The plant plasma membrane (PM) regulates the exchange of information and substances between the cell and its environment. Many of the functions are plant-specific, such as cell wall assembly and the response to pathogens, and involve PM-specific proteins. To date, however, only a few PM proteins have been characterized. Even in the model plant Arabidopsis thaliana, under 50 gene products have been identified among an estimated population of at least 500 polypeptides ( Masson & Rossignol 1995). In addition, these proteins fall into less than 20 multigene families of integral membrane proteins, mostly isoforms of a few extensively characterized transport systems ( Logan et al. 1997a ). Among other PM-resident proteins, there is almost no data available concerning peripheral proteins. Finally, nearly nothing is known about proteins with multiple locations in plants ( Robinson 1996), including those that are reversibly associated with the membrane after post-translational modifications ( Park et al. 1997 ;Xing et al. 1997 ).
Despite the availability of approximately 36 000 Arabidopsis ESTs in the databases, it is difficult to identify potential PM proteins. There is a clear need to develop methods to identify novel PM proteins and to link genes systematically to their products. In the last few years, approaches aimed at constructing links between proteins and the genome have become popular and led to the term ‘proteome’ to characterize the protein complement of a genome ( Wasinger et al. 1995 ). In yeast, with the availability of the complete set of genes, such proteome-based approaches immediately emerged as the most efficient way for systematic identification of gene products ( Shevchenko et al. 1996 ). Since the Arabidopsis genome sequencing will soon be complete, similar efficiency can be expected provided that the relevant proteome information has been made available.
In plants, relatively little effort has been devoted to proteomes and concerned total proteins to date ( Kamo et al. 1995 ). However, when performed on subcellular fractions, proteomes might allow identification of the intracellular location of unknown proteins. They may also reveal unexpected localizations of proteins already characterized elsewhere in the cell, as recently suggested in animals ( Fialka et al. 1997 ;Scianimanico et al. 1997 ). The aim of this work was to assess the feasibility of the proteome approach to tag PM proteins. The generated information was stored in an accessible database to constitute a resource allowing (i) direct comparison of homemade protein 2D gel maps with annotated reference gels1; (ii) recovery of the corresponding gene sequences; and (iii) identification of probable intracellular location.