Labeling of HER3 and HSA

Biotinylation of recombinant human HER3-Fc chimera (R&D Systems, Minneapolis, MN) was performed using the Biotin-XX Microscale Protein Labeling Kit (Invitrogen, Carlsbad, CA) according to the supplier’s recommendations. The concentration of labeled protein was determined using amino acid analysis. Recombinant human HER3 (ECD, Sino Biological Inc., Beijing, China) was conjugated with biotin carboxylic acid, succinimidyl ester in NaHCO₃ (0.1 M, pH 8.5) for 1.5 hours. Subsequently, glycine was added to stop the reaction followed by buffer exchange to PBS using a PD MiniTrap G-25 column (GE Healthcare, Uppsala, Sweden) according to manufacturer’s recommendations. Labeling of human serum albumin (HSA) with Alexa Fluor 647 succinimidyl ester (Invitrogen, Carlsbad, CA, USA) was performed according to supplier’s recommendations.

Alanine Scanning Mutagenesis

Vector pSCZ₀₅₄₁₆ [14] was used as template for construction of the thirteen mutants. PCR amplification was performed with oligonucleotides encoding alanine replacements at residues 9, 10, 11, 13, 14, 17, 18, 24, 25, 27, 32 and 35 of the HER3-binding Affibody molecule Z₀₅₄₁₆ [14]. Amino acid A28 in the Affibody molecule was substituted to a valine by the same means. The PCR products were digested with NheI and XhoI restriction enzymes (New England Biolabs, Beverly, MA, USA) and ligated to the staphylococcal display vector pSCZ1 [25], digested with the same enzymes, using T4 DNA ligase (New England Biolabs) according to supplier’s recommendations. The E. coli strain RR1ΔM15 was used as host for plasmid construction and preparation was performed with a JETSTAR Kit (Genomed, Bad Oeynhausen, Germany) according to the supplier’s recommendations. BigDye Thermo Cycle Sequencing reactions and an ABI Prism 3700 instrument (Applied Biosystems, Foster City, CA) were used to verify the sequences. The constructs were transformed to electrocompetent S. carnosus TM300, according to previously described protocol [26].

Flow-cytometric Analysis of Alanine Mutants

Staphylococcal cells displaying the respective mutants were inoculated to 10 ml tryptic soy broth supplemented with yeast extract (TSB+YE; Merck, Darmstadt, Germany) and 10 µg/ml chloramphenicol and grown overnight at 37°C and 150 rpm. 10⁶ cells from overnight cultures were washed with 800 µl phosphate-buffered saline supplemented with 0.1% Pluronic® F108 NF Surfactant (PBSP; pH 7.4; BASF Corporation, Mount Olive, NJ). The cells were pelleted by centrifugation (15000 g, 4°C, 6 min) and resuspended in 50 µl of PBSP containing 5 nM biotinylated HER3-Fc (R&D systems). The cells were incubated at room temperature for 2 hours with gentle mixing. The cells were washed with 180 µl ice-cold PBSP, followed by incubation in 200 µl ice-cold PBSP containing Streptavidin, Alexa Fluor 488 conjugate (Invitrogen, Carlsbad, CA) and 150 nM Alexa Fluor 647-HSA conjugate for 40 min in the dark. Following one wash with 180 µl ice-cold PBSP, cells were resuspended in 200 µl ice-cold PBSP prior to flow-cytometric analysis. The mean fluorescence intensity (MFI) was measured using an FACS Vantage SE (BD Biosciences, San Jose, CA) flow cytometer. The experiment was carried out in duplicate on different days using freshly prepared solutions.

Grafting of Z05416 to an Optimized Affibody Scaffold

The gene sequence encoding the amino acids in positions 9, 10, 11, 13, 14, 17, 18, 24, 25, 27, 28, 32 and 35 of the HER3-binding Affibody molecule Z₀₅₄₁₆ was grafted to an optimized Affibody scaffold [27]. The gene sequence was digested with SacI and XhoI restriction enzymes (New England Biolabs) and ligated to a staphylococcal display vector (digested with the same enzymes) using T4 DNA ligase (New England Biolabs) according to supplier’s recommendations. Plasmid construction and preparation, as well as transformation to electrocompetent S. carnosus TM300, was performed as described above.

Library Design and Cloning

A SlonoMax library was obtained from Sloning BioTechnology GmbH (Pucheim, Germany), encoding helix 1 and 2 of the Affibody molecule with a XhoI and a SacI site for subcloning to the vector (5′-CTC GAG GCG GAA GCC AAA TAC GCC AAA GAA XXX XXX XXX GCG XXX XXX GAG ATC XXX XXX TTA CCT AAC TTA ACC XXX XXX CAA XXX XXX GCC TTC ATC XXX AAA TTA XXX GAT GAC CCA AGC CAG AGC TCT-Ć; X denotes a randomized nucleotide position). The library was PCR amplified in 11 cycles using Phusion DNA polymerase (Finnzymes, Espoo, Finland) followed by purification of the PCR product using QIAquick PCR purification kit (Qiagen GmbH, Hilden, Germany). Subsequently, the library oligos were digested by XhoI and SacI-HF restriction enzymes (New England Biolabs) and purified by preparative gel electrophoresis (2% agarose gel) using QIAquick gel extraction kit (Qiagen GmbH). A modified version of the S. carnosus expression vector pSCZ1 [28] was restricted by XhoI and SacI-HF enzymes and purified by preparative gel electrophoresis as described above. The library oligos were ligated into the vector using T4 DNA ligase (New England Biolabs) at a 1∶5 molar ratio of vector to insert, followed by phenol-chloroform extraction and ethanol precipitation for purification and concentration of DNA fragments. Next, the library-encoding plasmids were transformed into electrocompetent E. coli SS320 by electroporation and individual clones where sequenced for library validation by BigDye Thermo Cycle Sequencing using an ABI Prism 3700 instrument (Applied Biosystems). The library plasmids were subsequently isolated using a Jetstar Maxi Kit (Genomed, Bad Oeynhausen, Germany), purified by phenol-chloroform extraction and concentrated by isopropanol precipitation. Finally, the library (hereafter denoted Sc:Z_HER3LIB2) was transformed by electroporation into electrocompetent S. carnosus as previously described [26].

Cell Labeling and FACS

Sc:Z_HER3LIB2 was inoculated in TSB+YE supplemented with chloramphenicol (10 µg/ml) for overnight growth at 37°C and 150 rpm. The following day, cells were harvested by centrifugation (15000 g, 6 min, 4°C) and washed in PBSP before addition of biotinylated HER3. Cells were incubated with gentle mixing at room temperature prior to washing with ice-cold PBSP and labeling with Streptavidin, R-Phycoerythrin conjugate (SAPE; Invitrogen) as well as HSA conjugated with Alexa Fluor 647 for 30 minutes on ice at a concentration of 5 µg/ml and 300 nM, respectively. Cells were washed and resuspended in ice-cold PBSP. The library was labeled and sorted in four rounds using a MoFlo Astrios flow cytometer (Beckman Coulter, Indianapolis, IN). In sort 1 and 2, the library was labeled with 5 nM of biotinylated HER3. In sort 3 and 4, the library was subjected to off-rate selections. The library was first labeled with biotinylated HER3. After incubation and washing to remove unbound biotinylated HER3, the library was incubated with 5 nM of non-biotinylated HER3 (for 1 hour at room temperature in sort 3 and for 4 hours at room temperature in sort 4) prior to labeling with SAPE and HSA-Alexa Fluor 647. For each round of sorting, approximately ten times the library size was analyzed in the flow cytometer and the top fraction of cells (approximately 0.1–0.5%), with high HER3-binding to cell surface expression ratio, was gated and sorted into an eppendorf tube with TSB+YE. Subsequently, sorted cells were inoculated in TSB+YE with chloramphenicol (10 µg/ml) for overnight amplification prior to the next sorting round. Finally, isolated cells were spread on agar plates containing chloramphenicol and individual colonies were picked for BigDye Thermo Cycle Sequencing reactions using an ABI Prism 3700 instrument (Applied Biosystems, Foster City, CA).

On-cell Screening for HER3 Binding

Individual staphylococcal clones from the sorting were inoculated in TSB+YE with chloramphenicol (10 µg/ml) and grown overnight at 37°C and 150 rpm. Cells were pelleted by centrifugation and washed in PBSP before resuspension in 0.5 nM and 2 nM of biotinylated HER3, respectively. After 1-hour incubation at room temperature with gentle mixing, cells were washed with ice-cold PBSP and labeled with SAPE at a concentration of 5 µg/ml and HSA-Alexa Fluor 647 at a concentration of 300 nM for 30 minutes on ice. Finally, cells were washed and resuspended in ice-cold PBSP. All samples were ranked based on the ratio between mean fluorescence intensities (MFI) from HER3-binding and cell surface expression signals in a Gallios flow cytometer (Beckman Coulter). In addition, the original HER3-binder Z₀₅₄₁₇ was analyzed for comparison.

Expression and Purification of Soluble Affibody Molecules

Ten different HER3-binding Affibody molecules were produced and purified for further characterization. The Affibody-encoding DNA sequences were amplified from colonies by PCR, using primers that introduced NdeI and XhoI restriction sites. The Affibody sequences were subsequently inserted into the NdeI-XhoI restricted expression vector pET26b+ (Novagen, Merck KGaA, Darmstadt, Germany), generating monomeric Affibody constructs with a C-terminal His₆-tag (H₆). Rosetta (DE3) E. coli cells were transformed with the plasmids by standard heat-shock treatment. Cells were cultured in TSB+YE at 37°C and Affibody expression was induced by the addition of IPTG to a final concentration of 1 mM when the OD_{600 nm} had reached approximately 1. After incubation over night at 25°C, the cells were harvested by centrifugation at 6000 g for 8 minutes at 4°C. The cell pellets were resuspended in lysis buffer (7 M Guanidinium chloride, 47 mM Na₂HPO₄, 2.65 mM NaH₂PO₄, 10 mM Tris-HCl, 100 mM NaCl) and incubated for 2 hours at 37°C and 150 rpm. Subsequently, cell debris was removed by centrifugation at 4°C. The supernatant was isolated and Affibody molecules were purified by IMAC using a HisPur™ Cobalt resin (Thermo Scientific, Rockford, USA) under denaturing conditions. Buffer was exchanged to PBS by dialysis using Slide-A-Lyzer dialysis cassettes, 3.5 kDa cut-off (Thermo Scientific). The molecular weight and the purity of the purified Affibody molecules were verified by LC/MS (Agilent Technologies 6520 ESI-Q-TOF) and SDS-PAGE respectively. The protein concentration was determined by spectrophotometric measurement at 280 nm.

Biosensor Off-rate Ranking

All biosensor assays were performed on a ProteOn XPR36 instrument (Bio-Rad Laboratories, CA, USA) using PBST as running buffer and 15 mM NaOH for regeneration. In all experiments, subtraction of responses from each sample over a blank surface was performed to minimize buffer contributions. Recombinant human and murine HER3-Fc (R&D systems) respectively, was immobilized on a GLC chip (Bio-Rad Laboratories) using standard sulfo-NHS/EDAC amine coupling chemistry. The ligands were diluted to a final concentration of 10 µg/ml in NaAc (10 mM, pH 4.5) and final immobilization levels were approximately 3000 RU, respectively. The purified Affibody molecules were injected at a concentration of 25 nM over the immobilized ligands at a flow rate of 100 µl/min. Association and dissociation time was 120 and 1800 seconds, respectively.

Biosensor Analysis of HER3-binding before and after Heat Treatment

Affibody molecules, at a concentration of 25 and 50 nM, were subjected to heat treatment at 90°C for 15 minutes. Binding to HER3 was evaluated by injection of 400 µl of each sample before and after heat treatment over immobilized human and murine HER3 at a flow rate of 100 µl/min.

Circular Dichroism Spectroscopy

Affibody molecules (0.2 mg/ml) were subjected to circular dichroism at 195–250 nm and 20°C using a Jasco J-810 spectropolarimeter (Jasco Scandinavia AB, Mölndal, Sweden) in a cell with an optical path-length of 1 mm. In addition, the ellipticity at 221 nm was measured for each Affibody molecule while heating the sample from 20 to 90°C (5°C/min).

Biosensor Affinity Determination

For determination of the equilibrium dissociation constant (K_D), a GLC sensor chip (Bio-Rad Laboratories) was immobilized with human HER3-Fc (R&D systems) as described above, with a final immobilization level of approximately 650 RU. Duplicate injections of a two-fold dilution series of each Affibody molecule ranging from 1.8 to 7.2 nM for Z₀₈₆₉₈ and 1.7 to 6.8 nM for Z₀₈₆₉₉, was injected over immobilized human HER3 (protein concentrations were determined by amino acid analysis). The flow rate was set to 50 µl/min and the association and dissociation was followed for 300 seconds and 4 hours, respectively.

Materials for in vitro and in vivo Assays

High-quality Milli-Q water (resistance higher than 18 ) was used for preparing solutions. ^99mTc was obtained as pertechnetate from an Ultra-TechneKow generator (Covidien, Dublin, Ireland) by elution with sterile 0.9% NaCl. IsoLink labeling kits were kindly provided by Covidien. NAP-5 size exclusion columns were from GE Healthcare. Cells used during in vitro experiments were detached using trypsin-EDTA solution (0.25% trypsin, 0.02% EDTA in buffer, Biochrom AG, Berlin, Germany). Radioactivity was measured using an automated gamma-counter with a 3-inch NaI(Tl) detector (1480 WIZARD, Wallac Oy, Turku, Finland). The purity of radiolabeled Affibody molecules was determined by radio-ITLC (150–771 DARK GREEN, Tec-Control Chromatography strips from Biodex Medical Systems, New York, USA) and cross-validated by SDS PAGE. The distribution of radioactivity along the thin layer chromatography strips was measured on a Cyclone™ Storage Phosphor System and analyzed using the OptiQuant™ image analysis software (PerkinElmer). Data on cellular uptake and biodistribution were assessed by an unpaired, two-tailed t-test using GraphPad Prism (version 4.00 for Windows GraphPad Software, San Diego California USA) in order to determine significant differences (p<0.05).

Labeling and in vitro Cell Binding of ^99mTc(CO)3-Z₀₈₆₉₈-H₆ and ^99mTc(CO)3-Z₀₈₆₉₉-H₆

Labeling of Z₀₈₆₉₈-H₆ and -Z₀₈₆₉₉-H₆ with [^99mTc(CO)₃]⁺ was performed as described by Orlova and co-workers [29]. Briefly, 500 µl (200–320 MBq) of ^99mTcO₄⁻ -containing generator eluate were added to a vial with the IsoLink kit. The mixture was incubated at 100°C during 20 min. Thereafter, 40 µl of mixture were transferred to a tube containing 50 µg (∼6.8 nmol) Affibody molecule in 40 µl PBS and incubated at 50°C for 60 min. The labeling yield was measured by ITLC. When the ITLC strips are eluted with PBS, pertechnetate, as well as carbonyl and histidine complexes of ^99mTc migrate with the eluent front (R_f = 1.0), while Affibody molecules do not migrate under these conditions (R_f = 0.0). To determine the presence of reduced hydrolyzed technetium, ITLC was eluted with pyridine:acetic acid:water (5∶3:1.5). When this eluent is used, the technetium colloids stay at the application point (R_f = 0.0), while radiolabelled Affibody molecules, as well as pertechnetate and carbonyl complexes of ^99mTc migrate with the solvent front (R_f = 1.0). In addition, blank experiments were performed, where the Affibody molecules were omitted. The labeled Affibody molecules were purified using NAP-5 desalting columns, pre-equilibrated and eluted with PBS. The purity of each preparation was evaluated using ITLC.

The specificity of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ for binding to HER3-expressing cells was evaluated using LS174T colorectal carcinoma, NCI-N87 gastric carcinoma, MCF-7 breast carcinoma, LNCaP and DU-145 prostate cancer cell lines (American Type Tissue Culture Collection, ATCC, via LGC Promochem, Borås, Sweden.) An in vitro specificity test was performed according to the methods described earlier [30]. Briefly, a solution of radiolabelled Affibody molecules (at 1 nM) was added to 6 Petri dishes (around 2×10⁶ cells in each). For blocking, 0.7 µM of non-labeled Affibody molecule was added 15 min before radiolabeled conjugates to saturate the receptors. The cells were incubated during one hour in a humidified incubator at 37°C. Thereafter, the media was collected, the cells were detached by trypsin-EDTA solution, and the radioactivity in cells and media was measured to enable calculation of the fraction of cell-bound radioactivity.

In vivo Studies

All animal experiments were planned and performed in accordance with national legislation on laboratory animals’ protection and were approved by the Ethics Committee for Animal Research in Uppsala.

The goal of the biodistribution experiment in normal NMRI mice was to test specificity of in vivo accumulation of radiolabeled Affibody molecules in organs where HER3 is expressed (lung, liver, stomach, small intestines, salivary gland), as well as to study biodistribution of Affibody molecules. To evaluate if uptake in these organs was saturable, ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ Affibody molecule (65 kBq in 100 µl PBS per mouse) was intravenously injected in a group of four female NMRI mice (average weight 24.5±1.6 g). The injected protein dose was adjusted by dilution with non-labeled Affibody molecule to 1 µg (ca. 0.13 nmol) or 10 µg (ca. 1.3 nmol) per mouse. At 4 h pi, four mice were sacrificed by injection of a lethal dose of anesthesia (20 µl of Ketalar-Rompun per gram body weight: Ketalar (50 mg/ml, Pfizer), 10 mg/ml; Rompun, (20 mg/ml, Bayer) followed by heart puncture and exsanguination with a syringe rinsed with heparin (5000 IE/ml, Leo Pharma). Samples of blood, lung, liver, spleen, stomach, small intestines, kidney, uterus, salivary gland, muscle and bone were collected, weighed and their radioactivity was measured. ^99mTc radioactivity was measured in the energy range of 100–160 keV. The data were corrected for background. Tissue uptake was calculated as percent of injected radioactivity per gram (% IA/g). Radioactivity in carcass was calculated as % IA per whole sample.

To evaluate targeting of HER3-expressing tumors in vivo, mice bearing prostate cancer xenografts were used. LNCaP cells (around 6×10⁶) were implanted in right hind leg of male BALB/C nu/nu mice in 50% Matrigel. The biodistribution experiments were performed 4 weeks after implantation, when tumor weight was 0.8±0.4 g. The average animal weight was 20.1±0.6 g at the time of experiment. To evaluate if uptake in these organs was saturable, ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ (85 kBq in 100 µl PBS per mouse) was intravenously injected in a group of three mice. The injected protein dose was adjusted by dilution with non-labeled Affibody molecule to 0.1 µg (0.013 nmol) or 10 µg (1.3 nmol) per mouse. The animals were sacrificed at 6 h after injection, and biodistribution was measured as described above.

Inhibition of Cell Proliferation

The cell line MCF-7 was propagated as recommended in complete medium (RPMI 1640 medium with L-glut (Lonza, Basel, Switzerland)) supplemented with sodium pyruvate (Lonza), non-essential amino acids (Lonza), penicillin/streptomycin (Lonza) +10% fetal calf serum (FCS) (Gibco, Invitrogen). At the day of experiment, cells were washed twice in RPMI 1640 without supplements and resuspended in assay medium (RPMI 1640 medium with L-glut containing sodium pyruvate, non-essential amino acids, penicillin/streptomycin, +2% dialyzed FCS (Gibco)). The ability of the Affibody molecules (Z₀₈₆₉₈, Z₀₈₆₉₉ and Z₀₅₄₁₇) to block heregulin-induced proliferation was analyzed by mixing the binders with 200 pM heregulin (NRG1-β1/heregulin1-β1 EGF domain, R&D Systems) in assay medium. The molecules were titrated in a 5-fold dilution series with a fixed concentration of heregulin (200 pM). Furthermore, in an additional dilution series for each Affibody molecule, 9 µM recombinant human serum albumin (HSA, Novozymes, Bagsvaerd, Denmark) was added to the assay medium. The titration was performed in 96-well cell culture plates in a volume of 100 µl. 1500 cells were added per well (100 µl) and plates were incubated at 37°C, 5% CO₂ for five days. After incubation, determination of the number of living cells in each well was performed using cell counting kit-8 (CCK-8, Fluka, Sigma Aldrich). 19 µl of CCK-8 cell proliferation reagent diluted two-fold in RPMI 1640 medium was added per well and absorbance was measured after 4 hours at 450 nm using a microplate reader (Victor3, Perkin Elmer). The data on cell growth was assessed by non-linear regression, and the half maximal inhibitory concentration (IC50) was determined using GraphPad Prism (GraphPad Software, San Diego California USA).

Inhibition of Receptor Phosphorylation

Inhibition of receptor phosphorylation was essentially performed as described earlier [15] with a fixed concentration of heregulin (4 nM) and Affibody molecule concentration of 40 and 400 nM.

Alanine Scanning

Alanine scanning mutagenesis was used to determine the contributions from individual amino acids of an Affibody molecule in binding to the extracellular domain of HER3. The thirteen originally randomized residues in the Z-scaffold were substituted to alanine (except for position 28 that was mutated from alanine to valine), respectively, and each construct was subcloned into a staphylococcal display vector for subsequent transformation and expression on the staphylococcal host. Thirteen clones of staphylococcal cells, each displaying one of the thirteen mutants, were incubated with labelled HER3-Fc and the relative impact of each mutation on HER3-binding was analyzed using flow cytometry. Variants with alanine mutations at positions 9, 10, and 17 demonstrated substantially lower signals compared to the control, indicating a reduced affinity for HER3 and that the corresponding original residues were contributing to the interaction (Figure 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Alanine scanning and library design.

A. Alanine scanning of Z₀₅₄₁₆. Histogram showing the results from flow-cytometric analysis of the thirteen staphylococcal-displayed mutants. The residues in the Affibody molecule that were mutated are represented on the x-axis, and a ratio of the mean fluorescence intensity (MFI) corresponding to HER3 binding and the MFI corresponding to surface expression level (monitored by HSA binding) is represented on the y-axis. B. Library design. Schematic overview showing the randomized positions in helix 1 and 2, the relative importance of the respective residue for HER3 binding (as determined by alanine scanning), the amino acids in each position in the original HER3 binders, the percentage of the codon corresponding to the original amino acid and the percentage of each of 15 or 16 (depending on the position) mutation codons.

https://doi.org/10.1371/journal.pone.0062791.g001

Grafting of Z₀₅₄₁₆ to a New Affibody Scaffold

A second-generation Affibody scaffold with improved properties was recently described [27]. The optimized scaffold contains eleven amino acid substitutions in the non-binding surface and demonstrates an increased thermal and chemical stability, as well as more efficient chemical synthesis [27]. We sought to base our maturation library on this improved scaffold. However, prior to synthesizing the affinity maturation library, we wanted to confirm that the HER3-specific Affibody molecule was compatible with the eleven mutations in the framework. The selected sequence of the HER3-specific Affibody molecule (Z₀₅₄₁₆) was therefore grafted into the optimized scaffold and the new variant was subcloned to a staphylococcal display vector and expressed on staphylococci. Staphylococcal cells displaying the grafted Affibody variant as well as the original Affibody molecule (Z₀₅₄₁₆) were incubated with labeled HER3-Fc and analyzed using flow cytometry. The HER3-specific Affibody molecule that was grafted to the optimized Z-scaffold demonstrated a similar HER3-binding signal as well as cell surface expression as the binder in the old scaffold, thus indicating full compatibility (data not shown).

Library Design

The affinity maturation library was based on the amino acid sequences of the parental Z₀₅₄₁₆ and Z₀₅₄₁₇ as well as the results from the alanine scanning. The intention of the library design was to mimic the mutagenesis output from an error-prone PCR amplification, but restrict the mutations to 13 surface-exposed residues on helix 1 and 2. In order to achieve this, each randomized position was spiked with the codon that was encoding the original amino acid in addition to a mix of mutation codons that was encoding the rest of the amino acids (except aspartic acid, cysteine and proline). We decided to aim at an average mutation frequency of three mutations per molecule. Since we targeted thirteen positions for randomization, the average proportion of mutation codons in each position was therefore 3/13 • 100 = 23%, and thus an average proportion of 77% of the original codon. The theoretical binomial distribution of mutation frequencies in the library is shown in Figure S1A. The production of the randomized DNA was performed using the Slonomics technology from former Sloning BioTechnology GmbH (now part of the MorphoSys Group), which is similar to trinucleotide synthesis and enabled the design of the randomizations without biases associated with degenerative codons. To potentially increase the functionality of the final library, the mutation frequency in each position was normalized with the corresponding alanine scanning data, i.e. a position that was demonstrated to be important for the interaction was mutated at a lower frequency and vice versa. The final library design is outlined in Figure 1B.

Library Construction

The library was subcloned into a staphylococcal surface display vector and transformed into S. carnosus, yielding approximately 6.7 • 10⁷ individual clones. Theoretical calculations on library diversity, library coverage and combinatorics of the library design are shown in Figure S1B. The library was analyzed by DNA sequencing, which verified an amino acid distribution in acceptable accordance with the library design and a low percentage of frame shifts (1.8%, data now shown). However, the mutation frequency was somewhat lower than the theoretical design, with an average of approximately 2 mutations per clone (data not shown). Flow-cytometric analysis of Sc:Z_HER3LIB2 revealed a high fraction of full-length library-expressing clones (around 80%), as shown by the HSA-binding signal, and demonstrated retained binding of a fraction of library clones to HER3 (Figure 2), which was expected due to the relatively moderate mutation frequency in the library.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Fluorescence-activated cell sorting of Sc:Z_HER3LIB2.

Density dot plots of the affinity maturation library displayed on S. carnosus, showing the HER3-binding signal on the y-axis and the surface expression level on the x-axis. The dot plots show cells from the unsorted library as well as cells isolated from the first, second, third and fourth sorting round, respectively. For comparison of HER3-binding signals, a dot plot showing Z₀₅₄₁₇ is included.

https://doi.org/10.1371/journal.pone.0062791.g002

Flow-cytometric Sorting of Affinity Maturation Library

Fluorescence-activated cell sorting (FACS) is an efficient method for isolation of cells with desirable traits from combinatorial libraries. As we have shown previously, flow-cytometric sorting in combination with staphylococcal display enables identification of high-affinity binders towards many different targets of interest [14], [28]. In order to isolate Affibody molecules with high affinity towards HER3, Sc:Z_HER3LIB2 was sorted in four rounds against biotinylated HER3. Detection of receptor binding in each sorting cycle was performed using Streptavidin-conjugated R-Phycoerythrin and the target-binding signal was normalized against the surface expression level through labeling of cells with HSA-Alexa Fluor 647 conjugate. In each sorting round, the top fraction of cells, showing high HER3 binding compared to surface expression level, was isolated and amplified by overnight growth prior to the next round of sorting. The selection stringency was modified throughout the selection process by changing the sorting parameters and gates, and also by decreasing the target concentration as well as incorporating off-rate selections in later sorting rounds. As shown by flow-cytometric analysis after each selection cycle, the sorting procedure resulted in successful enrichment of HER3-binding clones (Figure 2). In addition, enrichment of binders with improved HER3-binding signals compared to the original binder Z₀₅₄₁₇ could be observed throughout the sorting. After the third and the final sorting round, individual colonies were sequenced for identification, yielding 107 unique sequences out of 272 total reads. In summary, these results suggested a successful library design and selection of affinity-matured Affibody molecules targeting HER3.

On-cell Affinity Ranking

After the sorting of Sc:Z_HER3LIB2 by FACS, isolated clones were analyzed individually in order to identify Affibody molecules with the highest affinity towards HER3. Altogether 40 unique clones from both the third and the fourth sorting round were affinity-ranked by flow cytometry, by determining the ratio between the HER3 binding signal and the surface expression level (FL-2/FL-6). The vast majority of analyzed clones showed improved affinity towards HER3 compared to the original HER3-binder Z₀₅₄₁₇ (data not shown) and ten top candidates (hereafter denoted Z₀₈₆₉₄–Z₀₈₇₀₃) were subcloned and purified from E. coli cell extracts as C-terminally His₆-tagged Affibody molecules.

Biosensor Off-rate Ranking

All ten purified Affibody molecules were subjected to an off-rate ranking by biosensor analysis in order to further narrow down the number of HER3-binding variants to those demonstrating the highest affinity towards the receptor. All ten Affibody variants were injected over a sensor chip surface immobilized with human HER3-Fc and dissociation was followed during 30 minutes. All affinity-matured Affibody molecules generated similar binding curves to HER3 with significantly reduced off-rates compared to the original HER3-specific Affibody (Figure 3A). For further evaluation of the HER3-specific Affibody molecules, we focused on the two candidates, Z₀₈₆₉₈ and Z₀₈₆₉₉, which demonstrated the slowest off-rates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Characterization of affinity-matured Affibody molecules.

A. Biosensor off-rate ranking of ten affinity-matured HER3-binders. Sensorgrams of ten different Affibody molecules injected over immobilized human HER3-Fc (all affinity-matured clones are shown in black, except for Z₀₈₆₉₈ and Z₀₈₆₉₉, which are shown in blue and red, respectively). For comparison, the HER3-binder Z₀₅₄₁₇ was included in the analysis (grey curve). B. Analysis of refolding capacity and thermal stability of affinity-matured HER3-binders evaluated by SPR and CD. Sensorgrams showing injections of 50 nM Z₀₈₆₉₈ and Z₀₈₆₉₉, respectively, before and after heat treatment at 90°C over immobilized human HER3-Fc. C. Variable temperature measurement (VTM) spectra obtained at 221 nm while heating the HER3-specific Affibody molecules Z₀₈₆₉₈ and Z₀₈₆₉₉ from 20 to 90°C. D. CD spectra of Z₀₈₆₉₈ and Z₀₈₆₉₉ at wavelengths ranging from 195 to 250 nm at 20°C.

https://doi.org/10.1371/journal.pone.0062791.g003

Biosensor Analysis of HER3-binding before and after Heat Treatment

Therapeutic effect of the HER3-specific Affibody molecules could potentially be mediated by the ability to block heregulin-induced tumor growth, but as an alternative, the small and stable Affibody molecules could easily be conjugated with different payloads, such as radionuclides. Radionuclide labeling is also employed when the binders are used as tracers in molecular imaging. Such labeling often requires heating of the protein to high temperatures, which motivated the evaluation of the refolding capacity of the affinity-matured Affibody molecules. Therefore, the interaction between HER3 and the Affibody molecules (Z₀₈₆₉₈ and Z₀₈₆₉₉) before and after heat treatment at 90°C was evaluated by SPR technology. The Affibody molecules, before and after heating, were injected over immobilized human and murine HER3-Fc for comparison of generated sensorgrams. As shown in Figure 3B, the sensorgrams from both Affibody molecules overlapped before and after heat treatment, demonstrating that the binding capacity of the Affibody molecules was unaffected by the thermal denaturation and refolding. The same was observed for binding to the murine HER3 receptor before and after heating (data not shown). These results were encouraging but not surprising due to the generally high solubility and efficient refolding of Affibody molecules [16].

Circular Dichroism Spectroscopy

For further investigation of the thermal stability, the Affibody molecules were subjected to circular dichroism spectroscopy. The melting temperatures of both binders were determined by measuring the ellipticity at 221 nm while heating the proteins from 20 to 90°C. The obtained melting curves corresponded to T_m values of 64 and 65°C for Z₀₈₆₉₈ and Z₀₈₆₉₉, respectively (Figure 3C), demonstrating improved thermal stabilities compared to the original HER3-specific Affibody molecules. Additionally, spectra at wavelengths ranging from 195–250 nm were obtained before and after the variable temperature measurement in order to assess the reversibility of the unfolding after heat treatment (Figure 3D). The results showed that both affinity-matured HER3-binders maintained the alpha-helical content and completely refolded after heating to 90°C, as shown by the overlap between spectra generated before and after heating.

Biosensor Affinity Determination

The equilibrium dissociation constant (K_D) of the affinity-matured Affibody molecules was determined using biosensor technology. A dilution series of each binder was injected over immobilized human HER3-Fc and the rate constants of the interactions were determined by non-linear regression using a one-site binding model and used to calculate the K_D (data not shown). The obtained affinities were estimated to 50 and 21 pM for Z₀₈₆₉₈ and Z₀₈₆₉₉, respectively (Table 1), representing approximately a 33-fold affinity improvement of the strongest binder Z₀₈₆₉₉ compared to Z₀₅₄₁₇. However, due to that the observed on-rates (8.3 • 10⁵ and 1.9 • 10⁶ M⁻¹s⁻¹, respectively) and off-rates (4.1 • 10⁻⁵ and 3.9 • 10⁻⁵ s⁻¹, respectively) of Z₀₈₆₉₈ and Z₀₈₆₉₉ were close to the limit of accurate determination of the biosensor instrument, these values should be regarded as approximations. Still, these results are encouraging, demonstrating that the staphylococcal display system indeed is a powerful method for isolation of affinity molecules of low picomolar affinities.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Affinities of HER3-specific Affibody molecules determined by SPR.

https://doi.org/10.1371/journal.pone.0062791.t001

Labeling and in vitro Cell Binding of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆

To confirm that the new binders could target the HER3 receptor on cancer cell lines, the Affibody molecules were radiolabeled using ^99mTc(CO)₃ on the His-tag. The radiolabeling provided a yield of 43±6% for ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and 73±12% for ^99mTc(CO)₃-Z₀₈₆₉₉-H₆. Purification using disposable size-exclusion chromatography NAP-5 columns provided purity of 98.1±0.1% for ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and 99±1% for ^99mTc(CO)₃-Z₀₈₆₉₉-H₆. Binding specificity tests were performed to assess if the binding of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ to HER3-expressing cells was receptor-mediated. Saturation of the receptors by pre-incubation with non-labeled Affibody molecules significantly (p<0.05) decreased the binding of the radiolabeled Affibody molecule, which suggested that the binding was specific (Figure 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Specificity of binding of labeled Affibody molecules to HER3-expressing cells in vitro.

Cells were incubated with 1 nM of radiolabeled conjugates for 1 h. For pre-saturation (blocking) of antigens, 0.7 µM unlabeled non-radioactive Affibody molecules was added. Data are presented as mean values of percent of added radioactivity that is cell-bound from three cell dishes and standard deviations. The difference between uptake by non-blocked and blocked cells was statistically significant (p<0.05).

https://doi.org/10.1371/journal.pone.0062791.g004

In vivo Studies

The biodistribution of the two ^99mTc -labeled Affibody molecules at 4 h after injection in female NMRI mice is presented in Table 2. Both conjugates were rapidly cleared from blood and non-HER3-expressing tissues, such as bone and muscle. A high accumulation in kidneys indicated predominately renal excretion of the Affibody molecules. There was a significant difference in uptake of both Affibody molecules in HER3-expressing tissues (i.e. liver, stomach, small intestines and salivary gland) after injection of relatively low (1 µg, 0.13 nmol) compared with high (10 µg, 1.3 nmol) amount of the proteins. The radioactivity uptake in these tissues was lower after injection of higher protein amount (more unlabeled Affibody molecule relative to radiolabeled), suggesting that uptake was saturable and HER3-mediated. The concentration of radioactivity in blood was also lower after injection of the larger protein doses. In both cases, there was increase of renal radioactivity after injection of 10 µg of Affibody molecules. There were observable difference in biodistribution between ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ after injection of 1 µg. The blood concentration and uptake in lung, stomach, kidneys and muscle were significantly lower, but uptake in small intestines was significantly higher in the case of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆. At an injected dose of 10 µg, the only difference between two conjugates was a lower uptake of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ in uterus.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Biodistribution of ^99mTc(CO)₃-Z₀₈₆₉₈-H₆ and ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ in female NMRI mice 4 h after injection.

https://doi.org/10.1371/journal.pone.0062791.t002

The data concerning biodistribution of ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ in nude mice bearing LNCaP xenografts are presented in Table 3. In agreement with data for NMRI mice, increase of protein dose caused a reduction of the uptake in blood, liver and salivary gland. The tumor uptake did not change significantly, although there was a tendency to increased uptake with the higher protein dose. Increase of the protein dose resulted in significantly higher tumor-to-blood, tumor-to-salivary gland, tumor-to-liver and tumor-to-bone ratios.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Biodistribution of ^99mTc(CO)₃-Z₀₈₆₉₉-H₆ in male BALB/C nu/nu mice bearing LNCaP prostate cancer xenografts at 6 h after injection.

https://doi.org/10.1371/journal.pone.0062791.t003

Inhibition of Cell Proliferation and Receptor Phosphorylation

The half maximal inhibitory concentration (IC50) of the Affibody molecules in fusion to an affinity-matured and deimmunized albumin-binding domain (ABD) (based on ABD₀₃₅ for future potential in vivo half-life extension [31]) was determined in a heregulin-induced proliferation assay. A dilution series of each binder was mixed with a fixed concentration of heregulin and incubated with MCF-7 cells for five days, both in presence or absence of HSA, respectively. After incubation, viable cell concentration was measured using a colorimetric assay. The IC50 values were determined by non-linear regression of the data to a dose-response equation. In presence of HSA, the obtained IC50 values for Z₀₈₆₉₈ and Z₀₈₆₉₉ were around 0.15 nM and 1.2 nM, respectively, representing an improvement in growth inhibition efficiency of approximately 280-fold and 34-fold, respectively, compared to Z₀₅₄₁₇ (Table 4, Figure 5). Without HSA, the corresponding IC50 values were 0.9 nM and 1.2 nM for Z₀₈₆₉₈ and Z₀₈₆₉₉, respectively (Table 4, data not shown). To investigate potential improvements on inhibition of receptor signaling, an in vitro phosphorylation assay was conducted. The affinity-matured binders, Z₀₈₆₉₈ and Z₀₈₆₉₉, demonstrated a concentration dependent inhibition of receptor phosphorylation and a 90% inhibitory effect was obtained with the highest concentration (data not shown). There was hence an approximately 20-fold improvement in the blocking capacity with the affinity-maturated binders compared to the previously obtained data [15] for Z₀₅₄₁₆ and Z₀₅₄₁₇.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Inhibition of heregulin-induced cell proliferation.

MCF-7 cells were treated with heregulin, HSA and a titration series of respective Affibody molecule. Plots showing the normalized viable cell concentration (measured using a CCK-8 kit) on the y-axis and the Affibody concentration on the x-axis. A. Z₀₅₄₁₇ and negative control Z_Taq. B. Z₀₈₆₉₈ and negative control Z_Taq. C. Z₀₈₆₉₉ and negative control Z_Taq. D. Comparison of Z₀₅₄₁₇ and Z₀₈₆₉₈. The signals corresponding to the viable cell concentrations were normalized with the average of the signals from samples treated with the negative control Affibody molecule (Z_Taq). The experiment was performed in duplicate on different days using freshly prepared solutions and each sample was analyzed at least in triplicate in each experiment. Data is represented as means with indicated standard deviations.

https://doi.org/10.1371/journal.pone.0062791.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. IC50 values of HER3-specific Affibody molecules.

https://doi.org/10.1371/journal.pone.0062791.t004