What unleashes aberrant gene regulation in acute promyelocytic leukemia?

In this issue of Blood, Yuan and colleagues report on an extensive transcriptional analysis of the normal differentiation of murine myeloid cells in culture and compare these gene expression patterns with those of mice engineered to express the PML-RARα oncoprotein. They found that the patterns of gene expression of normal and malignant promyelocytes were quite similar. Major shifts in gene expression occurred only when frank leukemia developed.

The PML-RARα protein is an aberrant form of the retinoic acid receptor (RAR) that recruits corepressor complexes to inhibit gene transcription. Repression is relieved by pharmacologic doses of retinoic acid that release corepressor complexes and facilitate degradation of the oncoprotein. Cell line models showed that PML-RARα represses genes critical for myeloid differentiation (C/EBPβ, C/EBPϵ, and PU.1) and stimulates proliferation genes such as cyclin A1, γ-catenin, and notch.¹  These results suggest that PML-RARα leads to both the differentiation block of acute promyelocytic leukemia (APL) as well as increased self-renewal of leukemic cells.

Yuan and colleagues used microarrays to document the transcriptional program of the murine APL model and contrasted this to that of normal myeloid differentiation. The malignant promyelocytes resembled normal promyelocytes in their general pattern of gene expression; however, about one-third of genes characteristic of mid-stage myeloid differentiation were significantly deregulated. Mining these data, the authors deduced that many of the global changes of gene expression might be attributed to inappropriate overexpression of the transcription factors Fos, Jun, and Egr-1, all known as genes rapidly induced by serum. How and why these genes are up-regulated in murine APL is uncertain. PU.1 positively regulates c-Jun; given that PML-RARα down-regulates PU.1, decreased rather than increased Jun levels would be expected.²  Furthermore, Fos, Jun, and Egr-1 stimulate myeloid differentiation,^2,3  and their up-regulation in differentiation-blocked leukemic cells seems counterintuitive.

In mouse models of APL, the first manifestation of the PML-RARα oncoprotein is a modest expansion of the myeloid compartment, and a decrease in granulocyte marker expression. Most aspects of normal hematopoiesis can persist in the presence of the transgene. Only after a latent period, which can be shorted by coexpression of other oncogenes, do mice develop APL. The most striking finding of the work of Yuan and colleagues arose from the analysis of preleukemic promyelocytes. The gene-expression profile of these cells was almost identical to that of normal promyelocytes. This presents a huge continuing paradox in the field. While in cell lines PML-RARα readily blocks differentiation and alters gene expression; if fails to do so in mice until a presumed second hit has occurred. However, the differentiation of murine APL induced by retinoic acid indicates that PML-RARα is important not only in disease initiation but in maintenance of the malignant phenotype.

Why can't PML-RARα initiate major genotypic and phenotypic changes on its own? Prior work by the authors suggested that proteolytic cleavage of PML-RARα, perhaps altering its function, is required for leukemia development.⁴  A second possibility is that the shifts in gene expression demonstrated in murine APL are unrelated to PML-RARα. PML-RARα might evoke relatively subtle changes in cell survival or self-renewal, while the second lesion fixes the differentiation block. Flt3 mutations, common in APL, cooperate with PML-RARα to induce leukemia and could activate early growth response genes. Alternatively, recurrent cytogenetic anomalies found in murine APL might explain altered gene expression.⁵  A hybrid explanation is that in vivo, PML-RARα function depends on the action of another gene altered during disease evolution. Cell lines in which PML-RARα can readily inhibit gene expression might already contain such mutations.

Further resolution of this issue can be obtained by determining the cause of the preleukemic-to-leukemic transition in mice (for example, searching for kinase mutations found only in leukemic cells). It will also be important to determine if RAR target genes are indeed repressed in the presence of PML-RARα in preleukemic versus leukemic cells, and whether chromatin precipitation experiments can demonstrate the occupancy of the oncoprotein on such promoters. These data show that our understanding of the pathogenesis of APL remains incomplete due, in part, to a reliance on cell-line models. Animal models of leukemia remain powerful tools that will both challenge and inform our understanding of disease.