Airway microbiota associated D-phenylalanine promotes non-small cell lung cancer metastasis through epithelial mesenchymal transition

Abstract

Background

Lung cancer is the leading cause of cancer-related death worldwide, and patients with distant metastasis have a poor prognosis. Various studies have reported that microbiota and metabolites significantly differ between healthy individuals and lung cancer patients. However, the effects of metabolites on tumor formation and metastasis are unclear. Therefore, our study aimed to determine the correlation between airway metabolites and microbiota, along with their respective roles in lung cancer metastasis.

Methods

Bronchoalveolar lavage fluid (BALF) samples were collected from 30 non-small cell lung cancer (NSCLC) patients, including 11 patients without metastasis (M0) and 19 patients with metastasis (M1). Integrated pathogenic metagenomic and Liquid chromatography-mass spectrometry (LC‒MS) analyses were employed to explore differences between two groups. The omics data were analyzed and integrated via Spearman’s correlation coefficient. Specific metabolites were subsequently used to intervene in lung cancer cells and animal models to assess their influence on tumor metastasis.

Results

A total of 801 metabolites were identified in the BALF of all patients. Compared with those in the M0 group, 48 metabolites in the M1 group were significantly different. D-phenylalanine was notably upregulated in M1 and was positively related to Metamycoplasma salivarium. Intranasal administration of D-phenylalanine promoted tumor intrapulmonary metastasis and induced epithelial mesenchymal transition (EMT) process in NSCLC mouse models. Moreover, D-phenylalanine promotes the proliferation of non-small cell lung cancer cells and facilitates their migration and invasion via EMT.

Conclusion

The airway microbiota associated D-phenylalanine could promote lung cancer metastasis via EMT, which could be a new predictor for the diagnosis of tumor metastasis in NSCLC patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-06701-1.

Introduction

Lung cancer is one of the most malignant tumors with the highest morbidity and mortality rates worldwide and is the leading cause of cancer death in China [1, 2]. Lung cancer metastasizes insidiously, leading to a poor prognosis [3]. Unfortunately, more than 30% of Chinese NSCLC patients have distant metastasis at the time of diagnosis, rendering them ineligible for surgery [4]. The identification of new diagnostic markers or therapeutic targets is essential to improve the prognosis of lung cancer patients.

There are billions of microorganisms distributed in the intestines, skin, vagina and respiratory tract, which are also known as “hidden organs“ [5]. Microbiota dysbiosis is closely related to various diseases, such as cardiovascular diseases, tumors, diabetes mellitus, obesity and cancer [5]. Studies have shown that the lung microbiota can promote inflammation and cell proliferation by stimulating myeloid cells and lung Vγ6⁺ Vδ1⁺ γδT cells to produce IL-1β, IL-23, IL-17 A and other effectors [6]. In addition, the increased abundance of Veillonella parvula in lower respiratory tract is strongly associated with poor prognosis in patients with lung cancer [7]. Lung cancer patients with a history of broad-spectrum antibiotic use have shorter metastasis-free survival, which may result from the impairment of T-cell immune function in NSCLC patients, suggesting that dysbiosis of lung microbiota may lead to the metastasis of lung cancer [8]. Compared with that in normal lung tissue, the alpha diversity of the microbiota in tumor tissues was significantly lower. In addition, the abundances of the genera Thermus and Legionella are greater in the advanced stage [9]. A study revealed that the microbiota in the sputum and BALF of NSCLC patients was correlated with pathological type, clinical stage and tumor metastasis [10]. Unfortunately, comprehensively elucidating the mechanism by which the airway microbiota affects metastasis in patients with lung cancer is difficult.

Metabolomics is usually applied for lung cancer risk assessment, therapy efficacy and prognosis prediction. Serum metabolic profiles, including amino acid, lipid and organic acid, significantly differ between healthy individuals and NSCLC patients [11, 12]. Significant changes in gut microbes and associated metabolites have been observed in both lung cancer patients and healthy controls, as well as before and after antitumor therapy [13–16]. In a Lewis lung cancer mouse model, intestinal Akkermansia muciniphila exerts anticancer effects by regulating the reprogramming of glycolytic metabolism, glutamine metabolism, and purine and pyrimidine metabolism in mice through interactions with the tumor microbiota [17].

Ginseng polysaccharide enhances the antitumor effect of PD-1 monoclonal antibody by modulating the ratio of the microbial metabolite kynurenine/tryptophan and increasing the ratio of CD8⁺/CD4⁺ T cells in peripheral and tumor tissues [18]. Various differentially abundant metabolites, which could be potential biomarkers for lung cancer diagnosis, have been identified in BALF [19, 20]. In another study, BALF samples from tumor-loaded lung segments and paracancerous tissues of 28 patients revealed a decrease in microbiota diversity in lung tumor-loaded segments, where short-chain fatty acids were positively correlated with Brachyspira hydrosenteriae and negatively correlated with the genus Pseudomonas [21]. In summary, lung microbiota and metabolites may play key roles in the pathogenesis of lung cancer and are expected to serve as biomarkers and therapeutic targets for lung cancer.

Phenylalanine is an essential amino acid that the human body cannot synthesize. Disorders of amino acid metabolism are associated with various diseases, including several metabolic diseases, cardiovascular and cerebrovascular diseases, immune diseases and tumors [22]. Amino acids are classified into two types, the D and L isomers, which can be transformed reversibly [23]. A study reported that some free D-amino acids were significantly increased in the intestines of SPF mice compared with those of germ-free mice, which suggested that D-phenylalanine, a subtype of D-amino acid, may be related to the microbiota [24]. Phenylalanine, D-phenylalanine, phenylacetylglutamine and phenylalanine metabolism are significantly elevated in the mortality subgroup of patients with acute respiratory distress syndrome [25]. However, the role of D-phenylalanine in NSCLC metastasis remains unclear.

In summary, serum, sputum, or lung tissue metabolites are closely associated with the clinical features of patients with lung cancer, as well as BALF. In our previous study, we detected microbiomes in BALF of NSCLC patients via 16S rRNA sequencing and reported that the composition of the airway microbiome in BALF was significantly different between the M0 and M1 groups [10]. However, how airway microbiota affects tumor metastasis remains unknown. In this study, we integrated pathogenic metagenomic and nontargeted metabolomic sequencing technologies to elucidate the differences in airway metabolites and microbiota between metastatic and nonmetastatic NSCLC patients. Further analysis revealed that D-phenylalanine was positively correlated with airway microbiota and may be a predictive biomarker of lung cancer metastasis and a potential therapeutic target in NSCLC.

Materials and methods

Patient samples

30 BALF samples from NSCLC patients were collected from the Department of Respiratory and Critical Care Medicine of Nanfang Hospital from July 2022 to December 2022. The exclusion criteria for patients were as follows: (1) Patients with acute oral or lung infection, acute bronchitis, acute exacerbation of COPD, or acute exacerbation of bronchial asthma within 4 weeks; (2) History of use of broad-spectrum antibiotics, oral hormones, immunosuppressants or microbiological agents within 4 weeks; (3) Combination of other malignant tumors or previous chemotherapy, radiotherapy, targeted therapy, anti-vascular therapy or immunotherapy for lung cancer; (4) Patient with metabolic or autoimmune diseases. All procedures in this study were carried out under the ethical protocols of the Ethics Committee of Nanfang Hospital (NFEC-2022-313, Guangzhou, China).

Clinical sample sequencing

The BALF sample sequencing protocol is described in the Supplementary Methods.

Cell culture

The mouse NSCLC cell line Lewis lung cancer cell (LLC) and the human NSCLC cell lines A549 and H1299 were maintained in our laboratory. A549 and H1299 cells were cultured in RPMI-1640 (Gibco, USA) supplemented with 10% fetal bovine serum (NEWZERU, New Zealand) and 1% penicillin‒streptomycin (Gibco, USA). LLC cells were cultured in DMEM (Gibco, USA) supplemented with 10% FBS and 1% penicillin‒streptomycin solution. All the cells were cultured in an incubator at 37 °C with 5% CO₂.

Mouse models

All animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Nanfang Hospital, and mouse studies were carried out at the Experimental Animal Center of Nanfang Hospital. To create a metastatic model, 1 × 10⁶ LLC-luciferase cells were resuspended in 150µL phosphate buffer saline (PBS) and injected into the tail vein of C57BL/6J mice (SPF grade, 4–5 weeks old, female). After tumor cell injection, the mice were randomly divided into 3 groups (n = 5 per group): the control group, 0.5 mg group and 1 mg group. D-phenylalanine was given by intranasal administration. After 3 weeks, the mice were injected with D-luciferin (LUCK-1G, Goldbio, USA) intraperitoneally. The luciferase signal was recorded via an in vivo imaging system (Spectral Instruments Imaging, United States). All the animals were subsequently euthanized. To further identify the characteristics of lung metastasis, tumor tissues were embedded for histochemical and immunohistochemistry evaluation.

In vivo Dose Selection: To determine the appropriate doses of D-phenylalanine in mice, we initially tested 6 different doses (0, 0.1, 0.25, 0.5, 1, 2 mg/mouse/day). The results showed that 0.5 mg and 1 mg group exhibited more tumor colonization in lung, while the 2 mg group had 80% mortality in 2 weeks, indicating a high level of toxicity. Based on the pre-experiment, we selected 0.5 mg and 1 mg as the final doses.

Cell counting Kit-8 (CCK-8) assay

A CCK-8 (EK-5103, ECOTOP, China) was used to detect cell viability and D-phenylalanine dose. The cells were seeded into 96-well plates at a density of 2 × 10³ cells per well with different concentrations (0 µM, 10 µM, 100 µM and 1000 µM) of D-phenylalanine. At 24 h and 48 h after the cells were plated, 100 µL of freshly prepared CCK-8 solution was added to each plate, and the cells were incubated at 37 °C for 2 h. A microplate reader (TECAN, Switzerland) was then used to measure the optical density (OD) at 450 nm. It showed 10 and 100 µM group promoted cell proliferation compared with control, whereas cell proliferation in the 1000 µM group was less than 100 µM group. Therefore, we chose 0, 10 and 100 µM for subsequent in vitro experiment.

EdU assay

Cell proliferation ability was measured with a BeyoClick™ EdU Cell Proliferation Kit with Alexa Fluor 594 (C0078S; Beyotime Biotechnology, Shanghai, China). After the cells were washed with PBS 3 times, EdU solution was used to incubate them for 2 h. After incubation, the fixed cells were washed with PBS and stained with Alexa Fluor 594 according to the manufacturer’s instructions. Images were captured with an inverted Olympus IX73 Inverted Microscope.

Wound healing assay

Single A549 and H1299 cell suspensions were prepared and seeded into 6-well plates overnight to reach 100%. A sterile pipette tip was used to scratch a line in 6-well plates. The cell debris was washed away with PBS, and serum-free medium supplemented with different concentrations of D-phenylalanine was added. The scratch area was captured at 0 h and 24 h. Tound area was measured using Image J software, and the percentage of area recovery was calculated.

Migration and invasion assay

Transwell chambers (Costar, Corning, United States) with a pore size of 8.0 μm were used for the cell migration and invasion assays. Transwell membranes were first hydrated with basal medium. The Matrigel (356234, Corning, USA) was diluted with basal medium at a ratio of 1:9. Then, the Matrigel was spread on the upper chamber surface and polymerized into a gel state after 15 min at 37 °C. No matrix gel was required for testing cell migration ability. The cells were starved for 24 h, digested and resuspended in serum-free medium at a density of 5 × 10⁵/ml containing different concentrations of D-phenylalanine. Then, add 100uL into the upper chamber. 600ul medium containing 10% FBS was added to the lower chamber. Transwell chambers were fixed in 4% paraformaldehyde methanol for 30 min and stained with Giemsa stain for 10 min. The upper chambers of non-migrated cells were gently wiped off with a cotton swab. The images were taken by a microscope and analysed by Image J.

Western blot analysis

NSCLC cells were collected, washed with PBS 3 times, and incubated with RIPA lysis buffer at 4 °C for 30 min to extract total protein. The samples were then transferred to EP tubes and centrifuged at 4 °C at 12,000 × g for 15 min. The supernatant was removed, the appropriate amount of loading buffer was added, and the mixture was heated at 100 °C for 10 min. Separate the protein samples by SDS-PAGE gel electrophoresis, then transfer them onto PVDF membranes (Merck Millipore, USA), blocked with 5% BSA, and incubated with the primary antibody overnight at 4 ℃. The membranes were incubated with secondary antibodies (RS23920, LICOR, USA) for 1 h, and the fluorescence intensity was detected with an Odyssey imaging system (LICOR, USA). Antibodies were used for the experiment: E-cadherin (3195, CST, USA), N-cadherin (22018-1-AP, Proteintech, China), Vimentin (10366-1-AP, Proteintech, China), Snail (13099-1-AP, Proteintech, China) and Twist (25465-1-AP, Proteintech, China).

Immunofluorescence assay

NSCLC cells were seeded into round coverslips in 12-well plates. After 24 h, remove medium and wash cells by PBS 3 times. Then fix cells by 4% paraformaldehyde for 30 min and then washed 3 times with PBS. The cells were blocked with 5% bovine serum albumin (BSA) for 30 min. The cells on the round coverslip were incubated with E-cadherin (3195, CST, USA) and N-cadherin (22018-1-AP, Proteintech, China) overnight at 4℃. After being washed by PBS 3 times, secondary antibodies (SA00013-4, proteintech, China) were added, and the samples were incubated for 2 h at room temperature. Finally, the cells were stained with DAPI to stain the nuclei, the coverslips were inverted on slides, and the fluorescence intensity was observed via fluorescence microscopy.

H&E staining

Paraffin-embedded mouse lung tissues were cut into 4 μm sections on slides. Heat slides for 2 h at 65℃ in a dry oven to remove paraffin. Then, the tissues were deparaffinized in xylene and rehydrated through a graded alcohol series. The tissue sections were stained with hematoxylin and eosin. The slices were photographed with Olympus light microscope.

Immunohistochemical staining (IHC)

Paraffin-embedded mouse lung tissues were deparaffinized and rehydrated as previously described. Submerge slides in citric-acid buffer and maintain a continuous boil for 20 min. Cool at room temperature and wash 3 times by PBS for 5 min. The tissues were covered with endogenous peroxidase blockers (PV-9001, ZSGB-BIO, China) for 10 min. Then, the tissues were incubated overnight with E-cadherin (3195, CST, USA), N-cadherin (22018-1-AP, Proteintech, China), Vimentin (10366-1-AP, Proteintech, China) and Ki67 (28074-1-AP, Proteintech, China) at 4℃, reaction strengthening fluid for 20 min, and secondary antibodies for 2 h in at room temperature. Freshly prepared DAB (ZLI-9018, ZSGB-BIO, China) solution was added to the tissues to detect positive signals, and the nuclei were counterstained with hematoxylin. All the antibodies were diluted at a ratio of 1:100. Images were captured with an Olympus microscope.

Statistical analysis

The experimental results were statistically analyzed via GraphPad Prism 9.5 (San Diego, CA, USA). The data are expressed as the mean ± standard deviation (SD). Comparisons between groups were made using the unpaired two-tailed Student’s test. Data between multiple groups were compared via one-way analysis of variance (ANOVA). Patient characteristic data were tested using Fisher’s exact test. P values < 0.05 were considered to indicate significant differences.

Results

Characterization of clinical data for NSCLC patients

We recruited 30 NSCLC patients, 11 with nonmetastatic disease and 19 with metastatic disease, who were diagnosed after a biopsy at Nanfang Hospital (Guangzhou, China) from July 2022 to December 2022. The clinical characteristics of the 30 NSCLC patients included in the study, including age, sex, smoking history and histology type are shown in Table 1. Patients with pleural/pericardial nodules, pleural/pericardial effusion, contralateral/bilateral pulmonary nodules or distant (extrathoracic) metastasis were classified into the M1 group [26]. The other patients were in the M0 group. No significant differences in baseline were observed between the two groups.

Table 1.

The characteristic of age, gender, histology, and smoking history between the M0 and M1 groups of NSCLC patients

Metabolites in the BALF were significantly altered in the M1 group

Through LC‒MS/MS detection, we ultimately identified 405 metabolites in positive ion mode and 396 metabolites in negative ion mode from 30 samples. The PLS-DA analysis revealed distinct metabolic clusters between the M0 and M1 groups in positive and negative ion modes (Fig. 1A), which revealed that the metabolic patterns were remodeled in M1 patients. A total of 48 metabolites were significantly altered in M1 compared with M0, among which 32 metabolites were increased in positive ion mode and 16 were increased and 3 were decreased in negative ion mode (Fig. 1B). In terms of FC, D-phenylalanine was the most significantly upregulated metabolite in the M1 group in positive ion mode (Fig. 1C), which implied that D-phenylalanine plays an important role in tumor metastasis. The heatmap revealed that the M1 group had a different metabolic pattern from the M0 group on the basis of 32 metabolites in positive ion mode and 16 metabolites in negative ion mode. (Fig. 1D). The results of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed that the differentially abundant metabolites in M0 and M1 patients were enriched mainly in phenylalanine metabolism (Fig. 1E), including phenylacetylglycine, D-phenylalanine and phenylpyruvic acid. In addition, the levels of the derived metabolites phenylalanine, homovanillic acid and epinephrine were also significantly increased in the M1 group (Fig. 1E). In summary, untargeted metabolomics revealed that the metabolic pattern in the M1 group was obviously altered, among which D-phenylalanine and phenylalanine metabolism may have potential clinical relevance in NSCLC metastasis.

Fig. 1.

Metabolite levels in the BALF were significantly altered in the M1 group. (A) PLS-DA plot of airway metabolites in positive (pos) and negative ion modes (neg); red: nonmetastasis group (M0); green: metastasis group (M1). (B) Volcano plot of differentially abundant metabolites in pos and neg ion modes. The horizontal coordinate is log₂ (fold change), and the vertical coordinate is -log₁₀(P). The size of the dot represents the VIP value. Red represents upregulation, and green represents downregulation. (C) Stem plots of the top 20 differentially abundant metabolites in pos and neg ion modes. The horizontal coordinate represents log₂(fold change), and the vertical coordinate represents the name of the metabolite. The size of the dot represents the VIP value. (D) Heatmap of differentially abundant metabolites in pos and neg ion mode; red: high expression; blue: low expression. (E) KEGG pathway enrichment of differentially abundant metabolites in pos and neg ion modes. The horizontal coordinate represents the degree of enrichment of differentially abundant metabolites; the color of the bubble represents the P value, and the size represents the number of differentially abundant metabolites in the pathway

Alterations in BALF metabolites are associated with the airway microbiota

As demonstrated previously, alterations in metabolites might be associated with the microbiota and its metabolic pathways. We compared the composition of the lung microbiota between M0 and M1 at the phylum and genus levels (Fig. 2A). The relative abundances of the top 20 species are shown in Fig. 2B. In terms of microbiota community diversity, no significant difference in alpha diversity was detected between the M0 and M1 groups (Fig. 2C). However, a significantly different microbiota composition was observed between the two groups via principal component analysis (Fig. 2D). LEfSe revealed the dominant microbiota in the M0 group (Fig. 2E). To further disentangle the interplay between the microbiota and metabolites in BALF, we correlated the significantly differentially abundant microbial species and metabolites in M1 group compared with M0 group (Fig. 2). D-phenylalanine was most significantly upregulated in the M1 group and was positively correlated with the abundance of Metamycoplasma salivarium (Fig. 2F). D-phenylalanine was upregulated 38-fold in the M1 group (Fig. 2G). The area under the ROC curve (AUC) of D-phenylalanine was 0.708, suggesting that D-phenylalanine has the potential for the prediction of NSCLC metastasis (Fig. 2H). These results further support the associations between the lung microbiota and airway metabolites. Moreover, these findings suggest that the airway microbiota associated D-phenylalanine may play a role in NSCLC metastasis.

Fig. 2.

Altered metabolites were associated with airway microbiota. (A) Bar chart of the airway microbiota composition at the phylum (left) and genus (right) levels. (B) Heatmap of the top 20 microbiota. (C) α diversity between the M0 and M1 groups. Left: Chao1 index; Right: Simpson index. (D) β diversity between the M0 and M1 groups. (E) LEfSe analysis of the microbiota between the M0 and M1 groups (LDA > 2.0). (F) Correlation heatmap of differentially abundant metabolites and airway microbes; Spearman analysis, *P < 0.05, **P < 0.01. (G) Quantitative value of D-phenylalanine in the M0 and M1 groups. (H) ROC plot of D-phenylalanine

D-phenylalanine promotes NSCLC cell proliferation, migration and invasion

D-phenylalanine was significantly upregulated in lung cancer patients with metastasis. To further understand its effect on tumor cells, we selected the lung cancer cell lines A549, H1299 and LLC for in vitro experiments. During cell culture, we supplemented various concentrations of D-phenylalanine in the cell culture medium. Compared with 0 µM D-phenylalanine, D-phenylalanine promoted A549, H1299 and LLC cell proliferation, and 100 µM D-phenylalanine had a more obvious effect on cell proliferation, compared with 10 and 1000 µM group (Fig. 3A). The possible reason may be that high concentrations of D-phenylalanine induce cell damage or apoptosis, which needs further verification. EdU assays also revealed that NSCLC cells in the 100 µΜ group had the strongest proliferative ability compared with 0 and 10µΜ groups (Fig. 3B and D). These results suggested that D-phenylalanine could promote tumor cell proliferation. We also found that the wound healing capacities of A549, H1299 and LLC cells were significantly enhanced by 10 µM and 100 µM D-phenylalanine (Fig. 3C and E). Transwell assays further showed that the number of migrating cells in the 10 µM and 100 µM D-phenylalanine groups was significantly greater than that in the control group, whereas the 100 µM D-phenylalanine group had a greater number of migrating cells (Fig. 3F, H). In addition, the degradation ability of the extracellular matrix and transwell penetration ability of the D-phenylalanine treatment group were significantly increased, and the number of invading tumor cells was the highest in the 100 µM treatment group (Fig. 3G, I). These results demonstrated that D-phenylalanine promoted the proliferation, migration and invasion ability of NSCLC cells.

Fig. 3.

D-phenylalanine promotes NSCLC cell proliferation, migration and invasion (A) The proliferative activity of A549, H1299 and LLC cells treated with different concentrations of D-phenylalanine, as determined by a CCK8 assay. (B) The proliferation ability of A549, H1299, and LLC cells after 48 h of D-phenylalanine treatment by EdU (scale bar = 200 μm); *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (C) Wound healing assay results showing the migration ability of A549, H1299, and LLC cells after 24 h of D-phenylalanine treatment (scale bar = 100 μm). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (D) Statistical analysis of the number of EdU-positive cells. (E) Statistical analysis of the scratch area. (F-G) Migration (F) and invasion (G) abilities of A549, H1299, and LLC cells after 48 h of treatment with D-phenylalanine in transwells (scale bar = 100 μm). (H-I) Statistical analysis of the number of migrated and invaded cells

D-phenylalanine promotes the epithelial mesenchymal transition of NSCLC cells

The invasive and distant metastatic abilities of tumor cells are closely related to epithelial mesenchymal transition (EMT). Thus, we examined EMT-related markers in cells after treatment with D-phenylalanine. Using western blotting, we found that the expression of E-cadherin decreased under treatment with 10 µM and 100 µM D-phenylalanine, whereas the expression of N-cadherin, Vimentin, Snail, and Twist increased (Fig. 4, A-B), indicating that D-phenylalanine promoted the EMT process in A549 and H1299 cells. The expression of E-cadherin and N-cadherin in A549 and H1299 cells was further detected by immunofluorescence, and the results revealed that D-phenylalanine inhibited the expression of E-cadherin and promoted the expression of N-cadherin (Fig. 4C-D). D-phenylalanine can promote EMT in NSCLC cells, and this function may promote tumor metastasis.

Fig. 4.

D-phenylalanine promotes NSCLC cell epithelial mesenchymal transition. (A-B) Western blot detection of the expression of E-cadherin, N-cadherin, Vimentin, Twist and Snail in A549 and H1299 cells treated with different concentrations of D-phenylalanine. (C-D) The expression of E-cadherin and N-cadherin in A549 and H1299 cells treated with different concentrations of D-phenylalanine was detected by immunofluorescence (scale bar = 10/50 µm)

D-phenylalanine promotes lung tumor metastasis in vivo

To explore the effect of D-phenylalanine on lung tumor metastasis in vivo, we injected LLC cells into C57BL/6 mice via the tail vein to establish a tumor metastasis model. The model mice were intranasally administered D-phenylalanine on day 6, administered D-phenylalanine daily and killed on day 21 (Fig. 5A). With increasing doses of D-phenylalanine, the number of tumor nodules in the lungs of mice increased (Fig. 5B), and the fluorescence intensity in the mouse lungs increased, as shown by in vivo small animal imaging (Fig. 5C). Compared with that in the control group, the lung tumor weight increased significantly in D-phenylalanine treatment group (Fig. 5D). Moreover, HE staining revealed that D-phenylalanine promoted tumor growth and colonization in the lungs (Fig. 5E). IHC staining of lung tumor tissues revealed that D-phenylalanine decreased the expression level of E-cadherin and increased the expression of N-cadherin, Vimentin and Ki67, suggesting that intranasal administration of D-phenylalanine promoted lung tumor proliferation and EMT in vivo (Fig. 5F). Taken together, D-phenylalanine may promote lung tumor metastasis in mice via EMT in vivo.

Fig. 5.

D-phenylalanine promotes lung tumor metastasis via EMT in vivo. (A) Schematic diagram of the establishment of the lung metastasis model. (B-C) Results of lung tumor and in vivo imaging in the 0 mg, 0.5 mg and 1 mg groups (n = 5). (D) Lung tissue weights of the 0 mg, 0.5 mg and 1 mg groups (n = 5). (E) Representative images of HE staining in the 0 mg, 0.5 mg and 1 mg groups (scale bar = 250 μm). (F) Representative images of IHC staining for E-cadherin, N-cadherin, Vimentin and Ki67 in lung tumor tissues (scale bar = 50 μm)

Discussion

Several studies have reported that the microbiota located in the respiratory tract differ between healthy individuals and NSCLC patients. Through pathogenic metagenomic and non-targeted metabolomic sequencing, we compared airway microbiota and metabolites in the M0 and M1 groups of NSCLC patients. D-Phenylalanine was the most significantly upregulated metabolite in M1 group, which was significantly and positively correlated with the airway microbiota.

D-phenylalanine, a D-type amino acid, is derived mainly from food or the microbiota or converted from L-amino acids [27]. A previous study reported that D amino acids can be oxidized by D amino acid oxidase to generate hydrogen peroxide, which may induce an imbalance in oxidative stress and lead to lung cancer metastasis [24, 28]. D-phenylalanine also induces neutrophil chemotactic activity through the G protein-coupled receptor GPR109B and leads to local inflammation, which is one of the most important causes of lung cancer metastasis [29]. In addition, higher levels of D-phenylalanine have been observed in patients who died from acute respiratory distress syndrome than in those survived, suggesting a correlation between D-phenylalanine and lung injury [25].

In our study, 3 differentially abundant metabolites, namely, phenylacetylglycine, D-phenylalanine and phenylpyruvic acid, were enriched in the phenylalanine metabolism pathway. Phenylpyruvic acid, also known as phenylpyruvate, is mainly converted from phenylalanine [30]. Significant phenylpyruvic acid accumulation, which enhances NLRP3 protein stability, decreases lysosomal degradation, and promotes NLRP3 inflammasome activation and the release of inflammatory factors, such as interleukin (IL)-1β, is observed in diabetic foot ulcers [31]. Phenylpyruvic acid is also a metabolite risk factor for prostate cancer [32]. Compared with normal ovaries, primary epithelial ovarian cancer and metastatic ovarian cancer, phenylpyruvic acid levels were higher in the tumor tissue of primary epithelial ovarian cancer [33]. An increase in phenylpyruvic acid in M1 group was also observed in our study. These different findings in different tumors may be associated with sample types and tumor heterogeneity. The level of phenylacetylglycine in the plasma of a colorectal cancer model decreased, which can be reversed by antitumor treatment with a secondary saponin in Lysimachia capillipes [34]. In addition, homovanillic acid and epinephrine, both of which are derivative metabolites of phenylalanine, were also upregulated in the M1 group. Homovanillic acid is positively associated with the risk of developing colon cancer in serum [35]and epinephrine can promote cell proliferation, migration, anti-apoptosis, EMT, and the acquisition of angiogenic and immunosuppressive phenotypes by activating β-adrenergic receptors on lung cancer cells [36]. The above studies suggest that the phenylalanine metabolism pathway may affect NSCLC metastasis, in which D-phenylalanine may play the most important role.

According to previous studies, the sources of human metabolites can be classified into 3 categories: self-synthesis, microbial synthesis, and exogenous intake. D-phenylalanine, an essential amino acid, may be derived mainly from the microbiota since differences are detected in BALF. The lung microbiome can affect tumor development by inducing DNA damage in tumor cells, modulating tumor-associated immune responses, and promoting local inflammatory responses through microbiota-associated metabolites [37]. The airway microbiota showed no significant differences in the α diversity of the different groups in our study, whereas PCA revealed compositional structural differences in β diversity. Acinetobacter, Rothia, Streptococcus, Burkholderia, and Pseudomonas were the top 5 bacteria in NSCLC BALF. The level of Acinetobacter was greater in lung cancer tissues than in paracancerous tissues [38]. Rothia was positively correlated with survival time when 0.9-year survival was used as a cutoff, and tumors with high levels of Rothia presented a reduction in cell proliferation; increased infiltration of CD8⁺ T cells, CD4⁺ T cells, monocytes and NK cells; and a high interferon-gamma response, T-cell receptor richness and cytolytic activity, which all indicate a favorable tumor immune microenvironment [39]. Veillonella and Streptococcus are more enriched in the lower respiratory tract of lung cancer patients and are positively correlated with the ERK and PI3K signaling pathways, which are strongly associated with tumor cell proliferation, migration, and invasion [40, 41]. The level of Pseudomonas increases gradually from the oral cavity to the lung and is negatively correlated with short-chain fatty acids [21, 42]. In addition, compared with those in healthy controls, the abundance of Neisseria in the saliva of patients with lung adenocarcinoma and squamous carcinoma is lower [43]. The abundance of Neisseria was also lower in the M1 group than in the other groups in our study, suggesting a dynamic alteration in the microbiota during the development, progression and metastasis of lung cancer.

Lung cancer metastasis is a complex process [3]of which EMT is a critical step. During this process, tumor cells acquire mesenchymal cell properties, with enhanced migratory and invasive abilities, which promote tumor cell escape from the primary lesion, migration, and colonization at distant sites [44, 45]. The process of EMT is accompanied by the activation of transcription factors, including the Snail, Twist and ZEB families [44]. Snail is the first transcription factor to inhibit the expression of E-cadherin, a cell adhesion molecule on the surface of epithelial cells, and induces the expression of the mesenchymal markers fibronectin and ZEB1; Twist acts similarly to Snail [46]. E-cadherin, a key epithelial cell marker, plays a crucial role in maintaining intercellular adhesion capacity. In contrast, N-cadherin, a marker of mesenchymal cells, has a significant effect on cell morphology shaping as well as the regulation of cell motility processes. D-phenylalanine may enhance the migratory and invasive abilities of lung cancer cells by promoting EMT.

Several limitations may influence the outcomes of this study. First, the small number of participants in a single center likely increased the bias in enrolment, leading to distinct findings. Additionally, no specific microorganisms were identified and confirmed to be directly related to D-phenylalanine by genomic testing or microbial culture. Furthermore, the mechanism by which D-phenylalanine affects tumor metastasis has not been explored in depth.

Conclusion

In conclusion, via pathogenic metagenomic and non-targeted metabolomic sequencing technologies, we found that there were significant differences in airway metabolites and microbiota between the M0 and M1 groups of NSCLC patients. The airway microbiota associated D-phenylalanine was notably increased in M1 group, which could promote NSCLC metastasis through EMT, suggesting that the airway microbiota and D-phenylalanine may be potential biomarkers for the prediction of lung cancer metastasis.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

LG and HL designed and performed the experiments and analyzed the data. LG and YC collected the clinical samples and analyzed the sequencing data. LG, HL, CY, LH, MX and JD performed the experiments. JZ and DH provided technical guidance. LG and HL wrote the manuscript draft and contributed equally to this work. HD and SC supervised the study, interpreted the data and provided funding for this study. HD and SC are the co-corresponding authors of this paper. All the authors have read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82170032, No. 82470058 and No. 82270024), President Foundation of The Fifth Affiliated Hospital, Southern Medical University (YZ2022ZX04), the Guangdong Basic and Applied Basic Research Foundation, China (Grant No. 2023A1515110216) and the China Postdoctoral Science Foundation (Certification Number: 2023M731546).

Data availability

The datasets supporting the conclusions of this article are included within the article.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Animal Care and Use Committee (IACUC) of Southern Medical University, and mouse studies were carried out at the Experimental Animal Center of Southern Medical University (IACUC: LAC-20230918-006). All clinical data collection and use in this study were carried out in accordance with the ethical protocols of the Ethics Committee of Nanfang Hospital (NFEC-2022-313, Guangzhou, China).

Consent for publication

All contributing authors agreed to the publication of this article.

Competing interests

The authors declare that they have no competing interests.