Construction of bacterial expression plasmid
Plasmid pMBP-DGAT1-His was engineered to express the full-length Vernicia fordii (tung tree) DGAT1 in an E. coli protein expression system. The recombinant protein MBP-DGAT1-His (rDGAT1) contained a MBP (maltose binding protein) at the amino terminus and 6 histidine residues (His) at the carboxyl terminus (Figure 2). A PreScission protease cleavage site was engineered between the MBP and DGAT1 fusion partners. The "Methods" section describes the details of the plasmid construction.
Degradation of recombinant DGAT1 in the soluble fraction of E. coli
The expression of rDGAT1 was induced by IPTG in E. coli strain BL21(DE3) and detected by immunoblotting using anti-MBP-hTTP and anti-MBP-mTTP polyclonal antibodies, which were raised in rabbits against purified recombinant human and mouse TTP proteins fused to MBP [2, 3]. These antibodies have been shown to react with MBP and MBP fusion proteins with high specificity [2, 3].
Anti-MBP-mTTP antibodies detected a number of bands ranging from approximately 40 kDa (corresponding to that of MBP, 42 kDa) to 60 kDa and a trace amount of protein corresponding to the full-size rDGAT1 (909 amino acid residues, 102 kDa) (Figure 3A). There was no significant difference in protein expression levels among different starting colonies of the bacterium. The expression levels, judged from the overall full-length and degradation products, were decreased under longer expression time tested (Figure 3A). Less but similar sizes of the full-length and degradation products were detected in the cells without IPTG induction (Figure 3A), indicating a high basal level expression of the recombinant protein under uninduced conditions. The 60 kDa band corresponds to protein smaller than the 80-kDa-sized MBP-mTTP fusion protein (a positive control for the antibodies in lane 2 of Figure 3A). Recombinant DGAT1 expression was not significantly different when protein expression was induced at 25°C or 37°C, nor under culture medium with or without 0.2% glucose (data not shown). Similar results were obtained using anti-MBP-hTTP antibodies and the commercial anti-MBP antibodies (data not shown). In agreement with the immunoblotting results shown in Figure 3A, a protein staining gel showed that no distinct band with a molecular mass corresponding to the full-length rDGAT1 was observed in the 10,000g supernatant from the uninduced and IPTG-induced cells (Figure 3B).
As a control experiment, plasmid pMAL-c2X vector was transformed into the same type of E. coli. Like rDGAT1 expression in the uninduced cells described above, MBP protein was clearly produced in the uninduced cells (Figure 3C). In contrast to the minimal expression of rDGAT1 in E. coli (Figure 3A and 3B), MBP was massively induced by IPTG in the same type of cells (Figure 3C), which was specifically recognized by the commercial anti-MBP antibodies [1] and the same anti-MBP-hTTP antibodies [2] and anti-MBP-mTTP antibodies [3] used in the current study. These results suggest that rDGAT1 was expressed well but was extensively degraded in E. coli under the experimental conditions.
Localization of recombinant DGAT1 in the insoluble fraction and membranes of E. coli
Amylose resin affinity chromatography was used to identify the trace amount of the full-length protein as rDGAT1 shown in Figure 3. Based on the expression studies described above, DGAT1 fusion protein expression in E. coli was scaled up. The 10,000g supernatant was applied onto an MBPTrap HP column. After extensive washing, the bound proteins were eluted with 20 mM maltose. However, no clear protein peak was eluted with maltose as judged from UV absorbance at 280 nm (data not shown). Immunoblotting confirmed that the great majority if not all of the immuno-reactive proteins did not bind to amylose resin and was enriched in the unbound fractions (Figure 4A).
To investigate if the recombinant protein was in the insoluble fraction, the 10,000g pellet was sonicated extensively in the same homogenization buffer. This supernatant was used for MBPTrap column purification. The great majority of the "released" recombinant DGAT1 did not bind to the column (as in the case with the soluble rDGAT1) and was mainly recovered in the unbound fractions (Figure 4B). However, immunoblotting showed that the full-length rDGAT1 was much more enriched in the pellet than in the soluble fraction of E. coli and some full-length protein was eluted from the column (Figure 4A versus 4B, right lanes).
To localize the trace amount of the recombinant protein in the supernatant shown in panel A, the 10,000g supernatant was centrifuged at 100,000g. Immunoblotting showed that rDGAT1 was mainly associated with the membrane fraction and only a smaller fraction of rDGAT1 in the cytosol (Figure 4C).
Figure 4A and 4B showed that rDGAT1 bound to amylose resin poorly because little rDGAT1 was recovered in the eluted fractions. To provide a positive control for amylose resin affinity purification, E. coli extract with overexpressed MBP shown in Figure 3C was applied to amylose resin. Significant amounts of MBP bound to the affinity resin and was purified to near homogeneity by the affinity chromatography, although lots of the recombinant MBP did not bind to the same affinity beads (Figure 4D). These results suggest that rDGAT1 may be folded in a way to prevent MBP fusion partner from binding to the affinity resin.
Purification of recombinant DGAT1 with Ni-NTA affinity chromatography
Ni-NTA beads were used to test the purification of the recombinant protein using the 10,000g supernatant. The bound proteins were eluted with successively increasing imidazole concentrations ranging from 50 to 1000 mM. SDS-PAGE showed that the purified fractions contained a number of proteins as shown by Coomassie blue staining (Figure 5A). The majority of the bound recombinant protein (as detected by immunoblotting) was eluted with 200-250 mM imidazole in the elution buffer and no recombinant protein was detected in the washes (Figure 5B). rDGAT1 only partially bound to Ni-NTA beads because a significant amount of the full-length rDGAT1 was recovered in the unbound fraction (data not shown). As expected, MBP and most of the degradation products were not bound to the Ni-NTA beads because the immunoreactive bands on the blot corresponded to the full-length rDGAT1 (Figures 3, 4 vs. 5B). A protein band with approximately twice the size (>150 kDa) of the full-length rDGAT1 was seen in the eluted fractions on the immunoblot after SDS-denatured gel separation (Figure 5B). A similar immunoreactive band with the higher molecular mass was also observed on other immunoblots (see below).
Purification of recombinant DGAT1 with tandem Ni-NTA and amylose resin affinity chromatography
The proteins eluted by 250 mM imidazole from Ni-NTA affinity beads (Figure 5A) were pooled and centrifuged at 10,000g. The supernatant was loaded onto an amylose resin affinity column. FPLC chromatogram showed that the great majority of proteins were washed off the column and little protein was bound to the MBPTrap column (data not shown), similar to those observed using the 10,000g supernatant (Figure 5A). Coomassie blue staining showed that the additional affinity purification step did not improve the purity (Figure 5C). Immunoblotting showed that part of the recombinant protein was precipitated (Figure 5D, lane P vs. line S) and rDGAT1 was detected in the unbound fractions (Figure 5D) but undetectable in the eluted fractions (data not shown).
SDS solubilization and purification of recombinant DGAT1 from insoluble fraction
Immunoblotting showed that the great majority of rDGAT1 was recovered in the 10,000g pellet (Figure 4B). Therefore, SDS was used to solubilize rDGAT1 from 10,000g as well as the 25,000 g pellet fractions followed by purification with Ni-NTA affinity chromatography. Immunoblotting showed that full-length rDGAT1 was barely detectable in the 25,000 g supernatant and the wash (Figure 6A). Some full-length rDGAT1 and several degraded protein bands were detected in the unbound fraction and the major protein band was detected in the eluted fraction (Figure 6A). The full-length rDGAT1 was stable in cells under IPTG induction for 2-24 h as demonstrated in the purified fractions from the SDS-solubilized pellet (Figure 6B).
Detergent and urea solubilization and purification of recombinant DGAT1 from insoluble fraction
As the results indicated that rDGAT1 was associated with the insoluble pellet, an attempt was made to solubilize the recombinant protein from the 10,000g pellet with seven different detergents (Brij 35, CHAPS, NP-40, SDS, Triton X-100, Tween 20 and Tween 80) and urea followed by purification with Ni-NTA affinity chromatography. Immunoblotting showed that SDS was the most effective detergent for rDGAT1 solubilization, while only small amounts of the full-length rDGAT1 was detected in the solubilized fractions by other detergents (Figure 7A, lanes 2-7). The optimal concentrations for SDS solubilization were 0.3-1% (Figure 7A, lanes 11-13). Ni-NTA affinity purification showed that solubilization by both SDS and Triton X-100 resulted in the highest yields of full-length rDGAT1 (Figure 7B, lanes 5-6 and 10-13). Therefore, various concentrations of Triton X-100 were tested for the solubilization of rDGAT1 from the pellet. Immunoblotting showed that 0.3-1% Triton X-100 were the optimal concentrations for the extraction (Figure 7C, lanes 2-8) and purification (Figure 7D, lanes 2-8). Urea at 4 M and 6 M also solubilized the fusion protein to a significant level (Figure 7C, lanes 9-12). However, Ni-NTA affinity chromatography indicated that rDGAT1 from urea-solubilized samples bound to the affinity beads poorly because much less of the full-length protein was obtained in the urea-solubilized protein solution (Figure 7D, lanes 9-12). Silver staining of the purified fractions indicated multiple proteins co-purified with the recombinant protein (data not shown). Extensive efforts were directed to purify rDGAT1 from the Ni-NTA purified proteins following SDS and Triton X100 solubilization. However, the protein could not be purified to near homogeneity by amylose resin affinity, Superose 12 size exclusion, and Mono Q anion exchange chromatography (data not shown).
PreScission protease digestion of recombinant DGAT1
Immunoblotting results showed that MBP-TTP antibodies detected multiple protein bands from soluble fraction (Figure 3), insoluble fraction (Figure 4B), membrane fraction (Figure 4C), Ni-NTA and amylose resin affinity-purified fractions (Figure 5), and detergent- and urea-solubilized fractions (Figures 6 and 7). To confirm the identity of the full-length rDGAT1, PreScission protease digestion was performed using rDGAT1 purified by Ni-NTA affinity chromatography (Figure 5A) because a PreScission protease cleavage site was engineered between MBP and DGAT1 fusion partners (Figure 2). The rDGAT1 sample was digested by the protease overnight followed by centrifugation at 10,000g. Immunoblotting showed that a protein band with the size of MBP was detected in the soluble fraction but not in the undigested protein sample (Figure 8, lane 1 vs. lane 2). It also showed that the great majority of rDGAT1 was recovered in the pellet following digestion (Figure 8, lane 2 vs. lane 3). These results confirmed the identity of rDGAT1 and showed extensive precipitation of rDGAT1 following purification.