PEDS Advance Access originally published online on May 22, 2008
Protein Engineering Design and Selection 2008 21(8):515-527; doi:10.1093/protein/gzn028
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Engineering of a femtomolar affinity binding protein to human serum albumin
1 Affibody AB, Voltavägen 13, SE-161 02 Bromma 2 Department of Molecular Biotechnology, Royal Institute of Technology (KTH), School of Biotechnology, Roslagstullsbacken 21, SE-106 91 Stockholm, Sweden
3 To whom correspondence should be addressed. E-mail: perake{at}biotech.kth.se (P-Å.N.)
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
We describe the development of a novel serum albumin binding protein showing an extremely high affinity (KD) for HSA in the femtomolar range. Using a naturally occurring 46-residue three-helix bundle albumin binding domain (ABD) of nanomolar affinity for HSA as template, 15 residues were targeted for a combinatorial protein engineering strategy to identify variants showing improved HSA affinities. Sequencing of 55 unique phage display-selected clones showed a strong bias for wild-type residues at nine positions, whereas various changes were observed at other positions, including charge shifts. Additionally, a few non-designed substitutions appeared. On the basis of the sequences of 12 variants showing high overall binding affinities and slow dissociation rate kinetics, a set of seven ‘second generation’ variants were constructed. One variant denoted ABD035 displaying wild-type-like secondary structure content and excellent thermal denaturation/renaturation properties showed an apparent affinity for HSA in the range of 50–500 fM, corresponding to several orders of magnitude improvement compared with the wild-type domain. The ABD035 variant also showed an improved affinity toward serum albumin from a number of other species, and a capture experiment involving human serum indicated that the selectivity for serum albumin had not been compromised from the affinity engineering.
Keywords: affinity/combinatorial protein engineering/femtomolar/human serum albumin/phage display
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Many Gram-positive bacteria produce cell surface-exposed proteins capable of binding to one or several proteins present in the serum of an infected host animal or human individual (Kronvall and Jonsson, 1999
). Some of these binding proteins, collectively termed ‘receptins’, have after their discovery been widely utilized as affinity tools in biotechnology and medicine, for applications where the specific binding properties are employed. For example, various portions of the immunoglobulin-binding staphylococcal protein A (SPA) have been extensively used for affinity-based purification, detection or removal of immunoglobulins (Linhult et al., 2005). Alternatively, the use of gene fragments encoding the immunoglobulin binding domains have been frequently used as gene fusion partners in the production of recombinant proteins allowing for affinity capture of the fusion proteins on immunoglobulin-containing matrices (Stahl and Nygren, 1997). SPA also serves as an example on how various protein engineering efforts have been performed to improve the properties of a natural receptin for certain applications, involving the introduction of changes in the amino acid sequence to, for example, improve the resistance of the protein to alkaline conditions (Linhult et al., 2004), to achieve altered elution properties (Gulich et al., 2000) or to completely switch the binding specificity toward other proteins (Nord et al., 1995).
Several receptins from a variety of different bacteria have been shown to contain albumin binding domains (ABDs), showing a high sequence homology to each other (Johansson et al., 2002). The binding strength between different ABDs and serum albumin from different species can differ widely, which has been investigated in several studies (Nygren et al., 1990; Falkenberg et al., 1992; He et al., 2006; Rozak et al., 2006). Using NMR or X-ray crystallography, the structures of a free ABD (denoted G148-GA3) derived from streptococcal protein G (Johansson et al., 2002) or a co-complex between an ABD (denoted PAB-GA) derived from Finegoldia magna and human serum albumin (HSA) (Lejon et al., 2004) have been determined showing upon a common three-helix bundle structure organization of the ABDs of which residues in helices 2 and 3 are directly involved in the binding. In humans, HSA is the most abundant of all serum proteins and has important functions as, for example, in the regulation of the osmotic pressure and as carrier protein for other molecules (Curry et al., 1998; Peters, 1985). Proteins capable of binding to serum albumin, including ABDs, peptides or anti-serum albumin antibodies, have been employed in several important applications within biotechnology and biotherapy. This includes, for example, use as serum albumin capture reagents in proteomics (depletion) (Darde et al., 2007; Klooster et al., 2007) and as affinity gene fusion partners to recombinant proteins, facilitating, for example, assay development (Baumann et al., 1998; Konig and Skerra, 1998; Stahl et al., 1989), albumin-affinity chromatography of the fusion protein (Nygren et al., 1988) or to provide a means to modulate the in vivo residence time and clearance routes of proteins for therapeutic or imaging applications. The latter is achieved via binding to circulating host serum albumin molecules characterized by having extended in vivo half-lives (ca. 19 days for HSA) (Nygren et al., 1991; Makrides et al., 1996; Dennis et al., 2002, 2007; Stork et al., 2007). Here, the affinity between the particular albumin binding protein employed and serum albumin is important. A study performed in rodents of the pharmacokinetics for a set of bifunctional constructs containing different albumin binding peptides genetically fused to a common Fab-fragment of the anti-HER2 antibody Herceptin showed a clear correlation between serum albumin affinity and serum half-life (Nguyen et al., 2006).
In this study, we describe efforts to use protein engineering for the development of ABD variants with very high affinity for HSA. In the work, the third ABD from streptococcal protein G (denoted G148-GA3) was recruited as template sequence. By a combination of combinatorial protein engineering, in vitro selection via phage display technology and rational design, the affinity was improved several orders of magnitude, from the parent affinity (KD) of 1 nM.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Bacterial strains
The Escherichia coli strain RR1
M15 was used for all phage work and for production of selected variants in initial binding experiments. The E.coli strain BL21 (DE3) was used for subsequent protein production.
A new phagemid vector (pAY1075) was constructed based on the phagemid pAffi1 used for the construction of the phage display library Zlib2002 (Grönwall et al., 2007). In pAY1075, the gene fragments encoding the Zwt and ABDwt fragment present in pAffi1 was replaced with each other. The replaced gene fragment encoding the wild-type ABD moiety (ABDwt), adapted for subsequent insertion of helix 2/helix 3-encoding library gene cassettes using introduced SacI and NheI restriction enzyme recognition sites. Phagemid pAY1075 DNA was prepared using a Qiagen plasmid preparation midi kit (QIAGEN). The plasmid was digested with SacI and NheI in NEB4 buffer for 3 h at 37°C (New England Biolabs). The solution was treated with phenol/chloroform, purified followed by an EtOH precipitation treatment and the vector part was thereafter gel purified from a 1% agarose gel using a QIAquick gel extraction kit (QIAGEN).
The 12 most promising clones from the initial screening and the ABDwt clone were selected for sub-cloning into the expression vector pAY442 (Grönwall et al., 2007). The genes for these ABD variants were cloned by AccI–NotI PCR sticky end cloning using the primer pairs AFFI-780 (5'-P-agacttagctgaagctaaagtcttagc-'3)/AFFI-898 (5'-gctttaaggtaatgcagctaaaat-'3) and AFFI-782 (5'-acttagctgaagctaaagtcttagc-'3)/AFFI-899 (5'-P-ggccgctttaaggtaatgcagctaaaat-'3) for PCR amplification of two products from their respective phagemid vectors for each variant, which after mixing, heating and re-annealing were designed to generate a cross-hybridization product with AccI–NotI overhangs (25% yield). The expression vector pAY442 was digested in a two steps at 37°C for 4 h using AccI and NotI (New England Biolabs) in NEB4 and NEB3 buffers, respectively, and dephosphorylated with calf intestinal alkaline phosphatase (CIAP) (New England Biolabs) for 1 h at 37°C. The cleaved plasmid was purified by QIAquick PCR purification kit (QIAGEN) according to the manufacturer's recommendations. The respective PCR fragments were ligated into AccI–NotI digested and dephosphorylated pAY442, for 1 h at room temperature (RT) using T4 DNA ligase (Fermentas). Part of the ligations were electroporated into E.coli BL21(DE3) cells using a 1 mm cuvette and an ECM 630 electroporator (settings: 1700 V, 200 
and 25 µF). The cells were plated on TBAB plates supplemented with 50 µg/ml kanamycin and incubated overnight (o/n) at 37°C. Correct clones were verified by DNA sequence analysis. Gene fragments encoding second generation ABD variants (variants no. 34–40, see text) were constructed using different variants from the first generation as templates in PCR amplifications using oligonucleotides AFFI-780 (5'-P-agacttagctgaagctaaagtcttagc-'3) and AFFI-1117 (5'-aggtaatgcagctaaaatatgcaatttaagggcctct-3') as primers, resulting in the introduction of codons for the amino acid sequence EALKL at positions 35–39. PCR products were purified using a QIAquick PCR purification kit (QIAGEN) according to the manufacturer's recommendations. The fragments were thereafter used as template for the AccI–NotI PCR sticky end cloning as described above. Variants ABD001 and ABD035 were selected for sub-cloning into the NheI–NotI digested expression vector pAY1448, a vector based on pAY442 and in which codons for the His6-tag was replaced with a codon for a cysteine tag, by NdeI–NotI PCR sticky end cloning using the primer pairs AFFI-1256 (5'-P-tatgtgcttagctgaagctaaagtc-'3), AFFI-1259 (5'-Biotin-gctttaaggtaatgcagc-'3) and AFFI-1257 (5'-Biotin-tgtgcttagctgaagctaaagtc-'3), AFFI-1258 (5'-P- ggccgctttaaggtaatgcagc-'3). The expression vector pAY1448 was digested at 37°C o/n using NdeI and NotI (New England Biolabs) in NEB3 buffer and dephosphorylated with CIAP (New England Biolabs) for 1 h at 37°C. The cleaved plasmid and fragments were purified by QIAquick PCR purification kit (QIAGEN) according to the manufacturer's recommendations. The PCR products and NdeI–NotI digested and dephosphorylated pAY1448 vector DNA were then used for sticky end cloning as described above.
Construction of libraries LibABDmat2005 A and B. The oligonucleotides AFFI-791 (5'-acagagagctcgacaaatatggag-3'), AFFI-792 (5'-cggaaagctagcaggtaatgcagc-3'), AFFI-794 (5'-ttgctagcaggtaatgcagctaaxxxxxxxxxtatxxxxxxxxxtacxxxxxxaacxxxxxxggcxxxgttgatxxxxxxcttgtaxxxgtcxxxxxxtactccatatttgtcgag-3') and AFFI-793 (5'-ttgctagcaggtaatgcagctaaxxxxxxxxxtatxxxxxxxxxtacxxxxxxaacxxxxxxggcxxxgttgatxxxxxxcttgtaxxxgtcxxxtactccatatttgtcgag-3') used for library construction were ordered from SGS DNA (Köping, Sweden), and ‘x’ represents a nucleotide position included in a codon subjected to variegation as summarized in Fig. 1. In two assembly reactions, the oligos AFFI-791 and AFFI-793 or AFFI-794 were annealed and extended with Taq DNA polymerase (Applied Biosystems). A PCR reaction using the external primers AFFI-791 and AFFI-792 was performed to amplify the fragment, resulting PCR products were purified using QIAquick PCR purification kit (QIAGEN) according to the manufacturer's recommendations.
|
Purified PCR products were digested with SacI and NheI in NEB4 buffer (New England Biolabs) for 3 h at 37°C. The DNA fragments were subsequently purified from a 1% agarose gel using QIAquick gel extraction kit, according to manufacturer's recommendations (QIAGEN), and ligated, into SacI and NheI-cleaved pAY1075 phagemid DNA for 1 h at RT using T4 DNA ligase (Fermentas). The ligations were purified by phenol/chloroform extraction, EtOH precipitated and resolved in a smaller volume of 10 mM Tris pH 8.0. Electrocompetent E.coli RRI
M15 cells were transformed with 60 sub-aliquots of the ligated material of each sub-library using 0.2 cm gap size cuvettes in an ECM 630 set electroporator (2.5 kV, 125 
and 50 µF). Transformed cells were grown in SOC medium for 50 min and transferred to 10 Ehrlenmayer flasks, each containing 1 l of TSB+YE supplemented with 2% glucose and 100 µg/ml ampicillin and grown o/n at 37°C. The cells were centrifuged at 6000g, following resuspension in PBS/glycerol solution to a final approximate concentration of 20% glycerol. The cells were then aliquoted and stored at –80°C. The number of cells after electroporation, amplification and resuspension in glycerol stocks was titrated on TBAB plates supplemented with 200 µg/ml ampicillin. DNA sequencing of phagemid (pAY1075) inserts was performed after library construction and ELISA. Picked colonies were subjected to PCR amplification using oligonucleotides AFFI-21 (5'-tgcttccggctcgtatgttgtgtg-3') and AFFI-22 (5'-cggaaccagagccaccaccgg-3') as primers. Sequencing of amplified fragments was performed using BigDyeTM Terminator v3.0 Cycle Sequencing Kit according to the manufacturer's recommendations (Applied Biosystems) and with the biotinylated oligonucleotide AFFI-72 (5'-biotin-cggaaccagagccaccaccgg-3'). The sequencing reactions were purified by binding to paramagnetic streptavidin-coated beads (Dynal) using a Magnatrix 8000 robot (Magnetic Biosolutions AB, Sweden) and analyzed on an ABI PRISM® 3100 Genetic Analyser (Applied Biosystems). Sub-cloned DNA fragments were verified using the same procedure, but with oligonucleotides AFFI-69 (5'-gtgagcggataacaattcccctc-'3) and AFFI-70 (5'-cagcaaaaaacccctcaagaccc-'3) for amplifying PCR fragments and AFFI-202 (5'-biotin-gtgagcggataacaattcccctc-'3) as sequencing primer.
Lyophilized HSA (Sigma) was dissolved in PBS to a final concentration of 10 mg/ml. EZ-link Sulfo-NHS-LC-Biotin (Pierce) was dissolved in water to a final concentration of 1 mg/ml and a 5-fold molar excess was added to 500 µg of serum albumin in a total volume of 0.5 ml. The mixtures were incubated at RT for 30 min. Unbound biotin was removed by dialyzing against PBS using a Slide-A-Lyzer (10 kDa cut-off, Pierce).
Preparation of phage stocks from the library and between selections was performed according to previous described procedures (Nord et al., 1997; Grönwall et al., 2007) using the helper phage M13KO7 (New England Biolabs, Beverly, MA, USA). PEG/NaCl (polyethylene glycol with NaCl) was used for phage precipitation, routinely yielding phage titers of about 1013 plaque forming units (pfu) per ml.
A phage library stock aliquot (ca. 1 ml) was PEG/NaCl-precipitated and dissolved in 1 ml PBS. Phages were pre-incubated with SA-beads in a 0.1% gelatin solution containing biotinylated bungaro toxin (20 nM) (Sigma) for 1 h at RT prior to the first selection round. The purpose of the pre-selection was to make a negative selection against SA-beads and an irrelevant biotinylated protein. All tubes and beads used in the selection procedure were pre-blocked in 0.5% gelatin for 30 min under gentle agitation at RT and subsequently left with no agitation over o/n at 4°C. The selection solutions contained biotinylated HSA, phages, Na-azid (0.02%), Tween 20 (0.1%) and gelatin (0.1%) in PBS. The selection parameters used under each cycle are presented in Table I. The phages were incubated with biotinylated HSA target for 1 h under agitation at RT before adding blocked SA-beads for capture of phage-target complexes during 15 min at RT under agitation. For the first, third and fifth panning rounds, the RT incubation was preceded by an additional incubation with target o/n at 4°C, and for the fourth cycle an additional incubation for 3 days at 4°C. The SA-beads were washed with 1 ml portions of selection buffer (0.1% of gelatin and Tween 20) and PBS for various numbers of total times depending on selection round and the next last wash in cycles 4 and 5 lasted for 2 h as described in Table I. Phages were eluted with 1000 µl of 0.05 M Glycine–HCl, pH 2.2, for 10 min at RT, followed by immediate neutralization using 900 µl PBS supplemented with 100 µl 1 M Tris–HCl, pH 8.0. The eluted phages (3/4 of the volume) were used to infect 50 ml log phase E.coli RR1M15 cells after each round of panning. After 30 min incubation with gentle agitation and 30 min with rigorous agitation at 37°C, the cells were centrifuged and the pellet was dissolved in a smaller volume and spread on TSB+YE plates finally incubated o/n at 37°C.
|
Screening of HSA binding clones after selection
Ninety-three ABD variants from each selection plus the ABDwt clone were expressed as Zwt fusion proteins as encoded by their respective phagemid vectors and analyzed for HSA binding activity using an ELISA set-up. The ELISA was optimized for detection of binders with better affinities than 10 nM against serum albumin. Randomly picked colonies were expressed in 96 deep-well plates (Nunc) by inoculating each colony into 1 ml TSB+YE medium supplemented with 100 µg/ml ampicillin and 1 mM IPTG and grown o/n at 37°C. Cells were pelleted by centrifugation at 3000g for 15 min, re-suspended in 400 µl PBS-T and frozen at –80°C. Frozen samples were thawed in a water bath, and cell debris was pelleted at 3700g for 40 min. Supernatants containing released ABD variant-Zwt fusion proteins were collected and stored at 4°C until used in ELISA.
Microtiter wells (Costar) were coated o/n at 4°C with 100 µl of HSA and with the controls RatSA (well F12), HSA (G12) and MSA (H12), at a concentration of 0.4 µg/ml in 0.1 M sodium carbonate, pH 9.5. The wells were blocked in 2% low fat milk (Semper) in PBS-T for 2 h at RT. A volume of 100 µl of the prepared ABD-Zwt containing samples was added to each well and the plates were incubated for 1.5 h at RT. Biotinylated IgG (Biovitrum) at a concentration of 0.5 µg/ml in PBS-T was added to the wells and incubated for 1.5 h. Bound IgG was detected using HRP-conjugated streptavidin (Dako), diluted 1:5000 in PBS-T and incubated for 1 h at RT followed by incubation with a TMB substrate A and B mix (Pierce) of which 100 µl was added to each well. After 30 min incubation in darkness, 100 µl of stop solution was added and the plates were read at A450 in a standard ELISA spectrophotometer. Prior to addition of each new reagent, four washes were done with PBS-T. The 59 clones from the ELISA analysis showing highest signals were sequenced as described above.
Biosensor-based ranking of off-rate kinetics
The 55 unique clones received were thereafter ranked on the basis of their respective off-rate kinetics using a Biacore 2000 instrument (Biacore AB). HSA and hIgG protein were immobilized by amine coupling onto the carboxylated dextran layer on surfaces of CM-5 chips (research grade), according to the manufacturer's recommendations. One surface on the chip was activated and deactivated and used as reference cell during injections. The periplasmic fractions of ELISA positive clones were diluted 2–8 times in HBS-EP and injected at a constant flow rate of 10 µl/min for 10 min, followed by injection of HBS-EP for 10 min. The surfaces were regenerated with two injections (20 µl) of 0.05% SDS and one injection (20 µl) of 15 mM HCl.
Protein expression and purification of His6-tagged ABD molecules
ABD variants (001, 007–018) were expressed in E. coli BL21 (DE3) as fusions to an N-terminal His6-tag and purified by IMAC using Biorobot 3000 system (QIAGEN) or Gravitrap columns (GE Healthcare). A colony of each ABD variant was used to inoculate 5 ml TSB medium supplemented with 50 µg/ml kanamycin. The cultures were grown o/n at 37°C. The following day, 50 µl of each o/n culture were inoculated separately to 100 ml TSB+YE medium supplemented with 50 µg/ml kanamycin in a 1 l flask. The cultures were grown at 100 rpm at 37°C to an OD600 between 0.7 and 1, following addition of IPTG to a final concentration of 0.5 mM and incubated at RT o/n at 100 rpm. Cultures were harvested by centrifugation at 8000g for 5 min and pellets were stored in the freezer until protein preparation. For the initial characterization of the 12 ABD variants, a Biorobot 3000 system (QIAGEN) was used, and the His6-tagged proteins were IMAC-purified under denaturing conditions using Ni-NTA Superflow columns and QIAsoft 4.1 software according to the manufacturer's recommendations.
ABD variants no. 001 (wt), 011, 013 and 034–040 were expressed in E.coli BL21(DE3) as described above. The His6-tagged proteins were IMAC purified under denaturing conditions using His-GravitrapTM kit (GE Healthcare). The pellets were resupended (vortexed) in 20 ml of the denaturation buffer B-7M (100 mM NaH2PO4, 10 mM Tris·Cl pH 8.0, 7 M urea, pH 8.0) and 8 µl Benzonase solution (MERCK) was added. The solutions were incubated for 30 min at RT and 200 rpm. An additional 20 ml buffer B-7M was added and sonicated on ice. Cell debris was removed by centrifugation at 25 000g for 40 min. The GravitrapTM columns were equilibrated with buffer B-7M and the samples were applied. The columns were then washed with 10 ml buffer B-7M, 20 ml sodium phosphate buffer (20 mM NaH2PO4, 500 mM NaCl, 20 mM imidazole) and finally with 10 ml sodium phosphate buffer with 60 mM imidazole. The ABD variants were eluted with 3 ml elution buffer (20 mM NaH2PO4, 500 mM NaCl, 500 mM imidazole). A buffer exchange to PBS using a Slide-A-Lyser (3.5 kDa cut-off, Pierce) was made by dialyzing against 5 l of PBS, pH 7.2 for 2 h at RT followed by an additional dialysis o/n at 4°C. The proteins purified on Gravitrap had an additional buffer exchange to PBS, pH 6.0 using PD 10 columns (GE Healthcare). Protein concentration was determined using Abs280. The purity of the proteins was analyzed by SDS–PAGE on 4–12% Novex gels (Invitrogen) and stained with Coomassie Brilliant Blue R (Amersham Biosciences).
Biosensor-based binding studies of first generation ABD variants
Binding studies to HSA were performed using a Biacore 2000 system. HSA was immobilized onto a CM-5 sensor chip surfaces as described above. For an initial analysis of the 12 selected variants, purified ABD samples (Biorobot 3000) were diluted in HBS-EP to 25 nM and injected at a constant flow rate of 25 µl/min for 10 min, followed by injection of HBS-EP for 30 min. The surfaces were regenerated with two injections of 20 µl of 15 mM HCl and one injection of 20 µl with 0.05% SDS in-between.
Kinetic analyses on ABD001 and high affinity ABD variants
Binding studies to HSA were performed using a Biacore 2000 system. Approximately 200 RU of HSA was immobilized onto a CM-5 sensor chip surfaces as described above. For kinetic analyses, purified His-ABD samples (GraviTrap) from variants 001, 011, 013 and 035 were diluted in HBS-EP to either 4, 10, 40, 100 and 400 nM (ABD001), 0.2, 0.8, 2, 5 and 20 nM (ABD011 and ABD013) or 0.2, 0.5, 1, 2 and 5 (ABD035). The samples were injected at a constant flow rate of 25 µl/min for 10 min, followed by injection of HBS-EP for 3 h. The surfaces were regenerated with two injections each of 20 µl of 5 and 10 mM HCl. The off-rate kinetics (kd) for ABD035 was further examined on a new HSA sensor chip surface. The ABD035 molecule (10 nM) was injected for 10 min at a constant flow rate of 75 µl/min and the off-rate kinetics monitored for 24 h.
Biosensor analysis of binding to other serum albumins
Biosensor analyses were performed on a Biacore 2000 instrument (GE Healthcare) using CM-5 chip (research grade) sensor chip surfaces containing MSA, HSA, RatSA, CSA, RabSA or BSA, immobilized by standard amine coupling. One cell surface on each chip was activated and deactivated and used as reference cell during injections. The concentrations of injected His6-ABD001, His6-ABD011 and His6-ABD035 samples were either 5 nM or 1 µM as indicated in the text. The samples were injected at a constant flow rate of 25 µl/min for 10 min, followed by injection of HBS-EP for 3 h. The surfaces were regenerated with two 20 µl injections of 10 mM HCl.
Circular dichroism measurements
Purified His6-ABD molecules (1, 11, 13, 34–40) were dialyzed as described earlier to 100 mM phosphate buffer at pH 6.0 and then diluted to 0.30 mg/ml. Samples were subjected to circular dichroism (CD) analysis using a Jasco J-810 spectro polarimeter. In all CD analyses, a cell with an optical path length of 1 mm was used. Secondary structure content data were obtained by measuring the ellipticity from 250 to 195 nm at 20°C. To obtain thermal denaturation data (Tm), the signal at 220 nm was monitored when increasing the temperature from 20°C to 90°C at 5°C/min.
Expression and purification of cysteine-tagged ABD001 and ABD035 variants
Variants ABD001 and ABD035, equipped with an N-terminal cysteine residue, were expressed as described for His6-tragged constructs (see above). The cell pellets were resuspended in 30 ml of TST buffer, 8 µl Benzonase solution (MERCK) was added and the sample was vortexed and put on ice until sonication. Sonication was preformed on ice. The lysate was centrifuged at 25 000g for 1 h and filtrated through a 0.45 µm filter (Millipore). The protein was purified using HSA-affinity chromatography. HSA-sepharose (10 ml) was equilibrated with 30 ml of TST buffer. The sample was applied and the column washed with 15 ml TST buffer and 15 ml 5 mM NH4Ac pH 5.5 to remove unbound proteins before elution of ABD molecules with 9 x 1 ml of 0.5 M HAc, pH 2.8 fractions. Protein content was spectrophotometrically determined by absorbance at A280 and relevant fractions were pooled. The proteins were further purified by reverse phase chromatography (RPC) on Äkta explorer (GE Healthcare) using a RESOURCETM RPC column packed with 1 ml SOURCETM 15RPC (GE Healthcare). The flow rate was 1 ml/min and the protein was eluted using linear gradient by 0–100% under 30 min with solvent A containing 0.065% trifluoroacetic acid and 2% acetonitrile in water and solvent B containing 0.05% trifluoroacetic acid and 80% acetonitrile in water. The eluate was fractionated in 1 ml fractions throughout the purification. Relevant fractions were analyzed by SDS–PAGE on 4–12% Novex gels and stained with Coomassie Brilliant Blue R (Amersham Biosciences) and fractions containing the ABD variants were pooled. A buffer exchange of the samples to PBS pH 6.0 using a Slide-A-Lyzer (3.5 kDa cut-off, Pierce) was performed according to the manufacturer's recommendations and the protein concentration was determined using Abs280.
Two hundred micrograms each of variants ABD001 and ABD035 equipped with an N-terminal cysteine were separately coupled to 1 ml SulfoLink® Coupling Gel (Pierce) using standard coupling conditions according to manufacturer's instructions, giving theoretical maximum HSA binding capacities of 2 mg/column. Coupling efficiencies were thereafter determined on SDS–PAGE gel with silver staining. Prepared matrices were packed in a TricornTM 5/20 column (GE Healthcare) and connected to an Äkta explorer system (GE Healthcare). Approximately 1 ml human serum (10 times diluted in PBS; in total containing 
5 mg HSA) was applied via a super loop to the two different columns. The columns were washed with three column volumes (CV) and eluted with one CV. The experiments were done with a flow rate of 0.2 ml/min and using PBS as running/washing buffer and 0.5 M HAc pH 2.8 for elution. The flow-through (0.1% of total amount applied on the gel) and elution (0.2% of total amount applied on the gel) fractions together with human serum sample and HSA standard (Sigma, 10 µg) were analyzed by SDS–PAGE on 4–12% Novex gels and stained with Coomassie Brilliant Blue R (Amersham Biosciences).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Design of the ABD library
With the objective to investigate the possibility of development of ligands showing ultra-high affinity for HSA through improvement of a naturally occurring variant, the streptococcal protein G-derived serum albumin binding domain denoted G148-GA3 was subjected to an affinity maturation strategy. This involved the use of combinatorial protein engineering for the construction of a library of domain variants which were individually displayed on bacteriophage particles for functional selection of variants showing affinity for HSA, used as target protein during repeated rounds of selections.
The library design was based on several sources of information related to both functional and structural studies of the G148-GA3 and other related ABDs, including an analysis of the structures of the G148-GA3 domain in solution (Johansson et al., 2002) (PDB entry 1GJS
[PDB]
) and the crystal structure of the complex between HSA and the homologous ALB8-GA ABD from F.magna (Lejon et al., 2004) (PDB entry 1TFO), information from sequence homologies between G148-GA3 and other ABDs (Johansson et al., 2002) and between serum albumins from different species, a mutational study of the G148-GA3 (Linhult et al., 2002) and binding data to serum albumins form different species for different ABDs (Falkenberg et al., 1992; Johansson et al., 2002).
In total, 15 of the 46 amino acid positions in G148-GA3, with surface exposed side chains located to helices 2 and 3 of the three-helix bundle structure, were selected for various degrees and characters of randomization based on considerations related to the nature of the original residue, its assumed involvement in complex formation to different serum albumins as interpreted from the HSA co-complex structure for the homologous ALB8-GA module, expected structural importance, the likelihood of some positions to be recruited for additional productive contacts with HSA and limitations related to gene synthesis (Figs 1 and 2).
|
Briefly, residues Y20, L24, L37 and I41, all highly conserved among the proteins homologous to G148-GA3 (Johansson et al., 2002
) and involved in the hydrophobic central contact interface with serum albumin, were relatively conservatively randomized for an investigation of a replacement of these for other hydrophobic residues possibly could improve the affinity. At position Y20, a conservative 50/50 randomization between only Tyr and Phe was designed based on the proposed importance of an aromatic residue in the binding interface. At positions 18 and 30, the influence from other polar side chains than the native serine and threonine, respectively, was investigated. Positions E32 and G33 were fully randomized (all 20 amino acids allowed), although for position E32, the codon was doped such that the wild-type glutamic acid would statistically occur in 56% of the variants before selection.
Residues N23, N27, K29 and E40 were interpreted to either already be involved in electrostatic interactions (i.e. attraction or repulsion) or to be possible to engineer for an improved contribution to the binding affinity. Both at positions 23 and 27, the amino acids arginine, lysine (present in several related ABDs at position 27) and serine were therefore allowed to compete with the native asparagine. At position 29, a solvent-exposed salt bridge is formed between K29 in ALB8-GA and an aspartate in HSA; therefore, only a small doping of arginine (10%) at position 29 was designed. A limited variegation of position 40, occupied by a glutamic acid in the wild-type domain, which in the ALB8-GA–HSA complex forms a salt bridge with a lysine in HSA bound to fatty acids, but also lies in the vicinity of a negatively charged surface in HSA, was investigated. Position 36 is occupied by an alanine in G148-GA3, and was allowed to be occupied also by serine (as seen in other ABDs) and threonine. The variation among the homologous proteins in positions K35 and D39 is extensive, and these positions were therefore fully randomized.
All of these variations were included in the design of the oligonucleotide used for library construction at the genetic level. In addition, compared with G148-GA3, several other ABDs contain an extra amino acid (threonine or serine) at a position before the serine at position 18. To account for this, an alternative library oligonucleotide was designed and used for the construction of a second library, which in addition to the variation described above also contained this extra codon which was designed to encode for either serine or threonine (50/50). The theoretical number of possible protein variants from the randomization was approximately 3 x 1011. However, due to the biased design of codons in several of the positions, variants representing combinations of favored triplets would be expected to be present at considerably higher relative frequencies than others.
Library construction, selection and characterization of isolated variants
The two G148-GA3 sub-libraries were separately assembled in the phagemide vector pAY1075, designed for monovalent display of the variants of M13 filamentous phage particles as fused to an IgG binding Z domain (Nilsson et al., 1987) followed by a truncated version of M13 phage coat protein III, and electroporated into E.coli host cells. Both sub-libraries had sizes of approximately 5 x 108 members and after pooling, the final library used for selections thus had a size of approximately 1 x 109 members. DNA sequencing of 96 clones from each sub-library showed that approximately 87% contained inserts with open reading frames as designed.
Five rounds of selections were performed using a soluble biotinylated HSA target protein for incubation with the phage library, followed by capture of target-phage complexes onto streptavidin beads. In the procedure, the selection stringency was gradually increased via a step-wise decrease of the target concentration from 50 nM in the first two cycles down to either 500 or 50 pM in cycle 5, in parallel with an increase in the number and duration of washing steps (Table I).
For an initial analysis of the HSA binding properties of selected variants, 93 clones each were randomly picked from plates resulting from the two variant conditions used in the parallel selection rounds IV-A and V-A. In the analysis, HSA was immobilized in microtiter plate wells and the binding of phagemid-encoded G148-GA3 library members fused to the Z fusion partner in the periplasmic fractions of the 186 clones investigated via the use of biotinylated IgG and a streptavidin-horse radish peroxidase (SA-HRP) conjugate as secondary reagents. As reference, a clone producing a wild-type G148-GA3-Z fusion protein was employed and the resulting ELISA signals from wells coated with either mouse (MSA), rat (RatSA) or HSA used as a measure for selecting clones for further analyses. In the analysis, 59 clones were found to generate signals in parity with or exceeding the signals obtained for the reference wild-type G148-GA3-Z fusion protein in its binding to RatSA (data not shown), previously shown to display the highest known affinity to the G148-GA3 domain of any of the serum albumins investigated (Nygren et al., 1990; Johansson et al., 2002). A later performed biosensor-based analysis of relative expression levels indicated that these variants had somewhat lower-than-wild type expression levels (data not shown) which showed, in retro-perspective, that none of the 59 selected clones with high ELISA signals had artifactually been chosen because of higher concentrations of recombinant protein in the extract than the reference.
The 59 selected G148-GA3 library members selected after the ELISA analysis were subjected to DNA sequencing and some clones appeared more than once leading to 55 unique clones. The resulting amino acid occupancies found in the respective variegated positions of these selected variants were compared with the individually designed input randomization strategies for the same positions and the results summarized in Fig. 2. A first and interesting observation was that none of the 55 selected variants was found to originate from the sub-library containing the additional codon (denoted codon ‘18’). Further, at nine of the 15 randomized positions the wild-type amino acid was found to re-occur in the majority of the clones, as seen for positions S18 (100% Ser), N23 (76% Asn), L24 (78% Leu), K29 (89% Lys), T30 (100% Thr), E32 (98% Glu), G33 (100% Gly), L37 (100% Leu) and I41 (100% Ile). Exception to this observations was seen for (i) position Y20, where a bias toward the alternative aromatic amino acid Phe was seen (75% Phe); (ii) position N27, at which a noteworthy and complete switch toward basic amino acids was seen (76% Arg and 24% Lys); (iii) position K35 at which the original basic residue was completely absent in selected variants, in favor for a strong bias toward polar or acidic residues (49% Ser/Asn/Gln, 36% Asp/Glu); (iv) position I38 at which a significant number of clones showed to contain Thr (13%) or Lys (5%), although this position had not been subjected to any intended variegation; (v) position D39, showing a very diverse amino acid content and (vi) position E40, at which His appeared in no less than 47% of the clones.
To further characterize the variants, an off-rate binding kinetics analysis was performed using crude periplasmic extracts prepared from the different clones. A biosensor chip was coated with HSA, and the extracts (at a single concentration) were injected over the surface, allowing (after normalization) for an approximative ranking of the variants in terms of their off-rate kinetics. Interestingly, the resulting sensorgrams showed that all 55 unique clones appeared to have slower off-rate kinetics than the wild-type G148-GA3 domain. From the results of this analysis, a set of 12 variants, denoted ABD007–ABD018 (Table II), were chosen for further studies based on their apparent capability of rapidly saturate the HSA surface combined with their capability to remain bound through long periods of washing (i.e. slow off-rate kinetics) (Fig. 3). These 12 variants were subjected to sub-cloning into an alternative expression vector for intracellular E.coli production as fused to an N-terminally located hexahistidyl (His6) tag allowing affinity purification by IMAC. To obtain a ranking with regard to the HSA-affinity for purified preparations of the variants, a biosensor-based binding study including all 12 variants was performed using IMAC purified proteins (data not shown). In this study, the G148-GA3 variants ABD011 and ABD013 showed the highest apparent affinities for HSA and these variants were subsequently characterized more thoroughly. This analysis revealed upon affinities (KD) in the 3.8–5.0 pM range (using a 3 h off-rate measurement), corresponding to up to a 315-fold increase in affinity for these ‘first generation’ affinity-matured variants compared with the wild-type G148-GA3 domain (KD = 1.2 nM) (Table II). Interestingly, an analysis of the sequences of the ABD011 and ABD013 variants showed that both variants, among other substitutions, were the only variants containing Leu residues in position 39. In addition, they both contained either of the two spontaneously occurring substitutions observed in position 38 (Lys or Thr) (Table II).
|
|
An analysis by CD spectroscopy of the secondary structure content at 20°C of the ABD011 and ABD013 variants, containing six or five amino acid substitutions, respectively, in comparison with the wild type G148-GA3 domain showed that both variants displayed CD spectra very similar to that of the wild-type G148-GA3 domain under these conditions. All three proteins showed marked minima at 207 and 221 nm, characteristic for proteins with a high content of
-helical secondary structure elements (data not shown). Studies performed to compare the relative stability toward thermal denaturation (at pH 6.0) showed a lower thermal melting point for the ABD011 (Tm
48°C) compared with the wild type G148-GA3 domain (Tm70°C) (Fig. 4). However, whereas both the wild type G148-GA3 domain and the ABD011 variant showed full refolding capability upon cooling after having experienced an incubation at 90°C, the ABD013 variant showed to be significantly affected and could only partially refold (data not shown).
|
Design and characterization of second generation variants
To investigate if novel variants showing even further improved HSA binding and good thermal stability properties could be constructed from combining sequence features seen in individual variants, a set of seven ‘second generation’ variants were constructed, expressed and characterized. These variants denoted ABD034–ABD040 were all designed to have a common C-terminal segment (residues 30–46) based on a combination of the sequences seen in variants ABD011 and ABD013. In this designed ‘consensus’ segment, the substitutions K35E (the ‘charge switch’ substitution seen in the variant ABD013, showing a slow off-rate), I38K (the spontaneous mutation occurring in the variant ABD011, showing a fast on-rate and thermo stability) were combined with leucine at position 39 (only seen in the high affinity variants ABD011 and ABD 013) and histidine at position 40 (seen in 11 out of the 12 variants selected after the off-rate ranking). This common aa 30–46 end sequence was genetically grafted into seven template sequences having different sequences in positions 20–29, corresponding to already sequenced and characterized clones to yield the seven ‘second generation’ variants (Table II).
For six of the seven variants (ABD034, 036, 037, 038, 039 and 040), tendencies of aggregation in solution at RT were seen and CD measurements showed upon limited capabilities for these variants to refold after incubation at elevated temperatures (data not shown). In contrast, the variant denoted ABD035, containing seven substitutions compared with the G148-GA3 domain, showed no aggregation tendencies, and was found to have excellent thermal refolding properties, a wild-type-like secondary structure content at RT and a relatively high Tm (Tm58°C) (Fig. 4). Interestingly, biosensor-based studies showed that this variant showed an exceptionally high affinity for HSA (Fig. 5). Large effects from mass transfer limitations were observed during injections of lower analyte concentrations, indicative of very fast association rate kinetics (Fig. 5A and B). However, to rule out that the observed effects were due to artifactually low analyte concentrations due to extensive adsorption of analyte protein to container walls, control experiments involving low analyte protein concentrations mixed with a high concentration of an irrelevant ‘sacrifice’ protein were performed with similar results (data not shown). Fitting of the data to an one-to-one Langmuir interaction model, taking into account also mass-transfer effects, showed upon an association rate constant (ka) of approximately 3 x 107 s–1 M–1 and a dissociation rate constant (kd) of approximately 1.5 x 10–5 s–1 (based on a 3 h measurement), resulting in an overall affinity (KD) of 500 fM. This value corresponds to a 10-fold improvement compared with the ABD011 variant and a 2400-fold improvement of the affinity compared with the wild type G148-GA3 domain (Table II and Fig. 5). However, motivated from the observed very low value of the dissociation rate constant, it was decided to monitor the dissociation phase for a longer period to possibly obtain a more reliable value. Interestingly, an analysis of the data from an experiment involving a 20 h dissociation phase (Fig. 5C; period of 4–24 h) suggested a further 10-fold lower value for the dissociation rate constant of 1.5 x 10–6 s–1, resulting in an overall affinity of ABD035 for HSA of 50 fM (Table II).
|
The dramatic increase in affinity for HSA displayed by the ABD035 variant compared with the wild-type G148-GA3 domain is illustrated in Fig. 6A, showing overlaid sensorgrams obtained from injections of identical concentrations of the two different domains over a HSA sensorchip surface. In addition, also the sensorgram obtained from injection of the first generation variant ABD011 is shown, indicating the improved performance seen for the second generation ABD035 variant in regard to the behavior during both the binding and the dissociation phases.
|
) protein, the ABD035 variant (•) or the ABD011 variant (
only in A) were injected over sensorchip surfaces containing comparable amounts of different serum albumin from different species as indicated. (A) HSA, 910 RU. Inset shows a magnification of the responses obtained during the initial phase; (B) Rat serum albumin (RatSA), 1170 RU; (C) Mouse serum albumin (MSA), 970 RU; (D) Cynomolgus serum albumin (CSA), 1050 RU; (E) rabbit serum albumin, RabSA (1230 RU) and (F) bovine serum albumin (BSA), 860 RU. Over the HSA, RatSA, MSA and CSA surfaces, 5 nM concentrations of protein were injected, whereas over the RabSA and BSA surfaces, 1 µM concentrations were injected. ABD011 was only injected over HSA surface.To investigate if the library work and subsequent engineering leading to the increased affinity to HSA displayed by the ABD035 variant also had affected the binding properties toward serum albumins from other species, a biosensor-based binding study was performed. Here, sensor chip surfaces containing immobilized serum albumin from either rat (RatSA, 1170 RU), mouse (MSA, 970 RU), cynomolgus monkey (CSA, 1050 RU), rabbit (RabSA, 1230 RU) or cow (BSA, 860 RU) were used (Fig. 6B–F). Comparisons of sensorgrams obtained for the wild-type G148-GA3 domain or the second generation variant ABD035 injected at either 5 nM (Fig. 6B–D) or 1 µM (Fig. 6E and F) concentrations, showed that the affinities for both RatSA and MSA had been significantly increased by the engineering (Fig. 6B and C). The binding profiles observed for the two investigated domains when injected over the primate-derived serum albumin CSA surface were very similar to the profiles obtained for HSA, thus showing upon a very much higher affinity for the ABD035 variant to CSA than displayed by the wild-type domain. For RabSA, the very low affinity displayed by the wild-type ABD domain was only marginally increased for the ABD035 variant. Interestingly, whereas the wild-type ABD did not show any detectable binding to BSA, the ABD035 variant did show a binding response. As indicated by the response value reached during the injection (1 µM concentration), corresponding to approximately half the theoretical RUmax value, the affinity (KD) for BSA displayed by the ABD035 variant appeared to be in the micromolar range.
The serum albumin specificity studies described above were performed using purified preparations of the respective serum albumins. To investigate if the engineering had affected the ability to selectively bind to serum albumin in a complex background containing also many other proteins of different characters, the capability of selective capture of serum albumin from human serum samples was investigated using affinity columns separately prepared from directed coupling to an activated gel matrix of either the original G148-GA3 domain or the ABD035 variant, recruiting unique cysteine residues introduced in their respective N-terminal by genetic engineering. The results from an SDS–PAGE analysis of both flow-through and eluate fractions showed that no apparent differences in the capability to selectively bind to serum albumin could be detected between the two affinity ligands under these conditions, indicating that the engineering had neither resulted in any detectable cross-reactivity toward any other human serum protein or in an increased general stickiness (Fig. 7).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
This work describes the development of binding proteins showing very high affinities toward HSA, based on the recruitment of a small (5 kDa) and naturally occurring HSA binding non-cysteine three-helix bundle protein (G148-GA3) as starting template for applying a combinatorial protein engineering strategy. Stringent functional selection by phage display technology for isolation of improved variants was followed by construction of second generation variants via recombination of obtained variants. The two pooled sub-libraries used in the selections were constructed to direct varying degrees of randomization at no less than 16 of the 47 positions, chosen on the basis of previous sequential, structural and biochemical characterizations of the serum albumin binding properties of the G148-GA3 domain or homologues thereto. In spite of this relatively extensive randomization, the DNA sequencing of the 59 variants chosen for analysis after phage selection and the initial ELISA screening step showed upon some variants containing additional and non-designed mutations occurring at positions L24 (L24Y: one clone) and I38 (I38K: three clones; I38T: seven clones). With regard to position 38, this position is indeed already occupied by lysine in several of the other naturally occurring ABDs. Furthermore, a strong selection for lysine at position 38 was observed when selecting for variants capable of binding to both HSA and guinea pig serum albumin in a protein library-based study investigating the molecular reasons to observed differences between homologous serum albumin binding domains (ABDs) in regards to their preferences to serum albumin from different species (Rozak et al., 2006
). However, the libraries used in the present study were designed with a focus on surface-located residues and the position 38, partly buried in the three-dimensional structure, was therefore not included for randomization.
The dissociation rate kinetics screening step using a biosensor and involving periplasmic cell lysates provided an efficient means to identify prime candidates for further investigations. In a similar approach performed with Fab antibody fragments produced in E.coli, off-rate values obtained with crude samples were found to differ only 2-fold to values obtained with purified proteins (Steukers et al., 2006).
In the present study, second generation clones were constructed via gene fusion of different individual candidate sequences to a common ‘consensus’ segment designed from hypotheses of sequence–structure–function relationships. Although a very high affinity variant was found among these clones, alternative means to combine the sequences of the first generation binders could be considered. The use of DNA-shuffling principles (Stemmer, 1994), using either the set of 55 clones from the ELISA experiment or the 12 clones chosen on the basis of their slow off-rate kinetics as input genes should provide secondary libraries in which different permutations of positive traits could be more extensively investigated. However, such an approach involving a large number of clones to be analyzed would need to be combined with a stringent selection or screening step, rather than the clone-by-clone analyses performed here. Six of the seven second generation clones analyzed in the present study were found to show tendencies of aggregation. Interestingly, the single clone which did not show such tendencies was the ABD035 clone, which was the only one containing the positively charged residue arginine at position 23.
A number of artefacts may contribute to misinterpretations of the interaction characteristics, when analyzing high affinity biomolecular interaction using the biosensor system. Such measurements typically involve both a very extended monitoring of the dissociation phase and a monitoring of binding responses after injections of sub-nanomolar concentrations of analytes. During off-rate determinations, in addition to kinetics-driven dissociation of bound ABD ligands, spurious loss of immobilized ligand molecules (i.e HSA) or thereto attached substances (e.g. fatty acid molecules) could contribute to an over-estimation of the dissociation rate kinetics. During injections of low concentration analyte solutions, adsorption to container or instrument tubings as well as mass transport limitations could result in an under-estimation of the association rate kinetics (slower than actual). However, these effects, if present during the measurements, would all contribute to an under-estimation of the interaction affinity. Mass transfer effects were indeed seen during the association phase for the determination of the affinity for the ABD035 variant, most pronounced at sub-nanomolar analyte concentrations. Analysis of the dissociation phase showed a bi-phasic behavior; a faster dissociation rate was observed during the first 3 h after which a stable and 10-fold lower value for the dissociation rate constant could be determined. It is possible that a fraction of the total number of immobilized ligand molecules, saturated with ABD035 domain proteins at the start of the analysis, had been affected by the covalent coupling to the chip surface influencing their ability to bind the ABD035 variant. A faster dissociation-rate for such ligand–analyte complexes would thus contribute to an apparently faster overall dissociation rate value during the initial period of measurements. Nevertheless, even if the higher dissociation rate constant value of 1.5 x 10–5 s–1 is used for the determination of the affinity for ABD035 variant, the resulting value of 500 fM corresponds to more than 2400-fold improvement of the affinity of the original G148-GA3 domain.
Interestingly, the ABD035 variant shows improved affinities also for a panel of three non-human serum albumins tested (RSA, MSA and CSA) which are also recognized by the original G148-GA3 domain. A fourth tested serum albumin (BSA) is not recognized by the G148-GA3 domain but was found to be bound by the developed ABD035 domain, although with a low affinity. The improved affinity obtained for an earlier described engineered serum albumin binding domain (PSD-1) containing a corresponding I38K substitution was attributed to an overall structural stabilization of the domain (He, et al., 2006; Rozak, et al., 2006). In contrast, the observed lower thermal stability for the ABD035 compared to its ancestor suggests that the improved, or in one case newly gained, binding affinities are not due to an overall structural stabilization of the domain.
The selective capture of serum albumin from unfractionated human serum showed that the engineering efforts resulting in the high affinity variant ABD035 did not compromise selectivity for serum albumin. The exceptionally high affinity for HSA displayed by ABD035 approaches the strength of the interaction between biotin and streptavidin (both having a KD in the femtomolar range), which has been widely utilized as a robust anchor within biotechnology. On the basis of the off-rate, a complex formed between ABD035 and HSA has a half-life of several days and may thus in many applications be treated as a ‘quasi-covalent’ complex. This opens up for a broadening of the utilization of the physiological properties of serum albumin in biomedical applications, including efficient extravascularization and distribution in the body of fusion proteins and conjugates containing ABD035.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
The Royal Institute of Technology (KTH) in Stockholm is acknowledged for support.
|
Footnotes |
|---|
Edited by Arne Skerra
|
Acknowledgement |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
We are grateful to Per Jonasson, Barbro Baastrup and Tove Eriksson at Affibody AB for help with some analyses and for general advice.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Baumann S., Grob P., Stuart F., Pertlik D., Ackermann M., Suter M. J. Immunol. Methods (1998) 221:95–106.[CrossRef][Web of Science][Medline]
Curry S., Mandelkow H., Brick P., Franks N. Nat. Struct. Biol. (1998) 5:827–835.[CrossRef][Web of Science][Medline]
Darde V.M., Barderas M.G., Vivanco F. Methods Mol. Biol. (2007) 357:351–364.[Medline]
Dennis M.S., Zhang M., Meng Y.G., Kadkhodayan M., Kirchhofer D., Combs D., Damico L.A. J. Biol. Chem. (2002) 277:35035–35043.
Dennis M.S., Jin H., Dugger D., Yang R., McFarland L., Ogasawara A., Williams S., Cole M.J., Ross S., Schwall R. Cancer Res. (2007) 67:254–261.
Falkenberg C., Bjorck L., Akerstrom B. Biochemistry (1992) 31:1451–1457.[CrossRef][Web of Science][Medline]
Grönwall C., Jonsson A., Lindström S., Gunneriusson E., Ståhl S., Herne N. J. Biotechnol. (2007) 128:162–183.[CrossRef][Web of Science][Medline]
Gulich S., Uhlen M., Hober S. J. Biotechnol. (2000) 76:233–244.[CrossRef][Web of Science][Medline]
He Y., Rozak D.A., Sari N., Chen Y., Bryan P., Orban J. Biochemistry (2006) 45:10102–10109.[CrossRef][Web of Science][Medline]
Johansson M.U., et al. J. Biol. Chem. (2002) 277:8114–8120.
Klooster R., Maassen B.T., Stam J.C., Hermans P.W., Ten Haaft M.R., Detmers F.J., de Haard H.J., Post J.A., Theo Verrips C. J. Immunol. Methods (2007) 324:1–12.[CrossRef][Web of Science][Medline]
Konig T., Skerra A. J. Immunol. Methods (1998) 218:73–83.[CrossRef][Web of Science][Medline]
Kronvall G., Jonsson K. J. Mol. Recognit. (1999) 12:38–44.[CrossRef][Web of Science][Medline]
Lejon S., Frick I.M., Bjorck L., Wikstrom M., Svensson S. J. Biol. Chem. (2004) 279:42924–42928.
Linhult M., Binz H.K., Uhlen M., Hober S. Protein Sci. (2002) 11:206–213.[CrossRef][Web of Science][Medline]
Linhult M., Gulich S., Graslund T., Simon A., Karlsson M., Sjoberg A., Nord K., Hober S. Proteins (2004) 55:407–416.[CrossRef][Web of Science][Medline]
Linhult M., Gulich S., Hober S. Protein Pept. Lett. (2005) 12:305–310.[CrossRef][Web of Science][Medline]
Makrides S.C., Nygren P.A., Andrews B., Ford P.J., Evans K.S., Hayman E.G., Adari H., Uhlen M., Toth C.A. J. Pharmacol. Exp. Ther. (1996) 277:534–542.
Nguyen A., Reyes A.E. 2nd, Zhang M., McDonald P., Wong W.L., Damico L.A., Dennis M.S. Protein Eng. Des. Sel. (2006) 19:291–297.
Nilsson B., Moks T., Jansson B., Abrahmsen L., Elmblad A., Holmgren E., Henrichson C., Jones T.A., Uhlen M. Protein Eng. (1987) 1:107–113.
Nord K., Nilsson J., Nilsson B., Uhlen M., Nygren P.A. Protein Eng. (1995) 8:601–608.
Nord K., Gunneriusson E., Ringdahl J., Stahl S., Uhlen M., Nygren P.A. Nat. Biotechnol. (1997) 15:772–777.[CrossRef][Web of Science][Medline]
Nygren P.A., Eliasson M., Abrahmsen L., Uhlen M., Palmcrantz E. J. Mol. Recognit. (1988) 1:69–74.[CrossRef][Medline]
Nygren P.A., Ljungquist C., Tromborg H., Nustad K., Uhlen M. Eur. J. Biochem. (1990) 193:143–148.[Web of Science][Medline]
Nygren P.-Å., Flodby P., Andersson R., Wigzell H., Uhlén M. In Vivo Stabilization of a Human Recombinant CD4 Derivative by Fusion to a Serum Albumin-Binding Receptor (1991) New York: Cold Spring Harbor Laboratory Press.
Peters T. Jr. Adv. Protein Chem. (1985) 37:161–245.[Web of Science][Medline]
Rozak D.A., Alexander P.A., He Y., Chen Y., Orban J., Bryan P.N. Biochemistry (2006) 45:3263–3271.[CrossRef][Web of Science][Medline]
Stahl S., Nygren P.A. Pathol. Biol. (Paris) (1997) 45:66–76.[Medline]
Stahl S., Sjolander A., Nygren P.A., Berzins K., Perlmann P., Uhlen M. J. Immunol. Methods (1989) 124:43–52.[CrossRef][Web of Science][Medline]
Stemmer W.P. Nature (1994) 370:389–391.[CrossRef][Medline]
Steukers M., Schaus J.M., van Gool R., Hoyoux A., Richalet P., Sexton D.J., Nixon A.E., Vanhove M. J. Immunol. Methods (2006) 310:126–135.[CrossRef][Web of Science][Medline]
Stork R., Muller D., Kontermann R.E. Protein Eng. Des. Sel. (2007) 20:569–576.
Received January 28, 2008; revised April 8, 2008; accepted April 19, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Stork, E. Campigna, B. Robert, D. Muller, and R. E. Kontermann Biodistribution of a Bispecific Single-chain Diabody and Its Half-life Extended Derivatives J. Biol. Chem., September 18, 2009; 284(38): 25612 - 25619. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Ren, R. Zhang, Z. Liu, J. M. Webster, Z. Miao, S. S. Gambhir, F. A. Syud, and Z. Cheng A 2-Helix Small Protein Labeled with 68Ga for PET Imaging of HER2 Expression J. Nucl. Med., September 1, 2009; 50(9): 1492 - 1499. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||