Barnase:Barstar affinity tag system

Affinity chromatography is based on a pair of molecules that ideally strongly bind to each other in a highly specific manner. One of the most prominent affinity pairs is streptavidin binding to the small molecule biotin. Streptavidin is a 16kDa protein produced by the bacteriaStreptomyces avidinii and was first described by Tausig and Wolf (1964). It binds biotinylated proteins with extremely high affinity of KD=10^-14M (Green, 1990) and once formed the bond can resist broad ranges of pH, temperatures, organic solvents or dena-turing compounds. However, besides these advantageous features, the affinity pair also has severe disadvantages. (1) Biotin is present in all living cells at least in small amounts.

Hence, whenever using streptavidin resins to extract proteins from a lysate, endogenous biotin will cause background binding, which might interfere with proper evaluation of re-sulting data. (2) Streptavidin itself is rather small, but it forms a homotetramer with a size of around 60kDa. Coupling the tetramer to the resin causes two major problems:

either the tetramer is attached by four streptavidin subunits building up a meshwork which might interfere with binding and accessibility of large molecules to the resin, or, if a tetramer is attached by fewer than four subunits, a leakage of streptavidin from the matrix is possible. (3) Streptavidin itself is difficult to express and to purify in large amounts fromE.coli. Even shortened versions, that are reported to show a higher level of

3.3 Novel tools for protein purification: a MADA resin and Barnase:Barstar as high affinity pair 94

expression as the wild type (Sanoet al., 1995), have severe solubility problems. (4) Strep-tavidin recognizes biotinylated proteins, a modification, that is not efficiently introduced to proteins containing a biotinylation site when expressed inE.coli. Thus, the protein has to be either co-expressed with the biotinylation enzyme BirA, or a chemical or enzymatic modification has to be performedin vitroafter purification. In both cases, the production of biotinylated proteins that can be recognized by streptavidin, requires additional hands on time as compared to simply expressing a fusion protein.

The extracellular RNase Barnase and its intracellular inhibitor Barstar show advantages over the streptavidin:biotin affinity pair in all four aspects. The pair is formed by two polypeptide chains, thus a target protein must be fused to either Barstar or Barnase but no additional modification is required. The small size of the proteins and their monomeric nature enable the direct coupling of defined amounts to a stationary phase without the risk of forming an intertwined layer that is prone to interfere with proper target protein bind-ing or leakage of non covalently bound subunits. The strong interaction between Barnase and Barstar (K_D= 10^-14, Schreiber and Fersht, 1993) that even tolerates ionic strengths of

∼5M NaCl, prevents leakage of the fused target protein once bound to the affinity partner.

This however requires elution of the target protein by protease cleavage and not by any means of competition as it is the case for His-tagged proteins and imidazole, respectively.

Elution by protease cleavage is highly specific. Only proteins containing the protease recognition site are efficiently cleaved off the resin whereas proteins unspecifically binding the resin remain unaffected. This is beneficial in sample preparation, mainly for subse-quent experiments that either depend on sensitive detection such as mass spectrometry or require highly homogenous samples, e.g. dynamic light scattering or protein crystal-lization. In addition, also unspecific binding of endogenous proteins to the immobilized affinity protein is very low as the Barnase:Barstar interaction is highly specific. Barstars only function in the bacterial cell is to bind and thus sequester the enzymatic activity of barnase. Barnase however functions extracellularly.

As previously described, it must be possible to efficiently express and purify both proteins of the affinity pair also in absence of the binding partner. Barnase, which hydrolyzes ri-bonucleotides by an acid-base mechanisms using residues Glu73and His102 (Mossakowska et al., 1989), causes lethality when expressed inE.coli. Hence, the active center was mu-tated by replacing His₁₀₂ with aspartic acid. The introduced negative charge prevents the catalytic mechanism and barnase can be expressed to high levels in E.coli. Although the wild type barstar protein is able to bind the mutated barnase protein (Hartley, 1993), a compensatory positive charge was introduced by replacing Cys42to a lysine. The obtained crystal structure at 1.98 ˚Aresolution revealed that the aspartic acid and lysine side chains are in very close proximity (2.7 ˚A) probably forming an additional shielded salt bridge,

3.3 Novel tools for protein purification: a MADA resin and Barnase:Barstar as high affinity pair 95

which makes the complex comprised of this mutant protein pair extremely stable.

Barnase seems to have a very stable fold. It can be expressed in a soluble manner to high amounts and it supports other proteins solubility when used as N-terminal fusion tag as in case of full length streptavidin (preliminary data, not shown). By contrast, Barstar has severe problems in solubility when expressed alone. Although its solubility is supported by ZZ-fusion tags, H14-brSUMO or H14-brSUMO-MBP tags do not prevent insolubility of the protein. Barstar contains a parallel β-sheet located on the opposite site of the Barnase:Barstar interface, which seems to be imperfectly packed. Hydrophobic residues are solvent exposed which makes the entire protein more prone to aggregation. By mu-tating the β-sheet region, solubility could be slightly increased, however it still did not result in an overall stable fold. Thus it is not suitable as N-terminal fusion tag, which was also experimentally shown by expression of a Barstar-brSUMO-GFP construct resulting in insoluble protein (data not shown).

Interestingly, although much is known about the binding and thus inhibition of Barnase by Barstar, the question what happens to the complex in the cell in terms of degradation remains unanswered. We speculate that the imperfect folding of the Barstar molecule, which might even be enhanced upon barnase binding, is beneficial for a short protein half-life and thus a quick degradation of the Barnase:Barstar complex, removing a highly toxic protein from the cellular interior. How this degradation occurs, is however unclear.

Recently, an ubiquitin - like system was described for Mycobacterium tuberculosis (re-viewed in Burns and Darwin, 2010). A prokaryotic ubiquitin-like protein (Pup) can be attached to lysine residues of the target protein with the help of two additional proteins, Dop and PafA, thus targeting the protein to the protease complexes, ClpP respectively.

TheB.amyloliquefaciens Barstar proteins has two consecutive N-terminal lysine residues, that are also fully exposed when looking at the structure, thus being putative degradation targeting sites.

The Barstar solubility problem was solved by switching from the B.amyloliquefaciens se-quence to the sese-quence identified from the thermophilic Bacillus species G.thermoglucosida-sius, which is known to ideally grow at temperatures ranging from 40℃ to 70℃ (Nazina et al., 2001). Proteins from thermophilic organisms tend to have a more stable fold and thus could result in stable and soluble barstar protein when expressed and purified alone.

Comparing complexes containing different barnase and barstar proteins by Thermofluor revealed that the pair containing not only Barstar but also the corresponding Barnase protein from G.thermoglucosidasius has the highest melting temperature and thus is the most stable complex tested. The elevated melting temperature of the complex can again be explained by the increased stability of the individual proteins. When comparing the Barstar sequences of the two organism, a significant N-terminal extension can be observed