Introduction
Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of lymphoma, which accounts for about 30–58% of NHL [
1]. In the era of new drugs for tumor treatment, such as anthracyclines and monoclonal anti-CD20 antibody, the prognosis of DLBCL patients has been significantly improved [
2,
3]. However, a recent large-scale survey has found the cure rate of DLBCL range from only 52.8–68.9% in the USA, and about one-third of DLBCL patients eventually become relapsed [
4,
5]. Therefore, novel molecular targets and treatment strategies are urgently needed to improve the prognosis of DLBCL [
6].
Translation initiation factors, which include at least 10 kinds in eukaryotes, play key roles in translation process [
7]. Among them, eukaryotic initiation factor 3 (eIF3) participates in the cap-dependent translation initiation and depolymerization of 80 S ribosomes during the termination of translation process, which prevents the premature binding of 40 S subunit to 60 S subunit, and regulates the initial stage of translation [
8,
9]. eIF3a, also known as p150 and eIF3-p170, is the largest subunit and functional core of eIF3, which was firstly purified from rabbit reticulocyte lysate and mainly distributed in the cytoplasm [
10]. Recent investigations have implicated that eIF3a was upregulated in several solid tumors, such as urinary bladder cancer (UBC) and hepatocellular carcinoma (HCC) [
11‐
13], eIF3a expression has been found to be associated with prognosis of tumor patients [
12,
14]. However, the role of eIF3a in tumor development is controversial. Previous studies showed that the stable suppression of eIF3a could increase cell doubling time and change cell sensitivity to various cycle regulators [
15], indicating the pro-oncogenic role of eIF3a in tumor development [
12,
16,
17]. In contrast, other research found that eIF3a could protect low-grade malignancies from progressing to high-grade types [
14,
18], suggesting the potential tumor-suppressive role of eIF3a. However, the expression level and biological role of eIF3a in hematological malignancies, especially in DLBCL have not been elucidated yet.
Our study aimed to evaluate the expression and biological function of eIF3a in DLBCL. We demonstrated that eIF3a was overexpressed in DLBCL compared with healthy control. DLBCL patients with high eIF3a expression displayed worse overall survival (OS). The expression of eIF3a protein was significantly increased in DLBCL cell lines compared with peripheral blood mononuclear cells (PBMCs) from healthy donors. eIF3a knockdown could inhibit the proliferation of DLBCL cells, and increase the cell apoptosis rate. In addition, function analysis revealed eIF3a had a close linkage to cell cycle and tumor immune. Mutation analysis found there were great discrepancy in several pathways and druggable categories between eIF3a-related differentially expressed genes (DEGs) mutation groups and non-mutation groups. Moreover, eIF3a expression was also found to be associated with chemosensitivity in DLBCL. These results indicate that eIF3a might be a promising prognostic biomarker and therapeutic target in DLBCL treatment.
Materials and methods
Data acquisition
The study datasets contained GSE25638 (
n = 97, including 26 DLBCL and 6 B cell samples from healthy controls), GSE31312 (
n = 498), GSE23501 (
n = 69) and GSE181063 (
n = 1037) from Gene Expression Omnibus (GEO) database (
https://www.ncbi.nlm.nih.gov/geo/), and TCGA-DLBCL (
n = 47) from The Cancer Genome Atlas (TCGA) database (
https://www.cancer.gov/). Pan-cancer dataset from UCSC (
https://xenabrowser.net/) and TCGA were utilized to explore the expression and prognostic value of eIF3a in other malignancies.
Patient samples
This study was approved by the Biomedical Research Ethic Committee and the consent form was signed. Paraffin-embedded specimens were obtained from 44 newly diagnosed DLBCL patients and 32 reactive hyperplasia lymphoid (RHL) patients. Patients who met the following criteria were included: pathologically confirmed diagnosis of DLBCL; with complete clinical data and follow-up information. Exclusion criteria contained: with other malignancies; die from other causes. All patients received standard treatment according to NCCN Clinical Practice Guidelines [
19]. Histological diagnoses were established according to the 2022 WHO classification [
20]. Clinical data of newly diagnosed DLBCL patients were collected simultaneously.
Cell culture and transfection
Human DLBCL cell lines LY1, LY3, U2932 and LY8 were cultured in Iscove-modified Dulbecco medium (IMDM, Gibco, Life Technologies, Carlsbad, CA, USA) with 10% fetal bovine serum (HyClone, Logan, UT, USA). PBMCs were obtained from two healthy donors, and isolated by the Ficoll-Hypaque density gradient centrifugation method. All cell lines were incubated in an atmosphere containing 5% CO2 at 37℃. All cells were examined for short tandem repeat (STR) and mycoplasma infection periodically. Green fluorescent protein (GFP) labeled lentiviral vector either eIF3a-konckdown or empty was purchased from Genechem (Shanghai, China). LY1 and LY8 cell lines were transfected with lentivirus according to the manufacturer’s instructions and screened with 5 µg/mL puromycin (Amresco, USA). The infection efficiency was assessed by qRT-PCR and western blot.
Quantitative real-time PCR
Total RNA was extracted by RNAiso Plus (TaKaRa, Dalian, China), according to the manufacturer’s instruction, and cDNA was synthesized by reverse transcription reagent (TaKaRa). We used SYBR Green Master Mix (TaKaRa) in LightCycler 480II (Roche, Basel, Switzerland) to achieve amplification reaction. The eIF3a specific primers (Biosune, Shanghai, China) were as follows: forward, 5’-AAAACAACACCATCCTCCGC-3’ and reverse 5’-AACTCAGGGTCCGAGAAGTG-3’. The reference gene is GAPDH (TaKaRa), for-ward, 5’-GCACCGTCAAGGCTGAGAAC-3’ and reverse 5’-TGGTGAAGAC-GCCAGTGGA-3’. The quantification of mRNA was calculated using the 2-ΔΔCt method [
21]. Real-time PCR for each gene of each cDNA sample was assessed in triplicate.
Western blot
Proteins were extracted using radio-immunoprecipitation assay buffer (RIPA, Shenergy Biocolor, Shanghai, China) and 1% phenyl methyl sulfonyl fluoride (PMSF, Shenergy Biocolor), and protein concentration was detected using the BCA assay (Shenergy Biocolor). A total of 30 µg protein extract from every sample was then electrophoretically separated on a 7.5% polyacrylamide gel (Bio-Rad, USA) and transferred to PVDF membranes (Millipore, Billerica, MA, USA), which were blocked in 5% skimmed milk for 1 h. The blots were cut before hybridisation with antibodies, then incubated in primary antibody overnight at 4℃. The PVDF membranes were incubated in an HRP-conjugated secondary antibody (Zhongshan Goldenbridge, Beijing, China, 1:5000 dilution) for 1 h. Ultimately, the blots were detected using the electro-chemiluminescence kit (Millipore) by Amersham Imager 680 imaging system (General Electric, USA). The primary antibodies included eIF3a (ab128996, Abcam, Cambridge, UK, 1:2000 dilution), Cyclin D1 (26939-1-AP, Proteintech, USA, 1:5000 dilution), CDK4 (11026-1-AP, Proteintech, 1:1000 dilution), CDK6 (14052-1-AP, Proteintech, 1:1000 dilution) and GAPDH (TA309157, Zhongshan Goldenbridge, 1:1000 dilution). Image J software 1.44 was used to quantify protein band signals.
Immunohistochemistry (IHC)
The paraffin-embedded tissues were sliced into 4 μm thick and then deparaffinized and hydrated. Antigen retrieval was performed using a 1 × EDTA antigen retrieval solution (Solarbio, Beijing, China). Endogenous peroxidase was blocked by 3% hydrogen peroxide and slides were incubated in goat serum (Solarbio) to block non-specific binding. The specimens were then incubated with anti-eIF3a antibody (ab128996, Abcam, 1:100 dilution) overnight at 4℃. Slides were incubated at 37℃ for 30 min, then handled with goat anti-rabbit biotinylated antibody and incubated with strept avidin-horseradish peroxidase complex (SABC). Finally, a diaminobenzidine (DAB) kit was used to stain the specimens, and hematoxylin was used to counterstain. Two independent observers who were blinded to patient’s clinical data scored the staining results. Five microscopic fields were observed at × 400 magnification. We calculated the IHC score by multiplying the expression level (0, 0–19%; 1, 20–49%; 2, 50–69%; 3, 70–100%) and staining intensity (0, negative; 1, weak; 2, moderate; 3, strong) of tumor cells. Scores of 0–3 were considered as negative expression, and 4–9 as positive expression [
22].
Cell proliferation assay
Cell proliferation was assessed using the cell counting kit-8 assay (CCK-8, Dojindo, Japan). LY1 and LY8 cells were seeded in 96 well plates at a 10,000 cells/well density for 24–96 h. After that, the cells were incubated with CCK-8 at 10 µL/well for 2 h at 37℃. Then, the cell proliferation was calculated by an absorbance value at 450 nm using Multiskan GO Microplate Reader (Thermo Scientific, Rockford, IL, USA).
Apoptosis analysis
Annexin V-PE/7AAD apoptosis detection kit (BD Biosciences, Bedford, MA, USA) was employed to detect cell apoptosis. DLBCL cells with indicated treatments were harvested and resuspended in 1× binding buffer, followed by the addition of 5 µL of Annexin V-PE and 5 µL of 7-AAD. After gentle vibration and incubation for 15 min at room temperature in the dark, cells were subjected to the flow cytometry (FACS navios flow cytometer, Beckman Coulter, CA, USA) according to the manufacturer’s instructions.
Identification and analysis of DEGs between two clusters based on eIF3a expression
The mRNA sequencing data of DLBCL patients in GSE23501 were employed to identify DEGs between two clusters based on eIF3a expression, with absolute log2 fold change (|log2FC|) > 1 and Wilcoxon test P value < 0.05. GO analysis of DEGs was performed via “clusterProfiler” and shown through “ggplot2” R package.
Immune profile evaluation between two clusters based on eIF3a expression
22 immune cells of DLBCL samples in GSE23501 was estimated with “CIBERSORT” package in R. The discrepancy of several vital immune checkpoints between the two groups was calculated through “vioplot” and “ggplot2” packages in R to compare the immunity in tumor environment.
Somatic mutation data such as variant classification, variant type of TCGA-DLBCL (n = 47) was employed for mutation analysis through “maftools” R package. All these samples were histologically diagnosed as DLBCL. Oncoplot was used to show the mutation profile of DEGs. Fisher’s exact test was calculated and displayed by correlation heatmap, and P < 0.05 was defined to have significance. Subsequently, all 47 samples were divided into two groups based on DEGs mutation. Mutational alteration of several vital pathways and potential druggable categories between DEGs mutation group and non-mutation group were also explored through “drugInteractions” algorithm.
Pan-cancer analysis of eIF3a
Sequencing dataset from UCSC were employed to perform pan-cancer analysis of eIF3a. Differential expression of eIF3a between normal and tumor samples in other 34 types of tumors were calculated by Wilcoxon test and
P < 0.05 was considered to be significant. The prognostic value of eIF3a was then calculated in 44 tumors from UCSC and previous study [
23].
Chemosensitivity analysis
Samples in GSE23501 were divided into two groups based on eIF3a expression, the “oncoPredict” R package was utilized to assess half of the maximum inhibitory concentration (IC50) of 198 kinds of drugs from Genomics of Drug Sensitivity in Cancer (
http://www.cancerrxgene.org/). “ggpurb” R package was used to show the difference of IC50 in the two groups.
Statistical analysis
The data of cell experiments were expressed as mean ± standard deviation, and SPSS 20.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 7.0 (San Diego, CA, USA) were used for statistical analysis. The t-test or one-way analysis of variance (ANOVA) was used to assess the differences between quantitative variables. The chi-square test was used to analyze the association between clinical data and eIF3a expression in DLBCL patients. Kaplan-Meier analysis was performed to assess the prognostic role of eIF3a expression in DLBCL patients. All experiments were conducted at least 3 times. Statistical significance was considered when P < 0.05 (* P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001).
Discussion
In this study, we for the first time elucidated that eIF3a was upregulated in DLBCL tissues and cell lines, which was positively correlated with clinical features of DLBCL patients. DLBCL patients with positive eIF3a expression had shorter OS. eIF3a knockdown significantly inhibited cell proliferation and increased apoptosis rate in DLBCL cells. Moreover, 114 DEGs were identified in the two clusters based on eIF3a expression, which had a close linkage to cell cycle and tumor immune. Cytotoxic immunocytes were verified to be significantly decreased in high-expressed eIF3a clusters, consistently to poorer prognosis in DLBCL patients. The mutation characteristics of eIF3a and DEGs had the potential to predict drug sensitivity in DLBCL. eIF3a expression was also clarified to be associated with chemosensitivity to several anti-tumor drugs. These results indicated that eIF3a might become a promising prognostic biomarker and therapeutic target in DLBCL treatment.
Till now, the correlation between eIF3a expression and prognosis in tumors is controversial. Recent studies have shown that eIF3a was highly expressed in esophageal squamous cell carcinoma (ESCC), colorectal cancer and thyroid cancer (TC) [
16‐
18], but lower in clear cell renal cell carcinoma (ccRCC) [
24]. Moreover, eIF3a expression was also related to prognosis of tumor patients. In patients with ESCC, ccRCC, low-grade UBC and oral squamous cell carcinoma (OSCC), high levels of eIF3a were associated with favorable prognosis [
11,
18,
24,
25]. Similarly, breast cancer patients with high eIF3a expression were more sensitive to anthracycline drugs and displayed better prognosis [
26]. However, Haybaeck J and Mei C et al. found that higher expression of eIF3a was associated with high tumor proliferation rate, distant metastasis rate and worse prognosis of colorectal carcinoma [
17,
27]. Correspondingly, our findings revealed that eIF3a expression was obviously increased in DLBCL, and has a close linkage with IPI score and Ann-Arbor stage, suggesting that eIF3a may be related to the progression of DLBCL. Kaplan-Meier analysis further proved the association between high expression of eIF3a and shorter OS in DLBCL patients. Overall, these findings suggested that eIF3a might serve as a promising biomarker for prognostic evaluation in DLBCL patients, which needs to be validated in larger cohort of tumor patients.
Published studies have demonstrated that eIF3a impacted tumor development through regulating several biological processes, including cell proliferation, apoptosis, colony formation and migration [
11,
28]. One study reported that eIF3a overexpression resulted in a higher proliferation rate, enhanced clonogenicity, resistance to apoptosis, and malignant transformation of immortal fibroblasts [
29]. Besides, in TC and HCC tissues, high level of eIF3a had correlation with acceleration of cell proliferation and apoptotic decrease [
12,
16]. eIF3a knockdown was found to reverse the malignant phenotype of lung and breast cancer cells [
30]. Downregulation of eIF3a also impaired the proliferation, colony formation, and migration of UBC cells [
11]. Previous studies found that eIF3a participated in many tumor development processes, such as maintaining the stem cell-like characteristics, mitochondrial dysfunction and glycolysis activation in tumor cells, which further contributed to cell cycle and proliferation [
31,
32]. Consistently, our results showed that eIF3a knockdown could cause the inhibited cell proliferation and enhanced cell apoptosis in DLBCL cell, indicating the key role of eIF3a in the development of DLBCL. The regulatory roles and molecular mechanisms of eIF3a on biological processes in tumor development need to be further clarified.
CCND1 has been identified as a G1 to S cell cycle regulator and about 2.1% of DLBCL patients with positive CCND1 expression, especially in younger male patients [
33]. The association between CCND1 expression and prognosis remained controversial in DLBCL patients [
33,
34]. Our study found that the CCND1 protein expression in DLBCL cell lines with eIF3a knockdown was significantly reduced, which suggested that reducing the expression of eIF3a would inhibit the proliferation of DLBCL cells via inducing cell cycle arrest. Interestingly, there was no statistical significance in CCND1 RNA level in our study. The possibilities for this difference in protein and RNA levels may be concluded that eIF3a impacted the translation process, which has been reported in other tumors [
17].
Furthermore, CDKs are a group of protein kinases that play essential role in controlling cell division and cell cycle [
35]. Among them, CDK4 and CDK6 have similar biochemical and biological characteristics, and both of them could promote the synthesis of D-type cyclins, thereby promoting the process of cell cycle. It has been confirmed that CDK4 and CDK6 are closely related to tumorigenesis [
36,
37], and inhibition of CDK4/6 could bring an anti-tumor effect in several tumors [
38,
39], including lymphoma [
40,
41]. To date, the U.S. Food and Drug Administration (FDA) has approved three selective CDK inhibitors (palbociclib, ribociclib, and abemaciclib) for breast cancer, and 15 CDK4/6 inhibitors are in clinical trials [
42]. Our study showed that knockdown of eIF3a reduced the CDK4 and CDK6 expression in DLBCL cells, suggesting that targeting eIF3a has a potential anti-tumor effect in DLBCL.
Apoptosis, a meaningful way to remove abnormal cells, is necessary for the body metabolism and homeostasis. Currently, many new drugs play a therapeutic role in lymphoma by promoting the apoptosis of tumor cells [
43‐
45], such as LW-213, BM-1197, 1, 2-diazole. Previous studies have found that there was no correlation between eIF3a expression and p53, who participates in the intrinsic and extrinsic cell death pathways, indicating the eIF3a-related apoptosis may through p53-independent pathway [
14]. Our results found that eIF3a knockdown could increase the apoptosis rate of DLBCL cells, indicating the regulatory role of eIF3a in cell apoptosis, and further studies are needed to explore underlying mechanism.
Point mutations of certain genes have been demonstrated to affect cell behavior and tumor progression [
46,
47]. For example, eIF3a mutation was found to be related to acquired chemotherapy resistance in lung cancer [
48], and a low-frequency missense variant in SPOCD1 was associated with reduced risk of gastric cancer [
49]. In our study, eIF3a and DEGs mutation were widely observed in DLBCL samples, and predicted drug sensitivity, suggesting that deep exploration of these mutations could further reveal the heterogeneity, and provide novel targets for precision treatment in DLBCL.
Previous studies have widely demonstrated that immune function was restricted in malignancies, such as inferior immune status and aberrant immune checkpoints, which was closely related to prognosis [
50,
51]. Immune profiling analysis found that several cytotoxic immunocytes such as CD8
+ T cells and γ/δ T cells were verified to be significantly accumulated in low-expressed eIF3a clusters, which were associated with improved survival outcome [
52,
53]. Moreover, we found less enriched immunocytes and more inactive components in high eIF3a subgroup, indicating that these patients may tend to be resistant to immunotherapeutic interventions. Immune checkpoints such as TIGIT and CTLA4 were denser in low eIF3a group, indicating that eIF3a may not only provide therapeutic target for DLBCL patients but also act as potential predictors of immunotherapy responses.
High heterogeneity of DLBCL forced deeper understandings of individualized treatment regime. We performed chemosensitivity analysis in different groups based on eIF3a expression, finding that eIF3a expression had a close linkage to several drugs including Vincristine and novel inhibitors (Wee1 inhibitor, YK-4-279, MK-1775 et al.). Previous studies have revealed that Wee1 inhibitor had great anti-tumor potential together with other G2/M arresting or DNA damaging therapeutic compounds in DLBCL [
54]. YK-4-279 was also found to exert anti-lymphoma activity via blocking the protein-protein interaction with RNA helicases [
55]. These results indicated that eIF3a has great potential to act as a biomarker for chemosensitivity in DLBCL.
In conclusion, our findings firstly identified that eIF3a was highly expressed in DLBCL, which was correlated with clinical characteristics and worse prognosis. eIF3a knockdown could inhibit the growth of DLBCL cells through regulating cell proliferation and apoptosis. Further function and immune analysis revealed that high eIF3a expression tends to be associated with immunosuppressive status and poorer prognosis. eIF3a and DEGs mutations were found to be correlated to chemosensitivity of some anti-tumor drugs and vital signal pathways. eIF3a expression was also demonstrated to be associated with chemosensitivity to several anti-tumor drugs. These findings suggest that eIF3a might serve as an indicator of prognosis and drug sensitivity for DLBCL treatment. Nevertheless, further researches are needed to explore the molecular mechanism of eIF3a in DLBCL development, which might provide more novel ideas for DLBCL treatment.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.