Michelle Kelliher received her B.A. in Biology from Smith College and a M.S. degree in Biology from Yale University. She received her Ph.D. in Immunology from Tufts University Sackler School of Biomedical Sciences and completed her post-doctoral training in the Department of Genetics at Harvard Medical School. In 1998, Dr. Kelliher joined the UMMS Faculty and is currently a Professor in the Molecular, Cellular and Cancer Biology Department. Dr. Kelliher is a Scholar and Stohlman Scholar of the Leukemia and Lymphoma Society of America and a Sidney Kimmel Cancer Scholar.  Her research is supported by the NIH (NCI and NIAID), a Hyundai Hope Grant for Pediatric Cancers and an Alex’s Lemonade Stand Innovator Award.

Visit the Kelliher Lab website: http://labs.umassmed.edu/kelliherlab/

Mechanisms of Leukemogenesis

Genetically engineered mouse model (GEMM) of pediatric T-ALL

T cell acute lymphoblastic leukemia (T-ALL) is largely caused by the activation of TAL1, LMO1/2 and the NOTCH1 oncogenes. To model the disease in mice, we ectopically expressed the basic-helix-loop-helix (bHLH) transcription factor TAL1 and its binding partner LMO2 in developing mouse thymocytes. These transgenic mice develop T-ALL that resembles the human disease. We have shown that the DNA binding activity of TAL1 is not required to induce leukemia in mice and demonstrated that E2A or HEB heterozygosity accelerates TAL1-mediated disease (O’Neil et al., 2001; 2004).  These studies revealed that TAL1 transforms in part, by interfering with the E47/HEB bHLH heterodimer required for the expression of T cell differentiation genes (Figure?). In collaboration with the Look lab, we performed ChIP-seq analyses that demonstrated that TAL1 also participates in a TAL1-GATA3-RUNX1 autoregulatory loop that induces the expression of stem cell genes such as MYB in leukemia (Sanda et al., 2012).

Identifying cooperating oncogenes

We have used our GEMM and retroviral insertional mutagenesis (RIM) to identify genes that cooperate with TAL1 to cause leukemia in mice. The RIM screens revealed recurrent retroviral insertions in the Notch1, Myc and Ikaros loci (Sharma et al., 2006). Consistent with these data, NOTCH1 mutations were discovered in 54% of T-ALL patients and shown to develop spontaneously in our TAL1 transgenic mice (O’Neil et al., 2006). These mouse T-ALLs also express dominant-negative forms of Ikaros, indicating that Ikaros acts as a tumor suppressor in the disease. We then discovered that NOTCH1 contributes to leukemogenesis by directly regulating the expression of MYC (Sharma et al., 2006).

NOTCH1-MYC axis mediates L-IC activity

Leukemia-initiating cells (L-IC) are hypothesized to exhibit extensive proliferative and self-renewal capabilities and thereby mediate relapse. Using our GEMM of T-ALL, we identified the DN3 progenitor population as enriched in L-IC activity. Since NOTCH1 is important in thymic progenitor expansion, we hypothesized that the L-IC may depend on the NOTCH1-MYC pathway for their activity.  We found that treatment with a gamma-secretase inhibitor (GSI) which prevents NOTCH1 activation or the bromodomain 4 (BRD4) inhibitor JQ1, which targets MYC, significantly reduces or eliminates the L-IC population and prevents disease initiation (Tatarek et al., 2011; Roderick et al., 2014). These studies suggested that NOTCH and BRD4 inhibition may target the L-IC and prevent relapse.

Establishing patient-derived xenografts from relapsed pediatric T-ALL and ETP-ALL patients

To translate our findings to the human disease, we generated patient-derived xenografts (PDX) from pediatric T-ALL patients at the time of diagnosis and upon induction failure or relapse. We are using these models to study disease heterogeneity and to test the efficacy of combination targeted therapies. We found that GSI-JQ1 combination therapy was effective in vivo, significantly prolonging survival in PDX models of relapsed pediatric T-ALL (Knoechel et al., 2014).

We are one of a few labs world wide that have also successfully generated PDX models from patients with early thymic progenitor (ETP)-ALL, a particularly treatment resistant ALL subtype. In collaboration with the Letai laboratory, we demonstrated that ETP-ALL is uniquely BCL2-dependent and highly sensitive to treatment in vivo with the BCL2 inhibitor ABT-199 (Chongaile et al., 2014). Our current goals are to examine human L-IC activity in these models and to optimize lentiviral-mediated transduction and CRISPR/Cas9 screening in primary human leukemic cells.

RIP Kinases in Cell Death and Inflammation

My laboratory has had a long-standing interest in RIP Kinases and their role in TNF- and TRIF-dependent signaling.  We demonstrated that a RIPK1 deficiency in the mouse results in neonatal lethality due to extensive TNF-induced cell death and inflammation (Kelliher et al., 1998; Cusson et al 2002). We showed that RIPK1 is recruited to the TNF receptor 1 (Tnfr1) and the Toll-like receptors (TLR) 3 and 4 via the adapter TRIF and is stably ubiquitin-modified with K63-linked polyubiquitin chains (Meylan et al., 2004; Lee et al., 2004). We have since established that in addition to TNF-induced apoptosis, RIPK1 regulates a form of programmed necrosis called necroptosis. Necroptosis is thought to require the kinase activities of RIPK1, RIPK3 and MLKL (Figure?) and induces an inflammatory form of cell death. We provide genetic evidence that in the absence of RIPK1, tissues undergo apoptosis and RIPK3-mediated necroptosis. Unlike Ripk1^-/- or Ripk1^-/-Tnfr1^-/- mice, which die during the postnatal period, Ripk1/Tnfr1/Ripk3 triple knock out mice survive to adulthood (Dillon et al., 2014). These in vivo studies demonstrate that RIPK1 is a master regulator of cell death and inflammation.

To identify the cell types and tissues that depend on RIPK1 for survival, we developed Ripk1 conditional mice and RIPK1 kinase inactive (D138N) mice.  These mouse models allow us to examine the role of RIPK1 in tissue homeostasis and inflammation. Our published work demonstrates critical survival roles for RIPK1 in the intestinal epithelium, keratinocytes and cells of the hematopoietic lineage (Dannappel, et al., 2014;Roderick et al., 2014). We also demonstrated that RIPK1D138N mice are completely protected from TNF-induced hypothermia and shock in vivo (Polykratis et al., 2014).  These studies indicate that RIP kinase-dependent necroptotic death mediates shock and may contribute to sepsis, raising the possibility that RIPK inhibitors may have clinical utility in these patients and in chronic inflammatory disease.