3.1 Expression analysis of ZNF32 in pan-cancer

To examine ZNF32 levels in pan-cancer, the TCGA and GTEx databases were utilized. We found that ZNF32 expression levels had significant differences in 24 cancers compared with paired normal tissues, which included ACC, BRCA, CESC, CHOL, COAD, DLBC, ESCA, LAML, LGG, LIHC, GBM, KICH, KIRC, LUSC, OV, PAAD, PCPG, READ, TGCT, THYM, UCEC, and UCS (Fig. 1A).

Fig. 1

The expression levels of ZNF32 differ amongst cancer types. A Comparative analysis of ZNF32 expression between cancerous (red box plots) and non-cancerous (blue box plots) samples. B Box plot data showing the differences in ZNF32 expression from matching TCGA data. C Box plot data showing the differences in ZNF32 expression from GTEx data. D Box plot data showing the differences in ZNF32 expression from CCLE data. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant

Based on the TCGA database, the expression level of ZNF32 in tumor tissues was generally increased. ZNF32 expression was lowest in KICH and highest in LGG (Fig. 1B). Our study examined the ZNF32 expression in normal tissues (Fig. 1C). ZNF32 expression levels were higher in the normal human uterus, ovary, and prostate. We also compared the ZNF32 expression using the data directly from the tumor cell lines in CCLE. The ZNF32 expression levels in various tumor cell lines were respectively presented (Fig. 1D).

To further investigate the difference between the expression of ZNF32 in tumor tissue and adjacent tissue, we further used the UALCAN database to conduct the research. ZNF32 expression levels were significantly up-regulated in CHOL, COAD, ESCA, LIHC and PCPG, but down-regulated in CESC, KICH, KIRC, KIRP, LUAD, THCA, and UCEC (Figure S1A-B). Further research demonstrated a strong association between ZNF32 expression and different stages of cancer (Figure S2). Changes in ZNF32 expression in cell lines and cancer tissues may affect how cancer cells grow and spread.

3.2 Genetic changes and epigenetic modification of ZNF32

Since the expression of ZNF32 is quite different in tumors, the cBioPortal was used to analyze its genetic alterations and epigenetic regulatory modifications. We also noticed that ZNF32 mRNA expression showed a positive correlation with the CNA of the ZNF32 in LGG (r = 0.55, p < 0.05), SARC (r = 0.53, p < 0.05) and other 28 types of cancer (Fig. 2A). Meanwhile, it showed a negative correlation between the DNA methylation standard and ZNF32 mRNA expression in DLBC (r = − 0.56, p < 0.05), STAD (r = − 0.55, p < 0.05) and other 20 types of cancer (Fig. 2B).

Fig. 2

ZNF32 DNA methylation and CNA in pan-cancer. A The correlation between ZNF32 and CNA. Results with significance (p < 0.05) are highlighted in red. B The correlation between ZNF32 and DNA methylation. Results with significance (p < 0.05) are highlighted in blue. CNA, copy number alteration

3.3 Survival analysis related to ZNF32 expression in pan-cancer

Our study examined the value of ZNF32 expression for predicting OS, DFI, DSS, and PFI across cancers by using a univariate Cox regression model (Fig. 3A–D). As displayed in Fig. 3A, improved OS was significantly correlated with elevated ZNF32 expression in LIHC (p = 0.03, HR = 1.540), and plays a protective role in LGG (p < 0.001, HR = 0.201), MESO (p < 0.001, HR = 0.419), PAAD (p < 0.001, HR = 0.501), OV (p = 0.005, HR = 0.861), CESC (p = 0.019, HR = 0.642) and GBM (p = 0.033, HR = 0.727). For DSS, an elevated level of ZNF32 was significantly correlated with shorter DSS in LIHC (p = 0.03, HR = 1.750), PCPG (p = 0.004, HR = 12.209), BRCA (p = 0.026, HR = 1.658), while ZNF32 expression has a protective impact on LGG (p < 0.001, HR = 0.182), PAAD (p < 0.001, HR = 0.470), MESO (p = 0.002, HR = 0.433), OV (p = 0.003, HR = 0.851), GBM (p = 0.046, HR = 0.729), and LUSC (p = 0.05, HR = 0.735) (Fig. 3B).The analysis results of DFI also demonstrated that high ZNF32 expression has a low DFI rate in LIHC (p = 0.008, HR = 1.425), ACC (p = 0.021, HR = 4.396), and high DFI in LGG (p = 0.005, HR = 0.189), THCA (p = 0.020, HR = 0.373), PAAD (p = 0.029, HR = 0.358), and UCEC (p = 0.032, HR = 0.611) (Fig. 3C). Additionally, an increased level of ZNF32 expression was strongly associated with worse PFI in LIHC (p < 0.001, HR = 1.4), but associated with better PFI in LGG (p < 0.001, HR = 0.318), PAAD (p < 0.001, HR = 0.397), UCEC (p = 0.010, HR = 0.661) (Fig. 3D). According to the results of the KM analysis of OS, ZNF32 plays a risk factor for patients with BRCA, COAD, DLBC, LIHC, PCPG, and THCA, but a favorable prognosis was found in CESC, GBM, LGG, LUSC, MESO, OV, PAAD, PRAD, TGCT and UCS (Fig. 4). The results showed a significant link between ZNF32 expression and outcomes in many types of cancer, suggesting that ZNF32 could be a new biomarker for prognosis.

Fig. 3

Forest plots of hazard ratios of ZNF32 in pan-cancer. A Association between ZNF32 expression and OS; B Association between ANLN expression and DFF; C Association between ANLN expression and DFI; D Association between ZNF32 expression and PFI. Cox regression analysis. p < 0.05 was considered to be significant

Fig. 4

Patient overall survival Kaplan–Meier curves by ZNF32 expression level in BRCA, COAD, DLBC, LIHC, PCGC, THCA, CESC, GBM, LGG, LUSC, MESO, OV, PAAD, PRAD, TGCT and UCS

3.4 Gene set variation analysis (GSVA) of ZNF32 in pan-cancer and HNSC

The possible pathways of ZNF32 were found by using GSVA to find the hallmark pathways linked to ZNF32 expression. We found that ZNF32 was positively correlated with the first three pathways in multiple cancers, including “DNA REPAIR”, “OXIDATIVE PHOSPHORYLATION” and “SPERMATOGENESIS”. (Fig. 5A).

Fig. 5

A Correlation of ZNF32 with 50 HALLMARK terms in pan-cancer. B Results of GSVA analysis in HNSC. Blue bars represent the 29 pathways exhibiting a negative correlation, while yellow bars denote the 7 pathways demonstrating a positive correlation

HNSC’s top three upregulated pathways associated with ZNF32 were “SPERMATOGENESIS”, “OXIDATIVE PHOSPHORYLATION”, and “DNA REPAIR”. Conversely, the top 3 downregulated pathways associated with ZNF32 were “P53 PATHWAY”, “PI3K AKT MTOR SIGNALING” and “HEME METABOLISM” (Fig. 5B). ZNF32 may promote cancer by regulating the tumor microenvironment and immune and inflammatory responses.

3.5 Relationship between ZNF32 expression and tumor microenvironment

The expression of ZNF32 was associated with the TME and immune cell infiltration. Increasing evidence suggests that the TME is a complex environment composed of extracellular matrix, immune cells, growth factors, fibroblasts, and cancer cells, which is essential for multidrug resistance, tumor growth, and metastasis [18]. We utilized the ESTIMATE algorithm to analyze the association between ZNF32 and TME composition. Our result showed that the ZNF32 expression was associated with TME, including the top three pathways nucleotide excision repair, base excision repair, and DNA replication (Fig. 6A). Additionally, 12 TME-related pathways in HNSC were examined, encompassing immune, stromal, and DNA repair pathways. ZNF32 expression was closely linked with DNA repair pathways, such as mismatch repair, nucleotide excision, base excision, DNA replication, and DNA damage response (Fig. 6B). All the details about observed associations were listed in Supplementary Table 1.

Fig. 6

Expression of ZNF32 is associated with TME. A Relationship between ZNF32 expression and twelve TME processes. Red indicates a positive correlation (correlation coefficient > 0), while blue indicates a negative correlation (correlation coefficient < 0) . B The CIBERSORT method statistical chart demonstrates the difference in TME levels between ZNF32 high and low expression groups in HNSC, with red representing high ZNF32 expression and yellow representing low expression

Surrounding cells in the TME have impacts on the behavior of tumors. In addition, we estimated Tumor purity and TME-related scores utilizing ESTIMATE. By exploring the connection of ZNF32 expression with Tumor purity, our study showed a positive connection in GBM, BRCA, THYM, UCEC, BLCA, PRAD, SARC, MESO, LG, and THCA. ZNF32 expression was negatively related to ESTIMATE scores in LUSC, GBM, BRCA, THYM, UCEC, PCPG, BLCA, PRAD, SARC, MESO, LGG, and THCA. We also ascertained that ZNF32 expression was linked to Stromal Score in TGCT, KIRC, LUSC, GBM, BRCA, UCEC, PCPG, BLCA, PRAD, SARC, MESO, LGG, and THCA. Moreover, low ZNF32 expression was associated with high immune scores in TGCT, STAD, CESC, LUSC, BRCA, THYM, UCEC, BLCA, PRAD, ACC, SARC, LGG and THCA (Fig. 7A, B).

Fig. 7

Correlation of ZNF32 expression with the tumor microenvironment in pan-cancer. A ZNF32 correlated Stromal Score, ESTIMATEScore, ImmuneScore, and Tumor purity across cancers. B Correlation between ZNF32 and Tumor Purity, ESTIMATE Score, Immune Score, and Stromal Score in pan-cancer using radar plots

3.6 ZNF32 is associated with the tumor immune microenvironment and impacts the effectiveness of immunotherapy

Immunotherapy has recently become known as a promising strategy in tumor therapy, leading us to examine ZNF32 expression and immune factors, such as immune cells and biomarkers. We initially examined the connection between ZNF32 and immune cells using 3 separate data sources. ZNF32 exhibited a robust association with a large number of immune cells across cancers. In the ImmuCell AI database, naive CD8 T cells and effector memory T cells were identified as the predominant immune cells associated positively with ZNF32 across most tumors, whereas macrophages were the most prevalent negatively correlated cells (Fig. 8A). The conclusion was confirmed in meanwhile utilizing the TIMER2 database, and the results largely agreed with the ImmuCell AI database (Fig. 8B). Overall, these databases revealed a strong relationship between ZNF32, naïve CD8 T cells, and effector memory T cells in pan-cancer.

Fig. 8

The role of ZNF32 in TME. A The role of ZNF32 on immune cell infiltration using data from ImmuCellAI database. B The role of ZNF32 on immune infiltration using the data from the TIMER2 database

Immune checkpoint blockade (ICB) has become a significant anticancer treatment, showcasing unparalleled survival advantages [19]. Preliminary findings indicated that ZNF32 significantly affected immune cell infiltration within the TME, and the efficacy of immunotherapy was strongly correlated with the immune checkpoints of these immune cells. Figure 9A–C showed three groups of patients who were treated for NSCLC, UCC, and KIRC. To get a better understanding of how ZNF32 expression was linked to the prognosis of these tumor patients undergoing immunotherapy, we analyzed the differences in therapeutic response and prognosis across ZNF32 expression subgroups. As expected, high expression of ZNF32 was closely linked to better therapeutic response and prognosis in tumor patients receiving immunotherapy. These findings suggested that ZNF32 could be a predictive biomarker for response to immunotherapy. Further clinical validation was required to confirm the association between ZNF32 and immunotherapy response found in this study.

Fig. 9

Low expression of ZNF32 has a negative impact on the prognosis and tumor immunotherapy response. A Kaplan–Meier analysis of OS between low- and high-ZNF32 groups of late-stage non-small cell lung cancer patients in the GSE61676 dataset. B Kaplan–Meier analysis comparing OS in urothelial carcinoma patients from the TCGA database between groups with high and low ZNF32. C Kaplan–Meier analysis of PFS in renal clear cell carcinoma patients with anti-PDL1 in the high- and low-ZNF32 groups

3.7 ZNF32 affects the biological function of HNSC cells

To validate the function of ZNF32, we chose human head and neck cancer for our further investigation. We established stable ZNF32 overexpressing cell lines in FaDu and CAL27 cells (Fig. 10A, B). Overexpression of ZNF32 would promote cell viability of FaDu and CAL27 cells by CCK-8 assay (Fig. 10C). Similarly, we used colony formation assay to find that the ZNF32-overexpressed group induced the proliferation capacity of FaDu and CAL27 cells (Fig. 10D). Next, we investigated ZNF32's possible function in the migration of CAL27 and FaDu cells using Transwell (Fig. 10E). The ZNF32-overexpressed group considerably enhanced the cell migration capacity of FaDu and CAL27 cells. We also found that ZNF32 increased the invasive abilities of HNSC cells (Fig. 10F). Further, ZNF32 influenced the proliferation of HNSC cells by increasing the proportion of cells in the S phase using flow cytometry (Fig. 10G). In vitro experiments indicated that ZNF32 was a risk factor for HNSC and the overexpression of ZNF32 would promote the progression of HNSC. Finally, we preliminarily explored the underlying mechanisms based on previous bioinformatics results. Our findings revealed that overexpressed ZNF32 significantly enhanced the PD-L1 and CTLA-4 expression in cells, although no statistically significant difference was observed in the expression of PD-L2 (Fig. 10H). Particularly, T cells and tumor immunity were strongly correlated with both CTLA-4 and PD-L1.

Fig. 10

ZNF32 promotes the malignant behavior of HNSC cells. A, B Overexpression of ZNF32 affects associated with mRNA expression in FaDu and CAL27 cells. C CCK8 cell proliferation assay results for CAL27 and FaDu cells. D Cell clone formation in 4 cell lines and their statistical analysis. E Four cell lines migration assay results and statistical analysis. F Four cell lines invasion assay results and statistical analysis. GThe cell cycle was characterized using Flow cytometry. H ZNF32 overexpression on linked mRNA expression in FaDu and CAL27 cells and statistical results. *p value < 0.05; **p value < 0.01; ***p value < 0.001;****p value < 0.0001