the expression of p-PI3K and p-AKT was significantly downregulated (Figure 6D-G). When tumor cells were treated with the AKT inhibitor (AZD5363), PDGF-BB expression was downregulated considerably (Figure 6H-J). To further verify the relationship between this pathway and PDGF-BB, we utilized an AKT activator (SC79), which led to a significant upregulation of PDGF-BB. Additionally, co-treatment with the AKT activator and DHA/ARM caused a downregulation of PDGF-BB expression in tumor cells (Figure 6K-M). These results suggest that artemisinin derivatives can inhibit tumor cell PDGF-BB expression by targeting the PI3K-AKT pathway.

Figure 6.

Artemisinin derivatives interfere with the PI3K-AKT pathway to inhibit tumor cell PDGF-BB expression. Transcriptomic data analysis after DHA (8 μM) treatment of tumor cells for 48 hours (n = 3) (A-C). Volcano plot of differentially expressed genes (P < 0.05, Log10 FC < -1) (n = 3) (A). WikiPathway enrichment (n = 3) (B). KEGG pathway enrichment (n = 3) (C). Western blotting and quantification results of PI3K-AKT proteins after DHA and ARM treatment of tumor cells (n = 3) (D-G). Western blot of PI3K, p-PI3K, AKT, p-AKT, and PDGF-BB after DHA and ARM treatment of tumor cells (n = 3) (D). Quantification of p-PI3K/PI3K (n = 3) (E). Quantification of p-AKT/AKT (n = 3) (F). Quantification of PDGF-BB (n = 3) (G). Expression of p-AKT, AKT, and PDGF-BB after AKT inhibitor (AZD5363) treatment of tumor cells (n = 3) (H-J). Western blot of p-AKT, AKT, and PDGF-BB after AKT inhibitor (AZD5363) treatment of tumor cells (n = 3) (H). Quantification of p-AKT/AKT (n = 3) (I). Quantification of PDGF-BB (n = 3) (J). Expression of p-AKT, AKT, and PDGF-BB after AKT activator (SC79)/DHA and ARM treatment of tumor cells (n = 3) (K-M). Western blot of p-AKT, AKT, and PDGF-BB after AKT activator (SC79)/DHA and ARM treatment of tumor cells (n = 3) (K). Quantification of p-AKT/AKT (n = 3) (L). Quantification of PDGF-BB (n = 3) (M). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Artemisinin derivatives effectively inhibit tumor progression at safe doses

DHA and ARM are derivatives of artemisinin, which, as first-line antimalarial drugs, have been widely recognized for their safety [23]. However, the drug concentrations and treatment durations used in antitumor therapies differ significantly from those used in antimalarial treatments [41]. First, we screened for safe doses of DHA and ARM to assess the safety and antitumor efficacy of artemisinin derivatives in cancer therapy. The mice were categorized into three dose groups (high, medium, and low dose groups), and the CMC-Na group (high dose: DHA 50 mg/kg, ARM 75 mg/kg; medium dose: DHA 25 mg/kg, ARM 35 mg/kg; low dose: DHA 10 mg/kg, ARM 20 mg/kg; control group: CMC-Na) for 21 consecutive days (Figure 7A). The results revealed that, in the high-dose group, the mice’s body, liver, and kidney weights significantly decreased (Figure 7B-D). Further biochemical analysis of liver and kidney functions indicated varying degrees of damage in the high-dose group (Figure 7E-I). In the medium-dose group, only a few parameters showed changes (Figure 7B-I). In the low-dose group, no significant differences were observed in physiological functions compared with the control group (Figure 7B-I). These findings established a medium dose (DHA, 25 mg/kg; ARM, 35 mg/kg) as the maximum safe dose. Subsequently, medium and low doses were administered to the tumor-bearing mice (Cal-27 model). At medium doses of DHA and ARM, tumor (Cal-27 model) proliferation was significantly inhibited, whereas no significant inhibitory effect was observed at the low dose (Figure 7K-P). These results suggest that when administered within a long-term, safe dosage range, artemisinin derivatives can effectively inhibit tumor proliferation.

Figure 7.

Artemisinin derivative safety dose screening and antitumor effects. Mice were administered DHA (50, 25, and 10 mg/kg), ARM (75, 35, and 20 mg/kg), and CMC-NA via gavage daily. Body weight was measured every three days until the end of the experiment, and mouse serum was collected to assess liver and kidney function-related biochemical parameters (n = 3) (B-I). Schematic diagram of the experimental design for safe dose screening of DHA and ARM in mice (A). Body weight change in mice (n = 3) (B). Liver weights of mice (n = 3) (C). Kidney weight of mice (n = 3) (D). Serum alkaline phosphatase (AKP) level (n = 3) (E). Serum alanine aminotransferase (GPT) levels (n = 3) (F). Serum uric acid (UA) level (n = 3) (G). Serum blood urea nitrogen (CRE) levels (n = 3) (H). Serum creatinine (BUN) level (n = 3) (I). Tumor-bearing mice (Cal-27 model) were gavaged daily with DHA (25 and 10 mg/kg), ARM (35 and 20 mg/kg), and CMC-Na, and body weight and tumor size were measured every 3 days until the end of the experiment (n = 4) (J-P). Schematic diagram of the experimental design for DHA and ARM treatment in tumor-bearing (Cal-27) mice (J). Tumor images (K). Representative histological images of tumor sections stained with hematoxylin and Eosin (HE) and Ki67 immunohistochemistry (IHC) (scale bar: 100 μm; n = 4) (L). Graph of changes in tumor size (n = 4) (M). Graph showing changes in mouse body weight (n = 4) (N). Tumor weight (n = 4) (O). Proportion of Ki67 immunohistochemically positive cells in tumor sections (n = 4) (P). The data were expressed as the mean ± standard deviation (B-I, M-P). p values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Fibroblast-to-CAFs transition promotes the progression of OTSCC

To explore the impact of CAFs on OTSCC progression and determine if OTSCC can transform hOMF into CAFs, we developed a tumor model with both hOMF/CAFs and Cal-27 tumor cells implanted together (Figure 8A). The findings indicated that the tumor size in the group where CAFs and Cal-27 were co-implanted was considerably larger than in the group with only Cal-27 (Figure 8B-J). Additionally, beginning on day 15 after implantation, a clear difference in tumor size emerged between groups co-implanted with hOMF/Cal-27 and those with just Cal-27 (Figure 8C). Moreover, lactate concentrations were significantly elevated in the co-implanted groups compared to the Cal-27-only group (Figure 8G). Immunohistochemistry showed substantially higher α-SMA expression in the hOMF/Cal-27 co-implantation group relative to the Cal-27-only group. The fraction of Ki67-positive cells also increased significantly (Figure 8E, 8H, 8I). Additionally, no significant Tunel-positive apoptotic cells were observed (Figure 8E, 8J). These results collectively suggest that fibroblasts (hOMF) and CAFs do not induce apoptosis in tumor cells in vivo. On the contrary, CAFs play an active role in promoting tumor progression, particularly after hOMF are converted into CAFs in the tumor microenvironment, enhancing tumor growth by releasing lactate.

Figure 8.

CAFs promote the progression of oral squamous cell carcinoma. Schematic diagram of the experimental design for co-transplantation hOMF/CAFs with Cal-27 in tumor-bearing mice (A). Tumor pictures (n = 4) (B). Graph of changes in tumor size (n = 4) (C). Graph of changes in mouse body weight (n = 4) (D). Representative histological images of tumor sections (hOMF/Cal-27) stained with hematoxylin and eosin (HE), α-SMA, Ki67, and TUNEL (scale bar: 100 μm; n = 4) (E). Tumor weight (n = 4) (F). Serum lactate levels in the tumor-bearing mice (n = 4) (G). Proportion of α-SMA immunohistochemical staining in tumor sections (n = 4) (H). Proportion of Ki67 immunohistochemically positive cells in tumor sections (n = 4) (I). The proportion of TUNEL negative cells in tumor sections (n = 4) (J). The data are expressed as mean ± standard deviation (C, D, F-J). P values were calculated using a two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Artemisinin derivatives as potential targeted therapies for disrupting tumor-stroma interactions in the OSTCC

To clarify the impact of artemisinin derivatives on modifying the tumor microenvironment (TME), we noted that small doses of DHA and ARM (10 mg/kg and 20 mg/kg) showed no substantial inhibitory effects on tumor growth (Cal-27 model) when compared to the control group (CMC-Na) (Figure 7J-P). Testing these concentrations in the tumor model (Cal-27) affirmed these consistent findings, with no notable alterations in tumor size (Figure 9A-D, 9F). Immunohistochemical evaluation of the tumor samples revealed no significant difference in the ratio of Ki67-positive cells between the groups treated and the control (CNC-Na) (Figure 9E, 9H). Further analysis using Tunel staining showed that there was no effective induction of apoptosis in tumor cells in vivo at this concentration (Figure 9E, 9I). However, at these doses, the level of PDGF-BB in the serum of treated mice was significantly reduced, indicating that DHA and ARM altered the tumor’s secretion behavior in vivo (Figure 9G). In the drug sensitivity experiments, the IC₅₀ values for DHA on Cal-27 cells and hOMF were 27.14 μM and 71.95 μM, respectively, while for ARM, these values were 151.6 μM and 284.8 μM (Figure 9J, 9K). These findings reveal that hOMF cells demonstrate a notably reduced sensitivity to DHA and ARM compared to Cal-27 cells. Therefore, when doses of DHA and ARM at 10 and 20 mg/kg were administered, there was no observed effect on hOMF cell proliferation in tumors co-transplanted with Cal-27. We investigated the antitumor mechanisms of artemisinin derivatives at these low levels by constructing a co-transplantation model combining hOMF with Cal-27. This study showed that DHA and ARM notably impeded tumor progression at doses of 10 mg/kg and 20 mg/kg (Figure 10A-D, 10F). The tumors displayed a pronounced decrease in serum PDGF-BB levels among the mice bearing tumors (Figure 10J). Immunohistochemical analyses indicated a marked downregulation of α-SMA and reduced counts of Ki67-positive cells (Figure 10E, 10G, 10H). TUNEL staining showed no significant positive cells (Figure 10E, 10I). Furthermore, mice in the treatment groups showed significantly lower serum lactate levels (Figure 10K). These outcomes imply that artemisinin derivatives impede tumor development by interfering with interactions between cancer cells and stromal components.

Figure 9.

Artemisinin derivatives inhibit tumor secretion in vivo. Tumor-bearing (Cal-27) mice were orally administered DHA (10 mg/kg) and ARM (20 mg/kg) daily. Mouse body weight and tumor size were measured every three days until the end of the experiment (n = 3). Schematic diagram of the experimental design for DHA and ARM treatments in tumor-bearing (Cal-27) mice (A). Images of tumors (Cal-27) (n = 3) (B). Graph of changes in tumor size (Cal-27) (n = 3) (C). Graph of changes in mouse body weight (Cal-27) (n = 3) (D). Representative histological images of tumor sections (Cal-27) stained with hematoxylin and eosin (HE), Ki67, and TUNEL (scale bar: 100 μm; n = 3) (E). Tumor weight (Cal-27) (n = 3) (F). ELISA measurement of PDGF-BB levels in the serum of tumor-bearing mice (n = 3) (G). Proportion of Ki67 immunohistochemically positive cells in tumor sections (n = 3) (H). The proportion of TUNEL negative cells in tumor sections (n = 3) (I). Dose-response curves for DHA on Cal-27 and hOMF cell proliferation with IC₅₀ values (n = 3) (J). Dose-response curves for ARM on Cal-27 and hOMF cell proliferation with IC₅₀ values (n = 3) (K). The data were expressed as the mean ± standard deviation (C, D, F-K). P values were calculated using a two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Figure 10.

Artemisinin derivatives are potential targeted drugs for interfering with tumor-stroma therapy in OSTCC. hOMF/Cal-27 co-implanted tumor-bearing mice were orally administered DHA (10 mg/kg) and ARM (20 mg/kg) daily. Mouse body weight and tumor size were measured every three days until the end of the experiment (n = 4). Schematic diagram of the experimental design for DHA and ARM treatment in hOMF/Cal-27 co-implanted tumor-bearing mice (A). Tumor pictures (hOMF/Cal-27) (n = 4) (B). Graph of changes in tumor size (hOMF/Cal-27) (n = 4) (C). Graph of changes in mouse body weight (n = 4) (D). Representative histological images of tumor sections (hOMF/Cal-27) stained with hematoxylin and eosin (HE), α-SMA, Ki67, and TUNEL (scale bar: 100 μm; n = 4) (E). Tumor weights of the mice (n = 4) (F). The proportion of α-SMA-positive cells in tumor sections (hOMF/Cal-27) by immunohistochemistry (n = 4) (G). Proportion of Ki67-positive cells in tumor sections (hOMF/Cal-27) by immunohistochemistry (n = 4) (H). The proportion of TUNEL negative cells in tumor sections (n = 4) (I). ELISA measurement of PDGF-BB levels in the sera of tumor-bearing mice (n = 4) (J). Serum lactate levels in the tumor-bearing (hOMF/Cal-27) mice (n = 4) (K). The data were expressed as the mean ± standard deviation (C, D, F-K). P values were calculated using two-sided Student’s t-test (#P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Discussion

PDGF-BB, an important component of the PDGF family, is elevated in numerous cancer types and is intricately linked with cell growth, invasion, and cancer spread [16]. In non-small cell lung cancer (NSCLC), both PDGF-BB and PDGF-R-α are highly expressed and linked to poor prognosis [42]. Our research verified the link between high PDGF-BB expression and unfavorable outcomes in HNSC patients. The PDGF-BB/PDGF-Rβ signaling pathway induces the transformation of pericytes into CAFs and promotes tumor invasion and metastasis [43]. We found that CAFs can originate from NFs, which are directly activated by PDGF-BB secreted by tumor cells, acquiring a characteristic CAF phenotype (e.g., upregulation of α-SMA, changes in cell morphology, and altered secretory functions). The altered secretory function of CAFs directly leads to increased lactate production. Further experiments demonstrated that fibroblasts treated with PDGF-BB- or Cal-27-conditioned media produced supernatants and lactate, which significantly enhanced the proliferative capacity of Cal-27 cells. Notably, unlike immune cells, normal fibroblasts do not possess the ability to kill tumor cells directly. However, when fibroblasts are reprogrammed by the tumor microenvironment and acquire cancer-associated features, they participate in tumor growth and metastasis [44]. Our experiment found that hOMF did not significantly promote tumor proliferation or induce apoptosis in tumor cells. Compared to CAFs (cancer-associated fibroblasts), normal fibroblasts have weaker tumor-promoting effects. In other words, normal fibroblasts might remain neutral in influencing tumor cell biological behavior [45-48]. This provides a theoretical basis for targeting CAF generation to offer anti-tumor strategies.

As a critical component of the tumor stroma, CAF promotes the formation of complex network systems in the tumor microenvironment, which further contributes to tumor invasion and metastasis [9]. Although CAF plays a crucial potential role in cancer therapy, targeted therapy against CAF in the clinic still faces numerous challenges. Current treatment strategies include reducing the number of CAFs, normalizing CAFs, and preventing their production [49,50]. However, despite the development of drugs targeting CAFs based on these strategies, issues such as side effects and poor targeting specificity present limitations [51,52]. Research has demonstrated that artemisinin derivatives exhibit antitumor effects and can block PDGFR/ERK-induced pulmonary fibrosis [22,53]. In fibrotic diseases, the matrix released by fibroblasts does not independently block fibrosis, and the transformation of fibroblasts into CAFs in the tumor microenvironment is mainly irreversible [9,54]. Therefore, exploring the interference of artemisinin derivatives in fibroblast-to-CAF transformation may provide a theoretical direction for CAF-targeted therapy. Our experiments showed that DHA and ARM inhibited fibroblast-to-CAF transformation by suppressing PDGF-BB secretion from cancer cells, maintaining normal fibroblast function. Moreover, DHA and ARM inhibit tumor proliferation by maintaining normal fibroblast function and exert direct inhibitory effects on tumor cells (Figure 11). We further explored the effects of artemisinin derivatives on CAF deactivation and PDGF-BB signaling blockade, and the results indicated that artemisinin derivatives had no significant impact. These results suggest that artemisinin derivatives primarily regulate tumor cells rather than acting directly on CAFs. This phenomenon can be attributed to the more pronounced inhibitory effect of artemisinin derivatives on tumor cells than on normal cells [29]. We also found that hOMF cells were less sensitive to artemisinin derivatives than Cal-27 cells were during drug concentration screening. The selective cytotoxicity of artemisinin derivatives toward tumor cells minimizes their impact on normal cells, offering a potential research direction to address the current challenges of drug side effects in tumor-targeted therapies.

Figure 11.

Artemisinin derivatives maintain fibroblast normalization by targeting tumor cells (Graphical abstract).

The safety of artemisinin derivatives as frontline antimalarial drugs has been widely recognized [55]. Studies have indicated that the drug concentrations required for antitumor and antimalarial effects differ significantly [41]. To address this, we determined safe dosages of DHA (dihydroartemisinin) and ARM (artemether) in BALB/c nude mice. Using a BALB/c nude mouse model of OTSCC, we observed that artemisinin derivatives, at safe dosages, significantly inhibited the progression of OTSCC. To further investigate the tumor burden induced by CAFs, we used a co-transplanted tumor model combining OTSCC cells and CAFs in BALB/c nude mice. This model revealed that CAFs significantly enhanced tumor volume. Further analyses demonstrated that artemisinin derivatives inhibited the secretion of PDGF-BB by cancer cells, normalizing fibroblasts and suppressing tumor progression in vivo. In addition, artemisinin derivatives have been found to inhibit tumor progression by directly suppressing the proliferation of tumor cells. Unlike conventional therapies that target stromal cells, artemisinin derivatives exert their effects by directly targeting tumor cells, thereby influencing tumor-stroma interactions. This approach differs from current drug development strategies aimed at stromal cells, which often focus on mechanisms involving CAF-specific markers, such as fibroblast activation protein (FAP), or signaling pathways involving cytokines, such as transforming growth factor beta (TGF-β) and epidermal growth factor (ERGF) [56-58]. However, the regulatory role of tumor cells within the microenvironment appears to be overlooked, which may contribute to the clinical bottleneck faced by targeted therapies [59].

Artemisinin derivatives exhibit high selectivity for specific proteins in their biological activity. Network pharmacology studies have found that artemisinin derivatives can tightly bind to AKT1, EGFR, and NFKB1, exerting cytotoxic effects [60]. Related research indicates that artemisinin derivatives target and inhibit heat shock protein 90 (Hsp90), suppressing AKT and ERK signaling and contributing to their anti-tumor activity [61]. In gastric cancer studies, DHA was found to downregulate fibroblast growth factor 2 (FGF2), thereby inhibiting angiogenesis [62]. Further research has shown that as suitable ferroptosis inducers, artemisinin derivatives can induce tumor ferroptosis through multiple targeted pathways. Specifically, artemisinin derivatives induce tumor ferroptosis by modulating the ERK signaling pathway and downregulating the STAT3 signaling pathway [63,64]. They can also trigger ferroptosis in primary hepatocellular carcinoma cells by upregulating CHAC1 [65]. Moreover, artemisinin compounds regulate iron homeostasis, making cancer cells more sensitive to ferroptosis [22]. These findings highlight the immense potential of artemisinin derivatives in cancer therapy. Studies suggest that artemisinin derivatives can improve polycystic ovary syndrome (PCOS) by directly targeting Lon protease 1 (LONP1) [66]. Artemisinin has also been reported to induce ferroptosis by regulating Atf3 or overexpressing Rock1, thereby reversing ovarian fibrosis (HF), further underscoring its potential as an anti-fibrotic agent [67].

Artesunate is a semisynthetic, water-soluble compound derived from the traditional Chinese medicine artemisinin, widely used in clinical settings to treat malaria [68]. A study involving 1,391 French patients showed a mortality rate of 4.1% among malaria patients treated with Artesunate, with 2% of the cases reporting adverse events related to death [69]. Currently, research on the clinical application of artemisinin derivatives in cancer therapy is limited. Still, small sample studies suggest that artemether does not exhibit significant side effects in the treatment of metastatic breast cancer [70]. Another study found that oral artemisinin effectively alleviated symptoms in patients with advanced metastatic cervical cancer [71]. However, it is essential to note that this conclusion does not rule out the potential side effects of artemisinin derivatives in cancer treatment. In a large number of clinical samples for antimalarial treatment, artemisinin showed eosinophilia during short-term treatment, but this side effect is reversible after discontinuation of the medication [72]. Encouragingly, some studies suggest that artemisinin derivatives can selectively kill tumor cells while offering some protective effects on normal cells [73]. Therefore, further exploration into the protective mechanisms of artemisinin derivatives on normal cells could provide valuable insights for reducing clinical side effects. Additionally, bioengineering materials may offer new possibilities for the clinical use of artemisinin derivatives.

In summary, this study found that artemisinin derivatives exert a dual anti-tumor effect by directly regulating or inhibiting tumor cell proliferation, providing a theoretical basis for developing targeted drugs based on artemisinin derivatives. Although this study offers valuable insights, it also has some limitations. The tumor microenvironment is a highly dynamic system that involves processes such as angiogenesis, immune cell infiltration, the formation of pre-metastatic niches, and extracellular matrix remodeling [74-77]. These components collectively shape the tumor’s biological characteristics and clinical manifestations. While indirect co-culture systems can somewhat simulate the signaling between different cell populations through solution exchange, they cannot replicate the direct contact and interactions between tumor cells and other cells. They cannot encompass the heterogeneity of CAFs (cancer-associated fibroblasts) in vivo [78]. This limitation restricts the accurate reproduction of complex cell-to-cell interactions within the tumor microenvironment. Although this study partially compensates for the shortcomings of indirect in vitro culture through in vivo models, tumors are complex three-dimensional structures composed of various cell types. Future research will employ more sophisticated 3D co-culture systems and tumor organoids to overcome these limitations, allowing for a more realistic simulation of the tumor microenvironment. Additionally, the sample size used in each animal model group in this study is relatively small, which may reduce the generalizability of the results. In the future, we will overcome the sample size limitation to obtain more robust evidence and further validate the efficacy of artemisinin derivatives in cancer.

Disclosure of conflict of interest

None.