ã 2002 Oxford University Press
Nucleic Acids Research, 2002, Vol. 30 No. 16 e88
Removal of impurities from transcription factor preparations that alter their DNA-binding properties Liping Sun and Thomas Kodadek* Department of Internal Medicine and Department of Molecular Biology, University of Texas±Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-8573, USA Received April 15, 2002; Revised May 20, 2002; Accepted June 27, 2002
ABSTRACT Biochemical studies of transcriptional activators are important for understanding their detailed mechanism of action. Such experiments generally employ chimeric constructs comprised of fused DNAbinding and activation domains that are expressed in, and puri®ed from, Escherichia coli, since full-length activators are usually dif®cult to express. We report here that such preparations contain chaperone impurities that affect the DNA-binding properties of the activator, for example sharply reducing the halflife of the protein±DNA complex. A simple method to remove these troublesome contaminants is described. INTRODUCTION The association of gene-speci®c transcription factors with sites in promoters is a key element in the regulation of eukaryotic gene expression. Therefore, there has been considerable interest in detailed studies of the DNA-binding properties of transactivators in vitro. Because many native activators are dif®cult to purify in quantities suf®cient for detailed biochemical studies, chimeric polypeptides are often used for this purpose. These species are comprised of a DNAbinding domain (DBD), often that of the potent yeast activator Gal4 (1,2), fused to an activation domain (AD). Our laboratory has employed some of these proteins in experiments directed towards understanding how activator binding to proteasomal proteins (3) and TBP (4,5) affect gene expression. During the course of these studies, we noticed that different fusion proteins puri®ed from Escherichia coli containing the Gal4 DBD had noticeably different DNA-binding properties, speci®cally regarding the half-life of the protein±DNA complex (see below). Furthermore, substantial variation between different preparations of the same protein was noted. We present evidence that one or more E.coli proteins that co-purify with the chimeric activators reduce the kinetic stability of their complexes with DNA in the presence of ATP. This raises a signi®cant and, to the best of our knowledge, previously unrecognized problem in detailed studies of recombinant transcription factors puri®ed from E.coli. A simple method of removing these contaminants from the activator preparations is presented.
MATERIALS AND METHODS DNA All of the DNA-binding studies employed the duplex oligonucleotide obtained from annealing 5¢-CCCCGGAGGACTGTCCTCCGCCC and 5¢-GGGCGGAGGACAGTCCTCCGGGG, the latter having been labeled with T4 polynucleotide kinase. This duplex contains the 17 bp consensus Gal4-binding site (1,6). Radioactive labeling of the 5¢-OH of one of the strands was done using standard methods (7). Proteins GST±Gal4±VP16 was puri®ed as described previously (3). His6-tagged Gal4(1±147) was obtained by cloning a PCR product encoding these residues into a pET vector (Novagen). The vector that expresses Gal4(1±93+768±881) with a His6 tag at the N-terminus was pRJR113 (8). The expression and puri®cation of His6±Gal4 DBD and His6±Gal4(1±93+ 768±881) was performed as follows. BL21 cells carrying the expression vector were grown to an OD600 of 1.0, then induced with 1 mM IPTG for 3 h. Cells were pelleted by low speed centrifugation at 5000 g, washed with cold, double-distilled H2O, frozen with liquid nitrogen and stored at ±80°C. Cells were thawed and resuspended in 3±5 vol resuspension buffer (10 mM Tris pH 8.0, 0.5 mM NaCl, 10% glycerol, 10 mM b-mercaptoethanol and 0.1% Tween-20). After sonication, the lysate was passed over a Ni±NTA column (Qiagen), then washed thoroughly with buffer A (50 mM NaH2PO4 pH 8.0, 300 mM NaCl and 20 mM imidazole). The tagged protein was eluted with 250 mM imidazole. Proteins were dialyzed and stored in buffer D (20 mM HEPES pH 7.5, 50 mM KCl, 1 mM b-mercaptoethanol, 20 mM ZnSO4, 0.1 mM EDTA, 10% glycerol and 5 mM MgCl2). Gel shift experiments Aliquots of 2 ml of 103 gel shift buffer (200 mM HEPES pH 7.5, 500 mM KCl, 50 mM MgCl2, 5 mM b-mercaptoethanol, 5 mM EDTA, 50% glycerol, 400 mM ZnCl2), 0.1 ml of 10 mg/ml BSA, 0.1 ml of 10 mg/ml carrier DNA, protein and radiolabeled DNA were added to a ®nal volume of 20 ml. The reactions were loaded onto a 7% native polyacrylamide gel that had been pre-run for 1 h. After electrophoresis, the gel was dried and the radioactive bands were visualized using a phosphorimager.
*To whom correspondence should be addressed. Tel: +1 214 648 1239; Fax: +1 214 648 1450; Email:
[email protected]
e88 Nucleic Acids Research, 2002, Vol. 30 No. 16
PAGE 2 OF 4
Figure 1. (A) Coomassie blue-stained gel showing the af®nity-puri®ed GST±Gal4±VP16 preparation. (B) Rapid dissociation of the GST±Gal4±VP16´DNA complex. Radiolabeled DNA (2 3 10±9 M) was titrated with GST±Gal4±VP16 to a point at which ~90% of the DNA was bound by the protein (lanes 1 and 3). The GST±Gal4±VP16´DNA complex was then challenged with a 10-fold excess of unlabeled DNA (lane 3) or a 5-fold excess of Gal4 DBD (residues 1±147). After a 5 min incubation, the amount of the original labeled complex remaining was assessed by native gel electrophoresis. Lanes 1 and 3 represent control experiments identical to those shown in lanes 2 and 4 except that the competing DNA or protein was omitted. (C) Effect of ATP concentration on the half-life of the GST±Gal4±Vp16´DNA complex. Samples were incubated for 5 min after the addition of a 5-fold excess of the Gal4 DBD (lanes 2±4). Lane 1 represents a control experiment in which the Gal4 DBD was omitted. The concentration of ATP in the solution is shown above each lane.
Treatment of immobilized GST±Gal4 AD Glutathione±agarose beads were saturated with GST±Gal4 AD (residues 841±875, puri®ed as described previously; 9) and these beads were then incubated with the activator preparation such that the GST AD protein was present in 50-fold excess over the chimeric activator. The beads were then pelleted by centrifugation at 12 000 r.p.m. for 1 min in a benchtop centrifuge and the supernatant removed using a micropipette. RESULTS AND DISCUSSION Recombinant GST±Gal4±VP16 was puri®ed from E.coli by glutathione±agarose af®nity chromatography. The preparation was ~90±95% pure, as determined by SDS±PAGE and staining with Coomassie blue (Fig. 1A). The recombinant protein, as expected, binds to a 21 bp DNA oligonucleotide bearing the 17 bp Gal4 consensus binding site (1,6), as visualized in a gel retardation assay (Fig. 1B, lane 1). In an experiment designed to measure the off-rate of GST±Gal4±VP16 under conditions similar to those used for in vitro transcription experiments (3,10), the protein was incubated with radiolabeled DNA for 1 h at room temperature in the presence of ATP. A 10-fold excess of unlabeled DNA or 5-fold excess of Gal4 DBD protein (residues 1±147) was then added to trap GST±Gal4±VP16 radiolabeled DNA complexes that dissociated with time. As shown in Figure 1B (lanes 2 and 4), most of the GST±Gal4±VP16´DNA complex dissociated within 5 min. This was surprisingly fast, given that the equilibrium dissociation constant (Kd) of Gal4 DBD´DNA complexes containing the consensus binding site has been reported to be in the low nanomolar range (6). Experiments
Figure 2. Treatment of the mini-Gal4 preparation with excess immobilized GST±Gal4 AD provides a preparation that forms DNA complexes of high kinetic stability. Ni±NTA-puri®ed His6±mini-Gal4 (residues 1±93+768±881) either was or was not treated with GST±Gal4 AD immobilized on glutathione±agarose. Each preparation was mixed with excess labeled DNA. Lanes 1 and 2 show that each preparation formed identical amounts of protein´DNA complex. Each complex was then challenged with 10-fold excess of unlabeled DNA and the amount of the labeled complex remaining was assayed 5 and 15 min after addition of the competitor.
using other Gal4 DBD-containing constructs showed that this was not due to the fusion of GST to the N-terminus of the Gal4 DBD. As shown in Figure 2, a `mini-Gal4' protein comprised
PAGE 3 OF 4
of the DBD (residues 1±93) fused to a C-terminal fragment (residues 768±881) (5,8) also dissociated from the consensus DNA binding site rapidly under the same conditions when challenged with excess unlabeled DNA (compare lanes 5 and 6 to lane 1). Finally, western blot analysis of the proteins at the end of the incubations did not reveal signi®cant degradation, eliminating the possibility that loss of the complex was due to proteolysis (data not shown). After exploring a number of experimental parameters, we found that the presence or absence of ATP had a major effect on the stability of the DNA´protein complexes. For example, as shown in Figure 1C, the amount of the DNA´GST±Gal4±VP16 complex remaining after a 5 min incubation in the presence of excess Gal4 DBD varied greatly when assayed at ATP concentrations from 0 to 0.8 mM. This was curious, since none of the Gal4 derivatives are ATPases nor are they known to bind ATP. Furthermore, we have previously performed experiments using the same mini-Gal4 construct and some preparations have exhibited much slower kinetics of dissociation from the same DNA site (4). Finally, as will be reported elsewhere, Gal4 DBD proteins lacking an activation domain exhibited kinetically stable binding to DNA that was consistent from preparation to preparation. Taken together, these observations suggest that the unexpected lability of the protein´DNA complexes was tied to the presence of an AD and ATP. One idea was that a bacterial ATPase associated with the hydrophobic AD might somehow act to destabilize the protein´DNA complex. As we will show elsewhere, yeast chaperonins associated with Gal4 through the AD can reduce the lifetime of its complexes with DNA (A.Delahodde, L.Sun, F.Gonzalez, A.Ferdous, T.Kodadek and S.A.Johnston, submitted for publication). If bacterial chaperones were loosely associated with the recombinant activator, this might explain the variability encountered in different preparations or between related constructs, since minor variations in the washing conditions, etc. might result in variable levels of contamination. Whether or not this interpretation is accurate, as a practical issue it was important to devise a simple method by which the putative contaminants could be removed, providing preparations that re¯ected the true DNA-binding properties of the recombinant proteins. After experimenting with several protocols, we found that the simplest and most effective procedure was to incubate the af®nity-puri®ed preparation with an excess of immobilized Gal4 AD. For example, the mini-Gal4 (residues 1±93+768±881) fraction puri®ed on Ni±NTA±agarose was incubated with GST±Gal4(841±875) (the 34 residue core Gal4 AD) (9) immobilized on glutathione±agarose. The beads were then pelleted and the chimeric activator in the supernatant was assayed. As shown in Figure 2 (lanes 3 and 4), little or no dissociation of the complex was observed over the course of 15 min after challenging with excess unlabeled DNA for the GST±Gal4 AD-treated sample. The same experiment using the preparation that had not been treated with the AD beads resulted in rapid dissociation of the complex (Fig. 2, lanes 5 and 6). The identities of the contaminating proteins responsible for this effect are unclear. To explore this issue, all of the bands visible in the Coomassie-stained gel (Fig. 1A) were excised, digested with trypsin and the resultant peptides were characterized by mass spectroscopy. All of the bands with a
Nucleic Acids Research, 2002, Vol. 30 No. 16 e88 higher mobility than the activator proved to be proteolytic degradation products that included the GST protein. However, the light band with an apparent molecular mass of ~90 kDa (Fig. 1A) proved to be bacterial DnaK (20 peptide matches; data not shown). This was interesting in that DnaK is an ATPdependent chaperone protein (11). In retrospect, it is not surprising that chaperones such as DnaK would co-purify to some extent with chimeric activators. Acidic activation domains such as those found in VP16 and Gal4 are unusual in having a wealth of hydrophobic side chains that are well displayed in solution due to the presence of several interspersed acidic residues (12,13). This is exactly the kind of peptide chaperones bind. Chaperonins can transiently unfold associated proteins and it is plausible to suggest that this could lead to disruption of the protein±DNA complex. This model would accommodate the strong ATP dependence of the effect we have observed. However, we stress that proof for this sort of model is lacking. We added recombinant DnaK to the ADtreated mini-Gal4 preparation and did not observe more rapid dissociation kinetics (data not shown). However, DnaK is known to require other proteins to function ef®ciently (14), so this could explain the negative result, or it could be that some other protein present at levels too low to be detected by mass spectrometry could be responsible for this effect. While it would be possible, in principle, to purify this factor using the dissociation effect as an assay to follow activity, we decided not to pursue this line of investigation since this is not a physiologically relevant event. The important point here is that preparations of these activators can contain impurities of bacterial protein(s), possibly chaperones, that can signi®cantly affect the DNA-binding chemistry of the recombinant protein in the presence of ATP. Therefore, it is important to pass these preparations over an AD column or employ some other means to reproducibly obtain preparations that re¯ect the intrinsic DNA-binding properties of the Gal4 DBD. REFERENCES 1. Carey,M., Kakidani,H., Leatherwood,J., Mostashari,F. and Ptashne,M. (1989) An amino-terminal fragment of GAL4 binds DNA as dimer. J. Mol. Biol., 209, 423±432. 2. Marmorstein,R., Carey,M., Ptashne,M. and Harrison,S.C. (1992) DNA recognition by GAL4: structure of a protein-DNA complex. Nature, 356, 408±414. 3. Ferdous,A., Gonzalez,F., Sun,L., Kodadek,T. and Johnston,S.A. (2001) The 19S regulatory particle of the proteasome is required for ef®cient transcription elongation by RNA polymerase II. Mol. Cell, 7, 981±991. 4. Xie,Y., Denison,C., Yang,S.-H., Fancy,D.A. and Kodadek,T. (2000) Biochemical characterization of the TATA-binding protein-Gal4 activation domain complex. J. Biol. Chem., 275, 31914±31920. 5. Xie,Y., Sun,L. and Kodadek,T. (2000) TATA-binding protein and the Gal4 transactivator do not bind to promoters cooperatively. J. Biol. Chem., 275, 40797±40803. 6. Vashee,S., Xu,H., Johnston,S.A. and Kodadek,T. (1993) How do Zn2Cys6 proteins distinguish between similar upstream activation sites? Comparison of the in vivo and in vitro DNA-binding speci®cities of the GAL4 protein. J. Biol. Chem., 268, 24699±24706. 7. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 8. Reece,R.J., Rickles,R.J. and Ptashne,M. (1993) Overproduction and single-step puri®cation of GAL4 fusion proteins from Escherichia coli. Gene, 126, 105±107. 9. Melcher,K. and Johnston,S. (1995) GAL4 interacts with TBP and coactivators. Mol. Cell. Biol., 15, 2839±2848.
e88 Nucleic Acids Research, 2002, Vol. 30 No. 16 10. Wootner,M., Wade,P.A., Bonner,J. and Jaehning,J.A. (1991) Transcriptional activation in an improved whole cell extract from Saccharomyces cerevisiae. Mol. Cell. Biol., 9, 4555±4560. 11. Lund,P.A. (2001) Microbial molecular chaperones. Adv. Microb. Physiol., 44, 93±140. 12. Cress,W.D. and Triezenberg,S.J. (1991) Critical structural elements of the VP16 transcriptional activation domain. Science, 251, 87±90.
PAGE 4 OF 4 13. Leuther,K.K., Salmeron,J.M. and Johnston,S.A. (1993) Genetic evidence that an activation domain of GAL4 does not require acidity and may form a beta sheet. Cell, 72, 575±585. 14. Yura,T., Nagai,H. and Mori,H. (1993) Regulation of the heat-shock response in bacteria. Annu. Rev. Microbiol., 47, 321±350.
