Deciphering the effects of PYCR family on cell function, prognostic value, immune infiltration in ccRCC and pan-cancer

doi:10.21203/rs.3.rs-4231917/v1

Download PDF

Research Article

Deciphering the effects of PYCR family on cell function, prognostic value, immune infiltration in ccRCC and pan-cancer

https://doi.org/10.21203/rs.3.rs-4231917/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Pyrroline-5-carboxylate reductase (PYCR) is pivotal in converting pyrroline-5-carboxylate (P5C) to proline, the final step in proline synthesis. Three isoforms, PYCR1, PYCR2, and PYCR3, existed and played significant regulatory roles in tumor initiation and progression.

Methods

In this study, we firstly assessed molecular and immune characteristics of PYCRs by a pan-cancer analysis, especially focusing on their prognostic relevance. Then, a kidney renal clear cell carcinoma (KIRC)-specific prognostic model was established, incorporating pathomics features to enhance predictive capabilities. The biological functions and regulatory mechanisms of PYCR1 and PYCR2 were investigated by in vitro experiments in renal cancer cells.

Results

The PYCRs’ expressions were elevated in diverse tumors, correlating with unfavorable clinical outcomes. PYCRs were enriched in cancer signaling pathways, significantly correlating with immune cell infiltration, tumor mutation burden (TMB), and microsatellite instability (MSI). In KIRC, a prognostic model based on PYCR1 and PYCR2 was independently validated statistically. Leveraging features from H&E-stained images, a pathomics feature model reliably predicted patient prognosis. In vitro experiments demonstrated that PYCR1 and PYCR2 enhanced the proliferation and migration of renal carcinoma cells by activating the mTOR pathway, at least in part.

Conclusion

This study underscores PYCRs' pivotal role in various tumors, positioning them as potential prognostic biomarkers and therapeutic targets, particularly in malignancies like KIRC. The findings emphasize the need for broader exploration of PYCRs' implications in pan-cancer contexts.

pyrroline-5-carboxylate reductase (PYCR)

pan-cancer analysis

prognostic model

kidney renal clear cell carcinoma (KIRC)

Pathomics

mTOR

Cancer stands as a predominant cause of mortality across the globe[1], and the intricate nature of the tumor microenvironment presents formidable challenges to cancer treatment. Despite ongoing advancements in therapeutic modalities, the worldwide incidence and mortality rates of cancer continue to escalate rapidly[2]. Given the intricate processes underlying cancer initiation, a comprehensive exploration of the biological characteristics of genes assumes paramount importance for clinical treatment and prognosis prediction. Through a systematic analysis of cancer gene expression and variations, leveraging the Cancer Genome Atlas (TCGA) database[3], we can delve into the exploration of new prognostic biomarkers and therapeutic targets. This approach aims to unravel the intricate relationship between specific genes and immunotherapy, potentially paving the way for innovative strategies in cancer treatment.

In recent years, the interconversion between proline, glutamate, and ornithine has emerged as a focal point in the regulation of tumor cell growth in the field of cancer metabolism. ∆1-pyrroline-5-carboxylate (P5C) serves as the direct precursor of proline and is also a product of proline degradation. The conversion of P5C to proline is catalyzed by PYCR, the final enzyme in proline biosynthesis [4] [5]. Three isoforms of PYCR have been identified—PYCR1, PYCR2, and PYCR3/PYCRL. Both PYCR1 and PYCR2 are located in the mitochondria, sharing 84% structural homology. They function similarly in the final step of the glutamate-P5C-proline pathway, exhibiting a preference for NADH as a cofactor [6]. PYCR3, on the other hand, lacks 40 amino acids at the C-terminus and is mainly situated in the cytoplasm. Additionally, PYCR3 prefers utilizing NADPH as a cofactor to catalyze the production of proline from ornithine [7]. Recent studies indicate that PYCRs are upregulated in various tumors and are associated with the development of certain cancers. For example, downregulating PYCR1 can inhibit the growth of LUAD cells by impacting the JAK/STAT signaling pathway [8]. In vitro and in vivo studies have demonstrated the ability to suppress the proliferation, invasion, epithelial-mesenchymal transition (EMT), and metastasis of HCC cells [9]. PYCRs can also modulate the STAT3-mediated p38 MAPK and NF-κB signaling pathways in CRC cells to inhibit proliferation, drug resistance, and the EMT signaling pathway [10]. Knocking out PYCR2 inhibits the proliferation, migration, and invasion of colon cancer cells[10, 11] and also influences the development of melanoma, bladder cancer, nasopharyngeal carcinoma, gastric cancer, and other tumors[12–15]. However, current research on PYCRs in a pan-cancer context is lacking, and there is a dearth of studies utilizing PYCRs to predict tumor prognosis.

Significant progress has been achieved in the application of artificial intelligence algorithms in the medical field, particularly in the diagnosis and treatment of cancer, with notable success in recent years [16, 17] Machine learning and deep learning have found extensive use in the medical image processing of various cancers, including breast cancer, lung adenocarcinoma, and neurogenic tumors [18–20]. By processing medical images at high throughput, artificial intelligence can extract a wealth of information beneficial for precision medicine[19]. Pathomics, employing artificial intelligence methods, extracts high-throughput features from pathology images stained with Hematoxylin and Eosin (H&E). This approach combines pathological morphological information with genetic data, uncovering information that is imperceptible to the naked eye and revealing patterns difficult to summarize through subjective experience[21]. By integrating microscopic-level image information at the cellular level, there is promising potential to significantly enhance the predictive efficiency of disease diagnosis.

In this study, we first analyzed the expression patterns and prognostic value of PYCRs in a pan-cancer context based on the TCGA database. Subsequently, we examined the correlation between PYCRs expression and tumor mutation burden (TMB), microsatellite instability (MSI), tumor immune infiltration, and potential biological pathways. Risk score models were constructed using PYCR1 and PYCR2 specifically for KIRC. We investigated the relationship between the risk score models and the clinical prognosis of overall survival in KIRC, followed by an analysis of the molecular and immune features of the risk score models. Furthermore, we utilized features extracted from H&E-stained images of KIRC to build a pathomics feature model predicting the prognosis risk score model we designed. Univariate and multivariate Cox regression analyses were employed to confirm the stability of the pathomics feature model in predicting the prognosis of KIRC patients. Finally, we explored the impact of PYCR1 and PYCR2 on cell growth and migration in renal cancer cells, revealing that PYCR1 and PYCR2 may exert their effects in renal cancer by activating the mTOR signaling pathway.

1. TCGA Pan-Cancer Data and Acquisition/Processing of H&E Pathology Images

RNA-Seq (RNASeqV2RSEM), clinical data, and mutation data for 33 cancers were obtained from the University of California, Santa Cruz (UCSC) Xena Browser (https://xena.ucsc.edu/). The R software (version 4.1.0; https://www.R-project.org) was employed to extract and integrate mRNA expression data of PYCRs and clinical data across the 33 cancers, followed by subsequent data analysis. Clinical data, corresponding sequencing data, and H&E pathology images of KIRC patients were downloaded from the TCGA database (https://portal.gdc.cancer.gov/).

2. Prognostic Analysis

We obtained prognostic information for TCGA tumor samples, including Overall Survival (OS) and Progression-Free Survival (PFS). The R package "survival" was employed for Cox analysis to calculate the correlation between gene expression levels and patient survival rates. The "forestplot" R package was used to create a forest plot. Kaplan–Meier (KM) method and log-rank tests were performed for survival analysis (p < 0.05), with the "survminer" and "survival" R packages used for plotting survival curves.

3. Immune Cell Infiltration Analysis and Gene Set Enrichment Analysis (GSEA)

The CIBERSORT algorithm was utilized to estimate the proportions of 22 immune cell infiltrations in samples from 33 cancer patients [22]. Visualization was conducted using the "ggplot2" R package. Gene Set Enrichment Analysis (GSEA) compared gene and functionally annotated gene sets based on the gene expression matrix, analyzing the enrichment differences between two groups of gene sets [23]. To explore the molecular mechanisms underlying prognostic differences between subgroups of the risk score model, we performed GSEA enrichment analysis on hallmark gene sets (h.all.v7.5.1.symbols.gmt) and Gene Ontology (GO) gene sets (c5.go.v7.4.symbols.gmt) using the "limma" and "clusterProfiler" R packages. A p-value < 0.05 was considered statistically significant.

4. Tumor Mutation Burden (TMB) and Microsatellite Instability (MSI) Analysis

Tumor Mutation Burden (TMB), defined as the number of somatic mutations per megabase of the genomic sequence in the germline of a tumor, varies across malignancies[24]. We calculated the mutation count for 33 cancers from the somatic mutation dataset (MAF data), assessed TMB based on a Perl script, and normalized it by dividing by the exon length. Microsatellite Instability (MSI) involves the spontaneous loss or gain of nucleotides in the repetitive DNA tract[25]. We obtained MSI scores for 33 cancers from the TCGA database. Spearman correlation analysis was conducted using the Spearman method to assess the correlation between PYCRs expression and TMB or MSI. Visualization of the results was performed using the "fmsb" R package, generating a radar plot.

5. Construction of the KIRC Prognostic Risk Score Model

Using the "caret" R package, we randomly divided 507 TCGA-KIRC patients into training (n = 254) and validation (n = 253) cohorts (Table S1). Leveraging the expression data of PYCR1 and PYCR2 in KIRC samples and overall survival (OS) data, a LASSO Cox regression analysis was conducted using the "glmnet" R package to construct a multi-gene risk score model. This model predicts the prognosis of KIRC patients, with the risk score established based on gene expression data and Cox regression coefficients, calculated by the formula: Risk Score = (0.050641 * Expression of PYCR1) + (0.051626 * Expression of PYCR2). The low-risk and high-risk groups were determined by the median of the risk scores. Cox regression coefficients and gene expression levels were utilized to compute the risk scores for each sample in the training and validation cohorts. Based on the median risk score, KIRC samples were categorized into high-risk (training cohort n = 141, validation cohort n = 112) and low-risk groups (training cohort n = 113, validation cohort n = 141). The Kaplan-Meier method was employed to assess the overall survival differences between high- and low-risk groups. Additionally, the "timeROC" R package was used to draw time-dependent ROC curves and calculate the area under the curve (AUC) to validate the accuracy and predictive ability of the model. To validate the independent prognostic value of the risk score model and explore potential clinicopathological features related to prognosis, univariate and multivariate Cox regression analyses were conducted.

6. Pathological Image Segmentation and Feature Extraction

Pathological Image Acquisition: We downloaded pathological images from The Cancer Genome Atlas (TCGA) database (https://tcga-data.nci.nih.gov/tcga/). These images are tissue slices embedded in formalin and paraffin, with an SVS format and a maximum magnification of 20× or 40× (H&E stained pathological tissue images at 20× or 40× magnification). Pathologists reviewed the images, excluding sub-images with poor quality (contamination, blur, or blank areas exceeding 50%). The OTSU algorithm (https://opencv.org/) was employed to obtain the tissue regions of pathological slices. The 40× images were segmented into multiple 1024×1024 pixel sub-images, while the 20× images were segmented into multiple 512×512 pixel sub-images, which were then upsampled to 1024×1024 pixels. Ten random sub-images were selected from each pathological image for subsequent analysis. The PyRadiomics open-source package (https://pyradiomics.readthedocs.io/en/latest/) was utilized to standardize and extract features from the sub-images. After removing unnecessary image features, the mRMR and RFE algorithms were employed to select the optimal features, resulting in the final choice of 5 quantitative image features, which were Z-score standardized.

7. Pathomics Feature Model Construction

The GMB algorithm was employed to construct the pathomics feature model. The probability predicted by the pathomics feature model for the prognosis risk score model was designated as the pathomics_score. To evaluate the pathomics feature model, Receiver Operating Characteristic (ROC) curves and Precision-Recall (PR) curves were plotted. Diagnostic performance was assessed using Area Under the Curve (AUC) and other diagnostic indicators, including Brier score, recall, accuracy, sensitivity, specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV). Calibration curves were used to demonstrate the calibration of the prediction model, and the goodness-of-fit was assessed using the Hosmer-Lemeshow test. Decision Curve Analysis (DCA) was employed to assess net benefit at different probability thresholds, evaluating clinical utility. Internal 5-fold cross-validation was conducted to validate the pathomics feature model. The sample cohort for building the prognosis model, after intersecting with KIRC samples with complete pathological image information, resulted in a total of 407 samples for pathological analysis. Using the 'caret' R package, these TCGA-KIRC 407 patients were randomly divided into training (n = 204) and validation (n = 203) cohorts (Table S2), and the pathomics features of the training and validation cohorts were Z-score standardized.

8. Cell Culture and Transfection

Human renal cell carcinoma cell lines, namely Caki-1, 786-O, and A498, were purchased from the National Collection of Authenticated Cell Cultures of Chinese Academy of Sciences and maintained in McCoy’s 5A, RPMI-1640 and EMEM (HyClone, Beijing, China) supplemented with 10% fetal bovine serum (Ausbian, Australia) at 37°C with 5% CO₂, respectively. Small interfering RNA (siRNA) targeting PYCR1 and PYCR2, along with the negative control (NC), were purchased from GenePharma (Shanghai, China). Transient transfection of PYCR1 or PYCR2 siRNA into Caki-1 cells was accomplished using Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA, USA) in OptiMEM® I Reduced Serum Medium (OPTI-MEM) (Gibco, Thermo Fisher Scientific, USA). The siRNA sequences are detailed in Table S3. PYCR1 and PYCR2 gene expression plasmids were constructed on the pBOB lentiviral vector and co-transfected with three viral packaging plasmids (pMDLg, pVSV-G, and pRSV-Rev) into 293T cells to generate complete infectious viral particles. In this study, lentivirus infection was employed in the establishment of gene overexpression models, and puromycin was utilized for the selection of stably expressing cells.

9. Protein Blotting

Proteins were extracted from cells using RIPA lysis buffer, which contained 1% phenyl methyl sulfonyl fluoride (PMSF, Sigma Aldrich, USA), 1% sodium orthovanadate (Sigma Aldrich, USA), and 1% aprotinin (Sigma Aldrich, USA). The protein concentrations were measured by BCA method. Protein samples (20 µg per well) were loaded onto 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Subsequently, proteins were transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore, Burlington, MA, USA). After blocking the membrane with 5% BSA at room temperature for 1 hour, it was incubated overnight at 4°C with primary antibodies. The membrane was then washed three times and incubated with secondary antibodies at room temperature for 1 hour. Finally, protein bands were visualized using the ImageQuant LAS 4000mini gel imaging system (General Electric Company, USA). The primary antibodies used in the immunoblotting studies were as follows: β-actin, GAPDH, α-tubulin, PYCR1, PYCR2, Vimentin, Bax ,Bcl2 (#Cat:20536-1-AP, 10494-1-AP, 11224-1-AP, 13108-1-AP, 55060-1-AP, 10366-1-AP, 50599-2-Ig, 50599-2-Ig, 68103-1-Ig, Proteintech Group, China), N-Cadherin, E-Cadherin (#PTM-5221, PTM-6222, PTM biolabs, China), CyclinD1, PCNA ,Stat3, PARP, Bcl-xl, mTOR, Phospho-mTOR (Ser2448), 4E-BP1, Phospho-4E-BP1 (Thr37/46), p70 S6 Kinase, Phospho-p70 S6 Kinase (Thr389) (#55506, #13110, #9139, #9532, #2764, #2972, #2971, #9644, #2855, #9202, #9205, Cell Signaling Technology, USA)

10. Cell Proliferation

Cell viability was assessed using the CCK-8 assay. Briefly, 2000 cells were seeded in a 96-well plate and cultured according to the recommended instructions. Then, 10 µl of CCK-8 reagent (Zomanbio, Beijing, China) was added to each well. After incubating for an additional 2 hours, the absorbance was measured at 450 nm using a microplate reader (Thermo Fisher Scientific). EdU incorporation assay was performed to detect cell proliferation. Caki-1 cells (1×10^5) were seeded in a 24-well plate, transfected with PYCRs siRNA or control siRNA for 48 hours, and then treated with 10 µM EdU for an additional 1.5 hours. Cells were fixed with 4% paraformaldehyde and stained with Azide 488 and Hoechst 33342 (Beyotime, Shanghai, China), followed by imaging using a fluorescence microscope.

11. Cell Migration

Wound Healing Assay: Cells were seeded in a 6-well plate and allowed to reach approximately 90% confluence. Using a sterile 200 µl pipette tip, a wound was created on the cell monolayer. The cells were then cultured in basal medium, and photographs were taken at 0 and 72 hours under a microscope to record the wound width. The cell migration rate was calculated using the formula: (Width at 0 hours - Width at 72 hours) / Width at 0 hours. Transwell Assay: Cells (5×10^4) were seeded in the upper chamber of Transwell inserts (8-µm pore size; Corning, NY, USA). DMEM containing 20% FBS was added to the lower chamber. After 24 hours of incubation at 37°C and 5% CO2, the migrated cells in the lower chamber were fixed with 4% paraformaldehyde and stained with crystal violet. Photographs of cells in five random fields under a microscope (20×) were taken, and the relative area ratio was calculated.

12. Drug treatment

The cytotoxicity of halofuginone (HF) (Sigma Aldrich, USA) on Caki-1 cells was determined by CCK8 method. Briefly, a total of 5000 cells were seeded in 96-well plates. Caki-1 cells were exposed to halofuginone of a variety of concentrations (0, 50, 100, 200, 400 nM) for 12, 24, 36, and 48 hours. Then, the cell viabilities were measured by CCK8 method. In colony formation assay, cells were pre-treated with different concentrations of HF (0, 50, 100, 200 nM) for 24 hours. Then, cells were seeded into 12-well plate with 5000 cells per well. Cells were cultured for 7 days and then fixed with 4% paraformaldehyde and stained with 1% crystal violet. Colony-forming ability was observed through both naked-eye examination and microscopy. In a separate experiment, Caki-1 cells were exposed to diverse HF concentrations (0, 50, 100, 200, 400 nM) for 24 hours, and cellular proteins were extracted for subsequent Western blot analysis. Caki-1 cells were exposed to diverse L-proline (Sigma Aldrich, USA) concentrations (0,1, 2, 4, 8, 16 mM) for 24 hours, and cellular proteins were extracted for subsequent Western blot analysis.

13. Statistical Analysis

Statistical analysis was performed using R language (version 4.1.0; https://www.R-project.org) and GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA). Wilcoxon signed-rank test was employed to compare the expression of PYCRs mRNA between tumor tissues and normal tissues. One-way analysis of variance (ANOVA) was used for comparisons among multiple groups. Spearman correlation analysis was applied to explore the correlation between continuous variables. Single-factor Cox proportional hazards regression model or log-rank test was used to assess the correlation between gene expression and overall survival rate of patients. A significance level of P < 0.05 was considered statistically significant.

1. Analysis of PYCRs in Pan-Cancer

We analyzed the mRNA expression levels of PYCRs in various cancer types and corresponding normal tissues using TCGA data. PYCR1 (Fig. 1A), PYCR2 (Fig.S1A), and PYCR3 (Fig.S2A) were upregulated in 18, 15, and 15 types of tumors, respectively, while PYCR3 showed decreased expression in KICH, KIRC, and THCA. Single-factor Cox regression analysis revealed that PYCR1 was a high-risk gene for overall survival (OS) and progression-free survival (PFS) in 8 cancer types (Fig. 1B). Similarly, PYCR2 and PYCR3 were identified as high-risk genes in multiple cancer types for OS and PFS (Fig.S1B, Fig.S2B).

GSEA enrichment analysis of PYCRs showed that in GBM, PRAD, and SKCM, PYCR1 negatively regulates TNF-α signaling and IL6-JAK-STAT3 signaling, while in ACC, KIRC, KIRP, and THCA, PYCR1 positively regulates G2M checkpoint, EMT, and E2F targets (Fig. 1C). PYCR2 exhibited similar enrichment patterns in pan-cancer (Fig.S1C). For PYCR3, it negatively regulates TNF-α signaling and IL6-JAK-STAT3 signaling in LGG, PAAD, PRAD, and STAD, while positively regulating G2M checkpoint and E2F targets in ACC, CHOL, ESCA, PAAD, and STAD (Fig.S2C).

We analyzed the relationship between the expression of PYCRs and tumor mutational burden (TMB) as well as microsatellite instability (MSI). The expression of PYCR1 was positively correlated with TMB in 18 cancer types and positively correlated with MSI in 9 cancer types, while it was negatively correlated with MSI in READ (Fig. 1D-E). The expression of PYCR2 was positively correlated with TMB in 6 cancer types and positively correlated with MSI in 8 cancer types, but negatively correlated with TMB in 4 tumor types and MSI in 3 tumor types (Fig.S1D-E). PYCR3 expression was positively correlated with TMB in 11 cancer types and positively correlated with MSI in 8 cancer types, while negatively correlated with TMB in COAD and MSI in both COAD and READ (Fig.S2D-E).

The results of immune cell infiltration analysis showed that in 8 types of tumors, including STAD and KIRC, the expression of PYCR1 was positively correlated with the infiltration of T cells CD4 memory activated. In 6 types of tumors, including LUSC and LUAD, the expression of PYCR1 was positively correlated with the infiltration of T cells CD8 (Fig. 1F). In 6 types of tumors, including STAD and SARC, the expression of PYCR2 was positively correlated with the infiltration of T cells CD8 (Fig.S1F). In 11 types of tumors, including LUAD and KIRC, the expression of PYCR3 was positively correlated with the infiltration of T cells CD8 (Fig.S2F).

2. Construction of OS-Related Prognostic Risk Score Model for PYCRs in KIRC

In the pan-cancer prognosis analysis, we observed a significant correlation between the entire PYCR family genes and the prognosis of KIRC (Fig.S3A). Utilizing the transcriptomic expression data of PYCRs and OS data in KIRC, we conducted LASSO regression analysis and identified PYCR1 and PYCR2 as two genes for constructing the prognostic risk model (Fig.S3B, 3C). Subsequently, a multifactor Cox regression analysis was performed to further confirm that PYCR1 and PYCR2 could be used to construct the risk score model (Fig.S3D). Using the median value of the risk score as the threshold, we divided the training and testing cohorts into high and low-risk groups. In the training cohort, the high-risk group showed significantly reduced OS (Fig. 2A), and the time-related ROC curve results indicated that the AUC of the risk score model for 1-year, 3-year, and 5-year OS were 0.725, 0.692, and 0.711, respectively (Fig. 2B). Additionally, we ranked the risk scores of each patient in the training cohort, with higher scores indicating shorter survival times and fewer surviving patients. The heatmap described the expression patterns of PYCR1 and PYCR2 in the high-risk and low-risk groups (Fig. 2C). The predictive ability of this risk score model was validated in the testing cohort, yielding results similar to the training cohort (AUC values for 1-year, 3-year, and 5-year were 0.713, 0.722, and 0.702, respectively) (Fig. 2D-F). This suggests that the risk score model we constructed exhibits high accuracy in predicting OS for KIRC patients. The results of univariate and multivariate Cox regression analyses assessing the impact of clinical-pathological factors and the risk score from the prognostic model on the OS of KIRC patients indicated that the risk score from the model is an independent prognostic factor for KIRC patients (Fig. 2G).

3. Molecular and Immune Features of Risk Score Model Subgroups

The results of GSEA enrichment analysis demonstrated significant enrichment of immune-related pathways in the GO gene set in the high-risk score group of KIRC, while the low-risk score group exhibited significant enrichment in spliceosome assembly-related pathways and pathways related to compound and metal ion reactions (Fig. 3A). In the Hallmark gene set, the high-risk groups of KIRC and KIRP showed significant enrichment in pathways such as E2F Targets, G2/M Checkpoint, and EMT, while the low-risk score group of KIRC exhibited significant enrichment in pathways related to bile, fat metabolism, and others (Fig. 3B). We conducted an analysis of gene mutations in the subgroups of the risk score model. In KIRC, the mutation rates in the high-risk group were slightly higher than those in the low-risk group. The most frequent mutation type was missense mutation, followed by frameshift deletion mutation. VHL, PBRM1, and TTN had mutation rates exceeding 10% in both high and low-risk groups of KIRC, and TTN had a mutation rate exceeding 10% in both high and low-risk groups of KIRP (Fig. 3C-D). We also explored the correlation between the risk score and tumor mutation burden (TMB), and the results showed a positive correlation between TMB and risk score in KIRC (Fig. 3E). The infiltration of immune cells was compared between subgroups of the risk score model. In KIRC, Plasma cells, T cells CD8, T cells CD4 memory activated, T cells follicular helper, T cells regulatory (Tregs), NK cells activated, and Macrophages M0 exhibited higher infiltration abundance in the high-risk score group. T cells CD4 memory resting, NK cells resting, Monocytes, Macrophages M1, Macrophages M2, and Mast cells resting showed higher infiltration abundance in the low-risk score group (Fig. 3F). This suggests that the risk score model we constructed can be used to reflect the state of the tumor microenvironment in KIRC.

4. Pathomics Feature Model Predicts Prognostic Risk Score Model

Through feature selection using mRMR and RFE algorithms, we ultimately selected 5 pathological features (Fig. 4A). A pathomics feature model was constructed using the GBM algorithm, and Fig. 4B displays the importance of the features selected in the GBM algorithm. The model's performance was comprehensively evaluated, and the indicators in the training set included sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and Brier score, which were 0.667, 0.857, 0.765, 0.815, 0.732, and 0.178, respectively. In the validation set, the indicators were 0.612, 0.876, 0.749, 0.822, 0.708, and 0.208 (Table S4). In the training dataset, the AUC values of the ROC curve were 0.821, and the AUC values of the PR curve reached 0.841. Calibration curves showed good fitting between our predictions and actual gene levels. Decision curve analysis indicated that our pathomics feature model achieved the maximum net benefit at a threshold probability of 0.2–1, demonstrating clinical utility (Fig. 4C). In the validation dataset, the AUC values of the ROC curve were 0.741, and the AUC values of the PR curve reached 0.736. Calibration curves showed good fitting between our predictions and actual gene levels. Decision curve analysis indicated that our pathomics feature model achieved the maximum net benefit at a threshold probability of 0.35–1, demonstrating clinical utility (Fig. 4D). The prognostic risk score probability, Pathomics score, outputted by the pathomics feature model was higher in the high-risk group than in the low-risk group in both the training set and the validation set (Fig. 4E).

Based on the median pathomics score of the GBM model, samples were divided into high-risk (n = 204) and low-risk groups (n = 203) for prognostic analysis of KIRC patients. The Kaplan-Meier curve showed that patients in the high-risk group had a poorer prognosis (Fig. 4F). Results from both univariate and multivariate Cox regression analyses indicated that a higher pathomics score was an independent prognostic factor for KIRC (Fig. 4G). Therefore, we believe that this pathomics model has potential clinical applications for assessing the prognostic risk of KIRC patients.

5. Impact of PYCRs Knockdown on In Vitro Cell Proliferation and Migration in Human Renal Cell Carcinoma Cell Line Caki-1

We validated the expression of PYCR1 and PYCR2 in human normal kidney tissue and renal cancer cell lines Caki-1, 786-O, and A498. The expression levels of PYCR1 and PYCR2 in renal cancer cells were significantly higher than those in normal kidney tissue (Fig. 5A), consistent with the results of the previous analysis. We knocked down PYCR1 and PYCR2 in the human renal cell carcinoma cell line Caki-1 to investigate their biological functions in KIRC cells (Fig. 5B). The CCK-8 assay revealed that the growth rate of Caki-1 cells was inhibited after knocking down PYCR1 and PYCR2 (Fig. 5C). EdU incorporation experiment results showed that knocking down PYCR1 and PYCR2 inhibited DNA replication in Caki-1 cells (Fig. 5D). Colony formation assay results indicated that the growth rate of Caki-1 cells was inhibited after knocking down PYCR1 and PYCR2 (Fig. 5E). Transwell migration assay demonstrated that the migration ability of Caki-1 cells decreased after knocking down PYCR1 and PYCR2 (Fig. 5F). Similarly, wound healing assay also indicated that knocking down PYCR2 inhibited the migration ability of Caki-1 cells (Fig. 5G). We further observed the expression of epithelial-mesenchymal transition markers (Vimentin), cell cycle protein D1 (Cyclin D1), proliferating cell nuclear antigen (PCNA), and epithelial calcium-binding protein (E-cadherin) in Caki-1 cells after PYCRs knockdown. The results showed that knocking down PYCR1 and PYCR2 downregulated the expression of Vimentin, Cyclin D1, and PCNA, while upregulating the expression of E-cadherin (Fig. 5H). In conclusion, these results indicate that inhibiting PYCR1 and PYCR2 in vitro can suppress the growth and migration of renal cancer cells.

6. Impact of PYCRs Overexpression on In Vitro Cell Proliferation and Migration in Human Renal Cell Carcinoma Cell Line Caki-1

We investigated the biological effects of overexpressing PYCR1 and PYCR2 in the human renal cell carcinoma cell line Caki-1 in vitro (Fig. 6A). The CCK-8 assay results showed that overexpression of PYCR1 and PYCR2 significantly promoted the growth rate of Caki-1 cells (Fig. 6B). EdU incorporation experiment confirmed that overexpressing PYCR1 and PYCR2 accelerated DNA replication in Caki-1 cells (Fig. 6C). Colony formation assay indicated that PYCR1 overexpression led to increased growth of Caki-1 cells (Fig. 6D). Transwell migration assay showed that overexpression of PYCR1 and PYCR2 enhanced the migration ability of Caki-1 cells (Fig. 6E). Wound healing assay also confirmed that PYCR1 overexpression accelerated the migration of Caki-1 cells (Fig. 6F). We further observed the expression of E-cadherin, N-cadherin, Vimentin, and STAT3 in Caki-1 cells after overexpressing PYCRs. The results showed that overexpression of PYCR1 and PYCR2 led to an upregulation of N-cadherin, Vimentin, and STAT3, while downregulating the expression of E-cadherin (Fig. 6G). Overall, these results suggest that overexpressing PYCR1 and PYCR2 in vitro can promote the growth and migration of renal cancer cells.

7. Regulation of the mTOR Signaling Pathway by PYCRs and Proline in Renal Cell Carcinoma

Given the impact of PYCR1 and PYCR2 knockdown and overexpression on the biological functions of renal cancer cells, we further investigated the potential molecular mechanisms involved in KIRC. We studied the influence of overexpressing PYCRs on the mTOR signaling pathway. The results showed that, compared to the control group, overexpression of PYCR1 and PYCR2 did not significantly alter the protein expression of mTOR and P-mTOR. However, downstream molecules of the mTOR signaling pathway, including p70S6K, P-p70S6K, and P-4EBP1, showed significant upregulation, while 4EBP1 expression was downregulated (Fig. 7A). As PYCR is the final enzyme in the proline biosynthetic pathway, we further investigated the impact of different concentrations of proline on the mTOR signaling pathway. The results revealed that with increasing proline concentrations, the protein expression of mTOR and P-mTOR did not show significant changes. However, the expression of p70S6K, P-p70S6K, and P-4EBP1 was significantly upregulated, while 4EBP1 expression was downregulated (Fig. 7B). These results suggest that PYCR1 and PYCR2 activate the mTOR signaling pathway by synthesizing proline, thereby influencing the proliferation and migration of renal cancer cells.

8. Impact and Mechanism Study of Downstream Pathways of PYCRs on the Survival of Renal Cancer Cells

Halofuginone (HF) is a prolyl-tRNA synthetase inhibitor, acting downstream of PYCR, preventing the insertion of proline into newly synthesized proteins, including proline-rich proteins such as collagen[26, 27]. In this section, we used HF to inhibit the insertion of proline into newly synthesized proteins, blocking the synthesis of proline-containing proteins (such as collagen), and observed the impact of this inhibition on the proliferation and apoptosis of renal cancer cells. The results of the CCK-8 cytotoxicity assay showed that with increasing concentrations of HF, cell viability gradually decreased (Fig. 7C). Colony formation experiments indicated that increasing HF concentrations inhibited the growth of Caki-1 cells (Fig. 7D). CCK-8 cell proliferation assay results showed that with increasing HF concentrations, cell activity decreased (Fig. 7E). We further observed the expression of E-cadherin, N-cadherin, PCNA, and Cyclin D1 in Caki-1 cells treated with different concentrations of HF. The results showed downregulation of N-cadherin, PCNA, and Cyclin D1 expression, while E-cadherin expression was upregulated (Fig. 7F). At the same time, we detected the expression of apoptosis-related proteins BCL-2, BAX, BCL-XL, and PARP. The results showed downregulation of BCL-2, BCL-XL, and PARP expression, while BAX expression was upregulated(Fig. 7G). These findings suggest that HF, by inhibiting the insertion of proline into newly synthesized proteins, exerts an inhibitory effect on the proliferation of renal cancer cells and may induce cell apoptosis by regulating the expression of related proteins. This provides an experimental basis for the potential application of HF in the treatment of renal cancer.

The study of the PYCR gene family in pan-cancer has not been reported. In this research, we initially analyzed the expression patterns of PYCRs based on the TCGA database. Compared to adjacent tissues, PYCRs were upregulated in various tumors, and prognosis analysis indicated that PYCRs serve as adverse factors for the overall survival (OS) in multiple cancers. This is consistent with previous studies reporting elevated expression of PYCR1 in cancers such as LIHC, BLCA, and KIRP, which can predict unfavorable prognosis outcomes[9, 12, 28]. In contrast to PYCR1, PYCR2 has been less studied in human tumors. Some research indicates high expression of PYCR2 in hepatocellular carcinoma and colorectal cancer, where it can serve as a prognostic biomarker[11, 29]. Currently, there is limited research on PYCR3 in the development of tumors, with only one report in nasopharyngeal cancer[30]. The development of immune checkpoint inhibitors (ICIs) represents a significant advancement in cancer treatment. However, many patients initially responding to ICIs eventually develop resistance[28]. Therefore, there is an urgent need to identify new biomarkers that can predict the responsiveness to ICI therapy. Increasing evidence suggests that Tumor Mutational Burden (TMB) can predict the effectiveness of ICI treatment, helping to identify patients who can benefit from immunotherapy[31–33]. Additionally, Microsatellite Instability (MSI) is also a biomarker for ICI treatment, and tumors with high MSI have the potential for sustained responses to PD-1/PD-L1 therapy[34, 35]. Our analysis reveals a correlation between the expression of PYCRs and TMB and MSI in various tumors. The tumor microenvironment (TME) plays a crucial role in the growth and metastasis of tumors[36]. We further discovered a close correlation between the expression of PYCRs and immune cell infiltration. Cancer cells are constantly under surveillance by immune cells throughout their lifespan. The development and progression of cancer occur when immune cells fail to destroy precancerous cells[37]. Therefore, high expression of PYCRs in most tumors may lead to a decrease in immune scores, thereby accelerating the development of cancer. These analyses suggest that the functions of PYCRs in tumors still have significant research potential. They not only have the potential to become prognostic biomarkers for various cancers but may also serve as potential targets for targeted cancer therapy. Furthermore, they could be used to predict patient responses to ICI treatment.

In various tumors, PYCRs positively regulate the G2M checkpoint and E2F targets pathways. Since both G2M checkpoint and E2F targets are involved in the positive regulation of the cell cycle[38, 39], we hypothesized that high expression of PYCRs would accelerate tumor growth. This prediction was confirmed in our subsequent cell experiments. Pathways such as TNF-α signaling via NF-κB, Interferon-gamma response, and Inflammatory response are negatively regulated by PYCRs in multiple tumors. TNF-α can induce apoptosis in various cancer cells[40], NF-κB activation plays a crucial role in macrophage polarization and inflammatory cytokine synthesis[37, 41], and Interferon-gamma (IFN-γ) regulates tumor immunity by controlling the transcription of immune-related genes, including inhibition of cell proliferation, promotion of tumor cell apoptosis, and immune regulation[42, 43]. Therefore, the high expression of PYCRs in tumor tissues may assist immune escape by inhibiting inflammatory responses and anti-tumor immunity, thereby promoting tumor development.

Renal cell carcinoma (RCC) is the most common solid lesion in the kidney, accounting for approximately 90% of malignant tumors in the kidney and 3% of all cancers. Among all histological subtypes of RCC, clear cell renal cell carcinoma (KIRC) is the most common, representing about 75% of all RCC cases[44]. In our pan-cancer analysis, we found that the expression and prognosis of the PYCR family in KIRC have significant biological significance. Therefore, we chose to construct prognostic risk models using PYCR1 and PYCR2 separately in KIRC to investigate whether PYCR1 and PYCR2 can serve as prognostic biomarkers for KIRC patients. Our results indicate that the prognosis of the low-risk score group in KIRC is significantly better than that of the high-risk group, and the risk score model can be an independent prognostic factor for KIRC patients. Gene Set Enrichment Analysis (GSEA) reveals significant enrichment of pathways such as E2F Targets, G2/M Checkpoint, EMT, and immune-related pathways in the high-risk score group. Our pan-cancer analysis also shows that PYCR1 and PYCR2 are enriched in E2F Targets, G2/M Checkpoint, and EMT pathways in various cancers. Studies have shown that knocking out PYCR1 can slow down the proliferation and migration of liver cancer cells and colon cancer cells, affecting the expression of EMT-related proteins[9, 10]. This also impacts the proliferation and migration of lung cancer cells and prostate cancer cells[45, 46]. These findings at the molecular level of tumor characteristics provide a rational explanation for the predictive value of the risk score model we constructed for the prognosis of KIRC patients.

Von Hippel-Lindau (VHL) is a tumor suppressor, and mutations in VHL occur in approximately 50% of KIRC patients[47]. In our analysis, the mutation rate of VHL in the high-risk score group was lower. Studies have reported that ccRCC patients with high VHL expression benefit from immunotherapy, suggesting that patients in the high-risk score group in KIRC may be more responsive to immunotherapy[47]. It has been reported that in cells with VHL loss and HIF activation, mTOR mutations lead to higher pathogenicity in clear cell renal cells, which may explain the poorer prognosis in the high-risk group of KIRC patients[48, 49]. We analyzed the infiltration of immune cells in the high and low-risk subgroups of the risk score model. In KIRC, CD4 + T cells and CD8 + T cells showed higher infiltration in the high-risk score group. Generally, the infiltration of CD4 + T cells and CD8 + T cells contributes to the immune response[50, 51]. Cancer immunotherapy aims to promote the activity of cytotoxic T lymphocytes (CTLs) within tumors, assist in initiating tumor-specific CTLs in lymphoid organs, and establish efficient and persistent anti-tumor immunity. The molecular mechanisms by which CD4 + T cells enhance CTL anti-tumor activity have been identified[52]. These results also suggest that patients in the high-risk score group in KIRC may have a better response to immunotherapy.

Precision medicine and genomic medicine combined with artificial intelligence have the potential to enhance patient healthcare[53]. The availability of high-dimensional datasets, coupled with advancements in high-performance computing and machine learning algorithms, has led to the widespread use of AI in various aspects of cancer research[16]. These applications include cancer detection and classification, molecular characterization of tumors and their microenvironment, drug discovery and repurposing, as well as predicting patient treatment outcomes[19]. In this context, we employed machine learning algorithms to construct an artificial intelligence model based on pathomics features, aiming to achieve accurate assessment of the prognosis risk in KIRC patients and predict their outcomes. The results indicate that our constructed pathomics feature model performed well in both the training and validation sets. The AUC values of the ROC curve and PR curve demonstrate the model's ability to provide excellent evaluation of the prognosis in KIRC patients. Furthermore, we stratified samples into high-risk and low-risk groups based on the model's pathomics score median, and Kaplan-Meier curves and Cox regression analysis illustrated that the model could be used for predicting the prognosis of KIRC patients. The construction and validation of this pathomics feature model provide feasible and effective evidence for the application of artificial intelligence in the prognosis assessment of renal cell carcinoma. Studies have reported that AI has fundamentally changed the diagnosis and subtype classification of tumors, improving survival prediction models[54]. Research by Chen et al. suggests that machine learning-based pathomics features can serve as novel prognostic markers for clear cell renal cell carcinoma patients[55]. Machine learning-based pathomics features have also been confirmed for use in bladder cancer diagnosis and survival prediction[56]. These comprehensive results strongly support the potential value of artificial intelligence in advancing breakthroughs and innovations in the field of oncology in the future.

Several studies have indicated that overexpression of PYCR1 and PYCR2 accelerates tumor growth[8–12]. In our experiments, silencing PYCR1 and PYCR2 in renal cancer cell line Caki-1 resulted in inhibited cell growth and migration, whereas overexpression of PYCR1 and PYCR2 promoted cell growth and migration. These findings align with our bioinformatics analysis results and existing research on the biological functions of PYCR1 in tumors such as colorectal cancer, bladder cancer, gastric cancer, among others[57–59], and PYCR2 in colorectal cancer, hepatocellular carcinoma, melanoma, and other cancers[11, 60, 61].

When considering the impact of PYCR1 and PYCR2 on the biological functions of renal cancer cells, we delved into the associated potential molecular mechanisms, with a particular focus on their regulation of the mTOR signaling pathway. The results indicated that, despite no significant changes in the protein expression of mTOR and P-mTOR, the upregulation of p70S6K, P-p70S6K, and P-4EBP1, along with the downregulation of 4EBP1, suggested alterations in the activity of the mTOR signaling pathway. Proline metabolism plays a crucial role in metabolic reprogramming, influencing the occurrence and development of cancer, and can impact cancer cell proliferation, invasion, and metastasis[62]. Further experiments with different concentrations of proline revealed similar changes, indicating that the increased proline concentration led to these alterations. This implies that PYCR1 and PYCR2 activate the mTOR signaling pathway by synthesizing proline, thereby affecting the proliferation and migration of renal cancer cells. These findings align with previous studies reporting that PYCR1 and PYCR2 influence the proliferation of melanoma cells, hepatocellular carcinoma cells, and colorectal cancer cells through the mTOR signaling pathway[15, 46, 63].

Halofuginone (HF), as an inhibitor of prolyl-tRNA synthetase, exerts multiple inhibitory effects by blocking the insertion of proline into newly synthesized proteins[26, 27]. Consistent results from cytotoxicity experiments, colony formation assays, and cell proliferation assays suggest that HF can effectively inhibit the survival, growth, and proliferation of renal cancer cells. Furthermore, observations of changes in the expression of proliferation- and migration-related proteins, along with results indicating alterations in apoptosis-related protein expression, suggest that HF may impact the biological characteristics of renal cancer cells by regulating the expression of proliferation- and migration-related proteins and inducing apoptotic pathways. Studies have reported that HF inhibits T-cell proliferation by blocking proline uptake and inducing apoptosis[27]. Considering the effects of PYCR1, PYCR2, and HF on the mTOR signaling pathway, these findings provide new perspectives for the treatment of renal cancer. Intervening in the proline metabolism pathway, especially through HF-mediated inhibition of proline insertion, may represent a potential therapeutic strategy for renal cancer treatment.

In summary, we conducted a comprehensive analysis of PYCRs in various cancers. We observed high expression of PYCRs in multiple tumors, which is correlated with patient prognosis, indicating the potential of PYCRs as independent prognostic factors in various cancers. Additionally, the expression of PYCRs is associated with TMB and MSI, impacting immune cell infiltration in multiple cancers. However, their role in tumor immunity varies depending on the cancer type. We constructed a prognostic risk model using the expression of PYCRs in KIRC. This model effectively predicts the prognosis, molecular characteristics, and immune features of KIRC patients. Extracting features from H&E-stained images of KIRC, we built a histopathomics feature model to predict the prognosis of KIRC patients, and single-factor and multi-factor Cox regression analyses confirmed that this histopathomics feature model can consistently predict the prognosis of KIRC patients. Our experiments in renal cancer cells further confirmed that PYCR1 and PYCR2 activate the mTOR signaling pathway by synthesizing proline, thereby influencing the proliferation and migration of renal cancer cells. In conclusion, our analysis and experiments contribute to elucidating the role of PYCRs in tumor development, providing new strategies for cancer treatment.

Funding

The research was supported by the grants from the Natural Science Foundation of Fujian Province (2023J01300), the National Natural Science Foundation of China (81773055) and the Fuzhou Science and Technology planning project(2021-S-173).

Disclosure Statement

The authors have no competing interests to declare that are relevant to the content of this article.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Author contributions

HC, QC, MZ, JC, YM and YQ designed the experiment and wrote the manuscript. HC, QC, LD, DY and WC performed the experiments, collected and analyzed the data. JC, XG, RL and WL analyzed the data. HC, MZ and YQ reviewed and modified the manuscript. All authors listed have approved the manuscript that is enclosed.

Availability of data and material

Publicly available datasets were analyzed in this study. This data can be found here: 33 types of cancers from the Xena browser (https://xena.ucsc.edu/).

Acknowledgements

Not applicable.

H. Sung, J. Ferlay, R.L. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal, F. Bray, Global Cancer Statistics 2020, GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries, CA: a Cancer. J. Clin. 71(3), 209–249 (2021). 10.3322/caac.21660
F. Bray, J. Ferlay, I. Soerjomataram, R.L. Siegel, L.A. Torre, A. Jemal, Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 68(6), 394–424 (2018). 10.3322/caac.21492
L. Ding, M.H. Bailey, E. Porta-Pardo, V. Thorsson, A. Colaprico, D. Bertrand et al., Perspective on Oncogenic Processes at the End of the Beginning of Cancer Genomics. Cell. 173(2) (2018). 10.1016/j.cell.2018.03.033
M. Chalecka, A. Kazberuk, J. Palka, A. Surazynski, P5C as an Interface of Proline Interconvertible Amino Acids and Its Role in Regulation of Cell Survival and Apoptosis. Int. J. Mol. Sci. 22(21) (2021). 10.3390/ijms222111763
A.N. Bogner, K.M. Stiers, J.J. Tanner, Structure, biochemistry, and gene expression patterns of the proline biosynthetic enzyme pyrroline-5-carboxylate reductase (PYCR), an emerging cancer therapy target. Amino Acids. 53(12), 1817–1834 (2021). 10.1007/s00726-021-02999-5
S.M. Patel, J. Seravalli, X. Liang, J.J. Tanner, D.F. Becker, Disease variants of human Delta(1)-pyrroline-5-carboxylate reductase 2 (PYCR2). Arch. Biochem. Biophys. 703, 108852 (2021). 10.1016/j.abb.2021.108852
M. Pietzner, E. Wheeler, J. Carrasco-Zanini, A. Cortes, M. Koprulu, M.A. Wörheide et al., Mapping the proteo-genomic convergence of human diseases. Science. 374(6569), eabj1541 (2021). 10.1126/science.abj1541
Y. Gao, L. Luo, Y. Xie, Y. Zhao, J. Yao, X. Liu, PYCR1 knockdown inhibits the proliferation, migration, and invasion by affecting JAK/STAT signaling pathway in lung adenocarcinoma. Mol. Carcinog. 59(5), 503–511 (2020). 10.1002/mc.23174
Y. Xu, W. Zuo, X. Wang, Q. Zhang, X. Gan, N. Tan et al., Deciphering the effects of PYCR1 on cell function and its associated mechanism in hepatocellular carcinoma. Int. J. Biol. Sci. 17(9), 2223–2239 (2021). 10.7150/ijbs.58026
K. Yan, X. Xu, T. Wu, J. Li, G. Cao, Y. Li, Z. Ji, Knockdown of PYCR1 inhibits proliferation, drug resistance and EMT in colorectal cancer cells by regulating STAT3-Mediated p38 MAPK and NF-κB signalling pathway. Biochem. Biophys. Res. Commun. 520(2), 486–491 (2019). 10.1016/j.bbrc.2019.10.059
S. Wang, L. Gu, L. Huang, J. Fang, Z. Liu, Q. Xu, The upregulation of PYCR2 is associated with aggressive colon cancer progression and a poor prognosis. Biochem. Biophys. Res. Commun. 572, 20–26 (2021). 10.1016/j.bbrc.2021.07.084
W. Song, K. Yang, J. Luo, Z. Gao, Y. Gao, Dysregulation of USP18/FTO/PYCR1 signaling network promotes bladder cancer development and progression. Aging (Albany NY). 13(3), 3909–3925 (2021). 10.18632/aging.202359
Z. Li, X. Zhou, J. Huang, Z. Xu, C. Xing, J. Yang, X. Zhou, MicroRNA hsa-miR-150-5p inhibits nasopharyngeal carcinogenesis by suppressing PYCR1 (pyrroline-5-carboxylate reductase 1), Bioengineered 12(2) (2021) 9766–9778. 10.1080/21655979.2021.1995102
S. Xiao, S. Li, Z. Yuan, L. Zhou, Pyrroline-5-carboxylate reductase 1 (PYCR1) upregulation contributes to gastric cancer progression and indicates poor survival outcome. Ann. Transl Med. 8(15), 937 (2020). 10.21037/atm-19-4402
Y. Ye, Y. Wu, J. Wang, Pyrroline-5-carboxylate reductase 1 promotes cell proliferation via inhibiting apoptosis in human malignant melanoma. Cancer Manag Res. 10, 6399–6407 (2018). 10.2147/CMAR.S166711
B. Bhinder, C. Gilvary, N.S. Madhukar, O. Elemento, Artificial Intelligence in Cancer Research and Precision Medicine. Cancer Discov. 11(4), 900–915 (2021). 10.1158/2159-8290.CD-21-0090
C.J. Haug, J.M. Drazen, Artificial Intelligence and Machine Learning in Clinical Medicine, 2023, The New England Journal of Medicine 388(13) (2023) 1201–1208. 10.1056/NEJMra2302038
P. Kickingereder, F. Isensee, I. Tursunova, J. Petersen, U. Neuberger, D. Bonekamp et al., Automated quantitative tumour response assessment of MRI in neuro-oncology with artificial neural networks: a multicentre, retrospective study. Lancet Oncol. 20(5), 728–740 (2019). 10.1016/S1470-2045(19)30098-1
P. Hamet, J. Tremblay, Artificial intelligence in medicine, Metabolism 69S (2017) S36-S40. 10.1016/j.metabol.2017.01.011
K.-H. Yu, C. Zhang, G.J. Berry, R.B. Altman, C. Ré, D.L. Rubin, M. Snyder, Predicting non-small cell lung cancer prognosis by fully automated microscopic pathology image features. Nat. Commun. 7, 12474 (2016). 10.1038/ncomms12474
D. Chen, M. Fu, L. Chi, L. Lin, J. Cheng, W. Xue et al., Prognostic and predictive value of a pathomics signature in gastric cancer. Nat. Commun. 13(1), 6903 (2022). 10.1038/s41467-022-34703-w
A.M. Newman, C.L. Liu, M.R. Green, A.J. Gentles, W. Feng, Y. Xu et al., Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods. 12(5), 453–457 (2015). 10.1038/nmeth.3337
A. Subramanian, P. Tamayo, V.K. Mootha, S. Mukherjee, B.L. Ebert, M.A. Gillette et al., Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U S A 102(43), 15545–15550 (2005). https://pubmed.ncbi.nlm.nih.gov/16199517
D. Sha, Z. Jin, J. Budczies, K. Kluck, A. Stenzinger, F.A. Sinicrope, Tumor Mutational Burden as a Predictive Biomarker in Solid Tumors. Cancer Discov. 10(12), 1808–1825 (2020). 10.1158/2159-8290.CD-20-0522
R.J. Hause, C.C. Pritchard, J. Shendure, S.J. Salipante, Classification and characterization of microsatellite instability across 18 cancer types. Nat. Med. 22(11), 1342–1350 (2016). 10.1038/nm.4191
T.L. Keller, D. Zocco, M.S. Sundrud, M. Hendrick, M. Edenius, J. Yum et al., Halofuginone and other febrifugine derivatives inhibit prolyl-tRNA synthetase. Nat. Chem. Biol. 8(3), 311–317 (2012). 10.1038/nchembio.790
T.L.H. Chu, Q. Guan, C.Y.C. Nguan, C. Du, Halofuginone suppresses T cell proliferation by blocking proline uptake and inducing cell apoptosis. Int. Immunopharmacol. 16(4), 414–423 (2013). 10.1016/j.intimp.2013.04.031
A.J. Schoenfeld, M.D. Hellmann, Acquired Resistance to Immune Checkpoint Inhibitors. Cancer Cell. 37(4), 443–455 (2020). 10.1016/j.ccell.2020.03.017
Q. Gao, H. Zhu, L. Dong, W. Shi, R. Chen, Z. Song et al., Integrated Proteogenomic Characterization of HBV-Related Hepatocellular Carcinoma, Cell 179(2) (2019) 561–577 e22. 10.1016/j.cell.2019.08.052
W. Li, Y. You, X. Zhang, Y. Song, H. Xiang, X. Peng et al., Amplification of chromosome 8q21-qter associated with the acquired paclitaxel resistance of nasopharyngeal carcinoma cells. Int. J. Clin. Exp. Pathol. 8(10), 12346–12356 (2015). https://pubmed.ncbi.nlm.nih.gov/26722421
T.A. Chan, M. Yarchoan, E. Jaffee, C. Swanton, S.A. Quezada, A. Stenzinger, S. Peters, Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann. Oncol. 30(1), 44–56 (2019). 10.1093/annonc/mdy495
A. Snyder, V. Makarov, T. Merghoub, J. Yuan, J.M. Zaretsky, A. Desrichard et al., Genetic basis for clinical response to CTLA-4 blockade in melanoma. N Engl. J. Med. 371(23), 2189–2199 (2014). 10.1056/NEJMoa1406498
N.A. Rizvi, M.D. Hellmann, A. Snyder, P. Kvistborg, V. Makarov, J.J. Havel et al., Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science. 348(6230), 124–128 (2015). 10.1126/science.aaa1348
A.M. Goodman, E.S. Sokol, G.M. Frampton, S.M. Lippman, R. Kurzrock, Microsatellite-Stable Tumors with High Mutational Burden Benefit from Immunotherapy. Cancer Immunol. Res. 7(10), 1570–1573 (2019). 10.1158/2326-6066.CIR-19-0149
E. Vilar, S.B. Gruber, Microsatellite instability in colorectal cancer-the stable evidence. Nat. Rev. Clin. Oncol. 7(3), 153–162 (2010). 10.1038/nrclinonc.2009.237
H.-Y. Tan, N. Wang, W. Lam, W. Guo, Y. Feng, Y.-C. Cheng, Targeting tumour microenvironment by tyrosine kinase inhibitor. Mol. Cancer. 17(1), 43 (2018). 10.1186/s12943-018-0800-6
H. Zhu, X. Hu, Y. Ye, Z. Jian, Y. Zhong, L. Gu, X. Xiong, Pan-Cancer Analysis of PIMREG as a Biomarker for the Prognostic and Immunological Role. Front. Genet. 12, 687778 (2021). 10.3389/fgene.2021.687778
N. Barnaba, J.R. LaRocque, Targeting cell cycle regulation via the G2-M checkpoint for synthetic lethality in melanoma. Cell. Cycle. 20(11), 1041–1051 (2021). 10.1080/15384101.2021.1922806
L.N. Kent, G. Leone, The broken cycle: E2F dysfunction in cancer. Nat. Rev. Cancer. 19(6), 326–338 (2019). 10.1038/s41568-019-0143-7
T.M. Young, C. Reyes, E. Pasnikowski, C. Castanaro, C. Wong, C.E. Decker et al., Autophagy protects tumors from T cell-mediated cytotoxicity via inhibition of TNFα-induced apoptosis. Sci. Immunol. 5(54) (2020). 10.1126/sciimmunol.abb9561
Y. Ye, T. Jin, X. Zhang, Z. Zeng, B. Ye, J. Wang et al., Meisoindigo Protects Against Focal Cerebral Ischemia-Reperfusion Injury by Inhibiting NLRP3 Inflammasome Activation and Regulating Microglia/Macrophage Polarization via TLR4/NF-κB Signaling Pathway. Front. Cell. Neurosci. 13, 553 (2019). 10.3389/fncel.2019.00553
K. Schroder, P.J. Hertzog, T. Ravasi, D.A. Hume, Interferon-gamma: an overview of signals, mechanisms and functions. J. Leukoc. Biol. 75(2), 163–189 (2004). https://pubmed.ncbi.nlm.nih.gov/14525967
F. Castro, A.P. Cardoso, R.M. Goncalves, K. Serre, M.J. Oliveira, Interferon-Gamma at the Crossroads of Tumor Immune Surveillance or Evasion. Front. Immunol. 9, 847 (2018). 10.3389/fimmu.2018.00847
J. Ferlay, M. Colombet, I. Soerjomataram, T. Dyba, G. Randi, M. Bettio et al., Cancer incidence and mortality patterns in Europe: Estimates for 40 countries and 25 major cancers in 2018. Eur. J. Cancer. 103, 356–387 (2018). 10.1016/j.ejca.2018.07.005
D. Wang, L. Wang, Y. Zhang, Z. Yan, L. Liu, G. Chen, PYCR1 promotes the progression of non-small-cell lung cancer under the negative regulation of miR-488. Biomed. Pharmacother. 111, 588–595 (2019). 10.1016/j.biopha.2018.12.089
F. Yin, X. Huang, Y. Xuan, Pyrroline-5-Carboxylate Reductase-2 Promotes Colorectal Cancer Progression via Activating PI3K/AKT/mTOR Pathway, Dis Markers 2021 (2021) 9950663. 10.1155/2021/9950663
S. Zhu, W. Ding, Y. Chen, W. Wang, R. Xu, C. Liu et al., High VHL Expression Reverses Warburg Phenotype and Enhances Immunogenicity in Kidney Tumor Cells. Genomics Proteom. Bioinf. (2021). 10.1016/j.gpb.2019.12.002
A.K. Murugan, mTOR: Role in cancer, metastasis and drug resistance. Semin Cancer Biol. 59, 92–111 (2019). 10.1016/j.semcancer.2019.07.003
J. Xu, C.G. Pham, S.K. Albanese, Y. Dong, T. Oyama, C.-H. Lee et al., Mechanistically distinct cancer-associated mTOR activation clusters predict sensitivity to rapamycin. J. Clin. Invest. 126(9), 3526–3540 (2016). 10.1172/JCI86120
H. Matsumoto, A.A. Thike, H. Li, J. Yeong, S.-L. Koo, R.A. Dent et al., Increased CD4 and CD8-positive T cell infiltrate signifies good prognosis in a subset of triple-negative breast cancer. Breast Cancer Res. Treat. 156(2), 237–247 (2016). 10.1007/s10549-016-3743-x
R. Huang, M. Mao, Y. Lu, Q. Yu, L. Liao, A novel immune-related genes prognosis biomarker for melanoma: associated with tumor microenvironment. Aging (Albany NY). 12(8), 6966–6980 (2020). 10.18632/aging.103054
J. Borst, T. Ahrends, N. Babala, C.J.M. Melief, W. Kastenmuller, CD4(+) T cell help in cancer immunology and immunotherapy. Nat. Rev. Immunol. 18(10), 635–647 (2018). 10.1038/s41577-018-0044-0
S. Quazi, Artificial intelligence and machine learning in precision and genomic medicine. Med. Oncol. 39(8), 120 (2022). 10.1007/s12032-022-01711-1
J.E. Knudsen, J.M. Rich, R. Ma, Artificial Intelligence in Pathomics and Genomics of Renal Cell Carcinoma. Urol. Clin. North. Am. 51(1), 47–62 (2024). 10.1016/j.ucl.2023.06.002
S. Chen, L. Jiang, F. Gao, E. Zhang, T. Wang, N. Zhang et al., Machine learning-based pathomics signature could act as a novel prognostic marker for patients with clear cell renal cell carcinoma. Br. J. Cancer. 126(5), 771–777 (2022). 10.1038/s41416-021-01640-2
S. Chen, L. Jiang, X. Zheng, J. Shao, T. Wang, E. Zhang et al., Clinical use of machine learning-based pathomics signature for diagnosis and survival prediction of bladder cancer. Cancer Sci. 112(7), 2905–2914 (2021). 10.1111/cas.14927
B. Zhou, Z. Mai, Y. Ye, Y. Song, M. Zhang, X. Yang et al., The role of PYCR1 in inhibiting 5-fluorouracil-induced ferroptosis and apoptosis through SLC25A10 in colorectal cancer. Hum. Cell. 35(6), 1900–1911 (2022). 10.1007/s13577-022-00775-5
S. Xiao, X. Yao, J. Ye, X. Tian, Z. Yin, L. Zhou, Epigenetic modification facilitates proline synthase PYCR1 aberrant expression in gastric cancer. Biochim. Biophys. Acta Gene Regul. Mech. 1865(6), 194829 (2022). 10.1016/j.bbagrm.2022.194829
S. Du, Y. Sui, W. Ren, J. Zhou, C. Du, PYCR1 promotes bladder cancer by affecting the Akt/Wnt/β-catenin signaling. J. Bioenerg Biomembr. 53(2), 247–258 (2021). 10.1007/s10863-021-09887-3
S. Wang, W. Yi, Z. Xu, M. Shi, PYCR2 promotes growth and aerobic glycolysis in human liver cancer by regulating the AKT signaling pathway. Biochem. Biophys. Res. Commun. 680, 15–24 (2023). 10.1016/j.bbrc.2023.09.007
R. Ou, X. Zhang, J. Cai, X. Shao, M. Lv, W. Qiu et al., Downregulation of pyrroline-5-carboxylate reductase-2 induces the autophagy of melanoma cells via AMPK/mTOR pathway. Tumour Biol. 37(5), 6485–6491 (2016). 10.1007/s13277-015-3927-8
P. Geng, W. Qin, G. Xu, Proline metabolism in cancer. Amino Acids. 53(12), 1769–1777 (2021). 10.1007/s00726-021-03060-1
J. Zhang, L. Shang, W. Jiang, W. Wu, Shikonin induces apoptosis and autophagy via downregulation of pyrroline-5-carboxylate reductase1 in hepatocellular carcinoma cells. Bioengineered. 13(3), 7904–7918 (2022). 10.1080/21655979.2022.2052673

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Deciphering the effects of PYCR family on cell function, prognostic value, immune infiltration in ccRCC and pan-cancer

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Materials and Methods

1. TCGA Pan-Cancer Data and Acquisition/Processing of H&E Pathology Images

2. Prognostic Analysis

3. Immune Cell Infiltration Analysis and Gene Set Enrichment Analysis (GSEA)

4. Tumor Mutation Burden (TMB) and Microsatellite Instability (MSI) Analysis

5. Construction of the KIRC Prognostic Risk Score Model

6. Pathological Image Segmentation and Feature Extraction

7. Pathomics Feature Model Construction

8. Cell Culture and Transfection

9. Protein Blotting

10. Cell Proliferation

11. Cell Migration

12. Drug treatment

13. Statistical Analysis

Results

1. Analysis of PYCRs in Pan-Cancer

2. Construction of OS-Related Prognostic Risk Score Model for PYCRs in KIRC

3. Molecular and Immune Features of Risk Score Model Subgroups

4. Pathomics Feature Model Predicts Prognostic Risk Score Model

7. Regulation of the mTOR Signaling Pathway by PYCRs and Proline in Renal Cell Carcinoma

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1