Karyoplasmic interaction selection strategy: a general strategy to detect protein-protein interactions in mammalian cells

Genetic and biochemical probes for protein-protein interactions

The two-hybrid system is a powerful approach for examining protein-protein interactions. Recently, the utility of the system has been extended to include the genome-wide mapping of protein-protein interactions and the identification of peptide inhibitors of protein interactions. In addition, immunophilins and their chemical ligands are providing useful reagents for generating conditional protein-protein interactions in vivo to dissect intracellular signaling pathways.

Detection of protein-protein interactions in the nervous system using the two-hybrid system

Specific interactions between proteins regulate nearly all cellular processes. In the nervous system specialized processes such as neuronal proliferation, differentiation and targeting, synapse formation and neurotransmitter release are all tightly controlled by cascades of protein-protein interactions. The extent and nature of these interactions is therefore a question of fundamental importance. The two-hybrid system, which is beginning to be widely applied in many other areas of cell biology, offers a novel and sensitive technique for the identification and analysis of these protein-protein interactions.

Converting cancer genes into killer genes

Over the past decade, it has become clear that tumorigenesis is driven by alterations in genes that control cell growth or cell death. Theoretically, the proteins encoded by these genes provide excellent targets for new therapeutic agents. Here, we describe a gene therapy approach to specifically kill tumor cells expressing such oncoproteins. In outline, the target oncoprotein binds to exogenously introduced gene products, resulting in transcriptional activation of a toxic gene. As an example, we show that this approach can be used to specifically kill cells overexpressing a mutant p53 gene in cell culture. The strategy may be generally applicable to neoplastic diseases in which the underlying patterns of genetic alterations or abnormal gene expression are known.

Zinc fingers 1-7 of EVI1 fail to bind to the GATA motif by itself but require the core site GACAAGATA for binding

EVI1 is a zinc finger oncoprotein that binds via fingers 1-7 to the sequence GACAAGATAA. The target genes on which EVI1 acts are unknown. This binding motif overlaps with that for the GATA transcription factors, (T/A)GATA(A/G), and GATA-1 can bind to and activate transcription via a GACAAGATAA motif. The possibility has been raised that, when overexpressed in leukemogenesis, EVI1 may function by interfering with the differentiation-promoting action of GATA factors. To explore this, we have assessed the affinity of EVI1 for the GATA binding sites derived from erythroid-specific GATA-1 target genes, and found only low affinity interactions. We examined the contacts between EVI1 and DNA by methylation interference studies, which revealed extensive contacts between EVI1 and its binding site. The importance of the contacts for high affinity binding was shown by in vitro quantitative gel shift studies and in vivo cotransfection studies. To examine what types of sequences from mouse genomic DNA bind to EVI1, we isolated and sequenced five EVI1-binding fragments, and each showed the GACAAGATA site. The data presented contribute to our knowledge of the binding specificity of EVI1, and yield a clearer picture of what sequences can, and cannot, act as targets for EVI1 action.

Dioxin-induced CYP1A1 transcription in vivo: the aromatic hydrocarbon receptor mediates transactivation, enhancer-promoter communication, and changes in chromatin structure

We have analyzed the dioxin-inducible transcriptional control mechanism for the mouse CYP1A1 gene in its native chromosomal context. Our genetic and biochemical studies indicate that a C-terminal segment of the aromatic hydrocarbon receptor (AhR) contains latent transactivation capability and communicates the induction signal from enhancer to promoter. Thus, transactivation and enhancer-promoter communication may be congruent functions of AhR. Both functions require heterodimerization between AhR and the AhR nuclear translocator (Arnt). Our findings also indicate that heterodimerization activates AhR's latent transactivation function and silences that of Arnt. Furthermore, removal of Arnt's transactivation domain does not affect dioxin-induced CYP1A1 transcription in vivo. In addition, our studies demonstrate that dioxin-induced changes in chromatin structure occur by different mechanisms at the CYP1A1 enhancer and promoter and that events at an enhancer can be experimentally dissociated from events at the cognate promoter during mechanistic analyses of mammalian transcription in vivo.

A link between increased transforming activity of lymphoma-derived MYC mutant alleles, their defective regulation by p107, and altered phosphorylation of the c-Myc transactivation domain

The c-Myc protein is a transcription factor with an N-terminal transcriptional regulatory domain and C-terminal oligomerization and DNA-binding motifs. Previous studies have demonstrated that p107, a protein related to the retinoblastoma protein, binds to the c-Myc transcriptional activation domain and suppresses its activity. We sought to characterize the transforming activity and transcriptional properties of lymphoma-derived mutant MYC alleles. Alleles encoding c-Myc proteins with missense mutations in the transcriptional regulatory domain were more potent than wild-type c-Myc in transforming rodent fibroblasts. Although the mutant c-Myc proteins retained their binding to p107 in in vitro and in vivo assays, p107 failed to suppress their transcriptional activation activities. Many of the lymphoma-derived MYC alleles contain missense mutations that result in substitution for the threonine at codon 58 or affect sequences flanking this amino acid. We observed that in vivo phosphorylation of Thr-58 was absent in a lymphoma cell line with a mutant MYC allele containing a missense mutation flanking codon 58. Our in vitro studies suggest that phosphorylation of Thr-58 in wild-type c-Myc was dependent on cyclin A and required prior phosphorylation of Ser-62 by a p107-cyclin A-CDK complex. In contrast, Thr-58 remained unphosphorylated in two representative mutant c-Myc transactivation domains in vitro. Our studies suggest that missense mutations in MYC may be selected for during lymphomagenesis, because the mutant MYC proteins have altered functional interactions with p107 protein complexes and fail to be phosphorylated at Thr-58.

Internal ribosome entry sites and dicistronic RNAs in mammalian transgenesis

Modification of the genetic content of cultured cells or of whole animals is now a key strategy in both basic biological research and applied biotechnology. Yet obtaining the desired level and specificity of expression of an introduced gene remains highly problematic. One solution could be to couple expression of a transgene to that of an appropriate intact genomic locus. The identification and functional characterization of RNA sequences known as internal ribosome entry sites now offer the possibility of achieving precise control of transgene expression through the generation of dicistronic fusion mRNAs.

The yeast two-hybrid system for studying protein-protein interactions

A great number of investigators are currently employing the yeast two-hybrid system to study protein-protein interactions. Recent applications and newer configurations of the technique include yeast strains with improved selectivity and screens for false-positive clones.

The LIM/double zinc-finger motif functions as a protein dimerization domain

Protein-protein interactions resulting in dimerization and heterodimerization are of central importance in the control of gene expression and cell function. Proteins that share the 52-residue LIM/double zinc-finger domain are involved in a wide range of developmental and cellular controls. Some of these functions have been hypothesized to involve protein dimerization. In the present report we demonstrate, using both in vitro and cell-based studies, that a representative LIM protein, human cysteine-rich protein (hCRP), can efficiently homodimerize. The dimerization ability of hCRP is mapped to the LIM domains, can be transferred to an unrelated protein by fusion of a single minimal LIM/double zinc-finger segment, occurs in the absence as well as the presence of DNA, and appears to depend on coordination of two zinc atoms in the finger doublet. These observations support a specific role for protein dimerization in the function of proteins containing the LIM/double zinc-finger domain and expand the general spectrum of potential interactions mediated by zinc-finger motifs.

Phenotypic mixing between different hepadnavirus nucleocapsid proteins reveals C protein dimerization to be cis preferential

Hepadnaviruses encode a single core (C) protein which assembles into a nucleocapsid containing the polymerase (P) protein and pregenomic RNA during viral replication in hepatocytes. We examined the ability of heterologous hepadnavirus C proteins to cross-oligomerize. Using a two-hybrid assay in HepG2 cells, we observed cross-oligomerization among the core proteins from hepatitis B virus (HBV), woodchuck hepatitis virus, and ground squirrel hepatitis virus. When expressed in Xenopus oocytes, in which hepadnavirus C proteins form capsids, the C polypeptides from woodchuck hepatitis virus and ground squirrel hepatitis virus, but not duck hepatitis B virus, can efficiently coassemble with an epitope-tagged HBV core polypeptide to form mixed capsids. However, when two different core mRNAs are coexpressed in oocytes the core monomers show a strong preference for forming homodimers rather than heterodimers. This holds true even for coexpression of two HBV C proteins differing only by an epitope tag, suggesting that core monomers are not free to diffuse and associate with other monomers. Thus, mixed capsids result from aggregation of different species of homodimers.

Mapping of interacting domains between the nucleocapsid protein and the phosphoprotein of vesicular stomatitis virus by using a two-hybrid system

Specific interaction between the nucleocapsid protein (N) and the phosphoprotein (P) of vesicular stomatitis virus (VSV), an important step in the life-cycle of the virus, was studied by using a two-hybrid system. Plasmids encoding P fused with the yeast GAL4 DNA-binding domain (pGALP) and N fused with the herpes simplex virus VP16 transactivating region (pVPN) were transfected into CHO cells along with a reporter plasmid encoding chloramphenicol acetyltransferase (CAT). The ability of N and P to associate in vivo was measured by activation of the CAT gene by the VP16 transactivating region. Transfection of plasmids pGALP and pVPN resulted in a high level of CAT activity, indicating that the N and P portions of the fusion proteins associated very strongly with each other. Progressive C-terminal deletions of the P protein revealed two regions that are important for association with the N protein: the N-terminal acidic domain and the C-terminal basic domain. Phosphorylation of P protein was not required for N-P association. Various deletions and mutations of the N protein revealed the C-terminal 5 amino acids (Val-Glu-Phe-Asp-Lys), in particular the amino acids Val-Glu-Phe, to be critical for N association with P. This two-hybrid system can be used in other viral systems to study the interaction between proteins involved in transcription and replication.

Evaluation of the contingent replication assay (CRA) and its application to the study of the general transcription initiation factor, TFIIF

The Contingent replication assay (CRA) is a rapid assay for the screening and isolation of cDNAs by protein-protein or protein-DNA interactions in mammalian cells. The method has been shown to enrich a plasmid containing a cDNA encoding the bacterial replication-related protein, R6K, from a mixture of two plasmids. In this report we present data illustrating the sensitivity and selectivity of the method. Using the small subunit of TFIIF (Rap30) as a target, we demonstrate the enrichment of a clone encoding the large subunit, Rap74, from a cDNA library. Additional cDNA clones including human Rap30 and an anonymous cDNA clone homologous to members of the human cdc2 kinase family were enriched and isolated by a modified screening approach. The structure of these additional clones suggest that the CRA enriches for products that interact not only directly with the target protein but also through bridging by endogenous proteins.

Strategies for the identification of interacting proteins

Many problems in modern biology involve complex arrays of interacting protein and, in some cases, RNA molecules. The initial challenge facing investigators is to identify the important players that drive the process under study. This difficult task is ameliorated somewhat by the development of methods designed to keep pace with the magnitude of this challenge. I have outlined a few of these approaches at the cutting edge of cloning interacting proteins. A perhaps more daunting prospect is to dissect the important molecules once they are in hand, to identify key interactions, and, ultimately, to move to an understanding of function in cells. For this, of course, all of the tools of genetics, biochemistry, and molecular biology, extant and yet to be developed, will have to be tapped.

Screening for in vivo protein-protein interactions

We describe an in vivo approach for the isolation of proteins interacting with a protein of interest. The protein of interest is "tagged" with a portion of the biotin carboxylase carrier protein (BCCP), encoded on a specially constructed plasmid, so that it becomes biotinylated in vivo. The "query" proteins (e.g., those in a cDNA library) are tagged by fusing them to the 3' end of the lacZ gene on a lambda vector in such a way that the beta-galactosidase activity is not disrupted. These phage are transfected into cells containing the plasmid encoding the BCCP-tagged protein. The infection lyses the cells and exposes the protein complexes. The BCCP-tagged protein and any associated protein(s) are "captured" by using avidin, streptavidin, or anti-biotin antibody-coated filters. The detection of bound protein is accomplished by directly assaying for beta-galactosidase activity on the filters. Positive plaques can be plaque-purified for DNA sequencing. We have tested this approach by using c-Fos and c-Jun as our model system. We show that avidin, streptavidin, or polyclonal anti-biotin (but not a monoclonal anti-biotin) antibody is capable of specifically capturing in vivo biotinylated beta-galactosidase and c-Jun and that this capture is dependent upon the presence of both avidin and the BCCP moiety. Further, complexes containing c-Jun and c-Fos can also be isolated in this manner, and the isolation of this complex is dependent on the presence of c-Fos, c-Jun, avidin, and the BCCP moiety. We discuss the possible uses and limitations of this technique for isolating proteins that interact with a known protein.