hAG-2 and hAG-3, human homologues of genes involved in differentiation, are associated with oestrogen receptor-positive breast tumours and interact with metastasis gene C4.4a and dystroglycan.

hAG-2 and hAG-3 are recently discovered human homologues of the secreted Xenopus laevis proteins XAG-1/2 (AGR-1/2) that are expressed in the cement gland, an ectodermal organ in the head associated with anteroposterior fate determination during early development. Although the roles of hAG-2 and hAG-3 in mammalian cells are unknown, both proteins share a high degree of protein sequence homology and lie adjacent to one another on chromosome 7p21. hAG-2 mRNA expression has previously been demonstrated in oestrogen receptor (ER)-positive cell lines. In this study, we have used real-time quantitative RT - PCR analysis and immunohistochemistry on tissue microarrays to demonstrate concordant expression of hAG-2 and hAG-3 mRNA and protein in breast tumour tissues. Tumour expression of both genes correlated with OR (hAG2, P=0.0002; hAG-3, P=0.0012), and inversely correlated with epidermal growth factor receptor (EGFR) (P=0.003). Yeast two-hybrid cloning identified metastasis-associated GPI-anchored C4.4a protein and extracellular alpha-dystroglycan (DAG-1) as binding partners for both hAG-2 and hAG-3, which if replicated in clinical oncology would demonstrate a potential role in tumour metastasis through the regulation of receptor adhesion and functioning. hAG-2 and hAG-3 may therefore serve as useful molecular markers and/or potential therapeutic targets for hormone-responsive breast tumours.

Protein interaction domain mapping for the selection of validated targets and lead compounds in the anti-infectious area.

Despite considerable progress in the analysis of microbial genomes, the number of validated targets suitable for the development of drugs acting on agents causing infectious diseases remains modest. The diversity of new chemical entities specific for such targets has almost not increased over the last years, while resistance to anti-infectious drugs has, in contrast, become a real threat in certain surroundings. New strategies are thus needed for selecting novel validated targets. We discuss here a combined approach which uses protein interaction mapping as the basic strategy to identify interacting domains which then serve to validate newly identified targets. The interactome of Helicobacter pylori is used as model to successively describe high-throughput protein interaction mapping, use of the H. pylori data to predict the interactome from other bacteria, analysis of interacting domains, and evaluation of the capacity of one such domain to block synthesis of flagella. The general applicability of this approach to target identification and validation, and to development of novel compounds is also discussed.

Characterization of the interaction partners of secreted proteins and chaperones of Shigella flexneri.

The type III secretion (TTS) system of Gram-negative pathogenic bacteria is composed of proteins that assemble into the TTS machinery, proteins that are secreted by this machinery and specific chaperones that are required for storage and sometimes secretion of these proteins. Many sequential protein interactions are involved in the TTS pathway to deliver effector proteins to host cells. We used the yeast two-hybrid system to investigate the interaction partners of the Shigella flexneri effectors and chaperones. Libraries of preys containing random fusions with fragments of the TTS proteins were screened using effectors and chaperones as baits. Interactions between the effectors IpaB and IpaC and their chaperone IpgC were detected by this method, and interaction domains were identified. Using a His-tagged IpgC protein to co-purify truncated IpaB and IpaC proteins, we showed that the chaperone-binding domain was unique and located in the N-terminus of these proteins. This domain was not required for the secretion of recombinant proteins but was involved in the stability of IpaC and instability of IpaB. Homotypic interactions were identified with the baits IpaA, IpaB and IpaC. Interactions between effectors and components of the TTS machinery were also selected that might give insights into regulation of the TTS process.

Identification of the Helicobacter pylori anti-sigma28 factor.

Flagellar motility is essential for colonization of the human gastric mucosa by Helicobacter pylori. The flagellar filament is composed of two subunits, FlaA and FlaB. Transcription of the genes encoding these proteins is controlled by the sigma28 and sigma54 factors of RNA polymerase respectively. The expression of flagellar genes is regulated, but no sigma28-specific effector was identified. It was also unclear whether H. pylori possessed a checkpoint for flagellar synthesis, and no gene encoding an anti-sigma28 factor, FlgM, could be identified by sequence similarity searches. To investigate the sigma28-dependent regulation, a new approach based on genomic data was used. Two-hybrid screening with the H. pylori proteins identified a protein of unknown function (HP1122) interacting with the sigma28 factor and defined the C-terminal part of HP1122 (residues 48-76) as the interaction domain. HP1122 interacts with region 4 of sigma28 and prevents its association with the beta-region of H. pylori RNA polymerase. Thus, HP1122 presented the characteristics of an anti-sigma28 factor. This was confirmed in H. pylori by RNA dot-blot hybridization and electron microscopy. The level of sigma28-dependent flaA transcription was higher in a HP1122-deficient strain and was decreased by the overproduction of HP1122. The overproduction of HP1122 also resulted in H. pylori cells with highly truncated flagella. These results demonstrate that HP1122 is the H. pylori anti-sigma28 factor, FlgM, a major regulator of flagellum assembly. Potential anti-sigma28 factors were identified in Campylobacter jejuni, Pseudomonas aeruginosa and Thermotoga maritima by sequence homology with the C-terminal region of HP1122.

Protein--protein interaction maps: a lead towards cellular functions.

The availability of complete genome sequences now permits the development of tools for functional biology on a proteomic scale. Several experimental approaches or in silico algorithms aim at clustering proteins into networks with biological significance. Among those, the yeast two-hybrid system is the technology of choice to detect protein-protein interactions. Recently, optimized versions were applied at a genomic scale, leading to databases on the web. However, as with any other 'genetic' assay, yeast two-hybrid assays are prone to false positives and false negatives. Here we discuss these various technologies, their general limitations and the potential advances they make possible, especially when in combination with other functional genomics or bioinformatics analyses.

The protein-protein interaction map of Helicobacter pylori.

With the availability of complete DNA sequences for many prokaryotic and eukaryotic genomes, and soon for the human genome itself, it is important to develop reliable proteome-wide approaches for a better understanding of protein function. As elementary constituents of cellular protein complexes and pathways, protein-protein interactions are key determinants of protein function. Here we have built a large-scale protein-protein interaction map of the human gastric pathogen Helicobacter pylori. We have used a high-throughput strategy of the yeast two-hybrid assay to screen 261 H. pylori proteins against a highly complex library of genome-encoded polypeptides. Over 1,200 interactions were identified between H. pylori proteins, connecting 46.6% of the proteome. The determination of a reliability score for every single protein-protein interaction and the identification of the actual interacting domains permitted the assignment of unannotated proteins to biological pathways.

Identification of novel Saccharomyces cerevisiae proteins with nuclear export activity: cell cycle-regulated transcription factor ace2p shows cell cycle-independent nucleocytoplasmic shuttling.

Nuclear export of proteins containing leucine-rich nuclear export signals (NESs) is mediated by the NES receptor CRM1/Crm1p. We have carried out a yeast two-hybrid screen with Crm1p as a bait. The Crm1p-interacting clones were subscreened for nuclear export activity in a visual assay utilizing the Crm1p-inhibitor leptomycin B (LMB). This approach identified three Saccharomyces cerevisiae proteins not previously known to have nuclear export activity. These proteins are the 5' RNA triphosphatase Ctl1p, the cell cycle-regulated transcription factor Ace2p, and a protein encoded by the previously uncharacterized open reading frame YDR499W. Mutagenesis analysis show that YDR499Wp contains an NES that conforms to the consensus sequence for leucine-rich NESs. Mutagenesis of Ctl1p and Ace2p were unable to identify specific NES residues. However, a 29-amino-acid region of Ace2p, rich in hydrophobic residues, contains nuclear export activity. Ace2p accumulates in the nucleus at the end of mitosis and activates early-G(1)-specific genes. We now provide evidence that Ace2p is nuclear not only in late M-early G(1) but also during other stages of the cell cycle. This feature of Ace2p localization explains its ability to activate genes such as CUP1, which are not expressed in a cell cycle-dependent manner.

From the analysis of protein complexes to proteome-wide linkage maps.

Recent advances in genomics have led to the accumulation of an unprecedented amount of data about genes. Proteins, not genes, however, sustain function. The traditional approach to protein function analysis has been the design of smart genetic assays and powerful purification protocols to address very specific questions concerning cellular mechanisms. Lately, a number of proteome-wide functional strategies have emerged, giving rise to a new field in biology, proteomics, that addresses the biology of a cell as a whole.

Genome-wide protein interaction maps using two-hybrid systems.

Automated sequence technology has rendered functional biology amenable to genomic scale analysis. Among genome-wide exploratory approaches, the two-hybrid system in yeast (Y2H) has outranked other techniques because it is the system of choice to detect protein-protein interactions. Deciphering the cascade of binding events in a whole cell helps define signal transduction and metabolic pathways or enzymatic complexes. The function of proteins is eventually attributed through whole cell protein interaction maps where totally unknown proteins are partnered with fully annotated proteins belonging to the same functional category. Since its first description in the late 1980's, several versions of the Y2H have been developed in order to overcome the major limitations of the system, namely false positives and false negatives. Optimized versions have been recently applied at multi-molecular and genomic scale. These genome-wide surveys can be methodologically divided into two types of approaches: one either tests combinations of predefined polypeptides (the so-called matrix approach) using various short-cuts to speed up the process, or one screens with a given polypeptide (bait) for potential partners (preys) present in complex libraries of genomic or complementary DNA (library screening). In the former strategy, one tests what one knows, for example pair-wise interactions between full-length open reading frames from recently sequenced and annotated genomes. Although based on a one-by-one scheme, this method is reported to be amenable to large-scale genomics thanks to multicloning strategies and to the use of small robotics workstations. In the latter, highly complex cDNA or genomic libraries of protein domains can be screened to saturation with high-throughput screening systems allowing the discovery of yet unidentified proteins. Both approaches have strengths and drawbacks that will be discussed here. None yields a full proteome-wide screening since certain proteins (e.g. some transcription factors) are not usable in Y2H. Novel two-hybrid assays have been recently described in bacteria. Applications of these time- and cost-effective assays to genomic screening will be discussed and compared to the Y2H technology.

Genome-wide protein interaction screens reveal functional networks involving Sm-like proteins.

A set of seven structurally related Sm proteins forms the core of the snRNP particles containing the spliceosomal U1, U2, U4 and U5 snRNAs. A search of the genomic sequence of Saccharomyces cerevisiae has identified a number of open reading frames that potentially encode structurally similar proteins termed Lsm (Like Sm) proteins. With the aim of analysing all possible interactions between the Lsm proteins and any protein encoded in the yeast genome, we performed exhaustive and iterative genomic two-hybrid screens, starting with the Lsm proteins as baits. Indeed, extensive interactions amongst eight Lsm proteins were found that suggest the existence of a Lsm complex or complexes. These Lsm interactions apparently involve the conserved Sm domain that also mediates interactions between the Sm proteins. The screens also reveal functionally significant interactions with splicing factors, in particular with Prp4 and Prp24, compatible with genetic studies and with the reported association of Lsm proteins with spliceosomal U6 and U4/U6 particles. In addition, interactions with proteins involved in mRNA turnover, such as Mrt1, Dcp1, Dcp2 and Xrn1, point to roles for Lsm complexes in distinct RNA metabolic processes, that are confirmed in independent functional studies. These results provide compelling evidence that two-hybrid screens yield functionally meaningful information about protein-protein interactions and can suggest functions for uncharacterized proteins, especially when they are performed on a genome-wide scale.

A genomic approach of the hepatitis C virus generates a protein interaction map.

The hepatitis C virus (HCV) causes severe liver disease, including liver cancer. A vaccine preventing HCV infection has not yet been developed, and, given the increasing number of infected people, this virus is now considered a major public-health problem. The HCV genome is a plus-stranded RNA that encodes a single polyprotein processed into at least 10 mature polypeptides. So far, only the interaction between the protease NS3 and its cofactor, NS4A, which is involved in the processing of the non-structural region, has been extensively studied. Our work was aimed at constructing a protein interaction map of HCV. A classical two-hybrid system failed to detect any interactions between mature HCV polypeptides, suggesting incorrect folding, expression or targetting of these proteins. We therefore developed a two-hybrid strategy, based on exhaustive screens of a random genomic HCV library. Using this method, we found known interactions, such as the capsid homodimer and the protease dimer, NS3-NS4A, as well as several novel interactions such as NS4A-NS2. Thus, our results are consistent with the idea that the use of a random genomic HCV library allows the selection of correctly folded viral protein fragments. Interacting domains of the viral polyprotein are identified, opening the possibility of developing specific anti-viral agents, based on their ability to modulate these interactions.

Characterization of Sm-like proteins in yeast and their association with U6 snRNA.

Seven Sm proteins associate with U1, U2, U4 and U5 spliceosomal snRNAs and influence snRNP biogenesis. Here we describe a novel set of Sm-like (Lsm) proteins in Saccharomyces cerevisiae that interact with each other and with U6 snRNA. Seven Lsm proteins co-immunoprecipitate with the previously characterized Lsm4p (Uss1p) and interact with each other in two-hybrid analyses. Free U6 and U4/U6 duplexed RNAs co-immunoprecipitate with seven of the Lsm proteins that are essential for the stable accumulation of U6 snRNA. Analyses of U4/U6 di-snRNPs and U4/U6.U5 tri-snRNPs in Lsm-depleted strains suggest that Lsm proteins may play a role in facilitating conformational rearrangements of the U6 snRNP in the association-dissociation cycle of spliceosome complexes. Thus, Lsm proteins form a complex that differs from the canonical Sm complex in its RNA association(s) and function. We discuss the possible existence and functions of alternative Lsm complexes, including the likelihood that they are involved in processes other than pre-mRNA splicing.

Yeast forward and reverse 'n'-hybrid systems.

Since its original description almost 10 years ago, the yeast two-hybrid system has been used extensively to identify protein-protein interactions from many different organisms. Simultaneously, a number of 'variations on a theme' based on the original concept have been described. In one set of variations, systems were developed to detect other macromolecular interactions: DNA-protein (one-hybrid), RNA-protein (RNA-based three-hybrid) and small molecule-protein interactions (ligand-based three-hybrid). These different versions are collectively referred to here as 'n-hybrid systems'. In another set of variations, the original configuration of the two-hybrid fusion proteins was modified to expand the range of possible protein-protein interactions that could be analyzed. For example, systems were developed to detect trimeric interactions, ligand-receptor interactions or interactions that require particular post-translational modifications. Finally, the original concept was turned upside down and 'reverse n-hybrid systems' were developed to identify mutations, peptides or small molecules that dissociate macromolecular interactions. These reagents can be used to validate, in the relevant biological systems, the potential interactions identified with the 'forward n-hybrid systems'. The powerful genetic selections of the forward and reverse n-hybrid systems are proving useful in proteomic projects aimed at generating macromolecular interaction maps.

Yeast RNase III as a key processing enzyme in small nucleolar RNAs metabolism.

The variety of biogenesis pathways for small nucleolar RNAs (snoRNAs) reflects the diversity of their genomic organization. We have searched for yeast snoRNAs which are affected by the depletion of the yeast ortholog of bacterial RNase III, Rnt1. In a yeast strain inactivated for RNT1, almost half of the snoRNAs tested are depleted with significant accumulation of monocistronic or polycistronic precursors. snoRNAs from both major families of snoRNAs (C/D and H/ACA) are affected by RNT1 disruption. In vitro, recombinant Rnt1 specifically cleaves pre-snoRNA precursors in the absence of other factors, generating intermediates which require the action of other enzymes for processing to the mature snoRNA. Most Rnt1 cleavage sites fall within potentially double-stranded regions closed by tetraloops with a novel consensus sequence AGNN. These results demonstrate that biogenesis of a large number of snoRNAs from the two major families of snoRNAs requires a common RNA endonuclease and a putative conserved structural motif.

Processing of a dicistronic small nucleolar RNA precursor by the RNA endonuclease Rnt1.

Small nucleolar RNAs (snoRNAs) are intron encoded or expressed from monocistronic independent transcription units, or, in the case of plants, from polycistronic clusters. We show that the snR190 and U14 snoRNAs from the yeast Saccharomyces cerevisiae are co-transcribed as a dicistronic precursor which is processed by the RNA endonuclease Rnt1, the yeast ortholog of bacterial RNase III. RNT1 disruption results in a dramatic decrease in the levels of mature U14 and snR190 and in accumulation of dicistronic snR190-U14 RNAs. Addition of recombinant Rnt1 to yeast extracts made from RNT1 disruptants induces the chase of dicistronic RNAs into mature snoRNAs, showing that dicistronic RNAs correspond to functional precursors stalled in the processing pathway. Rnt1 cleaves a dicistronic transcript in vitro in the absence of other factors, separating snR190 from U14. Thus, one of the functions of eukaryotic RNase III is, as for the bacterial enzyme, to liberate monocistronic RNAs from polycistronic transcripts.

Functional conservation of the transportin nuclear import pathway in divergent organisms.

Human transportin1 (hTRN1) is the nuclear import receptor for a group of pre-mRNA/mRNA-binding proteins (heterogeneous nuclear ribonucleoproteins [hnRNP]) represented by hnRNP A1, which shuttle continuously between the nucleus and the cytoplasm. hTRN1 interacts with the M9 region of hnRNP A1, a 38-amino-acid domain rich in Gly, Ser, and Asn, and mediates the nuclear import of M9-bearing proteins in vitro. Saccharomyces cerevisiae transportin (yTRN; also known as YBR017c or Kap104p) has been identified and cloned. To understanding the nuclear import mediated by yTRN, we searched with a yeast two-hybrid system for proteins that interact with it. In an exhaustive screen of the S. cerevisiae genome, the most frequently selected open reading frame was the nuclear mRNA-binding protein, Nab2p. We delineated a ca.-50-amino-acid region in Nab2p, termed NAB35, which specifically binds yTRN and is similar to the M9 motif. NAB35 also interacts with hTRN1 and functions as a nuclear localization signal in mammalian cells. Interestingly, yTRN can also mediate the import of NAB35-bearing proteins into mammalian nuclei in vitro. We also report on additional substrates for TRN as well as sequences of Drosophila melanogaster, Xenopus laevis, and Schizosaccharomyces pombe TRNs. Together, these findings demonstrate that both the M9 signal and the nuclear import machinery utilized by the transportin pathway are conserved in evolution.

Conservation of functional domains involved in RNA binding and protein-protein interactions in human and Saccharomyces cerevisiae pre-mRNA splicing factor SF1.

The modular structure of splicing factor SF1 is conserved from yeast to man and SF1 acts at early stages of spliceosome assembly in both organisms. The hnRNP K homology (KH) domain of human (h) SF1 is the major determinant for RNA binding and is essential for the activity of hSF1 in spliceosome assembly, supporting the view that binding of SF1 to RNA is essential for its function. Sequences N-terminal to the KH domain mediate the interaction between hSF1 and U2AF65, which binds to the polypyrimidine tract upstream of the 3' splice site. Moreover, yeast (y) SF1 interacts with Mud2p, the presumptive U2AF65 homologue in yeast, and the interaction domain is conserved in ySF1. The C-terminal degenerate RRMs in U2AF65 and Mud2p mediate the association with hSF1 and ySF1, respectively. Analysis of chimeric constructs of hSF1 and ySF indicates that the KH domain may serve a similar function in both systems, whereas sequences C-terminal to the KH domain are not exchangeable. Thus, these results argue for hSF1 and ySF1, as well as U2AF65 and Mud2p, being functional homologues.

Parallel analysis of genetic selections using whole genome oligonucleotide arrays.

Thousands of genes have recently been sequenced in organisms ranging from Escherichia coli to human. For the majority of these genes, however, available sequence does not define a biological role. Efficient functional characterization of these genes requires strategies for scaling genetic analyses to the whole genome level. Plasmid-based library selections are an established approach to the functional analysis of uncharacterized genes and can help elucidate biological function by identifying, for example, physical interactors for a gene and genetic enhancers and suppressors of mutant phenotypes. The application of these selections to every gene in a eukaryotic genome, however, is generally limited by the need to manipulate and sequence hundreds of DNA plasmids. We present an alternative approach in which identification of nucleic acids is accomplished by direct hybridization to high-density oligonucleotide arrays. Based on the complete sequence of Saccharomyces cerevisiae, high-density arrays containing oligonucleotides complementary to every gene in the yeast genome have been designed and synthesized. Two-hybrid protein-protein interaction screens were carried out for S. cerevisiae genes implicated in mRNA splicing and microtubule assembly. Hybridization of labeled DNA derived from positive clones is sufficient to characterize the results of a screen in a single experiment, allowing rapid determination of both established and previously unknown biological interactions. These results demonstrate the use of oligonucleotide arrays for the analysis of two-hybrid screens. This approach should be generally applicable to the analysis of a range of genetic selections.

Toward a functional analysis of the yeast genome through exhaustive two-hybrid screens.

The genome of the yeast Saccharomyces cerevisiae is now completely sequenced. Despite successful genetic work in recent years, 60% of yeast genes have no assigned function and half of those encode putative proteins without any homology with known proteins. Genetic analyses, such as suppressor or synthetic lethal screens, have suggested many functional links between gene products, some of which have been confirmed by biochemical means. Altogether, these approaches have led to a fairly extensive knowledge of defined biochemical pathways. However, the integration of these pathways against the background of complexity in a living cell remains to be accomplished. The two-hybrid method applied to the yeast genome might allow the characterization to the network of interactions between yeast proteins, leading to a better understanding of cellular functions. Such an analysis has been performed for the bacteriophage T7 genome that encodes 55 proteins and for Drosophila cell cycle regulators. However, the currently available two-hybrid methodology is not suitable for a large-scale project without specific methodological improvements In particular, the exhaustivity and selectivity of the screens must first be greatly improved. We constructed a new yeast genomic library and developed a highly selective two-hybrid procedure adapted for exhaustive screens of the yeast genome. For each bait we selected a limited set of interacting preys that we classified in categories of distinct heuristic values. Taking into account this classification, new baits were chosen among preys and, in turn, used for second-round screens. Repeating this procedure several times led to the characterization of the network of interactions. Using known pre-mRNA splicing factors as initial baits, we were able to characterize new interactions between known splicing factors, identify new yeast splicing factors, including homologues of human SF1 and SAP49, and reveal novel potential functional links between cellular pathways. Using different cellular pathways as anchor points, this novel strategy allows us to envision the building of an interaction map of the yeast proteome. In addition, this two-hybrid strategy could be applied to other genomes and might help to resolve the human protein linkage map.