Cobalt(III) as a Stable and Inert Mediator Ion between NTA and His6-Tagged Proteins**

Don't let go! The Co(3+)-mediated interaction between nitrilotriacetic acid (NTA) and the His6-tag is so stable and inert towards ligand exchange that it has a half-life of 7 days in the presence of imidazole and survives even under strongly chelating as well as reducing conditions, unlike the commonly used Ni(2+) or Co(2+) complexes. Possible applications include the separation of labeled proteins and the stable immobilization of proteins on surfaces.

defined as 100 % and all the His-GFP in solution and on the beads was calculated accordingly (Figure 2, Supporting Information, Figure S1). The solutions were added back to the beads after the fluorescence measurements and the fluorescence was measured again after 1, 3, 7, and 16 days to analyze the release of GFP-His6 from Co 3+ -NTA beads over time. The initial rate of Hi6-GFP release from the beads was fitted to 1 st order kinetics using the data from the samples that were treated with 5, 10 and 20 mM H 2 O 2 (Supporting Information, Figure S4). All experiments were done in duplicate.
To show the generality of the presented concept, aliquots of Co 2+ -NTA agarose beads, prepared as described above, were incubated with an equal volume of various His6-tagged protein solutions (10 µM).
100 µl aliquots of these beads were incubated with 20 mM H 2 O 2 for 1 hour before imidazole was added to a final concentration of 250 mM. The solution above the beads was analyzed for eluted protein by SDS-PAGE gel (Invitrogen, NuPage Bis-Tris 4-12 %) (Supporting Information, Figure S3).
Chemical stability measurements of the Co 3+ -NTA. To obtain Co 3+ -NTA agarose beads with immobilized His6-GFP, the above-described immobilized His6-GFP on Co 2+ -NTA agarose beads were reacted with 20 mM H 2 O 2 for 1 hour. Then the excess H 2 O 2 was removed by repeated washing of the beads with buffer A. Twice the bead volume of buffer A was then added to the beads and 150 µl aliquots of this suspension were used in each experiment. For comparison an equal amount of Co 2+ -NTA with GFP-His6 in an equal volume of buffer A was used as well. 50 µl of the appropriate chelators (final concentrations: 250 mM imidazole, 25 mM NTA and 25 mM EDTA) or reducing agents (final concentration: 1 mM) in combination with 250 mM imidazole was then added to the beads and left to react for 1 hour (Figure 3). 100 µl of the solution above the beads was then analyzed for the His6-GFP fluorescence with the same settings as above. The solutions were added back to the beads after the measurements and analyzed 24 hours later again (Supporting Information, Figure S5). All experiments were done in duplicate. Further, equal amounts of Co 2+ -NTA and Co 3+ -NTA beads were treated with 100 mM ascorbate, TCEP and DTT in the presence of 250 mM imidazole or put into pH 3.5 acetate buffer.
The amount of eluted protein after 1 hour and 1 day was quantified by His6-GFP fluorescence where possible and SDS-PAGE gels (Invitrogen, NuPage Bis-Tris 4-12 %) (Supporting Information, Figure S6). and Co 3+ centers in DMEM were added (250 µl DMEM with 50 µl bead volume). The medium with the beads was removed from the culture plate after 1, 6 and 24 hours, 250 mM imidazole was added and the His6-GFP fluorescence in the supernatant was measured. The fluorescence intensity measured for the Ni 2+ beads was set to 100 % (Supporting Information, Figure S7).

UV-Vis and ESI analysis of Co 2+ and Co 3+ complexes with His6-peptide and NTA
Different complexes were obtained by mixing the ligands in the corresponding stoichiometry with 1.5 mM CoCl 2 in buffer A to obtain [Co II (NTA)], [Co II (NTA)(Imidazole) 2 ] and [Co II (NTA)(His6-peptide)].
His6-peptide was purchased from Innovagen, with an acetyl functionalization on the N-terminus. To obtain the Co 3+ versions of these complexes 20 mM H 2 O 2 was added to these solutions and reacted for 1 hour. The UV-Vis spectra of these complexes were recorded and the maxima were determined (Supporting Information, Figure S2)

Separation of different His6-GFPspecies labeled via NTA by Ni-NTA columns
Co 3+ mediated labeling of His6-GFP with NTA. Aliquots of 250 µl with 20 µM His6-GFP and various amounts of [Co II NTA] (0-800 µM) were incubated with 10 mM H 2 O 2 for 1 hour before they were applied to a 5 ml Ni-NTA column connected to a FPLC system (AKTA Purifier). The column was run with 2.5 ml/min flow and 1.5 ml fractions were collected. The column was first washed with 10 ml buffer A and then a linear imidazole gradient up to 100 mM imidazole over 50 ml was used to elute different His6-GFP species. Then, 200 µl from each fraction was analyzed for Hi6-GFP fluorescence ( Figure 4). As controls His6-GFP alone and GFP-His6 in the presence of 800 µM [Co II NTA] were also analyzed in the same way as the H 2 O 2 treated samples (Supporting Information, Figure S9). The distribution of the peaks was analyzed by integrating the area of each peak and setting the total area under the peaks to 100 % for each run (Supporting Information, Figure S8). To ensure that His6-GFP is not damaged by Fenton reactions in the presence of [Co II NTA] (800 µM) before and after incubation with 10 mM H 2 O 2 for 1 hour, these samples are analyzed by MALDI-TOF (Supporting Information, Figure S12). As a positive control His6-GFP (20 µM) was incubated with 200 µM Co 2+ , 4.6 mM ascorbate and 10 mM H 2 O 2 , a conditions where Fenton reactions are known to take place. [2] Synthesis of coumarin-NTA. 7-hydroxycoumarin-3-carboxylic acid N-succinimidyl ester (5 mg, 16.5 nmol) was dissolved in 100 µl DMF and added dropwise to a solution of N α ,N α -bis(carboxymethyl)-Llysine hydrate (8.6 mg, 33 nmol) in 0.3 M HEPES pH 7.4. The reaction was stirred for 30 min. at RT and the purified by C18 reverse phase column (30 % Acetonitrile 0.1 % TFA, flow rate). The purified compound was lyophilized and the product was characterized by MALDI-TOF and 1 H-NMR. Scheme S1. Synthesis of coumarin-NTA.

References
Systematic errors can rise from the reverse reaction where the His6-GFP binds back on the beads, the stability of the protein over this long incubation time and the approximation that the initial rate is equal to the rate over the first 7 days, which were used in the fitting. little His6-GFP is eluted. Ascorbate is the only reducing agent that succeeds in reducing some of the Co 3+ centers on the beads so that more His6-GFP is eluted. Figure S6. Chemical reactivity of immobilized His6-GFP on Co 3+ -NTA agarose beads. The amount of His6-GFP in the supernatant, which eluted from the NTA agarose beads, was quantified by a) measuring the GFP fluorescence after 1 hour, 1 day and 3 days, and SDS-PAGE gel after b) 1 hour and c) 1 day.
NTA agarose beads with immobilized His6-GFP at Co 2+ (1, 3, 5, 7) and Co 3+ (2, 4, 6, 8) centers were incubated 1-2) with 250 mM imidazole, 3-4) in 100 mM pH 3.5 acetate buffer, 5-6) with 100 mM ascorbate in the presence of 250 mM imidazole and 7-8) with 100 mM TCEP in the presence of 250 mM imidazole. As can be observed both from the GFP fluorescence and in the SDS-PAGE gel 100 mM ascorbate can almost entirely reduce the Co 3+ centers, therefore the His6-GFP can be eluted with imidazole. While 100 mM TCEP partially can reduce the Co 3+ centers, acidification of the Co 3+ beads does not lead to protein elution. Figure S7. Stability of immobilized His6-GFP on Co 3+ -NTA agarose beads in cell culture. NTAagarose beads with His6-GFP immobilized at Ni 2+ , Co 2+ and Co 3+ centers were placed into cultures of adherent REF52 cell cultures for 1, 6 and 24 hours before the beads were taken out with the medium and 250 mM imidazole was added. The amount of His6-GFP in the supernatant, which eluted from the NTA agarose beads, was quantified by measuring the GFP fluorescence. Even after 24 hours in cell culture, the His6-GFP immobilized at the Co 3+ centers is still bound to the beads.  for 1 hour before all samples were run over a desalting column. Then the most concentrated protein fraction from the desalting column was applied to a Ni-NTA column, the protein was eluted with a linear imidazole gradient as described above and both the GFP and coumarin fluorescence was measured for each fraction. The GFP fluorescence is shown in green and the coumarin fluorescence in blue. Parallel to what is observed in Figure 4, the peak with the lowest affinity towards the Ni-NTA column gets bigger the higher the [Co II NTA-coumarin] concentration initially used is. High coumarin fluorescence is even observed without H 2 O 2 treatment in the early fractions in a) where there is no GFP signal. This indicates that there is some coumarin-NTA that binds nonspecifically to the protein and caries over even after the desalting column. His6-GFP and coumarin-NTA concentrations are calculated to be 96 µM and 91 µM respectively. Thus, the formed complex is approximately 1:1. b) The concentrated sample was run on a size exclusion column (CV= 25 ml, fraction size 1 ml) using detection at 280 nm, 400 nm and 490 nm. Afterwards all the fractions were also analyzed for their GFP and coumarin fluorescence. In the chromatogram signal from the coumarin and GFP were only detected at the same time. Thus, the Co 3+ mediated binding of coumarin-NTA to His6-GFP is stable. Figure S12. MALDI-TOF analysis of His6-GFP under the reaction conditions used for protein labeling. and 28567 g/mol respectively. In d) a shoulder peak is observed due to the cleavage of the His6-tag by Fenton reactions but this is not observed in the other sample despite the presence of [Co II NTA] and H 2 O 2 .