An ambitious EC-funded research initiative on epigenetics advancing towards systems biology 162

Karin Fellinger & Ulrich Rothbauer


Green fluorescent proteins (GFP) and derivates thereof are widely used to study protein localization and dynamics in living cells (Heim and Tsien, 1996; Tsien, 1998). The validation and interpretation of these data, however, requires additional information on biochemical properties of the investigated fluorescent fusion proteins e.g. enzymatic activity, DNA binding and interaction with other cellular components. For these biochemical analyses proteins are mostly fused with a small protein tag (e.g. Histidine-tag, c-Myc, FLAG or hemaglutinin). GFP, the most widely used labelling tag in cell biology is rarely used for biochemical analyses although various mono- and polyclonal antibodies are available (Cristea et al., 2005) (Abcam, Cambridge, UK; Sigma, St. Louis, USA.; Roche, Mannheim, Germany).

We recently generated a GFP-binding protein (GBP) based on a single domain antibody derived from Lama alpaca(Rothbauer et al., 2008). This GFP-binding protein is characterized by a small barrel shaped structure (13 KDa, 2.5 nm X 4.5 nm) and a very high stability (stable up to 70°C, functional within 2 M NaCl or 0.5% SDS).

The GFP-binding protein can be produced in bacteria and purified as a stable monomer. From detailed in vitro binding analysis we determined that one molecule GBP binds one molecule GFP in a stable stoichiometric complex. The dissociation constant (Kd) lies with 0.59 nm within the picomolar range comparable to conventional antibodies.

We further tested the binding properties of GBP to derivates of GFP but also to other fluorescent proteins derived from the red fluorescent protein DsRed(Baird et al., 2000; Campbell et al., 2002). Our binding analysis showed that GBP binds to wtGFP, eGFP and GFPS65T as well as to YFP and eYFP. Interestingly it does not bind to CFP, which is due to the fact, that CFP exhibit an amino acid exchange within the recognized epitope. In addition we could not detect any binding to red fluorescent proteins derived from DsRed.

From detailed crystal structure analysis of GBP-GFP complexes we determined the epitope which is recognized by the GBP. Interestingly GBP recognizes and binds a three dimensional epitope at the beta barrel structure of GFP (Kirchhofer et al., manuscript in preparation). This explains why the GBP does not recognise unfolded or denatured GFP (e.g. on immunoblots).

For immunoprecipitations of GFP fusion proteins we coupled the GBP covalently to monovalent matrices (e.g. agarose beads or magnetic particles) generating a so called GFP-Trap. A direct comparison of the GFP-Trap with conventional antibodies for immunoprecipitation of GFP from crude cell lysates reveal that the GFP-Trap allows a very fast (~ 5 ? 30 min) depletion of GFP from tested samples, which cannot be achieved with conventional antibodies even after 12 h of incubation. Moreover, after precipitation with the GFP-Trap only the antigen (GFP) was detectable on a coomassie gel whereas the typical antibody fragments (light chain, 25 kDa; heavy chain, 50 kDa) could be detected in the bound fraction after precipitation with conventional mono- and polyclonal antibodies (Figure 1) (Rothbauer et al., 2008). The lack of unspecific binding or antibody fragments is one major advantage of the GFP-Trap, because unspecific protein fragments in the bound fraction often interfere with subsequent mass spec analysis of interacting complex partners.

The high affinity binding of the GFP-Trap has also some kind of drawback. Elution of GFP can be only achieved either by applying hot sample buffer containing SDS and β-mercapto ethanol or by acidic elution using 0.1 M glycine pH 2.5 which might interfere with protein stability.

However, since our first findings 2007 we tested and used the GFP-Trap for a number of applications. We demonstrated that the GFP-Trap is a versatile tool to purify GFP-fusions and their interacting factors for biochemical studies including mass spectrometry and enzyme activity assays (Agarwal et al., 2007; Frauer and Leonhardt, 2009; Schermelleh et al., 2007; Trinkle-Mulcahy et al., 2008). Moreover the, GFP-Trap is also suitable for chromatin immunoprecipitations (ChIPs) in cells expressing fluorescent DNA binding proteins. In conclusion the protocol below can be used to perform immunoprecipitation of fluorescent fusion proteins from cell extracts to identify and map protein-protein interactions as exemplarily shown by the dimerization of Dnmt1 (Figures 2 and 3) (Fellinger et al., 2009).

PDF version

Karin Fellinger & Ulrich Rothbauer

Ludwig-Maximilians-University Munich / ChromoTek GmbH
Großhadernerstr. 2 – 82152 Planegg- Martinsried
München, Germany