USP44 regulates HEXIM1 stability to inhibit tumorigenesis and metastasis of oral squamous cell carcinoma

Abstract

Oral squamous cell carcinoma (OSCC) is the most frequent type of oral malignancy with high metastasis and poor prognosis. The deubiquitinating enzyme Ubiquitin Specific Peptidase 44 (USP44) regulates the mitotic checkpoint, and its deficiency leads to aneuploidy and increases tumor incidence. However, the role of USP44 in OSCC is not well understood. Herein, we analyzed mRNA sequencing data of OSCC samples downloaded from the TCGA and GEO databases and found that USP44 was decreased in human OSCC tissues and was positively correlated to the survival of OSCC patients. To investigate the biological impact of USP44, we used recombinant lentiviruses to overexpress or knockdown USP44 expression in OSCC cell lines, which were also injected subcutaneously or into the lateral tail vein of Male BALB/c nude mice to model tumorigenesis or lung metastasis in vivo, respectively. The results showed that overexpression of USP44 inhibited malignant cell phenotypes in vitro and suppressed tumor growth and lung metastasis in vivo, while its downregulation had the opposite effects. Comprehensive proteomic analyses through Co-IP mass spectrometry and label-free quantitative LC-MS/MS methods identified 112 differentially expressed proteins positively regulated by USP44, among which 13 were involved in cancer-related pathways including apoptotic signaling and cell cycle regulation. PPI analysis identified Hexamethylene Bis-Acetamide-Inducible Protein 1 (HEXIM1) as the hub protein. Upregulation of USP44 enhanced HEXIM1 protein stability, leading to its higher expression in OSCC cells. Silencing of HEXIM1 further enhanced the malignant phenotype of OSCC cells. At the same time, HEXIM1 knockdown reversed the antitumor effects of USP44. These findings demonstrated that USP44 acted as a critical tumor suppressor in OSCC by inhibiting cell proliferation and metastasis through the stabilization of HEXIM1 protein, suggesting that USP44-HEXIM1 axis is a promising target for OSCC therapy.

Graphical Abstract

Introduction

Oral squamous cell carcinoma (OSCC) originates from the squamous tissues in the oral cavity under the category of head and neck squamous cell carcinoma (HNSC) [1]. Despite the widespread use of surgery combined with adjuvant chemoradiotherapy or induction therapy with surgical salvage, the overall 5-year survival rate of OSCC patients has declined due to the delayed diagnosis over the last 20 years [2]. Tobacco and alcohol consumption, which result in a slight male predominance, are major factors in the high prevalence of OSCC [3]. In addition, OSCC can also arises from genetic alterations. Discovering driver genes responsible for the malignancy of OSCC could provide new therapeutic strategy for OSCC.

Ubiquitin Specific Peptidase 44 (USP44), a deubiquitinating enzyme, has been proven to be an important regulator of the mitotic checkpoint [4, 5]. USP44-depleted cells were unable to sustain activation of the mitotic checkpoint due to premature entry into anaphase of cell cycles [6]. Recent studies have shown that USP44 plays a dual role in cancer progression. Zhang et al. showed that USP44^–/– animals exhibited aneuploidy and increased tumor incidence, suggesting USP44 being a novel tumor suppressor [4]. In clinic, nasopharyngeal cancer patients with low USP44 expression had a poor prognosis and were prone to tumor relapse [7]. Upregulation of USP44 enhanced the radiotherapy sensitivity of nasopharyngeal cancer cells [7]. As a negative regulator of histone H2B ubiquitylation, USP44 suppressed embryonic stem cell differentiation by reversing the mono-ubiquitination of H2B [8]. Conversely, downregulation of USP44 suppressed the cancer stem cell-like behaviors of prostate cancer cells [9]. These findings suggest that USP44 functions as a key regulator in cancer development. However, its pathophysiological importance in OSCC is not well understood.

In this study, we explored the role and mechanism of USP44 in OSCC. We found the reduced USP44 expression in OSCC was closely associated with the poorer patient survival. USP44 functions as a tumor suppressor to inhibit cell proliferation and metastasis in OSCC both in vivo and in vitro. USP44 regulates the ubiquitination level and, consequently, the protein stability of the tumor suppressor Hexamethylene Bis-Acetamide-Inducible Protein 1 (HEXIM1), an inhibitor of positive transcription elongation factor b (P-TEFb), thereby inhibiting cell proliferation, migration, and invasion of OSCC. These results suggest that USP44, as well as HEXIM1, may serve as a promising target for OSCC therapy.

Materials and methods

Clinical OSCC samples

In this study, two clinical patient cohorts were collected from The First Affiliated Hospital of Zhengzhou University (Henan, China). Sample set 1: 30 pairs of fresh-frozen OSCC and adjacent normal tissues for RT-qPCR assay. Sample set 2: 86 paraffin embedded tissues for immunohistochemical detection. None of patients have received anticancer treatment before biopsy.

Cell lines, lentivirus infection and small interfering RNAs (siRNAs) transfection

Human immortal keratinocyte cell line HaCaT was a kind gift from Peking University School and Hospital of Stomatology (Beijing, China). The cell lines SCC-4, SCC-9, SCC-25, CAL27 and SAS were purchased from iCell Bioscience Inc (Shanghai, China) and Guangzhou Cellcook Biotech Co., Ltd. (Guangzhou, China), respectively. HaCaT and CAL-27 cells were cultured in Dulbecco’s modified Eagle medium (DMEM, Servicebio, Wuhan, China) supplemented with 10% fetal bovine serum (FBS). SCC-4, SCC-9, SCC-25 and SAS cells were cultured in DMEM/F12 (Biosharp life sciences, Hefei, China) supplemented with 10% FBS, respectively. Cells were cultured in an incubator of 5% CO₂ at 37 °C.

To construct a lentivirus-mediated USP44 silencing vector (pLKO.1-shUSP44-1 and pLKO.1-shUSP44-2), two shRNA sequences targeting USP44 (shUSP44-1 and shUSP44-2) were synthesized and cloned into the pLKO.1-EGFP-puro vector (FH1717, Fenghui Biotechnology Co., Ltd, Hunan, China) by Shanghai Genechem Co., Ltd. (Shanghai, China), so was the negative control sequence fragment (shNC). To construct a lentivirus-mediated USP44 overexpression vector (pLJM1-USP44), the full-length cDNA sequences of USP44 (Accession: NM_032147, CDS, 276–2414) were cloned into the pLJM1-EGFP-puro vector (ZT101, Fenghui Biotechnology Co., Ltd). The empty pLJM1-EGFP-puro was used as the negative control lentiviral vector (pLJM1-Ctrl). For stable transfection, CAL-27 and SCC-9 cells were transduced with lentiviruses packaged using plasmids pLKO.1-shUSP44-1, pLKO.1-shUSP44-2, pLKO.1-shNC, pLJM1-USP44 or pLJM1-Ctrl. After stable selection with puromycin (1.5 µg/ml), the cells were harvested and analyzed using quantitative Real-Time PCR (RT-qPCR) and western blotting assays to determine the expression levels of USP44. To knock down HEXIM1, siRNA targeting HEXIM1 (siHEXIM1) was synthesized by JTS scientific (Wuhan, China). The indicated cells were transfected with the siRNAs according to the manufacturer’s instructions. All the shRNA and siRNA sequences used in this study are as follows:

shNC 5’TTCTCCGAACGTGTCACGT3’;

shUSP44-1 5’GCTTCAAAGTGAAGATCAACT3’;

shUSP44-2 5’AGTGTATGATGTTATTCAAAT3’;

siNC (sense strand) 5’UUCUCCGAACGUGUCACGUTT3’, and (anti-sense strand) 5’ACGUGACACGUUCGGAGAATT3’;

siHEXIM1 (sense strand) 5’CGAUGACGACUUCAUGGAATT3’ and (anti-sense strand) 5’UUCCAUGAAGUCGUCAUCGTT3’.

Label-free quantitative proteomics and LC-MS/MS analysis

The label-free quantitative proteomics was performed by Beijing Qinglian Baiao Biotechnology Co., Ltd (Beijing, China). Briefly, total cellular protein was obtained by lysing the pLJM1-Ctrl or pLJM1-USP44 infected CAL-27 cells, and subsequently extracted peptide samples were analyzed by LC-MS/MS method. The chromatographic separation was performed on RIGOL L-3000 HPLC System (Rigol Technologies Co., Ltd, Beijing, China). The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid plus 80% acetonitrile in water (B) was delivered at a constant flow rate of 600 nl/min. The gradient elution program was adopted as follows: 8% B for 5 min, 12% B for 30 min, 30% B for 9 min, 40% B for 1 min and 95% B for 15 min. Data acquisition was performed on the Orbitrap Exploris™ 480 mass spectrometer (Thermo Fisher Scientific Inc., Pittsburgh, PA, USA) that was equipped with a Nanospray Flex™(NSI)ion source. Full scan range of mass spectrum was m/z 350–1500. The resolution was set to 120,000 (200 m/z) and 15,000 (200 m/z) for the primary and secondary mass spectra, respectively. Collision energy for peptide fragmentation was set to 33%. The raw MS/MS data was then exported for the protein identification in Proteome Discoverer 2.4 software. Dysregulated protein expression was considered significant at absolute fold change (|FC| ≥ 1.5) and P ≤ 0.05.

Bioinformatics analysis

The mRNA sequencing data of OSCC samples was downloaded from The Cancer Genome Atlas (TCGA, https://www.cancer.gov/ccg/research/genome-sequencing/tcga) database and GSE37991 data set from GEO database (https://www.ncbi.nlm.nih.gov/geo). The Gene Ontology (GO) analysis and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed in the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/home.jsp). The protein-protein interaction (PPI) network was analyzed by String (https://cn.string-db.org/). The structure of protein HEXIM1 was downloaded from AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/).

RT-qPCR and primers

Total RNA was extracted from fresh-frozen tissues or indicated cells using TRIpure reagent (RP1001, BioTeke Corporation, Beijing, China) following the manufacturer’s protocol, and cDNA was synthesized using random primers and the BeyoRT™ II M-MLV reverse transcriptase (D7160L, Beyotime Biotechnology, Shanghai, China). For RTqPCR, 1 µl cDNA was mixed with 1 µl primers (10 µM), 0.3 ul reaction buffer from SYBR Green kit (SY1020, Solarbio, Beijing, China) and 10 µl 2×Taq PCR MasterMix PCR amplification reagent (PC1150, Solarbio), followed by double-distilled H₂O replenishment to make a final volume of 20 µl. All reactions were performed on Exicycler 96 system (Bioneer Corporation, Korea) under the following conditions: 95˚C for 5 min, followed by 40 cycles of 95˚C for 10 s, 60˚C for 10 s and 72˚C for 15s. The target RNA expression was normalized to Actin Beta using the 2^−ΔΔCT method. The primer sequences for RT-qPCR are as follows:

USP44 (forward) 5’TGACGGTGCCCAATCTC3’, and (reverse) 5’ACTGAGCGAGCCCTTGT3’;

CD133 (forward) 5’CCAAGGACAAGGCGTTCA3’, and (reverse) 5’GCACCAAGCACAGAGGG3’;

Nanog (forward) 5’CACCTATGCCTGTGATTT3’, and (reverse) 5’CAGAAGTGGGTTGTTTGC3’;

OCT4 (forward) 5’AGGGCAAGCGATCAAGC3’, and (reverse) 5’GGAAAGGGACCGAGGAGTA3’;

SOX2 (forward) 5’ATGCACCGCTACGACGTGAG3’, and (reverse) 5’GCCCTGGAGTGGGAGGAAGA3’;

HEXIM1 (forward) 5’AAGAAGAAGCGGCATTG3’, and (reverse) 5’CTGGTGTCGTCGGATTT3’;

Actin Beta (forward) 5’GGCACCCAGCACAATGAA3’, and (reverse) 5’TAGAAGCATTTGCGGTGG3’;

Western blotting

To obtain the whole cell proteins, cells were lysed in ice-cold RIPA buffer (PR20001, Proteintech, Wuhan, China) mixed with 1% protease inhibitor (PR20032, Proteintech) for 30 min. Protein concentration was determined using BCA protein assay kit (PK10026, Proteintech) according to the manufacturer’s instructions. Subsequently, equal amounts of protein extracts (15 µg) were separated by SDS-PAGE gel electrophoresis and then transferred onto polyvinylidene fluoride (PVDF) membranes. After blocking in 5% skimmed milk solution for 2 h, the PVDF membranes were probed with diluted primary antibodies overnight at 4˚C, followed by incubation with HRP-conjugated Affinipure Goat Anti-Mouse IgG (1:10000, SA00001-1, Proteintech) and HRP-conjugated Goat Anti-Rabbit IgG (1:10000, SA00001-2, Proteintech) secondary antibodies for 40 min at 37˚C. The primary antibodies are as follows: anti-USP44 (1:400, sc-377203, Santa cruz, USA), Anti-HEXIM1 (1:3000, 15676-1-AP, Proteintech) and anti-Actin Beta (1:20000, 66009-1-Ig, Proteintech). The protein bands were detected using hypersensitivity ECL chemiluminescence detection kit (PK10003, Proteintech). The grayscale analysis of the target bands was performed on Gel-Pro-Analyzer software.

Cell proliferation analysis using CCK-8, colony formation and sphere formation assays

The proliferative capacity of the indicated CAL-27 and SCC-9 cells was assayed using a CCK-8 kit (KGA317, KeyGen Biotech, Jiangsu, China), according to manufacturer’s instructions. The absorbance of each well in 96-wells plate was determined using 800TS Microplate reader (BioTek, USA). For colony formation assay, cells were inoculated in Petri dishes (300 cells per dish) and incubated until colony formation (~ 2 weeks). The plates were then washed with PBS, fixed by 4% of paraformaldehyde and stained with Giemsa composite dye solution (KGA227, KeyGen Biotech). Colonies containing more than 50 cells were counted under an inverted IX53 microscope (OLYMPUS, Tokyo, Japan). For sphere formation assay, CAL-27 and SCC-9 cells were cultured for about 10 days in DMEM and DMEM/F12 mediums supplemented with 4 µg/mL insulin, B27 (1:50), 20 ng/mL EGF and 20 ng/mL basic FGF, respectively. The surviving colonies (spheroid diameter > 75 μm) were counted under IX53 microscope.

Transwell assays for cell migration and invasion

Cell migration and invasive ability was determined by Transwell assays. The invasion chambers (Corning, NY, USA) are covered by Matrigel gel (Corning) whereas the migration chambers are not. In brief, chambers with 800 µl culture medium containing 10% FBS in the lower chamber were placed onto 24-well plates. 200 µl cell suspensions (5 × 10⁴ cells/well for invasion assay and 5 × 10³ cells/well for migration assay) were placed into the upper chamber. After incubation at a 37℃ incubator with 5% CO₂, the chambers were washed twice with PBS, fixed with 4% paraformaldehyde (20 min) and then stained with 0.5% crystal violet (5 min) at room temperature. Cells invaded through the microporous membrane were counted under IX53 microscope.

Co-immunoprecipitation (Co-IP) assay

The endogenous binding between USP44 and HEXIM1 was detected in CAL-27 and SCC-9 cells. To verify the exogenous binding, 293T cells were transfected with ^Flag−USP44 and/or ^Myc−HEXIM1 plasmids. For the extraction of total protein, cells were lysed on ice with RIPA lysis buffer (PR20001, Proteintech). Co-IP assay was performed using the commercial Immunoprecipitation Detection Kit (PR40025, Proteintech) according to the instructions. Briefly, 500 µg total protein lysate was immunoprecipitated overnight at 4 °C with the indicated antibodies and 200 µl incubation buffer. 80 µl Protein G Magnetic Beads were then added into the immune complexes. After incubation at 4 °C for 2 h, the immunoprecipitation complex was washed with 1 × Washing buffer and then collected for SDS-PAGE separation. The USP44 and HEXIM1 in the Co-IP were detected by western blotting. The antibodies used are as follows: anti-USP44 (1:400, sc-377203, Santa Cruz), Anti-HEXIM1 (1:3000, 15676-1-AP, Proteintech), Anti-Myc-tag (1:3000, 60003-2-Ig, Proteintech) and Anti-Flag-tag (1:20000, 20543-1-AP, Proteintech).

Ubiquitination and protein stability detection of HEXIM1

To detect the effect of USP44 on the ubiquitination of HEXIM1 protein, 293T cells were co-transfected with ^Flag−USP44, ^Myc−HEXIM1 and ^HA−ubiquitin (Ub) plasmids. According to the instructions of the Immunoprecipitation Detection Kit (PR40025, Proteintech), the Co-IP assay was performed 48 h after transfection. The HEXIM1 and Ub in Co-IP were assessed by western blotting. The antibodies used are as follows: Anti-Myc-tag (1:3000, 60003-2-Ig, Proteintech) and Anti-HA-tag (51064-2-AP, Proteintech).

To evaluate the protein stability of HEXIM1, ov-NC or ov-USP44 plasmids-transfected CAL-27 cells were treated with 5 µg/ml cycloheximide (CHX) at 0, 0.5, 1, 3 and 6 h. The relative expression of HEXIM1 was measured by western blotting and normalized to Actin Beta expression.

Tumorigenesis and metastasis in vivo

Six weeks old male BALB/c nude mice were obtained from Jiangsu Huachuang Sino Pharmaceutical Technology Co., Ltd (Jiangsu, China). All mice were housed in a specific pathogenfree condition (12/12 h light/dark cycle) at 22 ± 1 °C with humidity of 45–55% and feeding ad libitum.

For subcutaneous tumor-forming model, mice were randomly assigned to 6 groups and administered a subcutaneous injection of 5 × 10⁶ CAL-27 or SCC-9 cells. The mice were sacrificed after 21 days of recording tumor volumes (volume = 0.5 × a × b², ‘a’ represents the longest diameter and ‘b’ represents the shortest diameter), and the tumor volume was measured every 3 days during this period. Tumor tissue was weighed and collected for immunohistochemical analysis. In our animal studies, the allowed maximal tumor size/burden (20 mm diameter) was not exceeded.

For establishing experimental lung metastases model, 2 × 10⁶ CAL-27 or SCC-9 cells were injected into the mouse lateral tail vein. Mice were sacrificed 8 weeks later, and the lungs were isolated and the metastatic pulmonary nodules were analyzed by H&E assay.

Immunohistochemical analysis

Immunohistochemical assay was performed to evaluate the expression of USP44, CD133 and Ki67 in tumor tissues. Specifically, the paraffin-embedded tumor tissues were cut into 5 μm sections. The sections were deparaffinized, rehydrated, and then placed in antigen repair solution for antigen repair under low heat. Afterwards, endogenous peroxidase was inhibited by incubation the tissue sections with 3% hydrogen peroxide for 15 min at room temperature. After blocking in 1% BSA for 15 min, the tissue samples were incubated with primary antibodies USP44 (1:50, A08401-2, Boster, Wuhan, China), CD133 (1:50, 18470-1-AP, Proteintech) and Ki67 (1:50, AF0198, Affbiotech, Changzhou, China) at 4˚C overnight, followed by secondary antibody HRP-conjugated Goat Anti-Rabbit IgG (1:500, #31460, ThermoFisher Scientific, Pittsburgh, PA, USA) for 60 min at 37˚C. Next, sections were stained with diaminobenzidine (DAB-1031, Maixin Biotechnology Co., Ltd., Fujian, China), counterstained with hematoxylin (H8070, Solarbio), and observed under an OLYMPUS BX53 microscope. The immunoreactions were evaluated according to the method described by Wang et al. [10]. Briefly, the staining intensity was categorized: 0 (no staining), 1 (weak), 2 (moderate), and 3 (strong). The percentage of cells strained was categorized: 0 (no positive cells), 1(< 25% positive cells), 2 (25 − 50% positive cells), 3 (50 − 75% positive cells), and 4 (> 75% positive cells). The final score was calculated by multiplying the intensity index with the percentage scale. The range of the final score was 0–12 and the median was 6. Tumors with scores ≤ 6 were considered as low expression, whereas those with scores > 6 were considered as high expression.

Statistical analysis

GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA) was applied to analyze and plot the data. All data were expressed as mean ± SD. Mean values between 2 groups were analyzed by Unpaired t test. One-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was used for analyzing statistical difference among multiple groups. Kaplan-Meier analysis followed by logrank test was applied for analyzing the prognostic performance of USP44. Chi-square tests were used for analyzing the correlation between USP44 expression and clinicopathological parameters of OSCC patients. P < 0.05 was considered as statistically significant difference.

Results

USP44 is downregulated in OSCC

OSCC-specific RNA-Seq data extracted from the TCGA-HNSC cohort and the GSE37991 dataset were analyzed by applying a threshold of | Log2 (fold change, FC) | >1 and P < 0.05 for significance. As shown in Figs. 1A, 1164 and 1906 differentially expressed genes (DEGs) were identified as downregulated in the GSE37991 and TCGA datasets, respectively. Conversely, 717 and 1881 DEGs were up regulated in these datasets. Notably, there were 685 downregulated and 407 upregulated DEGs common to both datasets (Fig. 1B). Kaplan-Meier analysis of these overlapping DEGs showed that USP44 was positively related to the survival of OSCC patients (Fig. 1C). Further, we analyzed the RNA-Seq data of UPS44 from GSE37991 and TCGA databases and found that the expression of UPS44 in the tumor samples was lower than that in the normal tissues (Fig. 1D). Consistently, we examined UPS44 expression in 30 paired fresh OSCC tumor and adjacent normal tissues and found a consistent reduction of USP44 expression in tumor tissues (Fig. 1E). Immunohistochemical staining of human OSCC tissues revealed USP44 expression in both the nucleus and cytoplasm of tumor cells (Fig. 1F). Intriguingly, patients with lower TNM stage (Fig. 1F(a-d)) exhibited higher expression of USP44 in tumor tissues, while those with higher TNM stage showed lower USP44 expression levels (Fig. 1F(e-h)). These results showed that the TNM stages are negatively correlated with USP44 expression (Table 1).

Fig. 1

Downregulation of USP44 in OSCC. (A) Volcano plot: all differentially expressed coding mRNAs from the GSE37991 (https://www.ncbi.nlm.nih.gov/geo) and The Cancer Genome Atlas (TCGA, https://www.cancer.gov/ccg/research/genome-sequencing/tcga) databases. (B) Venn diagrams: differentially expressed genes (DEGs) identified from GSE37991 and TCGA data sets. The overlaps display the up- or down-regulated DEGs common to both datasets. (C) Kaplan-Meier survival analysis of OSCC patients in TCGA data sets. (D and E) The expression levels of USP44 in OSCC tumor tissues and adjacent normal tissues using data from GSE37991, TCGA and samples from The First Affiliated Hospital of Zhengzhou University, respectively. (F) Representative immunohistochemical staining of USP44 in human OSCC tissues

Table 1 Correlation between USP44 expression and clinicopathological parameters of OSCC patients

USP44 inhibits proliferation, self-renewal, migration, and invasion of OSCC cells in vitro

We assessed the expression level of USP44 in human normal HaCaT cell line and other five OSCC cell lines (SAS, SCC-4, SCC-25, SCC-9 and CAL-27) by western blotting. Supplementary Fig. 1A shows that USP44 expression is significantly decreased in all OSCC cells compared to HaCaT cells. To explore the implications of USP44 downregulation in OSCC, we constructed CAL-27 and SCC-9 cells for stable overexpression or knockdown of USP44 (Supplementary Fig. 1B and 1 C). Downregulation of USP44 significantly promoted the viability and colony formation abilities of CAL-27 cells (Fig. 2A and C), whereas USP44 overexpressing cells exhibited a significant reduction in these characteristics (Fig. 2B and D). Furthermore, we found that knockdown of USP44 in CAL-27 and SCC-9 cells noticeably promoted cell migration and invasion, whereas its overexpression substantially impeded cell migration and invasion (Fig. 2E and H). These results demonstrated that downregulation of USP44 contributes to the enhanced malignancy of OSCC cells.

Fig. 2

USP44 inhibits OSCC cell proliferation, migration, and invasion in vitro. (A-B) CCK-8 assays showing that cell viability was enhanced in USP44-silenced OSCC cells but was decreased in USP44-overexpressed OSCC cells. (C-D) Clonogenic assays depicting the effects of USP44 on OSCC cell proliferation. (E-F) Transwell assays analyzing the migration and invasion capabilities of OSCC cells with altered USP44 expression. (G-H) Quantitation of Transwell assays in E and F

USP44 attenuates cancer stem cell-like characteristics in OSCC cells

To further validate the effect of USP44 on the malignant phenotypes of OSCC cells, the cancer stem cell markers including CD133, OCT4, SOX2 and Nanog in CAL-27 and SCC-9 cells were assessed by RT-qPCR. As described in Fig. 3A and B, the mRNA levels of these markers were significantly higher in OSCC spherical cells than those in their monolayer counterparts. Concurrently, RT-qPCR and western blotting assays showed that USP44 decreased in both CAL-27 and SCC-9 spherical cells (Fig. 3C and D). Besides, knockdown of USP44 enhanced the cancer stem cell-like behaviors in OSCC cells, while its overexpression inhibited the proportion of spherical cells (Fig. 3E). Consistently, there were increased expression of CD133, OCT4, SOX2 and Nanog genes upon USP44 knockdown but their decreased expression upon USP44 overexpression in CAL-27 and SCC-9 cells (Fig. 3F). Collectively, these results indicated an essential role of USP44 in the stemness of OSCC cells.

Fig. 3

USP44 inhibits stemness of OSCC cells. (A-B) Representative microscopy images of monolayer and spherical OSCC cells. The mRNA levels of stem cell markers CD133, OCT4, SOX2 and Nanog were detected by RT-qPCR. (C-D) The expression of USP44 in monolayer and spherical OSCC cells, respectively, using RT-qPCR and western blotting assays. (E) Representative microscopy images of spheroid formation in OSCC cells. (F) RT-qPCR quantification of CD133, OCT4, SOX2 and Nanog mRNA levels in the indicated OSCC cells. mc, monolayer cell; sc, spherical cell

USP44 inhibits growth and lung metastasis of OSCC cells in vivo

Next, the effects of USP44 on OSCC cancer cell growth were validated in xenograft mouse model using CAL-27 and SCC-9 cells. The results showed that downregulation of USP44 significantly promoted tumor growth, while its overexpression was associated with tumor size reduction (Fig. 4A and B). Immunohistochemical assay showed that USP44 expression was inversely correlated with the expressions of CD133 (a cancer stem cell marker) and Ki67 (a marker for cell proliferation) (Fig. 4C and D). These findings further supported the anti-tumor effect of USP44 in OSCC. Additionally, the effect of USP44 on lung metastasis was investigated using an experimental lung metastases model (Fig. 5A). H&E staining of lung tissues showed downregulation of USP44 significantly increased the number of metastatic pulmonary nodules (Fig. 5B). Conversely, the number of pulmonary nodules was decreased in USP44 overexpression group (Fig. 5C and D). Together, these results highlighted that USP44 is effective in mitigating lung metastasis in OSCC.

Fig. 4

USP44 suppresses tumorigenic abilities of OSCC cells in vivo. (A-B) Images of subcutaneous tumors from mice. The tumor growth curves (volume) over the course of the study were also shown. (C-D) Representative immunohistochemical staining images of tumor tissues for USP44, CD133, and Ki67

Fig. 5

USP44 reduces lung metastasis of OSCC cells in vivo. (A) Images of lung tissues from mice. (B) Representative H&E staining of lung tissues. (C-D) Quantification of pulmonary nodules in histograms

USP44 induces proteomic changes in CAL-27 cells

RIGOL LC-MS/MS system for label-free quantitative proteomics was carried out to dissect the effect of USP44 overexpression on proteome profile of CAL-27 cells. PCA analysis showed a clear separation between the control group (pLJM1-Ctrl, left-hand cluster) with negative PC1 scores and the pLJM1-USP44 group (right-hand cluster) with positive scores. The variance was accounted for by PC1 as 46.9% and PC2 as 14.4% (Fig. 6A). The clear separation suggested distinct proteomic profiles between the pLJM1-USP44 and the control groups. As shown in Fig. 6B, a total of 409 downregulated and 276 upregulated differentially expressed proteins (DEPs) were identified against the Proteome Discoverer database (Version 2.4) when applying a threshold of FC ≥ 1.5 and P ≤ 0.05. Further GO and KEGG pathway analyses were performed to explore the biological function of these DEPs (Fig. 6C and D). Additionally, Co-IP mass spectrometry was performed to identify potential proteins bound to USP44 (Fig. 6E). Integrative analysis identified 112 shared proteins that were upregulated in both the proteomics study and the Co-IP mass spectrometry analysis (Fig. 6E). These 112 USP44 overexpression-responsive proteins were annotated and categorized through GO-BP analysis. As shown in Figs. 6F and 13 proteins involved in complex BPs, including immune system, apoptotic signaling pathway, and cell cycle regulation, have attracted our interest. Through PPI analysis of these 13 proteins, we identified HEXIM1 as a pivotal protein for further exploration of its role in USP44-mediated biological functions (Fig. 6G).

Fig. 6

Label-free quantitative proteomics and CO-IP mass spectrometry analyses of USP44 effects in OSCC cells. (A) PCA plot illustrating expression data variance between pLJM1-Ctrl and pLJM1-USP44 overexpression in CAL-27 cells. (B) Volcano plot showing differentially expressed proteins in OSCC cells by the label-free quantitative proteomics and LC-MS/MS analysis. (C) GO (gene ontology) analysis showing the enrichment of biological processes among proteins differentially expressed in pLJM1-Ctrl- versus pLJM1-USP44-CAL-27 cells. (D) Circle plot showing KEGG pathway enrichment for differentially expressed proteins in pLJM1-Ctrl- versus pLJM1-USP44-CAL-27 cells. (E) Venn diagrams in the left panel comparing the proteins identified from proteomics and CO-IP mass spectrometry analyses, respectively. Overlaps showing the F 112 proteins common to both analyses. Heatmap in the right panel showing expression profiles of the 112 proteins. () GO-BP analysis of the 112 proteins. The 9 most significant biological processes are listed. GO terms are represented by filled circles where size is proportional to the significance. (G) The protein-protein interaction (PPI) network analyzed using String (https://cn.string-db.org/). The structure of protein HEXIM1 was downloaded from AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/). GO and KEGG pathway analyses were performed using the DAVID databease (Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/home.jsp)

USP44 enhances the stability of HEXIM1 by inhibiting its ubiquitination

We found that USP44 knockdown significantly reduced HEXIM1 protein levels in CAL-27 and SCC-9 cells, whereas USP44 overexpression led to an increase in HEXIM1 expression (Fig. 7A). Co-IP further verified both endogenous and exogenous interactions between USP44 and HEXIM1 (Fig. 7B and C). Besides, we found that overexpression of USP44 inhibited the degradation of endogenous HEXIM1 in the presence of cycloheximide (CHX), suggesting that USP44 could extend the half-life of HEXIM1 protein (Fig. 7D). Given USP44 is a deubiquitinating enzyme and usually stabilizes target proteins through the de-ubiquitination process (11), we hence examined the effects of USP44 on HEXIM1 ubiquitination. Consistently, overexpression of exogenous USP44 markedly decreased the ubiquitination level of HEXIM1 (Fig. 7E).

Fig. 7

USP44 enhances the stability of HEXIM protein. (A) Western blotting showing the expression of HEXIM in CAL-27 and SCC-9 cells. (B) Co-IP with anti-USP44 or anti-HEXIM1 antibody in CAL-27 and SCC-9 cells showing the endogenous interaction between USP44 and HEXIM1. (C) Co-IP with anti-Flag or anti-Myc antibody showing the interaction between Flag-USP44 and Myc-HEXIM1 overexpressed in 293T cells. (D) Effect of overexpression of USP44 on HEXIM1 stability. CAL-27 cells were transfected with either empty vector or USP44-expressing vector and then treated with CHX for the indicated periods of time. (E) Effects of USP44 overexpression on HEXIM1 ubiquitination. 293T cells transfected with Flag-USP44, Myc-HEXIM1 or HA-ubiquitin (Ub) plasmids were subjected to IP with anti-Myc antibody and then immunoblotted with the anti-HA or anti-Myc antibody

Silencing of HEXIM1 reverses the anticarcinogenic effect of USP44 in vitro

Given that USP44 inhibits the ubiquitination and degradation of HEXIM1, we investigated whether HEXIM1 could mediate the effects of USP44 on OSCC. CAL-27 cells were transfected with HEXIM1 siRNA and achieved significant silencing of HEXIM1 (Fig. 8A). Next, we found that silencing of HEXIM1 significantly enhanced the malignant phenotype of CAL-27 cells, manifested in an increase in cell viability (Fig. 8B) and migration and invasion abilities (Fig. 8C). At the same time, CCK-8 and Transwell assays also revealed that while overexpression of USP44 (pLJM1-USP44) inhibited the proliferation, migration, and invasion of CAL-27 cells, these effects were notably reversed upon HEXIM1 knockdown (Fig. 8D and E). These observations suggest that USP44 might exert its anti-cancer effect by regulating HEXIM1 protein.

Fig. 8

HEXIM1 mediates the antitumor effects of USP44 in OSCC cells. (A) Expression of HEXIM1 in CAL-27 cells transfected with the indicated shRNA plasmids and detected using Western blotting and RT-qPCR, respectively. (B) CCK-8 assays showing that cell viability was enhanced in HEXIM1-silenced CAL-27 cells. (C) Transwell assays analyzing the migration and invasion capabilities of CAL-27 cells with HEXIM1 silence. Quantification of migration and invasion cells was shown in the histograms. (D) Effect of HEXIM1 silencing on cell viability in CAL-27 cells overexpressing USP44. Cells were transfected with the indicated plasmids and the viability was determined by CCK-8 assay. (E) Effect of HEXIM1 silencing on cell migration and invasion capabilities in CAL-27 cells overexpressing USP44. Cells were transfected with the indicated plasmids. The migration and invasion capabilities were determined using Transwell assays. Quantification of migration and invasion cells was shown in the histograms

Discussion

The high metastasis and recurrence rates are associated with the high mortality rate of OSCC [11]. To develop effective therapeutic drugs for OSCC, efforts have been devoted to explore the key molecular characteristics and functional proteins involved in its recurrence and metastasis. In this study, the downregulation of the deubiquitinase USP44 in OSCC patients was verified to be positively related to poor prognosis. Overexpression of USP44 inhibited OSCC cell proliferation and metastasis both in vitro and in vivo. Furthermore, the label-free quantitative proteomics and Co-IP mass spectrum analysis revealed the potential proteins regulated by USP44. Importantly, the anticancer role of USP44 in OSCC via improving the protein stability of HEXIM1 was demonstrated.

Given the specific regulatory role of deubiquitinating enzymes in polyubiquitination cascade reactions, their dysfunction can result in different biological outcomes, such as aberrant cell proliferation, DNA damage [12] and consequent cancer occurrence and metastasis [13, 14]. Studies have revealed the tumor-suppressive role of USP44 in nasopharyngeal carcinoma [7], pancreatic ductal adenocarcinoma [15] and colorectal cancer [16]. However, the role of USP44 in cancer progression is controversial. For example, USP44 was demonstrated as an oncogenic factor in the progression of breast cancer, in that knocking down USP44 inhibited tumor angiogenesis and impaired tumor cell aggressiveness [17, 18]. Also, the promoting tumorigenesis role of USP44 has similarly been reported in prostate cancer [9]. These contradictory results indicate that USP44 is a multifunctional factor in cancer development, and its function may depend on the complex tumor microenvironment in different cancer subtypes. To our knowledge, the expression and biological function of USP44 in OSCC remain to be elucidated. In this research, we found that USP44 was significantly downregulated in OSCC samples compared to adjacent normal tissues, which was also confirmed by analysis of TCGA and GSE37991 datasets. Functional assays showed that increased USP44 expression notably suppressed OSCC cell proliferation, migration, invasion and cancer stem cell-like behaviors, whereas its decreased expression had the opposite effects. Consistent results were also observed in xenograft mouse model and experimental lung metastasis model. These findings suggested that USP44 is a cancer promoting factor in OSCC development.

Deubiquitinating enzymes are essential for regulating protein degradation and their ultimate biological function depends on the cellular function of their target proteins [19]. USP44 has been implicated in the turnover of ubiquitinated proteins and regulates various processes such as sister chromatid separation, stem cell differentiation, and tumor progression [20, 21]. Therefore, identifying downstream proteins targeted by USP44 can undoubtedly elucidate its anti-tumor function in OSCC. Our label-free quantitative proteomics and Co-IP mass spectrometry analyses identified 13 DEPs (SOD2, SCAF8, HEXIM1, SCAF8, IFI16, SUPT5H, BAX, DIABLO, SFN, RIOK3, HMOX1, BANF1, UBQLN1 and C1QBP) involved in tumorigenesis processes including immune system process, apoptotic signaling pathway and cell cycle regulation. For example, BAX, a major proapoptotic protein [22], is upregulated in USP44-overexpressing OSCC cells, implying that USP44 may inhibit OSCC progression by promoting apoptosis. However, these findings were obtained through theoretical bioinformatics analysis. The functions of these 13 potential targets in OSCC remain to be investigated in a follow-up studies. Given PPI analysis revealed that HEXIM1 to be the hub protein. This study mainly focused on the USP44-HEXIM1 axis in OSCC.

As a well-known inhibitor of positive transcription elongation factor b (P-TEFb) [23], HEXIM1 is involved in 60–70% of mRNA synthesis [24] and plays an important role in inflammation, acquired immunodeficiency syndrome and cancers [25,26,27]. Studies have reported that HEXIM1 functions as an estrogen receptor α (ERα)-binding protein in ERα-mediated primary breast cancer development [28] and suppresses metastatic breast cancer via indirectly action in multiple cell type, such as endothelial cells and bone marrow derived cells [29]. HEXIM1 was shown to assume a central role in the anticancer activity exerted by BET inhibitors [30] and a cancer suppressive factor in prostate cancer [31], melanomas [32], and AML [33]. In the present study, we found that USP44 interacts with HEXIM1, leading to enhanced HEXIM1 protein stability, suggesting a novel mechanism by which USP44 may inhibit OSCC progression in part by increasing HEXIM1 protein stability.

In conclusion, our findings support the role of USP44 in suppressing OSCC cell proliferation and metastasis by upregulating the stability of HEXIM1 protein, which pave the way for the development of targeted therapies in OSCC by targeting the USP44-HEXIM1 axis.