We collected HCC tissues and their paired adjacent tissues from 80 HCC patients who underwent radical HCC surgery at Ningbo Medical Center Lihuili Hospital from December 2021 to August 2022. Inclusion criteria were preoperative clinical diagnosis or pathological diagnosis of primary HCC and complete clinical data. Exclusion criteria were concurrent carcinomas in situ and absence of paracancerous tissues. All liver cancer tissues and their paired paracancerous tissues were collected during surgery. The specimens were partly snap-frozen in liquid nitrogen, stored at −80°C, and partly fixed in formalin and embedded in paraffin to create tissue wax blocks. The Ethics Committee of Ningbo Medical Center Lihuili Hospital approved the study (number: KY2021PJ231); patients provided informed written consent.

We downloaded the RNA sequencing (RNAseq) data in the Fragments Per Kilobase per Million (FPKM) format from the HCC Project from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/). The RNAseq data in the FPKM format was converted into the transcripts per million reads format and log2-transformed. The Entrez IDs were converted to gene symbols using the “org.Hs.eg.db” (v3.10.0) R package.

Total RNA from tissues was extracted using a TransZol UP Plus RNA Kit (TransGen Biotech, Beijing, China), and the extracted total RNA (OD 260 nm/OD 280 nm > 1.8) was reverse transcribed into cDNA using a TransScript All-in-One First-Strand cDNA Synthesis Supermix for QPCR kit (TransGen Biotech, Beijing, China). Real-time fluorescence quantification was performed on an ABI 7500 PCR instrument with PerfectStart^® Green qPCR SuperMix (TransGen Biotech, Beijing, China). The following primers were PER2 forward, 5′-GAC​ATG​AGA​CCA​ACG​AAA​ACT​GC-3’; PER2 reverse, 5′-AGG​CTA​AAG​GTA​TCT​GGA​CTC​TG-3’; GAPDH forward, 5′- GCA​CCG​TCA​AGG​CTG​AGA​AC-3′, and GAPDH reverse, 5′-TGG​TGA​AGA​CGC​CAG​TGG​A-3’. The 2^−ΔΔCt method was used to compare the mRNA levels of liver cancer samples and their paired adjacent tissues.

Total protein was extracted as protein lysates, then protein quantification and sample preparation were performed. Protein samples were separated using 7.5% Express Cast polyacrylamide gel electrophoresis (NCM Biotech, Suzhou, China) and transferred to polyvinylidene fluoride membranes (Millipore Company, United States). The membranes were blocked with quick blocking buffer (NCM Biotech, Suzhou, China) at room temperature for 10 min, then incubated overnight at 4°C with a primary antibody working solution (PER2 Mouse monoclonal antibody, 67513-1-Ig, 1:5000, Proteintech Company, United States; β-tubulin rabbit monoclonal antibody, A12289, 1:1000, ABclonal, China). The next day, the membranes were washed three times with Tris-buffered saline-Tween was incubated with a secondary antibody working solution (HRP-conjugated Affinipure Goat Anti-Mouse IgG (H + L), SA00001-1, 1:2000, Proteintech Company, United States; HRP-conjugated Affinipure Goat Anti-Rabbit IgG (H + L), SA00001-2, 1:2000, Proteintech Company, United States) for 2 h at room temperature, then the membranes were washed three times with Tris-buffered saline-Tween. The chemiluminescence method was used to expose the membranes in Image Quant LAS 500 instrument. Finally, the gray value of protein bands was analyzed using ImageJ software, and protein expression levels in liver cancer samples and their paired adjacent tissues were compared.

The paraffin sections of specimens were stained with PER2 protein by Ningbo Pathological Center. Immunohistochemical staining score: In 400 times visual field, randomly select five visual fields, each counting 200 cells, and score the positive rate and intensity of staining. The PER2 staining positive rate score was obtained according to the proportion of positively stained cells. The score of the staining range was 0 (0%), 1 (1%–25%), 2 (26%–50%), 3 (51%–75%), and 4 (76%–100%). PER2 staining intensity score: negative 0, light yellow 1, light brown 2 and tan 3. The product of the positive staining rate score and staining intensity score was used to judge the positivity grade. Samples were classified as 0 negative (−), 1–4 weak positive (+), 5–8 moderate positive (++), or 9–12 strong positive (++). The average of five visual fields was recorded.

According to the results of immunohistochemical staining of PER2 protein in liver cancer tissues, liver cancer tissues were divided into a low expression group where PER2 was negative (−) or weakly positive (+) and a high expression group where PER2 was moderately positive (++) or strongly positive (++). We recorded clinical data corresponding to liver cancer tissues, including gender, age, operation mode, smoking and drinking habits, body mass index (BMI), serum α-fetoprotein (AFP) level, whether they are infected with hepatitis B virus, tumor number, tumor maximum diameter, histological grade, microvascular invasion risk grading, Child-Pugh grading, China liver cancer staging (CNLC) to calculate correlations between PER2 expression and clinicopathological parameters and oncology behavior in patients with HCC.

HCC patients were divided into high PER2 expression and low PER2 expression groups. DEGs between the two groups were analyzed using the “DESeq2” (V1.26.0) R package (Love et al., 2014), and the threshold parameters of differential analysis were set to |log2(FC)| >1.5 and p. adj <0.05. We used the “ggplot 2"(V3.3.3) R package to visualize DEGs in volcano and heat maps.

The “ClusterProfiler” (v3.14.3) R package (Yu et al., 2012) was used for the functional annotation and Gene Set Enrichment Analysis (GSEA) of the DEGs. The curated reference genesets from the MsigDB file (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp): c2. cp.v7.2. symbols.gmt were selected for GSEA (Subramanian et al., 2005). The analysis results were significantly enriched with |NES| > 1, p. adj <0.05, and we used q-value <0.25 as thresholds.

The ssGSEA algorithm in the “GSVA” (v1.34.0) R package (Hänzelmann et al., 2013) evaluated the tumor infiltration of 24 immune cell types in HCC tissues. These 24 immune cell markers were plasmacytoid dendritic cells (pDC), cytotoxic cells, dendritic cells (DC), T cells, B cells, CD56 bright natural killer cells (NDC), CD56 bright natural killer cells (NK CD56bright cells), gamma delta T cells (Tgd), immature dendritic cells (iDC), macrophages, effector memory T cells (Tem), regulatory T cells (TReg), type 1 helper cells (Th1 cells), neutrophils, CD56 dim natural killer cells (NK CD56dim cells), mast cells, T follicular helper cells (TFH), activated dendritic cells (aDC), type 2 helper cells (Th2 cells), CD8 T cells, natural killer cells (NK cells), eosinophils, Type 17 helper cells (Th17 cells), Central memory T cells (Tcm), and helper T cells (T helper cells) (Bindea et al., 2013). The immune infiltration in HCC tissues was assessed using the StromalScore, ImmuneScore, and ESTIMATEScore algorithms in the “estimate” (v1.0.13) R package. The relationship between PER2 expression, immune cell infiltration status, and immune cell markers was determined using Spearman correlation analysis.

Spearman correlation analysis determined the relationship between PER2 expression level and core circadian rhythm genes, immune checkpoint genes, and TP53. The core circadian rhythm gene (Jiang et al., 2021) and immune checkpoint gene (Hu et al., 2021) were obtained from a literature review. The correlation was considered significant, with p < 0.05 as the threshold. We used the “ggplot 2"(V3.3.3) R package to visualize the heat maps and scatter plots analysis results.

The DNA methylation status in the CpG sites of the PER2 gene was analyzed in TCGA using the MetSurv database (https://biit.cs.ut.ee/methsurv/). The prognostic value of the CpG methylation status of PER2 was evaluated in the HCC samples.

To measure mRNA and protein expression levels of the circadian gene PER2 in HCC tissues and paired paracancerous tissues, we randomly selected liver cancer tissues and paired paracancerous tissues from 30 liver cancer patients from our liver cancer sample library and measured PER2 mRNA expression levels using qRT-PCR (Figures 1A, B), randomly selected liver cancer tissues and paired paracancerous tissues from 54 liver cancer patients. We measured PER2 protein expression levels using Western blotting (Figures 1C, D). The mRNA and protein expression levels of PER2 in HCC tissues were significantly lower than in the paired paracancerous tissues (p < 0.001).

We measured the expression of PER2 protein in 80 pairs of HCC tissues and paired paracancerous tissues using immunohistochemical staining. The Ningbo Pathology Center performed paraffin tissue sections of the specimens and immunohistochemical staining scoring. A negative (−) or weakly positive (+) staining result indicated low PER2 expression; a moderately positive (++) or strongly positive (++++) staining result indicated high PER2 expression. In liver cancer tissues, high expression of PER2 protein accounted for 62.5%, and low expression of PER2 protein accounted for 37.5%. In paracancerous tissues, high expression of PER2 protein accounted for 97.5%, and low expression of PER2 protein accounted for 2.5% (Figure 2). There was significantly less expression of PER2 protein in liver cancer tissues than in paracancerous tissues (p < 0.05) (Table 1).

To analyze the relationship between the circadian rhythm gene PER2 expression level in liver cancer tissues and the clinicopathologic and oncological characteristics, we collected clinical data from 80 patients with liver cancer corresponding to liver cancer tissues. We recorded sex, age, operation mode, smoking and drinking habits, BMI, serum AFP level, whether they were infected with hepatitis B virus, tumor number, maximum tumor diameter, histological grading, microvascular invasion risk grading, Child-Pugh grading, and CNLC staging. Samples were divided into high and low PER2 expression groups, with 50 people in the high-expression group and 30 in the low-expression group (Table 2). PER2 expression significantly correlated with nerve invasion (p = 0.017), Child-Pugh grading (p = 0.004), and CNLC staging (p = 0.004).

Based on the transcriptome data of HCC in TCGA, 424 HCC patients were divided into two groups according to the median expression value of the PER2 gene, and the differential expression between the two groups was analyzed. Taking |log2(FC)| >1.5 and p. adj <0.05 as threshold parameters, DEGs related to PER2 expression were obtained. Compared with the low expression group, there were 365 DEGs in the high PER2 expression group, of which 295 were upregulated, and 70 were downregulated (Figure 3A). Fourteen crucial DEGs were presented as a co-expression heat map (Figure 3B).

To study the biological function and signal pathways related to the PER2 gene in HCC, we used the transcriptome data from TCGA to group samples according to the median expression value of PER2 and obtained the DEGs related to PER2. We used the “Cluster Profiler” R package and the GSEA of PER2-related DEGs in HCC patients; the possible functions or pathways involved in PER2 gene expression were inferred. The predefined gene set used in GSEA came from the c2. cp.v7.2. symbols.gmt gene set in the MSigDB database. The PER2-related DEGs were enriched in mitochondrial oxidative phosphorylation, transcription and translation, amino acid metabolism, and other related pathways (|NES| > 1, p. adj <0.05, q value < 0.2) (Figure 3C).

Transcriptome data from HCC in TCGA were used to evaluate the infiltration status of 24 immune cell types using the ssGSEA algorithm in the “GSVA” R package. The correlation between PER2 expression and immune cell infiltration was calculated using Spearman correlation analysis. The expression of PER2 mRNA was correlated with many immune cells, including plasma cell-like pDC, cytotoxic cells, DC, T cells, B cells, NK CD56 bright cells, Tgd, iDC, eosinophils, Th17 cells, Tcm, and helper cells (Figure 4A; Table 3). We also used the “ESTIMATE” Score package to score the immune infiltration of HCC tissues. The results showed that the low expression of PER2 was significantly correlated with the immune infiltration score (p < 0.001) and with the immune and matrix comprehensive score (p < 0.01) (see Figure 4B). PER2 expression in HCC is related to immune cell infiltration.

A literature review revealed that there are a least 15 core circadian rhythm genes: PER1, PER2, PER3, CLOCK, CRY1, CRY2, ARNTL/BMAL1, TIMLESS/TIM, RORA, RORB, RORC, NPAS2, NR1D1, NR1D2, and CSNK1E/CKIε (Jiang et al., 2021). Using the transcriptome data of HCC in TCGA, the correlation between the PER2 gene and other core circadian rhythm genes was analyzed and expressed as a thermogram (Figure 5C). PER2 expression correlated with many core circadian rhythm genes (p < 0.05), including PER3, CLOCK, and NR1D2 (correlation >0.5) (Table 4).

Given the broad application of immune checkpoint inhibitors in anti-tumor treatment, we studied the relationship between PER2 expression and immunosuppression checkpoints. A literature review revealed 79 immune checkpoint genes (Hu et al., 2021). We analyzed the transcriptome data of HCC in TCGA to calculate the correlation between PER2 and immune checkpoint gene expression. PER2 positively correlated with 32 immune checkpoint genes and negatively correlated with one immune checkpoint gene (p < 0.05). These correlated immune checkpoint genes were presented as heat maps (Figure 5A).

TP53 is a tumor suppressor gene with low expression in normal cells and high expression in malignant tumors (Marei et al., 2021). Using correlation analysis of transcriptome data of HCC in TCGA, we found that PER2 expression positively correlated with TP53 expression (Figure 5B).

The DNA methylation level in the PER2 gene and the prognostic value of each CpG locus in the PER2 gene were analyzed using the MetSurv tool. The thermogram results revealed the methylation levels of 21 CpG sites in HCC (Figure 6A). The methylation levels of eight CpG sites (cg03004097, cg04169774, cg20070418, cg22879834, cg21315421, cg24831107, cg12308675 and cg06259818) were significantly correlated HCC outcomes (p < 0.05) (Table 5). Patients with hypermethylation of cg03004097 had better survival than patients with hypomethylation (p < 0.01, hazard ratio [HR] = 0.579), while patients with hypermethylation of cg06259818 had significantly worse survival outcomes than those with hypomethylation (p < 0.01, HR = 1.982) (Figures 6B, C). These findings suggest that the methylation status of the PER2 gene is related to HCC outcomes.