Proteomic analyses in Waldenstrom's macroglobulinemia and other plasma cell dyscrasias

Proteomic Analyses in Waldenstrom’s Macroglobulinemia and Other Plasma Cell Dyscrasias Constantine S. Mitsiades, Nicholas Mitsiades, Steven P. Treon, and Kenneth C. Anderson The proteomic analysis of tumor cells emerges as a key complement to gene expression profiling, primarily because regulation of protein expression (at the translational and post-translational levels) can buffer the magnitude of changes occurring at the gene transcription level, in order to fine tune cellular functions. Herein we describe the concept of proteomic analysis of the signaling state of tumor cells, as well as its application in the study of signaling pathways in plasma cell dyscrasias, such as Waldenstrom’s macroglobulinemia (WM) and multiple myeloma (MM). Comparative studies of WM versus MM cells at baseline and in the setting of drug treatment reveals proteomic profiles of the signaling state with significant overlap (that could reflect a putative B-cell lineage–related proteomic signature), but also distinct differences, possibly associated with differential features in the biologic behavior and drug sensitivity of these diseases. These proteomic studies pave the way for a more comprehensive insight into the molecular basis of WM versus other B-cell malignancies. Semin Oncol 30:156-160. © 2003 Elsevier Inc. All rights reserved.

N

OTWITHSTANDING its contributions to molecular classification and prognostication of neoplastic disorders,1 gene expression profiling alone cannot comprehensively characterize the functional aspects of molecular interactions in tumor cells. On one hand, changes in mRNA levels (eg, upon tumor cell stimulation with cytokines/ growth factors or antitumor agents) do not necessarily lead to stoichiometric changes at the protein level: due to complex regulatory mechanisms, protein translation not only mediates but also buffers

From the Jerome Lipper Multiple Myeloma Center, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA; and the Department of Medicine, Harvard Medical School, Boston, MA. Supported by the International Waldenstrom’s Macroglobulinemia Foundation (C.S.M), Multiple Myeloma Research Foundation (C.S.M and N.M.), Lauri Strauss Leukemia Foundation (C.S.M), and by a Doris Duke Distinguished Clinical Scientist Award (K.C.A). Address reprint requests to Kenneth C. Anderson, MD, Jerome Lipper Multiple Myeloma Center, Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, 44 Binney St, Mayer Building, Room M555, Boston, MA 02115. © 2003 Elsevier Inc. All rights reserved. 0093-7754/03/3002-3031$30.00/0 doi:10.1053/sonc.2003.50066 156

the transmission of biologic information from the transcriptome to the proteome, to optimize cell function. On the other hand, proteins are subject to post-translational modifications (eg, phosphorylation/dephosphorylation), which can fluctuate very rapidly, for example, in the context of signal transduction cascades, to modify the functional state of these proteins, independently of changes in corresponding mRNA levels. Proteomic profiles of tumor cells are therefore important to complement the information derived from gene expression studies. This is particularly important for diseases such as Waldenstrom’s macroglobulinemia (WM), where no curative therapy is currently available and significant advances in our understanding of the biology of this disease are warranted. METHODOLOGIES IN PROTEOMIC STUDIES

Two-dimensional (2-D) gel analyses of protein samples, which involve an initial isoelectric focusing and a conventional sodium dodecyl sulfate (SDS)-polyacrylamide gel (PAGE) electrophoresis, yield 2-D maps of protein spots, which capture the entire cellular proteome. However, the proteomic studies require more than just images of protein spots, but also identification of amino acid sequence for each spot and quantitative comparisons of protein levels in different gels. This lack of direct annotation/association of detected signals with the identity of corresponding protein(s) requires complicated and thorough biochemical studies (eg, radioimmunoassays, microsequencing, and mass spectrometry) to confirm the precise identity for each of the hundreds of protein dots in the 2-D gels, thus severely limiting the mainstream use of 2-D gels in a high-throughput fashion to study the proteomic profile of tumor cells. Recent applications of mass spectrometry– based proteomic studies (eg, matrix-assisted laser desorption ionization-time of flight [MALDI-TOF]) have allowed for higher degree of automation in proteomic profiling of tumor cells.2,3 These profiles consist of sequential spectrometric peaks, the position and magnitude of which reflect the molecSeminars in Oncology, Vol 30, No 2 (April), 2003: pp 156-160

PROTEOMICS IN PLASMA CELL DYSCRASIAS

ular weight and concentration, respectively, of the corresponding proteins. However, these analyses still lack direct correspondence of each spectrometric peak with its respective protein(s). To bypass the need to biochemically confirm the precise identity of each detected protein, efforts have been made to develop “protein microarrays,” which would include multiple series of antibodies coating the surface of a “protein chip”4 (in analogy to oligonucleotide or cDNA microarrays) in order to capture fluorescent-tagged proteins of analyzed samples and detect their binding to antibodies on the chip. Because of direct annotation of the position on the chip of each fluorescent signal with its corresponding protein identity, such studies can theoretically allow for high-throughput analyses of the tumor cell proteome. However, such chips, which would include antibodies with high sensitivity and specificity against all known proteins, are not currently available. Until such technologies are readily available, we have opted to perform proteomic analyses using a multiplex immunoblotting platform that allows for direct annotation to the identity of detected proteins. These studies are focused on a set of known, well-characterized proteins (for which prevalidated antibodies are widely available), implicated in signal transduction cascades (such as the Akt, Raf/mitogen-activated protein kinase [MAPK], protein kinase C [PKC], and NF-␬B pathways) regulating cell proliferation, survival, transcription, translation, etc. These signaling pathways not only transmit extracellular and intracellular stimuli that the cell needs to process in order to respond with fine-tuned coordinated modifications in its biologic functions, but also play a focal role in tumor cell biology and are frequently targeted by conventional, as well as novel, anticancer therapies.5-10 These proteomic analyses of the signaling state of WM and multiple myeloma (MM) cells were performed within the context of previously described studies both from other groups,11-13 as well as our center6,14 (C.S. Mitsiades, manuscript in preparation). In brief, total lysates of tumor cells from WM or MM cells were prepared in previously described lysis buffer,12 resolved on 13% SDS-PAGE gels, transferred to nitrocellulose membranes, and incubated with various primary antibodies followed by relevant horseradish peroxidase– conjugated secondary antibodies. The blots were developed with enhanced chemiluminescence (ECL) reagent

157

and signals were captured by multi-imager and quantified with image analysis software (Bio-Rad, Hercules, CA). The analyses were performed with mixes of prevalidated primary antibodies (from commercial sources) detecting protein targets of nonoverlapping molecular weights (to facilitate signal detection and analysis) and the application of each mix into a separate lane of a 20-lane multiblotter (Immunetics, Cambridge, MA), generating a setting resembling a 2-D gel. Detected protein signals in each sample are deployed in the two dimensions of the immunoblotting membrane (one for each experimental condition): the first dimension is created by separating the sample proteins according to their molecular weights (as in conventional immunoblottings), while the second dimension is generated by compartmentalization of the nitrocellulose membrane by the multiblotter; since each lane of the multiblotter is incubated with a separate mix of antibodies, each multiple protein per sample can be detected in one gel. Quantitative changes in signaling state of cells can be assessed by processing multiple samples (from different experimental conditions) with the same proteomic protocol and (similarly to comparative analyses between different 2-D gels) detection, in the different samples, of several proteins signals, which do not change across samples (and serve as reference values across different gels), as well as on normalization of the integrated staining intensity of each protein band by dividing its value by the sum of the intensities on the same gel of all proteins under study. Quantitative results were processed for hierarchical clustering and functional clustering analyses, as previously described.6,10,14 Pronounced changes in expression and/or phosphorylation of target proteins with potential pathophysiologic implications are confirmed with time-course and/or dose-response analyses by conventional immunoblotting. For the study of WM, we performed such analyses (1) to compare the signaling state of multiple myeloma (MM) versus WM cells; (2) to study in each tumor model the effect of tumor cell stimulation with cytokines, for example, stimulation of MM cells with insulin-like growth factor-1 (IGF-1) or interleukin-6 (IL-6),5 or antitumor agents, including the proteasome inhibitor PS34114 or the ansamycin inhibitors of the heatshock protein 90 (hsp90) molecular chaperone5;

158

and (3) to compare these drug-induced proteomic profiles in WM versus MM cells. While ongoing efforts are expanding these analyses to tumor cells freshly isolated from MM and WM patients, the original context of these proteomic studies involved cell lines, for example, the WM-WSU cell line model of WM, which has been previously characterized.15 The comparative proteomic analyses of signaling state in MM versus WM revealed, four clusters of proteins: (1) those expressed in MM, but not WM; (2) proteins present in WM, but not MM; (3) proteins expressed in both diseases; and (4) molecules that appear absent from both neoplasias (Fig 1). While ongoing mechanistic studies are addressing the biologic relevance of individual components of these clusters, the overlapping proteomic features of WM and MM cells conceivably reflect, in both diseases, key components of a pro-

MITSIADES ET AL

teomic signature of the B-cell lineage, and in particular the late stages of the B-cell differentiation, to which the normal counterpart(s) of MM and WM cells apparently correspond. For instance, the absence, in both MM and WM cells of the germinal center kinase (GCK), a serine/threonine protein kinase preferentially expressed in germinal center B lymphocytes and with a putative role in differentiation and selection of B cells in the germinal center, may be considered indicative of the post– germinal center ontogeny of both MM and WM. On the other hand, the qualitative and/or quantitative differences between MM and WM cells in terms of several individual signaling proteins possibly reflects their derivation from distinct levels of the post– germinal B-cell differentiation process. These distinct molecular backgrounds may also be implicated on the differential responsiveness of MM versus WM to various therapeutic

Fig 1. Comparative functional clustering analyses of proteomic profiles of PS-341–treated WM-WSU and MM-1S cells. The quantitative data of protein concentration and/or phosphorylation signals are represented visually on a spectrum of color changes (where red corresponds to higher and green to lower levels of expression), as generated with the Rainbow software for functional clustering analyses (T.A. Liberman & C. Bailey, Beth Israel Deaconess Medical Center, Boston, MA).

PROTEOMICS IN PLASMA CELL DYSCRASIAS

agents (eg, antitumor activity of nucleoside analogs in WM, but not MM). This partial overlap in MM versus WM in signaling proteomic signatures is also evident in comparative studies of the cell signaling profiles of cells treated with conventional or novel antitumor agents, including the proteasome inhibitor PS341. It might have been expected that this (highly active in MM) agent16 would lead to indiscriminate accumulation of undegraded proteins. Instead, PS-341 affects distinct groups of proteins implicated in regulation of proliferation, apoptosis/ survival and drug-sensitivity, eg, in both MM and WM cells, PS-341 induces suppression of expression DNA-PK, a kinase with key role in DNA repair, consistent with the recently documented potent chemosensitizing activity of PS-341 in MM.14 While certain quantitative differences were noted in the effect of PS-341 on certain signaling proteins in WM versus MM, an intriguing feature of these comparisons was that in tumor cells from both diseases, the magnitude of PS-341–induced changes at the protein level was less pronounced than those changes induced at the transcriptional level, as studied by gene expression profiling.10 In addition, hierarchical clustering analysis of the signaling state profiles of MM and WM cells showed that PS-341–treated cells from each disease clustered in the same branch of the dendrogram with their own respective baseline (untreated) cells, rather than with their drug-treated counterparts, ie, the similarities between untreated and PS-341–treated cells of the same disease were too potent to be overridden by drug-induced effects on the proteomic profiles (C.S. Mitsiades, manuscript in preparation). This observation, along with our recently described results showing coordinated patterns of transcriptional changes upon PS-341 treatment of MM cells, suggests that despite predictions to the contrary, PS-341 appears to induce fairly specific patterns of molecular and biologic sequelae. Importantly, these findings reinforce the notion that, at the molecular level, WM and MM share significant common denominators, but also have key differences. While the former ones may explain the overlap in the therapeutic options available for these diseases, the latter ones indicate that novel therapies rationally designed to capitalize on the distinct molecular features of WM may be

159

required to improve the outcome of patients with this presently incurable disease. CONCLUSIONS AND FUTURE DIRECTIONS

The studies of transcriptional and proteomic profiles of tumor cells in WM should not be viewed as conflicting, but as mutually complementing approaches to understand disease pathophysiology. Ongoing studies are attempting to define cascades mediating proliferation, survival, and drug resistance in WM, to expand the comparative proteomic profiling studies not only to WM and MM cells, but to other B-cell malignancies as well (lymphomas), to prepare the framework for therapeutic combinations tailored to the signaling state of tumor cells in individual patients and, finally, to define select group of molecular markers for diagnosis, prognostication and treatment monitoring in WM that can be studied by more widely available conventional techniques. REFERENCES 1. Golub TR, Slonim DK, Tamayo P, et al: Molecular classification of cancer: Class discovery and class prediction by gene expression monitoring. Science 286:531-537, 1999 2. Petricoin EF, Ardekani AM, Hitt BA, et al: Use of proteomic patterns in serum to identify ovarian cancer. Lancet 359:572-577, 2002 3. Petricoin EF, 3rd, Ornstein DK, Paweletz CP, et al: Serum proteomic patterns for detection of prostate cancer. J Natl Cancer Inst 94:1576-1578, 2002 4. Sreekumar A, Nyati MK, Varambally S, et al: Profiling of cancer cells using protein microarrays: Discovery of novel radiation-regulated proteins. Cancer Res 61:7585-7593, 2001 5. Mitsiades CS, Mitsiades N, Poulaki V, et al: The hsp90 molecular chaperone as a novel therapeutic target in hematologic malignancies. Blood 11S:377a, 2001 (suppl 1, abstr) 6. Mitsiades CS, Mitsiades N, Poulaki V, et al: Highthroughput global proteomic analysis of the signaling state of human multiple myeloma cells. Blood 98:3058a, 2001 (suppl 1, abstr) 7. Mitsiades CS, Mitsiades N, Poulaki V, et al: Activation of NF-␬B and upregulation of intracellular anti-apoptotic proteins via the IGF-1/Akt signaling in human multiple myeloma cells: Therapeutic implications. Oncogene 21:5673-5683, 2002 8. Mitsiades N, Mitsiades CS, Poulaki V, et al: Apoptotic signaling induced by immunomodulatory thalidomide analogs (IMiDs) in human multiple myeloma cells: Therapeutic implications. Blood 99:4525-4530, 2002 9. Mitsiades N, Mitsiades CS, Poulaki V, et al: Biologic sequelae of nuclear factor-kappaB blockade in multiple myeloma: Therapeutic applications. Blood 99:4079-4086, 2002 10. Mitsiades N, Mitsiades CS, Poulaki V, et al: Molecular sequelae of proteasome inhibition in human multiple myeloma cells. Proc Natl Acad Sci USA, 99:14374-14379, 2002

160

11. Zhang L, Pelech SL, Mayrand D, et al: Bacterial heat shock protein-60 increases epithelial cell proliferation through the ERK1/2 MAP kinases. Exp Cell Res 266:11-20, 2001 12. Zhang H, Shi X, Hampong M, et al: Stress-induced inhibition of ERK1 and ERK2 by direct interaction with p38 MAP kinase. J Biol Chem 276:6905-6908, 2001 13. Zhang H, Shi X, Zhang QJ, et al: Nocodazole-induced p53-dependent c-Jun N-terminal kinase activation reduces apoptosis in human colon carcinoma HCT-116 cells. J Biol Chem 46:43648-43658, 2002

MITSIADES ET AL

14. Mitsiades N, Mitsiades CS, Richardson PG, et al: The proteasome inhibitor PS-341 potentiates sensitivity of multiple myeloma cells to conventional chemotherapeutic agents. Blood (in press) 15. Al-Katib A, Mohammad R, Hamdan M, et al: Propagation of Waldenstrom’s macroglobulinemia cells in vitro and in severe combined immune deficient mice: Utility as a preclinical drug screening model. Blood 81:3034-3042, 1993 16. Mitsiades CS, Mitsiades N, Richardson PG, et al: Novel biologically based therapies for Waldenstrom’s macroglobulinemia. Semin Oncol 30:309-312, 2003