2.1 Gene Expression Profiling Interactive Analysis
We compared LPXN expression between esophageal malignancies and adjacent normal tissues using GEPIA. The transcriptional data of 9,736 tumors and 8,587 normal samples from the Cancer Genome Atlas (TCGA)15 and the Genotype–Tissue Expression project were evaluated using the GEPIA web-based tool (http://gepia.cancer-pku.cn/index.html). Tumor and normal differential expression analysis, profiling by cancer type or pathological stage, patient survival analysis, dimensionality reduction analysis, comparable gene detection, and correlation analysis are features that can be modified to meet user requirements. The fold-change cutoff was 0.5, and the p value cutoff was 0.05.
2.2 UALCAN Omnibus
Using clinical data and RNA sequencing (seq-RNA) from 31 different cancer types from TCGA, UALCAN was identified as an interactive and user-friendly web resource. We used UALCAN to evaluate changes in LPXN expression during the disease and to confirm that its expression in ESCC differs from that in healthy tissues.
2.3 GSE53625 from the GEO database
GSE53625, downloaded from GEO, comprises the clinical data and RNA sequencing results of 179 cases of ESCC. The samples submitted by the published literature and used in the GSE53625 dataset were obtained from Chinese patients.17
2.4 STRING database
Known and expected protein–protein interactions (PPIs) were obtained from the STRING database (https://string-db.org/).18 We analyzed the PPI networks of the co-expressed genes in LPXN to identify the functions of the individual genes in this network. With Homo sapiens as the selected species, statistical significance was determined as total score > 0.4. In this database, nodes represent proteins, and edges represent the relationships between them.
2.5 GeneMANIA and TCGA databases
The GeneMANIA database (http://genemania.org/), used to examine the links between gene sets,19 was utilized to construct the LPXN gene interaction network.
RNA was sequenced from 80 patients with ESCC based on information retrieved from TCGA website (http://tcga-data.nci.nih.gov/tcga).20 TCGA contains information on the genomic, epigenomic, transcriptomic, and proteomic characteristics of 33 different cancer types and their matched normal samples. Correlations were tested using Spearman’s correlation; statistically significant differences were defined as p < 0.05 and coefficient >0.7,. Based on the genes whose expression levels are strongly correlated with LPXN expression, GO and KEGG analyses were conducted using clusterProfiler and KEGG.db, respectively.
The stromal and immunological scores in each ESCC sample in TCGA database were determined by enrichment analysis using a gene set of a single sample (ssGSEA) and the GSVA R package. Stromal cells and immune infiltration in distinct tumor tissues were evaluated using these scores.21 We investigated the relationship between LPXN expression and ratings using the Spearman correlation test. Using features gleaned from the transcriptome and epigenetics of tumor samples from TCGA, stem cell-like properties of tumors can be quantified.22 The link between the infiltrating immune subtype and LPXN expression was investigated using one-way analysis of variance (ANOVA). Using Spearman’s correlation, the relationships between LPXN and PD-L1/PD1, MRP, and 47 other widely known immune checkpoint genes were assessed. Statistical significance was set at p < 0.05.
2.6 TIMER interface
The infiltration of many immune cells and their clinical effects were considered23 using the intuitive TIMER web interface (https://cistrome.shinyapps.io/timer/). B-cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells are among the six key analytical modules that communicate through this interface. The relationship between immune infiltration and integrin α (ITGA) gene expression depth in ESCC was visualized using scatterplots, with the ITGAs serving as inputs for the ‘Gene’ module.
2.7 Chemotherapy sensitivity analysis
CellMiner was used to gain entry into NCI-60 (https://discover.nci.nih.gov/cellminer), using nine different tumor types comprising 60 unique cancer cell lines. The link between drug sensitivity and LPXN expression was analyzed using Pearson’s test. We compared 263 medicines that received FDA clearance or had undergone clinical studies for efficacy (Supplementary Table 1).
2.8 Quantitative polymerase chain reaction
Twenty samples were obtained from the Chongqing University Cancer Hospital, including 10 pairs of tumor and healthy esophageal tissues, 14 immune-responsive ESCC tissues, and 8 immune-resistant ESCC tissues. Twenty-two patients with ESCC who had been treated with adjuvant immune therapy were included. Nanoparticle albumin-bound paclitaxel, cisplatin, and pembrolizumab were used in a four-cycle therapy plan for patients with ESCC in compliance with the 2021 Chinese Society for Clinical Oncology guidelines. Patients’ responses to immune treatment were measured based on the iRECIST criteria.24 Chongqing University Cancer Hospital ethics board approved this study. We analyzed LPXN expression in ESCC and normal esophageal tissue, as well as immune-resistant ESCC tissues and LPXN expression using q-PCR. RNA from each sample was acquired independently using TRIzol (Servicebio). Using a Thermo Revert Aid First Strand cDNA Synthesis Kit and reverse transcription, RNA was converted to cDNA. Step One Plus (Applied Biosystems) q-PCR analysis was performed using the Fast Start Universal SYBR Green Master Mix (Roche). The 2−ΔΔCT technique was used for data analysis, and GAPDH was used as the endogenous control. The primer sequences for LPXN and ATCB are as follows: LPXN forward, 5′-CCACCACCTTCTAAAACGTCAG-3′, reverse, 5′-CCCAAGCATTGAGTCCAG-GG-3′; ATCB forward 5′-GCGTGACATTAAGGAGAAGC-3′, reverse, 5′-CCACG-TCACACTTCATGATGG-3′.
2.9 Immunohistochemistry staining, assessment, and scoring
Forty-two clinical samples were obtained from Chongqing University Cancer Hospital, including 10 pairs of ESCC tissues and adjacent non-cancerous tissues, and 14 immune-responsive ESCC tissues and 8 immune-resistant ESCC tissues. The acquisition of these samples was approved by the Institutional Review Board of the hospital.
All specimens were immersed in paraffin, fixed in 10% formalin at 25 ℃, and sectioned at 4 mm thickness for IHC. Antigen recovery was performed by boiling the dewaxed hydrated tissue slices in 10 mmol/L citrate solution (pH 6.4) for 10 min. Using methanol, hydrogen peroxide 3% was used to inactivate endogenous peroxidase, and citric acid buffer (pH 6.0) was used to treat the slices to recover as much antigen as possible. Bovine serum albumin (1% concentration) was treated with a phosphate buffer for 30 min to reduce non-specific binding. The slices were treated with primary antibody for 12 h at 4 ℃. The sections were then rinsed thrice in saline phosphate buffer for 5 min each. The slices were then incubated for 50 min with a 1:200 dilution of secondary antibody (horseradish peroxidase polymer). The samples were sealed and examined under a light microscope. The primary antibodies used were anti-LPXN (PA597933, Thermofisher) and anti-PD-L1 (PA528115, Thermofisher). Supplementary Table 2 lists the principal antibodies.
To evaluate the IHC results, we followed an established scoring method25. PFN2 expression was assessed using a semi-quantitative approach based on the percentage of positive cells and the intensity of cytoplasmic/nuclear staining. The percentage of positive cells was categorized as follows: 0 (<5% positive cells), 1 (6–25% positive cells), 2 (26–50% positive cells), 3 (51–75% positive cells), or 4 (>75% positive cells). Cytoplasmic/nuclear staining intensity was classified as 0 (negative), 1 (light yellow), 2 (yellow), or 3 (brown). The product of the percentage of positive cells and staining intensity yielded the immunohistochemistry score (IS) for each patient. Two independent pathologists assessed all immunostaining results and arrived at a consistent score. Disagreement between the two pathologists regarding IHC results was resolved by consulting a third pathologist. Therefore, the IS ranged from 0 to 12, reflecting the scoring system used.
2.10 Small hairpin RNAs (shRNAs) and lentiviruses
The shRNA polynucleotides shLPXN-1 (CCAACGACTACCACCAACTTT) and shLPXN-2 (GCACTTCTTCTGCTCTCACTG) targeting human LPXN were synthesized, annealed, and cloned into the lentiviral vector GV493 carrying a green fluorescent protein expression cassette (GeneChem Co, Ltd., Shanghai, China). An empty GV493 vector served as the control shRNA. The models were tested using these sequences.
2.11 Western blotting
We conducted western blotting using cell lysates obtained from KYSE-150 human ESCC cell lines and ESCC tissues. An equal amount of protein lysate was loaded and separated by 10% SDS-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride membrane (Millipore, USA). The membranes were blocked with 5% fat-free milk in TBS buffer and incubated overnight at 4 ℃ with anti-LPXN (PA597933, Thermofisher) and anti-PD-L1 (PA528115, Thermofisher). Bound antibodies were detected using a secondary antibody (horseradish peroxidase polymer) at room temperature for 2 h. After washing, the resulting bands were visualized using the standard Electron Cryo-Microscopy (ECL) procedure (Kangwei, Beijing). The acquired images were analyzed for grayscale values using the image acquisition system (Bio-Rad, USA) and ImageJ (National Institutes of Health, USA).
2.12 Cell culture and transfection
KYSE-150 human ESCC cell line cultures obtained from GeneChem Co., Ltd. (Shanghai, China) were maintained in 6 cm plates with 3 mL complete media in a 37 ℃, 5% CO2 incubator. Tests were conducted using logarithmically growing cells. shRNAs (PSC107025 and PSC110829) and negative control sequence viruses were transfected into KYSE-150 cells cultured in 6-well plates or 60 mm2 dishes using the same method as described for lentivirus generation. The efficacy of the infection was assessed 72 h after transfection using a fluorescence microscope (IX71; Olympus, Japan). Blotting was performed using RT-PCR to analyze the mRNA and protein expression of LPXN in cells infected with shLPXN-1, shLPXN-2, or shCtrl to assess the effectiveness of knockdown.
2.13 Celigo image cytometry assay
Logarithmic-phase experimental cells were dissociated with trypsin and collected via centrifugation. In 96-well plates, the cells were resuspended, quantified, and seeded at a density of 2 × 103 cells/well in 100 μL growth medium. Seeding was performed to obtain the same number of cells from each well. The cells were cultured at 37 ℃ and 5% CO2. The fluorescence intensity of the cells was measured daily for five consecutive days using the Celigo Imaging Cytometer System (Nexcelom Bioscience, St. Lawrence, MA, USA), and the cell number was automatically calculated. A proliferation curve was created using the number of green fluorescent cells.
2.14 Flow cytometry analysis of apoptosis and cell cycle
A single-cell suspension of trypsin-digested logarithmic-phase cells that had undergone two cold PBS washes was administered to each test subject group. Ten microliters of Annexin V-APC (eBioscience, San Diego, CA, USA) was used to stain the cell suspensions at room temperature for 10–15 min in the dark. A flow cytometer (BD Biosciences, San Jose, CA, USA) was used to calculate the percentage of dead cells.
2.15 Colony-forming assay
Each cell was plated at a density of 500 in well plates for 14 days in full medium and observed every 3 days. After incubation in 4% paraformaldehyde for 30–60 min, the wells were washed with PBS to remove any remaining fixative. Filtered crystal violet staining solution (1000 μL) was used to stain the cells for 10–20 min. After multiple ddH2O washes and drying at room temperature, visible colonies were counted using the Image-Pro Plus software (Version 6.0.0; Media Cybernetics Inc.).
2.16 3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide (MTT) assay
Trypsin was used to loosen the developing cells in the logarithmic phase, which were then separated by centrifugation. Cells at a density of 2 × 103 cells/well were resuspended and seeded in a 96-well plate. After 2 days, cells in each group were treated with sterile MTT (5 mg/mL) at 37 ℃ for 4 h on days 3, 4, and 5. Dimethyl sulfoxide was added after the culture medium was discarded, and the mixture was stirred for 2–5 min in a vortex mixer. An enzyme-linked immunosorbent assay reader (M2009PR; Tecan Infinite, Switzerland) was used to measure absorbance at 490 nm. All experiments were performed at least in triplicate.
2.17 Celigo scratch assay
After resuspending the cells in a complete medium, the logarithmic phase experimental cells were counted after digestion with trypsin. The cells were seeded at a density of 5 × 103 cells/well and cultivated in 100 μL medium per well, three replicates per group, 37 ℃, 5% CO2 culture condition. The following day, a scratch tool was used to form a small indentation at the top-center of each well. At 0 h, the plates were scanned after being gently rinsed twice or thrice with PBS, followed by adding 1% FBS serum-containing medium. The plates were incubated at 37 ℃ for 24 h at 5% CO2 and scanned with Celigo to detect the extent of migration.
2.18 Cell invasion and migration assays
The experimental cells in every group were dislodged with trypsin, centrifuged for collection, reconstituted in a serum-free medium, and counted. In the top chamber, cells were injected into a serum-free medium with or without a Matrigel matrix membrane. Cells (5 × 105) were seeded into each well, and 750 μL of 30% FBS culture medium was added to the lower chamber. The steps for developing the cells followed a prescribed pattern. The non-invasive cells were removed from the inner wall of the incubator at 37 ℃ with a moist brush. The outer cell walls were gently washed with PBS, stained with crystal violet, and fixed in 4% paraformaldehyde. The number of cells in each of the nine distinct minute visual areas was measured under a microscope. Each experimental group was tested in triplicate.
2.19 Statistical analysis
The Mann–Whitney U test was used to compare LPXN immunological pathways; p-values were adjusted using the Bonferroni technique. The variation in OS across groups was investigated using Kaplan–Meier plots. Using Spearman’s or Pearson’s correlation analyses, we investigated the relationships between LPXN expression and stemness, stromal score, drug resistance genes, regulators of immunity, and drug sensitivity. We compared the experimental groups using one-way ANOVA and Bonferroni post-hoc tests. Charts were created using the R programming language (version 4.2.1) and the ggplot2, pheatmap, ggpubr, corrplot, and Survminer applications. A two-tailed p < 0.05 was utilized to indicate significance.