The expression of AURKA and TRIM28 in pan-cancer

In order to explore the expression of AURKA and TRIM28 in cancer, we analyzed their expression levels by using TIMER database and found that higher expressions of AURKA and TRIM28 in tumor tissue compared to that in normal tissue in HNSCC data set respectively (Fig. 1A and B). We next analyzed the correlations between AURKA expression and TRIM28 expression by using GEPIA. BioGRID and Cytoscape database showed Ten proteins (TRIM28, GMNN, NDN, PML, CHUK, BRCA1, XPO1, CCNE1, TRRAP and NCOR2) are related proteins of AURKA regulating cell cycle transcription factor E2F4 (Fig. 1C).

Fig. 1

TRIM28 expression is higher in HNSCC cancer tissue in different cancer types including HNSCC cancer and AURKA positively correlated TRIM28. A, B Differential of AURKA and TRIM28 mRNA expressions between tumor tissue and adjacent normal tissue in TCGA cancer types using TIMER, Gene_DE module. Distributions of gene expression levels are displayed using box plots. C Ten proteins are related proteins of AURKA regulating cell cycle transcription factor E2F4

TRIM28 was highly expressed in LSCC and correlated to TNM stage

As Fig. 2A and B showed that TRIM28 was highly expressed in LSCC compared to normal tissues in GEO (GSE59102) and TCGA-LSCC data sets. By using GEPIA database, the result was showed that TRIM28 was also highly expressed in HNSCC compared to normal tissues (Fig. 2C). In order to further explore the correlation of TRIM28 and LSCC, the results of UALCAN database indicated that the high expression of TRIM28 was correlated to TNM stage (Fig. 2D and E).

Fig. 2

The relationship between the mRNA expression of TRIM28 and the prognosis of patients with LSCC. Kaplan–Meier survival curve comparison of the higher and lower expression of TRIM28. A, B The expression of TRIM28 is significantly higher in LSCC tissue than normal tissue in GEO (GSE59102) and TCGA-LSCC data sets. C GEPIA analysis on differential expression of TRIM28 mRNA between tumor tissue and adjacent normal tissue in HNSCC data set of TCGA. D TRIM28 expression in different N stages. E TRIM28 expression in different tumor grades

Metabolic process was the most significantly GO biological process enriched with genes related to TRIM28

By using Linked Omics database in HNSCC, we found that 249 genes were positively correlated with TRIM28, 53 genes were negatively correlated TRIM28 (Fig. 3A). Then, we used Metascape to analyze the main enriched Gene Ontology (GO) biological processes of TRIM28 and 302 related genes. They were mainly focused on metabolic process, cellular process, response to stimulus (Fig. 3B). A similar effect was seen in protein–protein interaction (PPI) networks identified by functional cluster analysis (Fig. 3C and D).

Fig. 3

GO and PPI analyses of TRIM28-correlated genes. A The genes highly correlated with TRIM28 identified by Pearson correlation analysis in the HNSCC cohort. B Main GO biological processes of all 302 genes correlated with TRIM28. C The PPI networks determined by functional cluster analysis. The different colors represent different functional clusters. D The PPI networks determined by functional cluster analysis. The different colors represent different P values

TRIM28 is highly expressed in LSCC tissues and cells

In order to explore the role of TRIM28 in promoting the metastasis of LSCC, the expression of TRIM28 was investigated between laryngeal tumor and non-tumor tissues. Results of qRT-PCR were shown that the mRNA level of TRIM28 was observably higher in tumor tissues than non-tumor tissues (**P < 0.01, Fig. 4A, B). Furthermore, 22 of 30 LSCC tissues were categorized as TRIM28 protein strong expression, whereas 8 LSCC tissues and 30 adjacent normal tissues were categorized as TRIM28 protein weak expression. Typical immunostaining of TRIM28 is shown in Fig. 4C. Furthermore, the mRNA and protein levels of TRIM28 were examined in LSCC cells line (Hep2, TU686, TU212 cells). The results shown that the mRNA and protein levels of TRIM28 were higher in TU686 and TU212 cells, whereas lower in Hep2 cells (Fig. 4D–F). All observations indicated that TRIM28 is highly expressed in LSCC tissues and cells.

Fig. 4

TRIM28 overexpresses in human LSCC tissues and cells. A, B Expression of TRIM28 mRNA in 30 pairs of LSCC tissues and adjacent non-tumor laryngeal tissues was quantified by qRT-PCR. Data are shown as 2 − ∆Ct (**P < 0.01). C IHC staining of TRIM28 in LSCC tissues (magnification: × 10). D Patients with TRIM28 weak staining had a significantly benign prognosis than those with strong staining, P = 0.0056. E The TRIM28 mRNA expression level in LSCC cell lines was quantified by qRT-PCR. F The TRIM28 protein expression level in LSCC cell lines was examined by Western Blot. G Protein ratio in LSCC cells

AURKA may regulate TRIM28 in LSCC cells

Considering that TRIM28 was closely related to the occurrence and development of LSCC, the Co-IP assay indicated AURKA could interact with TRIM28 (Fig. 5A, B). And the expression of TRIM28 was reduced when the expression of AURKA was knockdown (Fig. 5C, D), while the expression of AURKA was not altered when the expression of TRIM28 was knockdown (Fig. 5E, F), which indicated AURKA could regulate TRIM28 in LSCC cells.

Fig. 5

AURKA regulates TRIM28 in LSCC cells. A, B Co-IP assay showed AURKA interacted with TRIM28. C The expressions of AURKA and TRIM28 were reduced in TU686/si-AURKA and TU212/si-AURKA cells. D Protein ratio in cells. E The expression of TRIM28 was reduced in TU686/si-TRIM28 and TU212/si-TRIM28 cells. F Protein ratio in cells

TRIM28 promoted dormancy-activated effects induced by AURKA

To further investigate TRIM28 played an important role in reviving dormant LSCC cells, biological cytology experiments were performed. Results showed TU686/si-TRIM28 and TU212/si-TRIM28 cells possess the lower ability of proliferation when compared to control groups (TU686/si-NC and TU212/si-NC cells, *P < 0.05, **P < 0.01, Fig. 6A, B). Moreover, the results of a colony formation assay exhibited the number of colonies of TU686/si-NC (401 ± 55.5) and TU212/si-NC cells (677 ± 110.0) were more than TU686/si-TRIM28 (156 ± 46.5) and TU212/si-TRIM28 cells (285 ± 56.0, *P < 0.05, Fig. 6C, D). And as flow cytometry showed that the percentage of G0/G1 cells was significantly higher in TU686/si-TRIM28 and TU212/si-TRIM28 cells (*P < 0.05, Fig. 6E, F). As Fig. 6G and H showed the dormancy-related proteins were also altered. The expressions of E2F4 and P130 were dramatically increased, and the expression of P107 was observably decreased in TU686/si-nc and TU212/si-nc cells. Finally, Co-IP assay demonstrated the E2F4-P130 complex, unique in quiescent cells, was expressed in TU686/si-TRIM28 cells (Fig. 6I, J). All results suggested that TRIM28 may promote dormancy-activated effects induced by AURKA in LSCC.

Fig. 6

TRIM28 promotes dormancy-activated effects induced by AURKA. A, B Knockdown of TRIM28 reduced cell proliferation of TU686 and TU212 cells (*P < 0.05, **P < 0.01). C, D Knockdown of TRIM28 induced the clone formation of TU686 and TU212 cells (*P < 0.05). E, F The percentage of G0/G1 cells was significantly higher in TU686/si-TRIM28 and TU212/si-TRIM28 cells (*P < 0.05). G The dormancy-related proteins were examined by Western Blot. The expressions of E2F4 and P130 were increased, the expressions of E2F4 and P130 were reduced in TU686/si-TRIM28 and TU212/si-TRIM28 cells. H Protein ratio in cells. I, J The complex of E2F4-P130 was detected by Co-IP

Blocking TRIM28 impaired the metastasis of LSCC in nude mice

Furthermore, cells were inoculated into nude mice by tail vein injection. Four weeks after inoculation, TU686/si-TRIM28 (1.0 ± 1.0), TU212/si-TRIM28 cells (0.7 ± 0.6) with lower TRIM28 expression showed less frequent lung metastases as compared to TU686/si-nc (4 ± 1.5) and TU212/si-nc cells (4 ± 1.7, *P < 0.05, Fig. 7A–D). All above observations indicated that TRIM28 promoted the metastasis of LSCC in nude mice.

Fig. 7

TRIM28 induces the metastasis of LSCC in nude mice. A, C TU686/si-TRIM28, TU212/si-TRIM28, TU686/si-nc and TU212/si-nc cells were inoculated into nude mice and pulmonary nodules were observed after 28 days (N = 5/group). H&E stains of pulmonary nodules (100 ×). B, D Pulmonary tissue and nodules were quantified by H&E staining from TU686/si-TRIM28, TU212/si-TRIM28, TU686/si-nc and TU212/si-nc cells (*P < 0.05)

The dormancy-activated effects induced by AURKA in LSCC are mediated through the activation of Akt signaling pathway

The AKT signaling pathway played a crucial role in the migration and invasion of cancers [7,8,9,10,11]. We suggested that TRIM28 might mediate AKT signaling pathway to promote LSCC metastasis. To better illustrate the important role of AKT in LSCC, Triciribine [12], a AKT inhibitor, was used to treat TU686 and TU212 cells (TU686/Triciribine and TU212/Triciribine cells). As shown in Fig. 5A and B, levels of TRIM28 and p-AKT was lower, level of AKT was not altered in TU686/si-TRIM28 and TU212/si-TRIM28 cells. And level of p-AKT was lower, while levels of TRIM28 and AKT were not changed in TU686/Triciribine and TU212/Triciribine cells, which indicated Akt was the downstream molecules of TRIM28 (Fig. 8C, D).

Fig. 8

Akt signaling pathway plays an important role in the dormancy-activated effects induced by AURKA in LSCC. A The expressions of TRIM28 and p-Akt were decreased, the expression of Akt was not altered in TU686/si-TRIM28 and TU212/si-TRIM28 cells. B Protein ratio in cells. C The expression of p-Akt was decreased, the expressions of TRIM28 and Akt were not altered in TU686/Triciribine and TU212/Triciribine cells. D Protein ratio in cells

Downregulation of Akt expression reduces LSCC cell proliferation, migration, and invasion

Moreover, biological cytology experiments were performed to further identify the role of AKT in the metastasis of LSCC. Results indicated TU686/Triciribine and TU212/Triciribine cells possessed lower cells proliferation when compared to control groups (TU686/parental and TU212/parental cells, *P < 0.05, **P < 0.01, Fig. 9A). In wound-healing assays, TU686/Triciribine and TU212/Triciribine cells were less motile at 48 h compared with TU686/parental and TU212/parental cells (*P < 0.05, Fig. 9B). Moreover, the results of a colony formation assay exhibited the number of colonies of TU686/Triciribine (459 ± 84.5) and TU212/Triciribine cells (453 ± 77.7) were less than TU686/parental (758 ± 111.6) and TU212/parental cells (861 ± 114.3, *P < 0.05, **P < 0.01, Fig. 9C, D). In addition, less TU686/Triciribine (78 ± 11.0) and TU212/Triciribine cells (45 ± 10.5) migrated through transwell chambers compared with TU686/parental (123 ± 8.5) and TU212/parental cells (85 ± 9.3), and TU686/Triciribine (52 ± 4.2) and TU212/Triciribine cells (24 ± 6.7) invaded through matrigel less frequently than TU686/parental (86 ± 7.6) and TU212/parental cells (52 ± 11.8, *P < 0.05, **P < 0.01, Fig. 9E–H). All results suggested that the AKT signaling pathway may promote dormancy-activated effects induced by TRIM28.

Fig. 9

Downregulation of Akt expression reduces LSCC cell proliferation, migration, and invasion. A Knockdown of Akt reduced cell proliferation (*P < 0.05, **P < 0.01). B Knockdown of Akt reduced cell movement (*P < 0.05). C, D Knockdown of Akt induced the clone formation of TU686 and TU212 cells (*P < 0.05). E–H Knockdown of Akt induced the ability of migration and invasion of TU686 and TU212 cells (*P < 0.05, **P < 0.01)