Target selection of soluble protein complexes for structural proteomics studies
© Shen et al; licensee BioMed Central Ltd. 2005
Received: 24 December 2004
Accepted: 18 May 2005
Published: 18 May 2005
Protein expression in E. coli is the most commonly used system to produce protein for structural studies, because it is fast and inexpensive and can produce large quantity of proteins. However, when proteins from other species such as mammalian are produced in this system, problems of protein expression and solubility arise . Structural genomics project are currently investigating proteomics pipelines that would produce sufficient quantities of recombinant proteins for structural studies of protein complexes. To investigate how the E. coli protein expression system could be used for this purpose, we purified apoptotic binary protein complexes formed between members of the Caspase Associated Recruitment Domain (CARD) family.
A combinatorial approach to the generation of protein complexes was performed between members of the CARD domain protein family that have the ability to form hetero-dimers between each other. In our method, each gene coding for a specific protein partner is cloned in pET-28b (Novagen) and PGEX2T (Amersham) expression vectors. All combinations of protein complexes are then obtained by reconstituting complexes from purified components in native conditions, after denaturation-renaturation or co-expression. Our study applied to 14 soluble CARD domain proteins revealed that co-expression studies perform better than native and denaturation-renaturation methods. In this study, we confirm existing interactions obtained in vivo in mammalian cells and also predict new interactions.
The simplicity of this screening method could be easily scaled up to identify soluble protein complexes for structural genomic projects. This study reports informative statistics on the solubility of human protein complexes expressed in E.coli belonging to the human CARD protein family.
The aim of structural proteomics is to experimentally derive the 3-dimensional structure of all proteins in genomes through determination of the 3D structure of sufficient members in a protein family such that the remaining structures can be predicted accurately through computational approaches . The success rates in studies so far, mainly applied to prokaryotes, are still low. Effectively, 1 to 5% of structures are solved experimentally from a few thousand of Open Reading Frame (ORF) tested per study. The reasons for this limitation are primarily due to the fact that many proteins are insoluble when expressed in a heterologous system or when purified and concentrated at the levels necessary for crystallization studies. The success rate is even lower when applied to eukaryotes, where for example, it was shown recently that in C. elegans only 20% of the ORF cloned produced soluble recombinant protein suitable for structural studies. This rate is significantly lower than that of structural genomics studying prokaryotics ORF . In these studies, a tremendous amount of effort has been spent to improve protein solubility by switching the expression system, changing the fusion tags, replacing the expression host, generating a new structural variant by modifying the protein through genetic engineering including mutagenesis  and computational approaches, and finally by performing crystallization studies on a protein orthologs to the initial protein candidate [2, 5]. However, full automation of these approaches needs to be improved and new technologies are constantly being implemented in the pipelines to increase our success rates in structure determination. An even more challenging and important task for structural proteomics studies is to determine the 3D structure of all the protein complexes of an organism. Most proteins do not work as a monomer, but interact with other proteins to perform their functions in the form of stable or transient complexes. In addition, the knowledge of these interactions is of fundamental importance, since the genome complexity of an organism is not simply related to its number of genes but rather is more directly related to the complexity of its protein-protein interaction networks. Full automation processes are still difficult because the proteomics pipeline to obtain sufficient quantities of soluble protein complexes has to allow for the possibility of different purifications scenarios for each protein complex. Furthermore, protein complexes are generally difficult to reconstitute from individual protein components and usually it is almost impossible to purify sufficient quantities of in vivo protein complexes for structural studies. Finally, interactions between proteins in cells are effectively dynamic with a wide range of affinities. The half-life of protein complexes is quite diversified from stable complexes to transient interactions and can be dependent on post-translation modifications. The post-translational modification may not be reproduced during in vitro experiments or are inherently not suitable for crystallization studies since highly stable systems are necessary for successful crystallographic studies [6–8].
Three steps are needed to obtain sufficient protein complex in vitro to perform structural studies of protein complexes from individual components. (i) Produce soluble protein, (ii) make a soluble complex, and (iii) generate high-concentration and highly purified protein complexes to tackle the crystallization project. In this study, we focus on the first step, that of obtaining soluble protein complexes. A few structural genomics initiatives are currently trying to develop proteomics pipelines that includes the production in E.coli of recombinant protein complexes . We tried to identify an ideal protein family that could be used for this purpose. It should be a small soluble domain that can perform simple protein complexes and from which many protein complexes have been identified in vivo. As a result we choose the human CARD protein family. This protein family belongs to the death domain super-family consisting of the Death Domain (DD), Death Effecter Domain (DED) and the Pyrin, AIM (Absent-in-melanoma), ASC, Apoptosis-associated speck-like protein containing a caspase recruitment domain CARD, and Death-Domain (DD)-like (PAAD/DAPIN/PYRIN) domain subfamilies. The CARD domain family contains 35 members that are involved in apoptosis and share a domain of 85 amino-acids folded into a 6 helix bundle. The domain is a recruitment domain that has the ability to form hetero-dimers between members of the same family providing a structural framework to complete the apoptotic cascade leading to caspase activation and cell death. To date, two different structures of death domain complex have been solved by x-ray crystallography represented by the structure of the human CARD domains of Apoptotic Protein Activating Factor-1 (APAF-1) with caspase-9, and the death domain of Pelle with Tube in Drosophillia. Structural analysis of these two different protein complexes revealed different modes of protein-protein recognition while similar domains are involved in protein-protein interaction.
In this work, the first step of a structural proteomic project to obtain soluble protein complexes from the CARD domain is reported. From 25 CARD proteins cloned and expressed in E.coli, 14 different CARD soluble proteins were obtained. Protein purification of CARD recombinant proteins were performed and binary protein complexes were reconstituted systematically from purified individual components under native, denaturation-renaturation conditions and by co-expressing two different CARD proteins simultaneously in E. coli to identify stable protein-protein interactions.
Expression and solubility
To study the interaction between CARD-containing proteins, 14 CARD recombinant proteins were subjected to three different binding assays. First, all recombinant clones were characterized for their expression and solubility in E. coli. Comparison of the total cell lysate and soluble protein fractions revealed that most of the CARD proteins were expressed at medium or high levels (2–5 mg/liter). Only pET constructs (that add a poly-histidine to the N-terminus of the expressed protein of interest) containing the CARD domains of the cellular homolog of the equine herpesvirus-2 E10 gene containing an amino-terminal caspase recruitment domain (BCL10) and the apoptotic protein containing a CARD domain (CLAN) (pET-28b-BCL10 and pET-28b-CLAN) were producing a recombinant protein expressed at low level (less than 1 mg/ml). Protein expression was induced at two different temperatures (37°C and 30°C) and two different IPTG concentrations (1 mM and 0.1 mM) to find the optimal solubilization conditions.
Three pET-28b-constructs of the, CARD only protein (COP), NucleOLar protein 3 (NOL3), and the Apoptosis-associated Speck-like protein containing a CARD (ASC) (pET-28b-COP, pET-28b-NOL3, and pET-28b-ASC) produced in E.coli recombinant proteins were insoluble. Although some improvement of solubility was seen after induction at 30°C, we classified them as "not soluble" because of low yield (below 1 mg/L of culture). These insoluble proteins were purified from E. coli under denaturing conditions and then renatured back to the native state. The status of refolding was confirmed by circular dichroism (CD) experiments (data not shown).
Protein purification and direct native binding assay
Induced denaturation-renaturation binding assay
In this study, co-expression binding assay performed the best compared to the other 2 approaches tested. It was also the simplest assay system since it did not require prior purification of the binding partners before complex formation. In addition, we observed that some proteins were not soluble when expressed alone, but could form a soluble protein complex in co-expression experiments. For example, the CARD domain of ASC protein is barely soluble when expressed alone in E. coli, but its complex with the CARD domain of ICEBERG in co-expression assay is totally soluble.
Both, native binding assay and induced denaturation-renaturation binding assay were more sensitive to sample preparation. For the native assay, among most of the protein pairs tested the binding was not detected even when both proteins were well folded and stable. This could indicate that the binding conditions needs to be optimized for each complex or that assisted folding and post-translational modifications are necessary for optimal binding. For the induced denaturation – renaturation binding assay, since the co-incubated pairs of proteins are denaturated then renaturated in the same conditions to form complexes, those conditions should be such that some potential protein complex could not be formed between some of the pairs tested in this study [25, 26]. This may be the reason why three out of seven published CARD-CARD complexes could not be reproduced in this assay.
Our experimental results indicate that the co-expression system is recommended as the first choice for structural study of protein-protein complexes. This binding assay system is simple, inexpensive and can be easily automated for structural proteomics studies.
EST clones encoding CARD domain were searched in the NCBI database http://www.ncbi.nlm.nih.gov/entrez/query.fcgi. 14 EST spanning the CARD domain were found and used as templates for PCR amplification. The CARD domains were amplified by PCR and cloned into the expression vector pET or pGEX to obtain the following CARD-containing proteins: NAC (residues 1373–1473), TUCAN (residues 341–431), Apaf-1 (residues 1–108), caspase-9 (residues 1–111), CARD9 (residues 14–100), BCL10 (residues 1–104), RAIDD (residues 9–95), COP (residues 1–98), NOL3 (residues 1–97), CLAN (residues 1–87), RICK (residues 417–540), ICEBERG (residues 1–90), CARD10 (residues 31–117), and ASC (residues 92–195). The individual CARD domains (except RICK) were cloned into the Nde1-BamH1 and BamH1-EcoR1 restriction sites of pET-28b (Novagen) and pGEX-2T vectors, respectively. Since RICK had an internal BamH1 site in its sequence we cloned it into Nde1-EcoR1 sites of the pET28b vector.
Protein expression, solubility, and purification
pET-28b clones were transformed into E.coli BL21(DE3) (Novagen) and grown on petri dishes containing 50 μg/ml kanamycin. Cultures were grown from single colonies in 25 ml LB medium supplemented with kanamycin (40 μg/ml) at 37°C. His-tagged proteins were induced at an OD600 of 0.6 by the addition of IPTG to a final concentration of 1mM. After an additional 4 hours of growth at 30°C or 37°C, 2x1 mL aliquots of the culture were centrifuged to obtain cell pellets (5 min at 13000 rpm). One of the cell pellets were re-suspended in denaturing buffer and kept as the whole cell fraction. The other fractions were re-suspended in native lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole), sonicated and centrifuged (5 min at 13000 rpm) in Eppendorf tubes. The resulting supernatant representing the soluble protein fraction was compared against the whole cell fraction by SDS-PAGE to determine the size, and the relative expression and solubility levels of each protein.
Binding assay – native condition
For this assay the 14 pET-28b constructs were transformed into E. coli BL21 (DE3). Transformed cloned were inoculated into a 50 ml LB for overnight preculture. The non-saturated bacterial cultures were further diluted into 1 Liter of LB medium for protein expression. The cultures were grown at 37°C with shaking until an A600 of 0.6 was reached. Protein expression was induced by the addition of IPTG to a final concentration of 1 mM and the induction was allowed to take place for an additional 4 hours. The induced bacteria were collected by centrifugation (20 min at 4000 rpm). Purification was carried out by either native or denaturing conditions depending on the solubility status of the individual proteins. For purification under native conditions, the cells were resuspended in 15 ml 50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, 5 mM β-mercaptoethanol, and 0.1 mM PMSF at pH 8.0. The cells were lysed with a french pressure cell and centrifuged for 30 min at 4°C at 18,000 rpm. Ni-NTA superflow (Qiagen) was used in the His-tag purification of the induced protein. 1 ml 50% Ni-NTA slurry was used to pack the chromatography column. The purification of 6xHis-tagged protein was performed in this column using a 10 to 250 mM imidazole gradient. To facilitate the purification process, the columns were centrifuged for 1 min at 4°C at 800 rpm for the washing and elution steps. The purified proteins were then dialyzed against binding buffer containing 50 mM Tris, and 100 mM NaCl at pH 8.0. For purification under denaturing conditions, the cells were re-suspended in 15 ml of 100 mM NaH2PO4, 10 mM Tris-HCl, 8 M urea, pH 8.0. After loading the lysate onto the column the renaturation was performed using a linear 6 M – 1 M urea gradient in 50 mM Tris, 100 mM NaCl, 5% glycerol, 1 mM EGTA, 0.1 mM PMSF, 2 mM MgCl2, 14.4 mM β-mercaptoethanol, pH 8.0. Elution of the bound proteins was performed using a pH gradient from pH 8.0 to pH 4.5. After elution of the re-natured proteins the protein solution were dialyzed against a binding buffer containing 50 mM Tris, 100 mM NaCl, pH 8.0. An equal amount of the purified proteins (91 combinations) were incubated overnight at 4°C and the binding was analyzed by continuous non-denaturing polyacrylamide gel electrophoresis (10% PAGE) followed by Coomassie blue staining.
Binding assay – induced denaturation-renaturation
The purification of the 14 recombinant proteins were performed as described previously. An equal amount of the purified proteins (91 combinations) were dialyzed in a micro dialysis bag (100 μL of total volume used) against 8 M urea, 1 mM EGTA, 0.1 mM PMSF, 50 mM Tris, 100 mM NaCl, 14.4 mM β-mercaptoethanol, pH 8.0. The denatured proteins were then renaturated by dialyzing against a renaturation buffer containing 2 M urea, 1 mM EGTA, 0.1 mM PMSF, 50 mM Tris, 100 mM NaCl, 2 mM MgCl2, 5% glycerol, 14.4 mM β-mercaptoethanol. The protein complexes were detected by continuous non-denaturing polyacrylamide gel electrophoresis (10%) followed by Coomassie blue staining.
Binding assay – co-expression in E. coli
A mixture of 10 ng of pET-28b and pGEX-2T constructs (156 combinations) were transformed into E. coli BL21(DE3) and grown on petri plates containing 40 μg/ml kanamycin with 50 μg/ml ampicillin. Cells from 5 ml overnight cultures grown at 37°C were used to inoculate 100 ml LB medium containing 40 μg/ml kanamycin and 50 μg/ml ampicillin. Protein expression was induced at an OD600 of 0.6 with the addition of 1 mM of IPTG. After an additional 4 hours of growth at 37°C, the cells were collected by centrifugation (10 min at 4000 rpm). The induced cells were then re-suspended in 50 mM Tris (pH 8.0), 100 mM NaCl, 5 mM β-mercaptoethanol, 0.1 mM PMSF and lysis were performed with a french pressure cell. Cell lysate were centrifuged 30 min at 4°C at 25,000 g. Ni-NTA superflow (Qiagen) affinity gel was used for the His-tag purification of the complex. Elution of the proteins was performed using a 20 to 250 mM imidazole gradient. After washing extensively, the bound proteins were eluted in 500 μL aliquots. The Ni-NTA purified co-proteins and protein complexes were loaded on SDS-PAGE and stained with silver nitrate.
Binding assay in rabbit reticulocyte lysate
In order to verify if the positive in vitro interactions obtained by co-expressing both CARD protein partners in E. coli were also able to form protein complex in a mammalian system, equal amounts of the putative interacting proteins were incubated in 50 μL of Rabbit Reticulocyte Lysate (Promega) at 4°C for 1 hour.
The co-expressed proteins or protein complexes and the binding products in Rabbit Reticulocyte Lysate (Promega) were immunoprecipitated with anti-6xHis or anti-GST antibody. Protein-protein interactions were detected by SDS-PAGE, western blotting and anti-GST or anti-6xHIS.
List of abbreviations
Caspase Associated Recruitment Domain
Open Reading Frame
Apoptosis-associated speck-like protein containing a caspase recruitment domain CARD, and Death-Domain (DD)-like/ Domain, ApoPtosis, INterferon response/Pyrin
Open Reading Frame
Apoptotic Protein Activating Factor-1
Apoptosis-associated Speck-like protein containing a CARD
apoptotic protein containing a nucleotide binding domain and a CARD domain
receptor-interacting protein (RIP)-Associated ICH-1/CED-3-homologous protein with a Death domain
Death effector domain
P53-Induced protein with a Death Domain
Tumor Up-regulated CARD-containing Antagonist of caspase-Nine
cellular homolog of the equine herpesvirus-2 E10 gene containing an amino-terminal caspase recruitment domain
CARD Only Protein
NucleOLar protein 3 or Apoptosis Repressor with CARD domain (ARC)
apoptotic protein containing a CARD domain, a Leucine rich repeat domain (LRR) and a NACHT domain (initially found in the genes NAip, Ciita, Het-e and Tp1)
R IP-like i nteracting C LARP k inase
Luria Broth medium
Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis
This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), British Columbia Advanced System Institute (BC ASI), the MSFHR/CIHR training program in bioinformatics and the Canadian Foundation for Innovation.
- Mayer MR, Dailey TA, Baucom CM, Supernak JL, Grady MC, Hawk HE, Dailey HA: Expression of human proteins at the Southeast Collaboratory for Structural Genomics. J Struct Funct Genomics 2004, 5: 159–165. 10.1023/B:JSFG.0000029202.77832.34PubMedView ArticleGoogle Scholar
- Terwilliger TC: Structures and technology for biologists. Nat Struct Mol Biol 2004, 11: 296–297. 10.1038/nsmb0404-296PubMedView ArticleGoogle Scholar
- Finley JB, Qiu SH, Luan CH, Luo M: Structural genomics for Caenorhabditis elegans: high throughput protein expression analysis. Protein Expr Purif 2004, 34: 49–55. 10.1016/j.pep.2003.11.026PubMedView ArticleGoogle Scholar
- Yang JK, Park MS, Waldo GS, Suh SW: Directed evolution approach to a structural genomics project: Rv2002 from Mycobacterium tuberculosis. Proc Natl Acad Sci U S A 2003, 100: 455–460. 10.1073/pnas.0137017100PubMed CentralPubMedView ArticleGoogle Scholar
- Hewitt L, McDonnell JM: Screening and optimizing protein production in E. coli. Methods Mol Biol 2004, 278: 1–16.PubMedGoogle Scholar
- Garcia RA, Pantazatos D, Villarreal FJ: Hydrogen/deuterium exchange mass spectrometry for investigating protein-ligand interactions. Assay Drug Dev Technol 2004, 2: 81–91. 10.1089/154065804322966342PubMedView ArticleGoogle Scholar
- Lanman J, Prevelige PEJ: High-sensitivity mass spectrometry for imaging subunit interactions: hydrogen/deuterium exchange. Curr Opin Struct Biol 2004, 14: 181–188. 10.1016/j.sbi.2004.03.006PubMedView ArticleGoogle Scholar
- Pantazatos D, Kim JS, Klock HE, Stevens RC, Wilson IA, Lesley SA, Woods VLJ: Rapid refinement of crystallographic protein construct definition employing enhanced hydrogen/deuterium exchange MS. Proc Natl Acad Sci U S A 2004, 101: 751–756. 10.1073/pnas.0307204101PubMed CentralPubMedView ArticleGoogle Scholar
- Alexandrov A, Vignali M, LaCount DJ, Quartley E, De Vries C, De Rosa D, Babulski J, Mitchell SF, Schoenfeld LW, Fields S, Hol WG, Dumont ME, Phizicky EM, Grayhack EJ: A Facile Method for High-throughput Co-expression of Protein Pairs. Mol Cell Proteomics 2004, 3: 934–938. 10.1074/mcp.T400008-MCP200PubMedView ArticleGoogle Scholar
- Qin H, Srinivasula SM, Wu G, Fernandes-Alnemri T, Alnemri ES, Shi Y: Structural basis of procaspase-9 recruitment by the apoptotic protease-activating factor 1. Nature 1999, 399: 549–557. 10.1038/21124PubMedView ArticleGoogle Scholar
- Xiao T, Towb P, Wasserman SA, Sprang SR: Three-dimensional structure of a complex between the death domains of Pelle and Tube. Cell 1999, 99: 545–555. 10.1016/S0092-8674(00)81542-1PubMed CentralPubMedView ArticleGoogle Scholar
- Martinon F, Burns K, Tschopp J: The inflammasome: a molecular platform triggering activation of inflammatory caspases and processing of proIL-beta. Mol Cell 2002, 10: 417–426. 10.1016/S1097-2765(02)00599-3PubMedView ArticleGoogle Scholar
- Bertin J, Guo Y, Wang L, Srinivasula SM, Jacobson MD, Poyet JL, Merriam S, Du MQ, Dyer MJ, Robison KE, DiStefano PS, Alnemri ES: CARD9 is a novel caspase recruitment domain-containing protein that interacts with BCL10/CLAP and activates NF-kappa B. J Biol Chem 2000, 275: 41082–41086. 10.1074/jbc.C000726200PubMedView ArticleGoogle Scholar
- Chu ZL, Pio F, Xie Z, Welsh K, Krajewska M, Krajewski S, Godzik A, Reed JC: A novel enhancer of the Apaf1 apoptosome involved in cytochrome c-dependent caspase activation and apoptosis. J Biol Chem 2001, 276: 9239–9245. 10.1074/jbc.M006309200PubMedView ArticleGoogle Scholar
- Pathan N, Marusawa H, Krajewska M, Matsuzawa S, Kim H, Okada K, Torii S, Kitada S, Krajewski S, Welsh K, Pio F, Godzik A, Reed JC: TUCAN, an antiapoptotic caspase-associated recruitment domain family protein overexpressed in cancer. J Biol Chem 2001, 276: 32220–32229. 10.1074/jbc.M100433200PubMedView ArticleGoogle Scholar
- Damiano JS, Stehlik C, Pio F, Godzik A, Reed JC: CLAN, a novel human CED-4-like gene. Genomics 2001, 75: 77–83. 10.1006/geno.2001.6579PubMedView ArticleGoogle Scholar
- Geddes BJ, Wang L, Huang WJ, Lavellee M, Manji GA, Brown M, Jurman M, Cao J, Morgenstern J, Merriam S, Glucksmann MA, DiStefano PS, Bertin J: Human CARD12 is a novel CED4/Apaf-1 family member that induces apoptosis. Biochem Biophys Res Commun 2001, 284: 77–82. 10.1006/bbrc.2001.4928PubMedView ArticleGoogle Scholar
- Razmara M, Srinivasula SM, Wang L, Poyet JL, Geddes BJ, DiStefano PS, Bertin J, Alnemri ES: CARD-8 protein, a new CARD family member that regulates caspase-1 activation and apoptosis. J Biol Chem 2002, 277: 13952–13958. 10.1074/jbc.M107811200PubMedView ArticleGoogle Scholar
- Lee SH, Stehlik C, Reed JC: Cop, a caspase recruitment domain-containing protein and inhibitor of caspase-1 activation processing. J Biol Chem 2001, 276: 34495–34500. 10.1074/jbc.M101415200PubMedView ArticleGoogle Scholar
- Zhou P, Chou J, Olea RS, Yuan J, Wagner G: Solution structure of Apaf-1 CARD and its interaction with caspase-9 CARD: a structural basis for specific adaptor/caspase interaction. Proc Natl Acad Sci U S A 1999, 96: 11265–11270. 10.1073/pnas.96.20.11265PubMed CentralPubMedView ArticleGoogle Scholar
- Humke EW, Shriver SK, Starovasnik MA, Fairbrother WJ, Dixit VM: ICEBERG: a novel inhibitor of interleukin-1beta generation. Cell 2000, 103: 99–111. 10.1016/S0092-8674(00)00108-2PubMedView ArticleGoogle Scholar
- Stehlik C, Lee SH, Dorfleutner A, Stassinopoulos A, Sagara J, Reed JC: Apoptosis-associated speck-like protein containing a caspase recruitment domain is a regulator of procaspase-1 activation. J Immunol 2003, 171: 6154–6163.PubMedView ArticleGoogle Scholar
- Chou JJ, Matsuo H, Duan H, Wagner G: Solution structure of the RAIDD CARD and model for CARD/CARD interaction in caspase-2 and caspase-9 recruitment. Cell 1998, 94: 171–180. 10.1016/S0092-8674(00)81417-8PubMedView ArticleGoogle Scholar
- Tinel A, Tschopp J: The PIDDosome, a protein complex implicated in activation of caspase-2 in response to genotoxic stress. Science 2004, 304: 843–846. 10.1126/science.1095432PubMedView ArticleGoogle Scholar
- Vincentelli R, Canaan S, Campanacci V, Valencia C, Maurin D, Frassinetti F, Scappucini-Calvo L, Bourne Y, Cambillau C, Bignon C: High-throughput automated refolding screening of inclusion bodies. Protein Sci 2004, 13: 2782–2792. 10.1110/ps.04806004PubMed CentralPubMedView ArticleGoogle Scholar
- Tresaugues L, Collinet B, Minard P, Henckes G, Aufrere R, Blondeau K, Liger D, Zhou CZ, Janin J, Van Tilbeurgh H, Quevillon-Cheruel S: Refolding strategies from inclusion bodies in a structural genomics project. J Struct Funct Genomics 2004, 5: 195–204. 10.1023/B:JSFG.0000029017.46332.e3PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.