Analyzing myeloid differentiation on multiple fronts

Like most intricate biologic processes, the regulation of hematopoietic differentiation is poorly understood. It seems clear that focusing on individual gene products or taking advantage of individual technologies is not adequate to elucidate the intricate regulatory networks responsible for the control of blood cell production. Rather, as shown for less-complex regulatory circuits (such as bacterial chemotaxis), an understanding of biologic regulation requires a global approach. Such an approach begins with the comprehensive identification of regulatory network components and culminates with the assembly of such a molecular “parts list” into interactive networks and circuits. The behavior of these networks often displays non–intuitively obvious properties such as robustness or a relative resistance to potentially deleterious perturbations. A description of how a regulatory network evolves over time and as a function of biologic change is also necessary as a foundation of mechanistic understanding.

Lian and colleagues (page 513) have begun to provide such a foundation for the process of myeloid differentiation using a well-characterized cell line model system. In this system differentiation can be induced in a controlled and relatively synchronous manner. Therefore, as differentiation proceeds it is possible to measure sequential changes in the molecular differentiation program. Wisely, the authors have analyzed gene expression changes at the level of RNA transcripts and at the level of proteins. In addition, 2 very different methodologies for measuring transcript levels were employed together with sophisticated bioinformatic analyses. To date, these studies are the only examples where global transcript and protein level changes have been analyzed during the progression of a differentiation program. Although the data so far have not profoundly changed our view of myeloid differentiation, they are a wonderful example of the kinds of global approaches that will be necessary to approach complex biologic questions in the postgenome era. As such, the information collected by these investigators and provided online to the scientific community represents a “living resource” that will ultimately be integrated with numerous other data sets and newly emerging computational modes of analysis.

Finally, the apparently poor correlation between transcript and protein levels highlights the importance of posttranscriptional regulatory mechanisms and suggests that a degree of caution is necessary when interpreting microarray transcript analyses as indicators of functional gene-product levels. In a particularly relevant situation, while transcript profiling may be extremely useful in categorizing hematopoietic tumors, the selection of potential therapeutic targets will require a combination of genomic and proteomic strategies.