The author is in the Department of Orthopaedics and Rehabilitation, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06510. osteoblast-like cell lines activated with IL-1 and TNF-a all secrete IL-6 (8). The secretion of IL-6 induced in this way can be inhibited by treating the cells in vitro with E2 and, to a lesser extent, with testosterone and progesterone. Thus, cytokine secretion by authentic bone cells can be down-regulated by the addition of E2 directly to osteoblasts and stromal cells, a finding that is consistent with the expression of E2 receptors by osteoblasts (9). Furthermore, these data predict that one target for E2 is either osteoblasts or their precursors, possibly stromal cells, in contrast to the data described above, which suggest that the monocyte is the target. However, some investigators have been unsuccessful in reproducing the inhibitory effect of E2 on IL-6 secretion in vitro (10), possibly due to differ-ences among cell preparations or culture conditions. The observation that other steroid hormones also inhibit cytokine secretion suggest the presence of specific steroid DNA binding sites (steroid response elements), which confer hormone sensitivity to target genes. However, no steroid response elements, including those for E2, are apparent in the IL-6 promotor. It is possible that other tran-scription regulators may interact with various steroid receptors and that nonsteroid hormone response elements are involved. The IL-6 made by these stromal cells and osteoblasts is crucial in the recruitment arm of the bone remodeling circuit, that is, in causing the differentiation of osteoclast precursors to mature osteoclasts. Indeed, the stimulatory effect of TNF-a on osteoclast formation was inhibited by either E2 or a neutralizing antibody to IL-6. Thesefindings and the observation that IL-6 has colonystimulating activity reinforce the impor-tance of IL-6 in osteoclast differentiation (11). Evidence from experiments in whole animals is also consistent with this conclusion. The absence of E2 in ovariectomized mice leads to an increase in colony-forming units for granulocytes and macrophages, enhanced osteoclast development in vitro in bone marrow cell cultures, and an increased number of osteoclasts in vivo (12). These changes were inhibited by E2 or an antibody to IL-6. Thus, osteoblast precursors, possibly stromal cells, respond to the loss of E2 by secreting IL-6, which then induces osteoclastogenesis (Fig. 2). Although the studies with isolated cells suggest that a primary effect of E2 is to act on osteoblasts and stromal cells to modulate IL-6 secretion, they do not tell the whole story. Clearly, there are also effects ofE2 on the secre-tion of IL-1 and the other cytokines by PBM or similar cells (described above). These two separate effects of estrogen can be reconciled: An increase in IL-6 secretion from osteoblasts could result from a primary effect of E2 on IL-i or TNF-a production by PBM, since each of these factors potently induce IL-6. Or E2 could affect both PBM (monocyte-macrophage cells) and the osteoblast, each of which responds independently. The important implications of these in-vestigations are that the anti-osteoporotic effect of E2 may be explained by its fundamental ability to interact with bone cells and regulate the cytokine circuitry that controls bone remodeling. In the E2 replete state, the hormone functions as a governor to reduce cytokine production and thereby temper the rate or extent of osteoclast formation and activity. This represents normal remodeling. In the E2-deficient state, the governor is lost, allowing for increased cytokine secretion, leading to more osteoclast formation and in-creased bone resorption. Although the pre …