UHRF1 predicts poor prognosis by triggering cell cycle in lung adenocarcinoma

Abstract Accumulating evidence suggests that ubiquitin‐like with plant homeodomain and ring finger domains 1 (UHRF1) is overexpressed in non‐small cell lung cancer (NSCLC); however, the expression and function of UHRF1 in the subtype of NSCLC are still unclear. Here, we investigate the expression and prognosis traits of UHRF1 in large NSCLC cohorts and explore the molecular characters during UHRF1 up‐regulation. We find that UHRF1 is predominantly overexpressed in lung squamous cell carcinoma (SCC). Surprisingly, the up‐regulated UHRF1 is only associated with the overall survival of lung adenocarcinoma (ADC) and knockdown of UHRF1 dramatically attenuates ADC tumorigenesis. Mechanically, we identify a hub gene that includes a total of 55 UHRF1‐related genes, which are tightly associated with cell cycle pathway and yield to the poor clinical outcome in ADC patients. What's more, we observe knockdown of UHRF1 only affects ADC cells cycle and induces cell apoptosis. These results suggest that up‐regulated UHRF1 only contributes to lung ADC survival by triggering cell cycle pathway, and it may be a prognostic biomarker for lung ADC patients.

opportunity for computational prediction of biomarkers, drugs and its potential mechanisms. [12][13][14] By investigating the previously published gene expression microarray data, we set out to identify the expression profile and prognosis value of UHRF1 in NSCLC and its subtype, respectively, and further to elucidate the mechanism of how UHRF1 affects the subtype of NSCLC patients' prognosis.

| Cell culture and transfection
The lung adenocarcinoma cell line (H1975) and lung squamous

| Western blotting
Western blot analysis was performed using standard techniques as described before with the primary anti-UHRF1 (Boster) and anti-GAPDH (Proteintech).

| Cell proliferation assay
The UHRF1-siRNA transfected cells were plated in 96-well plates at 3000/well and allowed to adhere overnight. CCK-8 solution (10 µL) was added to each well of the plate and incubated for another 2 hours in indicated days. The absorbance was measured using a Multilabel Plate Reader (Monobind Inc) at 450 nm.

| Wound healing assay
The UHRF1-siRNA transfected cells were digested and proceed cell counting, cells were seeded in 6-well plates and cultured with DMEM. A 10 μL white micropipette tip was used to create vertical wound in the cell monolayer. Images of the wound edges were captured at time 0 and 72 hours using a SONY ILCE-A6000L/B camera.

| Transwell assays
Cell migration was investigated with modified Boyden chamber (Costar). The UHRF1-siRNA transfected cells were digested and proceed to cell counting. A total of 5 × 10 4 cells in 100 μL were added to the upper wells, the lower compartments were added with 500 μL 10% FBS medium allowed cell to migrate for 12 hours at 37°C. Using cotton swabs to remove the cells that remained in the upper chamber . The membrane of the upper chamber was fixed with methanol for 20 minutes and stained with 0.1% crystal violet solution for   15 minutes. Then, washed the membrane with PBS for three times, and pictures of the migrated cells were taken for counting by using ImageJ (National Institutes of Health).

| Cell apoptosis and cell cycle analysis
The UHRF1-siRNA transfected cells were digested and proceed cell

| Meta-analysis
We searched GEO database by the following keywords: ('NSCLC' OR 'Non-small cell lung carcinoma' OR 'Non-small cell lung cancer'). The inclusion criteria are the both NSCLC tissues/healthy control or two main histological types (ADC and SCC) were included in each dataset which contained more than 100 human samples. We extracted all data into a standardized form which included following items: the name of first author, publication year, country, number of patients and the mRNA levels of UHRF1.

| Survival analysis
We divided samples into two groups (UHRF1 high expression and

| Bioinformatics analysis
The limma package (https://bioco nduct or.org/packa ges/relea se/ bioc/html/limma.html) was used to obtain the differentially ex- and these genes were served as NSCLC's survival-related genes in this research. The GO analysis was performed on the DEGs. We also carried out KEGG analysis to find the obvious altered pathways (P < .05) during UHRF1 increase. Based on the interaction relationship from KEGG, we built pathway and gene regulatory networks.
The PPI data were extracted from the STRING database (https://strin g-db.org) which was based on the protein interactions and signalling pathways, and the network was built by Cytoscape 3.6.1 application.

| Statistical analysis
Results were expressed as means ± SD and data were analysed using a two-sided unpaired Student's t test. For all analyses, * and *** indicated P < .05 and P < .001, respectively.

| UHRF1 is overexpressed in NSCLC tissues
We found UHRF1 was up-regulated in multiple cancers compared with normal controls from ONCOMINE data ( Figure 1A) and obtained the similar results from TIMER data derived from TCGA clinical patients ( Figure 1B). Of interest, we focused on the role of UHRF1 in lung cancer; however, the UHRF1 expression in lung cancer was inconsistent from different groups. Thereby, we did a meta-analysis  (Table S1). The mRNA levels of UHRF1 in NSCLC tumour tissues were significantly higher than that in healthy tissues, which was shown by meta-analysis, and the pooled mean difference was 1.62 (11 datasets, 1735 patients, 95% CI 1.13-2.12, Z = 6.46, P < .00001, Figure 1C).
UHRF1 mRNA levels in ADC and SCC were also compared by meta-analysis, and a statistical significance was observed (10 articles, pooled mean difference −0.55, 95% CI −0.76 to −0.35, Z = 5.4, P < .00001, Figure 1D). As shown in our analysis, the UHRF1 expression in SCC was significantly higher than that in ADC. These results indicate the UHRF1 expression is significantly up-regulated in NSCLC, especially in SCC.

| UHRF1 level is associated with NSCLC prognosis
To identify whether UHRF1 expression affects the patient's survival, and SCC with patients' survival in these four GEO sets ( Figure S1).
In turn, we found that high expression of UHRF1 was firmly as-

| Knockdown of UHRF1 attenuates ADC tumorigenesis
The conclusion that UHRF1 level only affects ADC prognosis was truly inconsistent with our expectation. As far as we know, poor prognosis is mainly caused by tumour metastasis and therapy resistance. Tumour metastasis yields approximately 90% cancer-related death 15 and includes multiple biologic steps, such as those of cell proliferation, migration, invasion, circulation, extravasation and colonization. 16 Of interest, we investigated the biofunction of UHRF1 based on the knockdown of UHRF1 by using siRNA in ADC (H1975) and SCC (SK-MES-1) cells ( Figure 3A). Indeed, knockdown of UHRF1 dramatically impeded H1975 cells growth ( Figure 3B), but there was no any significant growth differences in SK-MES-1 cells ( Figure 3C), which was consistent with the wound healing results ( Figure 3D).
What is more, both UHRF1 siRNA notably inhibited H1975 cells migration, especially the siRNA UHRF1-2 inhibited around 50% of cell migration, but only the siRNA UHRF1-2 decreased around 20% of cell migration in SK-MES-1 ( Figure 3E). These results suggest knockdown of UHRF1 can dramatically impede ADC tumorigenesis when compared with SCC.  Figure 4B). The expression of cell cycle pathway in non-survivors was significantly higher than that in survivors, and the result was detected between high and low UHRF1 expression patients ( Figure 4C).

| UHRF1 triggers cell cycle to yield poor prognosis in ADC
After the discovery of the significantly varied pathways, protein and protein interaction (PPI) and correlation analysis were used to identify the interaction between these 55 proteins. We showed that one strong network among 55 proteins which had a stronger enriched network than that in random proteins, and these genes were especially associated with cell cycle pathway ( Figure 5A). The correlation matrix owing correlation coefficients between sets of genes were derived from ADC discovery cohort, and it revealed that most proteins from this network had strong positive correlations with each other ( Figure 5B). So, these genes established by UHRF1-associated genes, especially cell cycle-related genes have a strong interwork interaction and it could be used as a multi-gene biomarker for predicting the survival of NSCLC patients and hub genes for differentiating the survival status difference of high UHRF1 in ADC and SCC.

| The prediction score of hub genes predicts ADC clinical outcome
We The risk score of each patient was obtained from sum of multiplication of normalized expression value of the gene by its corresponding weight value). A higher risk score represented a worse clinical outcome. Our results focused on both the ADC discovery/validation cohort and SCC TCGA datasets. Interestingly, the scores from nonsurvivors were significantly higher than those of survivors both in the ADC discovery and validation cohort ( Figure 6A), there were not any significant differences in SCC cohort ( Figure S2). Same results were obtained in ADC (discovery P = .0011, validation P = .0095) and SCC cohort according to the pathological stage ( Figure 6B). The Kaplan-Meier plot confirmed the survival results in both discovery and validation cohorts ( Figure 6C,D). Notably, we observed the G1 phase was increased while G2 phase was decreased in H1975 after knockdown of UHRF1, but there was not any significant differences in SCC SK-MES-1 cells ( Figure 6E). In the meantime, we observed significant apoptosis of ADC H1975 cells after treatment with docetaxel in UHRF1-down-regulation group; however, there were no effects in SK-MES-1 cells ( Figure 6F). These results indicate that UHRF1 is likely to trigger cell cycle. Based on all these results, the 55 genes detected from UHRF1 expression had the statistical power to predict clinical outcome in ADC and thus elucidated why the high UHRF1 expression in ADC and SCC had diverse clinical outcome.  lines. In view of these results, we postulate that UHRF1 is essential for proliferation in human cancer cell.

| D ISCUSS I ON
In conclusion, our study indicates that UHRF1 is overexpressed in NSCLC, especially in SCC, but the up-regulated UHRF1 only contributes to ADC patients' survival by activating cell cycle pathway.
These findings could benefit the understanding of the effect of UHRF1 on NSCLC and reveal the potential targets for NSCLC subclasses individual prognosis and treatment.

ACK N OWLED G EM ENTS
This study was supported by National Nature Science Foundation of China (grant number 81802997, 81770180).

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.