Roger E. Bumgarner

Direct multiplexed measurement of gene expression with color-coded probe pairs

We describe a technology, the NanoString nCounter gene expression system, which captures and counts individual mRNA transcripts. Advantages over existing platforms include direct measurement of mRNA expression levels without enzymatic reactions or bias, sensitivity coupled with high multiplex capability, and digital readout. Experiments performed on 509 human genes yielded a replicate correlation coefficient of 0.999, a detection limit between 0.1 fM and 0.5 fM, and a linear dynamic range of over 500-fold. Comparison of the NanoString nCounter gene expression system with microarrays and TaqMan PCR demonstrated that the nCounter system is more sensitive than microarrays and similar in sensitivity to real-time PCR. Finally, a comparison of transcript levels for 21 genes across seven samples measured by the nCounter system and SYBR Green real-time PCR demonstrated similar patterns of gene expression at all transcript levels.

Gene expression in Pseudomonas aeruginosa biofilms

Bacteria often adopt a sessile biofilm lifestyle that is resistant to antimicrobial treatment. Opportunistic pathogenic bacteria like Pseudomonas aeruginosa can develop persistent infections. To gain insights into the differences between free-living P. aeruginosa cells and those in biofilms, and into the mechanisms underlying the resistance of biofilms to antibiotics, we used DNA microarrays. Here we show that, despite the striking differences in lifestyles, only about 1% of genes showed differential expression in the two growth modes; about 0.5% of genes were activated and about 0.5% were repressed in biofilms. Some of the regulated genes are known to affect antibiotic sensitivity of free-living P. aeruginosa. Exposure of biofilms to high levels of the antibiotic tobramycin caused differential expression of 20 genes. We propose that this response is critical for the development of biofilm resistance to tobramycin. Our results show that gene expression in biofilm cells is similar to that in free-living cells but there are a small number of significant differences. Our identification of biofilm-regulated genes points to mechanisms of biofilm resistance to antibiotics.

Bayesian model averaging: development of an improved multi-class, gene selection and classification tool for microarray data

MOTIVATION Selecting a small number of relevant genes for accurate classification of samples is essential for the development of diagnostic tests. We present the Bayesian model averaging (BMA) method for gene selection and classification of microarray data. Typical gene selection and classification procedures ignore model uncertainty and use a single set of relevant genes (model) to predict the class. BMA accounts for the uncertainty about the best set to choose by averaging over multiple models (sets of potentially overlapping relevant genes). RESULTS We have shown that BMA selects smaller numbers of relevant genes (compared with other methods) and achieves a high prediction accuracy on three microarray datasets. Our BMA algorithm is applicable to microarray datasets with any number of classes, and outputs posterior probabilities for the selected genes and models. Our selected models typically consist of only a few genes. The combination of high accuracy, small numbers of genes and posterior probabilities for the predictions should make BMA a powerful tool for developing diagnostics from expression data. AVAILABILITY The source codes and datasets used are available from our Supplementary website.

Integrating Large-Scale Functional Genomic Data to Dissect the Complexity of Yeast Regulatory Networks

A key goal of biology is to construct networks that predict complex system behavior. We combine multiple types of molecular data, including genotypic, expression, transcription factor binding site (TFBS), and protein–protein interaction (PPI) data previously generated from a number of yeast experiments, in order to reconstruct causal gene networks. Networks based on different types of data are compared using metrics devised to assess the predictive power of a network. We show that a network reconstructed by integrating genotypic, TFBS and PPI data is the most predictive. This network is used to predict causal regulators responsible for hot spots of gene expression activity in a segregating yeast population. We also show that the network can elucidate the mechanisms by which causal regulators give rise to larger-scale changes in gene expression activity. We then prospectively validate predictions, providing direct experimental evidence that predictive networks can be constructed by integrating multiple, appropriate data types.

Comparative hybridization of an array of 21 500 ovarian cDNAs for the discovery of genes overexpressed in ovarian carcinomas

Comparative hybridization of cDNA arrays is a powerful tool for the measurement of differences in gene expression between two or more tissues. We optimized this technique and employed it to discover genes with potential for the diagnosis of ovarian cancer. This cancer is rarely identified in time for a good prognosis after diagnosis. An array of 21,500 unknown ovarian cDNAs was hybridized with labeled first-strand cDNA from 10 ovarian tumors and six normal tissues. One hundred and thirty-four clones are overexpressed in at least five of the 10 tumors. These cDNAs were sequenced and compared to public sequence databases. One of these, the gene HE4, was found to be expressed primarily in some ovarian cancers, and is thus a potential marker of ovarian carcinoma.

Cellular transcriptional profiling in influenza A virus-infected lung epithelial cells: The role of the nonstructural NS1 protein in the evasion of the host innate defense and its potential contribution to pandemic influenza

The NS1 protein of influenza A virus contributes to viral pathogenesis, primarily by enabling the virus to disarm the host cell type IFN defense system. We examined the downstream effects of NS1 protein expression during influenza A virus infection on global cellular mRNA levels by measuring expression of over 13,000 cellular genes in response to infection with wild-type and mutant viruses in human lung epithelial cells. Influenza A/PR/8/34 virus infection resulted in a significant induction of genes involved in the IFN pathway. Deletion of the viral NS1 gene increased the number and magnitude of expression of cellular genes implicated in the IFN, NF-κB, and other antiviral pathways. Interestingly, different IFN-induced genes showed different sensitivities to NS1-mediated inhibition of their expression. A recombinant virus with a C-terminal deletion in its NS1 gene induced an intermediate cellular mRNA expression pattern between wild-type and NS1 knockout viruses. Most significantly, a virus containing the 1918 pandemic NS1 gene was more efficient at blocking the expression of IFN-regulated genes than its parental influenza A/WSN/33 virus. Taken together, our results suggest that the cellular response to influenza A virus infection in human lung cells is significantly influenced by the sequence of the NS1 gene, demonstrating the importance of the NS1 protein in regulating the host cell response triggered by virus infection.

Bacterial Communities in Women with Bacterial Vaginosis: High Resolution Phylogenetic Analyses Reveal Relationships of Microbiota to Clinical Criteria

Background Bacterial vaginosis (BV) is a common condition that is associated with numerous adverse health outcomes and is characterized by poorly understood changes in the vaginal microbiota. We sought to describe the composition and diversity of the vaginal bacterial biota in women with BV using deep sequencing of the 16S rRNA gene coupled with species-level taxonomic identification. We investigated the associations between the presence of individual bacterial species and clinical diagnostic characteristics of BV. Methodology/Principal Findings Broad-range 16S rRNA gene PCR and pyrosequencing were performed on vaginal swabs from 220 women with and without BV. BV was assessed by Amsel’s clinical criteria and confirmed by Gram stain. Taxonomic classification was performed using phylogenetic placement tools that assigned 99% of query sequence reads to the species level. Women with BV had heterogeneous vaginal bacterial communities that were usually not dominated by a single taxon. In the absence of BV, vaginal bacterial communities were dominated by either Lactobacillus crispatus or Lactobacillus iners. Leptotrichia amnionii and Eggerthella sp. were the only two BV-associated bacteria (BVABs) significantly associated with each of the four Amsel’s criteria. Co-occurrence analysis revealed the presence of several sub-groups of BVABs suggesting metabolic co-dependencies. Greater abundance of several BVABs was observed in Black women without BV. Conclusions/Significance The human vaginal bacterial biota is heterogeneous and marked by greater species richness and diversity in women with BV; no species is universally present. Different bacterial species have different associations with the four clinical criteria, which may account for discrepancies often observed between Amsel and Nugent (Gram stain) diagnostic criteria. Several BVABs exhibited race-dependent prevalence when analyzed in separate groups by BV status which may contribute to increased incidence of BV in Black women. Tools developed in this project can be used to study microbial ecology in diverse settings at high resolution.

Clustering gene-expression data with repeated measurements

Clustering is a common methodology for the analysis of array data, and many research laboratories are generating array data with repeated measurements. We evaluated several clustering algorithms that incorporate repeated measurements, and show that algorithms that take advantage of repeated measurements yield more accurate and more stable clusters. In particular, we show that the infinite mixture model-based approach with a built-in error model produces superior results.

Cellular Gene Expression upon Human Immunodeficiency Virus Type 1 Infection of CD4+-T-Cell Lines

ABSTRACT The expression levels of ∼4,600 cellular RNA transcripts were assessed in CD4+-T-cell lines at different times after infection with human immunodeficiency virus type 1 strain BRU (HIV-1BRU) using DNA microarrays. We found that several classes of genes were inhibited by HIV-1BRU infection, consistent with the G2 arrest of HIV-1-infected cells induced by Vpr. These included genes involved in cell division and transcription, a family of DEAD-box proteins (RNA helicases), and all genes involved in translation and splicing. However, the overall level of cell activation and signaling was increased in infected cells, consistent with strong virus production. These included a subgroup of transcription factors, including EGR1 and JUN, suggesting they play a specific role in the HIV-1 life cycle. Some regulatory changes were cell line specific; however, the majority, including enzymes involved in cholesterol biosynthesis, of changes were regulated in most infected cell lines. Compendium analysis comparing gene expression profiles of our HIV-1 infection experiments to those of cells exposed to heat shock, interferon, or influenza A virus indicated that HIV-1 infection largely induced specific changes rather than simply activating stress response or cytokine response pathways. Thus, microarray analysis confirmed several known HIV-1 host cell interactions and permitted identification of specific cellular pathways not previously implicated in HIV-1 infection. Continuing analyses are expected to suggest strategies for impacting HIV-1 replication in vivo by targeting these pathways.

Multiclass classification of microarray data with repeated measurements: application to cancer

Prediction of the diagnostic category of a tissue sample from its gene-expression profile and selection of relevant genes for class prediction have important applications in cancer research. We have developed the uncorrelated shrunken centroid (USC) and error-weighted, uncorrelated shrunken centroid (EWUSC) algorithms that are applicable to microarray data with any number of classes. We show that removing highly correlated genes typically improves classification results using a small set of genes.