FTH1 indicates poor prognosis and promotes metastasis in head and neck squamous cell carcinoma

Data acquisition and tissue specimens

A workflow framework of this research is presented in Fig. 1. The RNA-Seq and clinical data of HNSCC dataset (528 cases) were downloaded from The Cancer Genome Atlas (TCGA) database. A total of 499 HNSCC samples, with 41 cases of matched adjacent normal tissues, without missing expression and missing follow-up were selected for the subsequent analysis. Besides, 21 sets of HNSCC tissue sections including cancer and adjacent normal tissues used for Immunohistochemistry (IHC) were collected from the First Affiliated Hospital of the University of South China. None of the patients received any preoperative anticancer treatment before surgical procedures. Their pathological diagnosis was confirmed by at least two pathologists. This study was approved by the Medical Ethics Committee of The First Affiliated Hospital of University of South China (Number: 2021ll0916001) and the requirement of consent was waived for the retrospective analysis.

Immunohistochemistry (IHC)

IHC was performed as our previous studies (Deng et al., 2020; Gong et al., 2018), using the PV-9000 IHC Reagent (ZSGB-BIO, Beijing, China). In brief, tissue sections were dewaxed with turpentine and then hydrated with a gradient of decreasing concentration of ethanol. The antigen retrieval was executed by boiling 10 mmol/l citric acid buffer (pH 6.0) for 15 min. After endogenous peroxidase was inactivated, the sections were blocked with normal goat serum (ZSGB-BIO, Beijing, China) for 10 min. Then the sections were incubated overnight with anti-FTH1 antibody (Affinity; 1:200) at 4 °C, followed by the horseradish peroxidase labeled secondary antibody. For the negative control, the normal rabbit IgG was applied. Finally, the positive signals were visualized by chromogenic agent DAB and nuclei were counter-stained with hematoxylin. Signal strength was scored as follows: 0 (negative), 1 (weak), 2 (moderate) and 3 (strong). The staining distribution score is based on the percentage of positive cells: 0 (0–5%), 1 (5–24%), 2 (25–49%), 3 (50–74%) and 4 (75–100%).

Cell culture and transfection

HNSCC cell lines (Fadu, SCC4, Cal27, HN8 and Cal33), 293T and immortalized non-malignant cell line DOK were obtained from Shanghai Cell Bank (The Chinese Academy of Science, Shanghai) or ATCC. DOK cells were cultured with RPMI 1640 (Gibco). HNSCC cell lines and 293T cells were cultured in DMEM (Gibco, Waltham, MA, USA). The culture mediums were supplemented with 10% FBS (Gibco) and all cells were cultured at 37 °C with 5% CO₂.

The shRNA-encoding lentiviral vectors for FTH1 knockdown were purchased from GeneCopoeia (GeneCopoeia, Rockville, MD, USA). The shRNA target sequences for human FTH1 were as follows: shFTH1-1: 5′-CCATGTCTTACTACTTTGACC-3′; shFTH1-2: 5′-CCATCAAAGAATTGGGTGACC-3′. The procedures of lentivirus packaging were performed using Lenti-Pac™ HIV lentivirus packaging kit (GeneCopoeia, Rockville, MD, USA). The negative control, nominated as Lv-control, was packaged with the empty vector. After 48–72 h of transfection, lentiviral particles were harvested to infect cell lines. Stable transfected cells were selected with puromycin (GeneCopoeia; 2 µg/ml) for 2 weeks and the inhibitory efficiency of FTH1 was confirmed by Western blotting.

Cell migration and invasion

Cell migration and invasion assay was conducted using wound healing and Transwell invasion assays as reported in our previous studies (Liu et al., 2020; Yang et al., 2020). For wound healing assay, stable transfected cells were seeded in a 6-well plate and scratched using a 10 µl tip when cells reached 90% confluency and then cells were cultured using serum-free medium for 48 h. Transwell invasion assays were performed using Transwell chambers (Corning, NY, USA) pre-coated with 15% Matrigel (Corning, USA). Cells were seeded into the upper chamber (2  × 10⁴ cells/well), while the lower chamber was placed with the medium containing 10% FBS. After 48 h, the cells on the lower surface were fixed utilizing paraformaldehyde and stained using crystal violet. After removing the cells on the upper surfaces, the stained cells were counted under a microscope.

Gene set enrichment analysis (GSEA)

Gene set enrichment analysis (GSEA) was performed using GSEA 4.0.1 software to explore the potential functions enriched in subgroups of high or low expression of FTH1. The gene set ‘c2.cp.kegg.v7.4. symbols.gmt [curated]’ from the MSigDB database was used as reference for GSEA. The enriched sets with P-value < 0.05 and FDR (false discovery rate) < 0.25 were considered statistically significant.

Western blotting

Whole-cell proteins were extracted using RIPA lysis buffer (NCM Biotech, China) with the proteasome inhibitor (Beyotime Biotechnology, Jiangsu, China) and centrifugated to collect the supernatant. After determination of the protein concentration, the collected supernatant was added with SDS buffer, incubated for 10 min at 100 °C, separated by SDS-PAGE and then transferred to the PVDF membrane (Millipore, Bedford, MA, USA). The membranes blocked with 5% skimmed milk was incubated overnight with primary antibody against FTH1 (1:1000 dilution; Affinity), E-cadherin (1:2000 dilution; Proteintech), N-cadherin (1:1000 dilution; CST), Vimentin (1:1000 dilution; CST) at 4 °C. Finally, the antigen-antibody complexes were visualized using enhanced chemiluminescence reagents (Thermo Fisher Scientific).

Statistical analysis

All statistical analyses were performed using R software (version 4.1.0; R Core Team, 2021). Student’s t test and Wilcox test were used to evaluate differences between two groups. Kruskal Wallis test and one-way analysis were used to analyze differences among multiple groups. Overall survival (OS) and disease-free survival (DFS) analysis were conducted using the Kaplan–Meier method with the log-rank test. The chi-square test and Fisher’s exact test were utilized to analyze the correlation between FTH1 expression and the clinicopathologic parameters. Univariate and multivariate Cox regression analyses were performed to identify independent prognostic factors. A value of P < 0.05 indicates statistical significance. Asterisks (*, **, ***) represented P < 0.05, P < 0.01 and P < 0.001, respectively.

Differentially expression and prognostic value of FTH1 and FTL1 in HNSCC

The differential analysis of TCGA-HNSC transcriptome data showed that the expression levels of FTH1 and FTL1 in HNSCC samples were significantly higher than those in adjacent normal samples (Figs. 2A, S1A), which were also confirmed using paired sample analysis (Figs. 2B, S1B). FTH1 in HNSCC cells was overexpressed compared with DOK, an immortalized non-malignant cell (Fig. 2C). The IHC staining showed that FTH1 protein was highly expressed in HNSCC tissues compared with matched adjacent epithelial tissues (Figs. 2D–2F). Based on FTH1 and FTL1, the coding genes for the heavy chain and light chain of ferritin significantly highly expressed in tumor samples from HNSCC patients, and we further analyzed their clinical prognostic value. In the Kaplan–Meier survival estimate, we found that only high expression of FTH1 in tumors indicates worse OS probability (P = 1.39e−03) and DFS probability (P = 4.054e−05), while FTL1 has no prognostic value (Figs. 3A and 3B, Figs. S1C and S1D). From the above results, it can be indicated that the expression of FTH1 has the prognostic value of OS and DFS simultaneously and the clinical outcomes of patients with high FTH1 expression are worse.

Correlation between FTH1 and clinical parameters

We generalized these patients’ baseline parameters with age, gender, smoking, drinking history, pathological grade, clinical stage, T/N/M stage, OS and DFS according to the high and low expression levels of FTH1 (Table 1). The preliminary correlation between FTH1 expression and clinicopathological features was obtained by logical regression analysis, as shown in Table 2. It shows that FTH1 expression is associated with smoking (no vs. yes, P = 0.008), pathological grade (G3 & $ vs. G1 & @, P < 0.01), N stage (N+ vs. N0, P < 0.01) and clinical stage (Stage III & IV vs. Stage I & II, P  < 0.05). After that, we further performed a Wilcoxon signed-rank test on clinicopathological features that distribute in high and low FTH1 expression groups. In addition to the differences in FTH1 expression among patients with a history of drinking (P < 0.001), the results of smoking (no vs. yes, P < 0.05), pathological grading (G3 & $ vs. G1 & 2, P < 0.01), N staging (N+ vs. N0, P < 0.01) and clinical stage (Stage III & IV vs. Stage I & II, P < 0.05) conform to logistic regression (Figs. 4A–4I).

We further performed a stratified Kaplan–Meier analysis for OS and DFS in TCGA-HNSC patients with the differential clinical features related to FTH1 expression. The related results showed that FTH1 expression had the prognostic value of OS (P = 0.004) and DFS (P < 0.001) in the G1 & 2 stage, but not in the G3 & 4 group (Figs. 5A and 5B, Figs. 6A and 6B). FTH1 expression has prognostic value of OS (both P < 0.05) and DFS (both P < 0.05) whether lymphatic metastasis or not (Figs. 5C and 5D, Figs. 6C and 6D); FTH1 expression has a predictive value of DFS (P = 0.002) but no predictive value of OS (P = 0.072) in the early stage (Stage I & II) population, while in advanced patients, have prognostic value regardless of OS (P = 0.013) or DFS (P = 0.004) (Figs. 5E and 5F, Figs. 6E and 6F). Finally, we were pleasantly surprised to find that FTH1 expression has a good prognostic value for both OS and DFS in smokers (P < 0.001), while not in non-smokers (Figs. 5G and 5H, Figs. 6G and 6H). From the results of the stratified analysis, we guess that FTH1 expression has better prognostic value for smokers, advanced and pathologically well-differentiated patients, and its prognostic value is not affected by the status of lymph node metastasis.

Independent predictive power of FTH1

The Cox regression analysis results of FTH1 expression level and clinicopathological features for OS prognosis have been shown in Table 3. Univariate cox analysis showed that metastasis (P = 0.026, HR: 3.721, 95% CI [1.177–11.764]) and FTH1 expression (P < 0.001, HR: 1.646, 95% CI [1.254–2.161]) were statistically significant. Moreover, multivariate cox analysis showed that both of them still had good statistical significance (metastasis, P = 0.031, HR: 3.547, 95% CI [1.122–11.214]) and (FTH1 expression level, P < 0.001, HR: 1.658, 95% CI [1.252–2.194]). The above analysis results indicate that in addition to the independent prognostic value of metastasis, FTH1 expression level can also be used as an excellent independent prognostic factor.

GSEA reveal FTH1-related pathways and molecular functions

Through GSEA, we explored the differences in the downstream activated signaling pathways between the low and high FTH1 expression groups to search for its potential carcinogenic mechanism. After screening of high expression in the group in MSigDB gene set (c2. Cp. Kegg. V7.1. Symbols. gmt), we found that 24 significant enrichment pathways (FDR < 0.25, NOM p-value < 0.05) are mainly concentrated in six aspects between the degree of enrichment of pathways: energy metabolism, glycometabolism, protein and amino metabolism, other metabolisms, cell adhesion and motility and tumor-associated signal pathways (Figs. 7A–7D).

FTH1 knockdown suppresses HNSCC metastasis by attenuating EMT

As previously mentioned, FTH1 was associated with lymph node metastasis. Hence, wound-healing and Transwell assays were performed to identify FTH1 affecting migration and invasion capabilities of HNSCC cells. Attractively, FTH1 knockdown suppressed wound healing rates in Fadu and HN8 cells (Figs. 8A–8D). Similar results were obtained in the Transwell invasion assays (Figs. 8E and 8F). Furthermore, we investigated molecular markers of EMT by Western blotting. After FTH1 depletion, mesenchymal markers (N-cadherin and vimentin) are suppressed, while the epithelial marker E-cadherin was upregulated (Figs. 9A and 9B).