Robert W. Georgantas III, Vivek Tanadve1, Matthew Malehorn, Shelly Heimfeld, Chen Chen, Laura Carr, Francisco Murillo, Jeanne Kowalski, Greg Riggins, Katie Wartenby, Curt I. Civin.

Robert W. Georgantas III

Hematopoietic “Stemness” Genetic Profile in Normal Human Bone Marrow, Cord Blood, and Mobilized Peripheral Blood Stem Cells

International Society of Hematology

July 2003, Paris

The human CD34+/CD38-/Lin- cell subset (HSC) comprises ~1-10% of the CD34+ cell population and contains most if not all of the primitive in vivo engrafting stem cells, yet includes few to none of the less primitive hematopoietic single-lineage-committed progenitor cells. We analyzed gene expression of the CD34+/CD38-/Lin- cell populations isolated from bone marrow, placental/umbilical cord blood, and mobilized peripheral blood from normal donors. The transcriptomes of the CD34+/CD38-/Lin- cells from each tissue were determined, and 4746 genes were found to be expressed in cells from all three tissues. We also isolated and determined the transcriptomes of the stem cell-depleted, progenitor cell-enriched CD34+/[CD38/Lin]+ cell population from each tissue. Comparison of the transcripts expressed in CD34+/CD38-/Lin- versus CD34+/[CD38/Lin]+ cells from each tissue yielded 81 genes that were over-represented in all three of the CD34+/CD38-/Lin- cell populations. These transcripts include a number of known genes (e.g., transcription factors, receptors, signaling molecules, etc), many of which have been previously implicated in hematopoiesis. Interestingly almost half of the expressed genes are named genes of unknown function or completely unknown genes (i.e. ESTs and predicted/hypothetical genes). In summary, we have compared the transcriptomes of HSCs from three different hematopoietic tissues to uncover genes that may play roles in survival, self-renewal, differentiation and/or migration/adhesion of human lympho-hematopoietic stem cells.

Robert W. Georgantas III, Vivek Tanadve1, Matthew Malehorn, Shelly Heimfeld, Greg Riggins, Laura Carr, Francisco Murillo, Jeanne Kowalski, Katie Whartenby, Curt I. Civin

Robert W. Georgantas III

Microarray Analyses And Serial Analysis Of Gene Expression (SAGE) Revealed Large Numbers Of Novel Genes Differentially Expressed In Stem Cell-Enriched Versus Progenitor Cell-Enriched Populations

American Society of Hematology

December 2003, San Diego

The human CD34+/CD38-/Lin- cell hematopoietic stem cell-enriched subset (“HSCs”) comprises ~1-10% of the CD34+ cell population and contains most if not all of the primitive in vivo engrafting stem cells, yet includes few of the less primitive hematopoietic single-lineage-committed progenitor cells. We hypothesized that this rare HSC subset may express transcripts that have not been identified in commonly-studied tissues. We analyzed gene expression of the CD34+/CD38-/Lin- cell populations isolated from bone marrow, placental/umbilical cord blood, and mobilized peripheral blood from normal human donors with Affymetrix U133 arrays. 4746 genes were found to be expressed in HSCs from all three tissues. We also isolated and determined the transcriptomes of the stem cell-depleted, progenitor cell-enriched CD34+/[CD38/Lin]+ cell population from each tissue. Rigorous comparison of the transcripts expressed in CD34+/CD38-/Lin- versus CD34+/[CD38/Lin]+ cells from each tissue yielded 81 genes that were over-represented in all three of the CD34+/CD38-/Lin- cell populations. These transcripts include a number of known genes (e.g., transcription factors, receptors, signaling molecules, etc), many of which have been previously implicated in hematopoiesis, and others that be involved in the key characteristics (e.g., survival, self-renewal, differentiation, and/or migration/adhesion) of HSCs. Interestingly almost half of the HSC-over-expressed genes were named genes of unknown function or completely uncharacterized genes (i.e. ESTs and predicted/hypothetical genes). To uncover transcripts not present on the Affymetrix arrays and transcripts expressed at very low levels, we performed SAGE on bone marrow (BM) HSCs. Where 6366 transcripts were detected in the BM HSC by microarrays, SAGE identified ~9000 unique transcripts. It is therefore probable that as many as 30% of the genes expressed by HSCs are either unknown transcripts not represented in the current EST database and/or may be expressed at levels too low to be detected by microarray technology.

Charles G. Drake, Jeanne Kowalski, Ching-Tai Huang, Jonathan Powell, Drew Pardoll

Charles G. Drake

Application of the HAM (Hypothesis-Based Analysis of Microarrays) Algorithm to T Cell Anergy

Affymetrix Users’ Group Meeting

April 29 – May 1, 2003

When T cells encounter their cognate antigen, several distinctbiological outcomes are possible. Under certain conditions, i.e. if adequate costimulation is provided, the T cells display effector function, including division and cytokine production. However, when T cells encounter certain self-antigens, they develop an “anergic” phenotype – they become refractory to further stimuli. We have developed an in-vivo model of T cell anergy, in which a small population of T cells specific for hemagglutinin is adoptively transferred to an animal that expresses hemagglutinin in multiple epithelial tissues. We can recover a pure population of specific T cells using a combination of magnetic bead and FACS based sorting. In order to investigate the transcriptional profile of anergic T cells, we employed a unique statistical method termed HAM (hypothesis-based analysis of microarrays). This two stage method facilitates comparisons between microarray data obtained from a single sample under multiple experimental conditions.

Jeanne Kowalski

Nonparametric, Hypothesis-Based Analysis of Genetic Heterogeneity Associated with Phenotype

The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins

September, 2002

Advances in technology have led to an explosion of genetic research in many fields. Oncology researchers study molecular markers for diagnostic tools by relating expressions from thousands of genes to cancer status, while HIV researchers study drug resistance by relating genetic mutations to altered drug susceptibility. Both tasks include statistical issues of high dimensionality coupled with small sample sizes and thus preclude formal hypothesis testing based on conventional principles.

In this talk, I describe three novel inference-based approaches to analysis of genetic heterogeneity associated with phenotype. A common theme among them is the construction of testable hypotheses with assumptions that reflect the complex structure of genetic data. With a modest sample, I discuss a distance-based approach to analysis of genetic heterogeneity based on population sequence data. With two samples from a replicated microarray experiment, I discuss a multiple threshold approach to define reproducible, differentially expressed genes. With a single sample, I introduce a stochastic linear hypothesis approach to define the number of genes over-expressed, beyond experimental variation, among three phenotypes relative to a reference. In each setting, I also discuss bioinformatic approaches to characterize genes or locations and mutation patterns that depict phenotypes. As motivation for the methods, I examine three separate problems, one for relating differences in a region of the HIV genome to drug resistance, a second for associating gene expressions with thyroid cancer and a third for immunogenetic analysis of T cells.

Jonathan Powell, Jeanne Kowalski, L. Luu, C. Chen, R. Sharpf, C. Drake, D Pardoll, R. Schwartz

Jonathan Powell

Hypothesis Driven Differential Display as a means of Dissecting TCR induced activation and Tolerance

Keystone Symposia on Mechanisms & Application of Immune Tolerance 2002

2002

TCR signaling leads to both T cell activation and the upregulation of factors that serve to inhibit the immune response and induce tolerance. The precise pathways involved in positive and negative regulation of T cells have been elusive. To address this question we utilized gene chips to examine differential gene expression for multiple discreet activation conditions. In order to execute this analysis we devised a novel method of intersecting Hypothesis-based Analysis of Microarrays. This approach, for which comparisons are made based on a cutoff of 1, does not require the assumption that gene expressions are from some common distribution. As a result, this method facilitates union-intersection-type tests of hypotheses for multiple conditions and experiments. To this end gene chip analysis was performed on RNA isolated from T cell clones either: 1: unstimulated; 2: stimulated with anti-TCR to induce anergy; 3: Anti-TCR + CSA which inhibits anergy; 4: Anti-TCR + PKC inhibitor which inhibits anergy; 5: Anti-TCR + MEKinhibitor, p38 inhibitor, rapamycin which do not inhibit anergy. In order to elucidate genes/pathways involved in tolerance (anergy) we looked for the intersection of genes that were not present in resting cells, upregulated by TCR engagement, inhibited by CSA and PKCi and not donwregulated by MEK inhibitor, p38 inhibitor and rapamycin. Interestingly, approximately 1000/36,000 genes/Ests were upregulated in response to TCR stimulation (an exact p value of .0001 based on 5,000 permutations). This number was dramatically reduced to approximately 150-200 when conditions 2 and 3 were included. Condition 5 (genes still present in the presence of the 3 inhibitors) not only reduced this number to approximately 100 but also served to eliminate a number of genes clearly associated with T cell activation such as Mip1a, Interferon, GM-CSF and TNF. Amongst the genes that met all 5 criteria were many genes whose function make them strong candidates as mediators of the induction and maintenance of anergy. These genes include Notch-1, SOCs, SLAM, PD-1, PKC eta and EGR-2, and the biologic evaluation of these genes is underway. Currently, we are expanding our data base to include data from in vivo anergized cells. We believe that our strategy of intersecting hypotheses based analysis can be employed to evaluate large databases of gene chip data in order to identify genes/pathways involved in a multitude of discrete biologic functions.

HAM Home

Applications

Papers

Oncology Biostatistics Home