In 1975 Prouty et al. first described the formation of dense, amorphous intracellular granules in Escherichia coli cells grown in the presence of the amino acid analog canavanine. These granules, comprising primarily of polypeptide chains, could be solubilized by sodium dodecyl sulfate (SDS) and were not surrounded by any sort of membrane. For several years this observation was considered an aberrant, and rather irrelevant, cellular response induced by growth under nonphysiological conditions. It was not until much later that it became apparent that protein aggregation in vivo is a widespread phenomenon, manifested in cells overexpressing heterologous proteins or native proteins beyond a certain level, and in cells exposed to thermal or other kinds of physiological stress. In addition, mutations resulting in amino acid substitutions, deletions, or insertions can interfere with the folding of a polypeptide to the native state, causing the formation of protein aggregates. Intracellular protein aggregates form dense, electron-refracting particles that can be distinguished readily from other cell components by electron microscopy. For this reason, protein aggregates, at least those observed in microorganisms, are usually called inclusion bodies or, less often, refractile bodies. However, protein misfolding and self-association may occur even when inclusion bodies cannot be detected by electron microscopy or by following careful cell fractionation. Nonetheless, for practical purposes, it is safe to assume that the expression of proteins susceptible to aggregation at a level 2% or greater of the total cell protein will be accompanied by the appearance of readily identifiable inclusion bodies.

Many heterologous polypeptides fail to fold into their native state when expressed in Escherichia coli; instead, they are either degraded by the cellular proteolytic machinery or accumulate in insoluble form, typically as inclusion bodies. Misfolding is a particularly vexing problem in the expression of mammalian proteins, especially those that are composed of multiple subunits, have several disulfide bonds, or contain prosthetic groups. Fortunately, bacteria exhibit a remarkable physiological plasticity that can be successfully exploited to improve protein folding. Significant yields of active heterologous proteins have been obtained through strategies that include the co-expression of homologous or heterologous folding accessory proteins, the optimization of growth conditions, and the use of fusion proteins. A flood of recent reports documenting the successful production of complex eukaryotic proteins in active form have demonstrated that bacteria can provide the proper environment for the folding of the vast majority of recombinant polypeptides.

The enzyme TEM beta-lactamase has been used as a model for understanding the pathway leading to formation of inclusion bodies in Escherichia coli. The equilibrium denaturation of TEM beta-lactamase revealed that an intermediate that has lost enzymatic activity, native protein fluorescence, and UV absorption, but retains 60% of the native circular dichroism signal, becomes populated at intermediate (1.0-1.4 M) concentrations of guanidium chloride (GdmCl). This species exhibits a large increase in bis-1-anilino-8-naphthalene sulfonic acid fluorescence, indicating the presence of exposed hydrophobic surfaces. When TEM beta-lactamase was unfolded in different initial concentrations of GdmCl and refolded to the same final conditions by dialysis a distinct minimum in the yield of active protein was observed for initial concentrations of GdmCl in the 1.0-1.5 M range. It was shown that the lower reactivation yield was solely due to the formation of noncovalently linked aggregates. We propose that the aggregation of TEM beta-lactamase involves the association of a compact state having partially exposed hydrophobic surfaces. This hypothesis is consistent with our recent findings that TEM beta-lactamase inclusion bodies contains extensive secondary structure (Przybycien TM, Dunn JP, Valax P, Georgiou G, 1994, Protein Eng 7:131-136). Finally, we have also shown that protein aggregation was enhanced at higher temperatures and in the presence of 5 mM dithiothreitol and was inhibited by the addition of sucrose. These conditions exert a similar effect on the formation of inclusion bodies in vivo.

The secondary structure of proteins in E.coli inclusion bodies was investigated via Raman spectroscopy. Inclusion bodies were purified from cells expressing different forms of RTEM beta-lactamase and grown at either 37 or 42-degrees-C. All of the solid phase inclusion body samples examined gave amide I band spectra that were perturbed from that of the native, purified protein in both solution and powder forms; secondary structure estimates indicated significant decreases in alpha-helix and increases in beta-sheet contents in the inclusion body samples. The structure estimates for inclusion bodies isolated from 37-degrees-C cultures were similar, regardless of aggregate localization in the E.coli cytoplasmic or periplasmic spaces or beta-lactamase precursor content. Inclusion bodies obtained from 42-degrees-C cells exhibited a further reduction of alpha-helix and augmentation of beta-sheet contents relative to those from 37-degrees-C cultures. These results are consistent with the paradigm for inclusion body formation via the self-association of intracellular folding intermediates having extensive secondary structure content. Further, the overall secondary structure content of inclusion bodies is not significantly affected by subcellular compartmentalization, but may be altered at increased temperatures.

We have determined the macromolecular composition of inclusion bodies formed by overexpressing beta-lactamase from three different expression systems as a function of the growth conditions. The inclusion bodies were purified by differential gradient centrifugation and detergent extraction. Both the expression system and the growth conditions were shown to have a pronounced effect on inclusion body composition. Specifically, contaminating polypeptides ranged from less than 5% to over 50% of the total protein content. Phospholipids composed 0.5-13% of the inclusion bodies. Nucleic acids represented a minor impurity for both cytoplasmic and periplasmic inclusion bodies. Cytoplasmic inclusion bodies of the mature beta-lactamase had the lowest amount of impurities, irrespective of the growth conditions. On the other hand, large amounts of outer membrane proteins and phospholipids were observed in periplasmic inclusion bodies from cells grown at basic pH. Our results show that, at least under some growth conditions, protein aggregation in vivo is highly specific, and the presence of contaminating proteins in inclusion bodies is due to incomplete purification following cell lysis.

R(TEM) beta-lactamase was overexpressed in E. coli from three different plasmids resulting in the formation of cytoplasmic (plasmid pGB1) and periplasmic (plasmids pKN and pJG108) inclusion bodies. Previous work demonstrated that the inclusion bodies differ in structure and composition according to the cellular compartment in which aggregation occurred (17). In this study, we used inclusion bodies purified by sucrose density gradient centrifugation to investigate the effect of the in vivo aggregation environment on the properties of the protein within the inclusion bodies. Guanidine hydrochloride and pH solubilization experiments revealed important differences in die interactions involved in the stabilization of the aggregates. In addition, trypsin digestion results suggested a less ordered protein conformation in periplasmic inclusion bodies. The influence of the inclusion body origin on the renaturation of active protein was investigated in detail. The highest recovery was achieved with periplasmic inclusion bodies from RB791(pJG108). The yield of active beta-lactamase upon refolding the material obtained from solubilized inclusion bodies was between 20% and 40% of that obtained from the renaturation of the purified protein under identical conditions. Our results suggest that the presence of certain contaminants in the inclusion bodies enhance the reaggregation of the protein during the removal of the denaturant.

High levels of expression of the secreted protein RTEM beta-lactamase in Escherichia coli result in the formation of protein aggregates, or inclusion bodies, in the periplasmic space. The formation of inclusion bodies can be inhibited in cells grown in the presence of non-metabolizable sugars such as sucrose. Earlier work has shown that sucrose appears to exert a direct effect on the folding of beta-lactamase within the cell (14). In this study we have investigated the in vitro renaturation and aggregation of beta-lactamase from guanidine-HCl solutions. The equilibrium folding transitions monitored by difference spectroscopy, intrinsic fluorescence and activity measurements exhibited some degree of non-coincidence indicating the existence of at least one stable intermediate. Denaturation was not fully reversible at protein concentrations in excess of 4 mg/ml and resulted in the formation of visible aggregates. The extent of aggregation was greater when the protein was first completely unfolded in buffers containing 2.0 M or higher concentrations of guanidine-HCl, or in the presence of 5 mM dithiothreitol. Addition of sugars caused a shift of the equilibrium curves to higher guanidine-HCl concentrations and prevented the aggregation of beta-lactamase upon refolding. These results are consistent with the aggregation inhibition observed in vivo in the presence of saccharides.

High levels of expression of the secreted protein RTEM β-lactamase in Escherichia coli result in the formation of protein aggregates, or inclusion bodies, in the periplasmic space. The formation of inclusion bodies can be inhibited in cells grown in the presence of nonmetabolizable sugars such as sucrose. Earlier work has shown that sucrose appears to exert a direct effect on the folding of β-lactamase within the cell (14). In this study we have investigated the in vitro renaturation and aggregation of β-lactamase from guanidine-HCl solutions. The equilibrium folding transitions monitored by difference spectroscopy, intrinsic fluorescence and activity measurements exhibited some degree of non-coincidence indicating the existence of at least one stable intermediate. Denaturation was not fully reversible at protein concentrations in excess of 4 mg/ml and resulted in the formation of visible aggregates. The extent of aggregation was greater when the protein was first completely unfolded in buffers containing 2.0 M or higher concentrations of guanidine-HCl, or in the presence of 5 mM dithiothreitol. Addition of sugars caused a shift of the equilibrium curves to higher guanidine-HCl concentrations and prevented the aggregation of β-lactamase upon refolding. These results are consistent with the aggregation inhibition observed in vivo in the presence of saccharides.