We initially observed that recombinant MBP-CpFAS-ENR1 purified by conventional amylose-resin chromatography was able to specifically oxidize NADH (but not NADPH) even in the absence of the substrate crotonoyl-CoA (Figure 1A). This activity was not observed in the MBP-tag control groups, indicating that the activity was not a result of self-oxidation of NADH in the solution. The presence of crotonoyl-CoA at varied concentrations had no effect on the activity either, suggesting that the electrons released from NADH were not transferred to the substrate (Figure 1B). These preliminary observations indicated that the CpFAS-ENR1 domain expressed alone as a single protein was unable to use crotonoyl-CoA as a substrate (or the activity was extremely low), although it was functional when expressed together with other enzymatic domains in the modules (5). The unexpected NADH oxidation activity could originate from either an unreported biochemical feature of the recombinant CpFAS-ENR1 protein or from some copurified contaminants of bacterial origin.
To test these two possibilities, we first examined whether other MBP-fused proteins purified by conventional amylose resin chromatography also possessed this NADH oxidation activity. These proteins (CpACBP1, CpPKS-AL1, and CpTE1) were selected because they were known not to use NADH as a cofactor and could serve as authentic negative controls. To our surprise, all three recombinant proteins, together with MBP-CpFAS-ENR1, possessed the NADH oxidation activity at similar levels (Figure 2A). The artifactual activity was not only detected from fusion proteins expressed in Rosetta 2 cells, but also in BL21 cells (Figure 2B), suggesting that this unwanted NADH oxidation activity is an artifact associated with some conventionally purified MBP-fusion proteins expressed in E. coli.
To further test whether the artifactual NADH oxidation activity was specifically associated with the conventional amylose resin-based purification protocol, we reengineered the CpFAS-ENR1 protein to contain both an N-terminal MBP and a C-terminal His-tag (MBP-CpFAS-ENR1-His). This new fusion, together with CpTE1 that had already contained a C-terminal His-tag, could be purified by both amylose resin-based and Ni-NTA resin-based protocols. Subsequent biochemical assay clearly showed that NADH oxidation activity was undetectable (or at most at near background levels) in both proteins purified by Ni-NTA resin protocol, but detectable again in the same proteins purified by conventional amylose resin protocol (Figure 3A).
These observations confirmed that the artifactual NADH oxidation activity among MBP-fusion proteins originated from a yet undefined bacterial contaminant(s) copurified using the conventional amylose resin-based purification protocol. The problem was likely associated with foreign gene expression in E. coli, because the MBP-tag alone (which is of bacterial origin) that was expressed and purified under the same conditions displayed no such activity (Figures (123). The presence of bacterial contaminants in fusion proteins is not uncommon. Some contaminants are simply nonspecific proteins, while others may be specific proteins associated with the processing of fusion proteins in bacteria during expression, such as DnaK and flagellin (17, 18).
The conventional amylose resin-based purification protocol used relatively mild wash and elution conditions, which might not be strong enough to remove some minor but highly bioactive contaminants. We thus tested whether increasing the stringency in the amylose resin purification protocol could improve the removal of the contaminating activity. After testing several different combinations of reagents, we observed that the addition of 0.1% Triton X-100 and 2% glycerol in the column and elution buffers could actually eliminate the contaminating NADH oxidation activity (Figure 3B). The NADH oxidase activity appeared to be related to a very minor protein contaminant(s) that could not be easily identified by standard SDS-PAGE analysis, as the protein band patterns were virtually indistinguishable between modified and conventional protocols (Figure 3B, inset). We also tested whether the addition of Triton X-100 in the column and elution buffers could affect the native enzyme activity using MBP-CpTE1 as an example. The result showed that although Triton X-100 at either 0.01%, 0.05%, or 0.1% concentrations could eliminate the contaminating NADH oxidation activity (Figure 4 A), it also reduces CpTE1 activity by 15.3%, 23.8%, and 74.8%, respectively. When Triton X-100 is undesirable in the final preparations, the detergent could be removed by adding a second wash step with regular column buffer, followed by elution of fusion proteins with regular elution buffer (Figure 5). The fact that NADH oxidation activity was undetectable in the final preparation lacking detergent indicates that the contaminants were truly removed, rather than being suppressed by Triton X-100 (Figure 5, compare group 3 with group 4).