Most previous studies of the 0-phaseolin (phas) gene, which encodes the major storage protein in bean (Phaseolus vulgaris L.), have shown its expression to be rigorously confined t o the developing seed, both i n bean and transgenic tobacco (Nicotiana fabacum L. cv Xanthi) plants. To confirm unequivocally the lack of phas expression in vegetative tissues, we placed the diphtheria toxin A-chain (DT-A) coding region under the control of P-phaseolin promoter sequences. Tobacco plants transgenic for phas/DT-A were phenotypically normal until flowering, when they produced anthers that were externally normal but contained no viable pollen. Microscopic examination of immature anthers revealed a normal tapetum, but the pollen mother cells did not undergo meiosis and subsequently degenerated, resulting in male-sterile plants. This demonstration of phas expression during microsporogenesis was corroborated by the expression of 0-glucuronidase in pollen of plants transformed with comparable phas/uidA constructs. Although these findings suggested that similarities i n phas expression may exist between seed and pollen maturation, no phas activity could be detected in bean pollen. After fertilization of the DT-Atransformed plants with pollen from wild-type tobacco, 50% of the resulting embryos aborted at the heart stage, defining this as the earliest time for phas expression during embryogenesis.

The effect of flanking matrix attachment regions (MARs) on homology-dependent trans-silencing was tested using two strong trans-silencing loci. The transgenic tobacco line 271 carries at a single locus a p35S-RiN-tNos transgene which is able to silence, in trans and at the transcriptional level, the expression of any p35S-driven transgene irrespective of its position. The transgenic tobacco line 6b8 carries at a single locus a p35S-uidA-tRbcS transgene which is able to silence in trans, at the post-transcriptional level, the expression of any uidA-expressing transgene irrespective of its position. Various transgenic tobacco lines carrying a target p35S-uidA-tNos transgene, flanked on each side by MARs from chicken, bean, yeast or tobacco, were crossed with lines carrying the 271 and 6b8 loci. Expression of the target transgene was silenced in all hybrids, irrespective of the presence or absence of MAR sequences. These results therefore demonstrate that MARs are not able to protect transgene expression from strong silencing loci that act in trans.