Improved cell composition deconvolution method of bulk gene expression profiles to quantify subsets of immune cells

Desmocollin-3 and Bladder Cancer

BACKGROUND

Desmocollin3, a transmembrane protein, is expressed in the basal/suprabasal layer of normal stratified epithelium. DSC3 gene expression is described in muscle-invasive bladder cancer (MIBC). DSC3-protein-expressing recurrent non-muscle-invasive bladder cancer (NMIBC) had a durable response to CADI-03, a DSC3-specific active immunotherapy.

METHODS

We evaluated DSC3 protein expression and its correlation with tumor-infiltrating immune cells in bladder cancer. DSC3 gene expression and its correlation with 208 immune encoding genes, treatment outcome, and survival were evaluated using the "ARRAYEXPRESS" and "TCGA" datasets. Immune genes were grouped as tumor-controlling immune genes (TCIGs) and tumor-promoting immune genes (TPIGs) as per their functions.

RESULTS & CONCLUSIONS

NMIBC had higher DSC3 expression compared to MIBC. More immune genes were correlated with DSC3 in MIBC (21) compared to NMIBC (11). Amongst the TCIGs, six in NMIBC and one in MIBC had a negative correlation while two in NMIBC and nine in MIBC had a positive correlation with DSC3. Amongst the TPIGs, nine in NMIBC and five in MIBC had a negative correlation. Seven TPIGs had a positive correlation with DSC3 in MIBC and none in NMIBC. Of the T cell exhaustion markers, none were correlated with DSC3 in MIBC. Among NMIBC, CTLA4 and TIGIT were the only markers of exhaustion that demonstrated a negative correlation with DSC3. DSC3 expression was also higher in p53 mutant compared to wild p53, non-papillary MIBC compared to papillary MIBC, and in basal, squamous molecular subtype compared to luminal MIBC. MIBC with lower DSC3 expression had better outcomes (response, survival) compared to those with higher DSC3 expression.

Assessing transcriptomic heterogeneity of single-cell RNASeq data by bulk-level gene expression data

BACKGROUND

Single-cell RNA sequencing (sc-RNASeq) data illuminate transcriptomic heterogeneity but also possess a high level of noise, abundant missing entries and sometimes inadequate or no cell type annotations at all. Bulk-level gene expression data lack direct information of cell population composition but are more robust and complete and often better annotated. We propose a modeling framework to integrate bulk-level and single-cell RNASeq data to address the deficiencies and leverage the mutual strengths of each type of data and enable a more comprehensive inference of their transcriptomic heterogeneity. Contrary to the standard approaches of factorizing the bulk-level data with one algorithm and (for some methods) treating single-cell RNASeq data as references to decompose bulk-level data, we employed multiple deconvolution algorithms to factorize the bulk-level data, constructed the probabilistic graphical models of cell-level gene expressions from the decomposition outcomes, and compared the log-likelihood scores of these models in single-cell data. We term this framework backward deconvolution as inference operates from coarse-grained bulk-level data to fine-grained single-cell data. As the abundant missing entries in sc-RNASeq data have a significant effect on log-likelihood scores, we also developed a criterion for inclusion or exclusion of zero entries in log-likelihood score computation.

RESULTS

We selected nine deconvolution algorithms and validated backward deconvolution in five datasets. In the in-silico mixtures of mouse sc-RNASeq data, the log-likelihood scores of the deconvolution algorithms were strongly anticorrelated with their errors of mixture coefficients and cell type specific gene expression signatures. In the true bulk-level mouse data, the sample mixture coefficients were unknown but the log-likelihood scores were strongly correlated with accuracy rates of inferred cell types. In the data of autism spectrum disorder (ASD) and normal controls, we found that ASD brains possessed higher fractions of astrocytes and lower fractions of NRGN-expressing neurons than normal controls. In datasets of breast cancer and low-grade gliomas (LGG), we compared the log-likelihood scores of three simple hypotheses about the gene expression patterns of the cell types underlying the tumor subtypes. The model that tumors of each subtype were dominated by one cell type persistently outperformed an alternative model that each cell type had elevated expression in one gene group and tumors were mixtures of those cell types. Superiority of the former model is also supported by comparing the real breast cancer sc-RNASeq clusters with those generated by simulated sc-RNASeq data.

CONCLUSIONS

The results indicate that backward deconvolution serves as a sensible model selection tool for deconvolution algorithms and facilitates discerning hypotheses about cell type compositions underlying heterogeneous specimens such as tumors.

Psychiatric risk gene Transcription Factor 4 (TCF4) regulates the density and connectivity of distinct inhibitory interneuron subtypes

Transcription factor 4 (TCF4) is a basic helix-loop-helix transcription factor that is implicated in a variety of psychiatric disorders including autism spectrum disorder (ASD), major depression, and schizophrenia. Autosomal dominant mutations in TCF4 are causal for a specific ASD called Pitt-Hopkins Syndrome (PTHS). However, our understanding of etiological and pathophysiological mechanisms downstream of TCF4 mutations is incomplete. Single cell sequencing indicates TCF4 is highly expressed in GABAergic interneurons (INs). Here, we performed cell-type specific expression analysis (CSEA) and cellular deconvolution (CD) on bulk RNA sequencing data from 5 different PTHS mouse models. Using CSEA we observed differentially expressed genes (DEGs) were enriched in parvalbumin expressing (PV+) INs and CD predicted a reduction in the PV+ INs population. Therefore, we investigated the role of TCF4 in regulating the development and function of INs in the Tcf4+/tr mouse model of PTHS. In Tcf4+/tr mice, immunohistochemical (IHC) analysis of subtype-specific IN markers and reporter mice identified reductions in PV+, vasoactive intestinal peptide (VIP+), and cortistatin (CST+) expressing INs in the cortex and cholinergic (ChAT+) INs in the striatum, with the somatostatin (SST+) IN population being spared. The reduction of these specific IN populations led to cell-type specific alterations in the balance of excitatory and inhibitory inputs onto PV+ and VIP+ INs and excitatory pyramidal neurons within the cortex. These data indicate TCF4 is a critical regulator of the development of specific subsets of INs and highlight the inhibitory network as an important source of pathophysiology in PTHS.

Tumor microenvironment characterization in colorectal cancer to identify prognostic and immunotherapy genes signature

BACKGROUND

The tumor microenvironment (TME) plays a crucial role in tumorigenesis, progression, and therapeutic response in many cancers. This study aimed to comprehensively investigate the role of TME in colorectal cancer (CRC) by generating a TMEscore based on gene expression.

METHODS

The TME patterns of CRC datasets were investigated, and the TMEscores were calculated. An unsupervised clustering method was used to divide samples into clusters. The associations between TMEscores and clinical features, prognosis, immune score, gene mutations, and immune checkpoint inhibitors were analyzed. A TME signature was constructed using the TMEscore-related genes. The results were validated using external and clinical cohorts.

RESULTS

The TME pattern landscape was for CRC was examined using 960 samples, and then the TMEscore pattern of CRC datasets was evaluated. Two TMEscore clusters were identified, and the high TMEscore cluster was associated with early-stage CRC and better prognosis in patients with CRC when compared with the low TMEscore clusters. The high TMEscore cluster indicated elevated tumor cell scores and tumor gene mutation burden, and decreased tumor purity, when compared with the low TMEscore cluster. Patients with high TMEscore were more likely to respond to immune checkpoint therapy than those with low TMEscore. A TME signature was constructed using the TMEscore-related genes superimposing the results of two machine learning methods (LASSO and XGBoost algorithms), and a TMEscore-related four-gene signature was established, which had a high predictive value for discriminating patients from different TMEscore clusters. The prognostic value of the TMEscore was validated in two independent cohorts, and the expression of TME signature genes was verified in four external cohorts and clinical samples.

CONCLUSION

Our study provides a comprehensive description of TME characteristics in CRC and demonstrates that the TMEscore is a reliable prognostic biomarker and predictive indicator for patients with CRC undergoing immunotherapy.

Titanium Particles Modulate Lymphocyte and Macrophage Polarization in Peri-Implant Gingival Tissues

Titanium dental implants are one of the modalities to replace missing teeth. The release of titanium particles from the implant's surface may modulate the immune cells, resulting in implant failure. However, little is known about the immune microenvironment that plays a role in peri-implant inflammation as a consequence of titanium particles. In this study, the peri-implant gingival tissues were collected from patients with failed implants, successful implants and no implants, and then a whole transcriptome analysis was performed. The gene set enrichment analysis confirmed that macrophage M1/M2 polarization and lymphocyte proliferation were differentially expressed between the study groups. The functional clustering and pathway analysis of the differentially expressed genes between the failed implants and successful implants versus no implants revealed that the immune response pathways were the most common in both comparisons, implying the critical role of infiltrating immune cells in the peri-implant tissues. The H&E and IHC staining confirmed the presence of titanium particles and immune cells in the tissue samples, with an increase in the infiltration of lymphocytes and macrophages in the failed implant samples. The in vitro validation showed a significant increase in the level of IL-1β, IL-8 and IL-18 expression by macrophages. Our findings showed evidence that titanium particles modulate lymphocyte and macrophage polarization in peri-implant gingival tissues, which can help in the understanding of the imbalance in osteoblast-osteoclast activity and failure of dental implant osseointegration.

Predictive network analysis identifies JMJD6 and other potential key drivers in Alzheimer's disease

Despite decades of genetic studies on late-onset Alzheimer's disease, the underlying molecular mechanisms remain unclear. To better comprehend its complex etiology, we use an integrative approach to build robust predictive (causal) network models using two large human multi-omics datasets. We delineate bulk-tissue gene expression into single cell-type gene expression and integrate clinical and pathologic traits, single nucleotide variation, and deconvoluted gene expression for the construction of cell type-specific predictive network models. Here, we focus on neuron-specific network models and prioritize 19 predicted key drivers modulating Alzheimer's pathology, which we then validate by knockdown in human induced pluripotent stem cell-derived neurons. We find that neuronal knockdown of 10 of the 19 targets significantly modulates levels of amyloid-beta and/or phosphorylated tau peptides, most notably JMJD6. We also confirm our network structure by RNA sequencing in the neurons following knockdown of each of the 10 targets, which additionally predicts that they are upstream regulators of REST and VGF. Our work thus identifies robust neuronal key drivers of the Alzheimer's-associated network state which may represent therapeutic targets with relevance to both amyloid and tau pathology in Alzheimer's disease.

The Impact of YRNAs on HNSCC and HPV Infection

HPV infection is one of the most important risk factors for head and neck squamous cell carcinoma among younger patients. YRNAs are short non-coding RNAs involved in DNA replication. YRNAs have been found to be dysregulated in many cancers, including head and neck squamous cell carcinoma (HNSCC). In this study, we investigated the role of YRNAs in HPV-positive HNSCC using publicly available gene expression datasets from HNSCC tissue, where expression patterns of YRNAs in HPV(+) and HPV(−) HNSCC samples significantly differed. Additionally, HNSCC cell lines were treated with YRNA1-overexpressing plasmid and RNA derived from these cell lines was used to perform a NGS analysis. Additionally, a deconvolution analysis was performed to determine YRNA1’s impact on immune cells. YRNA expression levels varied according to cancer pathological and clinical stages, and correlated with more aggressive subtypes. YRNAs were mostly associated with more advanced cancer stages in the HPV(+) group, and YRNA3 and YRNA1 expression levels were found to be correlated with more advanced clinical stages despite HPV infection status, showing that they may function as potential biomarkers of more advanced stages of the disease. YRNA5 was associated with less-advanced cancer stages in the HPV(−) group. Overall survival and progression-free survival analyses showed opposite results between the HPV groups. The expression of YRNAs, especially YRNA1, correlated with a vast number of proteins and cellular processes associated with viral infections and immunologic responses to viruses. HNSCC-derived cell lines overexpressing YRNA1 were then used to determine the correlation of YRNA1 and the expression of genes associated with HPV infections. Taken together, our results highlight the potential of YRNAs as possible HNSCC biomarkers and new molecular targets.

Deciphering Tumour Microenvironment of Liver Cancer through Deconvolution of Bulk RNA-Seq Data with Single-Cell Atlas

Liver cancers give rise to a heavy burden on healthcare worldwide. Understanding the tumour microenvironment (TME) underpins the development of precision therapy. Single-cell RNA sequencing (scRNA-seq) technology has generated high-quality cell atlases of the TME, but its wider application faces enormous costs for various clinical circumstances. Fortunately, a variety of deconvolution algorithms can instead repurpose bulk RNA-seq data, alleviating the need for generating scRNA-seq datasets. In this study, we reviewed major public omics databases for relevance in this study and utilised eight RNA-seqs and one microarray dataset from clinical studies. To decipher the TME of liver cancer, we estimated the fractions of liver cell components by deconvoluting the samples with Cibersortx using three reference scRNA-seq atlases. We also confirmed that Cibersortx can accurately deconvolute cell types/subtypes of interest. Compared with non-tumorous liver, liver cancers showed multiple decreased cell types forming normal liver microarchitecture, as well as elevated cell types involved in fibrogenesis, abnormal angiogenesis, and disturbed immune responses. Survival analysis shows that the fractions of five cell types/subtypes significantly correlated with patient outcomes, indicating potential therapeutic targets. Therefore, deconvolution of bulk RNA-seq data with scRNA-seq atlas references can be a useful tool to help understand the TME.

Toward Precision Radiotherapy: A Nonlinear Optimization Framework and an Accelerated Machine Learning Algorithm for the Deconvolution of Tumor-Infiltrating Immune Cells

Tumor-infiltrating immune cells (TIICs) form a critical part of the ecosystem surrounding a cancerous tumor. Recent advances in radiobiology have shown that, in addition to damaging cancerous cells, radiotherapy drives the upregulation of immunosuppressive and immunostimulatory TIICs, which in turn impacts treatment response. Quantifying TIICs in tumor samples could form an important predictive biomarker guiding patient stratification and the design of radiotherapy regimens and combined immune-radiation treatments. As a result of several limitations associated with experimental methods for quantifying TIICs and the availability of extensive gene sequencing data, deconvolution-based computational methods have appeared as a suitable alternative for quantifying TIICs. Accordingly, we introduce and discuss a nonlinear regression approach (remarkably different from the traditional linear modeling approach of current deconvolution-based methods) and a machine learning algorithm for approximating the solution of the resulting constrained optimization problem. This way, the deconvolution problem is treated naturally, given that the gene expression levels of pure and heterogenous samples do not have a strictly linear relationship. When applied across transcriptomics datasets, our approach, which also allows the coupling of different loss functions, yields results that closely match ground-truth values from experimental methods and exhibits superior performance over popular deconvolution-based methods.

Niche deconvolution of the glioblastoma proteome reveals a distinct infiltrative phenotype within the proneural transcriptomic subgroup

Glioblastoma is often subdivided into three transcriptional subtypes (classical, proneural, mesenchymal) based on bulk RNA signatures that correlate with distinct genetic and clinical features. Potential cellular-level differences of these subgroups, such as the relative proportions of glioblastoma’s hallmark histopathologic features (e.g. brain infiltration, microvascular proliferation), may provide insight into their distinct phenotypes but are, however, not well understood. Here we leverage machine learning and reference proteomic profiles derived from micro-dissected samples of these major histomorphologic glioblastoma features to deconvolute and estimate niche proportions in an independent proteogenomically-characterized cohort. This approach revealed a strong association of the proneural transcriptional subtype with a diffusely infiltrating phenotype. Similarly, enrichment of a microvascular proliferation proteomic signature was seen within the mesenchymal subtype. This study is the first to link differences in the cellular pathology signatures and transcriptional profiles of glioblastoma, providing potential new insights into the genetic drivers and poor treatment response of specific subsets of glioblastomas.

Luminal androgen receptor breast cancer subtype and investigation of the microenvironment and neoadjuvant chemotherapy response

Triple-negative breast cancer (TNBC) is the most aggressive breast cancer subtype with low overall survival rates and high molecular heterogeneity; therefore, few targeted therapies are available. The luminal androgen receptor (LAR) is the most consistently identified TNBC subtype, but the clinical utility has yet to be established. Here, we constructed a novel genomic classifier, LAR-Sig, that distinguishes the LAR subtype from other TNBC subtypes and provide evidence that it is a clinically distinct disease. A meta-analysis of seven TNBC datasets (n = 1086 samples) from neoadjuvant clinical trials demonstrated that LAR patients have significantly reduced response (pCR) rates than non-LAR TNBC patients (odds ratio = 2.11, 95% CI: 1.33, 2.89). Moreover, deconvolution of the tumor microenvironment confirmed an enrichment of luminal epithelium corresponding with a decrease in basal and myoepithelium in LAR TNBC tumors. Increased immunosuppression in LAR patients may lead to a decreased presence of cycling T-cells and plasma cells. While, an increased presence of myofibroblast-like cancer-associated cells may impede drug delivery and treatment. In summary, the lower levels of tumor infiltrating lymphocytes (TILs), reduced immune activity in the micro-environment, and lower pCR rates after NAC, suggest that new therapeutic strategies for the LAR TNBC subtype need to be developed.

Hyperactive neutrophil chemotaxis contributes to anti-tumor necrosis factor-α treatment resistance in inflammatory bowel disease

BACKGROUND AND AIM

Anti-tumor necrosis factor-α (anti-TNF-α) agents have been used for inflammatory bowel disease; however, it has up to 30% nonresponse rate. Identifying molecular pathways and finding reliable diagnostic biomarkers for patient response to anti-TNF-α treatment are needed.

METHODS

Publicly available transcriptomic data from inflammatory bowel disease patients receiving anti-TNF-α therapy were systemically collected and integrated. In silico flow cytometry approaches and Metascape were applied to evaluate immune cell populations and to perform gene enrichment analysis, respectively. Genes identified within enrichment pathways validated in neutrophils were tracked in an anti-TNF-α-treated animal model (with lipopolysaccharide-induced inflammation). The receiver operating characteristic curve was applied to all genes to identify the best prediction biomarkers.

RESULTS

A total of 449 samples were retrieved from control, baseline, and after primary anti-TNF-α therapy or placebo. No statistically significant differences were observed between anti-TNF-α treatment responders and nonresponders at baseline in immune microenvironment scores. Neutrophil, endothelial cell, and B-cell populations were higher in baseline nonresponders, and chemotaxis pathways may contribute to the treatment resistance. Genes related to chemotaxis pathways were significantly upregulated in lipopolysaccharide-induced neutrophils, but no statistically significant changes were observed in neutrophils treated with anti-TNF-α. Interleukin 13 receptor subunit alpha 2 (IL13RA2) is the best predictor (receiver operating characteristic curve: 80.7%, 95% confidence interval: 73.8-87.5%), with a sensitivity of 68.13% and specificity of 84.93%, and significantly higher in nonresponders compared with responders (P < 0.0001).

CONCLUSIONS

Hyperactive neutrophil chemotaxis influences responses to anti-TNF-α treatment, and IL13RA2 is a potential biomarker to predict anti-TNF-α treatment response.

AdRoit is an accurate and robust method to infer complex transcriptome composition

Bulk RNA sequencing provides the opportunity to understand biology at the whole transcriptome level without the prohibitive cost of single cell profiling. Advances in spatial transcriptomics enable to dissect tissue organization and function by genome-wide gene expressions. However, the readout of both technologies is the overall gene expression across potentially many cell types without directly providing the information of cell type constitution. Although several in-silico approaches have been proposed to deconvolute RNA-Seq data composed of multiple cell types, many suffer a deterioration of performance in complex tissues. Here we present AdRoit, an accurate and robust method to infer the cell composition from transcriptome data of mixed cell types. AdRoit uses gene expression profiles obtained from single cell RNA sequencing as a reference. It employs an adaptive learning approach to alleviate the sequencing technique difference between the single cell and the bulk (or spatial) transcriptome data, enhancing cross-platform readout comparability. Our systematic benchmarking and applications, which include deconvoluting complex mixtures that encompass 30 cell types, demonstrate its preferable sensitivity and specificity compared to many existing methods as well as its utilities. In addition, AdRoit is computationally efficient and runs orders of magnitude faster than most methods.

Comparative Metabolomics Reveals the Microenvironment of Common T-Helper Cells and Differential Immune Cells Linked to Unique Periapical Lesions

Periapical abscesses, radicular cysts, and periapical granulomas are the most frequently identified pathological lesions in the alveolar bone. While little is known about the initiation and progression of these conditions, the metabolic environment and the related immunological behaviors were examined for the first time to model the development of each pathological condition. Metabolites were extracted from each lesion and profiled using gas chromatography-mass spectrometry in comparison with healthy pulp tissue. The metabolites were clustered and linked to their related immune cell fractions. Clusters I and J in the periapical abscess upregulated the expression of MMP-9, IL-8, CYP4F3, and VEGF, while clusters L and M were related to lipophagy and apoptosis in radicular cyst, and cluster P in periapical granuloma, which contains L-(+)-lactic acid and ethylene glycol, was related to granuloma formation. Oleic acid, 17-octadecynoic acid, 1-nonadecene, and L-(+)-lactic acid were significantly the highest unique metabolites in healthy pulp tissue, periapical abscess, radicular cyst, and periapical granuloma, respectively. The correlated enriched metabolic pathways were identified, and the related active genes were predicted. Glutamatergic synapse (16-20),-hydroxyeicosatetraenoic acids, lipophagy, and retinoid X receptor coupled with vitamin D receptor were the most significantly enriched pathways in healthy control, abscess, cyst, and granuloma, respectively. Compared with the healthy control, significant upregulation in the gene expression of CYP4F3, VEGF, IL-8, TLR2 (P < 0.0001), and MMP-9 (P < 0.001) was found in the abscesses. While IL-12A was significantly upregulated in cysts (P < 0.01), IL-17A represents the highest significantly upregulated gene in granulomas (P < 0.0001). From the predicted active genes, CIBERSORT suggested the presence of natural killer cells, dendritic cells, pro-inflammatory M1 macrophages, and anti-inflammatory M2 macrophages in different proportions. In addition, the single nucleotide polymorphisms related to IL-10, IL-12A, and IL-17D genes were shown to be associated with periapical lesions and other oral lesions. Collectively, the unique metabolism and related immune response shape up an environment that initiates and maintains the existence and progression of these oral lesions, suggesting an important role in diagnosis and effective targeted therapy.

Progress and challenge for computational quantification of tissue immune cells

Tissue immune cells have long been recognized as important regulators for the maintenance of balance in the body system. Quantification of the abundance of different immune cells will provide enhanced understanding of the correlation between immune cells and normal or abnormal situations. Currently, computational methods to predict tissue immune cell compositions from bulk transcriptomes have been largely developed. Therefore, summarizing the advantages and disadvantages is appropriate. In addition, an examination of the challenges and possible solutions for these computational models will assist the development of this field. The common hypothesis of these models is that the expression of signature genes for immune cell types might represent the proportion of immune cells that contribute to the tissue transcriptome. In general, we grouped all reported tools into three groups, including reference-free, reference-based scoring and reference-based deconvolution methods. In this review, a summary of all the currently reported computational immune cell quantification tools and their applications, limitations, and perspectives are presented. Furthermore, some critical problems are found that have limited the performance and application of these models, including inadequate immune cell type, the collinearity problem, the impact of the tissue environment on the immune cell expression level, and the deficiency of standard datasets for model validation. To address these issues, tissue specific training datasets that include all known immune cells, a hierarchical computational framework, and benchmark datasets including both tissue expression profiles and the abundances of all the immune cells are proposed to further promote the development of this field.