Genome-Wide Profiling of Endothelial Progenitor Cells in Multiple Myeloma: Disease-Relevant Pathways and Overlaps with Common Cancer Biomarkers

Background: The current study is based on recent findings from our and other laboratories demonstrating endothelial progenitor cells (EPCs) in multiple myeloma (MM) to be a key component of the tumor microenvironment, integral to the neoplastic process. EPCs contribute to tumor neoangiogenesis, and their levels covary with MM progression. Furthermore, MM EPCs are genetically unstable, as evidenced by restricted X-chromosome inactivation patterns; by immunoglobulin rearrangements identical to those harbored by tumor cells; and by a range of shared chromosomal gains or losses with tumor cells, shown by array comparative genomic hybridization (aCGH). The present study further characterized the EPC genomic profile in MM by comparison of EPC gene expression with that of tumor cells and control ECs using a bioinformatics approach that integrated cancer gene databases to prioritize key molecular MM biomarkers.

Methods: EPCs (> 98% vWF/CD133/KDR+/CD38–) from bone marrow aspirates of 22 untreated MM patients were outgrown on laminin-coated flasks. The fractions enriched for tumor cells were > 50% CD38+. For expression profiling, total RNA from EPCs, MM cells, and control ECs (EPCs and HUVECs) was hybridized to Affymetrix U133 Plus 2.0 arrays, and comparisons were made by ANOVA with corrections for multiple comparisons to achieve a false discovery rate < 5%. Functional enrichment of gene ontology (GO) categories was performed using DAVID (NIAID) and Ingenuity Systems software.

Results: A total of 334 genes were differentially expressed in MM EPCs versus control ECs (> 1.5-fold difference and P <. 05). Of these, 81% were over-expressed in MM EPC. Functional annotation clustering of all 334 genes into GO categories by DAVID focused similar annotations together into 16 significant clusters (P < .01). This analysis revealed top biological clusters to be differentiation and development (P = 8×10⁻⁸, 159 genes); extracellular matrix adhesion (P = 2×10⁹, 110 genes); bone formation (P = 4×10⁻⁵, 33 genes); and angiogenesis (P = 8×10⁻⁵, 35 genes). From these clusters, an expression profile was obtained consisting of the strongest 21 differentially expressed genes (versus control ECs; P < 1×10⁻⁵) characterized by highest expression of 4 genes associated with metastasis and tumor growth: COL1A1, LUM, CYP1B1, and GREM1. RT-PCR validation studies of these genes are ongoing. Moreover, there was a 34% greater overlap of MM EPC and tumor profiles versus control ECs and tumor profiles (P < .001). In comparing the 334-gene set to the tumor cell profile, 2 important EPC gene sets emerged. The first set included 158 MM EPC genes differing significantly from control ECs, but similar to tumor (Profile 1). We hypothesize that this profile is a consequence of the clonal identity previously reported between MM EPCs and tumor, and that these genes contribute to MM progression. The second profile included 39 EPC genes that differed significantly both from control ECs and tumor cells (Profile 2). This gene profile may confer a predisposition to clonal transformation of EPCs. When genes in both profiles were compared with published databases of cancer biomarkers (Campagne et al., BMC Bioinformatics 2006), significant overlaps (P < .05) were found. The largest similarities were observed with the HM200 gene list which comprises 200 genes most consistently differentially expressed in human/mouse cancers, and containing potentially key genes previously undescribed in MM, including FZD7, TWIST1, and CARD8. Current studies are aimed at integrating genes in EPC Profiles 1 and 2 with chromosomal copy number abnormalities found in EPCs, and with MM progression and survival.

Conclusions: Our results strongly indicate that EPCs are an integral part of the neoplastic process in MM. Their altered genomic profile compared to control ECs indicates pathogenic functions critical for MM evolution, including angiogenesis, adhesion, bone turnover, and cell differentiation. The high degree of commonly expressed EPC genes compared with MM and other tumors permits prioritization of candidate MM-endothelial biomarkers not yet defined in this disease.