Modulation of recombinant protein solubility in ClearColi cells
In order to study the performance of APP fused to recombinant proteins in the endotoxin free ClearColi™ expression system, two model soluble proteins; iRFP (near-infrared fluorescent protein) and GFP (green fluorescence protein) were selected as scaffolds and fused to the surfactant-like peptide L6K2, previously described as APP (Fig. 1a) [14].
In transformed ClearColi cells, recombinant H6iRFP protein was equally distributed in both soluble and insoluble cell fractions (Fig. 1b and 1c). As expected, upon L6K2 fusion, an altered distribution pattern was observed, with most of the protein located within the insoluble cell fraction, which suggests an increased aggregation tendency for this fusion protein (Fig. 1b and 1c). The change in solubility pattern was observed within 1 hour of induction and was maintained for up to 5 hours (Fig. 1b). In contrast, the recombinant H6GFPL6K2 was mostly detected in the soluble cell fraction (Fig. 1b and 1c). As a model APP, with an appreciable ability to enhance the aggregation tendency of recombinant proteins, VP1 from the capsid protein of the Foot-and-mouth disease virus [26,27] was fused to GFP (Fig. 1a). As expected, most of the protein signal in the sample was detected in the insoluble cell fraction (Fig. 1b and 1c). These results indicate that the aggregation propensity of a recombinant protein may be modulated by an APP, although the solubility tendency of the scaffold protein may play a role in counteracting this effect (compare the solubility of iRFP versus GFP when fused to L6K2 in Fig. 1b and Fig. 1c). In the case of VP1, the strong aggregation tendency overcame the high solubility of the GFP, whereas GFP solubility was not affected by the addition of L6K2 (compare the solubility of GFP when fused to VP1 or L6K2 in Fig. 1b and 1c).
Impact of APP length on protein solubility in ClearColi cells.
As the ability of L6K2 to reduce solubility of GFP was not significant while it was effective in iRFP, the effect of peptide length in the solubility of GFP was subsequently evaluated. To that end, the H6GFPL6K2 recombinant gene was redesigned into several different variations of length and to affix at the C-terminus of the GFP sequence (Fig. 2a). The aggregating potential of L6K2 peptide was amplified by reiteration of leucine and lysine repeats in different positions (see Table 1) and analyzed by AGGRESCAN software [8]. Selected peptides displayed higher hot spot area (HSA) than the original L6K2 peptide. However, only L12K4 and L18K6 showed increased normalized hot spot area (NHSA) and increased average aggregation-propensity hot spot (a4vAHS).
As previously observed, H6GFPL6K2 was detected in the soluble cell fraction of transformed E. coli cells (Fig. 2b and 2c) and consequently, emitted fluorescence that was homogenously distributed throughout the cytosol (Fig. 2d). The addition of the L6K2 derived peptides had a positive impact in protein aggregation. As expected, the distribution of fluorescence in the transformed cells was detected in protein clusters (IB; Fig. 2d and Additional file 1: Fig S1a). In fact, two different aggregation patterns in the L6K2 derived constructs were detected. The proteins containing serial L6K2 repeats ((L6K2)x2 and (L6K2)x3) were preferentially detected in periplasmic areas around the cells, while the constructs containing longer non-repetitive Leucine/Lysine tracks (L12K4 and L18K6) were detected as fluorescent cellular pole aggregates. Therefore, the serial L6K2 repeats acted both as APPs and periplasm localization signals, deduced from the fluorescence pattern which revealed signal clustering in discrete aggregates on the periphery of the cell cytoplasm. In addition, the aggregation tendency of L6K2 repeats increased with the number of repeats, while L12K4 presented an aggregation pattern similar to the one observed in cells expressing the positive control, VP1GFP. This aggregation tendency was not recorded in Western Blot analysis of the soluble and insoluble cell fractions (Fig. 2b and 2c), indicating that the aggregation modulatory effects of the L6K2-derived peptides may be sensitive to the tested experimental conditions of protein extraction. This was not the case of the aggregation pattern of VP1GFP construct that was perfectly replicated under confocal laser scanning microscopy and Western Blot analysis (Compare VP1GFP data in Fig. 2b, 2c and 2d).
Antimicrobial activity of L6K2-containing recombinant proteins
During recombinant gene expression experiments of L6K2-containing constructs, the growth of transformed E. coli cells was compromised, especially in the case of L18K6 (data not shown). At that point, it was questioned whether or not the peptides were toxic to the cell because of the previously discussed antimicrobial activity. The modeling of the L6K2 derived peptides with PEP-FOLD 3 [31–33] displayed amphipathic alpha helices in all cases (Additional file 1: Fig. S2a). The preferred conformation of the L6K2-containing peptides was maintained in the presence of the PT linker, which has been described as a flexible peptide for separating protein domains (Fig. S2b) [34]. This configuration has been described in naturally produced or synthetic cationic antimicrobial peptides (AMPs) which have been proposed as a potential new class of antimicrobial drugs [35].
However, the production of small peptides is difficult to achieve through recombinant technologies due to reduced stability, and alternative strategies have been utilized to overcome such a hinderance [36]. One possibility to obtain recombinant peptides is through fusion to partner proteins for a potential dual effect on the final product. On the one hand, the reduction of the toxicity of the AMP over the expressing host, and on the other, the improvement in the stability of the peptide in expression systems [37]. However, the study of the resulting biological activity when fused to reporter proteins by genetic engineering has not been explored in depth. Examples of this strategy include the fusion between GWH1 (an AMP with amphipathic alpha helical structure) [38] and GFPH6 (as partner protein) [14,39] and the secretory production of AMP-containing fusion partners [40]. In these studies, the fusion of the AMP to the N-terminus of recombinant proteins preserved the AMP bactericidal activity despite the C-terminus being anchored by the fusion protein. Therefore, the putative antimicrobial activity of the purified soluble versions of H6GFPL6K2 and H6GFP(L6K2)x2 proteins was analyzed and subsequently compared to the data that was obtained from the purified GWH1-GFPH6 in this work. The results indicated that the antimicrobial activity of L6K2-containing recombinant proteins is strain specific (Fig. 3), being comparable to the antimicrobial activity of GWH1 peptide fused to GFP in E. coli cultures (Fig. 3b). In addition, the position of the peptide at each end of the scaffold protein did not appear to be relevant to the antimicrobial activity. The incubation of S. aureus with the proteins containing amphipathic alpha-helices had only a slight effect on cell viability under the tested conditions (Fig. 3a). Interestingly, the antimicrobial activity of the recombinant proteins was completely different when Micrococcus luteus cells were challenged. The addition of the purified proteins had a positive effect on cell viability at lower concentrations while at the highest protein concentration (8 mmol/L) the cell viability dropped drastically (Fig. 3c).
As observed in Figure 3b, the antimicrobial activity of the L6K2-containing constructs was detected in E. coli cultures at low protein concentrations. This mechanism may explain the cell growth inhibition observed in ClearColi cultures transformed with expression vectors with cloned L6K2-derived genes. Small cationic or amphipathic molecules, similar to the ones described in this work, have been described as produced by prokaryotes and eukaryotic organisms as defense against infectious agents. These molecules belong to a non-specific ancient system of innate immunity and they perform their activity through direct interaction with membranes, nucleic acids, proteins or even activate autolysins [41–44]. In the case of membrane interaction, they cause the destabilization of the cytoplasmic membrane by forming pores or by their arrangement parallel to the membrane surface, disrupting the proton motive force and provoking the leakage of vital molecules which leads to cell death. However, even though their mechanism of action is nonspecific, a differential efficacy of the same antimicrobial peptide between Gram-negative and Gram-positive bacteria has been described [45,46]. In the case of Gram-positive bacteria, apart from membrane disruption, the reaction requires further interactions with the cell wall [45].
Pull-down effect on aggregation tendency of H6GFPL6K2
Aggregation of different proteins may be enhanced by the stereospecific interaction of APP in bacteria [26]. Therefore, the aggregation ability of a recombinant protein fused to L6K2 may enhance the aggregation tendency of H6GFPL6K2 when produced simultaneously in cells. For that purpose, a dual expression vector including the gene encoding for H6iRFPL6K2 was generated, which displayed a high tendency to aggregate beside the gene coding for H6GFPL6K2 to be simultaneously expressed. In cells expressing the aggregation prone H6iRFPL6K2 construct and the soluble H6GFPL6K2 construct at the same time, the fluorescence of the GFP shifted from the cytoplasm to polar protein aggregates (IBs) (Fig. 4a). The green fluorescence distribution in expressing cells was similar to the pattern observed when co-expressing VP1GFPH6 and H6iRFPL6K2 (Fig. 4b).
The change in the aggregation propensity of the H6GFPL6K2 seemed to be directed by the pull-down ability of the L6K2 peptide present in the H6iRFPL6K2 construct. The intermolecular interactions between L6K2 present in the two proteins enhances the aggregation tendency of GFP. In the expressing cells, the newly formed H6GFPL6K2, when interacting with H6iRFPL6K2 with a high tendency to aggregate was dragged to the insoluble cell fraction. Therefore, it may be hypothesized that when two different proteins share aggregation prone domains, even if one of the proteins is still soluble, the protein with the highly aggregation propensity may lead the accompanying soluble protein to the insoluble cell fraction through co-expression. However, although the secondary structures of the iRFP and GFP proteins are not similar (Additional file 2: Fig. S3), the effect of the iRFP scaffold protein in the aggregation enhancement of H6GFPL6K2 may not be ruled out. For that reason, a spectral variant of GFP (EBFP2; highly similar in amino acid sequence and secondary structure) was fused to VP1 domain generating VP1EBFP2H6 construct (Additional file 3: Fig. S4). Predictably, when produced recombinantly, this protein was mainly accumulated in the insoluble cell fraction (Additional file 5: Fig. S5).
The distribution of the GFP fluorescence in cells simultaneously transformed with plasmids coding H6GFPL6K2 and VP1EBFP2H6 was homogeneously distributed in the cytoplasm of the cells, in agreement with the data obtained in the expression experiment of H6GFPL6K2 alone (compare the distribution of GFP fluorescence in Fig. 2c, Fig. 5a and Additional file: Fig. S1b). On the other hand, the fluorescence emitted by EBFP2 fused to VP1 in those cells was mainly detected in polar IBs as expected. When VP1GFPH6 was expressed along with VP1EBFP2H6, the GFP fluorescence was located exclusively at the poles of the cells, as IBs (Fig. 5b and Additional file: Fig. S1b). The colocalization analysis of the fluorescence emission from both proteins indicated the preference of H6GFPL6K2 to aggregate in the presence of the same APP (Fig. 5c) ruling out an aggregating role of the scaffold protein in this process. Therefore, this result has a direct application for biopharmaceutical and biotechnological applications through protein engineering. In fact, these protein nanoclusters have been described as a source of soluble active protein obtained upon incubation in non-denaturing conditions [14–16] and have also been administered as biocompatible depots for tumor targeting of therapeutic proteins [17–21]. Furthermore, protein aggregation seems to be a common mechanism described in most of the expression systems [47–49] that opens up the possibility of expanding this type of strategy to proteins that are difficult to produce in prokaryotes. Therefore, the fusion of common APP to different therapeutic recombinant proteins can induce the colocalization of two recombinant proteins in IBs, obtaining protein formulations with potential synergic activities.