Prognostic significance of SH2D5 expression in lung adenocarcinoma and its relation to immune cell infiltration

Data collection

All mRNA transcriptome data and clinical data from TCGA, GEO and CCLE database. We attained the clinical data of 594 cases, consisting of 535 LUAD and 59 normal tissue samples. A total of three, 6-week-old mice were purchased from the Shanghai Jihui experimental animal breeding Co., Ltd. (Shanghai, China) (n = 3 in LUAD and paracancerous group). This study was approved by the Laboratory Animal Ethics Committee, Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences. Ethical Approval Number: 2022-B27.

UALCAN analysis

The clinical features of SH2D5 in LUAD were analyzed with the assistance of the UALCAN website (http://ualcan.path.uab.edu/index.html) (Chandrashekar et al., 2017). LUAD type-, gender-, lymph node metastasis (LNM)-, smoking-, and stage-based patient stratification was conducted.

GEPIA

The GEPIA database (http://gepia.cancer-pku.cn/) is a newly developed web server based on the TCGA and GTEx projects (Tang et al., 2017), comprising the RNA sequencing expression data of 9736 tumor samples and 8587 normal samples. This database was adopted for survival analysis concerning SH2D5 in this study.

Survival analysis

The prognosis of LUAD based on SH2D5 expression was analyzed by the Kaplan–Meier (KM) website (http://kmplot.com/) (Györffy et al., 2010). A significant difference was noted if p < 0.05.

Immune infiltration analysis

The infiltration status of 22 immune cell types in the LUAD tissues was assessed by ssGSEA. The CIBERSORT algorithm can detect the composition of invasive immune cells in each specimen. The CIBERSORT algorithm allows a machine learning approach to distinguish 22 human immune cell phenotypes with a high degree of specificity and sensitivity (Chen & Wang, 2020). We used the CIBERSORT algorithm to compare the relative percentage of immune cells between the high expression and low expression group. Then, Spearman’s correlation analysis between SH2D5 expression and immune checkpoints.

SH2D5 and drug response

The CellMiner database (https://discover.nci.nih.gov/cellminer/home.do) is developed to help integrate and study molecular and pharmacological data for 60 diverse cancer cell lines in human. SH2D5 expression data was extracted from the CellMiner, and the correlation coefficient between SH2D5 and drugs was calculated with the help of the R core function and visualized by R ggplot2.

Establishment of LUAD in nude mice

Three 6-week-old mice were procured from Shanghai Jihui Laboratory Animal Care Co., Ltd. (Shanghai, China), with LUAD (n = 3) and adjacent non-tumor tissues (n = 3) collected. All mice were kept in an animal facility under pathogen-free conditions. Animal room was maintained at 18–23 °C, 10 h–14 h light and dark cycle. All food, water and other items that meet the rats are sterile. ADCre recombinant adenovirus was supplied by microbix. After pentobarbital sodium anesthesia of 6-week-old mice (45 mg/kg ip), approximately 125 µL of AdCre:CaPi co-precipitates was dripped into the mouse nasal cavity (twice a day, every 5 days, 42 days in total). Mice were euthanized 42 days following induction. Animal euthanasia complies with the principles stipulated by China’s National Standards for Ethical Review of Animal Welfare. Recommended procedures for CO₂ euthanasia: the initial flow of CO₂ was maintained at 20% to 30% V/min until unconsciousness, and then the flow was intensified (Bremnes et al., 2016). Besides, the CO₂ flow should be maintained for at least 1 min after clinical death to avoid reversal. Lastly, cervical dislocation was performed after CO₂ euthanasia.

Real-time quantitative PCR (qRT-PCR)

The total RNA extracted from mouse tissues by Trizol (Invitrogen, Waltham, MA) was reverse-transcribed into cDNA with the utility of M-MLV reverse transcriptase (Promega, Madison, WI, USA) and miRNA 1st Strand cDNA Synthesis Kit (by stem-loop) (Vazyme, Nanjing, China). Amplification was conducted employing SYBR Premix Ex Taq kit (TaKaRa, Shiga, Japan) and Applied Biosystems (ABI) StepOnePlus real-time PCR system (ABI, USA). Actin served as a loading control, and qRT-PCR results were analyzed by the 2^−ΔΔCt method. The primers for experimental use were as follows: Rat forward: 5-CAAGTCTGAGGCGGAATT-3, reverse: 5-GCAGTGGAGTGATGGTAG-3.

Human SH2D5: forward: 5-TGTGGTTTGGTGGCTGCCTTG-3, reverse: 5-CGACGGTTCTGCGTGGACTGAC-3. miRNA-339-5p: forward: 5-AACACGTGTCCCTGTCCTCCA-3, reverse: 5-ATCCAGTGCAGGGTCCGAGG-3, RT:5-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCGTGAG-3. All qRT-PCR procedures were conducted in duplicate or more.

Western blotting

Total proteins from cell lines were extracted using a lysis solution (RIPA) bufer, phenylmethylsulfonyl fuoride (PMSF) in a ratio of 100:1. After the protein concentration was determined, the protein was transferred to polyvinylidene fluoride (PVDF) membrane (0.45 µm; Millipore, USA), activated by treatment with methanol for 5 min. A primary antibody (SH2D5, Immunoway, USA) was applied at 4 °C overnight, followed by a secondary antibody.

Immunohistochemical (IHC) assay

We selected three human LUAD tissue samples from Suzhou Science & Technology Town Hospital. This study was approved by the Ethics Committee of Suzhou Science&Technology City Hospital (Ethics Committee No.: IRB2021054). The patient agrees and signs the informed consent form. Primary antibodies used were as follows: A primary antibody (SH2D5, Immunoway, USA). The immunohistochemical results of SH2D5 were evaluated by intensity score combined with staining intensity and area. The intensity score range is 0–9, 0 is no expression, and 9 is high expression.

Statistical analysis

Data were processed employing GraphPad Prism 8 software and statistical analysis was implemented with R software v3.1.3. Measurement data were analyzed by t-test. Cox regression multivariate analysis was carried out for survival analysis. A value of p < 0.05 was indicative of a statistically significant difference.

Abundant SH2D5 expression in LUAD

The study design encompassed SH2D5 expression, SH2D5 characterization, and mechanistic analyses, with the schedules summarized in Fig. 1A. Sangerbox website and R language were applied for SH2D5 expression profiling in different cancers. Relative to non-cancerous tissues, SH2D5 expression was elevated in most tumors, such as bladder urothelial carcinoma, cholangiocarcinoma, esophageal carcinoma, head and neck squamous cell carcinoma, renal papillary cell carcinoma, LUAD, lung squamous cell carcinoma, etc. (Fig. 1B). R language analysis of the TCGA and GEO data exhibited a noticeable elevation of SH2D5 in the LUAD tissues versus normal tissues (Figs. 1C, 1D and 1E) and additional qRT-PCR assay, Western blot and IHC staining displayed consistent results that SH2D5 expressed at a higher mRNA level in mouse LUAD tissues and BEAS-2B and H1299 cell lines as compared to adjacent non-tumor tissues (Fig. 1F, Figs. S2A–S2C).

Relations of SH2D5 expression to clinical features and prognosis of LUAD

To further characterize the clinical significance of SH2D5, the SH2D5 expression in TCGA-LUAD samples was visualized by UALCAN website analysis, which revealed considerably higher SH2D5 expression in LUAD than in non-cancerous tissues (Fig. 2A, p = 3.03e−08). SH2D5 expression in normal and LUAD groups was suggested to share associations with gender, LNM, smoking, and stage (Figs. 2B–2E). Table 1 shows the comparison of different types of patients in LUAD, and there is no difference in his groups. Next, the relation of SH2D5 expression to the clinical overall survival (OS) of 316 LUAD patients was examined by uni-variable and multi-variable analyses. The results revealed that SH2D5 overexpression was related to worse OS (Table 2). Hence, SH2D5 shows promise as an independent prognostic factor for LUAD.

The prognostic significance of SH2D5 in LUAD

Given the differential expression of SH2D5 in LUAD, the study focus was shifted onto the relevance of SH2D5 expression to prognosis. LASSO regression analysis on SH2D5 expression was implemented with the utility of the TCGA dataset, and the patients were sub-classified into high and low expression groups. The gene expression, survival time, and survival status are depicted in Fig. 3A. KM survival analysis offered data showing the association between high SH2D5 expression and poor prognosis (Fig. 3B). After validation through the KM-plot, GEPIA, and UALCAN databases, consistent results were generated (Figs. S1A–S1D). Then, receive operating characteristic curve (ROC) involving SH2D5 were plotted and area under curve (AUC) values were estimated, with the results indicating certain diagnostic efficacy of SH2D5 for LUAD (Fig. 3C, Fig. S1E).

Prognostic value of SH2D5 expression correlates with immune infiltration in LUAD

The infiltration status of 22 immune cell types in the LUAD tissues was assessed by ssGSEA. The relationship between SH2D5 expression and immune cell infiltration was evaluated by Spearman correlation analysis. SH2D5 expression was negatively correlated with dendritic cells resting (p < 0.001), plasma cells (p < 1.001), Mast cells resting (p = 0.031) and T cells CD4 memory resting (p = 0.036), and positively correlated with macrophages M0 (p < 0.001), NK cells resting (p < 0.001), T cells CD4 memory activated (p < 0.001) and macrophages M1 (p = 0.001) (Fig. 4A). We then investigated the expression of SH2D5 in 22 immune cell types between high and low expression groups. The results showed plasma cells, T cells CD4 memory resting, T cells CD4 memory activated, NK cells resting, macrophages M0 were significant between high and low expression groups (Fig. 4B). Furthermore, Among the immune checkpoints, CD40LG and TNFSF15 presented a negative correlation with the immune infiltrates, while CD276 and CD70 presented a positive correlation (Fig. 4C). We demonstrate that SH2D5 expression levels correlate with immune cell infiltration in LUAD of the TCGA dataset (Fig. 4D).

Immune infiltration has shown a relation to the prognosis of lung cancer (Conway et al., 2016; Fan et al., 2021; Sun et al., 2020b). We have demonstrated the relation of elevated SH2D5 expression to poor prognosis and also found the association between SH2D5 and immune infiltration. Thus, we surmised that the prognostic performance of SH2D5 in LUAD was potentially relevant to immune infiltration. We subsequently performed a prognostic analysis employing the KM website to correlate SH2D5 expression with immune cell subsets in LUAD. The data illustrated a poor prognosis of LUAD with high SH2D5 expression and enrichment of B cells, CD4+ T cells, macrophages, and Treg cells (Figs. 5A–5H). The aforementioned data, therefore, suggest that SH2D5 abundance in LUAD may affects prognosis through immune infiltration.

Network enrichment analysis of SH2D5 in LUAD

For the purpose of assessing the biological significance of SH2D5 in LUAD, the co-expression patterns of SH2D5 in LUAD were examined in the CCLE dataset (Fig. 6A). Gene Ontology (GO) analysis of 12 co-expression genes exhibited their correlations with natural killer (NK) cells, mediation of cytotoxicity, activation of molecular adapters, etc. Through Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, the aforementioned genes were relevant to the positive regulation of natural killer cells, phagocytosis-mediated immune regulation, and regulation of phagocytosis (Fig. 6B). Then, we used GSEA between high and low SH2D5 expression datasets. 6 hallmark gene-sets (including KEGG_COLORECTAL_CANCER, KEGG_MAPK_SIGNALING_PATHWAY, KEGG_NON_SMALL_CELL_LUNG_CANCER, KEGG_PATHWAYS_IN_CANCER, KEGG_PRIMARY_IMMUNODEFICIENCY, KEGG_SMALL_CELL_LUNG_CANCER) were chosen for analysis (Fig. 6C). The results showed that non-small cell lung cancer, pathways in cancer, primary immunodeficiency are differentially enriched in SH2D5 high expression phenotype of LUAD.

To seek SH2D5 upstream molecules in LUAD, we predicted the microRNAs (miRNAs) binding to SH2D5 through miRDB and starbase tools and found SH2D5 to be the target of miRNA-339-5p (Fig. 6D). We further verified the expression levels of miRNA-339-5p in BEAS-2B and H1299 (Fig. S2D). Then, the databases miRDB (http://www.mirdb.org/) and RNA22 (https://cm.jefferson.edu/rna22v2.0/) (Miranda et al., 2006) were employed for prediction of miRNA-339-5p binding site within SH2D5 3′UTR (Fig. 6E). This provides a potential mechanism that miRNA-339-5p may target SH2D5.

SH2D5 and drug response

SH2D5 expression exhibited positive associations with the drug reactions of Staurosporine, while negative associations with anti-cancer drugs Crizotinib, Tamoxifen, Dolastatin 10, Vinorelbine, Vinblastine, Eribulin mesilate, Homoharringtonine, Tyrothricin (Fig. 7).