Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems

A 6xHis tag (HHHHHH, 0.84 kDa, usually six histidines but anywhere from 2 to 10) binds immobilized transition metal ions (Co2+, Ni2+, Cu2+, Zn2+) in immobilized metal-affinity chromatography because the imidazole rings of histidine form coordination bonds with the metal. A bound His6-tagged protein is eluted from Ni2+-NTA with 20-250 mM imidazole or by lowering the pH, and washing with a low imidazole concentration such as 0.8 mM reduces nonspecific binding of host proteins that contain histidines.

Retention of poly-histidine-tagged dihydrofolate reductase (DHFR) on Ni2+-NTA depends on the number of histidines and on the buffer. In 0.05 M phosphate buffer, (His)3-DHFR is retained at about 90% and (His)2-DHFR at about 30%. In 6 M guanidine hydrochloride, (His)6-DHFR is retained at more than 90% while (His)3-DHFR is retained at less than 10% and (His)2-DHFR at less than 5%, so His6 tags suit denaturing purification and His3 tags suit physiological conditions.

Co2+-carboxymethylaspartate (Co2+-CMA, TALON) resin releases His-tagged proteins under milder conditions than Ni2+-NTA and shows less nonspecific protein binding, giving enzyme preparations of more than 95% purity by SDS-PAGE. Imidazole used for elution from IMAC resins can interfere with NMR experiments, competition studies and crystallographic trials, and its presence often leads to protein aggregates.

A His tag rarely affects protein activity because of its small size and charge, but when it does, moving it to the opposite terminus or purifying under denaturing conditions often solves the problem. Purification of metal-containing proteins on Ni2+-NTA is not recommended because the NTA can absorb the metal, and purification under anaerobic conditions is not recommended because Ni2+-NTA is reduced.

The 19-residue HAT tag (KDHLIHNVHKEFHAHAHNK, a natural poly-histidine affinity tag) allows one-step purification on Co2+-CMA resin under mild conditions to more than 95% purity. N-terminally HAT-tagged chloramphenicol acetyltransferase, dihydrofolate reductase and green fluorescent protein were purified with 5 mM imidazole in the equilibration and loading buffer and 150 mM imidazole for elution, and elution also works at pH 5.0. Urea suppresses HAT-tag binding much more than guanidinium HCl does, and expression with the HAT tag has been tested only in bacteria.

The FLAG tag (DYKDDDDK, 8 residues, 1.01 kDa) is a hydrophilic peptide bound by the monoclonal antibody M1, and the monoclonal antibodies M2 and M5 recognize it with different binding characteristics. FLAG-tagged proteins are purified under non-denaturing conditions and released with EDTA (2-5 mM) or at pH 3.0, giving about 90% purity, but the monoclonal-antibody matrix is less stable than Ni2+-NTA or Strep-Tactin. Whether M1 binding depends on calcium is controversial, although BIACORE kinetics for FLAG-GFP binding are identical with and without Ca2+.

The 3x FLAG tag (DYKDHDGDYKDHDIDYKDDDDK, a 22-residue tandem FLAG epitope, 2.73 kDa) improves detection and can detect as little as 10 fmol of expressed fusion protein. The FLAG tag contains an internal enterokinase recognition site (its C-terminal DDDDK), so enterokinase cleavage can remove it from a fusion protein.

Strep-tag II (WSHPQFEK, 8 residues, 1.06 kDa) binds Strep-Tactin, a streptavidin variant mutated at positions 44, 45 and 47 that has a higher affinity for the peptide than native streptavidin. Strep-tagged proteins bind in the biotin pocket under physiological buffer conditions and are eluted gently with 2.5 mM desthiobiotin, and the matrix can be regenerated with 4-hydroxy azobenzene-2-carboxylic acid (HABA).

Strep-tag II purification tolerates chelating agents, mild detergents and salt up to 1 M in the buffer, and 6 M urea destroys the Strep-tag II/Strep-Tactin interaction without destroying Strep-Tactin. Biotinylated proteins such as the E. coli carboxyl carrier protein also bind Strep-Tactin, and free biotin or biotinylated proteins can be blocked with avidin. The system is recommended for purifying active fusion proteins under anaerobic conditions, for metal-containing enzymes, and for integrating tagged proteins into membranes, where untagged membrane protein subunits can be co-purified.

The c-myc tag (EQKLISEEDL) is an epitope of the human c-myc protein recognized by the murine monoclonal antibody 9E10, and it keeps this recognition when expressed in a different protein context. It is used for Western blotting, immunoprecipitation and flow cytometry, and c-myc-tagged proteins can be affinity-purified on 9E10 coupled to divinyl sulphone-activated agarose under physiological washing conditions with elution at low pH, which can reduce protein activity, so the tag is mostly a detection tool and rarely used for purification.

The S-tag (KETAAAKFERQHMDS, 15 residues, 1.75 kDa) is a soluble peptide derived from RNase A that binds the 103-residue S-protein, also derived from RNase A, and the complex restores ribonucleolytic activity, which allows a rapid homogeneous assay or colorimetric detection on Western blots. S-tagged proteins bind S-protein matrices, but elution is harsh (3 M guanidine thiocyanate, 0.2 M citrate pH 2, or 3 M magnesium chloride), so cleaving the tag with a protease is recommended to obtain functional protein, and the S-tag is often combined with a second tag.

Glutathione S-transferase (GST) from Schistosoma japonicum is a 26 kDa fusion partner that binds immobilized glutathione, so GST fusion proteins can be purified in one step from crude E. coli lysate and eluted with 10 mM reduced glutathione under non-denaturing conditions. GST fusions are mostly soluble and form dimers, can be detected by enzyme assay or immunoassay, and the tag can protect against intracellular proteases and stabilize the recombinant protein.

Insolubility of GST fusion proteins is associated in several instances with hydrophobic regions, and other insoluble GST fusions contain many charged residues or are larger than 100 kDa. Some insoluble GST fusions can still be purified by glutathione affinity chromatography after solubilization in 1% Triton X-100, 1% Tween, 10 mM dithiothreitol, 0.03% SDS or 1.5% sarcosyl, and sarcosyl inhibits co-aggregation of proteins with bacterial outer membrane components.

The GST tag is removed with a site-specific protease such as thrombin or factor Xa. PreScission protease is human rhinovirus 3C protease fused to GST, so after cleavage both the GST carrier and the protease are removed by affinity chromatography on glutathione agarose.

Maltose-binding protein (MBP, 40 kDa, encoded by malE of E. coli K12) fusions are purified by one-step affinity chromatography on cross-linked amylose and eluted with 10 mM maltose in physiological buffer, with micromolar binding affinity. Binding works at pH 7.0-8.5 and up to 1 M salt, denaturing agents cannot be used, and some fusions do not bind efficiently in 0.2% Triton X-100 or 0.25% Tween 20 while others are unaffected.

MBP increases the solubility of overexpressed fusion proteins in bacteria, especially eukaryotic proteins. A spacer of ten asparagine residues between MBP and the protein of interest increases the chance that a fusion binds tightly to amylose resin, an N-terminal MBP location can reduce translation efficiency, and the MBP tag must be cleaved with a site-specific protease.

Cleavage of MBP fusion proteins with factor Xa has sometimes been ineffective, and introducing a polyglycine region of five amino acids can increase cleavage. The MBP system is widely combined with a small affinity tag.

NusA, E. coli thioredoxin (TrxA) and DsbA are hydrophilic fusion partners that increase the solubility of heterologous proteins in E. coli and so help avoid inclusion bodies, with DsbA acting in both the cytoplasm and periplasm. Proteins with these tags cannot be purified on a specific affinity matrix, so the construct must also carry a small affinity tag, NusA is often combined with a His tag, and the fusion is cleaved with a site-specific protease whose site can serve as the linker peptide.

Large fusion partners such as GST (26 kDa) and MBP (40 kDa) can increase the solubility of the target protein but must be removed for applications such as crystallization or antibody production. Small peptide tags such as Arg5-6 (0.80 kDa), 6xHis (0.84 kDa), FLAG (1.01 kDa), Strep-tag II (1.06 kDa), c-myc (1.20 kDa) and S-tag (1.75 kDa) are less immunogenic and may not need removal, and their effect on tertiary structure and biological activity depends on the tag location and amino acid composition.

Short affinity tags can affect crystals differently. Pyrococcus furiosus maltodextrin-binding protein carrying an Arg-tag crystallized into crystals visually indistinguishable from untagged crystals but differing in mosaicity and diffraction, whereas the FLAG-tagged protein gave crystals of quality very similar to untagged protein, and His-tagged proteins may vary slightly in mosaicity and diffraction compared with the native protein.

Using two different affinity tags on one protein lets one tag serve purification and the other detection, or allows consecutive purification steps for higher purity, and placing two tags at opposite termini yields only full-length product after two affinity chromatography steps. The tandem affinity purification (TAP) tag consists of a calmodulin-binding peptide, a TEV protease cleavage site and protein A for immobilization, and was tested in yeast to purify protein complexes.

Enterokinase recognizes the five-residue sequence D-D-D-D-K-X1 and cleaves at the carboxyl side of the lysine, which makes it often the protease of choice for N-terminal fusions, and sporadic cleavage at other residues occurs at low levels depending on substrate conformation. In a GST-calmodulin fusion with the sequence GSDYKDDDDK-X1-ADQLTEEQIA digested with 0.2 U enterokinase for 16 h at 37 C, cleavage depends on residue X1: alanine 88%, methionine 86%, glycine 74%, tryptophan 67% and proline 61%.

TEV protease recognizes the seven-residue site E-X-X-Y-X-Q-S and cleaves between the conserved glutamine and serine, with E-N-L-Y-F-Q-S as the optimal sequence. Its high specificity, activity on a variety of substrates and efficient cleavage at low temperature make it a suitable tool for removing tags, cleavage is best when the site lies between two domains, and a short flexible linker can improve cleavage that is not optimal.

Thrombin cleaves at temperatures between 20 and 37 C within 0.3-16 h, and unlike enterokinase and factor Xa it leaves two extra amino acids on the target protein. The optimal alpha-thrombin site is X4-X3-P-R[K]-X1'-X2' with hydrophobic X4 and X3 and non-acidic X1' and X2', commonly L-V-P-R-G-S, L-V-P-R-G-F or M-Y-P-R-G-N, and cleavage at P-R-G is more efficient than at P-K-L.

Five glycine residues between the thrombin cleavage site and an N-terminal tag enhance thrombin cleavage, so less enzyme is needed for complete digestion and inappropriate cleavage may be avoided. Digestion is effective in pure Tris buffer at pH 8 and NaCl inhibits thrombin, and thrombin can be removed from the cleaved product by p-amino agarose, gel filtration on Superose 12 FPLC or benzamidine sepharose.

Factor Xa cleaves at the carboxyl side of the four-residue sequence I-E[D]-G-R-X1, where X1 can be any amino acid except arginine and proline, at temperatures from 4 to 25 C, and a factor Xa site between a tag and the protein of interest allows complete removal of N-terminal affinity tags. Factor Xa cleavage has sometimes been ineffective or nonspecific, which can be caused by insolubility of the fusion protein or the presence of denaturing reagents, and a polyglycine region of five amino acids can increase cleavage.

Calmodulin-binding peptide (26 residues, derived from the C-terminus of skeletal-muscle myosin light-chain kinase) binds calmodulin with nanomolar affinity in the presence of 0.2 mM CaCl2, and bound fusion proteins are eluted with EGTA, or EGTA with 1 M NaCl for a second elution. Recovery of fusion proteins is 80-90%, reducing agents and detergents up to 0.1% are compatible, and the system is specific in E. coli because no endogenous E. coli proteins interact with calmodulin, whereas purification in eukaryotic cells is not recommended because many endogenous proteins bind calmodulin in a calcium-dependent manner.

The SBP tag (38 residues, 4.03 kDa) is a streptavidin-binding peptide with a dissociation constant for streptavidin of 2.5 nM, and SBP-tagged proteins are purified on immobilized streptavidin and eluted under mild conditions with 2 mM biotin. C-terminally SBP-tagged proteins have been expressed in bacteria and purified successfully.

The 51-residue chitin-binding domain from Bacillus circulans is commonly combined with a self-splicing intein such as the 454-residue Saccharomyces cerevisiae VMA1 intein. Thiol reagents such as 30-50 mM dithiothreitol, beta-mercaptoethanol or cysteine induce self-cleavage of the thioester bond and release the target protein while the uncleaved precursor and intein tag remain bound to the chitin resin, from which they can be stripped with 1% SDS or 6 M guanidine HCl.

Cellulose-binding domains (CBDs) bind cellulose over a pH range of 3.5 to 9.5. Family I CBDs bind so strongly to crystalline cellulose that release requires urea or guanidine hydrochloride, which makes refolding of the target protein necessary, whereas fusions with family II and III CBDs elute gently from cellulose with ethylene glycol, which is removed by dialysis.