A novel method for removing contaminant Hsp70 molecular chaperones from recombinant proteins

Abstract

The production of recombinant proteins in bacteria has increased significantly in recent years, becoming a common tool for both research and the industrial production of proteins. One of the requirements of this methodology is to obtain the desired protein without contaminants. However, this goal cannot always be readily achieved. Multiple strategies have been developed to improve the quality of the desired protein product. Nevertheless, contamination with molecular chaperones is one of the recalcitrant problems that still affects the quality of the obtained proteins. The ability of chaperones to bind to unfolded proteins or to regions where the polypeptide chain is exposed make the removal of the contamination during purification challenging to achieve.

This work aimed to develop a strategy to remove contaminating DnaK, one of the homologous Hsp70 molecular chaperones found in Escherichia coli, from purified recombinant proteins. For this purpose, we developed a methodology that captures the DnaK from the contaminating proteins by co‐incubation with a GST‐cleanser protein that has free functional binding sites for the chaperone. The cleanser protein can then be easily removed together with the captured DnaK. Here, we demonstrated the utility of our system by decontaminating a Histidine‐tagged recombinant protein in a batch process. The addition of the GST‐cleanser protein in the presence of ATP‐Mg eliminates the DnaK contamination substantially. Thus, our decontaminant strategy results versatile and straightforward and can be applied to proteins obtained with different expression and purifications systems as well as to small samples or large volume preparations.

Abbreviations

Pep_(pET29b)

peptide encoded in pET29b plasmid

Trx_(pET32b)

thioredoxin protein and connector peptide from pET32b plasmid

TrxFNR

mature region of pea ferredoxin‐NADP⁺ reductase fused to thioredoxin and thrombin recognition site encoded in pET205 plasmid

GST_(pGEX‐3X)

glutathione S‐transferase protein and connector peptide encoded in pGEX‐3X plasmid

GST_(pGEX‐2T)

glutathione S‐transferase protein and connector peptide encoded in pGEX‐2T plasmid

E. coli ΔdnaK

E. coli BB1553 strain (deficient in DnaK chaperone)

GST‐DnaK‐free

GST protein encoded in pGEX‐2T plasmid purified from E. coli BB1553 cells

TN buffer

50 mM Tris–HCl, pH 8.0, 150 mM NaCl

TN‐ATP buffer

TN buffer containing freshly prepared 5 mM ATP‐Mg

Introduction

The production of recombinant proteins is an essential stage of many research projects and different biotechnological developments. The most common methods employ bacteria as expression systems and various types of fusion proteins for further purification.1, 2

The quality of the purified protein is always the object of attention.3 Proteins with persistent contaminants can give rise to erroneous results, complicate experiments, or hamper their industrial application.

A reiterated contaminant found largely in protein purifications are the molecular chaperones.4, 5, 6, 7, 8 Chaperones participate in protein folding by assisting in the self‐assembly process of polypeptides. Most of the bacterial proteins interact with one or more chaperones during their folding process. Likewise, molecular chaperones take part in protein refolding and disaggregation, in proteolytic degradation and other essential processes of the cell, such as the translocation of polypeptides through membranes and signaling.9 One of the ways in which chaperones fulfill their function is through interacting with certain polypeptide regions. Consequently, during heterologous proteins’ expression, chaperone molecules usually bind to the protein under study, in many cases, with very high affinity. Therefore, when purifying the desired protein, the chaperone remains attached to it until the last stages of the purification process, without being removed. In particular, Hsp70 molecular chaperones recognize and bind specific protein sequences abundantly distributed in almost all proteins.10 These polypeptides regions are generally displayed during the synthesis of the protein or remain exposed after folding in the interdomain areas or loops. This phenomenon is favored when protein fusions are used since the linker regions between the carrier and the target protein are generally accessible to other proteins.

The bound chaperone could generate several anomalies in the process of obtaining the protein with high purity, even when present in undetectable quantities. In cases where the chaperone has bound to regions previously exposed, it may prevent the protein from achieving its final folding state, thus generating dangerous heterogeneity amongst protein population. In some other cases, its binding can mask the cleavage site of the fusion protein to its specific protease, thereby restricting the scission process. On the other hand, Hsp70‐type chaperones are very immunogenic molecules. Consequently, the co‐purification of them with the sample can generate an unwanted antigenic response when applied to animals or humans. Finally, the chaperone could alter the activity or function performed when accompanying the protein under study. The Hsp70 molecules within the essay media might be released and eventually bind to another protein and/or receptor, altering the results. This scenario could be even worse in reactions involving ATP, as the chaperone can add an exogenous ATPase activity that may change concentrations of this nucleotide in the medium.

Several strategies have been used to remove chaperones of this type.4, 6, 7 Previously, we designed a system that makes use of an Escherichia coli heat denatured protein lysate, which is added to the sample together with ATP, to promote mobilization of the chaperones from the target protein to the unfolded E. coli proteins.5 The method was successfully applied to different protein purifications. Some examples of the usefulness of this method can be found in the literature.11, 12, 13, 14 However, some drawbacks inherent to this procedure could compromise its general application, for example, the addition of proteins in high quantity, the diversity of them and the possible existence of some still active proteins that have overcome the heat treatment, especially proteases. In this respect, we have designed an improved system for the removal of DnaK, one of the homologous E. coli Hsp70 molecular chaperones, from protein samples obtained by recombinant expression in E. coli.

The system is based explicitly on adding to the protein that needs to be purified a “cleanser protein” that has been chosen to include functional binding sites for the DnaK chaperone. This cleanser protein was obtained in an E. coli mutant strain that lacks DnaK, the main Hsp70 molecular chaperone on this organism, and can be easily eliminated after cleaning. In the presence of ATP, we verified that it effectively removes contaminating DnaK from the purified protein. The method is inexpensive and straightforward and, can be applied routinely in small and large purifications, both in batch and in column chromatography.

Results

As we previously demonstrated, the linker peptides and carrier proteins encoded in various commonly utilized E. coli expression plasmids contain binding sites to the DnaK that lead to the inevitable contamination of the produced recombinant proteins with this molecular chaperone.5 Moreover, natural proteins with unstructured regions and/or exposed amino or carboxy‐terminal region can interact with Hsp70 molecular chaperones. To overcome this drawback inherent to the purification process of recombinant and natural proteins, we developed a simple decontamination strategy. As a first approach to develop our DnaK molecular chaperone decontamination system, proteins expressed by pGEX and pET plasmids were chosen, taking into account the values of predicted DnaK affinities estimated by Rial et al.5 Then, the chosen proteins were expressed and purified to select those which have detectable binding of DnaK [Fig. 1(A–C)]. The polypeptide and GST protein purified using pET29b and pGEX‐2T plasmids for protein expression respectively, showed the highest levels of contamination with DnaK [Fig. 1(A–C); lanes and bars Pep_(pET29b) and GST_(pGEX‐2T)]. The contamination levels found in these purified proteins were approximately 7–14 times higher than those found in Thioredoxin (Trx) and GST proteins purified using the pET32b and pGEX‐3X plasmids for protein expression, respectively [Fig. 1(A–C); lanes and bars Trx_(pET32b) and GST_(pGEX‐3X)], and 2.3 times greater than that found in the FNR protein purified from pET205 plasmid [Fig. 1(A–C); lane and bar Trx FNR]. These observations demonstrate that the carrier proteins encoded in the pET29b and pGEX‐2T plasmids have functional binding sites for DnaK, being able to hold the chaperone attached throughout the purification process. The difference found in the contamination levels with DnaK between the GST proteins purified from pGEX‐2T or 3X plasmids [Fig. 1(A–C); lanes and bars GST_(pGEX‐2T) and GST_(pGEX‐3X)] may be attributed to linker region, thus demonstrating that the choice of the connector peptide and/or protease recognition sequence linking the carrier to the protein of interest can strongly impact the quality of a purification; as previously suggested.5 As shown above, when the carrier proteins encoded in pGEX‐2T and pET29b were expressed in E. coli cells and purified, the available DnaK binding sites of these polypeptides became occupied with this endogenous chaperone. Thus, to develop a DnaK decontaminant system, we have cloned and expressed the pGEX‐2T plasmid in an E. coli strain deficient in the DnaK chaperone (E. coli ΔdnaK hereafter) to obtain GST proteins in which the binding sites for DnaK are empty of this chaperone. As shown in Fig. 1(D and E) when the pGEX‐2T plasmid was expressed in E. coli JM109 cells (DnaK contamination control), the GST protein and connector peptide encoded in it (GST_(pGEX‐2T) hereafter) copurified with the endogenous DnaK chaperone [Fig. 1(D and E); lane E. coli JM109]. However, when the pGEX‐2T plasmid was expressed in our E. coli ΔdnaK strain, the DnaK chaperone was not detected on the purified GST_(pGEX‐2T) protein [Fig. 1(D and E); lane E. coli ΔdnaK], demonstrating that the functional binding sites for DnaK present in the GST_(pGEX‐2T) protein are unoccupied. Based on these results, the purified GST_(pGEX‐2T) protein from E. coli ΔdnaK cellular extracts (GST‐DnaK‐free hereafter) could be utilized as a “cleanser” protein to eliminate the DnaK contamination from recombinant proteins of interest. To verify this hypothesis experimentally, we chose the FNR protein encoded in the pET205 plasmid (TrxFNR hereafter) as the target protein being purified from DnaK contamination. FNR is a FAD‐containing hydrophilic protein of about 35 kDa present in chloroplasts and cyanobacteria that can be efficiently expressed in E. coli.15 Thus, we initially expressed the TrxFNR protein in E. coli BL21(DE3) cells. Then, we performed a Ni‐NTA‐based IMAC purification. Once the TrxFNR was bound to Ni‐NTA beads, and before to its elution, additional washing steps were performed using the same wash buffer supplemented with ATP‐Mg and the purified GST‐DnaK‐free protein in a molar ratio equivalent to 1:1 respect to the immobilized TrxFNR. The DnaK, as well as other chaperones, is an ATP‐dependent protein whose activity is associated with ATP hydrolysis. The ATP molecule induces a turnover in the sites for substrate binding on DnaK, producing an alteration between the ATP‐bound state with low affinity and a fast exchange rate for the substrate, and the ADP‐bound state with high affinity and slow exchange rates for substrate.16 Our results showed that the addition of the GST‐DnaK‐free protein in the wash buffer supplemented with ATP‐Mg decreased the amount of DnaK that copurified with the TrxFNR protein [Fig. 2(A–C)]. At the same time, an equivalent increase of DnaK chaperone was encountered in the respective recovered washes [Fig. 2(D–F); lanes and bars GST‐DnaK‐free recovered wash 0, 1, 2, 3, and 5]. As an additional control, the presence of DnaK was analyzed in the purified GST‐DnaK‐free protein previously to its incubation with the TrxFNR‐Ni‐NTA complex, detecting no DnaK band at all [Fig. 2(D, E); lane GST‐DnaK‐free]. It is known that ATP‐Mg triggers the release of the peptide substrate from DnaK.16 Accordingly, in experiments performed to compare the DnaK content of the TrxFNR subjected to up to five washes with TN buffer with respect to those conducted with TN‐ATP buffer, a greater release of the contaminating DnaK could be noticed in the presence of ATP. However, the experiments also showed that the addition of the GST‐DnaK‐free protein was substantial for the chaperone removal and that its presence increased the purification achieved (Fig. S1). Together, these results demonstrate that the DnaK chaperone can dissociate from the protein of interest and bound to the non‐retained GST‐DnaK‐free protein in the presence of ATP‐Mg. Moreover, our experiments in batch showed that the amount of contaminant DnaK in the purifications of TrxFNR became smaller as the number of washes with the GST‐DnaK‐free protein increased. In this sense, DnaK contamination was substantially removed after five washes with the GST‐DnaK‐free protein in the presence of ATP‐Mg [Fig. 2(C)].

Figure 1.

Preparation of recombinant proteins with different degrees of DnaK contamination. (A–C) Analysis of the levels of contamination with DnaK found in different purified carrier proteins commonly used in E. coli expression systems. Lane Pep_(pET29b): peptide encoded in pET29b plasmid; Lane Trx_(pET32b): thioredoxin protein and connector peptide from pET32b plasmid; Lane TrxFNR: thioredoxin protein fused to the mature pea ferredoxin‐NADP⁺ reductase encoded in the pET205 plasmid; Lanes GST_(pGEX‐3X) and GST_(pGEX‐2T): GST protein and connector peptide from pGEX‐3X and pGEX‐2T plasmids, respectively. The level of DnaK contamination in each purified carrier protein (C) was obtained by determining the intensity of the DnaK band in the immunoblot B as integrated optical density (IOD_DnaK) and expressed relative to the IOD of the corresponding Ponceau red stain band (IOD_{purified protein}) in A. (D and E) Western blot with anti‐DnaK antiserum against purified GST protein obtained from JM109 E. coli cells (E. coli JM109 lane), or from BB1553 E. coli cells (E. coli ΔdnaK lane), previously transformed with the pGEX‐2T plasmid. Purified proteins were run in 12% SDS‐PAGE, transferred to nitrocellulose membrane, stained with Ponceau red solution (A and D), and analyzed by Western blot using a specific anti‐DnaK antiserum (B and E).

Figure 2.

Removal of DnaK contamination from the Ni‐NTA linked TrxFNR protein by washes with the GST‐DnaK‐free protein. (A–C) Analysis of DnaK contamination in the TrxFNR eluates after consecutive washing steps with the GST‐DnaK‐free protein. (A) TrxFNR protein eluates were run in 12% SDS‐PAGE, transferred to nitrocellulose membrane, stained with Ponceau red solution, and (B) analyzed by Western blot with anti‐DnaK antiserum. Lane 0: TrxFNR washed without GST‐DnaK‐free protein; Lanes: 1, 2, 3, and 5: TrxFNR washed once, twice, three, or five times with GST‐DnaK‐free protein, respectively. A molar ratio 1:1 of GST‐DnaK‐free to TrxFNR proteins was used in each wash. (C) DnaK contamination was estimated by the integrated optical density of the DnaK band in the immunoblot (IOD_DnaK) relative to the corresponding IOD of the purified TrxFNR protein band in the Ponceau red stain (IOD_TrxFNR). The DnaK contamination in the TrxFNR protein eluate not subjected to the cleaning process with GST‐DnaK‐free (Lane 0) was assumed as 1 (100%). Results are expressed as mean ± SEM and are representative of three independent experiments. (D–F) Analysis of the washes: In each recovered wash the DnaK protein was evinced by Western blot with the anti‐DnaK antiserum. (D) Blotted nitrocellulose membrane stain with Ponceau red reagent. (E) Immunoblot. Lane 0: washed without the addition of GST‐DnaK‐free; Lanes: 1, 2, 3, and 5: first, second, third, and fifth washed with the GST‐DnaK‐free protein, respectively. Lane GST‐DnaK‐free: GST‐DnaK‐free protein before its incubation with the TrxFNR protein. (F) The amount of the DnaK protein extracted in each wash was estimated by the IOD measurement of the DnaK band in the immunoblot (IOD_DnaK). The amount of DnaK quantified in the wash without the addition of GST‐DnaK‐free was set as 1 (100%). Results are expressed as mean ± SEM; n = 3.

Finally, we determined the effect of the amount of GST‐DnaK‐free protein added to the washes on the purity of the protein of interest. In this case, the TrxFNR protein bound to the Ni‐NTA resin was washed three times with different quantities of GST‐DnaK‐free protein, equivalent to molar ratios of 0:1; 0.2:1; 0.5:1; and 1:1 of GST‐DnaK‐free protein to Ni‐NTA resin bound TrxFNR protein (Fig. 3). As previously demonstrated, the analyses of the TrxFNR eluates showed that the addition of the GST‐Dnak‐free protein in the wash buffer is an effective method to extract the DnaK protein from the TrxFNR–DnaK complex bound to Ni‐NTA beads. In this regard, the addition of the GST‐DnaK‐free protein plus ATP‐Mg in the wash buffer extracted consistently more DnaK from the TrxFNR eluates than the control in which the TrxFNR protein was washed with wash buffer supplemented only with ATP‐Mg without GST‐DnaK‐free addition (Fig. 3). However, no appreciable differences in the contamination levels with DnaK were observed between the eluates obtained after the washes with the different amounts of GST‐DnaK‐free protein analyzed. In this way, our results demonstrate that the addition of a molar ratio of 0.2:1 GST‐DnaK‐free/TrxFNR is enough to eliminate the contaminant DnaK protein that copurifies together with the protein of interest. Since our method is based on the addition of a foreign protein to the preparation, we carried out controls to verify that the GST‐DnaK‐free added does not affect the quality of the obtained protein. Figure S2 shows that the addition of 20 mM imidazole to the TN buffer in the final washing steps was enough to remove the added GST‐DnaK‐free protein entirely. Alternatively, the addition of a small amount of glutathione agarose resin before the final use of the purified protein thoroughly cleans it from GST‐DnaK‐free protein.

Figure 3.

Removal of DnaK contamination by washing with increasing amounts of GST‐DnaK‐free protein. (A) Purified TrxFNR proteins obtained after three washes with molar ratios 0:1; 0.2:1; 0.5:1; and 1:1 of GST‐DnaK‐free to Ni‐NTA linked TrxFNR. Samples were run in 12% SDS‐PAGE, transferred to nitrocellulose membrane and stained with Ponceau red solution (Lanes: 0:1; 0.2:1; 0.5:1; and 1:1, respectively). (B) DnaK contamination was evidenced by incubating the blotted membrane with anti‐DnaK antiserum. (C) In each purification, the DnaK contamination (IOD_DnaK) was estimated as described in Fig. 1 and was done relative to the integrated optical density of the corresponding TrxFNR band in the Ponceau red stain (IOD_TrxFNR). The amount of DnaK contamination in the 0:1 lane was set as 1 (100%).

Experiments performed with 0.8 mg TrxFNR showed essentially equivalent results (Fig. S3), indicating that the method can be scaled according to the production needs of the desired protein.

Discussion

Chaperone contamination is a common problem during the purification of many recombinant proteins of interest expressed in prokaryote systems. Previously, several experimental methodologies have been addressed for their elimination. In our laboratory, the washing of Ni or glutathione matrices containing the protein of interest bound with very low salt solutions (10 mM Tris–HCl, pH = 8.0) or diluted organic solvents (5% ethanol in 10 mM Tris–HCl, pH = 8.0) did not diminish the observed DnaK contamination of the purified protein. Attempts to remove DnaK using ion exchange chromatography or gel filtration were also unsuccessful because of the strong interaction between the chaperone and the target protein.5 As previously mentioned, the ATP induces a turnover between high and low affinity in the sites for substrate binding on DnaK. However, it has been shown that the addition of ATP to bacterial lysates containing fusion proteins did not reduce the DnaK contamination but usually increased it.5 The reason for this could be that the ATP probably induced the release of DnaK from bacterial unfolded proteins and thus allowed it to bind to other proteins, including the protein of interest. Other authors8 have developed an efficient and inexpensive method to eliminate the copurifying bacterial chaperonin GroEL from overexpressed GST‐fusion proteins using ATP and urea‐denatured bacterial lysates that could also be applied to remove DnaK. However, care should be taken because excessive amounts of urea may lead to denaturation and/or loss of the fusion protein. Ratelade et al.7 have constructed an E. coli BL21(DE3) derivate strain free from DnaK chaperone by inactivating the dnaK gene (EN2 strain; CNCM I‐3863; Pasteur Institute). These cells are able to express high levels of correctly folded, active and soluble recombinant protein under the control of the T7 promoter in the absence of endogenous DnaK chaperone. However, this approach obligatorily requires the EN2 strain for the synthesis of the recombinant protein, limiting the possibility of using other potentially more efficient E. coli strains.

Previously, we developed an approach to reduce the unwanted contamination of DnaK by washing the recombinant proteins bound to the purification resin with ATP‐Mg plus homologous soluble heat‐denatured proteins before the elution.5 However, as mentioned in the introduction, some drawbacks inherent to this procedure that could affect its general application, induced us to design an improved procedure for the removal of the DnaK chaperone from protein samples obtained by recombinant expression in E. coli. It serves for Ni‐NTA‐based IMAC purifications and consists in washing the protein of interest bound to the Ni‐NTA matrix with a GST protein containing functional free binding sites for DnaK. Only a small amount of GST‐cleanser protein was necessary to eliminate efficiently the DnaK attached to the recombinant protein of interest. It may be concluded that the number of DnaK binding sites of the cleanser protein, their affinities for DnaK and accessibility in the structure may influence the effectiveness of the chaperone removal process. In this sense, we have previously described a bioinformatic method17 which can be used to identify possible DnaK binding sites and their affinities using an algorithm described previously,10 thus improving the selection of a cleanser protein.

Another aspect to consider is the affinity for DnaK of the protein of interest and the protein used to remove the chaperones. Our method was applied to the thioredoxin‐FNR fusion protein which has a calculated minimum value of ΔΔG_K of −3.92.17 The GST‐DnaK‐free protein used for the removal of DnaK has a ΔΔG_K of −3.10 as calculated using the algorithm of Rüdiger et al.10 to predict DnaK affinities.17 Consequently, in our experiments, the protein used for DnaK removal has an estimated lower affinity for DnaK than the recombinant protein of interest. However, it was possible to remove the contaminating DnaK. These results could be explained by the fact that in the presence of ATP, DnaK adopts a conformation of low affinity for the protein substrate and exchanges it rapidly. This causes the chaperones to be released from the protein substrates. Subsequently, ATP turnover favors a high‐affinity of DnaK (ADP‐bound form) and a slow exchange rate of the polypeptides bound. During the low‐affinity conformation of DnaK, TrxFNR is released from the chaperone and binding to GST‐DnaK‐free protein may occur. After DnaK changes to a high‐affinity conformation (ADP bound), the slow exchange of the DnaK attached to the GST‐DnaK‐free protein may favor their elimination. Thus, not only affinities but also the time it takes to achieve equilibrium should be important for the capability of the GST‐DnaK‐free protein to remove contaminating DnaK. However, the possibility that a protein with high affinity for DnaK could not be efficiently cleaned with this system cannot be excluded.

All in all, we have developed a useful tool for solving the problem of DnaK contamination. The proposed method is simple, inexpensive, efficient for its purpose (it only requires the addition of an inert and easily removable protein), and essentially versatile, being able to be used to remove DnaK contamination from Hist‐tagged recombinants protein expressed in any E. coli strain. Even more, any already purified protein from recombinant or natural sources in which DnaK contamination is detected can be subjected to this cleaning procedure. Also, although we have only evaluated our new DnaK‐decontaminant strategy in a batch mode, there is no reason to think it should not work with equal efficiency for column purification systems. On the other hand, our cleaning method could be adapted to eliminate the contamination with DnaK in other E. coli protein‐tagged purification systems. In this regard, we were able to verify that the carrier peptide encoded in the pET29b plasmid (Hist‐tagged peptide) has functional binding sites for the DnaK chaperone [Fig. 1(A–C); lane and bar Pep_(pET29b)]. A protein fusion cloned, expressed and purified from the EN2 E. coli strain7 could be used to remove the contamination with DnaK from recombinant proteins tagged with GST or maltose binding protein.

Materials and Methods

DnaK binding sites prediction

Identification of possible DnaK binding sites on the translated linker regions of the different plasmids used in this paper, and their respective value of overall statistical energy contributions (ΔΔGk), was performed as described previously,17 using the algorithm of Rüdiger et al.10

Plasmids and strains

The pET205 plasmid contains the cDNA sequence coding for the mature region of pea ferredoxin‐NADP⁺ reductase (FNR) fused to thioredoxin (Trx) immediately after the thrombin recognition site in a modified pET‐32 plasmid (Novagen, USA). The construction of this plasmid is described elsewhere.5 Plasmids pET205, pET‐29b[+], and pET‐32b(+) were transformed into BL21(DE3) E. coli cells. The plasmids pGEX‐3X (GE Healthcare Life Sciences; UK) and pGEX‐2T (GE Healthcare Life Sciences; UK) codifying for GST (Xa factor) and GST (thrombin site) respectively, were transformed into JM109 E. coli cells. Also, the pGEX‐2T plasmid was isolated and purified from JM109 E. coli cells and transformed into BB1553 E. coli (dnaK deficient) cells. BB1553 E. coli cells are derivate from MC4100 E. coli cells18 and are available upon request.

Protein expression and purification

Expression and purification of the proteins GST (Xa factor), GST (thrombin site), TrxFNR (pET205 plasmid) and Trx (pET32b(+) plasmid), and the Hist‐tag fusion peptide encoded in pET29b(+) plasmid was carried out as described previously.5 For the expression of the GST protein (thrombin site) in BB1553 strain, cells were grown at 28°C in Luria–Bertani medium containing 10 μg/mL of chloramphenicol and 100 μg/mL of ampicillin antibiotics, until the suspension reached an optical density of 0.6 at 600 nm. Protein expression was induced by adding 0.75 mM IPTG and incubation was continued for an additional 16 h period at 22°C. Finally, cells were harvested and the GST‐DnaK‐free protein was purified as described.5 All purified proteins were aliquoted and stored frozen at −70°C until use.

Release of DnaK from resin‐bound TrxFNR protein

To investigate the effect of the number of washes on the release of the contaminating DnaK the following experiments were performed. A 250 μL Ni‐NTA agarose resin containing‐tube, previously equilibrated with TN buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl) was incubated with 500 μg of pure TrxFNR protein at 4°C for 60 min under gentle stirring. Then, the resin was washed twice with 10 bed volumes (BV) of TN buffer to eliminate the non‐retained TrxFNR protein and subsequently divided into five equal parts (50 μL of resin each). One of the fractions was washed with 8 BV of TN buffer containing freshly prepared 5 mM ATP‐Mg (TN‐ATP buffer hereafter) at 25°C for 10 min with gentle shaking. The other four fractions were washed once, twice, three or five times, respectively, with 8 BV of TN‐ATP buffer supplemented with 50 μg of pure GST‐DnaK‐free protein under the same conditions. All the fractions were then washed three times (10 BV each wash) with TN buffer supplemented with imidazole 20 mM at 4°C for 10 min. Each wash was performed with gentle shaking to eliminate the GST‐DnaK‐free protein previously added. The TrxFNR protein was finally eluted from the Ni‐NTA agarose resin with TN buffer containing 150 mM imidazole for 30 min with gentle shaking at 4°C. Eluted samples were analyzed by SDS‐PAGE and Western blot. To investigate the effect of the amount of GST‐DnaK‐free protein added to the washes we employed the same procedure described before with the exception that in this case the Ni‐NTA resin with the TrxFNR attached was divided into four equal fractions. Each fraction was washed three times with 8 BV of TN‐ATP buffer only or three times with TN‐ATP buffer (8 BV) containing 10 μg, 25 μg or 50 μg of GST‐DnaK‐free protein, respectively. The following steps were carried out as described above.

SDS‐PAGE and Western blot analysis

SDS‐PAGE was performed on 12% polyacrylamide gel.19 Gels were electro‐transferred to nitrocellulose membranes.20 The membranes were first blocked with 5% non‐fat dry milk, 0.05% Tween 20 in TBS buffer and incubated with the anti‐DnaK antiserum (1:3000 dilution). The E. coli DnaK protein was later identified using 1:8000 dilution of anti‐rabbit IgG alkaline phosphatase conjugated antibodies (Sigma; St. Louis, MO) and bromochloroindolyl phosphate and nitro blue tetrazolium substrates. Protein levels were estimated by the integrated optical densities (IOD) of bands after quantitative scanning of Ponceau red stained or Immunodetected blots using ImajeJ software.

Reagents

Hybond™‐ECL™ Nitrocellulose membrane was purchased from Amersham Biosciences (Buckinghamshire, UK). Nickel‐NTA agarose resin was purchased from QIAGEN (Valencia, CA). All other reagents were purchased from Sigma‐Chemical Co (St. Louis, MO).

Supplementary material

Supplementary Figures: Figure S1. Removal of DnaK contamination from the Ni‐NTA bound TrxFNR protein by washing with the GST‐DnaK‐free protein. Figure S2. Analysis of GST‐DnaK‐free protein cross‐contamination in the TrxFNR eluates. Figure S3. Evaluation of the removal of DnaK contamination from the Ni‐NTA bound TrxFNR protein by washes with GST‐DnaK‐free protein in large scale preparations. Filename: Supplementary figures Morales et al.pdf.

Supporting information

Figure S1 Removal of DnaK contamination from the Ni‐NTA bound TrxFNR protein by washing with the GST‐DnaK‐free protein. (A–C) Analysis of DnaK contamination in the TrxFNR eluates after consecutive washing steps with TN buffer; TN‐ATP buffer and the GST‐DnaK‐free protein in TN‐ATP buffer. Experiments were performed as described in Fig. 2. Analysis of TrxFNR eluates by Ponceau red stain (A) and Western blot with the anti‐DnaK antiserum (B) after one, two, three, or five washes (Lanes: 1, 2, 3, and 5, respectively) with TN buffer (‐ATP panel); TN‐ATP buffer (+ATP panel) or GST‐DnaK‐free protein in TN‐ATP buffer in Figure 2 (+ATP + GST‐DnaK‐free panel). (C) The DnaK contamination in the TrxFNR protein eluate subjected to one wash step with TN‐ATP buffer was assumed as 1 (100%). The fit lines are for indicative purposes only and do not respond to any mathematical function. Inset C' shows the DnaK contamination in the TrxFNR eluates obtained in an independent experiment after one wash step with TN‐ATP buffer (Lane 1) or the TN‐ATP buffer supplemented with the GST‐DnaK‐free protein (Lane 2; 1:1 TrxFNR/GST‐DnaK‐free molar ratio).

Figure S2. Analysis of GST‐DnaK‐free protein cross‐contamination in the TrxFNR eluates. Ni‐NTA resin‐bound TrxFNR protein was washed three times with the GST‐DnaK‐free protein (1:1 molar ratio of GST‐DnaK‐free to TrxFNR proteins) and then subjected to different final washing protocols. Three and five washes were performed with TN‐buffer. In the case of three washes, TN‐buffer was also supplemented with imidazole 20 mM. In addition, the effect of incubating the eluted sample with glutathione–agarose resin was analyzed. Finally, the TrxFNR eluates were run in 15% SDS‐PAGE, transferred to nitrocellulose membrane, stained with (A) Ponceau red solution and (B) analyzed by Western blot with anti‐GST antiserum. Lane: GST: purified GST protein (positive control).

Figure S3. Evaluation of the removal of DnaK contamination from the Ni‐NTA bound TrxFNR protein by washes with GST‐DnaK‐free protein in large scale preparations. (A and B) Analysis of DnaK contamination in the TrxFNR eluates (800 μg of TrxFNR protein) after of zero (one wash with TN‐ATP buffer only; control), one, two, three, or five washes steps with the GST‐DnaK‐free protein (0.5:1 molar ratio) in TN‐ATP buffer (lanes washes with GST‐DnaK‐free: 0, 1, 2, 3, and 5, respectively). A and B were performed as described in Figure 2.