In this study, we chose the M2 macrophage as the in vitro study object of TAMs. The steps of polarization are as follows: human monocytic cell line THP-1 (obtained from Type Culture Collection of the Chinese Academy of Sciences, Shanghai, China, passage number: 5–25) was stimulated with 50 mM phorbol 12-myristate 13-acetate (PMA, Beyotime, Haimen, China) for 24 h, and then differentiated into classical M2 macrophages by stimulation with 20 ng/mL interleukin-4 (IL-4) and interleukin-13 (IL-13, PeproTech, NJ, USA) for 72 h.

First, prepare the non-adherent coating culture plate: add agarose powder to an appropriate amount of cell culture medium, and heat in a water bath at 80°C for 30 min until fully dissolved. After dissolution, quickly add the agarose solution to the 96-well plate at 60 μL per well, and allow the agarose to solidify for approximately 30 min. Subsequently, add the cell suspension pre-mixed with Matrigel matrix gel into the agarose-coated 96-well plate, and centrifuge at 4°C, 1,000 × g for 10 min. After 24 h of culture, observe the formation of spherical cell structures under a microscope.

A cell suspension of tumor cells pre-stained with the green cell membrane dye DiO was added to a 96-well plate coated with agarose to prepare tumor microballs. After a 24-h incubation, monocytes (THP-1) pre-stained with the orange-red cell membrane dye Dil were introduced at a ratio of 1:4. Following a 7-day co-culture period, the tumor microballs were transferred to a new 96-well plate. After washing three times with PBS, fluorescence microscopy was performed for observation.

For 2D cultures, 5 × 10³cells/well were seeded in 96-well plates. After 24 h attachment, cells were treated with gefitinib in either standard or FMD medium. CCK-8 reagent was added at 24 and 48 h. Absorbance at 450 nm was measured after 2 h incubation using a microplate reader. For 3D spheroids, Spheroids were washed with PBS (3×), dissociated into single cells. Cell viability was measured via CCK-8 after 4 h incubation.

THP-1 cells are seeded in the upper chamber of the Transwell and treated with PMA, IL-4, and IL-13 to differentiate THP-1 into M2-type TAMs. Next, tumor spheroids cultured for 7 days are transferred to the lower chamber of the Transwell and co-cultured with TAMs for 48 h. At the end of the experiment, any remaining cells on the surface of the upper chamber of the Transwell are removed using a cotton swab. The cells are then fixed with cold methanol-acetic acid, stained with crystal violet, and counted.

The in vitro fasting-mimicking diet medium was prepared following established protocols (Qian et al., 2022; Huang et al., 2023). The Fasting-mimicking diet medium consists of glucose-free medium supplemented with 0.5 g/L glucose and 5% FBS.

Before starting the study, the mice were fed the standard AIN-93G diet for 2 weeks, and their daily energy intake was monitored twice a week to establish a baseline. The FMD cycles consisted of a 'Day 1′and 'Day 2–3′ diet each week, with 10% animal-based protein and 90% fat. These diets provided 50% and 10% of the total energy intake from the baseline feeding, with a return to the normal diet at all other times.

BALB/c nude mice (5 weeks old) were purchased from Vital River Laboratories (Beijing, China). Cal-27 cells (1 × 10⁷/100 µL/each mouse) were injected into the armpits of each mouse. When the tumor volume reached 100–200 mm³, mice with excessively large or small tumors were excluded. The remaining mice were then randomly divided into treatment groups (n = 5): control; FMD; gefitinib (75 mg/kg/day, oral gavage (i.g.), MedChemExpress, NJ, USA); FMD + gefitinib (75 mg/kg/day, i.g.).

GSVA was conducted using the R package “GSVA” (version 1.52.5) implemented in R version 4.3.0 to assess the correlation of genes with specific gene sets. The gene sets used for this analysis were obtained from the Gene Set Enrichment Analysis (GSEA) database (https://www.gsea-msigdb.org).

Quantitative data were presented as mean ± SD, and group comparisons were made using one-way analysis of variance (ANOVA) with Tukey’s post hoc test. The statistical analyses were conducted using SPSS/Win 13.0 software (SPSS Inc., Chicago, USA).

We investigated whether FMD medium enhances the effects of the EGFR inhibitor gefitinib on oral cancer in both 2D and 3D cell spheroid models. The results of the CCK-8 assay revealed that treatment with gefitinib in the 2D cell model led to a decrease in cell viability of Cal-27 cells in a dose-dependent and time-dependent manner (Figure 1A). Notably, with the survival rate of the combination group was normalized to the FMD-only group (not the control), gefitinib exhibited a markedly improved anti-tumor effect in the FMD-conditioned medium environment. Furthermore, similar observations were made in OECM-1 cells, where the combination of gefitinib and FMD demonstrated a synergistic effect (Figure 1B).

To further validate these findings, we constructed 3D tumor spheroids from both Cal-27 and OECM-1 cells and evaluated their proliferative capacity by CCK-8 assay. Consistent with the results from the 2D cell culture experiments, the combination of gefitinib and FMD significantly inhibited the growth of 3D tumor spheroids compared to either treatment alone (Figures 1C–F). These findings suggest that FMD acts synergistically with gefitinib to suppress the growth of oral cancer cells in vitro.

To investigate molecular mechanisms underlying the interaction between fasting-mimicking diet (FMD) and gefitinib, we conducted a multi-step bioinformatics analysis. First, cross-examination of Gene Expression Omnibus (GEO) datasets identified genes upregulated by gefitinib (GSE75307, GSE122005) and downregulated under nutrient deprivation (GSE62663). Venn analysis revealed CCL2 as the only overlapping candidate (Figure 2A), suggesting its potential role in mediating FMD-gefitinib crosstalk. This suggests that FMD interventions might reduce the gefitinib-induced upregulation of CCL2. Notably, CCL2 ranked among the top 10 upregulated genes in both gefitinib-treated datasets, indicating its consistent dysregulation under gefitinib treatment (Figure 2B). To further characterize CCL2’s functional relevance, protein-protein interaction (PPI) network analyses of gefitinib-upregulated genes from GSE75307 (Figure 2C) and GSE122005 (Figure 2D) consistently identified CCL2 as a central hub protein (Figure 2E). This network prominence implies that CCL2 may coordinate multiple downstream pathways activated by gefitinib.

Clinically, while GSVA scoring showed an inverse correlation between the gefitinib-induced gene set and poor prognosis in head and neck cancer (The Cancer Genome Atlas (TCGA) data, Figure 2F), CCL2 exhibited a distinct pattern: its elevated expression significantly associated with worse survival outcomes (Figure 2G) and enhanced macrophage infiltration (Figure 2H). This paradoxical observation suggests that although the broader gefitinib response signature may confer protective effects, CCL2 specifically could drive tumor-associated macrophage (TAMs) recruitment, potentially counteracting therapeutic efficacy by remodeling the tumor microenvironment.

To validate these computational predictions, we performed functional experiments in oral cancer cell models. LISA and RT-PCR assays demonstrated that gefitinib dose-dependently reduced CCL2 secretion in Cal-27 tumor spheroids (Figure 2I), while FMD significantly reversed gefitinib-induced CCL2 suppression at both protein (Figure 2I) and mRNA levels (Figure 2J). Similar trends were observed in OECM-1 cells (Figure 2K), confirming the inhibitory effect of FMD on gefitinib-mediated CCL2 modulation. These results mechanistically support the hypothesis that FMD may suppress gefitinib-induced TAMs recruitment by specifically targeting CCL2 production.

To functionally validate the CCL2-TAMs recruitment axis identified above, we developed a 3D Cal-27 tumor spheroid-macrophage co-culture system (Figure 3A). Quantitative fluorescence imaging demonstrated that gefitinib treatment significantly enhanced TAMs infiltration into tumor spheroids compared to controls (Figures 3B,C). Notably, both the neutralizing antibody targeting CCL2 and the FMD medium significantly inhibited gefitinib-induced macrophage recruitment and concurrently suppressed tumor spheroid growth. This functional interdependence between immune microenvironment remodeling and tumor growth kinetics suggests CCL2-mediated chemotaxis constitutes a critical resistance mechanism.

To dissect the mechanism underlying TAMs-mediated therapeutic resistance, we established a paracrine communication model using conditioned media exchange. Tumor spheroids exposed to TAMs-derived conditioned media exhibited significantly reduced gefitinib sensitivity (Figure 3D), indicating that TAMs-secreted factors confer drug resistance. Additionally, Transwell co-culture experiments showed that conditioned medium of gefitinib-treated Cal-27 significantly increased the recruitment of TAMs, while the CCL2 neutralizing antibody effectively countered this promotion (Figure 3E). Importantly, control experiments confirmed that gefitinib does not directly stimulate TAMs migration (Figure 3F).

Our integrated findings establish a feedforward resistance loop: Gefitinib treatment paradoxically enhances tumor cell CCL2 secretion, driving TAMs infiltration that subsequently secretes protective factors to diminish drug efficacy. FMD intervention disrupts this pathogenic cycle through dual mechanisms - directly impairing tumor cell CCL2 production and indirectly modulating TAMs effector functions - thereby resensitizing malignancies to EGFR-targeted therapy.

To evaluate the translational potential of our findings, we employed a Cal-27 xenograft model, selected for its sensitivity to EGFR signaling (Hu et al., 2024), to recapitulate disease progression. Treatment with FMD for a total of three cycles resulted in a notable reduction in tumor volume compared to the control group. Additionally, mice treated with a combination of FMD and gefitinib (75 mg/kg/day) showed a further decrease in tumor volume compared to those treated with gefitinib alone (Figures 4C–E).

Histopathological analysis revealed that combined therapy induced tumor stromal remodeling, evidenced by 74% reduction in Ki-67+ proliferating tumor cells (Figure 4F) and 65% decrease in CD163+ M2-like TAMs density (Figure 4G). Notably, ELISA results further demonstrated that FMD treatment significantly lowered the serum CCL2 levels in the gefitinib-treated mice (Figure 4H).

To elucidate the molecular mechanisms underlying FMD-mediated suppression of gefitinib-induced CCL2 expression, we employed a multi-modal analytical approach. Gene Set Enrichment Analysis (GSEA) of TCGA-HNSC patient data indicated that CCL2-related gene pathways include the STAT signaling pathway (Figures 5A,B), prompting subsequent prioritization of STAT3 over c-Jun for mechanistic investigation. Furthermore, gene set variation analysis (GSVA) demonstrated a positive correlation between CCL2 and gene sets associated with EGF, macrophages, gefitinib, and genes induced by gefitinib from datasets GSE75307 and GSE122005. Conversely, a negative correlation was observed with gene sets related to glucose starvation in head and neck squamous cell carcinoma samples (Figure 5C). These multi-omics findings collectively suggest STAT3 phosphorylation as a critical mediator of gefitinib-induced CCL2 upregulation.

To validate the above hypothesis, we observed the effects of gefitinib, FMD, and their combination on the phosphorylation of STAT3 in both oral cancer cell lines Cal-27 and mouse tumor tissues. Western blot results (Figures 5D,E) showed that at the cellular level and in mouse tumor tissues, gefitinib can induce STAT3 phosphorylation, while FMD inhibits gefitinib-induced STAT3 phosphorylation. Additionally, in the in vitro cell model, ELISA results indicated that the gefitinib-induced secretion of CCL2 was diminished by the STAT3 inhibitor Stattic, while the STAT3 agonist ML115 enhanced CCL2 secretion (Figure 5F). These dose-responsive, reciprocal effects confirm STAT3 activation status as a critical determinant of CCL2 expression levels in the combined effects of FMD and gefitinib.