Protein Refolding

The expression of proteins in a host cell has become the key technique which has driven vast increases in the understanding of protein structure and enzyme kinetics and mechanisms, through the availability of large amounts of soluble proteins from easy to culture organisms. Although the information necessary for a protein to fold into its native conformation is present in the primary structure of proteins (Anfinsen, 1973), the over-expression of recombinant proteins inE. coli often results in the accumulation of the recombinant protein in structures known as inclusion bodies, which contain a highly concentrated, non-soluble form of the protein. To generate soluble protein for the study of protein structure, folding and enzyme kinetics it would be desirable to obtain soluble, correctly folded protein from inclusion bodies, since they represent a high purity, easy to isolate and relatively inexpensive source of recombinant protein. To do this, it is necessary to

disaggregate the inclusion bodies into separated protein chains, and then induce these chains to adopt the correct conformation. This is a difficult problem and has been

solved by different methods for different proteins. To understand the difficulties in this procedure we must first understand how the inclusion bodies are formed, what is contained in inclusion bodies as well as their nature.

1.2.1 The Formation of Bacterial Inclusion Bodies

Bacterial inclusion bodies are formed on the over-expression of some

recombinant proteins inE. coli. Inclusion bodies can also be formed naturally under certain forms of cell stress, e.g. heat shock, but these types of inclusion bodies are not considered here. The formation of inclusion bodies for these proteins is dependent on the temperature of expression, or on the rate of expression, as

determined by the concentration of the inducer of expression. Inclusion bodies can be visualized inE. colicells by light microscopy. Inclusion bodies are refractile under phase contrast microscopy and grow continuously during the expression of the recombinant protein (Carrióet al., 1998). Analysis of the size of inclusion bodies, and the amount of protein accumulating in the inclusion bodies indicates that the density of protein in the inclusion bodies does not change through the process of the building of inclusion bodies. That is, the conformation of the protein in the inclusion bodies is the same regardless of the size of the individual inclusion bodies or the length of the expression of the recombinant protein (Carrióet al., 1998).

The bacterial inclusion bodies that are formed on the over-expression of recombinant proteins often contain more additional proteins as well as the recombinant protein, reducing the purity of the inclusion bodies. Several host

proteins have been shown to be present in purified inclusion bodies in addition to the recombinant protein. The building of inclusion bodies of hFGF-2 includes the

elongation factor Tu (EF-Tu). The chaperones DnaK and DnaJ were also found in hFGF-2 inclusion bodies in addition to certain metabolic enzymes fromE. coli namely LpdA, GatY and TnaA (Rinaset al., 2007). GroEL/ES has also been found in some bacterial inclusion bodies. The recombinant protein usually, however, accounts for more than 80% of the total protein in the inclusion bodies (Carrióet al., 1998), making inclusion bodies a high purity source of recombinant protein.

In early studies of bacterial inclusion bodies, it was assumed that the inclusion bodies represented a disordered aggregate of mis-folded proteins. However, more recent studies on inclusion bodies have revealed that this view of

proteins present in bacterial inclusion bodies using Fourier Transform Infrared (FT- IR) spectroscopy. In studying the secondary structure of human growth hormone (h-

GH) and human interferon-α 2b (IFNα-2b) inclusion bodies it was determined that a

significant amount of secondary structure was present. This secondary structure

resembled the native secondary structure. However, intermolecular β-sheet bridges

were also present, which are responsible for the insoluble and densely packed nature of the inclusion bodies. The secondary structure component of the purified inclusion bodies was also different from the secondary structure component of thermal

aggregates of the proteins studied. Other studies on bacterial inclusion bodies have established that the inclusion bodies of fluorescent proteins can fluoresce, indicating that these inclusion bodies contain proteins that are mostly folded (Garcia-Frutoset al.,2005). In addition, inclusion bodies of over expressed enzymes have been shown to retain enzymatic activity (Tokatlidiset al., 1991).

Bacterial inclusion bodies are often considered to be inert protein aggregates that grow by accumulation of mis-folded protein and may shrink by proteolysis of the inclusion body proteins. Carrió and Villaverde (2000) showed, however, that bacterial inclusion bodies are dynamic. Exploiting an expression system using a temperature sensitive phage repressor (cI857 phage repressor) to control expression

they were able to express β-galatosidase and P22-tailspike polypeptide as inclusion

bodies and subsequently halt expression whilst continuing the culture by means of a temperature shift. When this was performed, the amount of soluble protein increased, despite no further expression of the protein being performed. This indicates that the inclusion bodies formed were being disintegrated by cellular machinery and protein refolded.

This has given rise to a different view of inclusion bodies proposed by Ventura and Villaverde (2006) and Villaverde and Carrió (2003). This view understands inclusion bodies to be dynamic structures with the protein contained being conformationally diverse. They propose that inclusion bodies do not contain only mis-folded protein which is trapped in its mis-folded state, but rather, are transient reservoirs of protein which is under the quality control mechanisms of the bacterial cell, namely chaperones and proteases. The accumulation of proteins in inclusion bodies is driven by the balance between the translation of the protein from the mRNA and the speed and chaperone requirements of the folding of the

cell to fold the protein, the excess partially folded protein will be stored in inclusion bodies to be folded later, when the production of the protein has lowered or ceased. There are no defined sequence features that lead to the accumulation of protein in inclusion bodies, as evidenced by similar mutations in different proteins resulting in different behaviour of the proteins regarding their accumulation as inclusion bodies.

1.2.2 The Refolding Problem

The accumulation of recombinant protein as inclusion bodies presents an easy to access source of high purity protein. To make use of this source of protein however, the protein chains must be separated, and the protein induced to fold into its native conformation. Many proteins have been refolded from inclusion bodies for different purposes. Owenet al., (1995) refolded the kinase domain of phosphorylase kinase from inclusion bodies, and crystallised the refolded protein to obtain the structure of the phosphorylase kinase kinase domain.

Some proteins for therapeutic use are produced by refolding bacterial inclusion bodies, for example heterodimeric platelet-derived growth factor (Müller and Rinas, 1999). When refolding inclusion bodies on a larger scale like that used for the production of therapeutic proteins, studies by Mannallet al.(2007) have

confirmed that the rate of dilution of the inclusion bodies from the unfolded state into the renaturation conditions used is key for obtaining high yields of refolded protein.

A survey of the renaturation conditions that have been reported in the REFOLD database (Chowet al., 2006), indicates that the conditions which support the refolding of different proteins vary widely. This has lead to the adoption of refolding screens to rapidly identify the conditions that promote the refolding of the protein. Commercial screens have also been produced, such as the Novagen iFold screen. Commercial refolding additives have also been produced, with the aim of improving the refolding of most proteins from inclusion bodies, such as the novexin reagents. These refolding screens are, however, usually broad, including several different reductant, and also including redox couples, such as reduced and oxidized glutathione. Since the commercial screens include four or more different pHs, and both reducing conditions and redox couples, the number of additives that is examined is necessarily lower. The typical refolding volume is also low, usually 1

protein is low the manufactures of these screens usually recommend that a sensitive activity assay is used to assess the extent of refolding. This is problematic for kinases however, which typically require activation by an activating kinase, or other

regulatory partners for full activity. The combination of the requirement for an activity assay which is difficult for kinases in this format, the low coverage of the different types of additives, and the low amounts of refolded protein available for readout from the screen make these commercial screens unsuited to the identification of conditions for the refolding of protein kinases.

1.2.3 The Preparative Refolding of Proteins

Methods for supporting the refolding of proteins from inclusion bodies are widely variable. Firstly the methods employed for the disaggregating of the inclusion bodies are different depending on the nature of the protein that is being refolded. Inclusion bodies can be disaggregated by the use of chaotropic denaturants such as guanidine or urea or detergents such as SDS. Physical means of disaggregating inclusion bodies such as temperature or pressure can also be used, although these methods are usually used with small quantities of protein and are difficult to scale to larger refolding experiments.

Once the inclusion bodies have been disaggregated, the denatured protein produced may be purified in the denatured state if a purification tag is present which functions in the denatured state. Typically poly-histidine tags are used. Physical means of disaggregating inclusion bodies do not allow for the purification of the protein of interest from the contaminant proteins found in disaggregated inclusion bodies. The use of high purity denatured protein for refolding experiments is expected to improve the recovery of correctly folded protein (Bataset al.,1999).

To refold the protein from the disaggregated state the conditions which maintain the denatured, disaggregated state need to be removed to allow the protein to refold. When physical means have been used to disaggregate the inclusion bodies, the protein is refolded by the removal of the pressure or temperature change, in either a single step, in a series of steps or in a slow gradient. Where chemical denaturation has been used several strategies have been employed to allow and support the refolding of the disaggregated inclusion body proteins. Where a poly- histidine tag is present on the protein to be refolded, the denatured protein can be bound to an IMAC column and the chemical means of denaturation removed by

changing the buffer the column is equilibrated in. This method was used by Ryuet al.(2008) to obtain refolded extracellular superoxide dismutase. A similar strategy can be employed using size exclusion columns equilibrated with a gradient of the

denaturant used. This technique has been employed to obtain β-lactamase and B

lymphocyte stimulator from inclusion bodies (Harrowing and Chaudhuri, 2003; Cao et al., 2005).

The most common means of refolding the denatured protein is the rapid dilution of denatured protein into buffer. The buffer conditions, and the presence and concentration of a number of refolding additives, both chemical, such as arginine (Tsumotoet al., 2005) and poly-ethylene glycol and protein based, such as molecular chaperones (Wanget al., 1999) are used to support the refolding of the target

protein. Such additives can be used with the column approaches outlined above, but the larger amounts required usually rule out such an approach for cost reasons.

The final approach used for the refolding of inclusion body proteins is the removal of the disaggregating agent by dialysis. This strategy cannot be used when the disaggregating agent is a large molecule, such as a detergent, and so is usually used with chemical denaturants like guanidine and urea. The dialysis buffer used may contain refolding additives like those used in rapid dilution. This approach is typically attempted after the rapid dilution approach has failed, since the removal of denaturant is slower and therefore refolding is slower. The disadvantage of this approach to the refolding of proteins is that it is slower and when scaled to generate quantities of protein suitable for biophysical analysis may require several dialysis steps to completely remove the denaturant. This approach was used to generate adenylate kinase from inclusion bodies by Hibinoet al., (1994).