Purification of His-Tagged Proteins with [Desthiobiotin− BSA− EDTA

Bioconjugate Chem. 2008, 19, 673–679

673

Purification of His-Tagged Proteins with [Desthiobiotin-BSA-EDTA] Conjugates Exhibiting Resistance to EDTA Guy Patchornik* Affisink Biotechnology Ltd, 11 Hamaccabee St. Kiryat-Ono 55572, Israel. Received October 2, 2007; Revised Manuscript Received December 5, 2007

Two His-tagged proteins (His6-P38 and His6-Protein A) were purified by specific precipitation utilizing nonsoluble macrocomplexes composed of: BSA conjugates (modified with desthiobiotin-NHS and EDTA-dianhydride), tetrameric avidin, and Cu2+ ions. The generated pellets containing bound His-tagged proteins are washed with EDTA (25–100 mM) and then eluted in relatively high purity (g90%) devoid the macrocomplexes. Three different BSA conjugates were synthesized (DB-BSA-EDTA, DB-BSA-EDTA-A, DB-BSA-EDTA-B) and their adsorption capacities (3.8–6.4 µmol/g of BSA conjugate) as well as the recovery yields of His-tagged proteins obtained with them (44–84%) determined. The data demonstrate that capacity is dependent on the stochiometric ratio of modifying reagents (i.e., desthiobiotin-NHS and EDTA-dianhydride) used during the synthesis of the BSA conjugates. Copper ions were found to be significantly superior to Zn2+, Co2+, and Ni2+. BSA conjugates could be regenerated in moderate yields (74–83%) by incubating them at 88 °C in the presence of biotin (10 mM) at pH 7. The absence of resins leads to formation of small pellets (1–5 mg) and utilization of minute volumes of elution buffer (50–100 µL). Hence, concentrated preparations can be obtained, and a reconcentration step may be circumvented.

INTRODUCTION Immobilized metal-affinity chromatography (IMAC) has become a universal approach for isolation of native as well as recombinant proteins possessing surface-exposed amino acid residues capable of binding immobilized metal ions (Scheme 1, A). The approach was first demonstrated on metalloproteins (1) but became widespread after the extensive work of Porath (2–4) and Sulkowski (5–7). Specific adsorption is accomplished when stable ternary complexes are formed between electron donor groups from the protein and metal ions bound on different chromatographic supports (Scheme 1, A). Most of these commercial supports are modified with tridentate (e.g., iminodiacetic acid-IDA), tetradentate (e.g., nitriloacetic acid (NTA), carboxymethylated aspartic acid (CM-Asp)), or pentadentate (e.g., N,N,N′-triscarboxymethyl ethylene diamine (TED)) metal chelators. The terms tridentate, tetradentate, and pentadentate describe the number of occupied coordination bonds formed between a metal ion and a particular chelator; for example, a tridentate chelator would form three bonds with a metal ion leaving five additional coordinating sites (in the octahedral complex) available for binding. These would be generally occupied by water molecules but will be displaced by appropriate electron donor groups from a protein. Whereas tetradentate and pentadentate chelators bind metal ions with high affinity, exhibit low levels of leached metal, and possess low binding capacities (due to the limited number of coordination sites available for protein binding), tridentate chelators bind metal ions with lower affinity, lead to greater metal leakage, but result in greater binding capacities (8, 9). Although divalent metal ions are generally used (Cu2+, Ni2+, Zn2+, Co2+), trivalent metal cations were found useful as well. For example, Fe3+ ions enabled purification of phosphoproteins and organophosphates (10–12). Among the different amino acids capable of binding immobilized metal ions (Glu, Asp, Tyr, Cys, * Corresponding author. E-mail: [email protected], Tel: 972-89302575, Fax: 972-8-9302565.

Arg, His, Lys, Met, Trp, Tyr), His was found to be the major participant both in proteins (13, 14) and in peptides (15–17). Nevertheless, it has been shown that the presence of aromatic residues (Trp, Tyr, Phe) in proximity to the histidine residue contribute to the binding affinity (6, 18). The limitation of the approach to naturally surface exposed His residues was circumvented with the advent of recombinant DNA technology which provided a practical route to introduce His residues at defined sites. It was Hochuli et al. (19, 20) who demonstrated that expression of proteins with sequential His sequences (i.e., His-tag) at the N or C terminal, together with a resin modified with NTA and loaded with Ni2+ ions, leads to a general and efficient isolation process for His-tagged recombinant proteins. Moreover, the observation that the binding affinity increases with the number of water-exposed His residues (5) implied that recombinant His-tagged proteins would bind with high affinity and may allow the presence of weak chelators (e.g., imidazole) during the isolation process. The latter would suppress nonspecific binding of impurities containing naturally occurring His residues and increase the purity. In addition to the compatibility of the His-tag in most prokaryotic and eukaryotic expression systems (21), the IMAC approach was found to possess several major advantages over biospecific affinity chromatography: these include high adsorption capacities, high stability of the ligand, mild eluting conditions, applicability to proteins under denaturing conditions, and the capability to regenerate the columns at low cost (18). Scheme 1A illustrates the dependence of the IMAC approach with respect to water-insoluble resins which ideally would exhibit (a) high permeability to the target, (b) hydrophilic characteristics, (c) chemical inertness, and (d) microbiological resistance and cost effectiveness (4). Currently, no such ideal resin is available, and the ones which are exhibit some of the above characteristics but not all. For example, silica-based supports provide excellent mechanical properties but tend to lead to irreversible nonspecific adsorption (22).

10.1021/bc700368y CCC: $40.75  2008 American Chemical Society Published on Web 03/01/2008

674 Bioconjugate Chem., Vol. 19, No. 3, 2008

Patchornik

Scheme 1. Schematic Illustration of His-Tagged Protein Purification Utilizing Modified Resins (A) or Nonsoluble [Desthiobiotinylated-BSA-EDTA: Avidin: Cu2+] Macrocomplexes (B)

This state of affairs motivated us to initiate the presented study with the objective of developing an alternative platform for IMAC where no resins are utilized and thus no columns are required. The approach is based on specific precipitation of Histagged proteins together with insoluble macrocomplexes comprising modified BSA1 conjugates (desthiobiotinylated-BSAEDTA (DB-BSA-EDTA)), tetrameric avidin, and Cu2+ ions. Impurities present in the supernatant are removed by centrifugation; the pellet is washed with EDTA and the His-tagged protein is eluted with imidazole, whereas the [DB-BSA-EDTA: Avidin] macrocomplexes are not.

EXPERIMENTAL PROCEDURES Materials. For SDS-PAGE, prestained protein markers and Tris were from Bio-Rad. All other reagents were of analytical grade. Avidin, biotin, BSA, ethylenediaminetetraacetic dianhydride, octyl-β-D-glucopyranoside, n-dodecyl-β-D-maltopyranoside; urea, and EDTA were from Sigma. Desthiobiotin N-hydroxysuccinimidyl ester was from Berry and Associates, Michigan/Illinois. Recombinant His6-tagged protein A was from BioVision, Inc., Palo Alto, California. Recombinant His6-tagged P38 was expressed and purified according to Diskin et al. (23). Protein Analysis. Modified BSA conjugates concentration was determined by UV quantitative analysis at 280 nm. 1 Abbreviations: BSA, Bovine serum albumin; DB, Desthiobiotin; DB-BSA, Desthiobiotinylated BSA conjugate; DB-BSA-EDTA, Desthiobiotinylated-BSA-EDTA conjugate; DDM, n-dodecyl-β-D-maltopyranoside; EDTA, Ethylenediamine tetraacetic acid; HMW, high molecular weight; IDA, Iminodiacetic acid; MALDI-TOF, Matrix assisted laser desorption ionization time-of-flight; NTA, Nitriloacetic acid; OG, Octylβ-D-glucopyranoside; PAGE, polyacrylamide gel electrophoresis; SDS, sodium dodecyl sulfate; TED, Tris(carboxymethyl)ethylene diamine.

Electrophoresis and Densitometry. Samples were loaded onto a 10% Bis-Tris SDS-polyacrylamide gel (1 mm thickness) according to Laemmli (24) and developed for 1 h at constant 120 V. All gels were stained with Coomassie Brilliant Blue R250, and the intensity of bands was measured by densitometry using the Scion Image program. These measurements were used for determining purity and quantification of recovery yields of the target His-tagged proteins. Preparation of Desthiobiotin-BSA-EDTA (DB-BSAEDTA) Conjugates. Native BSA (10.6 mg/mL) was modified simultaneously with desthiobiotin N-hydroxysuccinimidyl ester (4.2 mg/mL) and ethylenediaminetetraacetic dianhydride (0.85 mg/mL) in the presence of 0.1 M NaPi at pH 8. Both desthiobiotin-NHS and EDTA-dianhydride were first dissolved in DMSO and only then was BSA added slowly to their mixture with constant vortexing. The stochiometric ratio among desthiobiotin-NHS, BSA, and EDTA-dianhydride was 80:1: 20, respectively. The mixture was rotated at room temperature for 4 h, the excess of reagents was removed by extensive dialysis in the presence of 50 mM Tris at pH 8, and the preparation was kept at -20 °C until use. Preparation of DB-BSA-EDTA-A and DB-BSA-EDTAB Conjugates. These were synthesized according to the above protocol used for DB-BSA-EDTA but with different stochiometric ratios. For the DB-BSA-EDTA-A conjugate, the stoichiometric ratio among the desthiobiotin-NHS, BSA, and EDTA-dianhydride was 50:1:50, and for the DB-BSAEDTA-B conjugate, the stochiometric ratio was 20:1:80, respectively. Preparation of Desthiobiotin-BSA (DB-BSA) Conjugates. This protocol is identical to the one described for DB-BSA-EDTA but was performed in the absence of EDTA-dianhydride.

Purification of His-Tagged Proteins with DB-BSA-EDTA

General Purification Protocol for His6-Tagged Proteins Utilizing DB-BSA-EDTA Conjugates. Purification requires four sequential steps. Step I. Preparation of water-insoluble [DB-BSA-EDTA: Avidin: Cu2+] macrocomplexes: Into a medium (100 µL total volume) containing 50 mM Tris at pH 8 and DB-BSA-EDTA (0.75–1.5 mg/mL), a freshly prepared avidin solution (4 mg/mL) was added with constant vortexing. Copper ions (50 mM) were added and a centrifugation step (14K, 1 min) was applied. The supernatant was excluded and the pellet was resuspended again in 50 µL of Tris 50 mM at pH 8. An additional spin (14K, 1 min) was performed, and the supernatant was removed. Step II. Binding: The pellet was resuspended in 50 mM Tris pH 8, 25 mM NaCl, and a cell lysate containing the target His-tagged protein in a total volume of 100 µL. After a short incubation (5–10 min), the supernatant was removed (14K, 1 min). Step III. Washing: The pellet was washed with 50 mM Tris pH 8, 25 mM NaCl, and 25–100 mM EDTA in a total volume of 100 µL. A spin was applied and the supernatant excluded. Step IV. Elution: The His-tagged protein was eluted from the pellet into 50–100 µL of 50 mM imidazole and 100 mM Tris pH 7 within 5–10 min. All steps were performed at 4 °C. Purity may be increased by washing the pellet twice, loading lower concentrations of copper ions during the preparation step of the macrocomplexes or by addition of mild detergents (e.g., 1% of βOG) during the binding and washing steps. Inhibition of Purification. The purification process was suppressed when (a) no Cu2+ was added to the medium, (b) imidazole (25 mM) was present during the binding step, or (c) the DB-BSA conjugate (0.75 mg/mL) was used instead of the DB-BSA-EDTA. Purification with Other Metal Cations. Isolation was performed according to the general purification protocol but in the presence of 50 mM of Zn2+, Ni2+, or Co2+ ions. The effect of Octyl-β-D-glucopyranoside (βOG) and Urea on Process Efficiency. Into the above general purification protocol, increasing concentrations of βOG (1%, 3%, and 6%) and urea (1 M, 3 M, and 8 M) were added. Both reagents were dissolved in 50 mM Tris pH 8 prior to their use. Regeneration of DB-BSA-EDTA, DB-BSA-EDTAA, and DB-BSA-EDTA-B Conjugates. All three BSA conjugates were regenerated in 74–83% yield by incubating the pellet at 88 °C for 5 min in 0.1 M Tris pH 7 and 10 mM of biotin. These recycling conditions are similar to those described earlier with other modified albumins (25). Electron Spray Analysis (ESI). Peptide fingerprinting was performed on a Bruker Reflex III matrix assisted laser desorption ionization time-of-flight mass spectrometer (Bruker, Bremen, Germany) equipped with a delayed extraction ion source, a reflector, and a 337 nm nitrogen laser.

RESULTS Purification of Recombinant His6-P38 with [DB-BSAEDTA: Avidin: Cu2+] Macrocomplexes. Isolation of Histagged proteins requires the synthesis of DB-BSA-EDTA conjugates from BSA, desthiobiotin-NHS, and EDTA-dianhydride. These conjugates were expected to possess higher molecular weight than the native BSA and to exhibit a different migration pattern. The results in Figure 1 (lanes 1–2) show that the majority of the native BSA was indeed converted into higher molecular weight entities. To bind and precipitate His6-P38, we prepared insoluble [DB-BSA-EDTA: Avidin: Cu2+] macrocomplexes and added them to the medium. After a short binding step and a spin, the generated pellets were washed with EDTA (25 mM) and the target was eluted with imidazole (50 mM) at pH 7. When we increased the EDTA concentration from 25 mM to 50, 75, and 100 mM, we found that the reductions in

Bioconjugate Chem., Vol. 19, No. 3, 2008 675

Figure 1. Purification of His6-P38 with [DB-BSA-EDTA: Avidin: Cu2+] macrocomplexes. Lane 1, native BSA; lane 2, DB-BSA-EDTA conjugates; lane 3, lysate containing His6-P38; lanes 4–7, recovered protein after washing the pellets with indicated EDTA concentrations followed by elution with 50 mM imidazole; lane 8, Bio-Rad prestained protein markers; lane 9, recovered protein from the same lysate utilizing a [Sepharose: Ni2+] column. The gel is Coomassie stained.

the yield relative to the one obtained in the presence of 25 mM EDTA were 23%, 34%, and 48%, respectively (Figure 1, lanes 4–7). Hence, the greater the EDTA concentration during the washing step, the greater the reduction in yield. Nevertheless, it was surprising to find out that, even in the presence of 100 mM of EDTA, 52% of the His6-P38 was still present in the pellet. It should be noted that when higher NaCl concentrations were employed (e.g., 50–200 mM) the efficiency of the process was decreased due to solubilization of the macrocomplexes (not shown). Yield may be improved by altering the avidin concentration; increasing the BSA conjugate concentration or inclusion of mild detergents (e.g., 1% of βOG) during the elution step. Purification of Recombinant His6-P38 with a [Sepharose: Ni2+] Column. When the same lysate containing His6-P38 was applied on a [Sepharose: Ni2+] column, a band with a similar migration pattern to the one observed with the macrocomplexes was observed (Figure 1, lane 9). This was the first indication suggesting that the same His-tagged protein was recovered by the commercial Ni 2+column and the macrocomplexes. Electron Spray Ionization (ESI) Analysis. Purified bands obtained via the [DB-BSA-EDTA: Cu2+] macrocomplexes or via the [Sepharose: Ni2+] column were analyzed by ESI. High sequence identity to the known sequence of the MK14_HUMAN Mitogen-activated protein kinase 14 (EC 2.7.11.24) was observed with both approaches (42% for the macrocomplexes and 46% for the [Sepharose: Ni2+] column), thus implying that the same His-tagged protein was isolated. Requirement for Cu2+ Ions and Covalently Bound EDTA. Two independent experiments demonstrated the participation of Cu2+ ions in the process: (a) Purification of His6P38 was achieved only when Cu2+ ions were added to the [DB-BSA-EDTA: Avidin] macrocomplexes (Figure 3A, lanes 3–4). (b) The presence of imidazole (25 mM) during the binding step inhibited the process. Imidazole was expected to compete on immobilized Cu2+ ions and inhibit the binding of His6-P38 to the macrocomplexes. The results obtained in Figure 3A, lane 5 support this speculation. The dependence of the approach on covalently bound EDTA was demonstrated by performing the process with a BSA conjugate (DB-BSA) lacking EDTA. As expected, no band corresponding to the target His6-P38 was obtained (Figure 3A, lane 6).

676 Bioconjugate Chem., Vol. 19, No. 3, 2008

Figure 2. Electron spray ionization (ESI) analysis. Purified proteins were analyzed by ESI, and the obtained peptide fragments were compared to the known sequence of the MK14_HUMAN Mitogenactivated protein kinase 14 (EC 2.7.11.24). Isolated proteins recovered via the presented approach (A) or via a [Sepharose: Ni2+] column (B) exhibited 42% and 46% sequence identity, respectively. Matched peptides are shown in bold red.

Comparing Process Efficiency with Other Divalent Cations. Four metal cations are generally utilized in purification of His-tagged proteins; these include Cu2+, Zn2+, Ni2+, and Co2+ (15). When we tested the above, we found that Cu2+ ions were significantly more efficient than Zn2+, Co2+, and Ni2+ (Figure 3B, lanes 3–6). Effect of Octyl-β-D-glucopyranoside (OG) and Urea. The gel in Figure 3C evaluates the effect of a mild detergent (OG) and urea on the process. Whereas the presence of 1–6% of OG did not show any significant effect (Figure 3C, lanes 4–6), the addition of 1 M of urea decreased the process efficiency (Figure 3C, lane 7). When urea concentration was raised to 8 M, the process was quantitatively abolished (Figure 3C, lane 9). This implies that the approach may not be applicable for purification of His-tagged proteins under denaturing conditions. Binding Capacities of DB-BSA-EDTA, DB-BSAEDTA-A, and DB-BSA-EDTA-B Conjugates. Capacity experiments were performed by precipitating increasing concentrations of the commercial highly purified His6-tagged protein A at constant concentrations of the three BSA conjugates (0.4 mg/mL). The presence of His6-tagged protein A in the supernatant after the precipitation step defined the maximum capacity of the particular BSA conjugate. The data suggest that the higher the EDTA-dianhydride concentration utilized during the synthesis of the BSA conjugates, the greater the capacity value (Table 1).

DISCUSSION This study investigates an alternative approach for immobilized metal affinity chromatography (IMAC) outlined in Scheme 1A, which does not entail the usage of resins. The approach is based on specific precipitation of His-tagged proteins together with insoluble macrocomplexes composed of BSA conjugates (DB-BSA-EDTA), avidin, and Cu2+ ions (Scheme 1B, I-II). The strategy requires preparation of modified BSA conjugates by simultaneous modification of BSA with desthiobiotin-NHS and EDTA-dianhydride. These conjugates possess two unique features: (a) the ability to become water-insoluble

Patchornik

in the presence of free avidin (this phenomenon, previously demonstrated with a series of albumins, derives from the capability of free tetrameric avidin to cross-link desthiobiotinylated albumins via their covalently bound desthiobiotins and to generate HMW macrocomplexes which precipitate (25)) (b) the capacity to bind His-tagged proteins via immobilized Cu2+ ions (Scheme 1B, I-II). Purification can then be accomplished in four sequential steps. Step I: Synthesized DB-BSA-EDTA conjugates are interconnected with native avidin to form insoluble macrocomplexes of [DB-BSA-EDTA: Avidin] which are then loaded with Cu2+ ions. The formation of insoluble macrocomplexes enables efficient removal of free or weakly bound Cu2+ ions (by centrifugation) and is equivalent to washing IMAC columns with binding buffer prior to the binding step. Step II: Addition of the [DB-BSA-EDTA: Avidin: Cu2+] macrocomplexes to a medium containing the His-tagged protein will allow specific binding, whereas impurities are removed by centrifugation. Step III: The pellet containing the adsorbed His-tagged protein is washed once to remove nonspecifically adsorbed contaminants. Step IV: The His-tagged protein is eluted in the presence of imidazole. Under these conditions, the [DB-BSA-EDTA: Avidin] macrocomplexes do not dissociate and are kept in the pellet. Indeed, incubation of [DB-BSA-EDTA: Avidin: Cu2+] macrocomplexes with a cell lysate containing His6-P38 leads to the recovery of a major band (Figure 1 lane 4) which was shown to be His6-P38 by two independent experiments: (a) The lysate containing the expressed His6-P38 was loaded on a commercial [Sepharose: Ni2+] column and a band with a similar migration pattern was observed (Figure 1, lane 9). (b) Electron spray analysis of the protein recovered with DB-BSA-EDTA conjugates (Figure 2A) exhibited high sequence coverage (42%) to the known sequence of the P38 protein kinase (MK14_HUMAN Mitogen-activated protein kinase 14 (EC 2.7.11.24)) and matched as well the amino acid sequence of the recovered protein obtained with the commercial [Sepharose: Ni2+] column (Figure 2B). Hence, both purification approaches recovered the same His-tagged protein. Unexpected results were observed while searching for conditions postulated to increase the purity of the recovered protein. We found that macrocomplexes with bound His6-P38 can be washed with EDTA solutions if applied after the precipitation step. For example, a decrease of 23%, 34%, and 48% in yield was observed when the pellets were washed with 50, 75, and 100 mM of EDTA, respectively, in comparison to 25 mM (Figure 1 lanes 4–7). These washing conditions are similar to those generally applied for regeneration of IMAC columns or for total elution of His-tagged proteins (4). We speculate that this resistance to EDTA may be the result of the following: I. Possible formation of two different metal chelators on the same BSA (structures 1–2 in Scheme 1B). Whereas structure 1 is generated when a single Lys residue attack the EDTA-dianhydride, structure 2 is obtained when two Lys residues from the same BSA participate in the nucleophilic reaction. Structure 1 is the chemical structure of the strong pentadentate chelating ligand TED, known to form octahedral complexes with divalent metal ions (22) and is used for isolation of His-tagged proteins (26), while structure 2 to the best of our knowledge is novel. It is therefore likely that at least a single strong metal chelator is present. II. Electrostatic effectssthe accessibility of EDTA to the ternary complex may be limited by electrostatic repulsions between the negatively charged residues of BSA (pI ∼ 4.6) and the carboxylates of EDTA. Moreover, the absence of linkers between the metal chelators and the BSA imply that negatively charged residues of BSA could be in proximity to the ternary complex and may be more efficient in limiting the access of

Purification of His-Tagged Proteins with DB-BSA-EDTA

Bioconjugate Chem., Vol. 19, No. 3, 2008 677

Figure 3. Characterization of BSA conjugates. A. Dependence of the process on Cu2+ ions and covalently bound EDTA. Lane 1, Bio-Rad prestained protein markers; lane 2, cell lysate containing His6-P38; lanes 3 and 4, recovered His6-P38 in the presence or absence of Cu2+ ions, respectively; lane 5, the effect of imidazole (25 mM) during the binding step; lane 6, the effect of replacing BD-BSA-EDTA with DB-BSA conjugates. B. Process efficiency with other metal cations. Lane 1, Bio-Rad prestained protein markers; lane 2, cell lysate containing His6-P38; lanes 3–6, recovered His6-P38 with indicated metal cations under identical purification conditions described in the Experimental Procedures. C. Effect of octyl-β-Dglucopyranoside (OG) and urea on the process. Lane 1, Bio-Rad prestained protein markers; lane 2, cell lysate containing His6-P38; lane 3, controlsrecovered His6-P38 in the absence of OG; lanes 4–6, effect of OG at indicated concentrations; lanes 7–9, effect of urea at indicated concentrations. The three gels are Coomassie stained. Table 1. Characteristics of Different BSA Conjugates yieldb at indicated His6-protein A concentration (mg/mL)

type of BSA conjugate

stochiometric ratios used in the synthesis Desthiobiotin-NHS: BSA:EDTA-dianhy dride

adsorption capacity for His6-protein A

avidin (mg/mL)

0.5

0.18

0.05

DB-BSA-EDTA DB-BSA-EDTA-A DB-BSA-EDTA-B

80:1:20 50:1:50 20:1:80

3.8 µmol/g 5.1 µmol/g 6.4 µmol/g

4 2 2.5

44–50% 59–66% 71–76%

59–62% 64–69% 70–73%

67–72% 74–81% 78–84%

a

a

Avidin concentration utilized during macrocomplex formation. b Range of recovery yields obtained from four independent experiments.

EDTA. III. Increased binding affinity due to additional interactions with BSA and or avidinsthis argument is supported by previous studies demonstrating the existence of four types of interactions between His-tagged proteins and metal ions immobilized on membranes (27, 28). We, therefore, speculate that similar interactions may occur between the bound His-tagged protein and residues of BSA and avidin. A series of experiments demonstrated the dependence of the approach on Cu2+ ions and covalently bound EDTA. When Cu2+ ions were not added to the medium during the binding step (Figure 3A, lanes 3 vs 4) or when 25 mM of imidazole were present in addition to Cu2+ ions, the process was quantitatively suppressed (Figure 3A, lanes 3 vs 5). Furthermore, the requirement for covalently bound EDTA was demonstrated by replacement of DB-BSA-EDTA with DB-BSA conjugates. The latter were prepared by modification of BSA only with desthiobiotin-NHS and represent BSA conjugates lacking covalently bound EDTA. These failed to recover His6-P38 (Figure 3A, lane 6) and thus provide direct evidence for the presence of covalently bound EDTA in DB-BSA-EDTA conjugates and its participation in the process. A more efficient process to the one presented could be achieved if the binding step would have been performed under homogeneous conditions. For such a scenario, the DB-BSA-EDTA conjugates, Cu2+ ions, and His6-P38 would have to be incubated in the absence of avidin, which would be added at the end of the binding step to initiate specific precipitation. Unfortunately, this preferable route utilizing different metal ions (Cu2+, Zn2+, Co2+, Ni2+) demonstrated that a high contamination background was unavoidable. The major contributors for this observation were free metal ions which promoted nonspecific precipitation even at relatively low concentrations (0.25-5 mM). Hence, formation of insoluble BSA conjugates, prior to the binding step, which

enable removal of free or weakly bound metal ions, seems to be a prerequisite for a successful process. Other metal ions (e.g., Zn2+, Co2+, Ni2+) commonly used in IMAC (15) were compared to the performance of Cu2+ ions and were found to be significantly less efficient (Figure 3B). The superiority of Cu2+ ions seems to be consistent with the order of metal affinities found, for example, with the tridentate chelator IDA (i.e., Cu2+ > Ni2+ > Zn2+ > Co2+) (22), but fails to explain the dramatic decrease in efficiency with the other cations. The applicability of the approach for isolation of membrane proteins was investigated by addition of mild detergents generally used in purification (29) or crystallization (30, 31) of membrane proteins. We found that 1–6% of octylβ-D-glucopyranoside (OG) had no significant effect on process efficiency (Figure 3C, lanes 4–6) and that n-dodecyl-β-Dmaltopyranoside leads to similar results (not shown). Therefore, utilization of mild detergents seems to be compatible with the approach. Conversely, high concentrations of urea (8 M) generally applied for purification of His-tagged proteins under denaturing conditions (32) had a pronounced negative effect on process efficiency (Figure 3C, lanes 6–9). Since similar results were observed as well with guanidinium–HCL (at 1 M), we conclude that the approach may not be applicable for isolation of His-tagged proteins under denaturing conditions. These results may not be surprising, as both urea and guanidinium–HCL are known chaotropic agents capable of increasing protein solubility (33–35) and, as such, can promote solubilization of the generated macrocomplexes composed primarily of avidin, BSA, and the His-tagged protein, thereby leading to suppression of the process. Utilization of proteins rather than synthetic polymers as the insoluble scaffold may possess several limitations deriving from the inherent chemical and physical instability of proteins in

678 Bioconjugate Chem., Vol. 19, No. 3, 2008

general (36). These could include deamidation, disulfide cleavage, oxidation, as well as aggregation, adsorption to surfaces, and protein unfolding. Therefore, one could argue that under harsh purification conditions or after prolonged storage periods, as well as repeated use of the macrocomplexes, the process efficacy would be significantly affected. Evaluation of the binding capacity of DB-BSA-EDTA conjugates was determined with a commercial highly purified (g99% by HPLC) recombinant His6-protein A. The data obtained suggest that each gram of the DB-BSA-EDTA conjugate is capable of adsorbing 3.8 µmol of His6-protein A (Table 1). In an attempt to increase the binding capacity, two additional BSA conjugates were synthesized: DB-BSA-EDTA-A and DB-BSA-EDTA-B. These were modified under conditions favoring the concentration of EDTA-dianhydride relative to that of the desthiobiotin-NHS (see Experimental Procedures). The data presented in Table 1 show that the binding capacity of the latter BSA conjugates increased in comparison to the DB-BSA-EDTA and that the values obtained (3.8–6.4 µmol/ g) are in the range of those generally observed in IMAC (4). In addition, the greater recovery yields and lower consumption of avidin of the latter BSA conjugates imply that they should be preferred over DB-BSA-EDTA (Table 1). Interestingly, whereas IMAC generally requires higher salt concentrations (0.1–1 M NaCl) to suppress nonspecific electrostatic binding (22), the presented approach is performed at low ionic strength (25 mM NaCl) and still leads to relatively high purity. This observation can be explained by the (a) absence of resins, (b) absence of linkers between the chelator and the BSA which could provide a site for nonspecific adsorption (37), (c) possibility of washing the pellets with EDTA (25–100 mM), or (d) reduction of free metal ions in the medium by BSA, which could suppress nonspecific precipitation. This argument is supported by the findings of Anderson et al. (26), who demonstrated the ability of different albumins to abstract immobilized metal ions from chromatographic supports (e.g., Zn2+-IDA or Co2+-IDA) and to lead to the metal ion transfer phenomenon.

CONCLUSIONS An alternative platform for purification of His-tagged proteins at low ionic strength has been presented. The high purity observed derives from the ability to wash the pellets with EDTA. The absence of resins leads to formation of small pellets (1–5 mg) and the requirement of minute volumes of elution buffer (50–100 µL). Therefore, concentrated preparations of purified His-tagged proteins can be obtained and a reconcentration step may be circumvented. It is hoped that the approach may provide an additional route for isolation of His-tagged proteins.

ACKNOWLEDGMENT We thank Dr. A. Shainskaya, Weizmann Institute of Science, for kindly providing and analyzing the ESI data.

LITERATURE CITED (1) Everson, R. J., and Parker, H. E. (1974) Zinc binding and synthesis eight-hydroxy-quinoline-agarose. Bioinorg. Chem. 4, 15–20. (2) Porath, J., Carlsson, J., Olsson, I., and Belfrage, G. (1975) Metal chelate affinity chromatography, a new approach to protein fractionation. Nature 258, 598–599. (3) Porath, J., and Olin, B. (1983) Immobilized metal ion affinity adsorption and metal ion affinity chromatography of biomaterials: serum protein affinities for gel-immobilized iron and nickel ions. Biochemistry 22, 1621–1630. (4) Porath, J. (1992) Immobilized metal ion affinity chromatography. Protein Expression Purif. 3, 263–281.

Patchornik (5) Sulkowski, E. (1985) Purification of proteins by IMAC. Trends Biotechnol. 3, 1–7. (6) Sulkowski, E. (1989) The saga of IMAC and MIT. Bioessays 10, 170–175. (7) Sulkowski, E. (1996) Immobilized metal-ion affinity chromatography: imidazole proton pump and chromatographic sequelae: I. Proton pump. J. Mol. Recognit. 9, 389–393. (8) Gaberc-Porekar, V., and Menart, V. (2005) Potential for using histidine tags in purification of proteins at large scale. Chem. Eng. Technol. 28, 1306–1314. (9) Suen, S. Y., Liu, Y. C., and Chang, C. S. (2003) Exploiting immobilized metal affinity membranes for the isolation or purification of therapeutically relevant species. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 97, 305–319. (10) Anderson, L., and Porath, J. (1986) Isolation of phosphoproteins by immobilized metal (Fe3+) affinity chromatography. Anal. Biochem. 154, 250–254. (11) Muszynska, G., Anderson, L., and Porath, J. (1986) Model studies on iron (III) ion affinity chromatography. Biochemistry 25, 6850–6853. (12) Dobrowolska, G., Muszynska, G., and Porath, J. (1991) Selective adsorption of phosphoproteins on gel-immobilized ferric chelate. J. Chromatogr. 541, 331–339. (13) Hemdan, E. S., Zhao, Y. J., Sulkowski, E., and Porath, J. (1989) Surface topography of histidine residues: a facile probe by immobilized metal ion affinity chromatography. Proc. Natl. Acad. Sci. U.S.A. 86, 1811–1815. (14) Sulkowski, E. (1987) Immobilized metal ion affinity chromatography of proteins, In Protein Purification: Micro to Macro Liss, A., Ed., pp 149-169. (15) Below, M., and Porath, J. (1990) Immobilized metal ion affinity chromatography, effect of solute structure, ligand density and salt concentration on the retention of peptides. J. Chromatogr. 516, 333–354. (16) Nakagawa, Y., Yip, T.-T., Belew, M., and Porath, J. (1988) High performance immobilized metal ion affinity chromatography of peptides: Analytical separation of biologically active synthetic peptides. Anal. Biochem. 168, 75–81. (17) Smith, M. C., Furman, T. C., Ingolia, T. D., and Pidgeon, C. (1988) Chelating peptide immobilized metal affinity chromatography. J. Biol. Chem. 2633, 7211–7215. (18) Arnold, F. H. (1991) Metal-affinity separations: a new dimension in protein processing. Biotechnology 9, 151–156. (19) Hochuli, E., Doebeli, H., and Shachar, A. (1987) A new metal chelate adsorbent selective for proteins and peptides containing neighboring histidine residues. J. Chromatogr. 411, 177–184. (20) Hochuli, E., Bannwarth, W., Doebeli, H., Gentz, R., and Stueber, D. (1988) Genetic approach to facilitate purification of recombinant proteins with a novel metal chelate adsorbent. Bio/ Technology 1321–1325. (21) Seidler, A. (1994) Introduction of a histidine tail at the N-terminus of a secretory protein expressed in Escherichia coli. Protein Eng. 7, 1277–1280. (22) Gaberc-Porekar, V., and Menrat, V. (2001) Perspective of immobilized-metal affinity chromatography. J. Biochem. Biophys. Methods 49, 335–360. (23) Diskin, R., Lebendiker, M., Endgelberg, D., and Livnah, O. (2007) Structures of p38 101 active mutants reveal conformational changes in L16 loop that induce autophosphorylation and activation. J. Mol. Biol. 365, 6676. (24) Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. (25) Patchornik, G. (2007) Double-modified albumins as a tool for antibody purification. J. Biochem. Biophys. Methods 70, 671–673. (26) Anderson, L., Sulkowski, E., and Porath, J. (1991) Immobilized metal ion chromatography of serum albumins. Bioseparation 2, 15–22. (27) Tsai, Y.-H., Wang, M.-Y., and Suen, S.-Y. (2002) Purification of hepatocyte growth factor using polyvinyldiene fluoride-based immobilized metal affinity membranes: equilibrium adsorption study. J Chromatogr., B 766, 133–143.

Purification of His-Tagged Proteins with DB-BSA-EDTA (28) Wu, C.-Y., Suen, S.-Y., Chen, S.-C., and Tseng, J-H. (2003) Analysis of protein adsorption on regenerated cellulose-based immobilized copper ion affinity membranes. J Chromatogr., A 996, 53–70. (29) Cohen, E., Goldshleger, R., Shainskaya, A., Tal, D. M., Ebel, C., Le Maire, M., and Karlish, S. J. (2005) Purification of Na+,K+-ATPase expressed in pichia pastoris reveals an essential role of phospholipid-protein interactions. J. Biol. Chem. 280, 16610–16618. (30) Michel, H. (1983) Crystallization of membrane proteins. Trends Biochem. Sci. 8, 56–59. (31) Ostermeier, C., and Michel, H. (1997) Crystallization of membrane proteins. Curr. Opin. Struct. Biol. 7, 697–701. (32) Nilsson, J., Ståhl, S., Lundeberg, J., Uhlén, M., and Nygren, P. Å. (1997) Affinity fusion strategies for detection, purification, and immobilization of recombinant proteins. Protein Expression Purif. 11, 1–16. (33) Hatefi, Y., and Hanstein, W. G. (1969) Solubilization of particulate proteins and nonelectrolytes by chaotropic agents. Proc. Natl. Acad. Sci. U.S.A. 62, 1129–1136.

Bioconjugate Chem., Vol. 19, No. 3, 2008 679 (34) Santoro, M. M., and Bolen, D. W. (1988) Unfolding free energy changes determined by the linear extrapolation method. 1. Unfolding of phenylmethanesulfonyl alpha-chymotrypsin using different denaturants. Biochemistry 27, 8063–8068. (35) Shortle, D., Meeker, A. K., and Gerrin, S. L. (1989) Effects of denaturants at low concentrations on the reversible denaturation of staphylococcal nuclease. Arch. Biochem. Biophys. 103– 113. (36) Jacob, S., Shirwaikar, A. A., Srinivasan, K. K., Alex, J., Prabu, S. L., Mahalaxmi, R., and Kumar, R. (2006) Stability of proteins in aqueous solution and solid state. Ind. J. Pharm. Sci. 68, 154– 163. (37) Lowe, C. R. (1977) The synthesis if several 8 substituted derivatives of adenosine 5′-monophosphate to study the effect of the spacer arm in affinity chromatography. Eur. J. Biochem. 73, 265–274. BC700368Y