|Home | About | Journals | Submit | Contact Us | Français|
Recombinant proteins typically include one or more affinity tags to facilitate purification and/or detection. Expression constructs with affinity tags often include an engineered protease site for tag removal. Like other enzymes, the activities of proteases can be affected by buffer conditions. The buffers used for integral membrane proteins contain detergents, which are required to maintain protein solubility. We examined the detergent sensitivity of six commonly-used proteases (Enterokinase, Factor Xa, Human Rhinovirus 3C Protease, SUMOstar, Tobacco Etch Virus Protease, and Thrombin) by use of a panel of ninety-four individual detergents. Thrombin activity was insensitive to the entire panel of detergents, thus suggesting it as the optimal choice for use with membrane proteins. Enterokinase and Factor Xa were only affected by a small number of detergents, making them good choices as well.
Modern recombinant protein expression constructs include one or more affinity tags to aid in purification and/or detection. After serving its requisite function(s), the tag is often removed so as not to (potentially) interfere with “downstream” protein applications such as functional or structural studies. Three-dimensional crystallization, for structure determination by x-ray crystallography, is often deleteriously affected by inclusion of the disordered or flexible affinity tag. An engineered site for a specific protease in the linker region between tag(s) and native protein is thus included to facilitate tag removal. Common proteases include Enterokinase , Factor Xa , Human Rhinovirus 3C Protease (1 HRV 3C) , SUMO Protease , Tobacco Etch Virus (TEV) Protease , and Thrombin [6, 7].Table 1 lists the canonical recognition sequences, and specific cut-sites, for each of these proteases. For constructs containing an N-terminal tag with a protease site in the linker, Enterokinase, Factor Xa, and SUMOstar will return the original (parent) protein, while HRV3C, Thrombin, and TEV leave several residues of the protease site. TEV is the most widely-used of these proteases [8–11]. In addition to its high specificity, TEV maintains activity in a wide range of buffer and solution conditions, and is readily capable of being produced in-house.
Several other considerations can influence the choice of protease for removal of affinity tags. Protease specificity can vary widely. Digestive and coagulation proteases can (and do) cleave proteins at sites other than the engineered “cut-site”; examples of this include non-specific proteolysis of recombinant proteins by enterokinase , thrombin  and factor Xa . To quote, “It is necessary to characterize the protein of interest after cleavage from the affinity label to assure that there are no changes in the covalent structure of the protein of interest .” Typically, this characterization method would be mass spectrometry, and reliable methods of sample preparation have been developed for integral membrane proteins . In contrast, viral proteases (e.g. HRV3C and TEV) are very highly specific [15, 16]. However, viral proteases typically possess turnover rates that are very much lower, as much as 104 lower, than those of non-viral proteases . The much lower activity of viral proteases is reflected, empirically, in the observation that those labs which utilize it for “large-scale” protein production (for x-ray crystallography or NMR spectroscopy) commonly make their own HRV3C and/or TEV for use. Therefore, the selection of non-viral vs. viral proteases, for removal of affinity tags from recombinant fusion proteins, is, essentially, a trade-off between specificity and activity.
Maintaining the solubility of integral membrane proteins in aqueous solution requires the presence of detergents or other surfactants . These detergents, present at concentrations above the critical micelle concentration (CMC), form a protein-detergent complex (PDC) with the membrane protein . Detergents can have inhibitory effects upon proteases; in one example, we previously demonstrated that several detergents inhibit TEV . The inability of TEV to efficiently remove an affinity tag in a particular detergent is troublesome and unfortunately precludes the universal use of TEV for membrane proteins. Many detergents and detergent mixtures are, in principle, possible candidates for use with membrane proteins. Also, as mentioned, multiple proteases besides TEV are commonly used. In practice, when a protease does not remove the affinity tag of a membrane protein, two possibilities (that are not mutually exclusive) for this failure exist. The tag could be sterically inaccessible to the protease because of the protein, the detergent, or both. Or, the protease could be inhibited by the detergent. In order to eliminate this situation of “one equation with two unknowns”, we characterized the sensitivities of a set of proteases (Enterokinase, Factor Xa, HRV 3C, SUMOstar, TEV, and Thrombin) to a large number (ninety-four) of individual detergents. This detergent panel was recently compiled in conjunction with our recent development of a high-throughput assay for screening the stability and size of a PDC in multiple detergents.
Enterokinase, Factor Xa, HRV 3C, and Thrombin along with their respective cleavage control proteins were purchased from EMD Biosciences; SUMOstar and its cleavage control protein were obtained from LifeSensors, Inc. We made TEV “in-house” using published methods ; the cleavage control protein is a protein domain on which we work, and its affinity tag is quantitatively removed by TEV. Detergents were from Anatrace, Avanti Polar Lipids, EMD Biosciences, or Bachem. Electrophoresis and blotting was performed with E-PAGE 48-well 8% gels and iBLOT nitrocellulose transfer stacks (Invitrogen), and visualized with colloidal gold total protein stain (Bio-Rad).
Enterokinase (1:50dilution, 4hr digest, 1μg control protein/well); Factor Xa (1:50dilution, 4hr digest, 2μg control protein/well); HRV 3C (1:50dilution, 4hr digest, 1μg control protein/well); SUMOstar (1:50dilution, 4hr digest, 2μg control protein/well); TEV (36ng/μl, overnight digest, 5μg control protein/well); Thrombin (1:35dilution, 4hr digest, 1μg control protein/well). The reaction and dilution buffers were made from the concentrated commercial stocks accompanying the proteases except for TEV where the buffer (100mM Tris pH 8, 1mM EDTA, 2mM DDT) was prepared. Two samples were made in 96 well PCR plates based on the above conditions. Plate1 contained control protein and the detergent while Plate2 consisted of the control protein, detergent, and 1μl of the diluted protease. The final volume of each well was 15μl. Plates were gently shaken at 25°C/300rpm in an Eppendorf Thermomixer. After the digestion was complete, 5μl of 4X E-PAGE loading dye was added to each plate. Samples were then loaded on a 48-well E-PAGE gel, blotted to a nitrocellulose membrane using the iBLOT apparatus, and visualized with colloidal gold stain.
In order to assess the activity of commonly used proteases in our detergent panel, we digested soluble proteins containing the appropriate protease cleavage site. The experimental design presented here is similar to our previous study of the detergent sensitivity of TEV .We note that a report from another laboratory utilized three different membrane proteins as test proteins . We have chosen to use soluble proteins for several reasons: 1) a test membrane protein would have to be stable in every detergent in the panel to be a reliable test protein, and 2) a protease site on a membrane protein might be occluded by the detergent of the PDC, while a soluble protein should not interact with detergent and is thus much less likely to have its protease site occluded by detergent. Moreover, in this present study, the use of vendor-supplied positive control proteins obviates the possibility of the protein occluding the cleavage site.
Figure 1 shows a composite of the protein gels used to evaluate the protease activity in the detergent panel. The relative activities of each protease were estimated from the amount of cleavage product observed on the protein gels and is summarized in Table 2. The best protease was Thrombin which has maximum activity in all of the detergents tested, followed closely by Enterokinase and Factor Xa while HRV 3C and SUMOstar were drastically affected by detergent. TEV possessed activity in most detergents, but at low levels in a large percentage of these detergents. Since TEV is typically made as a reagent in-house, more can be added to a cleavage reaction to possibly overcome the inhibitory effect of a particular detergent. The poor performance of SUMOstar was somewhat surprising, since this protease recognizes the tertiary structure of the large SUMOstar tag  compared to the short recognition sequences of the other proteases tested. The SUMOstar tag may be partially unfolded in detergent micelle solutions or may possibly insert into the micelle, making it unavailable for binding the SUMOstar protease.
Based upon our data, the activity of Thrombin is not significantly affected by any of the ninety-four detergents of our panel . This panel encompasses, as single detergents in individual solutions, nearly all of the detergents utilized in membrane protein biochemistry, biophysics and structural biology (at present). Therefore, we recommend the design and utilization of a thrombin cleavage site for protein expression constructs; this will provide for the most detergent-invariant affinity tag removal. Moreover, Enterokinase and Factor Xa were only affected by a small number of detergents, making them good choices as well. Additionally, removal of an N-terminal affinity-binding site by Enterokinase or Factor Xa produces the wildtype (or parent) construct protein free from any extraneous residues derived from the protease recognition site. This attribute may be (very) advantageous; for example, a crystal contact mediated through the N-terminus could be disrupted by the presence of these extra residues.
Funding for this research was provided by NIH Roadmap Grant 5R01 GM075931 (to M.C.W.).
1Abbreviations used: CMC, critical micelle concentration; DTT, Dithiothreitol; EDTA, Ethylenediamine-tetraacetic acid; HRV 3C, Human Rhinovirus 3C Protease; PAGE; Polyacrylamide gel electrophoresis; PDC, protein-detergent complex; SUMO, Small Ubiquitin-like Modifier; TEV, Tobacco Etch Virus Protease.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.