David Tuck

MSVM-RFE

MOTIVATION Given the thousands of genes and the small number of samples, gene selection has emerged as an important research problem in microarray data analysis. Support Vector Machine-Recursive Feature Elimination (SVM-RFE) is one of a group of recently described algorithms which represent the stat-of-the-art for gene selection. Just like SVM itself, SVM-RFE was originally designed to solve binary gene selection problems. Several groups have extended SVM-RFE to solve multiclass problems using one-versus-all techniques. However, the genes selected from one binary gene selection problem may reduce the classification performance in other binary problems. RESULTS In the present study, we propose a family of four extensions to SVM-RFE (called MSVM-RFE) to solve the multiclass gene selection problem, based on different frameworks of multiclass SVMs. By simultaneously considering all classes during the gene selection stages, our proposed extensions identify genes leading to more accurate classification.

Macrophages Directly Contribute to the Exaggerated Inflammatory Response in Cystic Fibrosis Transmembrane Conductance Regulator−/− Mice

Pulmonary infection with an exaggerated inflammatory response is the major cause of morbidity and mortality in cystic fibrosis (CF). The objective of this study was to determine whether differences in the innate immune system underlie the exaggerated immune response in CF. We established a model that recapitulates the exaggerated immune response in a CF mouse model by exposure to Pseudomonas aeruginosa LPS and assessed the pulmonary cellular and cytokine responses of wild-type (WT) and CF mice. Compared with WT mice, CF mice had increased numbers of neutrophils and increased proinflammatory cytokines in their bronchoalveolar lavage fluid after LPS exposure. Based on the increased levels of IL-1alpha, IL-6, granulocyte colony-stimulating factor (G-CSF), and keratinocyte chemoattractant, all of which are known to be produced by macrophages, we tested whether two populations of macrophages, bone marrow-derived macrophages and alveolar macrophages, directly contribute to the elevated cytokine response of CF mice to LPS. After in vitro stimulation of bone marrow-derived macrophages and alveolar macrophages with LPS, IL-1alpha, IL-6, G-CSF, and monocyte chemoattractant protein-1 were higher in CF compared with WT cell supernatants. Quantitative analyses for IL-6 and keratinocyte chemoattractant revealed that LPS-stimulated CF macrophages have higher mRNA and intracellular protein levels compared with WT macrophages. Our data support the hypothesis that macrophages play a role in the exuberant cytokine production and secretion that characterizes CF, suggesting that the macrophage response may be an important therapeutic target for decreasing the morbidity of CF lung disease.

Differentiated cells are more efficient than adult stem cells for cloning by somatic cell nuclear transfer

Since the creation of Dolly via somatic cell nuclear transfer (SCNT), more than a dozen species of mammals have been cloned using this technology. One hypothesis for the limited success of cloning via SCNT (1%–5%) is that the clones are likely to be derived from adult stem cells. Support for this hypothesis comes from the findings that the reproductive cloning efficiency for embryonic stem cells is five to ten times higher than that for somatic cells as donors and that cloned pups cannot be produced directly from cloned embryos derived from differentiated B and T cells or neuronal cells. The question remains as to whether SCNT-derived animal clones can be derived from truly differentiated somatic cells. We tested this hypothesis with mouse hematopoietic cells at different differentiation stages: hematopoietic stem cells, progenitor cells and granulocytes. We found that cloning efficiency increases over the differentiation hierarchy, and terminally differentiated postmitotic granulocytes yield cloned pups with the greatest cloning efficiency.

Loss of the retinoblastoma binding protein 2 (RBP2) histone demethylase suppresses tumorigenesis in mice lacking Rb1 or Men1

Aberrations in epigenetic processes, such as histone methylation, can cause cancer. Retinoblastoma binding protein 2 (RBP2; also called JARID1A or KDM5A) can demethylate tri- and dimethylated lysine 4 in histone H3, which are epigenetic marks for transcriptionally active chromatin, whereas the multiple endocrine neoplasia type 1 (MEN1) tumor suppressor promotes H3K4 methylation. Previous studies suggested that inhibition of RBP2 contributed to tumor suppression by the retinoblastoma protein (pRB). Here, we show that genetic ablation of Rbp2 decreases tumor formation and prolongs survival in Rb1+/− mice and Men1-defective mice. These studies link RBP2 histone demethylase activity to tumorigenesis and nominate RBP2 as a potential target for cancer therapy.

MicroRNA signatures differentiate melanoma subtypes

Melanoma is an aggressive cancer that is highly resistance to therapies once metastasized. We studied microRNA (miRNA) expression in clinical melanoma subtypes and evaluated different miRNA signatures in the background of gain of function somatic and inherited mutations associated with melanoma. Total RNA from 42 patient derived primary melanoma cell lines and three independent normal primary melanocyte cell cultures was evaluated by miRNA array. MiRNA expression was then analyzed comparing subtypes and additional clinicopathologic criteria including somatic mutations. The prevalence and association of an inherited variant in a miRNA binding site in the 3’UTR of the KRAS oncogene, referred to as the KRAS-variant, was also evaluated. We show that seven miRNAs, miR-142-3p, miR-486, miR-214, miR-218, miR-362, miR-650 and miR-31, were significantly correlated with acral as compared to non-acral melanomas (p < 0.04). In addition, we discovered that the KRAS-variant was enriched in non-acral melanoma (25%), and that miR-137 under expression was significantly associated with melanomas with the KRAS-variant. Our findings indicate that miRNAs are differentially expressed in melanoma subtypes and that their misregulation can be impacted by inherited gene variants, supporting the hypothesis that miRNA misregulation reflects biological differences in melanoma.

Distinct human papillomavirus type 16 methylomes in cervical cells at different stages of premalignancy.

Human papillomavirus (HPV) gene expression is dramatically altered during cervical carcinogenesis. Because dysregulated genes frequently show abnormal patterns of DNA methylation, we hypothesized that comprehensive mapping of the HPV methylomes in cervical samples at different stages of progression would reveal patterns of clinical significance. To test this hypothesis, thirteen HPV16-positive samples were obtained from women undergoing routine cervical cancer screening. Complete methylation data were obtained for 98.7% of the HPV16 CpGs in all samples by bisulfite-sequencing. Most HPV16 CpGs were unmethylated or methylated in only one sample. The other CpGs were methylated at levels ranging from 11% to 100% of the HPV16 copies per sample. The results showed three major patterns and two variants of one pattern. The patterns showed minimal or no methylation (A), low level methylation in the E1 and E6 genes (B), and high level methylation at many CpGs in the E5/L2/L1 region (C). Generally, pattern A was associated with negative cytology, pattern B with low-grade lesions, and pattern C with high-grade lesions. The severity of the cervical lesions was then ranked by the HPV16 DNA methylation patterns and, independently, by the pathologic diagnoses. Statistical analysis of the two rating methods showed highly significant agreement. In conclusion, analysis of the HPV16 DNA methylomes in clinical samples of cervical cells led to the identification of distinct methylation patterns which, after validation in larger studies, could have potential utility as biomarkers of neoplastic cervical progression.

GPHMM: an integrated hidden Markov model for identification of copy number alteration and loss of heterozygosity in complex tumor samples using whole genome SNP arrays

There is an increasing interest in using single nucleotide polymorphism (SNP) genotyping arrays for profiling chromosomal rearrangements in tumors, as they allow simultaneous detection of copy number and loss of heterozygosity with high resolution. Critical issues such as signal baseline shift due to aneuploidy, normal cell contamination, and the presence of GC content bias have been reported to dramatically alter SNP array signals and complicate accurate identification of aberrations in cancer genomes. To address these issues, we propose a novel Global Parameter Hidden Markov Model (GPHMM) to unravel tangled genotyping data generated from tumor samples. In contrast to other HMM methods, a distinct feature of GPHMM is that the issues mentioned above are quantitatively modeled by global parameters and integrated within the statistical framework. We developed an efficient EM algorithm for parameter estimation. We evaluated performance on three data sets and show that GPHMM can correctly identify chromosomal aberrations in tumor samples containing as few as 10% cancer cells. Furthermore, we demonstrated that the estimation of global parameters in GPHMM provides information about the biological characteristics of tumor samples and the quality of genotyping signal from SNP array experiments, which is helpful for data quality control and outlier detection in cohort studies.

MicroRNA Signatures Differentiate Uterine Cancer Tumor Subtypes

OBJECTIVE Endometrial cancer (EC) is the most common gynecologic malignancy. Type I EC has a favorable prognosis, while type II ECs account for half of all treatment failures. Little knowledge of the biological differences is available to predict EC outcomes besides their pathological distinctions. MicroRNAs (miRNA) are a family of non-translated RNAs important in regulating oncogenic pathways. Mis-expression patterns of miRNAs in EC, as well as differences in miRNA expression patterns between the subtypes of EC, has not been previously evaluated. Our purpose was to identify miRNA profiles of EC subtypes, and to identify miRNAs associated with these subtypes to ultimately understand the different biological behavior between these subtypes. METHODS Ninety-five fresh/frozen and paraffin-embedded samples of endometrial type I and II cancer, carcinosarcomas and benign endometrial samples were collected. MiRNA expression profiles were evaluated by microarray analysis. Statistical analysis was performed. RESULTS Distinct miRNA signatures in tumor versus normal samples and in endometrioid vs. uterine papillary serous carcinomas exist. Additionally, carcinosarcomas have a unique miRNA signature from either the type I or II epithelial tumors. CONCLUSIONS We hypothesize that further understanding the miRNAs that separate these subtypes of EC will lead to biological insights into the different behavior of these tumors.

Overcoming Therapeutic Resistance in HER2-Positive Breast Cancers with CDK4/6 Inhibitors

Using transgenic mouse models, cell line-based functional studies, and clinical specimens, we show that cyclin D1/CDK4 mediate resistance to targeted therapy for HER2-positive breast cancer. This is overcome using CDK4/6 inhibitors. Inhibition of CDK4/6 not only suppresses Rb phosphorylation, but also reduces TSC2 phosphorylation and thus partially attenuates mTORC1 activity. This relieves feedback inhibition of upstream EGFR family kinases, resensitizing tumors to EGFR/HER2 blockade. Consequently, dual inhibition of EGFR/HER2 and CDK4/6 invokes a more potent suppression of TSC2 phosphorylation and hence mTORC1/S6K/S6RP activity. The suppression of both Rb and S6RP enhances G1 arrest and a phenotype resembling cellular senescence. In vivo, CDK4/6 inhibitors sensitize patient-derived xenograft tumors to HER2-targeted therapies and delay tumor recurrence in a transgenic model of HER2-positive breast cancer.

Training the Next Generation of Informaticians: The Impact of “BISTI” and Bioinformatics—A Report from the American College of Medical Informatics

In 2002-2003, the American College of Medical Informatics (ACMI) undertook a study of the future of informatics training. This project capitalized on the rapidly expanding interest in the role of computation in basic biological research, well characterized in the National Institutes of Health (NIH) Biomedical Information Science and Technology Initiative (BISTI) report. The defining activity of the project was the three-day 2002 Annual Symposium of the College. A committee, comprised of the authors of this report, subsequently carried out activities, including interviews with a broader informatics and biological sciences constituency, collation and categorization of observations, and generation of recommendations. The committee viewed biomedical informatics as an interdisciplinary field, combining basic informational and computational sciences with application domains, including health care, biological research, and education. Consequently, effective training in informatics, viewed from a national perspective, should encompass four key elements: (1). curricula that integrate experiences in the computational sciences and application domains rather than just concatenating them; (2). diversity among trainees, with individualized, interdisciplinary cross-training allowing each trainee to develop key competencies that he or she does not initially possess; (3). direct immersion in research and development activities; and (4). exposure across the wide range of basic informational and computational sciences. Informatics training programs that implement these features, irrespective of their funding sources, will meet and exceed the challenges raised by the BISTI report, and optimally prepare their trainees for careers in a field that continues to evolve.