Direct delivery of DNA into embryogenic pollen was used to produce transgenic plants in tobacco. A plasmid bearing the ß-glucuronidase (GUS) marker gene in fusion with the 35S-promoter was introduced by microprojectile bombardment into mid-binucleate pollen of Nicotiana tabacum that had been induced to form embryos by a starvation treatment. In cytochemical expression assays, 5 out of 10(4) pollen grains were GUS(+). Visual selection by staining with a non-lethal substrate for GUS was used to manually isolate transformed embryos. From the initial population of embryogenic GUS(+) pollen, 1-5% developed into multicellular structures and 0.02% formed regenerable embryos. Two haploid transformants were regenerated. GUS expression was detected in different parts of the plants, and Southern analysis confirmed stable integration of the foreign DNA. Diploidisation was induced by injection of colchicine into the stem near adventitious buds. Offspring from selfings and backcrosses of one transformant were tested for GUS expression and by Southern blots. All F1-plants were transgenic, in accordance with Mendelian inheritance.

A transgenic reconstruction experiment has been performed to determine the feasibility of male gametophytic selection to enhance transmission of genes to the next sporophytic generation. For tobacco pollen from a transgenic plant containing a single hygromycin-resistance (hygromycin phosphotransferase, hpt-) gene under control of the dc3 promoter, which is active in both sporophytic and gametophytic tissues, 3 days of in vitro maturation in hygromycincontaining medium was sufficient to result in a 50% reduction of germinating pollen, as expected for meiotic segregation of a single locus insert. Pollination of wild-type plants with the selected pollen yielded 100%o transgenic offspring, as determined by the activity of the linked kanamycin-resistance gene-present within the same transferred T-DNA bordersunder control of the nos promoter. This is direct proof that selection acting on male gametophytes can be a means to alter the frequency of genes in the progeny.The plant life cycle consists of two distinct phases, the diploid sporophytic"and the haploid gametophytic phase. In angiosperms, the haploid phase is reduced, and the male and female gametophytes are enclosed in the respective sex organs of the sporophyte. Male and female gametophytes (i.e., pollen and embryo sacs) are formed from haploid spores, the direct products of meiosis, and the sporophytic phase starts again after the fusion of the two pollen-contained sperms with the egg and polar cells of the embryo sac in a double fertilization event. In spite of the large morphological and physiological differences, a large overlap of gene expression has been demonstrated between the sporophytic and gametophytic phase on the protein and mRNA level (for reviews, see refs. 1 and 2).These data provided the theoretical basis for the early pollen selection experiments (for review, see ref.2). The idea was that if such an overlap exists, it should be possible to exert selection pressures on pollen to increase the frequency of responsive genes in the subsequent sporophytic generation. Angiosperm pollen has some features that makes pollen selection a very attractive method of plant breeding: large population size, haploid nature of the cells, and independence from the maternal plant (3).A variety of selection pressures such as drugs, heat, water, cold temperature, and other stresses have been applied to mature pollen just before pollination, and nonstressed plants have been pollinated with the selected pollen in anticipation of obtaining progeny in which an increased frequency of resistant individuals can be found (for review, see ref. 4). Despite a few positive reports (5-9), the feasibility of pollen selection is still under question. This is mainly due to the problem of discriminating between gametophytic and sporophytic effects on pollen development (resulting from pollen-and antherexpressed genes, respectively; refs. 2 and 9) and a lack of appropriate selection schemes. The developmental windowThe publication costs of this article were defrayed in part by...