Systematic analysis of mutation distribution in three dimensional protein structures identifies cancer driver genes

Role of RNA Splicing Mutations in Diffuse Large B Cell Lymphoma

Ribonucleic acid splicing is a crucial process to create a mature mRNA molecule by removing introns and ligating exons. This is a highly regulated process, but any alteration in splicing factors, splicing sites, or auxiliary components affects the final products of the gene. In diffuse large B-cell lymphoma, splicing mutations such as mutant splice sites, aberrant alternative splicing, exon skipping, and intron retention are detected. The alteration affects tumor suppression, DNA repair, cell cycle, cell differentiation, cell proliferation, and apoptosis. As a result, malignant transformation, cancer progression, and metastasis occurred in B cells at the germinal center. B-cell lymphoma 7 protein family member A (BCL7A), cluster of differentiation 79B (CD79B), myeloid differentiation primary response gene 88 (MYD88), tumor protein P53 (TP53), signal transducer and activator of transcription (STAT), serum- and glucose-regulated kinase 1 (SGK1), Pou class 2 associating factor 1 (POU2AF1), and neurogenic locus notch homolog protein 1 (NOTCH) are the most common genes affected by splicing mutations in diffuse large B cell lymphoma.

Faltwerk: a library for spatial exploratory data analysis of protein structures

Summary

Proteins are fundamental building blocks of life and are investigated in a broad range of scientific fields, especially in the context of recent progress using in silico structure prediction models and the surge of resulting protein structures in public databases. However, exploratory data analysis of these proteins can be slow because of the need for several methods, ranging from geometric and spatial analysis to visualization. The Python library faltwerk provides an integrated toolkit to perform explorative work with rapid feedback. This toolkit includes support for protein complexes, spatial analysis (point density or spatial autocorrelation), ligand binding site prediction and an intuitive visualization interface based on the grammar of graphics.

Availability and implementation

faltwerk is distributed under the permissive BSD-3 open source license. Source code and documentation, including an extensive common-use case tutorial, can be found at github.com/phiweger/faltwerk; binaries are available from the pypi repository.

A New View of Activating Mutations in Cancer

A vast effort has been invested in the identification of driver mutations of cancer. However, recent studies and observations call into question whether the activating mutations or the signal strength are the major determinant of tumor development. The data argue that signal strength determines cell fate, not the mutation that initiated it. In addition to activating mutations, factors that can impact signaling strength include (i) homeostatic mechanisms that can block or enhance the signal, (ii) the types and locations of additional mutations, and (iii) the expression levels of specific isoforms of genes and regulators of proteins in the pathway. Because signal levels are largely decided by chromatin structure, they vary across cell types, states, and time windows. A strong activating mutation can be restricted by low expression, whereas a weaker mutation can be strengthened by high expression. Strong signals can be associated with cell proliferation, but too strong a signal may result in oncogene-induced senescence. Beyond cancer, moderate signal strength in embryonic neural cells may be associated with neurodevelopmental disorders, and moderate signals in aging may be associated with neurodegenerative diseases, like Alzheimer's disease. The challenge for improving patient outcomes therefore lies in determining signaling thresholds and predicting signal strength.

HNF1A POU Domain Mutations Found in Japanese Liver Cancer Patients Cause Downregulation of HNF4A Promoter Activity with Possible Disruption in Transcription Networks

Hepatocyte nuclear factor 1A (HNF1A) is the master regulator of liver homeostasis and organogenesis and regulates many aspects of hepatocyte functions. It acts as a tumor suppressor in the liver, evidenced by the increased proliferation in HNF1A knockout (KO) hepatocytes. Hence, we postulated that any loss-of-function variation in the gene structure or composition (mutation) could trigger dysfunction, including disrupted transcriptional networks in liver cells. From the International Cancer Genome Consortium (ICGC) database of cancer genomes, we identified several HNF1A mutations located in the functional Pit-Oct-Unc (POU) domain. In our biochemical analysis, we found that the HNF1A POU-domain mutations Y122C, R229Q and V259F suppressed HNF4A promoter activity and disrupted the binding of HNF1A to its target HNF4A promoter without any effect on the nuclear localization. Our results suggest that the decreased transcriptional activity of HNF1A mutants is due to impaired DNA binding. Through structural simulation analysis, we found that a V259F mutation was likely to affect DNA interaction by inducing large conformational changes in the N-terminal region of HNF1A. The results suggest that POU-domain mutations of HNF1A downregulate HNF4A gene expression. Therefore, to mimic the HNF1A mutation phenotype in transcription networks, we performed siRNA-mediated knockdown (KD) of HNF4A. Through RNA-Seq data analysis for the HNF4A KD, we found 748 differentially expressed genes (DEGs), of which 311 genes were downregulated (e.g., HNF1A, ApoB and SOAT2) and 437 genes were upregulated. Kyoto Encyclopedia of Genes and Genomes (KEGG) mapping revealed that the DEGs were involved in several signaling pathways (e.g., lipid and cholesterol metabolic pathways). Protein–protein network analysis suggested that the downregulated genes were related to lipid and cholesterol metabolism pathways, which are implicated in hepatocellular carcinoma (HCC) development. Our study demonstrates that mutations of HNF1A in the POU domain result in the downregulation of HNF1A target genes, including HNF4A, and this may trigger HCC development through the disruption of HNF4A–HNF1A transcriptional networks.

Novel ratio-metric features enable the identification of new driver genes across cancer types

An emergent area of cancer genomics is the identification of driver genes. Driver genes confer a selective growth advantage to the cell. While several driver genes have been discovered, many remain undiscovered, especially those mutated at a low frequency across samples. This study defines new features and builds a pan-cancer model, cTaG, to identify new driver genes. The features capture the functional impact of the mutations as well as their recurrence across samples, which helps build a model unbiased to genes with low frequency. The model classifies genes into the functional categories of driver genes, tumour suppressor genes (TSGs) and oncogenes (OGs), having distinct mutation type profiles. We overcome overfitting and show that certain mutation types, such as nonsense mutations, are more important for classification. Further, cTaG was employed to identify tissue-specific driver genes. Some known cancer driver genes predicted by cTaG as TSGs with high probability are ARID1A, TP53, and RB1. In addition to these known genes, potential driver genes predicted are CD36, ZNF750 and ARHGAP35 as TSGs and TAB3 as an oncogene. Overall, our approach surmounts the issue of low recall and bias towards genes with high mutation rates and predicts potential new driver genes for further experimental screening. cTaG is available at https://github.com/RamanLab/cTaG .

Enhanced interpretation of 935 hotspot and non-hotspot RAS variants using evidence-based structural bioinformatics

In the current study, we report computational scores for advancing genomic interpretation of disease-associated genomic variation in members of the RAS family of genes. For this purpose, we applied 31 sequence- and 3D structure-based computational scores, chosen by their breadth of biophysical properties. We parametrized our data by assembling a numerically homogenized experimentally-derived dataset, which when use in our calculations reveal that computational scores using 3D structure highly correlate with experimental measures (e.g., GAP-mediated hydrolysis R_Spearman = 0.80 and RAF affinity R_spearman = 0.82), while sequence-based scores are discordant with this data. Performing all-against-all comparisons, we applied this parametrized modeling approach to the study of 935 RAS variants from 7 RAS genes, which led us to identify 4 groups of mutations according to distinct biochemical scores within each group. Each group was comprised of hotspot and non-hotspot KRAS variants, indicating that poorly characterized variants could functionally behave like pathogenic mutations. Combining computational scores using dimensionality reduction indicated that changes to local unfolding propensity associate with changes in enzyme activity by genomic variants. Hence, our systematic approach, combining methodologies from both clinical genomics and 3D structural bioinformatics, represents an expansion for interpreting genomic data, provides information of mechanistic value, and that is transferable to other proteins.

DriverML: a machine learning algorithm for identifying driver genes in cancer sequencing studies

Although rapid progress has been made in computational approaches for prioritizing cancer driver genes, research is far from achieving the ultimate goal of discovering a complete catalog of genes truly associated with cancer. Driver gene lists predicted from these computational tools lack consistency and are prone to false positives. Here, we developed an approach (DriverML) integrating Rao’s score test and supervised machine learning to identify cancer driver genes. The weight parameters in the score statistics quantified the functional impacts of mutations on the protein. To obtain optimized weight parameters, the score statistics of prior driver genes were maximized on pan-cancer training data. We conducted rigorous and unbiased benchmark analysis and comparisons of DriverML with 20 other existing tools in 31 independent datasets from The Cancer Genome Atlas (TCGA). Our comprehensive evaluations demonstrated that DriverML was robust and powerful among various datasets and outperformed the other tools with a better balance of precision and sensitivity. In vitro cell-based assays further proved the validity of the DriverML prediction of novel driver genes. In summary, DriverML uses an innovative, machine learning-based approach to prioritize cancer driver genes and provides dramatic improvements over currently existing methods. Its source code is available at https://github.com/HelloYiHan/DriverML.

eVIDENCE: a practical variant filtering for low-frequency variants detection in cell-free DNA

Plasma cell-free DNA (cfDNA) testing plays an increasingly important role in precision medicine for cancer. However, circulating cell-free tumor DNA (ctDNA) is highly diluted by cfDNA from non-cancer cells, complicating ctDNA detection and analysis. To identify low-frequency variants, we developed a program, eVIDENCE, which is a workflow for filtering candidate variants detected by using the ThruPLEX tag-seq (Takara Bio), a commercially-available molecular barcoding kit. We analyzed 27 cfDNA samples from hepatocellular carcinoma patients. Sequencing libraries were constructed and hybridized to our custom panel targeting about 80 genes. An initial variant calling identified 36,500 single nucleotide variants (SNVs) and 9,300 insertions and deletions (indels) across the 27 samples, but the number was much greater than expected when compared with previous cancer genome studies. eVIDENCE was applied to the candidate variants and finally 70 SNVs and 7 indels remained. Of the 77 variants, 49 (63.6%) showed VAF of < 1% (0.20–0.98%). Twenty-five variants were selected in an unbiased manner and all were successfully validated, suggesting that eVIDENCE can identify variants with VAF of ≥ 0.2%. Additionally, this study is the first to detect hepatitis B virus integration sites and genomic rearrangements in the TERT region from cfDNA of HCC patients. We consider that our method can be applied in the examination of cfDNA from other types of malignancies using specific custom gene panels and will contribute to comprehensive ctDNA analysis.

Functional characterization of 3D protein structures informed by human genetic diversity

Sequence variation data of the human proteome can be used to analyze 3D protein structures to derive functional insights. We used genetic variant data from nearly 140,000 individuals to analyze 3D positional conservation in 4,715 proteins and 3,951 homology models using 860,292 missense and 465,886 synonymous variants. Sixty percent of protein structures harbor at least one intolerant 3D site as defined by significant depletion of observed over expected missense variation. Structural intolerance data correlated with deep mutational scanning functional readouts for PPARG, MAPK1/ERK2, UBE2I, SUMO1, PTEN, CALM1, CALM2, and TPK1 and with shallow mutagenesis data for 1,026 proteins. The 3D structural intolerance analysis revealed different features for ligand binding pockets and orthosteric and allosteric sites. Large-scale data on human genetic variation support a definition of functional 3D sites proteome-wide.

Clinical utility of androgen receptor gene aberrations in circulating cell-free DNA as a biomarker for treatment of castration-resistant prostate cancer

The therapeutic landscape of castration-resistant prostate cancer (CRPC) has rapidly expanded. There is a need to develop noninvasive biomarkers to guide treatment. We established a highly sensitive method for analyzing androgen receptor gene (AR) copy numbers (CN) and mutations in plasma circulating cell-free DNA (cfDNA) and evaluated the AR statuses of patients with CRPC. AR amplification was detectable in VCaP cell line (AR amplified) genomic DNA (gDNA) diluted to 1.0% by digital PCR (dPCR). AR mutation were detectable in LNCaP cell line (AR T878A mutated) gDNA diluted to 0.1% and 1.0% by dPCR and target sequencing, respectively. Next, we analyzed AR status in cfDNA from 102 patients. AR amplification and mutations were detected in 47 and 25 patients, respectively. As a biomarker, AR aberrations in pretreatment cfDNA were associated with poor response to abiraterone, but not enzalutamide. In serial cfDNA analysis from 41 patients, most AR aberrations at baseline diminished with effective treatments, whereas in some patients with disease progression, AR amplification or mutations emerged. The analysis of AR in cfDNA is feasible and informative procedure for treating patients with CRPC. cfDNA may become a useful biomarker for precision medicine in CRPC.

A CATH domain functional family based approach to identify putative cancer driver genes and driver mutations

Tumour sequencing identifies highly recurrent point mutations in cancer driver genes, but rare functional mutations are hard to distinguish from large numbers of passengers. We developed a novel computational platform applying a multi-modal approach to filter out passengers and more robustly identify putative driver genes. The primary filter identifies enrichment of cancer mutations in CATH functional families (CATH-FunFams) – structurally and functionally coherent sets of evolutionary related domains. Using structural representatives from CATH-FunFams, we subsequently seek enrichment of mutations in 3D and show that these mutation clusters have a very significant tendency to lie close to known functional sites or conserved sites predicted using CATH-FunFams. Our third filter identifies enrichment of putative driver genes in functionally coherent protein network modules confirmed by literature analysis to be cancer associated. Our approach is complementary to other domain enrichment approaches exploiting Pfam families, but benefits from more functionally coherent groupings of domains. Using a set of mutations from 22 cancers we detect 151 putative cancer drivers, of which 79 are not listed in cancer resources and include recently validated cancer associated genes EPHA7, DCC netrin-1 receptor and zinc-finger protein ZNF479.

Analyzing Biomolecular Ensembles

Several techniques are available to generate conformational ensembles of proteins and other biomolecules either experimentally or computationally. These methods produce a large amount of data that need to be analyzed to identify structure-dynamics-function relationship. In this chapter, we will cover different tools to unveil the information hidden in conformational ensemble data and to guide toward the rationalization of the data. We included routinely used approaches such as dimensionality reduction, as well as new methods inspired by high-order statistics and graph theory.

Defining HIV-1 Envelope N-Glycan Microdomains through Site-Specific Heterogeneity Profiles

The HIV-1 envelope (Env) glycans shield the surface of Env from the immune system and form integral interactions important for a functional Env. To understand how individual N-glycosylation sites (NGS) coordinate to form a dynamic shield and evade the immune system through mutations, we tracked 20 NGS in Env from HIV-transmitted/founder (T/F) and immune escape variants and their mutants involving the N262 glycan. NGS were profiled in a site-specific manner using a high-resolution mass spectrometry (MS)-based workflow. Using this site-specific quantitative heterogeneity profiling, we empirically characterized the interdependent NGS of a microdomain in the high-mannose patch (HMP). The changes (shifts) in NGS heterogeneity between the T/F and immune escape variants defined a range of NGS that we further probed for exclusive combinations of sequons in the HMP microdomain using the Los Alamos National Laboratory HIV sequence database. The resultant sequon combinations, including the highly conserved NGS N262, N448, and N301, created an immune escape map of the conserved and variable sequons in the HMP microdomain. This report provides details on how some clustered NGS form microdomains that can be identified and tracked across Env variants. These microdomains have a limited number of N-glycan-sequon combinations that may allow the anticipation of immune escape variants.IMPORTANCE The Env protein of HIV is highly glycosylated, and the sites of glycosylation can change as the virus mutates during immune evasion. Due to these changes, the glycan location and heterogeneity of surrounding N-glycosylation sites can be altered, resulting in exposure of different glycan or proteoglycan surfaces while still producing a viable HIV variant. These changes present a need for vaccine developers to identify Env variants with epitopes most likely to induce durable protective responses. Here we describe a means of anticipating HIV-1 immune evasion by dividing Env into N-glycan microdomains that have a limited number of N-glycan sequon combinations.

Survival associated alternative splicing events in diffuse large B-cell lymphoma

Growing evidence has revealed that the initiation of various malignancies is closely associated with alternative splicing (AS) events in certain key oncogenes. However, in diffuse large B-cell lymphoma (DLBCL), there is still a great deal to learn about AS variants. In this study, 33,724 AS variant profiles were obtained from 16,278 genes in 48 DLBCL cases. A total of 10 AS variants were identified as overall survival (OS)- related events via multivariate Cox regression analysis. Notably, alternative donor (AD) sites in AS events in the low-risk group showed a significantly better outcome in DLBCL patients than in the high-risk group (P=0.0002). The area under the curve (AUC) of the receiver-operator characteristic curve (ROC) for ADs in DLBCL was 0.746. Furthermore, 66 related splicing factors were obtained to investigate their potential correlations with AS events. Factors SF1, HNRNPC, HNRNPD, and HNRNPH3 were significantly involved in different OS-related AS variants. Collectively, we constructed valuable prognostic predictors for DLBCL patients and mapped novel splicing networks for further investigation of the underlying mechanisms related to AS variants in DLBCLs.

Loss-of-function mutations in Zn-finger DNA-binding domain of HNF4A cause aberrant transcriptional regulation in liver cancer

Hepatocyte nuclear factors (HNF) are transcription factors that crucially regulate cell-specific gene expression in many tissues, including the liver. Of these factors, HNF4A acts both as a master regulator of liver organogenesis and a tumor suppressor in the liver. In our whole genome sequencing analysis, we found seven somatic mutations (three Zn-finger mutations, three deletion mutants, and one intron mutation) of HNF4A in liver cancers. Interestingly, three out of seven mutations were clustered in its Zn-finger DNA-binding domain; G79 and F83 are positioned in the DNA recognition helix and the sidechain of M125 is sticking into the core of domain. These mutations are likely to affect DNA interaction from a structural point of view. We then generated these mutants and performed in-vitro promoter assays as well as DNA binding assays. These three mutations reduced HNF4 transcriptional activity at promoter sites of HNF4A-target genes. Expectedly, this decrease in transcriptional activity was associated with a change in DNA binding. RNA-Seq analysis observed a strong correlation between HNF4A expression and expression of its target genes, ApoB and HNF1A, in liver cancers. Since knockdown of HNF4A caused a reduction in ApoB and HNF1A expression, possibly loss of HNF4 reduces the expression of these genes and subsequently tumor growth is triggered. Therefore, we propose that HNF4A mutations G79C, F83C, and M125I are functional mutations found in liver cancers and that loss of HNF4A function, through its mutation, leads to a reduction in HNF1A and ApoB gene expression with a concomitant increased risk of liver tumorigenesis.

Structural dynamics is a determinant of the functional significance of missense variants

Discrimination of clinically relevant mutations from neutral mutations is of paramount importance in precision medicine and pharmacogenomics. Our study shows that current computational predictions of pathogenicity, mostly based on analysis of sequence conservation, may be improved by considering the changes in the structural dynamics of the protein due to point mutations. We introduce and demonstrate the utility of a classifier that takes advantage of efficient evaluation of structural dynamics by elastic network models.

Accurate evaluation of the effect of point mutations on protein function is essential to assessing the genesis and prognosis of many inherited diseases and cancer types. Currently, a wealth of computational tools has been developed for pathogenicity prediction. Two major types of data are used to this aim: sequence conservation/evolution and structural properties. Here, we demonstrate in a systematic way that another determinant of the functional impact of missense variants is the protein’s structural dynamics. Measurable improvement is shown in pathogenicity prediction by taking into consideration the dynamical context and implications of the mutation. Our study suggests that the class of dynamics descriptors introduced here may be used in conjunction with existing features to not only increase the prediction accuracy of the impact of variants on biological function, but also gain insight into the physical basis of the effect of missense variants.

Cancer genetics meets biomolecular mechanism-bridging an age-old gulf

Increasingly available genomic sequencing data are exploited to identify genes and variants contributing to diseases, particularly cancer. Traditionally, methods to find such variants have relied heavily on allele frequency and/or familial history, often neglecting to consider any mechanistic understanding of their functional consequences. Thus, while the set of known cancer-related genes has increased, for many, their mechanistic role in the disease is not completely understood. This issue highlights a wide gap between the disciplines of genetics, which largely aims to correlate genetic events with phenotype, and molecular biology, which ultimately aims at a mechanistic understanding of biological processes. Fortunately, new methods and several systematic studies have proved illuminating for many disease genes and variants by integrating sequencing with mechanistic data, including biomolecular structures and interactions. These have provided new interpretations for known mutations and suggested new disease-relevant variants and genes. Here, we review these approaches and discuss particular examples where these have had a profound impact on the understanding of human cancers.