Protein Domain Fusion

Ivan Sadowski, University of British Columbia, Vancouver, Canada

Proteins are composed of multiple functional and structural domains. Using recombinant DNA techniques, it is possible to create chimaeric proteins with novel functions for research or biotechnology purposes, and for the development of novel therapies.

Introductory article
Article Contents
. Introduction . Domain Organization of Proteins . Mapping Functional Properties of Proteins by Domain Swaps . Identifying Functional Domains Using Chimaeric Proteins

Introduction
Advances in recombinant DNA techniques over the past two decades have made it possible for researchers to produce designer proteins that have properties tailored for specic purposes. Strategies for producing and expressing chimaeric protein molecules are greatly simplied by powerful techniques for manipulation of DNA in vitro, such as the polymerase chain reaction (PCR) and oligonucleotide-directed mutagenesis, as well as automated DNA sequencing techniques. Combined with the understanding that many proteins are comprised of multiple independently folded functional domains, it is relatively simple to use in vitro DNA manipulation techniques to construct genes encoding proteins with completely novel desired functions using combinations of domains from dierent proteins. Most protein domain fusions that have been constructed to date have been exclusively used for research purposes, but the enormous potential of these technologies will undoubtedly drive development of the next generation of therapeutic agents. Indeed, a number of very promising strategies involving recombinant antibody fusions that target specic cancers are already in clinical trials. In this article, some of the experiments and observations that were signicant in establishing protein domain fusions as a useful and accepted means of manipulating protein function and cell regulatory events are reviewed. Some of the protein domains and functional elements that are commonly used for various domain fusion strategies are also summarized. It should be recognized that this summary is far from complete because of the vast number of dierent functional and structural domains that have been characterized already, and the rapid pace at which new domains are discovered. The usefulness of this vast array of building blocks for producing designer proteins appears to be limited only by the researchers imaginations.

. Signal Transduction Studies with Fusion Proteins . Epitope Tagging . Protein Purification Tag Fusions . Protein Cleavage Sites . Fusions for Monitoring Intracellular Protein Trafficking . Antibody and Bacterial Toxin Fusions . Antiviral Fusion Proteins . Manipulating Cell Regulation using Protein Fusions

Domain Organization of Proteins

Many proteins are comprised of two or more separately folded structures joined together by linker sequences. These protein substructures, called modules or domains, generally confer specic functional or structural properties

to the protein. Protein domains may function by directing specic interaction with another macromolecule or ligand, or may confer a specic catalytic function. From the accumulated DNA sequences that are available, it has become obvious that the repertoire of proteins produced by dierent organisms is built from a nite set of protein domains organized in dierent combinations. This process is thought to have been accomplished throughout evolution by shuing of protein-encoding DNA sequences, or exons, through recombination of the intervening intron DNAs. One theoretical study suggests that the genomes of complex eukaryotes are composed of less than 7000 exons. Consistently, comparison of all available protein sequences identied approximately 6000 domains that occur in two or more proteins. These gures may increase substantially as more DNA sequences are determined for more diverse species. One well-characterized example of a functional domain that occurs in a variety of proteins is the Src-homology 2 (SH2) domain. SH2 was originally identied as a region of sequence identity that was shared amongst nonreceptor protein tyrosine kinases. Many dierent classes of proteins involved in eukaryotic signal transduction were subsequently found to possess SH2 domains, whereas the protein tyrosine kinase domain (SH1) is only found within proteins possessing tyrosine kinase catalytic function. Structural analysis revealed that the SH2 domain represents an independently folding protein subunit that mediates proteinprotein interactions by binding specic phosphotyrosine-containing protein sequences.

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

Mapping Functional Properties of Proteins by Domain Swaps

Families of related proteins can be identied by sequence identity of their structural domains. Individual members of protein families often have related domains which have evolved amino acid sequence dierences that confer separate distinct specicities. One approach for identifying regions of proteins that confer specicity involves swapping domains from related proteins with dierent specicity. Such an experiment was rst used to demonstrate that the DNA-binding specicity of lamboid phage repressor proteins was conferred by an a-helical recognition sequence within the N-terminal DNA-binding domain. In this experiment, an Escherichia coli phage 434 repressor a helix was substituted with a corresponding a helix from Salmonella phage P22. The resulting chimaeric 434P22 repressor protein was found to bind to DNA with the specicity of the P22 repressor, demonstrating that the substituted a helix represented the DNA-recognition motif. This type of strategy has been emulated many times since to identify regions of proteins that confer specic interactions.

Identifying Functional Domains Using Chimaeric Proteins

Most functional domains of proteins can work even when fused to completely dierent proteins. This is not surprising, considering that this is how evolution is thought to have produced our existing proteins. Construction of chimaeric protein molecules using parts of dierent proteins is commonly used to characterize and identify the function of protein domains. Proteins that regulate transcription in eukaryotes have been extensively studied using this approach. A key experiment which demonstrated that positively acting gene regulatory proteins possessed transcription activation functions, separate from their ability to bind DNA, was performed by creating a fusion protein between LexA, a repressor protein from E. coli, and GAL4, a yeast transcriptional activator. GAL4 activates transcription when bound to a specic DNA sequence upstream of the genes it regulates. The portion of GAL4 that mediates specic binding to DNA, or the DNA-binding domain, is contained within the N-terminus. This portion of GAL4 was replaced with the LexA protein by an in-frame fusion. The resulting LexAGAL4 chimaeric protein, when expressed in yeast, was found to activate transcription of articial genes which had upstream binding sites for LexA protein. This experiment demonstrated that the ability of GAL4 to activate transcription must be mediated by a region within its C2

terminus that can be dissociated from the DNA-binding domain. A similar chimaeric protein was constructed to demonstrate that the C-terminal region of the herpesvirus structural protein VP16 possessed a transcriptional activation function. Upon infection, VP16 is responsible for activating transcription of the herpes immediate early genes. However, unlike many transcriptional activator proteins, VP16 does not bind directly to DNA, but rather interacts with viral promoters by binding to cellular proteins that specically bind sequences upstream of the immediate early genes. Interaction of VP16 with the cellular proteins causes strong immediate early gene transcription. To demonstrate that a C-terminal region of VP16 was causing activation of transcription, this protein segment was fused onto the DNA-binding domain of GAL4 to create a GAL4VP16 chimaera. The GAL4 DNA-binding domain can eciently bind its site on DNA, but is unable to activate transcription because it lacks its Cterminal-activating regions (see above). However, the GAL4VP16 chimaeric protein was found to activate transcription very strongly in both yeast and mammalian cells, which demonstrates that the C-terminal region of VP16 must possess a strong transcriptional activation function. Construction of chimaeras with the GAL4 DNAbinding domain and LexA protein are now commonly used to identify activation domains from other eukaryotic transcriptional activator proteins. This strategy is also used to identify regulatory elements of transcription factors. For example fusion of the C-terminal ligandbinding domains of steroid hormone receptors onto the GAL4 or LexA DNA-binding domains invariably creates chimaeras that are capable of activating transcription of genes that have binding sites for GAL4 or LexA in response to the steroid ligand. GAL4 DNA-binding domain fusions with various signal transduction-responsive transcription factors are also commonly used to monitor activation of specic protein kinases in vivo (see below).

Signal Transduction Studies with Fusion Proteins

Higher eukaryotic cells have numerous signal transduction mechanisms involving networks of proteinprotein interactions and sequential protein kinase activation cascades. Mapping the eect of agonists on specic signal transduction events is often complicated by the fact that it is dicult to assay the activity of individual protein kinases within the whole-cell environment. To address this problem, a strategy has been devised involving chimaeric transcription factors whose activities are responsive to specic signal-activated protein kinases. The transcriptional acti-

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

vation functions of the Jun, Elk and CREB (cAMP response element-binding protein) transcription factors are dependent upon specic phosphorylations mediated by the JNK (Jun N-terminal kinase), MAPK (mitogenactivated protein kinase), and PKA (protein kinase A) kinases, respectively. When expressed as fusions with the DNA-binding domain of GAL4, the activation domains from these transcription factors cause induction of GAL4dependent reporter genes in response to signals that stimulate JNK, MAPK and PKA activity. Therefore, using this system, the activity of these three protein kinases can be measured in vivo by simply assaying expression of a GAL4-dependent reporter gene.

Epitope Tagging
The characterization of proteins in vivo requires the use of specic antibodies which can be used for immunoprecipitation, Western blotting or immunouorescence. Monoclonal or polyclonal antibodies have traditionally been raised against individual proteins of interest by immunization of animals with puried protein, which is often produced by expression in E. coli (see below). While it is relatively simple to produce polyclonal antibodies against specic proteins in this way, this approach is timeconsuming, normally requiring three to four months, and there is no guarantee that the resulting antibody preparations will be highly specic. Production of monoclonal antibodies usually guarantees highly specic antibody preparations, but this technique is also much more technically challenging and labour intensive. A strategy for immunological detection of proteins in vivo that circumvents problems associated with producing individual specic antibodies involves creating a fusion of the protein of interest with a short peptide epitope for which a highly specic monoclonal antibody is available. Epitope-tagged proteins are normally produced by fusing the coding sequence of a protein of interest in-frame with the coding sequence for one or more copies of the peptide epitope in a plasmid vector that can be used to express the fusion gene. When expressed in vivo, the epitope-tagged fusion protein can be immunologically detected with a

monoclonal antibody that recognizes the fused epitope. Epitope tags can be fused to the N- or C-terminus of proteins, but it is generally considered that the epitope is more likely to be exposed when fused at the C-terminus. The obvious advantage of this approach is that proteins can be detected in vivo using highly specic monoclonal antibodies without expending the time and labour to produce an individual specic antibody preparation. A number of dierent monoclonal antibodies have been produced that recognize short, well-dened peptide epitopes. Table 1 summarizes some of the commonly used epitope tags for which monoclonal antibodies are available commercially; most of these epitopes are derived from viral gene products, and therefore background detection of proteins other than the tagged fusion is generally not a signicant problem. The sensitivity at which epitopetagged proteins can be detected can be increased by multimerizing the tag. For example, when tagging a protein with the HA epitope, it is common to fuse the protein-coding sequence to DNA that encodes three epitope sequences joined end to end. Epitope tagging is possible because many proteins can tolerate fusion of even multimerized epitope tags without a signicant eect on the protein function. However, because this is not true for all proteins, it is important to rigorously verify that the tag does not aect protein function when using this technique. Another drawback of using epitope tags is that the tagged protein has to be expressed from a vector in whichever cell type the researcher is interested. Therefore it is also important to determine that over- or underexpression does not aberrantly aect any relevant features of the proteins function. In addition to their use in detecting proteins in vivo using standard immunological techniques, epitope tags are sometimes used as anity purication tags (also see below). Proteins expressed as fusions with the FLAG epitope (see Table 1) can be puried using an anity resin produced by coupling anti-FLAG monoclonal antibody to Sepharose beads. FLAG-tagged fusion proteins bound to the anity column can be stringently washed to eliminate unbound proteins and then eluted by lowering the pH or adding competitor FLAG peptide. Other epitope tags can be used for anity purication as well, using the

Table 1 Commonly used epitope tags Tag Myc V5 HSV FLAG T7 HA Sequence (amino acid) EQKLISEEDL GKPIPNPLLGLDST QPELAPEDPED DYKDDDDK MASMTGGQQMG YPYDVPDYA Protein of origin c-myc Simian virus 5 RNA polymerase Herpes simplex virus glycoprotein D Enterokinase cleavage site Phage T7 capsid protein Inuenza virus haemagglutinin

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

corresponding monoclonal antibodies for preparation of immunoanity columns and free epitope peptide for elution. An advantage of using purication tags such as the FLAG epitope is that it is only eight amino acids long, and therefore the tag is less likely to interfere with subsequent biochemical analysis than larger purication tags.

Protein Purification Tag Fusions

Even when expressed at high levels in E. coli, purication of recombinant proteins to near homogeneity may require at least three or four purication steps. A number of protein domains have been identied that specically bind small ligand molecules with high anity (Table 2). These domains have become useful for simplifying purication of recombinant fusion proteins. The rationale is similar to that described above using the FLAG monoclonal antibody epitopes as a purication tag. Proteins of interest are expressed as in-frame fusions with the purication tag domain. The fusion protein can be puried by specic
Table 2 Commonly used anity purication tags Tag 6-His Size (amino acids) 6 Purication

binding to an anity resin composed of the ligand for the purication tag coupled to Sepharose beads. The fusion protein is puried by binding to the anity resin; unbound proteins can be washed away, and the fusion protein is eluted by addition of buer containing free ligand. Using anity purication tags, fusion proteins can usually be puried to greater than 80% purity from crude extracts in a single step. Even when expressed in E. coli at high levels, none of the purication tags are guaranteed to give homogeneous protein preparations from a single anity purication step, primarily because there will always be background proteins that bind to the anity resins. Therefore, if a completely homogeneous preparation is required, it is useful to combine the anity step with at least one other purication step. Because binding of the purication tag to its ligand requires proper folding of the binding domain, most of the fusion tags listed in Table 2 require purication in nondenaturing conditions. This usually means that the protein must be puried from the soluble fraction of cell lysates. A problem commonly encountered when expressing proteins in recombinant form is that much of the protein can form insoluble aggregates, termed inclusion

Notes
21

ions; elute with Binds to chelated Ni imidazole, or lowered pH Binds to polyarsine oxide; resistant to high-temperature incubations; released from E. coli by osmotic shock Binds to reduced glutathione Binds cellulose; elute in low salt

Thioredoxin

109

Six consecutive histidine residues; native or denaturing conditions Produced in soluble form; highly thermostable; accumulates in adhesion zones

Glutathione Stransferase (GST) Cellulose-binding domain (CBP) Maltose-binding protein (MBP) Chitin-binding protein (CBD) Biotinylated peptide S-tag

220 185

390 50 75 15

Binds amylose resin; elute with maltose Binds chitin Binds avidin; elute with biotin Binds S-protein tag (KETAAAKFERQHIDS) binds to S-protein to form ribonuclease S Fusion proteins accumulate in E. coli inclusion bodies. A monoclonal is available for anity purication Binds calmodulin (calcium dependent); elute with a calcium chelator Binds immunoglobulin G Sepharose Binds streptavidin

Binding domain from prokaryotic cellulase enzymes From E. coli MalE protein

Peptide

TrpE

320

E. coli anthrinylate synthetase From myosin light-chain kinase

Calmodulin-binding peptide Protein A Strep-tag

26 508 9

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

bodies. Since inclusion bodies are formed almost exclusively from the expressed recombinant protein, isolation of this insoluble material generally yields nearly pure preparations. These insoluble protein preparations are ideal for immunization of animals for production of antibodies. The TrpE fusion expression system was designed with this feature in mind (Table 2); TrpE fusions are usually expressed at high levels as insoluble inclusion bodies in E. coli and generally make an excellent source of antigen for production of polyclonal antibodies. Proteins expressed as 6-His tagged fusions (Table 2) can be puried from either the soluble or insoluble fraction, because this tag can bind to chelated nickel ions even in the presence of denaturing agents such as guanidinium or urea. Therefore, insoluble recombinant material can be solubilized in urea or guanidinium, and puried by anity on chelated nickel resin. An advantage of this approach is that many proteins can be refolded while bound to the nickel resin by gradually reducing the concentration of denaturant, and then eluted from the column in nondenaturing buer with imidazole.

Table 3 Commonly used proteinase cleavage sites Endoproteinase Enterokinase Factor Xa Thrombin Intein Recognition sequence DDDDK/ IEGR/ LVPR /S Self-cleaving fragment of yeast VMA1 protein.

at the N-terminus of the intein by incubation at low temperature in the presence of thiols.

Fusions for Monitoring Intracellular Protein Trafficking

Analysis of intracellular protein localization has traditionally been examined using immunouorescence of permeabilized cell preparations xed to glass slides. This approach requires a specic and sensitive antibody that recognizes the protein of interest. As indicated above, immunouorescence for analysis of intracellular protein localization can also be performed with epitope-tagged proteins, using monoclonal antibodies that recognize the tag. However, this requires construction and expression of a tagged version of the protein of interest, and is subject to all of the caveats of using epitope tags as detailed above. Detection of protein fusions in vivo is greatly simplied by using the jellysh Aequoria victoria green uorescent protein (GFP). GFP is a 27-kDa monomeric protein that naturally emits green light when exposed to long-wave ultraviolet light. Importantly, GFP can tolerate fusion to either the C- or N-termini of proteins; this feature makes it very useful as a reporter for gene expression, and for examining the intracellular organization of fusion proteins. The intracellular localization of GFP fusions can be examined in live cells by uorescence microscopy. Variants of GFP have also been developed that uoresce more intensely, or emit blue or yellow light instead of green. These variants allow comparison of protein localization using double labelling experiments. As with all fusion proteins, it is important to verify that the GFP-tagged protein of interest reiterates the known functions of the corresponding unfused protein.

Protein Cleavage Sites

Purication of recombinant proteins is greatly simplied by the use of anity purication tags. However, one problem with this approach is that the N-terminal tag may interfere with the function of the recombinant protein. In cases where it is necessary to study a recombinant protein in its native form, it is possible to produce a fusion with a specic endoproteinase cleavage site inserted between the anity tag domain and the sequence of the protein of interest. Following purication of the fusion, the recombinant protein can be cleaved from the anity tag by digestion with the specic endoproteinase (see Table 3). Since the endoproteinase cleaves the peptide sequence at a precise location relative to the recognition sequence, it is possible to design the fusion construct so that cleavage produces an N-terminus closely resembling the native protein. An important consideration when using protease cleavage to release recombinant proteins from their tags, is that the protein of interest must not itself have a recognition sequence for the endoproteinase. Once cleaved from the recombinant protein, the anity tag can easily be removed from the preparation by passing over the anity resin a second time. An interesting variation on this strategy involves the use of a modied form of a self-splicing protein element (intein) from the yeast VMA1 gene (Table 3). The protein of interest is expressed as a tripartite fusion in E. coli with the intein and an anity purication tag at the C-terminus. The fusion protein is puried from extracts using the appropriate anity resin. The recombinant protein can be released from the anity column by inducing self-cleavage

Antibody and Bacterial Toxin Fusions

Most of the eorts to use fusion proteins as therapeutic agents involve production of recombinant antibody molecules. Antibody production technologies have progressed very rapidly over the past decade, and using
5

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

domain swapping it is possible to produce humanized antibodies using variable (Fv) regions from mouse monoclonal antibodies. Production of recombinant antibodies in E. coli by expression of cDNAs is also simplied by the development of single-chain Fv (sFv) proteins, in which the Fvs from the heavy and light chains are joined on the same polypeptide by a short linker. Continuing in this vein, recombinant bispecic antibodies have also been produced by fusing Fvs from two separate cloned Fvs. For therapeutic use, recombinant fusion antibodies are being used to produce immunocytokines, immunotoxins and for antibody-directed enzyme prodrug therapy (ADEPT). Immunocytokines are fusions between a humanized sFv that recognizes a tumour-specic antigen, and a cytokine, such as interleukin 2 (IL-2). The rationale for this strategy is that direction of IL-2 to the surface of a tumour cell elicits a directed cytotoxic T-cell response. This strategy has been successful in the treatment of metastatic melanoma in animal model systems. Immunotoxins are similar, except that the tumour-specic sFv is fused to a protein that is itself toxic to the tumour cell, such as Pseudomonas exotoxin A, ricin A chain, diphtheria toxin, or tumour necrosis factor (TNF). ADEPT is a related strategy that involves tumour-specic sFv proteins fused to enzymes that convert a nontoxic prodrug into its highly toxic form. The rationale for this strategy is that since prodrugs would be activated only in the vicinity of tumour cells, compounds could be used for treatment that would otherwise be too toxic. Diphtheria toxin and Pseudomonas exotoxin have also been targeted to specic cells by fusion to cytokines such as epidermal growth factor (EGF) and IL-2, for treatment of malignancies that overexpress their receptors. Targeting of toxins to human immunodeciency virus type 1 (HIV-1)infected cells has also been attempted using fusions with CD4, which binds the viral gp120.

repeat (LTR), in a TAR-dependent manner. Similar inhibitor fusion proteins that target functions of other viruses are also likely to be developed. A signicant problem in the use of antiviral fusion proteins is that a mechanism needs to be developed for delivery of the fusions to infected cells.

Manipulating Cell Regulation using Protein Fusions

In this nal section several techniques involving protein fusions that can be used to manipulate critical cell functions are reviewed. A problem when working with mammalian cells is that there are very few natural functions that can easily be manipulated experimentally without globally aecting cell physiology. For example, the only naturally regulatable promoters in mammalian cells are those controlled by steroid hormone receptors. Therefore, for characterizing the eects of specic proteins on cell regulation, there is a need for ways to regulate gene expression and manipulate protein function using treatments that are otherwise as innocuous to the cell as possible. One example is the TET-OFF and TET-ON systems, which enable regulation of specic transgenes in cell cultures or even in whole organisms by tetracyclines. The system employs a fusion between the E. coli Tn10 Tet repressor (TetR) and the HSV VP16 transcriptional activation domain (see above). The wild-type Tet repressor protein is normally prevented from binding to DNA in the presence of tetracycline. Therefore in the TET-OFF system, addition of tetracyclines causes dissociation of the TetRVP16 fusion from its DNA operator sequence, and therefore recombinant genes that are activated by TetRVP16 are switched o in the presence of tetracycline. A variant Tet repressor (rTet) was identied that binds DNA in the presence of tetracyclines. This variant forms the basis of the TET-ON system; rTetVP16 fusion proteins cause induction of genes bearing upstream Tet operators in response to tetracyclines. Other fusions have been developed to manipulate the activity of specic proteins. Fusion of the hormonebinding domain (HBD) of steroid hormone receptors has been used in many instances to confer regulation by steroid hormone on proteins that are normally unregulated. In the absence of bound ligand, hormone receptors are held in an inactive form in a complex that includes HSP90. Addition of the cognate steroid ligand causes dissociation of the HBD from the complex, and causes activation of the receptor. Fusion of the HBD from various steroid hormone receptors, including the glucocorticoid, oestrogen and mineralocorticoid receptors to heterologous proteins in many cases results in hormone-regulated protein functions. A number of mammalian transcription factors, including fos, myc and myb have been converted to

Antiviral Fusion Proteins

There are currently very few therapies that have been developed which specically inhibit viral proteins and yet do not also aect host cell functions. It is likely that protein domain fusions will comprise some of the next generation of antiviral therapies because fusions can be designed that specically interact with viral proteins or inhibit viral gene expression. One example of such an approach involves a fusion between the HIV-1 TAT protein and the KRAB (Kruppel-associated box) domain. TAT is a virally encoded protein that activates transcription from the nascent HIV-1 TAR stemloop RNA. KRAB is a transcriptional repression domain found in a large family of metazoan zinc-nger proteins. Expression of the TAT KRAB fusion has been found to inhibit both basal and activated transcription from the HIV-1 long terminal
6

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Protein Domain Fusion

hormone-responsive factors in this way, as have the protein kinases Abl, Raf-1 and yeast STE11. This technique will undoubtedly be used in the future to regulate the functions of other classes of proteins. Some proteins are regulated by controlling their multimerization. For example, activation of at least some protein kinases involves the formation of dimers. To manipulate protein function experimentally, an ingenious strategy was developed that allows stimulation of dimer formation by addition of the drug coumermycin. A number of antibiotics, such as novobiocin and coumermycin, inhibit E. coli replication by tightly binding the type II topoisomerase gyrase B (GyrB). However, unlike novobiocin, coumermycin can simultaneously bind two GyrB proteins, thereby causing their dimerization. To demonstrate that the Raf-1 protein kinase could be activated by dimer formation, a Raf-1GyrB fusion protein was expressed in cells and was shown to be activated by addition of coumermycin, but not novobiocin. In view of these observations, GyrB fusions will probably be an important tool for manipulating regulation of proteins that are controlled by dimer formation. An additional technique for manipulating protein function in vivo involves controlling protein stability. This can be accomplished in several ways. Proteins that are very stable and long-lived in vivo can be made unstable simply by fusion of a destruction or PEST sequence which targets proteins for degradation by the ubiquitin-proteosome pathway. Similarly, specic proteins have been made unstable by direct fusion to a single ubiquitin peptide chain. A method for regulating degradation of specic fusion proteins has also been developed. A thermo-labile dihydrofolate reductase (DHFR) protein was developed that becomes rapidly degraded only at temperatures high enough to cause its partial unfolding. Fusion of this thermolabile DHFR protein to heterologous proteins causes their degradation at elevated temperatures. This technique, known as td (temperature-inducible degron) can be used to selectively eliminate fusion proteins from a cell by their degradation.

Further Reading
Brizzard BL, Chubet RG and Vizard DL (1994) Immunoanity purication of FLAG epitope-tagged bacterial alkaline phosphatase

using a novel monoclonal antibody and peptide elution. Biotechniques 16: 730735. Chong S, Mersha FB, Comb DG et al. (1997) Single-column purication of free recombinant proteins using a self-cleavable anity tag derived from a protein splicing element. Gene 192: 271281. Dohmen RJ, Wu P and Varshavsky A (1994) Heat-inducible degron: a method for constructing temperature-sensitive mutants. Science 263: 12731276. Farrar MA, Alberol I and Perlmutter RM (1996) Activation of the Raf-1 kinase cascade by coumermycin-induced dimerization. Nature 383: 178181. Gossen M, Freundlieb S, Bender G, Muller G, Hillen W and Bujard H (1995) Transcriptional activation by tetracyclines in mammalian cells. Science 268: 17661769. Hayden MS, Gilliland LK and Ledbetter JA (1997) Antibody engineering. Current Opinion in Immunology 9: 201212. Kim JS and Raines RT (1993) Ribonuclease S-peptide as a carrier in fusion proteins. Protein Science 2: 348356. LaVallie ER, DiBlasio EA, Kovacic S, Grant KL, Schendel PF and McCoy JM (1993) A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm. Biotechnology 11: 187193. Lin A, Minden A, Martinetto H et al. (1995) Identication of a dual specicity kinase that activates the Jun kinases and p38-Mpk2. Science 268: 286290. Maina CV, Riggs PD, Grandea AGd et al. (1988) An Escherichia coli vector to express and purify foreign proteins by fusion to and separation from maltose-binding protein. Gene 74: 365373. Nilsson J, Larsson M, Stahl S, Nygren PA and Uhlen M (1996) Multiple anity domains for the detection, purication and immobilization of recombinant proteins. Journal of Molecular Recognition 9: 585594. Pawson T (1995) Protein modules and signalling networks. Nature 373: 573580. Pengue G, Caputo A, Rossi C, Barbanti-Brodano G and Lania L (1995) Transcriptional silencing of human immunodeciency virus type 1 long terminal repeat-driven gene expression by the Kruppel-associated box repressor domain targeted to the transactivating response element. Journal of Virology 69: 65776580. Picard D (1993) Steroid-binding domains for regulating the functions of heterologous proteins in cis. Trends in Cell Biology 3: 278280. Sadowski I (1995) Uses for GAL4 expression in mammalian cells. Genetic Engineering 17: 119148. Sonnhamer ELL and Kahn D (1994) Modular arrangement of proteins as inferred from analysis of homology. Protein Science 3: 482492. Woodworth TG and Nichols JC (1993) Recombinant fusion toxins a new class of targeted biologic therapeutics. Cancer Treatment and Research 68: 145160. Zheng CF, Simcox T, Xu L and Vaillancourt P (1997) A new expression vector for high level protein production, one step purication and direct isotopic labeling of calmodulin-binding peptide fusion proteins. Gene 186: 5560.

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net