GFP has been widely used as fluorescent fusion tag in various applications [42–45]. Although several studies have reported the formation of active IBs for GFP-containing fusion proteins, there is always the involvement of another specific fusion partner rather than GFP to induce the aggregation of the target proteins [1, 5]. GFP fusion has not been considered as an IB-inducing approach under conditions where GFP alone can be expressed as soluble proteins. Our present study, in contrast, demonstrated for the first time that even when GFP is fused with a soluble fusion partner, active IBs can also be formed, mainly due to the aggregation of the GFP moiety. In this context, GFP is demonstrated to have the potential as a novel IB-inducing fusion partner for some well-folded proteins.
In this study, (His)6 tag was used for the easy purification of PhoC-GFP fusion proteins. Although the addition of small peptide to proteins might have IB-inducing effects for some proteins , (His)6 tag has been widely used to facilitate purification, where soluble target proteins can be expressed [31–35]. More importantly, we have confirmed that the addition of (His)6 tag to GFP or PhoC did not lead to the formation of IBs under the same conditions, thus the addition of (His)6 tag to the fusion protein PhoC-GFP is not an important factor that contributes to the formation of PhoC-GFP IBs , but only making the purification of PhoC-GFPs much easier in this study.
The solubilization of IBs by arginine has revealed that GFP in IBs maintains near-native folding, and thus the low GFP fluorescence in IBs should be attributed to the aggregation of GFP moiety [27, 36]. This aggregation has probably led to the formation of PhoC-GFP IBs. Supports for this mechanism can be found in other IB formation researches, where they showed that the intermolecular interaction between folding intermediates is the major cause for IB formation [5, 13, 27, 46].
After arginine was removed from solubilized PhoC-GFP by dialysis, no aggregation was observed for the fusion proteins (data not shown), thus we further speculated that aggregation of GFP moiety probably resulted from the hydrophobic interactions between the hydrophobic patches exposed on the surface of folding intermediates of GFP. It has been reported that fusion can reduce the folding yield and rate of the GFP moiety [25, 46, 47]. Similarly, the GFP domain in PhoC-GFP fusion protein can be reasonably supposed to fold less efficiently and rapidly than non-fusion GFP, presumably prolonging the intermediate-folding time [13, 46]. This also depletes the available molecular chaperones in the cell [48, 49]. Taken together, all these consequently facilitate the aggregation of the folding intermediates of GFP moiety in fusion proteins [39, 50, 51]. For PhoC domain, due to the intrinsic folding characteristics [23, 52], folding can be reasonably expected to be faster than GFP. On the other hand, as PhoC domain is translated prior to GFP domain in the fusion sequence in this study, the available folding time is longer than that of GFP . Therefore, when PhoC-GFP folding-intermediates were incorporated into IB nucleus, PhoC was probably folded well and retain the native or native-like structure in the IBs. This structure is less probable to interact with other PhoC moieties to form aggregates which would harm PhoC functions . This hypothesis can be supported by the fact that change in PhoC specific activities between soluble form and IB form is much smaller than that of GFP (Figure 3), suggesting similar folding state in IBs and in the soluble protein for PhoC, and no obvious aggregation in IBs.
In this study, we also exploited the linker sequence to modulate the activities of IBs for the first time. In fact, besides the improvement of activities shown in Results, we noticed that the effect of linker sequence is even more significant in IBs than in soluble proteins. For example, the differences in specific activities (GFP, phosphatase and phosphotransferase) of soluble fractions between PhoC-R-GFP and PhoC-F-GFP were 15.3%, 8.74%, and 1.61%, respectively, while the differences between their IBs were 30.7%, 10.4%, and 10.0%, respectively (Table 2), showing more significant effects of linker sequence for IBs than for soluble proteins. In addition, for the active IBs, the change of linker shows high statistical significance regarding the specific fluorescence, phosphatase and phosphotransferase activities (α = 0.05, Table 2), respectively. Whereas for the soluble proteins, the change of linker shows less statistical significance, which is in good agreement with our suggestion that linker would have more effect for the aggregated form proteins. In fact, studies on linker have pointed out their role in controlling the conformation of fusion proteins (e.g., the distance and orientation between domains, and folding of domains) [17–20], but mostly focusing on the study of soluble proteins. Regarding the spatial relationship of domains, on which the linker region exerts its influence, the role of linker can be expected to be comparable or even more significant in IBs than in soluble proteins, as the former is a much more crowded environment than the latter [1, 5, 22]. The effects of linker on aggregation of the fusion proteins is probably in the following two aspects: (i) linker sequence is believed to modulate the distance between PhoC and GFP, thus affecting the aggregation of GFP moieties; (ii) on the other hand, the linker sequence itself, which is directly linked to the target protein sequence, would possibly affect the folding of proteins, thus exposing hydrophobic regions susceptible for the aggregation of the fusion proteins. Therefore our results suggest that linker property (flexibility, length, hydrophobicity, etc.) can be a potential way to engineer IBs for desired characteristics of active IBs.
For targeted engineering of active IBs, much more effort would still be needed to study the relationship between IB characteristics and linker properties. For this purpose, a systematic study of IB variants with different linkers by using a novel linker library with widely controllable and traceable flexibility developed by our group is undergoing now, which would provide important clues for the design of IBs, and should be of general importance for their industrial applications.
The partial aggregation of the target protein would result in distribution of enzyme activities among soluble and insoluble fractions, which would probably hinder the activity yield of IBs and their application. However, our results have shown that by using linkers of different flexibility, the distribution of total activities can be altered, with over 80% PhoC activities in IBs for PhoC-R-GFP (Figure 4b). This high PhoC activities in IBs could benefit the total recovery efficiency and the reuse of enzyme by the simplified protein separation via IBs. Further optimization of linker could probably lead to even higher level and ratio of activities in IBs, which is indispensable for the bioprocess application of the active IBs.
With respect to the bioprocess application of active IBs, the stability of the active IBs and enzymatic activities is of general importance. In fact, IBs have been reported to be stable [53, 54]. Bioprocess studies on some active IBs have demonstrated the feasibility of repeated use of active IBs as biocatalysts [1, 55–58], indicating good stability of the active IBs and enzymatic activities. Systematic studies on the stability of active IBs in real bioprocess conditions were also reported [16, 59, 60]. Taken together, further elucidation of the effects of linker properties on the stability of IBs and enzymatic activities will be indispensable for application of IBs in bioprocesses.
In addition, growth conditions and medium composition are also important parameters for the formation of active IBs. To examine the stability and maintenance of the activity of IBs for bioprocess application, further study on the effects of culture conditions by using more systematic experiment design such as fed-batch cultivation should be carried out in the next step.