Howard Y. Chang

Research/Lab website:   Chang Lab, Stanford University

We are interested in two fundamental questions in epithelial biology: (1) the basis of positional identities in epidermal structures throughout the body, and (2) how those signals and boundaries may be abrogated to allow cancer metastasis. Although site-specific differences in epidermal structures (such as hairs on skin) on different anatomic sites are easily appreciated and are the basis of many diagnostic and treatment strategies in skin diseases, embryologic transplantation experiments have demonstrated that it is the underlying mesenchymal stroma that dictates the epithelial fates that develop. We have begun to define the organizational and developmental principles of stromal cells based on their global gene expression programs. For example, we discovered that endothelial cell (EC) diversity is primarily dictated by the vessel size of origin, and the Notch-Hey2 and left-right polarity signaling axis help to specify artery vs. vein fate. In contrast, fibroblasts from each anatomic site exhibit systematic and characteristic differences in gene expression and retain the embryonic anatomic expression pattern of Hox genes. Fibroblasts are thus excellent candidates as the bearer of positional memory in tissues and organs.

The unanticipated diversity and precision of fibroblast differentiation suggest much richer roles for stroma cells during development and diverse disease processes. Taking advantage of the unique capacities of skin for primary cell culture, gene transfer, and tissue reconstitution of site-specific features, we are pursuing the mechanisms by which the Hox code in fibroblasts specify epidermal fates, and how the embryonic Hox code is maintained in isolated adult fibroblasts.

In contrast to the orderly acquisition of positional identities in development, cancer cells can abrogate and override the positional cues in tissues and organs as they metastasize. We discovered that one way that cancer cells may accomplish this feat is by activation of an emergency “wound healing” program. The conserved genomic response of fibroblasts to serum is a wound-like gene expression signature that allows one to visualize the wound-like features in cancer samples. The genes that comprise the wound healing program include those that activate matrix synthesis, cell migration, angiogenesis, and cell survival—ideal genetic tools for cancer invasion and metastasis. We found that the wound signature is a universal and powerful prognostic predictor of metastasis in many human tumors. The wound signature thus presents previously unappreciated avenues for cancer diagnosis and therapy.

We are using a combination of computational modeling, functional genomics, molecular genetics, and animal models to dissect the wound signature. We seek to identify upstream regulators that turn on the wound-like program in cancers, the key effector genes that enhance metastasis, and how their actions can be blocked therapeutically. We are also investigating practical strategies for applying and interpreting the wound signature in real life clinical scenarios.