The expression profile of PSAT1 in NSCLC patients and corresponding clinical data were collected from Gene Expression Omnibus (GEO) datasets (https://www.ncbi.nlm.nih.gov/) [17] and The Cancer Genome Atlas (TCGA-NSCLC) (https://portal.gdc.cancer.gov/) [18] datasets.

TCGA-NSCLC cohorts were examined for PSAT1 mRNA expression levels and divided into two groups (PSAT1-high and PSAT1-low group) based on the median expression. Then, the relationship between the PSAT1 expression profile and TCGA-NSCLC was analyzed. In addition, the Kaplan–Meier Plotter tool (https://kmplot.com/analysis/) [19] was applied to validate the impact of PSAT1 on the OS of NSCLC patients in three independent GEO datasets (GSE37745, GSE19188, and GSE30219).

To explore the different biological functions and pathways between the PSAT1-high and PSAT1-low groups, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were conducted using gene set enrichment analysis (GSEA) [20]. Using the threshold at p 0.05 and false discovery rate (FDR) 0.25.

A total of 24 paired samples consisting of tumor and normal tissue obtained from patients that underwent lung cancer surgery at Sun Yat-sen University Cancer Hospital were included in the experimental analysis. Detailed information regarding these samples can be found in Supplementary Table 1. This study was carried out following ethical guidelines and was endorsed by the Ethics Committee of Sun Yat-sen University Cancer Hospital (Approval No. YB2018-85).

The expression profiles of PSAT1 were assessed by quantitative real-time PCR (qRT-PCR). Total RNA was isolated from cancer tissue samples and adjacent normal tissue samples with TRIzol (TIANGEN, Beijing, China). PrimeScript${}^{\text{TM}}$ RT Master Mix (ES Science, Shanghai, China) was used to reverse-transcribe complementary DNA. SYBR Green Master Mix (ES Science, Shanghai, China) was used to amplify the target gene. Each sample was subjected to three independent qRT-PCR reactions, each with a reaction volume of 10 µL. The following primer sequences were used for amplification: PSAT1, 5${}^{\prime }$-CAGGAAGGTGTGCTGACTATG-3${}^{\prime }$ (forward), 5${}^{\prime }$-CCCATGACGTAGATGCTGAAA-3${}^{\prime }$ (reverse); GAPDH, 5${}^{\prime }$-GATTCCACCCATGGCAAATTC-3${}^{\prime }$ (forward), 5${}^{\prime }$-GTCATGAGTCCTTCCACGATAC-3${}^{\prime }$ (reverse). The relative expression of PSAT1 was calculated using the comparative threshold cycle (2${}^{-\mathrm{△}\mathrm{△}Ct}$) method.

Two human-derived NSCLC cell lines (H1299 and PC9) were used in this study, purchased from the Xinyuan Biotech Co. Ltd. (Shanghai, China), and then frozen and stored in liquid nitrogen. Genotype was determined by STR DNA testing and was free of bacterial, mycoplasma, and fungal contamination. DMEM (Gibco, Grand Island, NY, USA) containing 10% fetal bovine serum (Sigma, USA) and 5% CO${}_2$ was used to culture H1299 and PC9 cells. si-PSAT1s and negative control (siRNA-nc) were acquired from Suzhou GenePharma, China. The selected sequences of siRNAs were as follows: si-PAST1-1: 5${}^{\prime }$-GCUGUGCUUCAUGAAGUUATT-3${}^{\prime }$, si-PSAT1-2: 5${}^{\prime }$-CAGGAACCUUGGAUUAUAUTT-3${}^{\prime }$. The sequences of siRNA targeting PSAT1 were cloned into H1299 and PC9 cells. Using Lipofectamine 3000 (Invitrogen, 5791 Van Allen Way, Carlsbad, CA, USA), the siRNA transfection process was conducted according to the manufacturer’s instructions.

The Cell Counting Kit-8 (CCK-8) from JingXin Biological Technology in Guangzhou, China, was utilized to measure cell proliferation. Cell suspensions of H1299 and PC9 cells were prepared at a density of 5 $\times$ 10${}^5$ cells/mL. A volume of 0.2 mL of the cell suspension was added to each well of a 96-well plate, with a concentration of 1 $\times$ 10${}^3$ cells per well. Plates were then placed in a 5% CO${}_2$ incubator at 37 °C and cultured for 1–6 days. Before measuring the absorbance, 10 µL of CCK-8 solution was added to each well. After incubation, the absorbance was measured at 450 nm using a microplate reader (Bio Tek Instruments, Inc., Vermont, USA).

For in vitro migration assays, Transwell plates with 8-µm pores (Corning, NY, USA) were used. A total of 5 $\times$ 10${}^4$ cells in 200 µL of serum-free DMEM medium were added to the upper layer of the chamber, whereas the lower chamber was filled with a complete culture medium. After a 36-hour incubation, the cells in the upper chamber were removed, and the cells in the lower chamber were stained and counted. For invasion assays, Matrigel-coated chambers (BD Biosciences, Franklin Lake, NJ, USA) were employed. Cells at a concentration of 1.5 $\times$ 10${}^5$ in 200 µL of serum-free DMEM medium were placed on the upper layer of the chamber, and the lower chamber was filled with a complete culture medium. Following a 36-hour incubation period, the cells present in the upper chamber were removed, stained, and subsequently counted in the lower chamber. Four visual fields were captured for each plate, and the cell counting was performed using Image J software (version 2.0, LOCI, University of Wisconsin, Madison, WI, USA). This experimental process was repeated three times.

Tumor Immune Single Cell Hub (TISCH) (http://tisch.comp-genomics.org) is a database focusing on single-cell RNA sequencing in tumor microenvironments (TME) [21]. In this study, the TISCH database was used to systematically study the expression of PSAT1 in various cell types of NSCLC. This study utilized the TISCH database to analyze the expression of PSAT1 across different cell types of NSCLC.

Using the “GSVA” package, 28 immune cells were scored in GSE37745 using single set gene set enrichment analysis (ssGSEA) [22]. In addition, the “estimate” package was used to calculate the tumor purity and TME score of each patient (including the ESTIMATE score, immune score, and stromal score) [23].

We downloaded immunophenoscores (IPS) from The Cancer Immunome Atlas (TCIA, https://www.tcia.at/) database [24] for TCGA-NSCLC patients, which is an indicator of the sensitivity of NSCLC patients to immunotherapy, including anti-PD-1 and CTLA-4 therapy. In addition, we used another immunotherapy dataset to assess the predictive immunotherapy sensitivity of PSAT1, including 173 patients with advanced clear-cell renal cell carcinoma (ccRCC) who received anti-PD-1 immunotherapy (Nivolumab) [25].

To evaluate the chemotherapy sensitivity differences between cells expressing high and low levels of PSAT1, the half-maximal inhibitory concentration (IC50) of commonly used drugs was determined using the “pRRophetic” package. This package utilizes the expression matrix and drug response data from the Cancer Genome Project (CGP) plan, which includes information on 138 anticancer drugs tested against 727 cell lines [26].

The gene expression data were subjected to analysis using the Student’s t-test. The analysis of variance (ANOVA) test was employed to detect the expression level of PSTA1 across different stages of lung cancer. The Kaplan–Meier method was utilized to estimate OS. Spearman correlation analysis was employed for all correlation analyses. Data processing and plotting in this study were conducted using GraphPad Prism 8.0 (San Diego, California, USA). Statistical significance was determined as a p-value 0.05.

Analysis of the NSCLC database from TCGA revealed a significant increase in PSAT1 expression in tumor tissues compared to adjacent normal tissues (Fig. 1A,B). These findings were further confirmed by qRT-PCR analysis using 24 paired NSCLC samples obtained from our center (Fig. 1C). In addition, the relationship between PSAT1 expression and clinical stages of NSCLC patients from TCGA was analyzed, demonstrating that PSAT1 was significantly upregulated in T2/T3, N1, and stage II-IV (Fig. 1D–F). Finally, GSE37745, GSE19188, and GSE30219 datasets verified that NSCLC patients with high expression of PSAT1 displayed inferior OS (Fig. 1G–I).

Fig. 1.

Characteristics of PSAT1 expression in NSCLC. (A) Expression of PSAT1 in the TCGA-NSCLC cohort. (B) Expression of PSAT1 based on the paired NSCLC data from TCGA. (C) qRT-PCR showed the expression of PSAT1 in the tumor and adjacent normal tissue in NSCLC, p 0.05. (D–F) The relationship between PSAT1 expression and T, N, and TNM stage is based on the TCGA-NSCLC cohort. (G–I) Three independent GEO datasets (GSE37745, GSE19188, and GSE30219) confirmed that PSAT1 is associated with poor OS of NSCLC. qRT-PCR, quantitative real-time PCR. No significant (ns), p $\ge$ 0.05; *p 0.05; **p 0.01; *** p 0.001; **** p 0.0001. TCGA, The Cancer Genome Atlas; NSCLC, non-small cell lung cancer; GEO, Gene Expression Omnibus; OS, overall survival.

To investigate the potential mechanism of PSAT1 in NSCLC, GSEA was performed between PSAT1-high and PSAT1-low expression groups in 196 cases of NSCLC patients from GSE37745. GO annotation showed significant enrichment in biological processes, including DNA recombination and repair in the PSAT1-high group, whereas the immune response was significantly enriched in the PSAT1-low group (Fig. 2A,B and Supplementary Table 2). KEGG pathway analysis indicated that DNA nucleotide biosynthesis and P53 signaling pathway were enriched in the PSAT1-high group, whereas complement and coagulation cascades were enriched in the PSAT1-low group (Fig. 2C,D and Supplementary Table 3).

Fig. 2.

Gene set enrichment analysis between two PSAT1 subgroups. Enrichment of GO in the PSAT1-high group (A) and PSAT1-low group (B). Enrichment of KEGG in the PSAT1-high group (C) and PSAT1-low group (D). GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

In CCK8 assays, the suppression of PSAT1 resulted in significant inhibition of H1299 and PC9 cell proliferation (Fig. 3A,B). Moreover, the knockdown of PSAT1 demonstrated a significant reduction in the migration and invasion abilities of NSCLC cells (Fig. 3C–F). These findings strongly suggest that PSAT1 may promote NSCLC proliferation, migration, and invasion.

Fig. 3.

Experimental validation of the role of PSAT1 in NSCLC cells. (A) The knockdown efficiency by specific siRNA against PSAT1 was confirmed by western blot. (B) The CCK8 assay measured the proliferation ability of H1299 and PC9 cells after knocking down PSAT1. H1299 and PC9 cell migration assay (C,D) and invasion assay (E,F) were performed in control and si-PSAT1 groups. si-nc, blank control; si-PSAT1-1 and si-PSAT1-2, knock down PSAT1 in NSCLC cells; ns, p $\ge$ 0.05; *p 0.05; **p 0.01; ***p 0.001; ****p 0.0001.

The expression of PSAT1 in the TME was analyzed using the dataset NSCLC_GSE148071 from the TISCH database, which revealed that PSAT1 was primarily expressed in malignant cells, with minimal expression in other cells (Fig. 4A,B).

Fig. 4.

PSAT1 Expression in TME-associated cells from GSE148071. (A) Annotation of all cell types in GSE148071. (B) PSAT1 expression is primarily in malignant cells. TME, tumor microenvironments.

A total of 28 types of immune cells were assessed using ssGSEA for differences between groups with high and low PSAT1 expression. Heatmap analysis using dataset GSE37745 showed significant infiltration of 28 types of immune cells in the TME (Fig. 5A). The 21 types of immune cells exhibited significant differences in abundance between both groups (Fig. 5B). In addition, higher immune infiltration levels were observed in the PSAT1-low group. According to ESTIMATE analysis, the PSAT1-low group revealed higher immune and stromal scores, whereas the PSAT1-high group displayed higher tumor purity scores (Fig. 5C–E). Collectively, these results highlight a better prognosis for the PSAT1-low group.

Fig. 5.

Analysis of TME cell infiltration. (A) Heatmap showing ssGSEA scores, ESTIMATE score, immune score, stromal score, and tumor purity between PSAT1-high and PSAT1-low group. (B) The differences in the proportions of 28 immune cells between PSAT1-high and PSAT1-low groups. Differences in the immune score (C), stromal score (D), and tumor purity (E) between PSAT1-high and PSAT1-low groups. *p 0.05; **p 0.01; ***p 0.001; ****p 0.0001. ssGSEA, single set gene set enrichment analysis.

The differences in the levels of 10 immune checkpoints were compared between groups with high and low PSAT1 expression to assess the possible correlation with immunotherapy. The results demonstrated that the PSAT1-high group was associated with significant upregulation of PD-L1, CD276, and PD-L2 in contrast with the PSAT1-low group (Fig. 6A). To investigate the relationship between PSAT1 expression and immunotherapy sensitivity, we recruited an anti-PD1 treatment cohort of ccRCC patients stratified into groups with high and low PSAT1 using the median expression of PSAT1 as the cutoff. Finally, the PSAT1-low group had relatively higher rates of complete or partial response (CR/PR), clinical benefit or no change (CB/NCB), and immune checkpoint blockade response or no change (ICB/NCB) than the PSAT1-high group (Fig. 6B,C). Similarly, downregulated expression of PSAT1 correlated with superior outcomes (Fig. 6D). In conclusion, it can be inferred that PSAT1 can be harnessed to predict the efficacy of immunotherapy.

Fig. 6.

The predictive value of PSAT1 for immunotherapy response in NSCLC patients. (A) Expression of 10 immune checkpoints between high- and low-risk groups. (B,C) Proportions of anti-PD-L1 immunotherapy response in high and low PSAT1 score groups. (D) KM survival analysis in anti-PD-1 immunotherapy cohort of NSCLC. The median value is represented by the line in the box. *p 0.05; **p 0.01. CB, clinical benefit; ICB, intermediate clinical benefit; NCB, no clinical benefit; CR, complete response; PR, partial response; PD, progressive disease; SD, stable disease.

Drug therapy remains the standard of care for patients with late-stage NSCLC. However, drugs are often empirically selected. In this study, we studied the relationship between the expression of PSAT1 in various cell lines and sensitivity to multiple chemotherapy drugs using CGP data. It was found that the PSAT1-low group was more sensitive to cisplatin, docetaxel, etoposide, gemcitabine, paclitaxel, vinblastine, gefitinib, and vinorelbine (Fig. 7A–H). However, the PSAT1-high group showed better sensitivity to lapatinib and erlotinib (Fig. 7I–J), suggesting that PSAT1 expression can be used to guide drug selection.

Fig. 7.

PSAT1 expression is associated with sensitivity to (A) Cisplatin. (B) Docetaxel. (C) Etoposide. (D) Gemcitabine. (E) Paclitaxel. (F) Vinblastine. (G) Gefitinib. (H) Vinorelbine. (I) Lapatinib. (J) Erlotinib. **p 0.01; ***p 0.001; ****p 0.0001.