Let me summarize by reviewing a model which is meant to raise as many questions as it answers (Fig. 2). What I have discussed today are data suggesting that during progression of solid tumors, like colon cancer, an increased cellular DNA methylating capacity characterizes the initial stages of multi-clonal hyperplasia. Despite this increase, the altered pattern of DNA methylation which subsequently emerges is largely manifest by a widespread hypomethylation of DNA. However, on a more regional basis, areas of hypermethylation appear which can affect strategic areas such as normally unmethylated CpG islands. These shifted DNA methylation patterns have the capacity to both follow, or cause, chromatin changes that can both directly silence genes critical for normal cell maturation--and/or participate in the structural chromosome changes which constitute genetic instability during tumor progression (Fig. 2). I suggest that one must view these changes as an interchangeable cycle of events during tumor progression. The chromatin changes and abnormal methylation patterns can drive one another with increasingly deleterious effects as the malignant phenotype emerges (reviewed in Baylin, 1991). What are the molecular events that would initiate the above dynamics? A working construct model is shown in Fig. 3. As discussed for the normal adult cell, there is a delicate balance between the strategic location of DNA MTase, regulation of this enzyme, and rate of DNA synthesis at replication forks (top panel, Fig. 3). In pre-neoplastic and cancer cells, perhaps failure of cells to exit the cell cycle and halt DNA replication, facilitates some sort of pressure to increase cellular DNA methyltransferase activity (bottom panel, Fig. 3). This increase may involve loss of feedback inhibition of the enzyme during the post DNA replication phase. There are also probable structural alterations in the nucleus which may alter the geographic relationship between the DNA replication fork and DNA MTase. In consequence, many DNA areas that should be getting methylated do not, and novel areas of methylation also arise. This cycle of events leads to the imbalance of DNA methylation that I have talked about. Future investigations of these possibilities, and of their specific consequences for alterations of gene expression and chromosome structure, may reveal a key molecular step underlying virtually all stages of tumor progression.