Although we cannot exclude the possibility that the expression of the transgene at the low level promoted bone formation at 4 months of age, this seemed to be unlikely because overexpression of BCL2 impaired osteoblast differentiation in a manner dependent on the expression levels of the transgene in vivo and in vitro, osteoblast density was Pimozide similar between BCL2 transgenic mice and wild-type mice at 4 months of age, and cortical bone in BCL2 transgenic mice was not further increased at 6 months of age irrespective of the similar level of transgene expression at 4 
and 6 months of age. The disturbed osteocyte network was gradually restored after 4 months of age, probably because the level of the transgene expression was not sufficient to reduce the number of osteocyte processes. These seemed to be the reasons why the enhanced bone formation was observed only at 4 months of age. Thus, our findings suggest that the reductions in the numbers of osteocytes and their processes and the accumulation of TUNEL-positive lacunae were followed by the activation of osteoblast function, leading to an increase in bone formation. As the distribution of Sost protein was interrupted in osteocytes and b-catenin protein was increased in osteoblasts in BCL2 transgenic mice at 4 months of age, the activation of Wnt signaling in osteoblasts by the reduction of disseminated Sost protein seems to be one of the causes for the increase in bone formation in BCL2 transgenic mice at 4 months of age. In mice with osteocyte ablation by diphtheria toxin, the suppression of bone formation seemed to be due to maturational inhibition of osteoblasts, which was shown by the reduction in Diperodon osteocalcin expression, and maturational inhibition may have been caused by a necrosisinduced inflammatory reaction. The frequencies of TUNEL-positive lacunae in the trabecular bone of BCL2 transgenic mice were less than those in the cortical bone, probably because trabecular bone is more extensively remodeled than cortical bone and the dead osteocytes in the trabecular bone are rapidly replaced with live osteocytes. Irrespective of the relatively low frequency of TUNEL-positive lacunae, however, the increase of bone formation and unresponsive to unloading were observed in the trabecular bone at 4 months of age. Therefore, the reduction in the number of osteocyte processes in addition to the reduction in the number of osteocytes may be sufficient for the disturbance of osteocyte network in trabecular bone. The number of osteoclasts was reduced in the cortical bone but not in the trabecular bone of BCL2 transgenic mice at 4 months of age. It may indicate that osteoclastogenesis in cortical bone is more dependent on the osteocyte network than that in trabecular bone, because the number of osteoclasts was already reduced in the cortical bone in BCL2 transgenic mice at 5�C6 weeks of age, when the frequency of TUNEL-positive lacunae was about 20%, an equivalent value detected in the trabecular bone at 4 months of age. The augmented function of the osteocyte network by unloading may be required for the enhancement of osteoclastogenesis by the osteocyte network in trabecular bone. Our model also may explain why exercise increases bone mass, which has been indicated by many clinical studies, because osteocytes decrease the inhibitory effects on bone mass by reducing the inhibitory effect on osteoblast function and the stimulatory effect on osteoclastogenesis in the loaded condition compared with the unloaded condition.