EDITORIALS Tissue Microarrays for Hypothesis Generation

Decades of hypothesis-driven research have led to the discovery of oncogenes and tumor suppressors that affect carcinogenesis. With the advent of powerful genomic and proteomic microarray techniques, a different scientific paradigm is emerging—one in which hypotheses are generated based on the evaluation of global gene expression in cells or tissues. Tissue microarrays are also useful to interrogate expression, at the mRNA or protein level, of a single or small set of gene products in normal, preneoplastic, or malignant tissues. In this issue of the Journal, Gurrieri et al. (1) used tissue microarrays to analyze comprehensively the mRNA and protein expression profiles of the tumor suppressor PML in diverse cancers of hematopoietic and solid tumor origin. They found frequent loss of PML protein but not of PML mRNA in cancers from multiple histopathologic sites. In a few cases, PML mutations or polymorphisms were identified. However, neither these changes nor loss of heterozygosity (LOH) could account for the loss of PML protein. These findings and those obtained after treatment with a proteasome inhibitor, point to a post-transcriptional mechanism for this PML regulation. PML is the most common translocation partner of the retinoic acid receptor alpha (RAR ) found in acute promyelocytic leukemia (APL) rearrangements (2–4). Originally called Myl (5), it was renamed PML when its cDNA sequence was reported (6,7). PML is an integral component of nuclear structures known as PML nuclear bodies that are disrupted in APL patients expressing the PML–RAR translocation product (8,9). Early work on PML highlighted its role as a growth-suppressive species in APL and potentially other tumor cell contexts (10,11). PML expression is repressed in small-cell lung cancers (12) as well as in other cancers (13). These findings were built on by in vitro studies and by engineering PML null mice, which turned out to be prone to tumor formation; PML null cells were resistant to apoptotic stimuli mediated through p53 and other tumor suppressors that could affect genomic stability (14–22). Loss of PML function impaired cellular senescence despite oncogenic signals (3). This work set the stage for comprehensive examination of PML expression and possible PML structural alterations in carcinogenesis, as in the study conducted by Gurrieri et al. (1). Such studies could implicate a functional role for PML in cancers beyond APL. Gurrieri et al. should be commended for undertaking these studies using tissue microarrays to explore patterns of PML mRNA and protein expression in non-APL hematopoietic and solid tumors. The findings confirmed and extended prior work (6–22) by excluding LOH as the cause of repression of PML protein in the tumor cells examined. Because PML mRNA expression was intact, the authors sought post-transcriptional mechanisms for instability of PML protein. This work is an example of using tissue microarrays for hypothesis generation. On the basis of the results from the tissue microarrays, the authors proposed an intriguing mechanism for PML regulation that involves the ubiquitin–proteasome degradation pathway. Treatment with a proteasome inhibitor restored PML protein expression in cell lines that expressed PML mRNA but not PML protein. This result is not surprising because post-translational modification by Sumo-1, the ubiquitin-like modifier, affects PML nuclear bodies (3). It is interesting to note that previous gene profiling studies revealed that the E1-like ubiquitinactivating enzyme UBE1L can be induced by all-trans-retinoic acid treatment of APL cells (23). UBE1L is a retinoic acid target gene that promotes apoptosis and PML–RAR degradation (24). Activation of the ubiquitin–proteasome degradation pathway also occurs during retinoid-induced tumor cell differentiation and chemoprevention (4). Cell cycle regulators were identified as molecular targets for these pharmacologic effects. PML protein may be directly affected by UBE1L or regulators of the proteasome degradation pathway, and future work should explore this possibility. Additional studies are needed to establish whether a proteasome-dependent mechanism was predominantly responsible for repression of the PML protein in the cancers studied by Gurrieri et al. (1). Nevertheless, there are therapeutic implications to finding that regulation of PML was proteasomedependent. Proteasome inhibition has clinical activity (25). Clinical use of a proteasome inhibitor might restore PML expression and thereby confer its activity as a tumor suppressor. Although the currently available proteasome inhibitors do not target PML specifically, clinical treatment with these agents would address whether PML expression can be restored in malignant tumors. The data in Gurrieri et al. (1) are consistent with an important role for PML repression in regulating the growth of cancers in patients. PML repression occurred early during prostate carcinogenesis, even at the transition from prostatic intraepithelial neoplasia to invasive prostate carcinoma. Statistically significant associations were also observed for PML repression in breast and central nervous system cancers. These data would have been strengthened by inclusion of paired normal and malignant tumor specimens from the same patients. Yet, these findings do suggest a regulatory role for PML in the growth of certain tumors. Future work should establish whether restoration of PML expression has therapeutic benefits. It is also important to consider other post-transcriptional mechanisms for PML silencing.