2Division of Craniofacial Medicine, Department of Pediatrics, University of Washington, Seattle, WA, USA
We describe a rapid, simple, and efficient method for recovering glutathione S-transferase (GST)- and His6-tagged maltose binding protein (MBP) fusion proteins from inclusion bodies. Incubation of inclusion bodies with 10% sarkosyl effectively solubilized >95% of proteins, while high-yield recovery of sarkosyl-solubilized fusion proteins was obtained with a specific ratio of Triton X-100 and CHAPS. We demonstrate for the first time that this combination of three detergents significantly improves binding efficiency of GST and GST fusion proteins to gluthathione (GSH) Sepharose.
Proteins are usually engineered to be overex-pressed in Escherichia coli as fusion proteins, commonly with glutathione S-transferase (GST) (1), His6 tag (2,3), small ubiquitin-like modifier (SUMO) (4,5), thioredoxin (6), and maltose binding protein (MBP) (3,7). These fusion systems promise increased solubility of target proteins, except for the His6 tag, with single-step purification efficiency. However, many recombinant proteins, especially those of eukaryotic origin, aggregate or become packaged into inclusion bodies (8-11). Refolding recombinant proteins from inclusion bodies can be challenging and yields of correctly folded proteins can be low (2,12).
We observed a number of proteins that readily formed inclusion bodies in E. coli, even with optimized conditions. The problem is amplified when the cells are grown in M9 minimal media, which is required to isotopically label proteins for nuclear magnetic resonance (NMR) and x-ray crystallographic studies. The increased insolubility when using M9 minimal media compared with Luria Bertani (LB) media may be the result of differences in the cellular environments of overexpressed proteins, or may be protein-specific (13,14).
Lysis buffers containing 0.3–2% sarkosyl (Cat. no. 61207-5000; Acros, Morris Plains, NJ, USA) have been used to solubilize GST and other proteins expressed in bacteria grown in LB media (10,11, 15-17), but are less effective when using minimal media. Three cysteine-rich zinc binding domains (RING, Bbox1, and Bbox2) from Midline1 (MID1), a microtubule-associated ubiquitin E3 ligase, readily form inclusion bodies when expressed as GST fusion proteins in M9. We found these typical amounts of sarkosyl were insufficient to solubilize these GST fusion proteins and a related MBP fusion protein, MBP-RBCC (RING-Bbox-coiled-coil domains of MID1) (Figure 1A). Therefore, we tried higher percentages of sarkosyl (up to 10%) in the lysis buffer, and obtained 40–70% of the GST fusion proteins in the solubilized extract based on the intensities of the protein b and following SDS-PAGE. However, we determined that soaking the insoluble pellet (containing essentially 100% insoluble GST fusion RING, Bbox1, Bbox2, or MBP-RBCC) from 5–10 g lysed cells in 2 mL ST buffer (50 mM Tris, 300 mM NaCl, 5 mM ZnCl2, 10 mM β-mercaptoethanol) with 10% (w/v) sarkosyl for 6–24 h effectively and efficiently solubilized >95% of the proteins from the pellet (Figure 1, A and B). We found 10% sarkosyl to be optimal because higher concentrations were too viscous, leading to difficulty in subsequent purification steps. The solubilizing effects of the sarkosyl actually decreased when the concentration of sarkosyl was >10% (data not shown). This simple approach of using 10% sarkosyl was effective in solubilizing at least six different proteins tested, all of which formed inclusion bodies even when fused to His6-MBP and His6 tag. Of note, we observed that the majority of other proteins found in the pellet were also solubilized with 10% sarkosyl (Figure 1, A and B), suggesting that this methodology is broadly applicable.
Even though sarkosyl can solubilize GST fusion proteins, purifying these proteins in the presence of the detergent can be challenging and difficult (12). Consistent with previous reports (10,11,15), we observed that GST and our GST fusion proteins could not be affinity-purified even in 0.3% sarkosyl. To overcome the problem of high sarkosyl concentrations, the 10% sarkosyl–solubilized pellet solution was diluted with the lysate to yield a 2% sarkosyl solution, or to 1% with a variety of common buffers. In each case, solubility of the overexpressed protein was maintained, although at lower sarkosyl concentrations (<<1%), some proteins began to precipitate. Proteins in the soluble extract with 2–10% sarkosyl can be stably stored at 4°C for a week before affinity purification.
To facilitate efficient glutathione (GSH) Sepharose (Cat. no. 17-5132-01; Sigma-Aldrich, St. Louis, MO, USA) affinity purification, 4% Triton X-100 and 40 mM CHAPS were added to the 2% sarkosyl solution (Figure 1C). To the solution that contained 1% sarkosyl, the addition of 2% Triton X-100 and 20 mM CHAPS resulted in similar binding efficiency (data not shown). While this ratio was used in these studies, 3% Triton X-100 and 30 mM CHAPS also worked well (data not shown). All three detergents resulted in significantly greater binding of GST to the GSH Sepharose (Figure 1C). Both Triton X-100 and CHAPS had a greater synergistic effect with sarkosyl than either alone, and neither worked alone to yield soluble GST fusion RING, Bbox1, or Bbox2 proteins.
To understand how these detergents work, we performed kinetic assays with GST (Figure 2A). The activity of purified GST (horse liver; Sigma Aldrich) at a concentration of 0.01 unit/µ1 in phosphate buffer (100 mM KH2PO4, 1 mM EDTA, pH 6.5) was assayed by its conjugation of GSH to 1-chloro-2,4-dinitrobenzene (CDNB), according to manufacturer's specification. The formation of GSH-CDNB adduct was measured by absorbance at 340 nm for 30 min. Detergents, either alone or in combinations of different ratios, were added to fresh assay solution to determine their effect on GST activity. In the presence of 0.3% sarkosyl—the least amount of sarkosyl used in previous reports (10,11, 15-17)—GST was inactive. Addition of 1% Triton X-100 and 10 mM CHAPS to GST with 1% sarkosyl rescued ~10% of the original activity, while 2% Triton X-100 and 20 mM CHAPS regained ~60% of GST activity and 3% Triton X-100 and 30 mM CHAPS yielded ~80% of the original enzymatic activity. Individually, CHAPS and Triton X-100 at these concentrations could only recover ~30–40% of the GST activity. In the control experiment, 1% Triton X-100 and 10 mM CHAPS, without sarkosyl, did not affect GST enzymatic activity. GSH Sepharose binding was also significantly enhanced in the presence of all three detergents with a binding affinity estimated to be three- to five-fold better in the three detergents compared with free GST.
We postulate that the sarkosyl molecules encapsulate proteins and disrupt aggregates. Triton X-100 and CHAPS, with critical micelle concentrations of 0.25 mM and 6–10 mM, respectively, form large mixed micelle or bicelle structures that incorporate sarkosyl molecules from the solution. In doing so, they decrease the apparent concentration of sarkosyl surrounding GST, potentially freeing active sites or facilitating proper protein refolding. The >80% yield of properly refolded proteins is significant when compared with yields obtained with other commonly employed methods.
Two-dimensional 1H-15N heteronu-clear single quantum coherence (HSQC) NMR spectroscopy (Varian Inova 600 MHz spectrometer, VarianInc, Palo Alto, CA, USA) was used to gain further insight into each detergent's mechanism of action. We observed that the 1H-15N signals of 15N-labeled MID1 Bbox2 were weaker in the buffer with 1% sarkosyl compared with those without detergents (Figure 2B). The weaker signals are due to increased resonance line broadening from the slower isotopic molecular tumbling rate of Bbox2, likely the result of being encapsulated by sarkosyl molecules. Viscosity was ruled out as a potential cause because the addition of 2% Triton X-100 and 20 mM CHAPS, which would have increased viscosity, resulted in 1H-15N signal intensities returning to values similar to those of detergent-free Bbox-2 (Figure 2C). The lack of large chemical shift changes that would imply a drastic structural change or collapsed peaks that would indicate unfolding of Bbox2 suggests that the tertiary structure of Bbox2 (a ββα-RING fold with two coordinated zinc ions) was intact in the presence of 1% sarkosyl.
The signals of the 2-D HSQC spectra of GST and MBP showed more collapsed signals in the presence of 1–2% sarkosyl, consistent with a molten globule state but not that of a denatured protein (data not shown). The spectra of these proteins in the presence of 1% sarkosyl, 2% Triton X-100, and 20 mM CHAPS were similar to natively folded GST and MBP. It is important to note that proteins denatured by urea, guanidine hydrochloride, or heat could not be refolded with just these three detergents. An intrinsically unstructured protein also remained unstructured in the three detergents (data not shown).
As the ingredients for minimal media are relatively expensive and yet essential for isotopically labeling proteins for structural studies, it is important to maximize, in milligrams amounts, the yield of soluble folded protein. We therefore tested the protocol with one His6-tagged and seven His6-MBP fusion proteins, including one that contained two disulfide bonds. Incubation of the His6-tagged FMN/NAD-dependent trehalose oxidoreductase from Sinorhizobium meliloti (His6-ThuB) with either 5% or 10% sarkosyl resulted in >75% (Figure 1D) and >95% soluble protein (data not shown), respectively. Subsequent dilution to 1% sarkosyl enabled efficient affinity purification of His6-ThuB with Ni2+ resin (Cat. no. 30410; Qiagen, Valencia, CA, USA). Similarly, the inter-leukin binding protein with disulfide bonds was also successfully folded in the presence of all three detergents (data not shown). While some of the His6-MBP fusion proteins required the pellets to be incubated with 10% sarkosyl, others were soluble with 1% sarkosyl in the lysis buffer. Even though the His6-MBP fusion proteins could be purified with Ni2+ resin in the presence of 1% sarkosyl, the addition of Triton X-100 and CHAPS increased the binding. Based on our NMR spectra of solubilized protein purified in this manner, we believe it is important to have all three detergents to maximize yields.
We thank Brian Krumm for helpful discussions, and for successfully applying our sarkosyl protocol to his His6-MBP-ILC4S protein. This work was supported in part by the Oklahoma State University Agricultural Experimental Station (Project no. 2527) and the NSF CAREER (no. 0546506) grants.
The authors declare no competing interests.
Address correspondence to Michael A. Massiah, Department of Biochemistry and Molecular Biology, Oklahoma State University, 246 Noble Research Center, Stillwater, OK 74078, USA. email: [email protected]