Proc. Nati. Acad. Sci. USA Vol. 86, pp. 4367-4371, June 1989 Biochemistry Dual affinity fusion approach and its use to express recombinant human insulin-like growth factor II (human serum albumin binding/IgG binding) BJORN HAMMARBERG*, PER-AKE NYGREN*, ERIK HOLMGRENt, ANETTE ELMBLADt, MICHAEL TALLYf, ULF HELLMAN§, TOMAS MOKS*, AND MATHIAS UHLtN*¶ *Department of Biochemistry, Royal Institute of Technology, S-100 44 Stockholm, Sweden; tKabigen AB, S-112 87 Stockholm, Sweden; tDepartment of Endocrinology, Karolinska Institutet, S-104 01 Stockholm, Sweden; and §Ludwig Institute for Cancer Research, Biomedical Center, S-75 123 Uppsala, Sweden Communicated by Peter Reichard, February 28, 1989 (received for review September 27, 1988) ABSTRACT A dual affinity fusion concept has been de- In this report, we describe the development of a concept in veloped in which the gene encoding the desired product is fused which the gene of interest is fused between two different between two flanking heterologous genes encoding IgG- and heterologous genes. After affinity purification of the dual albumin-binding domains. Using sequential IgG and serum affinity fusion protein by using both the N-terminal and the albumin affinity chromatography, a full-length tripartite fu- C-terminal domains, a full-length product is obtained, suit- sion protein is obtained. This approach was used to recover a able for structural and functional studies. The system can full-length fusion product in Escherichia coli containing the also be used to facilitate molecular studies of protein degra- human insulin-like growth factor II (IGF-ll). Surprisingly, the dation in vivo and in vitro, as both the C-terminal and the recombinant IGF-II showed increased stability against proteo- N-terminal regions can be recovered independently. lytic degradation in E. coli when produced as a dual affinity To develop an optimal dual affinity system, the two affinity fusion protein, as compared to an N-terminal fusion protein. tails should be small, soluble, and stable in various host After site-specific cleavage of the tripartite fusion protein, organisms and should preferably not form dimers or multi- IGF-ll molecules with immunological and receptor binding mers. It is advantageous for the tails to be secretion "com- activity were obtained without renaturation steps. The results petent" so that the product can be recovered from the culture demonstrate that proteins can fold into biologically active medium, which facilitates the protein purification and allows structures, even if provided with large flanking heterologous the formation of disulfide bonds. Finally, it is desirable that protein domains. The concept was further used to characterize the two affinity systems have similar binding strengths and do the degradation of recombinant IGF-II in this heter- not cross react. specific We have chosen a dual affinity fusion system based on the ologous host. IgG-binding domains of staphylococcal protein A and the albumin-binding domains of streptococcal protein G. This A large number of gene fusion systems to facilitate expres- system was used to express biologically active human insulin- sion and purification ofrecombinant proteins in heterologous like growth factor II (IGF-II) in Escherichia coli and also used hosts have been described (1). Most of these expression to characterize the specific degradation of recombinant IGF- systems have been designed for high expression levels to II in this heterologous host. yield insoluble material that accumulates as aggregates in the cytoplasm (1). The disadvantage of this "inclusion body" approach is that in vitro refolding is required to obtain a MATERIAL AND METHODS biologically active protein. Such refolding schemes are often Bacterial Strains and Vectors. E. coli HB101 (5) and RRI complex and demand specific and time-consuming optimiza- M15 (6) were used as hosts. Vectors used were phage tion for each gene product. M13mpl8 (Pharmacia) and plasmids pEMBL8, pEMBL9 (7), Alternative expression systems that yield a soluble gene pEZZ8 (8), pEZZT308 (9), pSPG2 (10), and pRIT18 (11). product with a native structure in vivo, therefore, have a great DNA Constructions. DNA work was carried out as de- advantage. In particular, protein engineering studies might scribed (5). The synthesis of oligonucleotides was performed benefit from an approach where a biologically active protein as described (12). DNA sequencing was performed as de- can be obtained directly. A soluble gene product also allows scribed (13). the assembly of fusion proteins containing an "affinity han- The plasmid pRIT24 was obtained by subcloning the al- dle" to facilitate the purification. The same general purifi- bumin-binding regions from streptococcal protein G derived cation scheme can thus be used for a variety ofgene products. from plasmid pSPG2. After digestion with EcoRI the ends For protein engineering studies this approach eliminates the were made blunt with the Klenow fragment of DNA poly- need for individual purification schemes for "mutant" pro- merase I, a synthetic Sal I linker (GGTCGACC; Pharmacia) teins even with altered biochemical properties (2). was added by ligation. Digestion with Sal I and Pst I yielded However, expression in heterologous hosts of soluble gene a 640-base-pair fragment that was isolated by agarose elec- products has been hampered by problems with proteolysis trophoresis and inserted between the same sites in pEMBL8. (3), in which a heterologous population of products is ob- The gene fragment encoding the albumin-binding domains tained even if an afflnity-purification approach is used (4). was isolated from this plasmid by digesting with EcoRI and The recovery of the full-length product from such a mixture HindIII, adjacent to the insertion sites. The expression requires additional purification steps and often gives low vector pEZZT308 was digested with EcoRI and HindIII and overall yield. This has emphasized the need for expression the fragment described above was inserted by ligation. The systems that ensure a full-length product. resulting plasmid, pRIT24, contains the staphylococcal pro- The publication costs of this article were defrayed in part by page charge Abbreviations: HSA, human serum albumin; IGF, insulin-like payment. This article must therefore be hereby marked "advertisement" growth factor; RRA, radio receptor assay. in accordance with 18 U.S.C. §1734 solely to indicate this fact. ITo whom reprint requests should be addressed. 4367 Downloaded by guest on September 29, 2021 4368 Biochemistry: Hammarberg et al. Proc. Natl. Acad. Sci. USA 86 (1989) tein A promoter and signal sequence followed by a gene ........ encoding a dual affinity fusion protein consisting of the ZZ A ,,,,,, ,,,,.....B... region derived from staphylococcal protein A and the B1B2 B affinity purification O region of streptococcal protein G. The IGF-II gene was . 1 assembled from synthetic oligonucleotides and inserted into . ',Is the EcoRI and HindIII sites of pEZZ8 to yield plasmid pRIT19 (B.H., unpublished results). This plasmid was di- A affinity purification gested with Not I and Msp I and the 870-base-pair fragment Bound Flow through containing the noncoding region upstream of the promoter, the promoter region, and the regions encoding S, ZZ, and -n..... IGF-II was isolated. This fragment was ligated with a syn- Site-specific & cleavage thetic linker (5'-CGGCGAAATCTGAAATGG and its com- plementary sequence, 5'-GATCCCATTTCAGATTTCGC) A andE''''''ssssssssssssss..........B affinity Q purification to change the stop codon to an ATG triplet encoding a methionine residue, which enables CNBr cleavage (see Fig. 2B). The linker starts with the Msp I site in the IGF-II gene a new the B1B2 FIG. 1. Dual affinity fusion concept. The protein of interest (X) and ends with BamHI site in frame with with a putative protease degradation site (solid arrow) is fused region. The fragment was recovered by cleavage with EcoRI between two different affinity tails (A and B). and BamHI and the shorter fragment was purified and ligated with the large EcoRI-BamHI fragment isolated from radation ofrecombinant proteins can be obtained by selective pEMBL8. After transformation to E. coli RR1 M15, blue affinity purification. The flow-through fraction of the tail A colonies were isolated. DNA prepared from one of these affinity purification step yields the C-terminal degradation colonies was cut with Xho I and BamHI, cloned into products (Fig. 1), which can be N-terminally sequenced. In M13mp18, and sequenced. The EcoRI-BamHI fragment con- addition, "nicked" proteins that are proteolytically cleaved sisting ofthe mutated IGF-II gene without the stop codon was but held together by disulfide bonds can be released by isolated and ligated to the isolated fragment from pRIT24 to reduction of the material obtained by the dual affinity steps. yield the IGF-II dual affinity fusion plasmid, pRIT25. This might provide additional information about the specific Expression and Purification of Proteins. Bacteria were degradation sites. harvested by an osmotic shock procedure (14). The shock Dual Affinity Vector System. Nygren et al. (9) showed that lysate was passed through an affinity column of human serum the B1B2 fragment of the streptococcal protein G receptor albumin (HSA)-Sepharose (9) or of IgG-Sepharose Fast Flow binds specifically to HSA and also demonstrated that this (Pharmacia) as described (15). Flow-through and fractions fragment could be used to purify a fusion protein by HSA eluted with 0.5 M HOAc (pH 2.8) were collected and lyo- affinity chromatography. The similarities in size (-60 amino philized. Eluted material was lyophilized and dissolved in acid residues per binding unit), binding and elution condi- PBST buffer (50 mM sodium phosphate, pH 7.1/0.9% NaCl/ tions, stability, and solubility make the B1B2 domain and the 0.05% Tween 20) for a second affinity chromatography step ZZ domain of protein A suitable fusion partners in a dual or in 70% (vol/vol) formic acid for treatment with CNBr as affinity system. Both receptors are monomeric and show no described (4). cross affinity between the ligands (9).