The early stage of pollen chamber development in ovule and the cytological mechanism of nucellar cell death were studied in Ginkgo biloba L. DNA ladder appearance and TUNEL assay demonstrated that the nucellar cell death, doomed to bring about pollen chamber formation, was a process of programmed cell death (PCD). A spatial distribution of PCD was visualized during the development of pollen chamber. Together with the observation under the scanning electron microscope, these results have revealed that the early developmental pattern of pollen chamber consists of four phases. Firstly, several layers of the nucellar cells at the micropylar end elongate longitudinally. Thereafter, the uppermost layer of the nucellar cells at the micropylar end initiate PCD; and the nucellar cell death extends in a basally and laterally oriented direction to form a cavity. Finally, the epidermal cells at the micropylar end detach from the other epidermis by dehiscence, bringing about the opening of the pollen chamber. The early development of pollen chamber begins sometime after the stage of megasporocyte and continues by the time of the formation of megaspore tetrad, and finally completes at the stage of development of female gametophyte. This shows a synchronous development of megaspore and pollen chamber.
The early stages of graft union, when male branch was grafted onto female branch in Ginkgo biloba L. by cleft graft, have been observed under light microscope in order to determine the origin of callus cells between the stock and scion. Pith parenchyma cells near the graft interface were the earliest cells in response to such method of grafting. These cells de differentiated and then divided within 7～12 days after grafting. A large number of callus cells extended from the pith into the space between the graft interface linking the stock with the scion about 18～20 days after grafting; and then continued to proliferate and extend outwards along the space. Cambium cells and immature vascular tissue near the graft interface dedifferentiated into callus rather late. The over all link between the stock and scion was completed in the sites 30 days after grafting. Callus cells were also produced from cortical parenchyma cells, but they were much limited in quantity. In conclusion, the graft interface may be considered as a “natural culture bed” after grafting, in which all undamaged, living cells are capable of dedifferentiation and producing callus cells for compatible graft union. In the case of G. biloba (male/female) it were the pith parenchyma cells that appeared first to form the callus cells and later extend to link the stock with the scion.
