Improved tandem affinity purification tag and methods for isolation of protein heterocomplexes from plants


For correspondence (fax +1 402 472 3139; e-mail


A synthetic gene encoding the tandem affinity purification (TAP) tag has been constructed, and the TAP tag assayed for its effects on expression levels and subcellular localization by fusion to green fluorescent protein (GFP) as well as for its effects on steroid-dependent translocation to the nucleus and transcription when fused to a hybrid glucocorticoid receptor. A nuclear localization signal (NLS) was detected in the calmodulin-binding protein (CBP) domain and removed by mutation to improve the usefulness of the TAP tag. Additionally, purification improvements were made, including inhibition of a co-purifying protease, and adding a protein cross-linking step to increase the recovery of interacting proteins. The improved synthetic TAP tag gene and methods were used to isolate proteins interacting with the hybrid glucocorticoid receptor and to identify them by mass spectrometry. The two proteins identified, HSP70 and HSP90, are known to interact with the glucocorticoid receptor in vivo in mammalian cells and in vitro in plants.


The tandem affinity purification (TAP) method provides a high level of purification of protein complexes containing the TAP-tagged protein (Puig et al., 2001; Rigaut et al., 1999). The TAP tag consists of tandem protein A domains that bind IgG–agarose, a tobacco etch virus (TEV) protease site for TEV protease-mediated cleavage of the fusion protein from the IgG–agarose, and a calmodulin-binding protein (CBP) domain, for reversible, calcium-dependent binding to calmodulin (CAM)–agarose. This TAP tag method has proven useful for mass spectrometry identification of proteins in heterocomplexes in yeast (Gavin et al., 2002), mammalian cells (Knuesel et al., 2003), Escherichia coli (Gully et al., 2003), and insect cells (Forler et al., 2003). However, the TAP tag method has only been used as an epitope for Western blots or pull down experiments in plants and has not been successfully used to fully purify protein complexes for mass spectrometry analysis of the proteins (Gao et al., 2003; Rivas et al., 2002). We have constructed a synthetic TAP tag gene for use in plants and assayed its effects on a GFP fusion protein for protein expression levels and subcellular localization, as well as for steroid-dependent gene expression, purification methods, and protein complex formation when fused to a hybrid glucocorticoid receptor transcription factor.

The hybrid transcription factor consists of the Gal4 DNA-binding domain, the VP16 transactivation domain, and the glucocorticoid hormone-binding domain (HBD) that binds steroid hormones and cellular proteins (GVG; Aoyama and Chua, 1997). The HBD has been shown to confer a cytoplasmic localization to fusion proteins in the absence of steroid hormone, and a nuclear localization in the presence of hormone in several plant systems (Brockmann et al., 2001; Hay et al., 2003; Kinkema et al., 2000; Spelt et al., 2000). The HBD is expected to be bound to a chaperone containing protein heterocomplex similar to those identified in mammalian, insect, and yeast cells (Pratt et al., 2001). These complexes have been shown to contain at least HSP90, HSP70, immunophilin, and p23, although in vitro plant complexes seem to lack p23, which accounts for their reduced stability (Pratt et al., 2001). In vitro studies with plant proteins have shown that the mouse glucocorticoid receptor binds to plant HSP70 and HSP90 proteins as efficiently as their mammalian counterparts, indicating a strong conservation of the function of the HSP70 and HSP90 chaperones between plants and animals (Dittmar et al., 1997; Stancato et al., 1996). In the current study, a TAP-tagged GVG fusion protein was tested for its ability to regulate a GUS reporter gene with a synthetic promoter containing Gal4 DNA-binding sites. Regulated GUS gene expression would indicate that the TAP-tagged GVG protein is initially in a protein heterocomplex in the cytoplasm, and upon hormone induction, is then translocated to the nucleus where it participates in an active transcription complex.

The GFP and GVG bioassays led to an improved TAP tag design that removes a nuclear localization signal (NLS) present in previous TAP tags, and now has an appropriate subcellular localization when fused to GFP and demonstrates a steroid-dependent induction of gene expression when fused to the GVG transcription factor. The TAP-tagged GVG protein was also used for optimizing the purification of a protein heterocomplex. The new TAP tag system appears very effective at isolating and purifying protein heterocomplexes from plants in amounts suitable for mass spectrometry identification of the associated proteins.


Synthetic TAP tag design

The peptide sequence of the TAP tag (Rigaut et al., 1999) was used to design a synthetic gene according to the principles used for the synthetic Bacillus thuringiensis insecticidal (Perlak et al., 1991) and synthetic GFP (Pang et al., 1996) genes, by third codon position changes to remove cryptic splice sites, polyadenylation sites, and AT- or GC-rich regions. Additionally, the duplicated 180-bp regions of the tandem protein A domains were made non-identical (76% homology) to reduce recombination and the possibility of repeat-induced gene silencing (Ma and Mitra, 2002). The nucleotide sequence of the synthetic TAP tag gene has 76% identity with the original TAP tag gene sequences while encoding the identical protein domains, with the exceptions of two amino acids added between the duplicated protein A domains and differences in the amino acids flanking the CBP region (Figure 1). These synthetic N-terminal or C-terminal TAP tag genes also include the castor bean catalase intron 1 (Tanaka et al., 1990), and were cloned into a binary vector containing lambda recombination sites (GATEWAY™, Invitrogen, Carlsbad, CA, USA), either N-terminal to the attR1 recombination site or C-terminal to the attR2 site (Figure 1). The castor bean catalase intron 1 has been shown to efficiently splice in monocot and dicot plants and to increase gene expression in rice plants (Tanaka et al., 1990). Its presence is useful in Agrobacterium-mediated transient assays to distinguish gene expression in bacteria from gene expression in plants (Vancanneyt et al., 1990), as well as for the intron-mediated increase in gene expression generally observed in monocots (Callis et al., 1987; Tanaka et al., 1990).

Figure 1.

Structure of the TAP tag expression cassettes and peptide sequence.

(a) Schematic of NTAPi and CTAPi expression cassettes. The expected mRNA is shown beneath the DNA regions. Target fusion genes will replace the region between the attR1 and attR2 recombination sites after LR reactions to create in-frame fusions with either the N-terminal (NTAPi) or C-terminal (CTAPi) TAP tag regions. The ECaMV 35S promoter (E35S) and CaMV 35S polyadenylation region (T35S) are used to express the TAP-tagged genes. ‘Tev-L’ designates the TEV untranslated leader.

(b) Peptide sequence of the NTAPi or CTAPi tags. The amino acids in the protein A (ProA), TEV protease site (TEV), CBP, and attB1 or attB2 protein regions are designated in capital letters, while small connecting peptide linker regions are designated in lowercase letters. The flexible linker regions separating the NTAPi or CTAPi tags from the target protein include the peptide encoded by the attB1 and attB2 regions (produced after the LR recombination reaction), and are designated ‘attB1’ or ‘attB2’, respectively. These regions consist primarily of the attB sites and the exons of the catalase intron 1, while the position the catalase intron 1 would occupy in the corresponding mRNA is designated ‘<intron>‘. The lysine or arginine amino acids changed to serine to remove the NLS are designated as ‘s’ in the CBP region.

Detection and mutation of an NLS in the TAP tag

The NTAP and CTAP tag fusions to GFP were tested for subcellular localization in Agrobacterium-mediated transient assays in Nicotiana benthamiana leaves (Llave et al., 2000). Inspection by confocal microscopy found that the NTAP–GFP protein was predominantly localized to the nucleus (Figure 2b), although this was not true of the GFP–CTAP protein (data not shown). These results suggest that a context-dependent NLS is active in the NTAP tag but masked in the CTAP tag. As described below, this NLS is present in the protein sequence of the original TAP tag and is not a result of our synthetic gene design. The presence of an NLS in a general-purpose affinity tag is undesirable and further experiments were carried out to remove it.

Figure 2.

Subcellular localization and binding properties of mutated CBPs.

(a–d) Confocal photomicrographs of TAP-tagged GFP proteins. Agrobacterium-mediated transient assays of N. benthamiana leaves were photographed 48 h after infection with: (a) GFP; (b) NTAP–GFP; (c) NTAPi–GFP; or (d) GFP–CTAPi.

(e) CBP mutations, localization, and binding. The peptide sequence of the original CBP and the position of the serine substitutions in the NTAP CBP mutants are shown. The nuclear to cytoplasmic ratio of GFP intensity (Nuc/Cyto) was measured from photomicrographs. Equivalent binding of the CBP mutants to CAM–agarose and IgG–agarose binding (CAM/IgG) is designated ‘+’, as measured by Western blot analysis. The ability of selected CBP mutants to be eluted from CAM–agarose with EGTA is shown as a ratio of CBP remaining bound in the presence of EGTA to the amount bound in the presence of Ca2+ (designated EGTA/Ca2+).

The NLS was determined to be in the CBP region of the NTAP tag by comparison of GFP fusions containing either the protein A and TEV protease site domains, or the CBP domain, catalase exons and attB1 regions, or only the attB1 region (data not shown). Mutations in the CBP region were then constructed to remove the NLS while retaining its CAM-binding properties. Ten different constructs containing two to four serine substitutions in seven of the eight basic residues in the CBP region were assayed for subcellular localization when fused to GFP, using transient expression in N. benthamiana leaves (Figure 2e). The NLS appears to be bipartite as mutations in either of the two basic regions greatly reduce the nuclear localization of GFP. NTAPi–GFP (Figure 2c) and NTAPi–GFP had nuclear to cytoplasmic ratios identical to GFP alone, while the other mutants had a slightly higher ratio, but considerably less than the original NTAP–GFP protein (Figure 2e).

The CAM-binding ability of the mutated CBP regions in the NTAP–GFP proteins was assayed by binding them separately to either IgG–agarose or CAM–agarose, followed by Western blot analysis to detect the protein A domains of the TAP tag. The Western blot showed equivalent binding to the IgG–agarose or CAM–agarose for all the mutated CBP constructs (Figure 2e). Five of the altered CBPs that showed the least amount of nuclear localization were assayed for EGTA-dependent release after binding to CAM–agarose and showed complete release (Figure 2e). Therefore, the various mutated CBP domains appear to have equivalent CAM-binding and EGTA release properties, despite the differences in the number and location of the serine substitutions. NTAPi, which contains three substitutions that are predicted by the analysis of the protein structure of a CAM:CBP complex to least alter CAM binding (Ikura et al., 1992), was chosen as the new TAP tag for further use (Figure 2c). CTAPi, which contains the modified CBP domain in the C-terminal TAP tag, also shows a normal subcellular distribution of protein when fused to GFP (Figure 2d).

Steroid regulation of a TAP-tagged GVG transcription factor

The GVG hybrid transcription factor consists of the Gal4 DNA-binding domain, the VP16 transactivation domain and the glucocorticoid regulatory region (Aoyama and Chua, 1997). N- or C-terminal TAP tag fusions were made to test the effect of the TAP tag on GVG's biological functions. The control GVG protein, NTAPi–GVG, and GVG–CTAPi fusion proteins all demonstrated similar hormone-dependent activation of a synthetic Gal4 promoter/GUS reporter gene (Table 1). The original NTAP–GVG fusion protein, containing an NLS, also showed regulated expression (data not shown), indicating that the NLS is overcome by the cytoplasmic localization signal(s) present in the protein heterocomplex. However, the strengths of cytoplasmic and nuclear localization signals in different protein heterocomplexes are unpredictable, and the lack of an NLS in the NTAPi and CTAPi tags should reduce potential problems. The regulated GUS gene expression indicates the formation of a cytoplasmic heterocomplex, as well as the steroid hormone-induced translocation of the TAP-tagged GVG protein into the nucleus and participation in a transcription complex, occurs normally in the NTAPi–GVG and GVG–CTAPi fusion proteins. These results indicate the NTAPi and CTAPi tags do not interfere, by either steric hindrance or subcellular localization signals, with the normal behavior of the GVG protein.

Table 1.  GUS assays of the ability of GVG proteins to induce the Gal4 synthetic promoter
GVG constructGUS unitsa (nmol MU min−1 mg−1 protein)
(−) DEX(+) DEX
  • a


  • ±

    ±SD of three independent samples.

GVG13 ± 5537 ± 175
NTAPi–GVG36 ± 7943 ± 43
GVG–CTAPi27 ± 7487 ± 118

Protease inhibitors

Prior experiments in Arabidopsis cells had indicated that an Arabidopsis protease binds to IgG beads and reduces the recovery of protease sensitive target proteins from the IgG–agarose during the TEV protease cleavage step. In particular, an Arabidopsis MAPKK (AtMKK2; At4g29810) fusion protein was sensitive to this endogenous protease that co-purified on the IgG–agarose and was active during the TEV protease cleavage step. We examined a number of protease inhibitors for their compatibility with the TEV cleavage of the fusion protein from the IgG–agarose, using TEV protease-mediated release of a GUS–CTAP protein bound to IgG–agarose (data not shown). The best of these protease inhibitors were then tested for improving the recovery of AtMKK2 from the IgG–agarose, and it was found that the E-64 cysteine protease inhibitor increased recovery (Figure 3). E-64 is now routinely included in the TEV protease cleavage step.

Figure 3.

TEV protease compatible protease inhibitors increase fusion protein recovery.

N-terminal TAP-tagged AtMKK2 (At4g29810) was bound to IgG–agarose and then the fusion protein was eluted by incubation with TEV protease. The size and recovery of the proteins was examined by Western blot analysis using biotinylated camodulin as the primary probe. The lanes are either the bound protein prior to cleavage (IgG), or TEV cleaved in the presence of E-64 (+E-64), or PMSF (+PMSF), or the TEV buffer only control (control).

TAP tag purification and protein cross-linking

The steroid-regulated GUS gene expression suggests the NTAPi–GVG protein is present in a cytoplasmic heterocomplex. A small-scale purification from Agrobacterium-infected N. benthamiana leaves was performed to determine the recovery of the TAP-tagged protein from the IgG–agarose and CAM–agarose, using Western blot analysis with biotinylated camodulin to detect the CBP domain (Figure 4a). The GVG portion of NTAPi–GVG protein has some non-specific affinity to the IgG–agarose, making the TEV elution step only about 25% efficient. This non-specific binding is protein dependent, as other proteins do not show this (data not shown). The IgG binding as well as the CAM–agarose binding and elution steps appear to be very efficient.

Figure 4.

Tandem affinity purification of NTAPi–GVG.

(a) CAM Western blot of the stages of TAP of NTAPi–GVG. Protein extracts from transiently infected N. benthamiana leaves were purified according to the TAP protocol and aliquots analyzed by Western blot using biotinylated CAM and alkaline phosphatase–streptavidin. The following aliquots are shown, with the percentage of the initial amount of total NTAP–GVG protein loaded in each lane in parentheses: IgG–agarose bound (10%); TEV eluate from IgG–agarose (18%); CAM–agarose bound (30%); and CAM–agarose EGTA eluate (42%).

(b) Visualization of protein bands after SDS–PAGE of the purified NTAPi–GVG complex. Extracts from transiently infected N. benthamiana leaves were purified by the TAP protocol, separated by SDS–PAGE, and visualized with a fluorescent dye. Lane 1, purification without prior protein cross-linking; lane 2, purification after an initial protein cross-linking with 1% formaldehyde. The protein cross-linking was reversed just prior to loading the SDS–PAGE. The bands designated ‘*’ are from reagent contaminants or keratins and are not plant proteins.

A larger scale TAP purification was performed, the proteins separated by SDS–PAGE, and visualized by fluorescence staining (Figure 4b). The bands observed were isolated and analyzed by tandem mass spectrometry after in situ digestion of the proteins with trypsin. The major band contained the NTAPi–GVG protein and the minor band contained a HSP70 protein (Figure 5a). The most closely related plant species in the GenBank database was Saussurea medusa as the Nicotiana sequences for this protein family are not available yet. The HSP70 protein sequences are sufficiently conserved for multiple identical peptides to be identified (Figure 5a). The other protein bands in the gel, marked with the asterisks (Figure 4b), are non-plant proteins. One of these is the bacterial ompF, which migrates at about 45 kDa and is presumably from the rTEV reagent that is produced from E. coli. Only the CAM–agarose purification step follows this TEV cleavage step, and the ompF protein may have some affinity for this resin. The other protein bands above the 108-kDa marker contain human keratins, and appear to be from handling of the plants in the greenhouse as these amounts are reproducibly reduced when the tissues are first cross-linked. Additionally, we have not detected significant keratin in the IgG beads, CAM beads, rTEV, prepared gels, or purification buffers when analyzed by mass spectrometry of the trypsin-digested samples.

Figure 5.

Identification of HSP70 and HSP90 proteins from mass spectrometry analysis.

The cross-species identification of the tobacco peptide sequences identified from tandem MS are shown (bold and underlined) for the top matches. The GenBank GI number is also shown for the matching gene.

The utility of formaldehyde protein cross-linking was investigated to determine if protein heterocomplex recovery was more efficient when cross-linked in intact tissues prior to making the protein extract. Initial titrations indicated a 1% formaldehyde concentration provided efficient cross-linking of proteins, as measured by a mobility shift on Western blots (Figure 6). This cross-linking-dependent mobility shift is a rapid method to verify that the tagged protein is present in a protein complex in vivo, although it does not distinguish homocomplexes from heterocomplexes.

Figure 6.

Reversible formaldehyde cross-linking of NTAPi–GVG in tobacco leaves.

Isolated tobacco leaves expressing NTAPi–GVG were cross-linked with the indicated concentrations of formaldehyde. The protein extracts were incubated with CAM–agarose and analyzed by Western blot with PAP reagent. ‘R’ designates samples where cross-linking was reversed by heating before loading the gel.

The TAP purification was performed on Agrobacterium-infected N. benthamiana leaves transiently expressing the NTAPi–GVG protein. The leaves were infiltrated with a 1% formaldehyde solution to cross-link the proteins prior to purification. The formaldehyde cross-linking of the purified proteins was reversed by heating, the proteins separated by SDS–PAGE and visualized by fluorescence staining (Figure 4b). The ratio of HSP70 to NTAPi–GVG was increased considerably when cross-linked with formaldehyde, and an additional band of about 80 kDa was also observed. This band was identified as HSP90 by tandem mass spectrometry analysis. The N. benthamiana HSP90 protein sequence was not available in GenBank and the most closely related protein was from Arabidopsis thaliana (Figure 5b). These results indicate that protein cross-linking increases the recovery of less stable protein complexes, as the HSP90 protein is expected to interact with the HBD portion of the GVG protein, based on results from other organisms and in vitro plant extracts (Pratt et al., 2001). The contaminating band marked with an asterisk is the E. coli ompF protein at about 45 kDa.


An ideal affinity purification tag should provide a high degree of purification and not affect the expression, localization, stability, or function of the target protein. The modified synthetic NTAPi and CTAPi tags appear to fulfill these goals as indicated by expression levels of the fusion proteins that are similar to the control proteins, a normal intracellular distribution of the NTAPi–GFP and GFP–CTAPi proteins, and the biological functionality of the NTAPi–GVG and GVG–CTAPi proteins when fused to the GVG hybrid transcription factor. Additional results supporting this conclusion are the reports that the C-terminally TAP-tagged Cf9 disease-resistance protein was functional in inducing a hypersensitive response in infected N. benthamiana leaves (Rivas et al., 2002), and the C-terminally TAP-tagged CTR1 protein was functional in a membrane-associated ethylene receptor complex (Gao et al., 2003). Presumably the NLS present in the CBP region is masked in these C-terminal TAP tags, as we observed for GFP–CTAP, or the strength of their membrane localization overcomes any NLS-mediated transport to the nucleus. Taken together, these results indicate that the TAPi tags should not systematically interfere with the biological function of fusion proteins in plants.

The presence of an NLS in the CBP domain was not predicted by examining known NLSs (Nair et al., 2003;, although the psort program predicts that most basic regions are potential NLSs (Nakai and Horton, 1999; Our mutation analysis indicates the CBP NLS is bipartite, as mutations in either of the two basic regions greatly reduce the nuclear localization of GFP fusion proteins. Surprisingly, all the CBP mutations retained sufficient CAM-binding ability to bind all the NTAP–GFP protein present in the extracts tested. The NTAPi CBP mutant was chosen because of the low degree of nuclear localization, and the structure of CBP bound to CAM indicated that these mutations should not alter binding (Ikura et al., 1992).

The TAPi tag provides a high degree of purification, as N. benthamiana or Arabidopsis proteins do not co-purify in sufficient quantities to produce a visible band on the fluorescently stained gels when mock, GFP–CTAP, or GUS–CTAP purifications are performed. Concerns about contaminating cellular proteins binding to the CAM–agarose (Knuesel et al., 2003) seem unwarranted in plants, as we have not observed this, presumably because these proteins do not bind to, or are not released from, the IgG–agarose in the initial purification step. The HSP70 and HSP90 proteins recovered have not been observed in the TAP purification of five other proteins in our laboratory, indicating the interaction is specific for the GVG protein and not the TAP tag, IgG, or CAM beads.

Protein cross-linking recently has been shown to identify interacting yeast proteins that were not detected without prior cross-linking in immunoprecipitation experiments (Hall and Struhl, 2002). The HSP90 subunit appears to fall into this category, as we only observe it when the samples are first cross-linked. The HSP90 subunit is known to rapidly dissociate from the glucocorticoid receptor in plant extracts as its stable association was not detected in vitro until mammalian p23 was added to the extracts (Hutchison et al., 1995). Thus, HSP90s dissociation when the NTAPi–GVG complex is isolated is consistent with this observation. HSP70 is known to be necessary for forming the p23–HSP90–glucocorticoid receptor complex in mammals and its persistence in the complex thereafter is protein dependent (Pratt et al., 2001). Our results indicate that HSP70 is present in the NTAPi–GVG heterocomplex isolated from plants, and its relative abundance increases when protein cross-linking is used. The other protein that might be expected in the heterocomplex would be an immunophilin, as these are known to bind to plant HSP90 heterocomplexes in vitro (Harrell et al., 2002). If immunophilins are present in the NTAPi–GVG heterocomplex in vivo, it might be possible to detect them with protein cross-linkers that are capable of longer molecular cross-linking distances.

A key difficulty with most protein interaction systems is determining whether the interaction occurs in the intact living cell. Interactions in yeast two-hybrid screens, or in any procedure that utilizes an initial crude extract, provide opportunities for proteins to interact that normally might be prevented from interacting in the intact host cell. The combination of cross-linking proteins in intact cells and a TAP tag for efficient purification provides a convenient solution to this problem. The most rigorous approach would be to demonstrate that the interacting proteins are cross-linked and not simply co-purifying. This could be accomplished by isolating the cross-linked complex at its higher molecular weight position on SDS–PAGE, followed by detection of the interacting proteins by tandem mass spectrometry identification. Proteins whose co-purification requires cross-linking, as we observed for HSP90, or whose relative abundance increases on cross-linking, as we observed for HSP70, are highly suggestive of interactions occurring in vivo.

The issue of competition between the introduced TAP tag fusion protein and the corresponding endogenous protein to participate in a heterocomplex is not relevant here as GVG is a foreign protein and there appears to be sufficient HSP70 and HSP90 protein present in the cell for heterocomplex formation. However, in most cases, heterocomplex formation should be increased if competition from endogenous proteins can be reduced through RNAi (Forler et al., 2003) or knockouts. Overexpression of the fusion protein is another approach for competing with the endogenous protein, as has been performed in yeast (Ho et al., 2002). Determining an optimum level of overexpression using bioassays or cross-linking assays may increase the utility of this approach. In the present case, the steroid-regulated expression of the GVG-controlled transcription of the GUS reporter gene provided a convenient bioassay of heterocomplex formation.

The NTAPi and CTAPi vectors are available upon request (maps and sequences are available in Supplementary Material) and should be generally useful for protein tagging in planta and subsequent protein purification. The improved synthetic N-terminal and C-terminal TAPi tags are in GATEWAY™ binary vectors that are most suitable for dicots, as the nopaline synthase (NOS) promoter expressing the Bar selectable marker gene is weakly expressed in monocots. The N- or C-terminal TAPi tag cassettes can be transferred to other vectors as the ECaMV 35S promoter and TAPi regions should express well in monocots because their synthetic gene design includes features for monocot gene expression (Pang et al., 1996), the catalase intron 1 increases expression in monocots (Tanaka et al., 1990), and the NTAP tag has been expressed well in transgenic rice plants when fused to several rice protein kinases and expressed from the maize ubiquitin promoter (P. Ronald and P. Canlas, personal communication).

Experimental procedures

Plasmids and genes

The synthetic NTAP and CTAP genes were constructed by overlapping PCR of multiple oligonucleotides, the castor bean catalase intron 1 fragment, and for the NTAP construct, the TEV leader (James et al., 2000; Niepel and Gallie, 1999). The CTAP construct does not contain the TEV leader. The PCR products were cloned into pDEST14 (Invitrogen), sequenced, and then the entire TAP and GATEWAY™ portions cloned as an XhoI to NheI fragments into XhoI- and XbaI-digested pPTN289. pPTN289 (Tom Clemente, unpublished data) is a pPZP200-based binary vector (Hajdukiewicz et al., 1994) that has an expression cassette consisting of the ECaMV 35S promoter, TEV 5′ untranslated leader sequence (James et al., 2000; Niepel and Gallie, 1999) and CaMV 35S polyadenylation region. The plasmid also has the NOS promoter expressing the Bar gene-selectable marker (D'Halluin et al., 1992) for plant transformation.

The sequences of the final NTAP and CTAP regions are available in GenBank (AY436344 and AY436346, respectively). The N-terminal TAP tag is designed to be in reading frame 1 (GATEWAY™, Invitrogen) of the downstream attR1 site (attB1 after the LR recombination reaction). This structure separates the AUG start codon of the target gene from the 3′ end of the CBP domain by the reading frame encoded by the exons flanking the castor bean catalase intron 1, junction regions, and the attB1 site, the combination of which when translated encodes 19 amino acids (Figure 1b, attB1 region). The CTAP tag is designed to be in reading frame 1 (GATEWAY™, Invitrogen) of the upstream attR2 site (attB2 after the LR recombination reaction). The translation of the attB2 junction region and catalase exons produces a 20-amino-acid hydrophilic linker (Figure 1b, attB2). These 19 or 20-amino-acid linkers serve as hydrophilic spacers between the TAP tags and the target proteins, increasing the steric availability of the CBP region for binding to CAM–agarose beads.

GFP6 (Cormack et al., 1996) and GVG (Aoyama and Chua, 1997) were PCR amplified to have attB1 and attB2 sites on their 5′ and 3′ ends, respectively, and cloned into pDONR207 (Invitrogen) using the BP recombination reaction (GATEWAY™, Invitrogen), and sequenced. Both the GFP6 and GVG constructs were made with or without stop codons, to allow for N-terminal or C-terminal fusions. The GFP6 signal peptide and endoplasmic reticulum retention signals were removed by PCR using primers that are designed to re-create the size of the original GFP protein, so that the resulting mGFP6 should be localized in the cytoplasm. The CBP mutations were incorporated into the NTAP or CTAP DNA by PCR, joined to the catalase intron 1 fragment by restriction and ligation of flanking restriction sites, and amplified by PCR. The NTAPi PCR product was digested with the XhoI and XbaI restriction enzymes and ligated into the modified pPTN289 vector and sequenced (Accession AY436345). The CTAPi PCR product was digested with SpeI and HindIII and cloned into a modified pDEST14 vector digested with the same enzymes, and sequenced (Accession AY436343). The entire CTAPi and GATEWAY™ region was isolated as an XhoI to NheI fragment and cloned into XhoI- and XbaI-digested pPTN289. LR recombination reactions (Invitrogen) produced in-frame fusions between the NTAPi tag and the coding regions of mGFP6 or the GVG hybrid transcription factor present in the pDONR207 vector, to produce NTAPi–GFP and NTAPi–GVG, respectively. An LR reaction between the C-terminal TAPi tag GATEWAY™ cassette and the pDONR207 plasmids containing the mGFP6 or GVG genes lacking stop codons, produced GFP–CTAPi and GVG–CTAPi, respectively. Plasmids were transformed into the Agrobacterium strain GV3101 by electroporation.

Agrobacterium transient assays

Agrobacterium strains were grown at 28°C and re-suspended in infiltration medium (10 mm 2-[N-morpholino]ethanesulfonic acid (MES), pH 5.6, 10 mm MgCl2, 100 µm acetosyringone) at room temperature for 3 h (Llave et al., 2000). The induced Agrobacterium cultures were injected into the leaves of N. benthamiana plants and maintained at 28°C and 60–70% relative humidity. Steroid induced plants were sprayed with 20 µm DEX (Sigma, St. Louis, MO, USA) solution containing 0.03% Silwet L-77 (Lehle Seeds, Round Roct, TX, USA) 48 h after infection. Leaves were harvested 12 h after spraying with DEX.

In vivo protein cross-linking

Agrobacterium-infiltrated leaves were harvested, cut into small pieces, and vacuum infiltrated in a 1% formaldehyde in ice-cold PBS buffer for 30 min (Hall and Struhl, 2002). The reaction was quenched by adding cold 300 mm glycine for 30 min. Cross-linked tissue was then washed with PBS and stored at −80°C.

Standard TAP purification

The TAP purification protocol (Rigaut et al., 1999) was used with some modifications. Protein extracts were prepared from Agrobacterium-infiltrated leaves (with or without a prior protein cross-linking treatment) in protein extraction buffer (20 mm Tris, pH 8.0, 150 mm NaCl, 0.1% IGEPAL (Sigma), 2.5 mm EDTA, 2 mm benzamidine, 10 mmβ-mercaptoethanol, 20 mm NaF, 1 mm phenylmethanesulfonylfluoride (PMSF), 1% Protease cocktail (Sigma), 10 µm leupeptin (Sigma), and 10 µm 3,4-dichloroisocoumarin (Sigma). The centrifuged supernatant was mixed with 50 µl of IgG Sepharose beads (Amersham Biosciences, Piscataway, NJ, USA) and incubated at 4°C for 2 h with gentle rocking. The mixture was loaded onto a disposable polyprep chromatography column (Bio-Rad Laboratories, Hercules, CA, USA) and washed with protein extraction buffer lacking protease inhibitors. The TAP-tagged proteins were released by digestion with 100 U of rTEV protease (Invitrogen) in TEV cleavage buffer containing 1 µm E-64 protease inhibitor for 1 h at 16°C. The eluate was then bound to the CAM agarose beads (Stratagene, La Jolla, CA, USA) in CAM-binding buffer (CBB; Rigaut et al., 1999) and eluted with buffer containing 2 mm EGTA. The eluate was TCA precipitated and loaded onto a 4–15% gradient polyacrylamide gel (Bio-Rad Laboratories) for SDS–PAGE. Formaldehyde cross-linking was reversed by boiling in loading buffer for 20 min prior to loading the gel. Proteins were visualized by fluorescence staining (Sypro Ruby; Bio-Rad Laboratories).

Mass spectrometry analysis of TAP-purified proteins

Protein bands separated with SDS–PAGE were excised and digested in situ using a slightly modified version of a published method (Shevchenko et al., 1996). Briefly, the samples were washed with 100 mm ammonium bicarbonate, reduced with 10 mm DTT, alkylated with 55 mm iodoacetamide, washed twice with 100 mm ammonium bicarbonate, and digested in situ with 10 ng µl−1 trypsin (Promega, Madison, WI, USA). Peptides were extracted with two 60 µl aliquots of 1 : 1 acetonitrile:water containing 1% formic acid. The extracts were reduced in volume to approximately 25 µl using a SpeedVac. Ten microliters of the extract solution was injected onto a trapping column (300 µm × 1 mm) in line with a 75 µm × 15 cm C18 reversed phase LC column (LC-Packings, Sunnyvale, CA, USA). Peptides were eluted from the column using a water + 0.1% formic acid (A)/95% acetonitrile:5% water + 0.1% formic acid (B) gradient with a flow rate of 270 nl min−1. The gradient was developed with the following time profile: 0 min, 5% B; 5 min, 5% B; 35 min, 35% B; 40 min, 45% B; 42 min, 60% B; 45 min, 90% B; 48 min, 90% B; and 50 min, 5% B.

The eluting peptides were analyzed using a Q-TOF Ultima tandem mass spectrometer (Micromass/Waters, Milford, MA, USA) with electrospray ionization. Analyses were performed using data-dependent acquisition (DDA) with the following parameters: 1 sec. survey scan (380–1900 Da) followed by up to three 2.4 sec MS/MS acquisitions (60–1900 Da). The instrument was operated at a mass resolution of 8000. The instrument was calibrated using the fragment ion masses of doubly protonated Glutamate (Glu)-fibrinopeptide. The MS/MS data were processed using masslynx software (Micromass) to produce peak lists for database searching. mascot (Matrix Science Ltd, London, UK) was used as the search engine. Data were searched against the National Centre for Biotechnology Information (NCBI) non-redundant database. The following search parameters were used: mass accuracy 0.1 Da, enzyme specificity trypsin, fixed modification carboxyamidomethylcysteine (CAM), variable modification oxidized methionine. Protein identifications were based on random probability scores with a minimum value of 25.

CBP and protein A Western blots

The CBP domain was detected by Western blots using biotinylated CAM (Safadi et al., 2000). The electroblotted membrane was blocked overnight in Tris-buffered saline (TBS) containing 1% BSA at 4°C. The membrane was washed in Buffer A (20 mm Tris–HCl, pH 7.5, 150 mm NaCl, 1 mm CaCl2, 50 mm MgCl2, 0.01% Tween 20) followed by Buffer B (same as Buffer A but lacking Tween 20) and probed with 450 ng ml−1 of biotinylated CAM (Calbiochem, La Jolla, CA, USA) in Buffer B containing 1% BSA for 2–3 h. The membrane was washed in Buffer A and incubated in a 1 : 3000 dilution of streptavidin–alkaline phosphatase (Amersham Biosciences) in Buffer B containing 1% BSA for 1 h at room temperature. After washing in Buffer B, the membrane was developed using 1-Step™ NBT/BCIP (Pierce, Rockford, IL, USA). The protein A domains of the TAP tag were detected with PAP conjugate (Sigma) as described by Rivas et al. (2002).

CAM bead assay

Protein extracts, prepared as above, were diluted threefold into CBB, bound to CAM–agarose beads (Stratagene) for 1 h, and washed in CBB. The beads were then incubated in CBB supplemented with 30 mg ml−1 gelatin and a 1 : 5000 dilution of PAP conjugate (Sigma) for 1 h at room temperature. The beads were then washed thrice with CBB and once with the substrate prior to incubation with the substrate (SUPERSIGNAL® WEST PICO Chemiluminescent Substrate; Pierce). The signal was detected in a luminometer (LUMISTAR; BMG Labtechnologies, Durham, NC, USA).


The authors wish to thank Hideaki Moriyama (University of Nebraska, Lincoln) for analysis of the camodulin–CBP protein structure; Tiera Gilder and You Zhou for technical assistance; Karin van Dijk, Alan Christensen, and Tom Clemente for critical reading of the manuscript. This work was supported by funding from the Nebraska Research Initiative and the National Science Foundation (DBI-0217312) to M.E.F.

Supplementary Material

The following material is available from

Figure S1. The plasmid maps for NTAPi are shown.

Abbreviations: Enhanced cauliflower mosaic virus 35S promoter (EcaMV 35S Pro); tobacco etch virus leader (TEV-L); chloramphenicol resistance (CAM R); polyadenylation region of cauliflower mosaic virus 35S transcript (T35S); nopaline synthase promoter (NOS Pro); bar gene selectable marker (BAR); polyadenylation region of the nopaline synthase gene (T-NOS); right T-DNA border (R Br); left T-DNA border (L Br); spectinomycin resistance gene (aadA Spec R).

Figure S2. The fasta files for the DNA sequences of these plasmids are also provided (NTAPi).

Figure S3. The plasmid maps for CTAPi are shown.

Abbreviations: Enhanced cauliflower mosaic virus 35S promoter (EcaMV 35S Pro); tobacco etch virus leader (TEV-L); chloramphenicol resistance (CAM R); polyadenylation region of cauliflower mosaic virus 35S transcript (T35S); nopaline synthase promoter (NOS Pro); bar gene selectable marker (BAR); polyadenylation region of the nopaline synthase gene (T-NOS); right T-DNA border (R Br); left T-DNA border (L Br); spectinomycin resistance gene (aadA Spec R).

Figure S4. The fasta files for the DNA sequences of these plasmids are also provided (CTAPi).

GenBank Accession numbers: AY436344, AY436346, AY436345, and AY436343