One method to discover new therapeutic targets in cancer is array-based comparative genomic hybridization (CGH) coupled with gene expression profiling. Previously, our laboratory used gene expression profiling to define molecular subtypes of diffuse large B cell lymphoma (DLBCL), termed germinal center B cell-like (GCB) DLBCL, activated B cell-like (ABC) DLBCL and primary mediastinal B cell lymphoma (PMBL). We determined regions of copy number gain, amplification and deletion using array-based CGH in 203 DLBCL biopsy samples. Using this approach, we uncovered a host of genetic aberrations that distinguished the subtypes from one another, providing compelling genetic evidence that these DLBCL subtypes should be considered as distinct disease entities. Gene expression profiling on the same samples was used to identify which genes in the affected genomic regions were altered in expression by the genomic aberrations. A region on chromosome 19 including the SPIB locus was gained or amplified in one quarter of ABC DLBCLs but not in GCB DLBCLs. SPIB was one of the most upregulated genes in cases with this genomic aberration. Moreover, our laboratory demonstrated that the SPIB gene is juxtaposed to the immunoglobulin heavy chain locus by a chromosomal translocation in an ABC DLBCL cell line, leading to pronounced overexpression of SPIB. SPIB is a lymphoid-restricted ETS family transcription factor that is required for optimal B cell activation and germinal center responses. RNA interference (RNAi)-mediated knock down of SPIB was lethal to all ABC DLBCL cell lines, establishing SPIB as a new oncogene and therapeutic target in this disease. Our laboratory has embarked on a new initiative to discover oncogenic somatic mutations in lymphoid malignancies by cancer gene resequencing. Previously, in an RNAi-based genetic screen, we discovered that a pathway involving CARD11, BCL10 and MALT1 was responsible for the constitutive NF-kB signaling in ABC DLBCL. In a clear validation of the RNAi-based genetic screen, our lab discovered recurrent somatic mutations in the CARD11 gene in ABC DLBCL tumor biopsies. All of the CARD11 mutations in DLBCL changed amino acids in one small domain that is predicted to adopt a coiled-coil structure. These CARD11 mutations created protein isoforms that constitutively engaged NF-kB signaling, apparently due to their ability to spontaneously form large cytosolic aggregates that colocalize with signaling proteins in the NF-kB pathway. Interference with the CARD11 coiled-coil domain was lethal to ABC DLBCL cells, thereby suggesting a method to attack CARD11 therapeutically. Most recently, our Achilles heel screens allowed us to define a chronic active form of B cell receptor (BCR) signaling that activates NF-kB in ABC DLBCLs with wild-type CARD11. Such ABC DLBCLs die upon knockdown of BCR signaling components, including subunits of the B cell receptor itself. ABC DLBCLs have prominent clusters of the BCR in the plasma membrane, similar to antigen-stimulated normal B cells. Cancer gene resequencing revealed that over one fifth of ABC DLBCLs have mutations in the CD79B or CD79A subunits of the BCR. The most common mutations, present in 18% of ABC DLBCLs, involved a single tyrosine of the BCR signaling subunit, CD79B. These mutations affect the critical ITAM signaling motif, generating BCRs that avoid negative autoregulation by the LYN tyrosine kinase. Importantly, the BCR pathway offers a wealth of targets that can be exploited therapeutically, including several protein kinases (SRC-family kinases, SYK, BTK, PKCbeta) as well as PI(3) kinase. Dasatinib, a clinically available kinase inhibitor that targets BTK and SRC-family kinases, kills ABC DLBCL cells by blocking their chronic active BCR signaling. Among the downstream targets of NF-kB signaling in ABC DLBCL are the cytokines IL-6 and IL-10. Both cytokines signal through cell surface receptors linked to JAK family kinases, which phosphorylate and activate the transcription factor STAT3. A gene expression signature of STAT3 activity was present in a subset of primary ABC DLBCL tumors but absent in GCB DLBCLs. From a therapeutic standpoint, ABC DLBCL cell lines were killed synergistically by a combination of a small molecule inhibitor of JAK family kinases with an IkB kinase beta inhibitor, which blocks NF-kB signaling. We have recently identified oncogenic signaling by the adapter protein MYD88 as the genetic basis for the JAK-STAT3 activation in ABC DLBCL. An RNAi screen in ABC DLBCL revealed the dependence of these cells on MYD88 and its downstream kinases IRAK1 and IRAK4. High-throughput RNA resequencing (RNA-seq) uncovered recurrent mutations in MYD88. MYD88 mutations are present in 39% of ABC DLBCLs, with 29% affecting changing one leucine in the MYD88 TIR domain to proline (L265P). The L265P mutant isoform spontaneously coordinates a signaling complexed involving IRAK1 and IRAK4, which turns on the NF-kB, JAK-STAT3 pathway, and type I interferon pathway. Small molecule inhibitors of IRAK4 kinase are selectively lethal to ABC DLBCL cells, offering new therapeutic prospects. ABC DLBCL cells are killed in a synergistic fashion by combined inhibition of the BCR and MYD88 pathways, suggesting the development of clinical trials combining drugs targeting both pathways. We have also conducted screens to identify genes and pathways that synergize or antagonize the action of therapeutic agents in cancer. We conducted screens for shRNAs that affect the response of ABC DLBCLs to a small molecule inhibitor of IKKbeta. Unexpectedly, this screen revealed that a distinct kinase, IKKalpha, contributes to classical NF-kB signaling when IKKbeta is inhibited. These findings suggest that therapeutic targeting of IKKbeta may be compromised by compensatory IKKalpha activation, lending impetus to the development of IKKalpha inhibitors. Soon after we recognized the importance of the NF-kB pathway to ABC DLBCL, we teamed with Wyndham Wilson and Kieron Dunleavy (Metabolism Branch) to evaluate the proteasome inhibitor bortezomib in DLBCL. Bortezomib inhibits NF-kB signaling by blocking the proteasomal degradation of IkappaBalpha, although it clearly has additional targets. Since NF-kB signaling inhibits the action of most cytotoxic chemotherapy, we hypothesized that combining bortezomib with multi-agent chemotherapy might be synergistic. We enrolled patients with relapsed and refractory DLBCL and obtained biopsy samples so that we could classify each case as ABC or GCB DLBCL by molecular profiling. Patients with ABC DLBCL had much more frequent complete and partial remissions than those with GCB DLBCL (p=0.0004) and had a superior overall survival (p=0.011). In multiple previous cohorts, the curative response to chemotherapy was significantly worse in ABC DLBCL, supporting our proposal that bortezomib synergizes with chemotherapy to improve the survival in ABC DLBCL. Based on these results, Millennium Pharmaceuticals is designing a new trial combining bortezomib with CHOP chemotherapy and Rituximab in previously untreated DLBCL, which will only enroll patients with molecularly documented ABC DLBCL. Given that bortezomib can have therapy-limiting toxicity (e.g. peripheral neuropathy), we are intent on searching for additional ways to inhibit the NF-kB pathway clinically. There are several new drugs entering early phase clinical trials that target the pathways we have implicated using our functional genomics methods. The B cell receptor signaling pathway affords many possible targets for the treatment of ABC DLBCL, notably BTK. We have initiated a phase I/II clinical trial of a small molecule BTK inhibitor in patients with ABC DLBCL, with gene expression profiling used for both diagnosis and pharmacodynamics.
Showing the most recent 10 out of 42 publications
