A reciprocal benefit between genomics and proteomics can be expected for both fast-evolving areas. The "Proteome" was recently described as "the protein complement of the genome" (Wassinger et al. 1995), giving rise to the novel discipline of proteomics whose aim is the systematic characterisation of the proteome in a given organism, tissue, cell or subcellular compartment. The initial concept of proteomics as the identification of gene products in a given proteome, often associated with two-dimensional electrophoresis, is experiencing a rapid evolution. Not only large improvements in separation and identification methods are being reported, but also new perspectives in quantification and characterisation of protein-protein interactions are providing a multidimensional perception of how proteins execute their functions within the cellular context.
One approach to illustrate putative functions for unknown gene products is to identify their protein interacting partners or "interologs" (Walhout et al. 2000a). The establishment of a map of protein interactions ("interactome") may help establish clusters of interacting gene products. Based on this kind of protein network organisation, the identification of the protein partners for an unknown protein may assign a putative functional category to the uncharacterised protein (Tucker et al. 2001). The original development of the yeast two-hybrid system (Fields and Song 1989) has allowed the validation and standardisation of an amenable technique to characterise protein-protein interactions in living yeast cells. Two separated modules of a transcription factor, the DNA binding domain (DBD) and the transcriptional activator domain (AD) are translationally fused to interacting proteins X and Y generating the "bait" and "prey" constructs respectively. Upon X-Y interaction within the nucleus of the yeast cell, the hybrid transcription factor will be functionally reconstituted with the subsequent biological readout as the expression of reporter genes containing promoter-binding sites for the transcription factor. The strength of the idea relies on the powerful yeast molecular genetics, which allows a rapid and easy identification of coding sequences for interacting proteins when using cDNA libraries as a prey (Walhout et al. 2000b). This powerful technology also contains some drawbacks and limitations. Among the most disturbing consequences, it is the appearance of false positives due to spurious interactions the one centring most attention. Therefore any protein-protein interaction found by the yeast two-hybrid technology should be treated with caution until proven by alternative in vivo demonstrations. Also false negatives can rise due to intrinsic features of the system: improper folding, incorrect subcellular localisation, instability of hybrid proteins or absence of required post-translational modifications in yeast may lead to lack of interaction.
An immediate extension from the original application to study interactions between two known proteins was the identification of novel potential interactions at a large-scale, using either cDNA libraries fused to AD (library approach) or full length cDNAs as both "bait" and "prey" (matrix approach). These two-hybrid variants have been applied to diverse biological systems. Examples of the matrix approach extend from limited protein number in Drosophila (Finley and Brent 1994) to covering the whole proteome for the vaccinia virus (McCraith et al. 2000) and the baker's yeast (Uetz et al. 2000, Ito et al. 2001). The use of library screenings in a wide-proteome approach was initiated with the T7 phage (Bartel et al. 1996) and continued with the hepatitis C virus (Flajolet et al. 2000) and the bacteria Helicobacter pylori (Rain et al. 2001). Although the use of two-hybrid system is already a routine for Arabidopsis researchers, no global "interactome" analysis has been reported to date for plant systems. Parallel uses of both matrix and library approaches (Flajolet et al. 2000, Walhout et al. 2000a) indicate that the library strategy yields higher number of protein interactions. Another conclusion of the high-throughput assays is that large numbers of false positives preclude identification of as much as 90% of previously reported interactions. Therefore it seems evident the unavoidable request of independent validation for any reported protein-protein interaction based on the two-hybrid system.
Among the different strategies to characterise protein interactions, the use of immunological methods is widely accepted. The coimmunopurification of proteins from total protein extracts represents unequivocal demonstration of in vivo protein associations. The same antibodies can provide information regarding tissue or subcellular localisation, data that may serve as an indirect evidence for protein association if overlapping signals are immunodetected.
The availability of appropriate antibodies is undoubtedly a crucial requirement for characterisation of protein associations by immunological procedures. One alternative to the laborious and uncertain practice of raising antibodies against purified proteins is the use of epitope-tagging methodology. This approach is particularly useful for available genomic coding sequences. In the highly favourable case of efficient yeast gene replacement, the complete genome sequence encouraged the development of successful chromosomal epitope tagging (Knop et al. 1999). The concept is based on the attachment of short genomic sequences coding for antigenic peptides (named "epitopes" of 6 to 20 amino acids length), to any cDNA of interest. The addition of a peptide sequence recognisable by a pre-existing antibody to a protein under study permits its surveillance by immunological methods readily after expressing the cloned coding sequence. In addition to the speed of the method, a series of additional advantages of epitope tagging to trace any protein of interest illustrate the power of this technique. For instance, the artificially attached epitope can be considered an outsider antigen within the total protein extract of cells that do not express the tagged protein. This represents an unbeatable negative control in terms of specificity when compared to the use of antibodies against peptides from the native protein. Moreover, the use of epitope tagging permits immunological discrimination between closely related proteins without risk of cross-reaction. Since the experimenter decides the location of the epitope within the investigated protein, suitable sites to avoid potential interferences in terms of localisation or function can be envisaged. Finally, the use of epitopes facilitates mild elution condition in protein purification methods by competition with the corresponding purified peptide. A large battery of peptide-antibody combinations is commercially available. For a recent review see Fritze and Anderson (2000). This panoply of suitable epitopes has been thoroughly exploited in yeast and animal cells (Jarvik and Telmer 1998).
In plant systems, the use of epitope-tagging technique has been modest (DeWitt and Sussman 1995, Boyes et al. 1998) mainly due to restrictions for working with transgenic plants. The recent development of an Agrobacterium-mediated technology to introduce intron-tagged epitope coding sequences in cultured Arabidopsis plant cells brings hope for a wider use of epitope-tagging technology in plant systems (Ferrando et al. 2000). This technology allows the detection, purification, and subcellular localisation of epitope-tagged proteins as early as five days after transformation. In addition, the high transformation rate achieved by means of Agrobacterium transformation, licenses cotransformation experiments whereby combinations of two differentially epitope-tagged proteins can be used for protein-protein interaction studies. This technique represents a fast and reliable means for the in vivo verification of protein interactions in plant cells (Ferrando et al. 2001). Production of transgenic cell lines expressing epitope-tagged proteins is also feasible by the use of intron-tagged epitope technology, thus providing unlimited source of transformed plant material for large-scale protein purification. In combination with the generation of large amount of tagged proteins, the purification of protein complexes by affinity chromatographic methods is a powerful technique to define multiproteic complex associations (Farras et al. 2001).
In the laboratory, we are using some of these proteomic approaches to elucidate the formation of protein complexes between enzymes involved in PA biosynthesis in Arabidopsis (Panicot et al. 2002). Spd synthase SPDS2 was used as a "bait" using a cDNA library from Arabidopsis cell suspensions. SPDS2 was shown to interact in yeast with the functional homolog Spd synthase SPDS1, in addition to a novel Spm synthase named SPMS. Only heterodimers between these enzymes were found in the yeast two-hybrid system. In plant cells, in vivo evidence was demonstrated in Arabidopsis with the use of the intron-tagged epitope technique. Coexpression of hemaglutinin and c-Myc epitope-labelled proteins confirmed the presence of coimmunoprecipitating SPDS1-SPDS2 and SPDS2-SPMS heterodimers. In addition, the epitope-labeled proteins copurified associated to protein complexes in the range of 650-700kD. We have therefore suggested the formation of a metabolon involving at least the last two-steps of PA biosynthesis in Arabidopsis (Panicot et al. 2002). Further analysis of the identified protein complexes by mass spectrometry is expected to provide information about yet unknown regulatory subunits of SPDS-SPMS metabolon in the PA biosynthesis pathway in Arabidopsis. The use of proteomic techniques to unravel the formation of protein complexes in plant secondary metabolism is in progress. As an example, similar proteomic approaches have been employed to unravel the multi-enzymatic association in the flavonoid biosynthetic pathway (Burbulis and Winkel-Shirley 1999). By means of two-hybrid complemented with coimmunoprecipitation and protein retention studies, interactions among chalcone synthase (CHS), chalcone isomerase (CHI) and dihydroflavonol reductase (DFR) have been described.
The full potential of this kind of approaches is achieved with the use of physico-chemical methods for peptide and protein identification. These combined approaches of affinity purification and protein identification by means of mass-spectrometry have been implemented for the characterisation of large multiproteic complex associations, as the spliceosome and the proteasome complexes (Neubauer et al. 1997, Verma et al. 2000). Compared to yeast or other eukaryotic systems, the proteomic analysis for plant systems is still in its infancy (van Wijk 2001). The most likely reason is the delayed completion of genome projects that are essential for protein identification. The fully sequenced Arabidopsis genome is giving boost to proteomic studies for higher plants, and the first reported proteomic analysis have appeared (Gallardo et al. 2001, Kruft et al. 2001, Peltier et al. 2001). Although limited, due to lack of complete genome sequences, studies in other plant systems are also feasible (Peltier et al. 2000, Huber et al. 2001).
In addition to the studies of physical protein interactions, the evidence of subcellular protein colocalisation can serve as a support to elucidate the formation of protein complexes. In this context, the epitope-tagging technology can also be exploited (Ferrando et al. 2000). The target proteins may also be fused to reporters such as fluorescent polypeptides to facilitate their cellular localisation (Quaedvlieg et al. 1998). A remarkable advance for protein interaction studies is the implementation of the fluorescence resonance energy transfer (FRET) technology as detailed by Kenworthy (2001) where the development of fluorescent proteins together with advances in confocal microscopy allow in vivo imaging of fusion-protein associations (Mas 2000, Kato et al. 2002).
Was this article helpful?