MOGSA: Integrative Single Sample Gene-set Analysis of Multiple Omics Data.

Gene-set analysis (GSA) summarizes individual molecular measurements to more interpretable pathways or gene-sets and has become an indispensable step in the interpretation of large-scale omics data. However, GSA methods are limited to the analysis of single omics data. Here, we introduce a new computation method termed multi-omics gene-set analysis (MOGSA), a multivariate single sample gene-set analysis method that integrates multiple experimental and molecular data types measured over the same set of samples. The method learns a low dimensional representation of most variant correlated features (genes, proteins, etc.) across multiple omics data sets, transforms the features onto the same scale and calculates an integrated gene-set score from the most informative features in each data type. MOGSA does not require filtering data to the intersection of features (gene IDs), therefore, all molecular features, including those that lack annotation may be included in the analysis. Using simulated data, we demonstrate that integrating multiple diverse sources of molecular data increases the power to discover subtle changes in gene-sets and may reduce the impact of unreliable information in any single data type. Using real experimental data, we demonstrate three use-cases of MOGSA. First, we show how to remove a source of noise (technical or biological) in integrative MOGSA of NCI60 transcriptome and proteome data. Second, we apply MOGSA to discover similarities and differences in mRNA, protein and phosphorylation profiles of a small study of stem cell lines and assess the influence of each data type or feature on the total gene-set score. Finally, we apply MOGSA to cluster analysis and show that three molecular subtypes are robustly discovered when copy number variation and mRNA data of 308 bladder cancers from The Cancer Genome Atlas are integrated using MOGSA. MOGSA is available in the Bioconductor R package "mogsa."

Quantitative proteome profiling of human myoma and myometrium tissue reveals kinase expression signatures with potential for therapeutic intervention.

Uterine leiomyomas are benign tumors affecting a large proportion of the female population. Despite the very high prevalence, the molecular basis for understanding the onset and development of the disease are still poorly understood. In this study, we profiled the proteomes and kinomes of leiomyoma as well as myometrium samples from patients to a depth of >7000 proteins including 200 kinases. Statistical analysis identified a number of molecular signatures distinguishing healthy from diseased tissue. Among these, nine kinases (ADCK4, CDK5, CSNK2B, DDR1, EPHB1, MAP2K2, PRKCB, PRKG1, and RPS6KA5) representing a number of cellular signaling pathways showed particularly strong discrimination potential. Preliminary statistical analysis by receiver operator characteristics plots revealed very good performance for individual kinases (area under the curve, AUC of 0.70-0.94) as well as binary combinations thereof (AUC 0.70-1.00) that might be used to assess the activity of signaling pathways in myomas. Of note, the receptor tyrosine kinase DDR1 holds future potential as a drug target owing to its strong links to collagen signaling and the excessive formation of extracellular matrix typical for leiomyomas in humans.

A multivariate approach to the integration of multi-omics datasets.

BACKGROUND: To leverage the potential of multi-omics studies, exploratory data analysis methods that provide systematic integration and comparison of multiple layers of omics information are required. We describe multiple co-inertia analysis (MCIA), an exploratory data analysis method that identifies co-relationships between multiple high dimensional datasets. Based on a covariance optimization criterion, MCIA simultaneously projects several datasets into the same dimensional space, transforming diverse sets of features onto the same scale, to extract the most variant from each dataset and facilitate biological interpretation and pathway analysis. RESULTS: We demonstrate integration of multiple layers of information using MCIA, applied to two typical "omics" research scenarios. The integration of transcriptome and proteome profiles of cells in the NCI-60 cancer cell line panel revealed distinct, complementary features, which together increased the coverage and power of pathway analysis. Our analysis highlighted the importance of the leukemia extravasation signaling pathway in leukemia that was not highly ranked in the analysis of any individual dataset. Secondly, we compared transcriptome profiles of high grade serous ovarian tumors that were obtained, on two different microarray platforms and next generation RNA-sequencing, to identify the most informative platform and extract robust biomarkers of molecular subtypes. We discovered that the variance of RNA-sequencing data processed using RPKM had greater variance than that with MapSplice and RSEM. We provided novel markers highly associated to tumor molecular subtype combined from four data platforms. MCIA is implemented and available in the R/Bioconductor "omicade4" package. CONCLUSION: We believe MCIA is an attractive method for data integration and visualization of several datasets of multi-omics features observed on the same set of individuals. The method is not dependent on feature annotation, and thus it can extract important features even when there are not present across all datasets. MCIA provides simple graphical representations for the identification of relationships between large datasets.

Mass-spectrometry-based draft of the human proteome.

Proteomes are characterized by large protein-abundance differences, cell-type- and time-dependent expression patterns and post-translational modifications, all of which carry biological information that is not accessible by genomics or transcriptomics. Here we present a mass-spectrometry-based draft of the human proteome and a public, high-performance, in-memory database for real-time analysis of terabytes of big data, called ProteomicsDB. The information assembled from human tissues, cell lines and body fluids enabled estimation of the size of the protein-coding genome, and identified organ-specific proteins and a large number of translated lincRNAs (long intergenic non-coding RNAs). Analysis of messenger RNA and protein-expression profiles of human tissues revealed conserved control of protein abundance, and integration of drug-sensitivity data enabled the identification of proteins predicting resistance or sensitivity. The proteome profiles also hold considerable promise for analysing the composition and stoichiometry of protein complexes. ProteomicsDB thus enables navigation of proteomes, provides biological insight and fosters the development of proteomic technology.

High-fat diet alters gut microbiota physiology in mice.

The intestinal microbiota is known to regulate host energy homeostasis and can be influenced by high-calorie diets. However, changes affecting the ecosystem at the functional level are still not well characterized. We measured shifts in cecal bacterial communities in mice fed a carbohydrate or high-fat (HF) diet for 12 weeks at the level of the following: (i) diversity and taxa distribution by high-throughput 16S ribosomal RNA gene sequencing; (ii) bulk and single-cell chemical composition by Fourier-transform infrared- (FT-IR) and Raman micro-spectroscopy and (iii) metaproteome and metabolome via high-resolution mass spectrometry. High-fat diet caused shifts in the diversity of dominant gut bacteria and altered the proportion of Ruminococcaceae (decrease) and Rikenellaceae (increase). FT-IR spectroscopy revealed that the impact of the diet on cecal chemical fingerprints is greater than the impact of microbiota composition. Diet-driven changes in biochemical fingerprints of members of the Bacteroidales and Lachnospiraceae were also observed at the level of single cells, indicating that there were distinct differences in cellular composition of dominant phylotypes under different diets. Metaproteome and metabolome analyses based on the occurrence of 1760 bacterial proteins and 86 annotated metabolites revealed distinct HF diet-specific profiles. Alteration of hormonal and anti-microbial networks, bile acid and bilirubin metabolism and shifts towards amino acid and simple sugars metabolism were observed. We conclude that a HF diet markedly affects the gut bacterial ecosystem at the functional level.

Global proteome analysis of the NCI-60 cell line panel.

The NCI-60 cell line collection is a very widely used panel for the study of cellular mechanisms of cancer in general and in vitro drug action in particular. It is a model system for the tissue types and genetic diversity of human cancers and has been extensively molecularly characterized. Here, we present a quantitative proteome and kinome profile of the NCI-60 panel covering, in total, 10,350 proteins (including 375 protein kinases) and including a core cancer proteome of 5,578 proteins that were consistently quantified across all tissue types. Bioinformatic analysis revealed strong cell line clusters according to tissue type and disclosed hundreds of differentially regulated proteins representing potential biomarkers for numerous tumor properties. Integration with public transcriptome data showed considerable similarity between mRNA and protein expression. Modeling of proteome and drug-response profiles for 108 FDA-approved drugs identified known and potential protein markers for drug sensitivity and resistance. To enable community access to this unique resource, we incorporated it into a public database for comparative and integrative analysis (http://wzw.tum.de/proteomics/nci60).

Comprehensive identification of proteins from MALDI imaging.

Matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI IMS) is a powerful tool for the visualization of proteins in tissues and has demonstrated considerable diagnostic and prognostic value. One main challenge is that the molecular identity of such potential biomarkers mostly remains unknown. We introduce a generic method that removes this issue by systematically identifying the proteins embedded in the MALDI matrix using a combination of bottom-up and top-down proteomics. The analyses of ten human tissues lead to the identification of 1400 abundant and soluble proteins constituting the set of proteins detectable by MALDI IMS including >90% of all IMS biomarkers reported in the literature. Top-down analysis of the matrix proteome identified 124 mostly N- and C-terminally fragmented proteins indicating considerable protein processing activity in tissues. All protein identification data from this study as well as the IMS literature has been deposited into MaTisse, a new publically available database, which we anticipate will become a valuable resource for the IMS community.

Diagnostic values of GHSR DNA methylation pattern in breast cancer.

DNA methylation patterns have been recognised as cancer-specific markers with high potential for clinical applications. We aimed at identifying methylation variations that differentiate between breast cancers and other breast tissue entities to establish a signature for diagnosis. Candidate genomic loci were analysed in 117 fresh-frozen breast specimens, which included cancer, benign and normal breast tissues from patients as well as material from healthy individuals. A cancer-specific DNA methylation signature was identified by microarray analysis in a test set of samples (n = 52, p < 2.1 × 10(-4)) and its performance was assessed through bisulphite pyrosequencing in an independent validation set (n = 65, p < 1.9 × 10(-7)). The signature is associated with SFRP2 and GHSR genes, and exhibited significant hypermethylation in cancers. Normal-appearing breast tissues from cancer patients were also methylated at these loci but to a markedly lower extent. This occurrence of methylated DNA in normal breast tissue of cancer patients is indicative of an epigenetic field defect. Concerning diagnosis, receiver operating characteristic curves and the corresponding area under the curve (AUC) analysis demonstrated a very high sensitivity and specificity of 89.3 and 100 %, respectively, for the GHSR methylation pattern (AUC >0.99). To date, this represents the DNA methylation marker of the highest sensitivity and specificity for breast cancer diagnosis. Functionally, ectopic expression of GHSR in a cell line model reduced breast cancer cell invasion without affecting cell viability upon stimulation of cells with ghrelin. Our data suggest a link between epigenetic down-regulation of GHSR and breast cancer cell invasion.

Discovery of O-GlcNAc-modified proteins in published large-scale proteome data.

The attachment of N-acetylglucosamine to serine or threonine residues (O-GlcNAc) is a post-translational modification on nuclear and cytoplasmic proteins with emerging roles in numerous cellular processes, such as signal transduction, transcription, and translation. It is further presumed that O-GlcNAc can exhibit a site-specific, dynamic and possibly functional interplay with phosphorylation. O-GlcNAc proteins are commonly identified by tandem mass spectrometry following some form of biochemical enrichment. In the present study, we assessed if, and to which extent, O-GlcNAc-modified proteins can be discovered from existing large-scale proteome data sets. To this end, we conceived a straightforward O-GlcNAc identification strategy based on our recently developed Oscore software that automatically analyzes tandem mass spectra for the presence and intensity of O-GlcNAc diagnostic fragment ions. Using the Oscore, we discovered hundreds of O-GlcNAc peptides not initially identified in these studies, and most of which have not been described before. Merely re-searching this data extended the number of known O-GlcNAc proteins by almost 100 suggesting that this modification exists even more widely than previously anticipated and the modification is often sufficiently abundant to be detected without enrichment. However, a comparison of O-GlcNAc and phospho-identifications from the very same data indicates that the O-GlcNAc modification is considerably less abundant than phosphorylation. The discovery of numerous doubly modified peptides (i.e. peptides with one or multiple O-GlcNAc or phosphate moieties), suggests that O-GlcNAc and phosphorylation are not necessarily mutually exclusive, but can occur simultaneously at adjacent sites.

Systematic identification of the HSP90 candidate regulated proteome.

HSP90 is a central player in the folding and maturation of many proteins. More than two hundred HSP90 clients have been identified by classical biochemical techniques including important signaling proteins with high relevance to human cancer pathways. HSP90 inhibition has thus become an attractive therapeutic concept and multiple molecules are currently in clinical trials. It is therefore of fundamental biological and medical importance to identify, ideally, all HSP90 clients and HSP90 regulated proteins. To this end, we have taken a global and a chemical proteomic approach in geldanamycin treated cancer cell lines using stable isotope labeling with amino acids in cell culture and quantitative mass spectrometry. We identified >6200 proteins in four different human cell lines and ~1600 proteins showed significant regulation upon drug treatment. Gene ontology and pathway/network analysis revealed common and cell-type specific regulatory effects with strong connections to unfolded protein binding and protein kinase activity. Of the 288 identified protein kinases, 98 were geldanamycin treatment including >50 kinases not formerly known to be regulated by HSP90. Protein turn-over measurements using pulsed stable isotope labeling with amino acids in cell culture showed that protein down-regulation by HSP90 inhibition correlates with protein half-life in many cases. Protein kinases show significantly shorter half lives than other proteins highlighting both challenges and opportunities for HSP90 inhibition in cancer therapy. The proteomic responses of the HSP90 drugs geldanamycin and PU-H71 were highly similar suggesting that both drugs work by similar molecular mechanisms. Using HSP90 immunoprecipitation, we validated several kinases (AXL, DDR1, TRIO) and other signaling proteins (BIRC6, ISG15, FLII), as novel clients of HSP90. Taken together, our study broadly defines the cellular proteome response to HSP90 inhibition and provides a rich resource for further investigation relevant for the treatment of cancer.

Quantitative chemical proteomics reveals new potential drug targets in head and neck cancer.

Tumors of the head and neck represent a molecularly diverse set of human cancers, but relatively few proteins have actually been shown to drive the disease at the molecular level. To identify new targets for individualized diagnosis or therapeutic intervention, we performed a kinase centric chemical proteomics screen and quantified 146 kinases across 34 head and neck squamous cell carcinoma (HNSCC) cell lines using intensity-based label-free mass spectrometry. Statistical analysis of the profiles revealed significant intercell line differences for 42 kinases (p < 0.05), and loss of function experiments using siRNA in high and low expressing cell lines identified kinases including EGFR, NEK9, LYN, JAK1, WEE1, and EPHA2 involved in cell survival and proliferation. EGFR inhibition by the small molecule inhibitors lapatinib, gefitinib, and erlotinib as well as siRNA led to strong reduction of viability in high but not low expressing lines, confirming EGFR as a drug target in 10-20% of HNSCC cell lines. Similarly, high, but not low EPHA2-expressing cells showed strongly reduced viability concomitant with down-regulation of AKT and ERK signaling following EPHA2 siRNA treatment or EPHA1-Fc ligand exposure, suggesting that EPHA2 is a novel drug target in HNSCC. This notion is underscored by immunohistochemical analyses showing that high EPHA2 expression is detected in a subset of HNSCC tissues and is associated with poor prognosis. Given that the approved pan-SRC family kinase inhibitor dasatinib is also a very potent inhibitor of EPHA2, our findings may lead to new therapeutic options for HNSCC patients. Importantly, the strategy employed in this study is generic and therefore also of more general utility for the identification of novel drug targets and molecular pathway markers in tumors. This may ultimately lead to a more rational approach to individualized cancer diagnosis and therapy.

Sulfite reductase defines a newly discovered bottleneck for assimilatory sulfate reduction and is essential for growth and development in Arabidopsis thaliana.

The role of sulfite reductase (SiR) in assimilatory reduction of inorganic sulfate to sulfide has long been regarded as insignificant for control of flux in this pathway. Two independent Arabidopsis thaliana T-DNA insertion lines (sir1-1 and sir1-2), each with an insertion in the promoter region of SiR, were isolated. sir1-2 seedlings had 14% SiR transcript levels compared with the wild type and were early seedling lethal. sir1-1 seedlings had 44% SiR transcript levels and were viable but strongly retarded in growth. In mature leaves of sir1-1 plants, the levels of SiR transcript, protein, and enzymatic activity ranged between 17 and 28% compared with the wild type. The 28-fold decrease of incorporation of (35)S label into Cys, glutathione, and protein in sir1-1 showed that the decreased activity of SiR generated a severe bottleneck in the assimilatory sulfate reduction pathway. Root sulfate uptake was strongly enhanced, and steady state levels of most of the sulfur-related metabolites, as well as the expression of many primary metabolism genes, were changed in leaves of sir1-1. Hexose and starch contents were decreased, while free amino acids increased. Inorganic carbon, nitrogen, and sulfur composition was also severely altered, demonstrating strong perturbations in metabolism that differed markedly from known sulfate deficiency responses. The results support that SiR is the only gene with this function in the Arabidopsis genome, that optimal activity of SiR is essential for normal growth, and that its downregulation causes severe adaptive reactions of primary and secondary metabolism.

Cross-species common regulatory network inference without requirement for prior gene affiliation.

MOTIVATION: Cross-species meta-analyses of microarray data usually require prior affiliation of genes based on orthology information that often relies on sequence similarity. RESULTS: We present an algorithm merging microarray datasets on the basis of co-expression alone, without any requirement for orthology information to affiliate genes. Combining existing methods such as co-inertia analysis, back-transformation, Hungarian matching and majority voting in an iterative non-greedy hill-climbing approach, it affiliates arrays and genes at the same time, maximizing the co-structure between the datasets. To introduce the method, we demonstrate its performance on two closely and two distantly related datasets of different experimental context and produced on different platforms. Each pair stems from two different species. The resulting cross-species dynamic Bayesian gene networks improve on the networks inferred from each dataset alone by yielding more significant network motifs, as well as more of the interactions already recorded in KEGG and other databases. Also, it is shown that our algorithm converges on the optimal number of nodes for network inference. Being readily extendable to more than two datasets, it provides the opportunity to infer extensive gene regulatory networks. AVAILABILITY AND IMPLEMENTATION: Source code (MATLAB and R) freely available for download at http://www.mchips.org/supplements/moghaddasi_source.tgz.

Decoding pooled RNAi screens by means of barcode tiling arrays.

BACKGROUND: RNAi screens via pooled short hairpin RNAs (shRNAs) have recently become a powerful tool for the identification of essential genes in mammalian cells. In the past years, several pooled large-scale shRNA screens have identified a variety of genes involved in cancer cell proliferation. All of those studies employed microarray analysis, utilizing either the shRNA's half hairpin sequence or an additional shRNA-associated 60 nt barcode sequence as a molecular tag. Here we describe a novel method to decode pooled RNAi screens, namely barcode tiling array analysis, and demonstrate how this approach can be used to precisely quantify the abundance of individual shRNAs from a pool. RESULTS: We synthesized DNA microarrays with six overlapping 25 nt long tiling probes complementary to each unique 60 nt molecular barcode sequence associated with every shRNA expression construct. By analyzing dilution series of expression constructs we show how our approach allows quantification of shRNA abundance from a pool and how it clearly outperforms the commonly used analysis via the shRNA's half hairpin sequences. We further demonstrate how barcode tiling arrays can be used to predict anti-proliferative effects of individual shRNAs from pooled negative selection screens. Out of a pool of 305 shRNAs, we identified 28 candidate shRNAs to fully or partially impair the viability of the breast carcinoma cell line MDA-MB-231. Individual validation of a subset of eleven shRNA expression constructs with potential inhibitory, as well as non-inhibitory, effects on the cell line proliferation provides further evidence for the accuracy of the barcode tiling approach. CONCLUSIONS: In summary, we present an improved method for the rapid, quantitative and statistically robust analysis of pooled RNAi screens. Our experimental approach, coupled with commercially available lentiviral vector shRNA libraries, has the potential to greatly facilitate the discovery of putative targets for cancer therapy as well as sensitizers of drug toxicity.