Bioinformatics analysis identified MMP14 and COL12A1 as immune-related biomarkers associated with pancreatic adenocarcinoma prognosis

Continually adapting pre-trained language model to universal annotation of single-cell RNA-seq data

MOTIVATION

Cell-type annotation of single-cell RNA-sequencing (scRNA-seq) data is a hallmark of biomedical research and clinical application. Current annotation tools usually assume the simultaneous acquisition of well-annotated data, but without the ability to expand knowledge from new data. Yet, such tools are inconsistent with the continuous emergence of scRNA-seq data, calling for a continuous cell-type annotation model. In addition, by their powerful ability of information integration and model interpretability, transformer-based pre-trained language models have led to breakthroughs in single-cell biology research. Therefore, the systematic combining of continual learning and pre-trained language models for cell-type annotation tasks is inevitable.

RESULTS

We herein propose a universal cell-type annotation tool, called CANAL, that continuously fine-tunes a pre-trained language model trained on a large amount of unlabeled scRNA-seq data, as new well-labeled data emerges. CANAL essentially alleviates the dilemma of catastrophic forgetting, both in terms of model inputs and outputs. For model inputs, we introduce an experience replay schema that repeatedly reviews previous vital examples in current training stages. This is achieved through a dynamic example bank with a fixed buffer size. The example bank is class-balanced and proficient in retaining cell-type-specific information, particularly facilitating the consolidation of patterns associated with rare cell types. For model outputs, we utilize representation knowledge distillation to regularize the divergence between previous and current models, resulting in the preservation of knowledge learned from past training stages. Moreover, our universal annotation framework considers the inclusion of new cell types throughout the fine-tuning and testing stages. We can continuously expand the cell-type annotation library by absorbing new cell types from newly arrived, well-annotated training datasets, as well as automatically identify novel cells in unlabeled datasets. Comprehensive experiments with data streams under various biological scenarios demonstrate the versatility and high model interpretability of CANAL.

AVAILABILITY

An implementation of CANAL is available from https://github.com/aster-ww/CANAL-torch.

CONTACT

dengmh@pku.edu.cn.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Journal Name online.

In situ grown magnetic COF@MOF with a phosphoserine anchor for in-depth N-glycopeptide analysis in serum

A hydrophilic phosphoserine-functionalized magnetic organic framework composite (termed Fe3O4@COF@MOF-PS) was synthesized by an in situ growth strategy for effective capture of N-glycopeptides. Fe3O4@COF@MOF-PS exhibited high sensitivity (0.2 fmol μL-1), outstanding exclusion of size capability (1 : 10 000), good selectivity (1 : 2000), and reusability (at least 10 times). It also exhibited remarkable performance in the N-glycopeptide analysis in complex biological samples. Via nano-LC-MS/MS analysis, a total of 223 N-glycopeptides with 161 glycosylation sites assigned to 91 glycoproteins and 331 N-glycopeptides with 243 glycosylation sites assigned to 134 glycoproteins were identified in sera from cervical cancer patients and normal controls, respectively. Biological processes and molecular functional analyses indicate that the captured glycoproteins are of significant relevance to cervical cancer, for example, gene coverage or expression of cell adhesion and extracellular matrix structural constituents. Thus, Fe3O4@COF@MOF-PS not only efficiently captures N-glycopeptides, but also has the possibility of screening potential disease markers and elucidating the process of cervical cancer development.

PABPC1 promotes cell proliferation and metastasis in pancreatic adenocarcinoma by regulating COL12A1 expression

BACKGROUND

The expression of cytoplasmic poly (A) binding protein-1 (PABPC1) has been reported in multiple cancer types. This protein is known to modulate cancer progression. However, the effects of PABPC1 expression in pancreatic adenocarcinoma (PAAD) have not been investigated. Here, we investigate the regulatory targets and molecular mechanisms of PABPC1 in PAAD.

METHODS

PABPC1 and collagen type XII α1 chain (COL12A1) expression in PAAD and their role in tumor prognosis and tumor stage were investigated using The Cancer Genome Atlas database analysis. After silencing PABPC1, messenger RNA sequencing and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were performed. The expression of differentially expressed genes (DEGs), cell viability, apoptosis, and cell migration and invasion were explored using reverse transcription-quantitative polymerase chain reaction, Cell Counting Kit-8 assay, flow cytometry assay, and transwell assay, respectively. The relationship between PABPC1 and COL12A1 expression was assessed by Pearson's correlation analysis. The regulatory function of COL12A1 in PABPC1-affected BXPC3 cell behavior was studied after COL12A1 was overexpressed.

RESULTS

PABPC1 and COL12A1 expression was upregulated in patients with PAAD and was linked to poor prognosis. Four hundred and seventy-four DEGs were observed in BXPC3 cells after PABPC1 silencing. GO and KEGG analyses revealed that the top 10 DEGs were enriched in cell adhesion pathways. Additionally, PABPC1 silencing inhibited cell viability, migration, and invasion and accelerated apoptosis in BXPC3 cells. PABPC1 silencing increased AZGP1 and ARHGAP30 expression and decreased CAV1 and COL12A1 expression in BXPC3 cells. PABPC1 positively mediated COL12A1 expression, whereas PABPC1 knockdown induced the inhibition of BXPC3 cell proliferation, migration, and invasion.

CONCLUSION

The results of this study indicate that PABPC1 may function as a tumor promoter in PAAD, accelerating BXPC3 cell proliferation and metastasis by regulating COL12A1 expression.

COL12A1 Acts as a Novel Prognosis Biomarker and Activates Cancer-Associated Fibroblasts in Pancreatic Cancer through Bioinformatics and Experimental Validation

Pancreatic cancer remains one of the most challenging malignancies to date and is associated with poor survival. Cancer-associated fibroblasts (CAFs) are key stromal cells in the tumor microenvironment (TME) that play a crucial role in tumor progression in pancreatic cancer. Thus, uncovering the key genes involved in CAF progression and determining their prognostic value is critically important. Herein, we report our discoveries in this research area. Analysis of The Cancer Genome Atlas (TCGA) dataset and investigation of our clinical tissue samples indicated that COL12A1 expression was aberrantly highly expressed in pancreatic cancer. Survival and COX regression analyses revealed the significant clinical prognostic value of COL12A1 expression in pancreatic cancer. COL12A1 was mainly expressed in CAFs but not in tumor cells. This was verified with our PCR analysis in cancer cells and CAFs. The knocking down of COL12A1 decreased the proliferation and migration of CAFs and down-regulated the expression of CAF activation markers actin alpha 2 (ACTA2), fibroblast activation protein (FAP), and fibroblast-specific protein 1 (FSP1). Meanwhile, the interleukin 6 (IL6), CXC chemokine Ligand-5 (CXCL5), and CXC chemokine Ligand-10 (CXCL10) expressions were inhibited, and the cancer-promoting effect was reversed by COL12A1 knockdown. Therefore, we demonstrated the potential prognostic and target therapy value of COL12A1 expression in pancreatic cancer and elucidated the molecular mechanism underlying its role in CAFs. The findings of this study might provide new opportunities for TME-targeted therapies in pancreatic cancer.

Epigenetic dysregulation-mediated COL12A1 upregulation predicts worse outcome in intrahepatic cholangiocarcinoma patients

BACKGROUND

Collagen type XII alpha 1 chain (COL12A1) is associated with human cancer progression. Nevertheless, the expression pattern and the function of COL12A1 in intrahepatic cholangiocarcinoma (iCCA) remain unknown. The present study was performed to assess the role of COL12A1 in iCCA.

RESULTS

A total of 1669 genes, differentially expressed between iCCA and nontumor liver tissue samples, were identified as potential tumor-specific biomarkers for iCCA patients. Of these, COL12A1 was significantly upregulated in clinical iCCA tissue samples and correlated with epithelial-mesenchymal transition gene set enrichment score and advanced tumor stage in clinical iCCA. COL12A1-high expression was associated with the poor prognoses of iCCA patients (n = 421) from four independent cohorts. Promoter hypermethylation-induced downregulation of miR-424-5p resulted in COL12A1 upregulation in clinical iCCA. Experimental knockout of COL12A1 inhibited the proliferation, invasiveness and growth of iCCA cells. MiR-424-5p had a therapeutic potential in iCCA via directly targeting COL12A1.

CONCLUSIONS

Promoter hypermethylation-induced miR-424-5p downregulation contributes to COL12A1 upregulation in iCCA. COL12A1 is a promising druggable target for epigenetic therapy of iCCA.

Clinical diagnosis and management of pancreatic cancer: Markers, molecular mechanisms, and treatment options

Pancreatic cancer (PC) is the third-leading cause of cancer deaths. The overall 5-year survival rate of PC is 9%, and this rate for metastatic PC is below 3%. However, the PC-induced death cases will increase about 2-fold by 2060. Many factors such as genetic and environmental factors and metabolic diseases can drive PC development and progression. The most common type of PC in the clinic is pancreatic ductal adenocarcinoma, comprising approximately 90% of PC cases. Multiple pathogenic processes including but not limited to inflammation, fibrosis, angiogenesis, epithelial-mesenchymal transition, and proliferation of cancer stem cells are involved in the initiation and progression of PC. Early diagnosis is essential for curable therapy, for which a combined panel of serum markers is very helpful. Although some mono or combined therapies have been approved by the United States Food and Drug Administration for PC treatment, current therapies have not shown promising outcomes. Fortunately, the development of novel immunotherapies, such as oncolytic viruses-mediated treatments and chimeric antigen receptor-T cells, combined with therapies such as neoadjuvant therapy plus surgery, and advanced delivery systems of immunotherapy will improve therapeutic outcomes and combat drug resistance in PC patients. Herein, the pathogenesis, molecular signaling pathways, diagnostic markers, prognosis, and potential treatments in completed, ongoing, and recruiting clinical trials for PC were reviewed.

Increased Expression of the RBPMS Splice Variants Inhibits Cell Proliferation in Ovarian Cancer Cells

RNA-Binding Protein with Multiple Splicing (RBPMS) is a member of family proteins that bind to nascent RNA transcripts and regulate their splicing, localization, and stability. Evidence indicates that RBPMS controls the activity of transcription factors associated with cell growth and proliferation, including AP-1 and Smads. Three major RBPMS protein splice variants (RBPMSA, RBPMSB, and RBPMSC) have been described in the literature. We previously reported that reduced RBPMS levels decreased the sensitivity of ovarian cancer cells to cisplatin treatment. However, little is known about the biological role of the RBPMS splice variants in ovarian cancer cells. We performed RT-PCR and Western blots and observed that both RBPMSA and RBPMSC are reduced at the mRNA and protein levels in cisplatin resistant as compared with cisplatin sensitive ovarian cancer cells. The mRNA and protein levels of RBPMSB were not detectable in any of the ovarian cancer cells tested. To better understand the biological role of each RBPMSA and RBPMSC, we transfected these two splice variants in the A2780CP20 and OVCAR3CIS cisplatin resistant ovarian cancer cells and performed cell proliferation, cell migration, and invasion assays. Compared with control clones, a significant reduction in the number of colonies, colony size, cell migration, and invasion was observed with RBPMSA and RBPMSC overexpressed cells. Moreover, A2780CP20-RBPMSA and A2780CP20-RBPMSC clones showed reduced senescence-associated β-galactosidase (β-Gal)-levels when compared with control clones. A2780CP20-RBPMSA clones were more sensitive to cisplatin treatment as compared with A2780CP20-RBPMSC clones. The A2780CP20-RBPMSA and A2780CP20-RBPMSC clones subcutaneously injected into athymic nude mice formed smaller tumors as compared with A2780CP20-EV control group. Additionally, immunohistochemical analysis showed lower proliferation (Ki67) and angiogenesis (CD31) staining in tissue sections of A2780CP20-RBPMSA and A2780CP20-RBPMSC tumors compared with controls. RNAseq studies revealed many common RNA transcripts altered in A2780CP20-RBPMSA and A2780CP20-RBPMSC clones. Unique RNA transcripts deregulated by each RBPMS variant were also observed. Kaplan-Meier (KM) plotter database information identified clinically relevant RBPMSA and RBPMSC downstream effectors. These studies suggest that increased levels of RBPMSA and RBPMSC reduce cell proliferation in ovarian cancer cells. However, only RBPMSA expression levels were associated with the sensitivity of ovarian cancer cells to cisplatin treatment.

A cancer-associated fibroblast gene signature predicts prognosis and therapy response in patients with pancreatic cancer

Pancreatic cancer is a lethal malignancy with a 5-year survival rate of about 10% in the United States, and it is becoming an increasingly prominent cause of cancer death. Among pancreatic cancer patients, pancreatic ductal adenocarcinoma (PDAC) accounts for more than 90% of all cases and has a very poor prognosis with an average survival of only 1 year in about 18% of all tumor stages. In the past years, there has been an increasing interest in cancer-associated fibroblasts (CAFs) and their roles in PDAC. Recent data reveals that CAFs in PDAC are heterogeneous and various CAF subtypes have been demonstrated to promote tumor development while others hinder cancer proliferation. Furthermore, CAFs and other stromal populations can be potentially used as novel prognostic markers in cancer. In the present study, in order to evaluate the prognostic value of CAFs in PDAC, CAF infiltration rate was evaluated in 4 PDAC datasets of TCGA, GEO, and ArrayExpress databases and differentially expressed genes (DEGs) between CAF-high and CAF-low patients were identified. Subsequently, a CAF-based gene expression signature was developed and studied for its association with overall survival (OS). Additionally, functional enrichment analysis, somatic alteration analysis, and prognostic risk model construction was conducted on the identified DEGs. Finally, oncoPredict algorithm was implemented to assess drug sensitivity prediction between high- and low-risk cohorts. Our results revealed that CAF risk-high patients have a worse survival rate and increased CAF infiltration is a poor prognostic indicator in pancreatic cancer. Functional enrichment analysis also revealed that "extracellular matrix organization" and "vasculature development" were the top enriched pathways among the identified DEGs. We also developed a panel of 12 genes, which in additional to its prognostic value, could predict higher chemotherapy resistance rate. This CAF-based panel can be potentially utilized alone or in conjunction with other clinical parameters to make early predictions and prognosticate responsiveness to treatment in PDAC patients. Indeed, it is necessary to conduct extensive prospective investigations to confirm the clinical utility of these findings.

Analysis of Yes-Associated Protein-1 (YAP1) Target Gene Signature to Predict Progressive Breast Cancer

Breast cancers are treated according to the ER/PR or HER2 expression and show better survival outcomes with targeted therapy. Triple-negative breast cancers (TNBCs) with a lack of expression of ER/PR and HER2 are treated with systemic therapy with unpredictable responses and outcomes. It is essential to investigate novel markers to identify targeted therapies for TNBC. One such marker is YAP1, a transcription co-activator protein that shows association with poor prognosis of breast cancer. YAP1 transcriptionally regulates the expression of genes that drive the oncogenic phenotypes. Here, we assess a potential YAP target gene signature to predict a progressive subset of breast tumors from METABRIC and TCGA datasets. YAP1 target genes were shortlisted based on expression correlation and concordance with YAP1 expression and significant association with survival outcomes of patients. Hierarchical clustering was performed for the shortlisted genes. The utility of the clustered genes was assessed by survival analysis to identify a recurring subset. Expression of the shortlisted target genes showed significant association with survival outcomes of HER2-positive and TNBC subset in both datasets. The shortlisted genes were verified using an independent dataset. Further validation using IHC can prove the utility of this potential prognostic signature to identify a recurrent subset of HER2-positive and TNBC subtypes.

Identification of Three Core Secretome Genes Associated with Immune Infiltration in High Tumor Mutation Burden Across 14 Major Solid Tumors

Background

Secretome genes, encoding proteins secreted from the cell, are involved in the tumor immune response and correlated with levels of tumor mutation burden (TMB) in multiple tumors. This study aimed to identify core secretome genes and their potential association with immunomodulators and immune infiltration in high TMB groups across 14 major solid tumors through bioinformatics analysis.

Methods

Multi-omics data for 14 major solid tumors were downloaded from The Cancer Genome Atlas (TCGA) database. Patients were divided into high TMB (TMB-high) and low TMB (TMB-low) groups using the median TMB values for each of the solid tumors. The CIBERSORT algorithm was conducted to estimate the proportion of 22 tumor-infiltrating immune cells (TIICs). Kaplan-Meier analysis and the log-rank test were utilized to screened prognosis-related genes. The correlations between core secretome genes and TIICs were analyzed using Spearman correlation coefficients.

Results

In TMB-high groups, multi-omics data analysis revealed that secretome genes were strongly associated with clinical characteristics, and 65 prognosis-related secretome genes were screened. Among the prognosis-related genes, 21 core secretome genes were identified, and strongly associated with five types of TIICs, namely activated NK cells, follicular helper T cells, CD8 T cells, and macrophages M0 and M2. Notably, three secretome genes (ADAMTS12, COL12A1, and COL5A2) were significantly related to immunomodulators and TIICs in multiple solid tumors. In addition, 12 core secretome genes were significantly differentially expressed between responding and non-responding patients receiving immunotherapy. Furthermore, core secretome genes may be involved in the PI3K/AKT signaling pathway.

Conclusion

We examined the prognostic significance of secretome genes and their potential association with immunomodulators and immune infiltration across 14 major solid tumors. In summary, three secretome genes (ADAMTS12, COL12A1, and COL5A2) may be pivotal mediators of immune infiltration in TMB-high patients, which may help to identify patients who could benefit from immunotherapy.