An efficient tandem affinity purification procedure for interaction proteomics in mammalian cells

Tilmann Bürckstümmer, Keiryn L Bennett, Adrijana Preradovic, Gregor Schütze, Oliver Hantschel, Giulio Superti-Furga & Angela Bauch

Research Center for Molecular Medicine (CeMM), Lazarettgasse 19/3, 1090 Vienna, Austria.

The TAP method involves fusion of the TAP tag to the target protein and introduction of the construct into the host cell. The yeast TAP tag (yTAP tag) consists of two immunoglobulin G (IgG)-binding units of protein A from Staphylococcus aureus, a cleavage site for the tobacco etch virus (TEV) protease and a calmodulin binding peptide (CBP; Fig. 1). The fusion protein and associated components are recovered from cell extracts by two sequential purification steps. In the first step, the protein A moiety of the TAP tag is bound to IgG sepharose, and complexes are released by TEV-protease cleavage. In the second step, the CBP moiety is bound to calmodulin sepharose in the presence of calcium and then eluted with EGTA. Although this technology has been applied successfully in mammalian cells^{12, 14}, it also has limitations. First, and most importantly, the overall yield of the process is very low. As a consequence, TAP requires large initial quantities of cells (typically 5 10⁸-1 10⁹ cells). Second, the technology has not been applied to highly differentiated cells (for example, neuronal cells). Third, primary cells have not been amenable to the TAP procedure because the availability of cellular material has been limiting.

Figure 1. Schematic representation of TAP. (a) Schematics of the four TAP tags that were obtained by permutation of protein A and protein G with CBP and SBP. (b) Overview of the TAP protocol. During the first step, TAP-tagged proteins are sequestered by IgG sepharose (1) and released by TEV-protease cleavage (2). TEV protease–cleaved proteins are then bound to streptavidin sepharose (3) and eluted with 1 mM biotin (4). (c) Nucleotide and amino-acid sequence of the GS-TAP tag which is comprised of protein G and SBP. Full Figure and legend (74K)

Here we compare the TAP tag originally developed for yeast with four newly designed TAP fusion protein cassettes encoding alternative affinity binding moieties optimized for use in mammalian systems. Using one of the optimized TAP tags, we obtained overall complex yields that were an order of magnitude higher as compared to the original TAP tag. This increase in affinity purification efficiency permitted the purification and identification of a protein complex (Ku70-Ku80) from 5 10⁷ HEK293 cells.

We compared the AC-TAP tag to the yTAP tag by fusing each to the N terminus of IKK as available data indicated that N-terminal TAP tag fusion proteins of IKK are functional¹². In general, it is advisable to incorporate all available information to decide whether to insert the tag at the N or the C terminus of a protein. In the absence of any information, it is recommended to generate both versions^{7, 15}. IKK is part of the multiprotein IB kinase (IKK) complex that mediates activation of the transcription factor NF-B¹⁶. The regulatory subunit IKK is known to copurify with the two catalytic subunits IKK and IKK¹². We monitored the efficiency of TAP assaying for purification of the bait (IKK) and one key interactor, IKK, by immunoblotting.

We introduced yTAP-IKK or AC-TAP–IKK into HEK293 cells by retroviral gene transfer. Both proteins were expressed at similar levels (Fig. 2); the humanized AC-TAP–IKK being only marginally better expressed. The overall purification process was comparable for the two constructs, both at the level of the bait and a key interactor (IKK; Fig. 2). The entire procedure, however, was very inefficient in both cases and only led to the identification of interacting proteins when using large quantities of cells (5 10⁸). As the AC-TAP tag performed at least as well as the original yTAP tag, we used the AC-TAP tag as a reference and basis for improvement.

Figure 2. Comparison of yTAP and AC-TAP. yTAP-IKK and AC-TAP–IKK were purified by TAP from 5 108 cells. Recovery of the bait (IKK) and one key interactor (IKK) was tracked throughout the entire TAP procedure by immunoblotting using IKK-specific antiserum and IKK-specific antiserum. XT, cell extract; SN1, supernatant after IgG binding; TEV, eluate after TEV-protease cleavage; SN2, supernatant after calmodulin binding; BB, boiled beads, that is, the final eluate. Numbers below the lanes indicate which percentage of the total sample was loaded on each lane. Full Figure and legend (25K)

Next we compared the efficiencies of the four TAP tag variants. We stably expressed TAP-tagged fusion proteins with IKK in HEK293 cells and subjected them to TAP. Whereas AC-TAP– and GC-TAP–tagged proteins were eluted by boiling in SDS sample buffer, AS-TAP– and GS-TAP–tagged proteins were eluted with 1 mM biotin (AS/Bio and GS/Bio). To monitor the efficiency of the biotin elution, we then boiled the streptavidin sepharose in SDS sample buffer (AS/BB and GS/BB). The TAP-tagged IKK fusion proteins were expressed at comparable levels (Fig. 3a). Notably, the tags gave different overall purification yields. Although the GC-TAP tag did not result in an improvement over the reference AC-TAP tag, both the AS-TAP and the GS-TAP tags led to a dramatic increase in the overall recovery of the bait (Fig. 3a). This is striking taking into account that both elution fractions should be added for quantitative comparison with the other tags. When calculating the overall yield for the procedure, it is important to consider that protein G and protein A differ in their ability to be detected by different antibody subtypes from different species (Supplementary Fig. 1). We estimate that the difference in recognition between the AC- and the GS-TAP tag amounts to a factor of 2. Using this factor, we calculated the differences in the overall recovery rates as exemplified for the comparison of the AC- with the GS-TAP tag: the difference in recovery (12) is divided by the difference in expression level (3.4) and multiplied by the recognition factor (2). Overall, we estimate an improvement in yield of bait recovery of about tenfold (Fig. 3a).

Figure 3. Comparison of the different TAP tag variants. (a) AC-TAP–, AS-TAP–, GC-TAP– and GS-TAP–tagged IKK were purified by TAP from 5 108 cells and compared with respect to the overall recovery of the bait (IKK; top) and one key interactor (IKK; bottom). Note that the protein G–containing constructs (GC, GS) reacted specifically with the mouse secondary antibody used for the detection of IKK (compare lower band). IKK levels in the top panel were quantified by the LI-COR Odyssey system (middle). (b) Purification of AC-TAP– and GS-TAP–tagged IKK including all purification steps. Sample nomenclature is the same as in Figure 2. (c) AC-TAP– and GS-TAP–tagged IKK were purified by TAP from 5 107 cells (10% of standard starting material) as described in a. Final eluates were compared to AC-TAP-tagged IKK purified from 5 108 cells (100% of standard starting material) by analyzing both recovery of IKK and IKK. Full Figure and legend (52K)

We not only detected the increased yield at the level of the bait (IKK), but also at the level of the key interactor IKK (Fig. 3a). There was no major difference in performance between the protein A– and the protein G–containing TAP tags (AC was similar to GC, AS was similar to GS; Fig. 3a). We obtained similar results for another bait (MyD88; Supplementary Fig. 1). For simplicity, we used the GS-TAP tag for all subsequent analyses.

One of the most cumbersome limitations of the present TAP protocol is the massive amount of cellular material required. Based on the improved efficiency of the GS-TAP tag, we investigated whether it was possible to reduce the quantity of starting material. We purified both AC-TAP and GS-TAP–IKK from 5 10⁷ cells (10% of the standard starting material). We compared the final eluates of the purifications to the final eluate of AC-TAP–IKK recovered from 5 10⁸ cells (100% of the standard starting material). Purification of GS-TAP–IKK from one tenth of the usual starting material retrieved as much IKK and interacting IKK as obtained using AC-TAP–IKK under standard conditions (Fig. 3c).

To circumvent this problem, we determined the protein input and output by using two different recombinant standards: GS-TAP–IKK and untagged IKK, both expressed and purified from Escherichia coli. We created a serial dilution of the cell extract from HEK293/GS-TAP–IKK (that is, the starting material) and compared it to a serial dilution of recombinant purified GS-TAP-IKK. Comparison of the dot blot signal intensities revealed that 100 g of cell extract contained approximately 169 ng of purified GS-TAP–IKK (Fig. 4a). In other words, 36 mg of cell extract (obtained from 5 10⁷ cells) contain 69 g (690 pmol) of purified GS-TAP–IKK.

Figure 4. Quantification of the overall process. (a) A twofold serial dilution of 100 g of cell extract (XT) from either parental HEK293 cells (–) or HEK293 cells expressing GS-TAP–IKK was compared to a serial dilution of 100 ng recombinant purified GS-TAP–IKK derived from E. coli. Spotted proteins were analyzed by immunoblotting (IKK). Signal intensities were quantified using the LI-COR Odyssey system and provided the basis for calculating the amount of GS-TAP-IKK that is present in the cell extract. (b) HEK293 cells expressing GS-TAP–IKK were subjected to TAP. Serial dilution of 5% of the final TAP eluate was compared to a serial dilution of 1 g of recombinant purified IKK derived from E. coli. Spotted proteins were analyzed as described in a. Full Figure and legend (21K)

We compared the quantity of IKK in the final TAP eluate to the amount of recombinant purified untagged IKK (Fig. 4b). Based on this comparison, 5% of the final TAP eluate contained approximately 98 ng of purified IKK. This means that 100% of the final TAP eluate contained the equivalent of 2 g (35 pmol) of IKK. In conclusion, the overall yield amounts to approximately 5% of the input material (35 pmol from 690 pmol). We obtained similar results for three other baits in two other cell lines (Supplementary Fig. 1).

Finally, we applied the improved procedure to purify a protein complex that had not yet been isolated by the TAP procedure. We chose the DNA-dependent protein kinase (DNA-PK), which consists of a catalytic subunit (DNA-PK_cs) and two cofactors (Ku70 and Ku80), as a model complex. The DNA-PK holoenzyme is of fundamental importance for DNA double-strand-break repair (nonhomologous end-joining) and for VDJ recombination (for review, see ref. 20). We expressed GS-TAP–Ku70 and GS-TAP–Ku80 in HEK293 cells. We assessed the functionality of GS-TAP–Ku70 by monitoring DNA-PK activity associated with Ku70 (Supplementary Fig. 2 online). We purified GS-TAP–Ku70 and GS-TAP–Ku80 from 5 10⁷ cells. We included untransduced HEK293 cells as negative control (Supplementary Fig. 2). Ku70 and Ku80 always copurified in equimolar amounts as judged by silver staining (Fig. 5a), indicating that Ku70 and Ku80 form a stoichiometric complex. This is in agreement with the observation that the biologically functional unit is the Ku70-Ku80 heterodimer²¹.

Figure 5. TAP of the Ku70-Ku80 complex. (a) GS-TAP–Ku70 and GS-TAP–Ku80 were purified from 5 107 HEK293 cells by TAP. Purified proteins were visualized by silver staining and subsequently identified by LC-MSMS. The final data set contains a total of 39 proteins that were grouped according to biological function (Table 1). (b) The iHOP database was searched to identify previously reported interactions within the Ku70-Ku80 interactome. Proteins were grouped in one module if there was evidence in the literature that the proteins interact with one another (solid lines). Interaction data stemming from yeast two-hybrid data are indicated with dashed lines. Full Figure and legend (50K)

Table 1. Specific interactors of Ku70 Full Table

	Top

Figure 6. Quantitative overview of the TAP procedure. A typical TAP purification is done from 5 107 cells, yielding 35 mg of total cell extract. Based on the quantification presented in Figure 4, the cell extract contains 70 g (700 pmol) of bait protein. The final eluate contains 2 g (35 pmol) of bait, which correspond to an overall yield of 5%. Depending on the stoichiometry of the interactions (2:1 to 1:20), 0.2–4 g (1.5–70 pmol) of interactors will be copurified. Full Figure and legend (32K)

Depending on the bait under consideration, the number of cells required ranges from 5 10⁷ to 1 10⁹. For HEK293 cells, 5 10⁷ cells are routinely used, and this typically yields 35 mg of total protein. For a standard bait stably expressed after retroviral gene transfer, this may correspond to approximately 700 pmol of TAP-tagged protein. The two subsequent steps (first affinity binding step and elution, and second affinity binding step and elution) show reproducible levels of efficiency. In the first step, one can expect to bind 40% of the bait and retrieve 30% thereof after TEV-protease cleavage. In the second step, approximately 75% of the remaining material is recovered. The biotin elution is the step that shows the greatest bait-dependent variability with an average of 50% recovery. If elution by boiling in SDS sample buffer is chosen, the yield can be increased at the expense of specificity. Overall, the TAP procedure recovers about 5% of the bait present in the lysate. Therefore, the overall amount of purified bait expected is in the picomole range (35 pmol in the example given above). If this is true for the bait, what about the interacting proteins? Depending on the stoichiometry of the interacting proteins (for example, 2:1 to 1:20), copurification of between 70 and 1.5 pmol of the interacting proteins can be envisaged. As a reference, the lower amount is visible by silver stain and is well within the limits of detection of nano liquid chromatography–tandem mass spectrometry (LC-MSMS) in a relatively complex sample.

	Top

We obtained the yTAP cassette from Euroscarf. The new TAP cassettes were synthesized by Midland Certified. We cloned all cassettes into a Moloney murine leukemia virus–based vector³⁴. We introduced an IRES-GFP derived from pBMN-I-GFP to monitor infection efficiency³⁵. We obtained the cDNAs for IKK, Ku70, Ku80, MyD88, RIG-I, Btk and Tbk1 from RZPD, amplified them by PCR and introduced them into the expression vector using the Gateway technology (Invitrogen). We stably transduced HEK293 cells by retroviral gene transfer¹².

We obtained the IKK-specific antiserum from Santa Cruz (FL-419). We obtained the IKK-specific antibody from Imgenex.

We lysed cells in lysis buffer (50 mM Tris-HCl (pH 7.5), 125 mM NaCl, 5% glycerol, 0.2% NP-40, 1.5 mM MgCl₂, 25 mM NaF, 1 mM Na₃VO₄ and protease inhibitors) and cleared the lysate by centrifugation. We incubated the lysate with rabbit-IgG agarose (Sigma) at 4 °C for 2 h. We washed bound proteins with lysis buffer and then with TEV-protease cleavage buffer (10 mM Tris-HCl (pH 7.5), 100 mM NaCl and 0.2% NP-40) and eluted by addition of 40 g TEV protease (16 °C for 1 h). We incubated the TEV-protease cleavage product with calmodulin sepharose (Amersham) in the presence of 2 mM CaCl₂ or with Ultralink Immobilized Streptavidin Plus (Pierce) at 4 °C for 1 h. For the CBP-containing tags, we eluted bound proteins by boiling in SDS sample buffer. For the SBP-containing tags, we eluted bound proteins with 1 mM D-biotin (Sigma). To assess the efficiency of the biotin elution, we boiled the remaining streptavidin beads in SDS sample buffer.

	Top

1. Alberts, B. The cell as a collection of protein machines: preparing the next generation of molecular biologists. Cell 92, 291–294 (1998). | Article | PubMed | ISI | ChemPort |
2. Papin, J.A., Hunter, T., Palsson, B.O. & Subramaniam, S. Reconstruction of cellular signalling networks and analysis of their properties. Nat. Rev. Mol. Cell Biol. 6, 99–111 (2005). | Article | PubMed | ISI | ChemPort |
3. Bauch, A. & Superti-Furga, G. Charting protein complexes, signaling pathways, and networks in the immune system. Immunol. Rev. 210, 187–207 (2006). | Article | PubMed | ChemPort |
4. Hinsby, A.M. et al. A wiring of the human nucleolus. Mol. Cell 22, 285–295 (2006). | Article | PubMed | ChemPort |
5. Fox, A.H. & Lamond, A.I. Nuclear processes controlled by molecular machines. Genome Biol. 3 REPORTS4016 (2002).
6. Aebersold, R. & Mann, M. Mass spectrometry-based proteomics. Nature 422, 198–207 (2003). | Article | PubMed | ISI | ChemPort |
7. Puig, O. et al. The tandem affinity purification (TAP) method: a general procedure of protein complex purification. Methods 24, 218–229 (2001). | Article | PubMed | ISI | ChemPort |
8. Rigaut, G. et al. A generic protein purification method for protein complex characterization and proteome exploration. Nat. Biotechnol. 17, 1030–1032 (1999). | Article | PubMed | ISI | ChemPort |
9. Gavin, A.C. et al. Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415, 141–147 (2002). | Article | PubMed | ISI | ChemPort |
10. Gavin, A.C. et al. Proteome survey reveals modularity of the yeast cell machinery. Nature440, 631–636 (2006). | Article | PubMed | ChemPort |
11. Ho, Y. et al. Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry. Nature 415, 180–183 (2002). | Article | PubMed | ISI | ChemPort |
12. Bouwmeester, T. et al. A physical and functional map of the human TNF-alpha/NF-kappa B signal transduction pathway. Nat. Cell Biol. 6, 97–105 (2004). | Article | PubMed | ISI | ChemPort |
13. Butland, G. et al. Interaction network containing conserved and essential protein complexes in Escherichia coli. Nature 433, 531–537 (2005). | Article | PubMed | ISI | ChemPort |
14. Brajenovic, M., Joberty, G., Kuster, B., Bouwmeester, T. & Drewes, G. Comprehensive proteomic analysis of human Par protein complexes reveals an interconnected protein network. J. Biol. Chem. 279, 12804–12811 (2004). | Article | PubMed | ISI | ChemPort |
15. Terpe, K. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol. 60, 523–533 (2003). | PubMed | ISI | ChemPort |
16. Ghosh, S. & Karin, M. Missing pieces in the NF-kappaB puzzle. Cell 109, S81–S96 (2002). | Article | PubMed | ISI | ChemPort |
17. Sjobring, U., Bjorck, L. & Kastern, W. Streptococcal protein G. Gene structure and protein binding properties. J. Biol. Chem. 266, 399–405 (1991). | PubMed | ISI | ChemPort |
18. Sauer-Eriksson, A.E., Kleywegt, G.J., Uhlen, M. & Jones, T.A. Crystal structure of the C2 fragment of streptococcal protein G in complex with the Fc domain of human IgG. Structure 3, 265–278 (1995). | Article | PubMed | ChemPort |
19. Keefe, A.D., Wilson, D.S., Seelig, B. & Szostak, J.W. One-step purification of recombinant proteins using a nanomolar-affinity streptavidin-binding peptide, the SBP-Tag. Protein Expr. Purif. 23, 440–446 (2001). | PubMed | ISI | ChemPort |
20. Collis, S.J., DeWeese, T.L., Jeggo, P.A. & Parker, A.R. The life and death of DNA-PK. Oncogene 24, 949–961 (2005). | Article | PubMed | ISI | ChemPort |
21. Gottlieb, T.M. & Jackson, S.P. The DNA-dependent protein kinase: requirement for DNA ends and association with Ku antigen. Cell 72, 131–142 (1993). | Article | PubMed | ISI | ChemPort |
22. Li, B. & Comai, L. Functional interaction between Ku and the werner syndrome protein in DNA end processing. J. Biol. Chem. 275, 28349–28352 (2000). | Article | PubMed | ISI | ChemPort |
23. Aoki, Y., Zhao, G., Qiu, D., Shi, L. & Kao, P.N. CsA-sensitive purine-box transcriptional regulator in bronchial epithelial cells contains NF45, NF90, and Ku. Am. J. Physiol. 275, L1164–L1172 (1998). | PubMed | ChemPort |
24. Crevel, G., Ivetic, A., Ohno, K., Yamaguchi, M. & Cotterill, S. Nearest neighbour analysis of MCM protein complexes in Drosophila melanogaster. Nucleic Acids Res. 29, 4834–4842 (2001). | Article | PubMed | ChemPort |
25. Fouraux, M.A., Bouvet, P., Verkaart, S., van Venrooij, W.J. & Pruijn, G.J. Nucleolin associates with a subset of the human Ro ribonucleoprotein complexes. J. Mol. Biol. 320, 475–488 (2002). | Article | PubMed | ISI | ChemPort |
26. Park, J., Wood, M.A. & Cole, M.D. BAF53 forms distinct nuclear complexes and functions as a critical c-Myc-interacting nuclear cofactor for oncogenic transformation. Mol. Cell. Biol. 22, 1307–1316 (2002). | Article | PubMed | ISI | ChemPort |
27. Krogan, N.J. et al. Global landscape of protein complexes in the yeast Saccharomyces cerevisiae. Nature 440, 637–643 (2006). | Article | PubMed | ChemPort |
28. Brown, D. & Superti-Furga, G. Rediscovering the sweet spot in drug discovery. Drug Discov. Today 8, 1067–1077 (2003). | Article | PubMed |
29. Fishman, M.C. & Porter, J.A. Pharmaceuticals: a new grammar for drug discovery. Nature 437, 491–493 (2005). | Article | PubMed | ISI | ChemPort |
30. Cheeseman, I.M. & Desai, A. A combined approach for the localization and tandem affinity purification of protein complexes from metazoans. Sci. STKE 2005, pl1 (2005). | PubMed |
31. Bertwistle, D., Sugimoto, M. & Sherr, C.J. Physical and functional interactions of the Arf tumor suppressor protein with nucleophosmin/B23. Mol. Cell. Biol. 24, 985–996 (2004). | Article | PubMed | ISI | ChemPort |
32. Canton, D.A. et al. The pleckstrin homology domain-containing protein CKIP-1 is involved in regulation of cell morphology and the actin cytoskeleton and interaction with actin capping protein. Mol. Cell. Biol. 25, 3519–3534 (2005). | Article | PubMed | ChemPort |
33. Hacker, H. et al. Specificity in Toll-like receptor signalling through distinct effector functions of TRAF3 and TRAF6. Nature 439, 204–207 (2006). | Article | PubMed | ISI | ChemPort |
34. Naviaux, R.K., Costanzi, E., Haas, M. & Verma, I.M. The pCL vector system: rapid production of helper-free, high-titer, recombinant retroviruses. J. Virol. 70, 5701–5705 (1996). | PubMed | ISI | ChemPort |
35. Kinsella, T.M. & Nolan, G.P. Episomal vectors rapidly and stably produce high-titer recombinant retrovirus. Hum. Gene Ther. 7, 1405–1413 (1996). | PubMed | ISI | ChemPort |

	Top

Competing interests statement:  The authors declare competing financial interests.