Co-production of GroELS discriminates between intrinsic and thermally-induced recombinant protein aggregation during substrate quality control.

BACKGROUND: The effects and effectiveness of the chaperone pair GroELS on the yield and quality of recombinant polypeptides produced in Escherichia coli are matter of controversy, as the reported activities of this complex are not always consistent and eventually indicate undesired side effects. The divergence in the reported data could be due, at least partially, to different experimental conditions in independent research approaches.
RESULTS: We have then selected two structurally different model proteins (namely GFP and E. coli β-galactosidase) and two derived aggregation-prone fusions to explore, in a systematic way, the eventual effects of GroELS co-production on yield, solubility and conformational quality. Host cells were cultured at two alternative temperatures below the threshold at which thermal stress is expected to be triggered, to minimize the involvement of independent stress factors.
CONCLUSIONS: From the analysis of protein yield, solubility and biological activity of the four model proteins produced alone or along the chaperones, we conclude that GroELS impacts on yield and quality of aggregation-prone proteins with intrinsic determinants but not on thermally induced protein aggregation. No effective modifications of protein solubility have been observed, but significant stabilization of small (encapsulable) substrates and moderate chaperone-induced degradation of larger (excluded) polypeptides. These findings indicate that the activities of this chaperone pair in the context of actively producing recombinant bacteria discriminate between intrinsic and thermally-induced protein aggregation, and that the side effects of GroELS overproduction might be determined by substrate size.

Introduction

Recombinant protein production is a leading methodological platform of biotechnology and biomedicine [1]. Protein misfolding, degradation, aggregation, inclusion body formation and low functional protein quality are among the main obstacles encountered when using conventional bacterial hosts such as Escherichia coli [2-4] as cell factories, these events being specially distressing when proteins are intended for therapeutic uses [5,6]. Apart from shifting to a growing number of alternative hosts, what might eventually improve protein yield and quality, several strategies have been implemented to favour recombinant protein yield and quality in bacteria, specially addressed to enhance proteolytic stability and minimize aggregation [3,7]. As the activities of the cell's quality control were identified ([8], and references therein), the co-production of chaperones along with the target protein was taken as a routine approach. Firstly individual chaperones, and later chaperone-co-chaperone pairs or larger chaperone sets, supplied from additional plasmid vectors, have been assayed as folding modulators under different experimental setting ups and for a large catalogue of proteins [9]. Despite reports of relevant successes, a consensus about final doses and composition of the best chaperone catalogue has not been reached, as the positive effects of this strategy seem to be highly dependent, among other potential factors, on the specific recombinant protein. In addition, externally supplied chaperones might show side effects on the whole physiology of recombinant cells and eventually compromise the quality, solubility or yield of the target recombinant protein [10,11]. Such undesired effects might limit the success of this method, and their occurrence can account, at least partially, for the lack of consistency of chaperone co-production as a generic strategy.

DnaK, a key negative regulator of the quality control system and the main cytosolic E. coli chaperone (acting together with its co-chaperone DnaJ) is a common component of the most successful chaperone sets [9,12]. However, an excess of DnaK is recognized to promote functional inactivation of the target protein and eventually its massive proteolysis, at least in given experimental setting ups [13-15]. Protein degradation mediated by DnaK seems to be mechanistically connected to the disaggregation process on the surface of inclusion bodies (where DnaK accumulates), in which by DnaK-DnaJ, IbpA-IbpB and ClpP participate [16]. DnaK-mediated degradation of recombinant polypeptides might represent a physiological replica of σ32 inactivation, through its DnaK-mediated delivery to the protease La during the regulation of the heat shock circuits [17]. Recombinant protein production in DnaK knock out mutants, in contrast, results in high yields of recombinant protein and larger inclusion bodies [18].

The activities of GroEL, the other main cytosolic chaperone in E. coli, on recombinant protein solubility, but specially their potential side effects on protein quality are much less known. This might be due to the fact that GroEL knock out mutants are not viable [19], and that studies on partially inactivating groEL mutations [20] or truncated versions of the chaperone [21] might render non conclusive results. GroEL co-production has been shown to enhance solubility and/or yield of different proteins, at very different extents (although sometimes with no perceivable or very mild effects) and under different production conditions [22]. GroELS, apart from acting in the sequential folding steps of nascent polypeptides [23-26], is a holding chaperone that can also promote natural and recombinant protein stability and the formation of large macromolecular complexes such as tobacco mosaic virus-like particles [27] and fully assembled P22 phage virions [28], by preventing aggregation and proteolysis of thermolabile intermediates. However, the effects of these chaperones on recombinant protein yield and functional quality are extremely difficult to summarize since diverse authors focus on different analytical parameters, with alternative criteria and under diverse experimental conditions. Also, the potential contribution of thermally induced aggregation of intrinsically aggregation-prone proteins in GroELS activities has been essentially neglected. Beside these findings and considerations, independent observations have again suggested instability of target proteins as potential, undesired effects associated to GroEL overproduction [10], in special, proteolysis of proteins deposited as inclusion bodies [29].

Results

To examine in a defined experimental setting up, how GroEL and its co-chaperone GroES can affect the yield, stability and quality of aggregation-prone polypeptides, and to identify mechanistic roles of this chaperone pair in the quality control of recombinant proteins we have selected as models two structurally dissimilar soluble proteins (namely GFP and E. coli β-galactosidase). These polypeptides were produced as non-fused, pseudo wild type versions or joined to a hydrophobic viral protein (the VP1 capsid protein of foot-and-mouth disease virus) that acts as an efficient aggregation tag. While GFP is a monomeric small protein (27 kDa) that gains activity through a maturation process in the last folding steps [30], β-galactosidase is a huge tetrameric protein (470 kDa) that becomes active once the correct contacts between the monomer's interfaces have been stabilized [31]. Point mutations in GFP have a dramatic impact on the aggregation rate of the protein [32] while disposition of both GFP (and related fluorescent proteins) and β-galactosidase as inclusion bodies is also sensitive to the amino acid sequence and position of end-terminal fused aggregation tags [33-35]. Aggregation-prone versions of both proteins have been shown to interact in vivo with GroEL [36-38].

To minimize the contribution of thermal stress on protein deposition we tested recombinant protein production at two suboptimal growth temperatures, namely 16°C and 27°C. The non-fused GFP was fully soluble when produced in E. coli at 16°C (Figure 1A). However, around 40% of recombinant GFP occurred in the insoluble cell fraction when the culture temperature was up-set to 27°C (Figure 1A). Co-production of GroELS did not improve GFP solubility in any of these production conditions (Figure 1A). Also, GFP yield was very similar in all cases, excluding the possibility of extended proteolysis mediated by the chaperones. In contrast, the aggregation-prone version of the same protein (VP1GFP) was mainly insoluble at both 16°C and 27°C (Figure 1B), indicating an intrinsic rather than environmental-induced trend to deposition mediated by the viral tag. The yield of VP1GFP at 27°C was significantly lower than that at 16°C, suggesting proteolysis of this protein. However, and in contrast with GFP, the co-production of GroELS dramatically enhanced the yield of VP1GFP (Figure 1B). Furthermore, VP1GFP solubility, expressed as the ratio of protein occurring in the soluble fraction over total protein, was not dramatically enhanced by the chaperones at 27°C (Figure 1B). The five-fold increase in solubility observed at 16°C might be a numerical rather than a factual issue linked to data scattering at low yield values.

Figure 1

Influence of GroELS co-production on yield and solubility of native and engineered GFPs. Yield of total (T) and soluble (S) GFP (A) and VP1GFP (B) produced alone or along with the chaperones GroELS (+). Only significant differences between relevant data pairs are indicated as *, 0.01

Influence of GroELS co-production on yield and solubility of native and engineered GFPs. Yield of total (T) and soluble (S) GFP (A) and VP1GFP (B) produced alone or along with the chaperones GroELS (+). Only significant differences between relevant data pairs are indicated as *, 0.01