ZNF71 was downregulated at the protein level in LSCC tissues

According to the results from the HPA website, the ZNF71 protein is mainly localized to the nucleoli (Figure S1). In the IHC staining of non-cancerous laryngeal tissues, we observed that the staining of the nucleus was positive (Fig. 1A, B), while in the IHC staining of LSCC tissues, the staining of whole tissues was negative (Fig. 1C, D). To further quantify the results, we calculated the statistical difference between the IHC scores of the LSCC and non-LSCC tissues, and the results showed that the expression of ZNF71 was downregulated in LSCC tissues (p < 0.0001, Fig. 1E). And the ROC curve indicated that the ZNF71 protein has great discriminatory ability between LSCC and non-LSCC tissues (p < 0.0001, Fig. 1F). Moreover, the correlation analysis between the protein expression of ZNF71 and clinicopathologic parameters were conducted. The results showed us that patients with different T stage had no difference of ZNF71 expression (p > 0.05, Fig. 1G). However, patients with nodal metastasis had lower protein expression level of ZNF71 than patients without nodal metastasis (p < 0.05, Fig. 1H).

Fig. 1

IHC staining of ZNF71 protein in laryngeal squamous cell carcinoma (LSCC) and non-LSCC tissues. A, B Expression level of ZNF71 protein in normal laryngeal tissues (left: ×200; right: ×400). C, D Expression level of ZNF71 protein in LSCC tissues (left: ×200; right: ×400). E Scatter plot of IHC score. F The ROC curve of ZNF71 expression. G, H Differentially expression analysis of ZNF71 protein between different T stage (G) and N stage (H). Note: ns: non-significant; *p < 0.05; **p < 0.01

ZNF71 was downregulated at the mRNA level in LSCC based on comprehensive analysis

We comprehensively analyzed the multiple datasets to clarify the exact expression status of ZNF71 in LSCC. The forest plot showed that ZNF71 was downregulated at mRNA level in LSCC (total SMD: −0.22, 95% CI, −0.42–−0.03, Fig. 2A). Sensitivity analysis (Fig. 2B) showed that the results from the included datasets were robust. We tested the publication bias, and the Begg’s test (p = 0.283, Fig. 2C) and Egger’s test (p = 0.438, Fig. 2D) showed that there was no publication bias. After comprehensive analysis, it was confirmed that the expression of ZNF71 was downregulated in LSCC tissues. Furthermore, we integrated the mRNA data and the protein data of in-house IHC, and we found the downregulation status of ZNF71 in LSCC tissues (total SMD: −0.40, 95% CI, −0.58–−0.21, Fig. 2E).

Fig. 2

Identification of ZNF71 mRNA expression between LSCC and non-cancerous laryngeal tissues based on comprehensive analysis. A Forest plot of ZNF71 expression of included datasets. B Sensitivity analysis of included analysis; Begg’s test (C) and Egger’s test (D) for publication bias test of included datasets. E Forest plot of ZNF71 expression integrating the mRNA data and protein data of in-house IHC

To explore the discriminatory ability of ZNF71 expression between LSCC and non-cancerous laryngeal tissues, we analyzed the ROC curves of all included datasets and calculated the AUC values (Fig. 3A). We further analyzed the sROC curve to describe this discriminatory ability more intuitively. The results of the sROC showed that ZNF71 has moderate discriminatory ability in LSCC (AUC = 0.71, 95% CI, 0.67–0.75, Fig. 3B). Deek’s test showed that the included datasets had no publication bias (p = 0.969, Fig. 3C). The total sensitivity of the included datasets was 0.50 (95% CI, 0.39–0.60, Fig. 3D), and the total specificity was 0.89 (95% CI, 0.78–0.95, Fig. 3E).

Fig. 3

Discriminatory ability analysis of ZNF71 between LSCC and non-cancerous laryngeal tissues. A The ROC curve of ZNF71 in each included dataset. B Summary ROC curve of ZNF71. C Deek’s test for publication bias test; sensitivity (D) and specificity (E) analysis of ZNF71 expression in LSCC and non-LSCC samples based on included datasets

We also analyzed the relationship between the expression of ZNF71 and clinicopathological parameters of LSCC patients. Male LSCC patients had lower mRNA expression level of ZNF71 than female LSCC patients (p = 0.0046, Fig. 4A). But there was no expression difference in non-cancerous samples across genders (p > 0.05, Figure S2). Among other clinicopathological parameters, there was no significant difference in the expression of ZNF71 (p > 0.05, Fig. 4B–G). And ZNF71 expression was not related to survival time (p > 0.05, Fig. 4H).

Fig. 4

Clinical significance analysis of ZNF71 expression and gender (A), T_clinic (B), N_clinic (C), M_clinic (D), stage_clinic (E), stage_pathologic (F), N_pathologic (G), and survival time (H)

ZNF71 was absent in LSCC in different subpopulations based on scRNA-seq analysis

In addition to the protein level and mRNA level, we further explored the expression of ZNF71 in LSCC at the cell level using scRNA-seq data. Firstly, after analyzing the scRNA-seq data and setting appropriate parameters, we divided the cells of LSCC into 15 cell clusters (Fig. 5A). Then, we analyzed and visualized the expression of ZNF71 in these cell clusters with the gene mapping map (Fig. 5B) and violin plot (Fig. 5C). The results showed that there was no expression of ZNF71 in various LSCC cells. Finally, we annotated the cell types according to the marker genes of these cell clusters and found that these cell types are cancer cells, plasma cells, macrophages, M2-macrophages, B cells, T cells, naive T cells, fibroblasts, epithelial cells, endothelial cells, and unidentified cell types (Fig. 5D).

Fig. 5

Expression status analysis of ZNF71 based on single cell RNA sequencing data. A Dimensionality reduction and clustering of screened single cells. B Expression status analysis of ZNF71 in identified clusters. C Violin plot of ZNF71 expression in each cluster. D Cell type annotation for each cluster

Potential molecular mechanism of ZNF71 in LSCC

To explore the potential molecular mechanism of ZNF71, we constructed a PPI network of ZNF71 and its PDCEGs and conducted enrichment analysis through the STRING website. The interactive relationship is presented in Fig. 6A. These PDCEGs of ZNF71 were mainly enriched in biological regulation, regulation of biological progress, and regulation of cellular progress in the biological process (BP). In cellular components (CC), they concentrated on the nucleus. In terms of molecular function (MF), they were mainly enriched in binding, iron binding, and organic cyclic compound binding. On the Reactome pathway, they were enriched in gene expression and the generic transcription pathway (Fig. 6B).

Fig. 6

Potential molecular mechanism analysis of ZNF71 in LSCC. A Protein-protein interactive network of ZNF71 and its positively and differentially co-expressed genes (PDCEGs). B Enrichment analysis of ZNF71 and its PDCEGs. C–F Biological process (BP), cellular component (CC), molecular function (MF), and KEGG pathway signaling analysis for downstream target genes of ZNF71

As ZNF71 is a transcription factor, we further predicted the downstream target genes of ZNF71 through Cistrome DB. We conducted enrichment analysis of ZNF71 and its target genes to explore the potential molecular mechanism of ZNF71 in the occurrence and development of LSCC. In the BP, these genes concentrated on protein targeting (Fig. 6C). In terms of CC, they were mainly enriched in the apical plasma membrane and the apical part of cells (Fig. 6D). On the MF, they were relative to heme binding and tetrapyrrole binding (Fig. 6E). On KEGG signaling pathways, they mainly participated in tight junctions (Fig. 6F). We then further analyzed the expression level of these tight junctions-related genes, and we found they were significantly downregulated in LSCC tissues (Figure S3). And many tight junction-related genes were positively correlated with ZNF71 expression (Table S5).

The correlation between the expression of ZNF71 and tumor immune infiltration

We further evaluated the relationship between ZNF71 expression and immune cells and stromal cells in the tumor immune microenvironment (TME) through the ESTIMATE algorithm. The expression of ZNF71 was positively correlated with the ESTIMATE comprehensive score (Pearson r = 0.2037, p = 0.0320, Fig. 7A). The expression of ZNF71 was positively correlated with immune cell score (Pearson r = 0.2433, p = 0.0101, Fig. 7B), but not with stromal cell score (Pearson r = 0.1198, p = 0.2103, Fig. 7C). Furthermore, based on the TISIDB online tool, we further analyzed the relationship between ZNF71 expression and 28 kinds of tumor-infiltrating lymphocytes. The results showed that the expression of ZNF71 was positively correlated with tumor-infiltrating lymphocytes, such as neutrophils, activated dendritic cells, type 17 helper cells, and monocytes (Fig. 7D).

Fig. 7

Correlation analysis between ZNF71 expression and ESTIMATE score (A), immune score (B), stromal score (C), and tumor infiltration lymphocytes (TILs, D)