Abbreviations and terminology

Abbreviations are appended to the full description of the chemicals when they first appear in the text, however it is important to note that the abbreviation PPEA-polyplex is a conjugate of polyethyleneglycol, polyethyleneimine, polyIC and the anti-EGFR affibody Z_EGFR1907’. When it is appropriate to emphasize that this polyplex contains polyIC, we also use the term PPEA-polyIC-polyplex for this conjugate. Z_{EGFR 1907’} is used as an abbreviation for Z_{EGFR 1907’} affibody.

The abbreviation sEGFR is used to describe extra-cellular fragments of the EGFR. We use two particular sEGFR fragments: residues 1–621 (EGFR_1–621) and residues 1–501 (EGFR_1–501).

We have analyzed cell lines with a wide range of EGFR expression; when cells have less than 5% of the level of receptors on DiFi cells they are described as low, cells with greater than 5 and less than 50% of the EGFR levels for Difi cells are described as moderate and cells with higher than 50% of the EGFR levels on Difi cells are described as high.

SDS-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of proteins

Samples of Ni-NTA–purified Z_EGFR-1907’ were diluted in SDS loading buffer with or without 1.43 M β-mercaptoethanol, heated to denature, and separated on NuPAGE 4–12% Bis-Tris gels (MES buffer; Invitrogen). Protein bands were stained with Coomassie Brilliant Blue R-250 (BioRad) (Fig 1Ai). The concentrated 1:1 Triconjugate Z_EGFR-1907’ affibody construct (see Synthesis of the LPEI-PEG-EGFR Affibody section below) was also analysed using SDS-PAGE. To visualise the sample bands, the gel was silver stained as follows: washed with 40 mM DTT, stained with 1 mg/ml AgNO₃, washed with de-ionized water, developed with 30 mg/ml Na₂CO₃ and 0.15% formaldehyde, reaction stopped with de-ionized water and gel fixed with 10% acetic acid (Fig 1C).

Mass spectrometry analysis

Protein bands responding to the Z_{EGFR 1907’}-affibody were extracted from a Coomassie stained SDS-PAGE gel and subjected to manual in-gel tryptic digestion. Briefly, the gel spots were de-stained and dehydrated with acetonitrile before being submitted to reduction with 50 mM TCEP solution (50 μL of 0.5 M TCEP / 125 μL of 200 mM TEAB / 325 μL de-ionized water) and alkylation with 100 mM Iodoacetamide solution in 50 mM TEAB buffer. Trypsin solution was prepared by adding 50 μL of 50 mM TEAB to each trypsin single vial (1 μg trypsin) and rehydrating dried gel plug/bands with approximately 10 μL trypsin working solution for 60 min at 4^oC. Excess trypsin solution was removed with a pipette before adding 35 μL of 50 mM TEAB to the sample vial to ensure proper hydration during digestion at elevated temperatures. Digestion was performed at 37^oC overnight. The digestion was stopped by adding 5 μL of 10% formic acid before collecting the supernatant. The generated tryptic peptides were concentrated to ~20 μL by centrifugal lyophilization for LC MS/MS analysis for peptide mapping mass spectrometry on LC/MS/MS.

Mass spectrometry was used to confirm protein purity, molecular mass and the protein sequence (Fig 1Aii). A spectrometer (Thermo Scientific) with a nanoESI interface in conjunction with an Ultimate 3000 RSLC nano HPLC (Dionex Ultimate 3000) was used. The LC system was equipped with an Acclaim Pepmap nanotrap column (Dionex-C18, 100 Å, 75 μm x 2 cm) and an Acclaim Pepmap RSLC analytical column (Dionex-C18, 100 Å, 75 μm x 15 cm). The tryptic peptides were injected into the enrichment column at an isocratic flow of 5 μL/min of 3% v/v CH3CN containing 0.1% v/v formic acid for 5 min applied before the enrichment column was switched in-line with the analytical column. The eluents were 0.1% v/v formic acid (solvent A) and 100% v/v CH3CN in 0.l% v/v formic acid (solvent B). The flow gradient was (i) 0–5 min at 3% B, (ii) 5–6 min, 3–6% B (iii) 6–18 min, 6–10% B (iv) 18–38 min, 10–30% B (v) 38–40 min, 30–45% B (vi) 40–42 min, 45–80% B (vii) 42–45 min at 80% B (viii) 45–46 min, 80–3% B and (ix) 46–53 min at 3% B. The LTQ Orbitrap Elite spectrometer was operated in the data-dependent mode with nano ESI spray voltage of 2.0 kV, capillary temperature of 250°C and S-lens RF value of 55%. All spectra were acquired in positive mode with full scan MS spectra scanning from m/z 300–1650 in the FT mode at 240,000 resolutions after accumulating to a target value of 1.0x10⁶. Lock mass of 445.120025 was used.

The top 20 most intense precursors were subjected to collision-induced dissociation (CID) with normalized collision energy of 30 and activation. The generated MGF files were uploaded to the Bio21 MSPF Pipeline (Bio21 Mass Spectrometry and Proteomics Facility Pipeline, http://proteomics.bio21.unimelb.edu.au/msile) and then searched using the MASCOT version 2.4.01 for algorithm against the murine Uniprot fasta database that included the added affibody fasta sequences (84562 sequences in total). The search parameters consisted of carbamidomethyl of cysteine as a fixed modification, NH2-terminal acetylation and oxidation of methionine as variable modifications. Up to three missed tryptic cleavage sites were allowed (trypsin/P) with a peptide mass tolerance of 20 ppm, a fragment mass tolerance of 0.8 Da, automatic decoy searches enabled with low decoy rate (< 5%), a significance threshold p < 0.05 and a Mascot score >30.

Biosensor analyses

SPR binding assays (Biacore S200, Cytiva) were performed to measure interactions of Z_{EGFR 1907’} and Cetuximab with sEGFR_1–501 and sEGFR_1–621. Proteins were immobilized on CM5 chips by amine coupling (pH 5.0) to ~1500 RU (sEGFR_1–501) and ~1700 RU (sEGFR_1–621). An initial pH scouting experiment was performed using 10 mM Acetate with pH 4.0, 4.5, 5.0, 5.5 and 6.0 for sEGFR proteins, sEGFR_1–501 and sEGFR_1–621. Samples were injected over the flow cell surface at 10 µL/min for 12 min. EGFR proteins were diluted in SPR buffer, 10 mM Hepes pH 7.4, 150 mM NaCl, 3 mM EDTA and 0.005%(v/v) Tween 20. EGFR proteins, sEGFR_1–501 and EGFR_1–621, were immobilized at 1500 RU and 1700 RU (resonance units) respectively to Series S Sensor Chip CM5 (Cytiva) by amine coupling at pH 5.0 (Table S1), according to the manufacturer’s instructions.

The conditioning cycle consisted of 0.1% SDS for 60 sec, 100 mM HCl for 60 sec, 10 mM NaOH for 60 sec and 50 mM glycine pH 2.5 for 60 sec. For binding kinetic studies, proteins were injected at 30 µL/min for 300 sec association over the sEGFR surfaces. Z_{EGFR 1907’} and Cetuximab concentrations ranged from 0 to 250 nM. This was followed by 800 sec dissociation phase. After each injection, the sensor surface was regenerated with SPR buffer supplemented with 50 mM glycine pH 2.5 for 120 sec between each cycle before repeating the sample injection. The dissociation equilibrium constant (K_D) was calculated using Biacore S200 Evaluation Software 1.1 (Cytiva). The curves were fitted using a local 1:1 binding model and a global heterogenous binding model.

The binding sites of the Z_{EGFR 1907’} and Cetuximab to sEGFR₁–₆₂₁ were explored. For A-B-A competitive binding assays, various concentrations and combinations of competitor sample solutions (B from A-B-A assay) and flanking competitor sample solutions (A from A-B-A assay) were tested. The assay consisted of 200 sec of flanking solution A, followed by 180 sec of sample solution B and further 60 sec of flanking solution A. Protein samples were pumped over the cell at a rate of 30 µL/min. The sensorgrams generated were subtracted from reference flow cell values. After each condition the flow cell was regenerated with SPR buffer supplemented with 50 mM glycine pH 2.5. A flow chart, summarizing the steps in the synthesis of the affibody-polyplex is presented in the supplementary information (Supplementary Figure S1). The details of each phase of the synthesis are documented in the sections below.

Synthesis of LPEI-PEG-OPSS thiol reactive co-polymer

Linear polyethyleneimine (LPEI) precursor, poly(2-ethyl-2-oxazoline), was synthesized following the methods published by Brissault and colleagues [35,36].

In brief, methyl p-toluenesulfonate (74 mg, 397 mmol) dissolved in 60 ml acetonitrile was added to 8 ml (79 mmol) of 2- ethyl-2-oxazoline. The reaction mixture was stirred under reflux for 2 hrs, during which time a precipitate formed. The crude product was dissolved in 100 ml of methylene chloride and precipitated in 400 ml of diethyl ether yielding 5.5 g of poly(2-ethyl-2-oxazoline) as a yellow powder. Synthesis of the LPEI (free base form) was prepared as described previously [36,37]. In brief, 5.5 g (0.11 mmol) of poly(2-thyl-2-oxazoline) was hydrolyzed with 68.8 ml of concentrated HCl (37%). The resulting air-dried LPEI hydrochloride salt was dissolved in 100 ml of water to yield 4 g of LPEI salt. This was made alkaline by adding 100 ml of 3 M NaOH. The lyophilised solid weighted ~1.25 g. LPEI-PEG_2k-OPSS diconjugates (1:1 and 1:3) were synthesized as previously described [37]. 174 mg (8 µmol) of LPEI in 2.7 ml of absolute EtOH was mixed with 79 mg of OPSS-PEG2k-CONHS (39.5 µmol) in 500 µl of anhydrous DMSO. The co-polymers conjugated with different molar ratios of PEG to LPEI were separated by cation exchange chromatography using a HR10/10 column packed with MacroPrep High S resin (BioRed).

Upon agitation for 3 hrs, 2 ml of 20 mM Hepes pH 7.4 were added to the reaction mixture and after few min, once the viscosity increased, the pH was adjusted to 7.4 with 1 M HCl. The column was equilibrated with 20 mM Hepes pH 7.4 buffer (buffer A). Separation was performed under a 20 mM Hepes pH 7.4; 3 M NaCl (buffer B) gradient. Sample was loaded at 1 ml/min for 7 min and for further 60 min with 0% buffer B. This was followed by isocratic elution at 20% buffer B until 76 min, followed by gradient elution up to 45% buffer B at 87 min and continuing at 45% buffer B until 110 min. Then, isocratic elution at 50% buffer B until 120 min was followed by gradual increase in gradient up to 100% buffer B until 180 min. Isocratic elution continued at 100% buffer B until 215 min, followed by a gradual decrease in gradient to 0% buffer B by 218 min. 2 ml fractions were collected between 0% − 50% Buffer A and 1 ml fractions were collected during the elution step (50% −100% buffer B). Sample separations were monitored at 220 nm, 280 nm and 343 nm, and all fractions from the three peaks were stored -20^oC.

Fractions from the 1:1 PEG to LPEI conjugation, which eluted during the 50% to 100% buffer B, were combined and desalted against 20 mM Hepes buffer pH 7.4 using a 20 ml HiTrap desalting column (4 x 5 ml Sephadex G25 columns). The purest fractions containing LPEI-free LPEI-PEG-OPSS were pooled and stored in the dark at -20^oC. A copper assay [38] was used to determine the concentration of di-conjugate 1:1, with molar ration of LPEI to PEG ~ 1:1. In brief, copper assay consisted of incubating copolymers (initially dissolved at 2x concentration with 100% methanol and then diluted to 1x with de-ionized water) with CuSO₄ (23 mg of CuSO₄.5H₂O in 100 ml of 0.1 M acetate buffer pH 5.4) for 20 min and measuring their absorbance at 285 nm.

Synthesis of the LPEI-PEG-EGFR affibody

The 1:1 diconjugate was mixed with EGFR affibody as described by Joubran et al. [37]. The reaction mixture was vortexed at ~800 rpm for 24 hrs and stored at -20^oC. Later, the reaction mixture was thawed and purified using cation exchange chromatography using a HR10/10 filled with MacroPrep High S resin (BioRad). 3 step gradient elution from 20 mM Hepes pH 7.4 reaching to 20 mM Hepes pH 7.4 with 3 M NaCl was used. Sample was loaded at 1 ml/min for 6 min and for further 10 min with 0% buffer B. Then isocratic elution with 33% buffer B until 15 min was followed by gradient elution reaching 66% buffer B at 30 min, isocratic elution from 30 min to 37 min, and gradient elution reaching 100% buffer B at 55 min. Finally, there was a gradual decrease down to 33% buffer B at 60 min. 1 ml fractions were collected between 0% − 100% Buffer B. Sample separation was monitored at 213 nm, 280 nm and 343 nm, and all fractions from the three peaks were stored at -20^oC. The 1:1 triconjugate fractions eluting between 25 min and 32 min (i.e., the last peak eluted by the gradient) were combined and desalted using the 4x5 ml Sephadex G25 (two runs) against PBS. Desalted fractions were combined, and 1 ml aliquots were frozen at -80^oC. The concentration of the 1:1 triconjugate was determined using the copper assay. The triconjugate fractions were pooled, concentrated and buffer exchanged to 20 mM Hepes pH 7.4, 5% glucose with RNase free de-ionized water (HBG) using a 3K MWCO Amicon Ultra15 Centrifugal filter unit. The presence of the affibody (~8 kDa) was confirmed under denaturing/reducing conditions (Fig 1C).

EGFR affibody polyIC-polyplex formation

The 1:1 Triconjugate affibody was complexed with poly (I:C) (Tocris, UK) in HBG buffer at a Nitrogen to Phosphate (N/P) ratio of 6 (where N = nitrogen from LPEI and P = phosphate from polyIC [37]). This ratio corresponds to a LPEI/polyIC weight ratio of 0.78 [39]. Both the 1:1 Triconjugate Affibody and polyIC were diluted up to twice the final volume using HBG buffer. Fresh polyplexes were made prior to the experiment by transferring equal volumes of 1:1 Triconjugate to the diluted polyIC and mixing by pipetting. The polyplexes were incubated at room temperature for 30 mins. The final 1:1 Triconjugate Affibody-polyIC-polyplex (Z_{EGFR 1907’-}polyIC-polyplex) was determined to be 100 µg/ml using the copper assay [36].

Polyplex size measurements

To determine the influence polyIC concentration has on the polyplexes formed, the particle size and its distribution were measured at 25^oC by dynamic light scattering using a Zetasizer µV (Malvern Instruments). The instrument was fixed with an 830 nm laser wavelength and the optic arrangement angle at which the measurements were performed was 90^o. Each polyplex sample was measured in triplicate.

Cell culture

EGFR-overexpressing A431 human epidermoid vulval carcinoma [40], SK-BR-3 [41], U87MG human glioma and its EGFR-overexpressing subline U87MGwtEGFR [42] cell lines were grown in Dulbecco’s modified Eagle’s medium (DMEM). U87MGwtEGFR cells were supplemented with geneticin sulfate (G418, Gibco, UK) at a final geneticin concentration of 500 µg/ml. MCF7, MDA-MB-231, MDA-MB-468 and BT-474 [43] cells were grown in Roswell Park Memorial Institute (RPMI). U138MG cells were grown in Eagle’s Minimum Essential Medium (MEM) supplemented with 2 mM L-glutamine, non-essential amino acids and 0.1 g/L sodium pyruvate. All cell culture media were supplemented with 10% fetal bovine serum (FBS),100 U/mL penicillin, and 100 µg/mL streptomycin and all cells were grown at 37 ºC in an incubator with humidified air equilibrated with 5% CO₂.

CRC cell lines: Difi, SW620, Lim2537, SW480, Lim2099, HCA7, Colo201, Colo320, HCT116, IS1, SW48, MC38, SW1116 and CX1 were obtained from colleagues either at WEHI or the Olivia Newton John Cancer Research Institute (ONJCRI). Other cell lines such as A431 (human squamous), were obtained from colleagues at WEHI, BT-20 and MDA-MB-468 were purchased from ATCC. All cell lines were grown in their recommended media (DMEM from Gibco, RPMI 1640 from Sigma and EMEM from ATCC) supplemented with 10%FCS, -/+ Adds (1.08% thioglycerol, 50 mg/ml hydrocortisone, 100 U/ml insulin), -/+ G418. All cell lines were grown to approximately 75% confluency before passaging. All cell lines except MC38 (mouse colon), BT-20 (human breast) and MDA-MB-468 (human breast) cell lines were maintained at 37^oC and 10% CO₂. MC38, BT-20 and MDA-MB-468 cell lines were grown at 37^oC and 5% CO₂. The original sources, current locations of these CRC cell lines and the characteristics of each cell line are documented in the Supplementary material of Wang et al. [44].

Cell viability assay

Briefly, the optimal seeding cell density for each cell line was determined in triplicate in a 96 well plate: Difi 3000 cells per 90ul, BT20 2000 cells per 90ul, MBA-MB-468-1500 cells per 90ul, Lim1215 2500 cells per 90ul and SW620 2000 cells per 90ul.

Using optimal seeding densities, cell viability assays were set up in triplicates in 96 well plates. After 24hr, cells were treated with the varying concentrations PPEA-polyplexes and a cytotoxic drug such as 2.5µM WEHI 7326 [45] or 1µM Bortezomib (Cell Signaling Technology), for total cell death control. Untreated cell viability was determined by adding an equivalent volume of the HBG buffer. Cells were treated with drugs for either 2hr, 72 hr or 96 hr. Cell survival was quantitated using CellTiter96Aqueous One Solution Cell Proliferation assay or CellTiter-Glo 2.0 Cell Viability assay (Promega, Madison) and luminescence plate reader (EnVision 2100 Multilabel reader) with Wallac EnVision Manager 1.12 software. Total absorbance for treated wells was calculated by subtracting the mean background (non-treated). Percentage of treated wells was calculated dividing total absorbance with mean values of non-treated minus total cell death. Mean percentage of treated wells was calculated by taking the average of percentage of the 3 treated cells and its SD. EC₅₀ was determined to have 50% of cell viability.

Flow cytometric EGF receptor quantitation

For flow cytometric analysis 0.5 x 10⁶ cells in 50 µl were seeded into U-bottom 96 well plates and the cell suspension was treated with 2.5 µg/ml mouse anti-human EGFR monoclonal antibody 528 (WEHI) and 0.4 µg/ml F(ab’)2-Goat anti-Mouse IgG (H+L) PE (eBioscience) or 1.5 µl of PE anti-human EGFR antibody (BioLegend). Cells were incubated on ice for 1hr for anti-mouse 528 and 30 mins for PE antibody or 45 min for PE anti-human EGFR antibody. All samples except control cell preparations were stained with either 5 µg/ml Propidium Iodide (PI) (Thermofisher) or 0.1 µg/ml DAPI (Thermofisher). Cells expressing fluorescent protein and/or cell surface markers were analysed using CytoFLEX (Beckmann Coulter) or Fortessa1 (BD Biosciences) flow cytometers. Each analysis consisted of at least 10,000 live cells (events). Spectral bleed-through between the PI channel and PE channel were corrected by compensation. A431 and/or Difi cells were run as positive controls every time. FlowJo 10 software (BD Biosciences) was used for flow cytometric data analyses. Cells were gated in order: live cells, single cells and PI or DAPI negative cells. The final sorted population (PI negative cells) was analysed for presence of GFP using the 525 nm channel and PE using the 585 nm channel, and calculations on fluorescence intensity performed. The median values for the 585 nm population in quadrant 3 were obtained by subtracting the median PE value from median 528 nm, i.e.,1^oAB + PE. The level of EGFR expression was expressed as a percentage of A431 or Difi EGFR expression. The error bars represent the standard deviation of the difference between sample medians (σ_d) [46]. σ_d was calculated using the formula = sqrt (σ₁² / n₁ + σ₂² / n₂) where σ₁ and σ₂ refers to the standard deviations from sample 1 (528 1^oAB + PE) and sample 2 (PE). n₁ and n₂ refers to the number of cells measured in each of these samples.

PBMC extraction

Peripheral blood mononuclear cells (PBMC) from healthy human donors were separated on Ficoll Plaque (Pharmacia). Briefly, cells were diluted in RPMI 1640 (x4) and 40 mL of diluted cell suspension was layered over 13 mL of Ficoll Plaque in 50 mL conical tubes. Following centrifugation at 400g for 30 mins at 20 ºC in a swinging bucket rotor without brakes, the mononuclear cell layer was transferred to a new 50 mL conical tube, washed twice with 50 ml RPMI 1640 medium, resuspended at a density of 4x10⁶ cells/ml and cultured in this RPMI 1640 medium supplemented with 10% fetal calf serum, 100 units/ml penicillin and 100 µg/ml streptomycin.

Chemotaxis assay

Chemotaxis was performed as described previously [27], except that chemotaxis plates were from Corning (HTS Transwell-96 well plate, 5 µm polycarbonate membrane, catalogue no.3388).

Cytokine measurements by ELISA

Supernatants from polyIC-treated cells were harvested and assayed for IP-10 and Gro-α using commercial ELISA kits (PeproTech) according to the manufacturer’s instructions.

Cell treatment and immunoblotting

MDA-MB-468 cells were seeded in 6-well plates (600,000 cells/well), allowed to grow for 24 hrs in compete medium and starved for 24 hrs in serum-free medium. PPEA was diluted in HBS (20 mM Hepes pH 7.4, 150 mM NaCl) and cells were treated in duplicates with 120 ng PPEA for different periods of time. As a positive control for EGFR kinase activity, cells were treated with 10 ng EGF for 5 min. Cells were lysed with boiling sample buffer (10% glycerol, 50 mmol/L Tris-HCl pH 6.8, 3% SDS, and 5% 2-mercaptoethanol). Protein determination was performed using the bound Coomassie Blue method, as described previously [47]. Lysates containing 60 µg of protein were resolved on 10% SDS-PAGE gel. Western blot analysis was conducted as described previously [48]. The following antibodies were used: Antibody specific for EGFR was purchased from Santa Cruz Biotechnology (catalogue no.sc-03, dilution 1:1000); Antibody recognizing phosphorylated EGFR (Tyr1068) was from Cell Signaling Technology (catalogue no. 2234S, dilution 1:1000); Anti-GAPDH was from Cell Signaling Technology (catalogue no. 2118, dilution 1:1000).

Treatment of subcutaneous xenografts

Female nude mice (NUDE-HSD: Athymic Nude-NU mice) aged 3–4 weeks were obtained from Harlan, Rehovot. All animal experiments were performed according to the Hebrew University Ethical Committee regulations. Humane endpoints required by HUJI Ethical committee to reduce risk of pain to the animals:

1. Tumor length reaches 15 mm.
2. 10% decrease in weight between 2 consecutive weighings or 20% between 1st and last weighings.
3. Slow movement, apathy

A total of 40 animals were used in this experiment. Suffering of the animals was reduced to the minimum and animals were checked at least 3 times/week until the tumor diameter reached 10 mm; after which the animals were checked daily. Where possible, mice were sacrificed before reaching one of the formal humane endpoints. The total duration of the experiment was 21 days.

For A431 xenografts, two million A431 cells resuspended in 0.2 ml of PBS were injected subcutaneously into the right flank of immune compromised female athymic nude mice. The volume of growing tumors was calculated as follows: V = LW²/2 (L = length and W = width). When the tumors reached an average volume of 137 mm³, mice were randomly divided into 5 groups (n = 7 or 8 per group), and the treatment was initiated. PolyIC formulated with PPEA in HBG buffer at w/w ratio of 0.78, which correlates with N/P ratio (molar ratio of nitrogen in PEI to phosphate in polyIC) of 6 as described previously [23,27], was injected. The polyplexes were administered via intravenous injection at doses of 0.75 mg/kg or 0.1 mg/kg every 24 hrs, 6 days per week, for a total of 9 injections over a 10-day period. Tumor volumes were measured on days 3, 6 and 10 after starting the intravenous injections.

Ethics statement

All animal procedures were approved by the [Hebrew University Ethical Committee / relevant authority] under protocol number NS-13-13709-5 and conducted in accordance with institutional and national guidelines for the care and use of laboratory animals. De-identified human PBMC were obtained from the Israel National Blood Bank.

Engineering and characterisation of triconjugate affibody-polyIC-polyplexes

We engineered an affibody-PEG-polyIC-polyplex to target the EGFR and deliver polyIC to the cytoplasm of cancer cells. An analogue of the anti-EGFR affibody [31], Z_{EGFR 1907’}-Cys was expressed in E.coli BL21 (DE3) and purified on a Ni-NTA column from lysates as described in the methods section. SDS-PAGE analysis of the sample detected two bands with apparent molecular weights ~8 kDa and 16 kDa under pseudo native conditions (Fig 1Ai; see Materials and methods). These bands were analysed by mass spectrometry and the resulting sequence for the Z_{EGFR 1907’}-Cys affibody is shown in Fig 1Aii. The 8 kDa and 16 kDa bands correspond to the monomer and dimer of the Z_{EGFR 1907’}-affibody. The monomeric form of the Z_{EGFR 1907’}-Cys affibody was purified by gel filtration (Fig 1Bi). All subsequent experiments were performed using the purified with the Z_{EGFR 1907’}-Cys affibody monomer.

Binding of the Z_{EGFR 1907’}-Cys affibody to the extracellular domain of the EGFR

Surface plasmon resonance analysis was used to measure the binding of the monomeric Z_{EGFR 1907’}-Cys affibody to two forms of the extracellular domain of EGFR: sEGFR_1–501 (a truncated form of the extracellular domain) [49] and sEGFR_1–621 (the full length extracellular domain) [50–52] (Fig 2A). Our binding kinetics data indicates that Z_{EGFR 1907’}-Cys affibody binds more tightly to EGFR_1–621 than EGFR_1–501 and has an average K_D value of 6.74 ± 0.91 nM SD (Supplementary Figure S3A and Table 1). Z_{EGFR 1907} has been reported previously to have a K_D value of 5.4 nM using a similar method [31], however, different EGFR analogues were immobilized onto the biosensor. The strong affinity of the Z_{EGFR 1907’}-Cys affibody to EGFR_1–621 is consistent with having a more accessible groove created by folding of domain I and III in an untethered confirmation by the full-length extracellular domain sEGFR_1–621 [53]. However, fitting the data to a global model suggests that the Z_{EGFR 1907’}-Cys affibody binds to two sites on the sEGFR_1–621 receptor with two different binding affinities (Table S2 and Supplementary Figures S2 and S3B). This was not evident when affibody bound to sEGFR_1–501. This finding supports the notion that sEGFR_1–621 adopts two conformations, tethered and untethered [54–56].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Binding affinities of Cetuximab and Z_{EGFR 1907’} affibody to the EGFR-extracellular domain (ECD).

https://doi.org/10.1371/journal.pone.0334584.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Biacore analysis of Z_{EGFR 1907’} binding to EGFR proteins.

SDS-PAGE analysis showing the purity of the proteins used for the Biacore analyses. Expected molecular weights: sEGFR_1–501 56kDA, sEGFR_1–621 69kDa, Z_{EGFR 1907’} 11kDa and Cetuximab 152kDa. (B) Binding kinetics for the Z_{EGFR 1907’} and Cetuximab interacting with sEGFR_1–501 and sEGFR_1–621, used to determine K_d values using Biacore S200. (C) Competitive binding kinetics of Z_{EGFR 1907’} and Cetuximab detected under various conditions shows Cetuximab and Z_{EGFR 1907’} appear to bind to the same EGFR binding site. Solid and dotted line graphs represent raw and fitted data respectively.

https://doi.org/10.1371/journal.pone.0334584.g002

Comparison of the binding of the ZEGFR 1907’ and Cetuximab to sEGFR

For comparison purposes we studied the binding kinetics of Cetuximab [57,58] binding to the EGFR extracellular domain. Binding kinetic curves for Cetuximab showed that once bound to either sEGFR proteins, Cetuximab is slow in coming off the receptor as indicated by the flat dissociation curves (Supplementary Figure S3B). Comparison between Z_{EGFR 1907’} and Cetuximab shows Cetuximab does not come off the EGFR sensor chips as easily as the Z_{EGFR 1907’} (Supplementary Figure S3A). The tight binding of Cetuximab to the EGFR sensor reduces the ability of the chip to reach equilibrium binding, so the K_D value determined in this study is not accurate. Our study determined Cetuximab to have an average K_D value of 5.3 pM ± 9.9 SD for EGFR_1–621 which is higher affinity than the reported K_D value of ~1 nM for Cetuximab binding to EGFR [59]. There was a significant difference between K_on and K_off values calculated for both forms of the sEGFRs under local 1:1 binding model fits (Table S3). It is possible that upon binding of Cetuximab to the EGFR surface, the conformation of the EGFR changes from a low affinity receptor to a high affinity binding state (Table 1). X-ray crystallographic and binding studies have shown that Cetuximab binds exclusively to domain III of soluble extracellular region of the EGFR (sEGFR) [53,60].

A competition screen between Z_{EGFR 1907’} affibody and Cetuximab was measured using an A-B-A assay format where a flanking solution was injected before and after the sample. We used this assay to investigate the binding site of the Z_{EGFR 1907’} affibody to the EGFR. As shown in Figs 2B, 2C, and S2B binding of Cetuximab blocks binding of Z_{EGFR 1907’} affibody to sEGFR_1–621. The slight dip in response units upon injection of the second component (B: Z_{EGFR 1907’} affibody) to Cetuximab suggests occurrence of some non-specific binding as Cetuximab is not expected to come off within 400 sec. Also, the relatively fast off rate for Z_{EGFR 1907’} affibody allows Cetuximab to bind to newly unbound sEGFR_1–621 sites, however when Cetuximab is mixed with Z_{EGFR 1907’} affibody in excess, this does not occur, indicating direct competition between the EGFR binders. Together these data show that the Z_{EGFR 1907’} affibody shares the Cetuximab binding site (i.e., EGFR-domain III).

EGFR expression and cell surface availability in colorectal and breast cancer cell lines

EGFR is overexpressed in majority of (50–80%) of CRCs(61) and these elevated levels of expression have been linked with tumor aggression and poor survival [14,61]. Cancer cells overexpressing EGFR appear to be ideal candidates for new targeted therapeutics such as the affibody polyplexes but was not clear that all the EGFR would be on the cell surface and accessible to the polyplexes. Consequently, we compared the published proteomic data from our laboratory, i.e., total EGFR levels [44] with the cell surface EGFR levels measured by flow cytometry for a range of CRC cell lines and some breast cancer cell lines. Cell lines with increased EGFR expression from the proteomic analyses also had more cell surface EGFR (Fig 3A, Spearman correlation coefficient (r_s) 0.75, p < 0.001). For example, Difi cells have the highest levels of EGFR expressed on the cell surface (Fig 3A). A small number of CRC cells, e.g., SW620 and Colo201 and SW480 showed little to no EGFR expression in the proteomic or flow cytometric studies. Thus, we used the SW620 cell line as an EGFR negative control in our experiments.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Relationship between cell surface EGFR and total EGFR in cancer cell lines.

(A) Cell lines were incubated with anti-EGFR-PE and the PE fluorescence was measured by flow cytometry. The median fluorescence intensity for the respective quadrant was used to calculate relative levels of EGFR expression on cell surface compared to the A431 cell line. Proteomic data was calculated from mass spectrometer peptide counts. (B) Relative levels of EGFR expression on cell surface of several CRC cell lines and breast cancer cell lines, BT-20 and MDA-MB-468, compared to the A431 cell line.

https://doi.org/10.1371/journal.pone.0334584.g003

Two triple-negative breast cancer (TNBC) cell lines, BT-20 and MDA-MB-468 overexpress EGFR [59,62]. At present, TNBC patients have limited options for treatment and we expected that the EGFR-directed polyplexes might be suitable for treating TNBC patients whose tumors overexpress the EGFR. Cell surface expression of EGFR levels on these cell lines was determined. Both breast cell lines have lower levels of surface EGFR compared to the Difi and A431 cell lines. MDA-MB-468 has slightly less than half the Difi EGFR levels (48%) and the BT-20 cell line had approximately a quarter of the Difi EGFR levels (21%) (Fig 3B). Similar results have been reported for total EGFR levels using Western blot analysis (Mohan et al., 2021). The BT-20 and MDA-MB-468 EGFR levels are similar to the EGFR levels on CRC cell lines C135 and Lim2405 respectively (see Fig 3A,3B). Thus, both BT-20 and MDA-MB-468 cell lines can be considered to have moderate levels of EGFR on the cell surface.

We also used three cell lines which express very low EGFR levels: BT-747 [63], MCF-7 [63] and SK-BR-3 [63]. Although we did not measure the EGFR levels on these cell lines each has been studied extensively and the flow cytometry data from Stanley, et al [63] shows they all have low EGFR levels. The other cell line used for the cytotoxicity studies MDA-MB-231 had low to moderate levels of the EGFR [63], i.e., EGFR was detectable by flow cytometry, but clearly the EGFR levels were only 5% the levels in the over-expressing cell line MDA-MB-468 [63].

In vitro cancer cell killing by PPEA-polyplexes

To assess the cytotoxic activity of the PPEA-polyplex on cells in culture, we used cell lines that expressed different levels of the receptor (from undetectable to more than 10⁶ EGFR per cell). After preparation of the PPEA-polyplex and before use, we checked the size of the polyplex using the zetasizer. The particle size of the PPEA-polyplex was 119 ± 26 nm. The PPEA triconjugate (i.e., the affibody-PEI-PEG complex without the polyIC) was used as a control for nonspecific toxicity to the cells. As illustrated in Fig 4A and 4B, the PPEA-polyIC-polyplex induced a strong killing of EGFR-over-expressing cells (1-2x10⁶ EGFRs/cell) as compared to a cell line devoid of EGFR (U138-MG). In contrast, treatment with the PPEA, polyI-polyplex did not result in cell killing in either of the cell lines tested (Fig 4C), indicating that growth inhibitory activity induced by PPEA treatment is polyIC-dependent. PPEA-polyIC-polyplexes also produced a significant growth inhibitory effect in the SK-BR-3, BT-474 and MDA-MB-231 cell lines, which express medium levels of the receptor (1x10⁵-3x10⁵ EGFRs/cell) (Fig 5). These results show that PPEA-polyIC-polyplexes inhibit the proliferation of tumor cell lines with medium to high levels of EGFR expression.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Anti-tumor activity of the PPEA-polyIC-polyplex in vitro.

(A) PPEA-polyplexes selectively kill cell lines overexpressing EGFR. Cells were seeded in duplicates into 96-well plates at a density of 5000 cells in 0.1 ml medium per well and grown overnight. Cells were then treated with polyIC at the indicated concentrations using the PPEA complex. PEI-PEG ratio = 1:1; w/w ratio PEI: polyIC = 0.78. U138MG cells do not express EGFR; U87MGwtEGFR cells express 1x10⁶, A431 express 2-3x10⁶ and MDA-MB-468 express 2x10⁶ EGFRs/cell. (B,C) In vitro anti-tumor activity of PPEA-polyIC-polyplex is polyIC-specific. A431 and U87MGwtEGFR cell lines were treated with same doses of PPEA-polyIC-polyplex (B) or PPEA polyI polyplex (C), which served as negative control. Viability was measured by the PrestoBlue Cell Viability Reagent (Invitrogen), according to the manufacturer’s instructions, at 72 hrs after treatment. These experiments were repeated three times with a representative experiment shown.

https://doi.org/10.1371/journal.pone.0334584.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Anti-tumor cell activity of PPEA-polyplex in cell lines with medium levels of EGFR.

https://doi.org/10.1371/journal.pone.0334584.g005

Cells were treated with polyIC at the indicated concentrations using PPEA. Cell viability was measured as described in Fig 4. MDA-MB-231 cells express 2.5-3x10⁵ EGFRs/cell, SK-BR-3 express 3x10⁵, BT-474 express 10⁵ EGFRs/cell and MCF7 cells express 5x10³ EGFRs/cell.

To investigate the relationship between the level of EGFR expression on cell surface and the potency of the PPEA-polyplex, we determined EC₅₀ values for two colon cancer cell lines SW480 and Difi and two breast cancer cell lines BT-20 and MDA-MB-468. Upon treatment with PPEA-polyplex (Table 2, Figure S4), the cell viability assays showed significant growth inhibition of the CRC cell lines with medium levels of EGFR overexpression such as LIM1215 ((EC₅₀ < 370 ng/ml) Figure S4A and S4B). Despite the very high levels of EGFR on Difi cells the PPEA-polyplex was no more potent, than it was on the LIM1215 cells (Table 2, Figures S4B and S4C).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Cytotoxicity of the PPEA-polyplex on selected cancer cell lines.

https://doi.org/10.1371/journal.pone.0334584.t002

The PPEA-polyplex was a potent inhibitor of the breast cancer cell lines BT-20 and MDA-MB-468 even though they have less than half the EGFR levels of A431 cells, indeed the BT-20 cell line was 2 orders of magnitude more sensitive to PPEA-polyplex killing than the other cells (Table 2).

PPEA-polyplex does not induce EGFR phosphorylation

The affibodies in the Z_EGFR1907 family have been reported to bind the EGFR without activating the kinase [29,30]. By treating MDA-MB-468 cell with the PPEA-polyplex and analyzing the autophosphorylation of EGFR we confirmed that the Z_EGFR1907’ affibody, when tethered to PEI-PEG, does not activate the EGFR kinase (Supplementary Figure S5), i.e., there was no increased EGFR phosphorylation in the cells treated with the PPEA-polyplex.

The PPEA-polyplex induces expression of chemotactic cytokines in cancer cells expressing high and moderate levels of EGFR

Tumor cells that internalize polyIC are induced to secrete cytokines that attract immune cells [64,65]. In our previous studies, we showed that polyIC targeted by PPE, induced expression of Interferon ɣ-induced protein 10 kDa (IP-10), and growth- regulated protein α (Gro-α) in EGFR-overexpressing cell lines, but not in cells devoid of EGFR [23,24]. Gro-α and IP-10 are chemokines that play an active role in the recruitment of leukocytes to the domain in which they are secreted [66,67]. To evaluate the ability of PPEA-polyplex to induce EGFR-expressing cells to secrete inflammatory cytokines, we collected cell culture supernatants 48 hrs after exposure to the PPEA-polyplex and assessed for the presence of cytokines Gro-⍺ and IP-10. As can be seen in Table 3, cells harboring high or moderate levels of EGFR secrete chemotactic cytokines following challenge with PPEA-polyplex. In contrast, no cytokine secretion was detected in U138-MG and MCF-7 cells that are devoid of (or express low levels of) the EGFR, respectively, following treatment with the PPEA-polyplex. Interestingly, we observed significantly higher amounts of IP-10 in supernatants collected from MDA-MB-468, SK-BR-3 and BT-474 cells, compared to A431 and MDA-MB-231 (Table 3). This observation points to possible differences in signalling mechanisms mediated by PPEA-polyplex treatment in the cell lines tested. Collectively, these data demonstrate that polyIC treatment using PPEA stimulates release of pro-inflammatory cytokines by tumor cells expressing high and moderate levels of EGFR.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Secretion of chemokines after transfection with PPEA-polyplex.

https://doi.org/10.1371/journal.pone.0334584.t003

PPEA-polyplex treated cancer cells activate human immune cells in vitro

We used peripheral blood mononuclear cells (PBMCs) to assess the ability of cytokine-enriched medium derived from PPEA-polyplex treated tumor cells to stimulate the immune cells [27]. PBMCs consist of several types of immune cells, including monocytes, macrophages, NK (natural killer) cells and T-cells. When stimulated, PBMCs produce an array of cytokines which can be conveniently quantified by ELISA [68].

We tested whether the chemotactic stimuli secreted by cells that internalized polyIC stimulated PBMC to migrate toward the secreting cells [69]. Using chemotactic chambers (see Materials and Methods) we showed that the cytokine-enriched medium from A431 cells treated with PPEA-polyplex stimulated chemotaxis of PBMCs (Fig 6). In contrast, medium of U138MG treated with PPEA-polyplex did not induce chemotaxis. Thus, PBMCs are attracted by medium from EGFR overexpressing cells, which have been treated with PPEA-polyplex. The role of polyIC would be clearer if we had used a PPEA-polyI control.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Activity of conditioned media from PPEA-polyIC-polyplex treated cancer cell lines on PBMC chemotaxis in vitro.

https://doi.org/10.1371/journal.pone.0334584.g006

The PPEA-polyIC-polyplex selectively kills cell lines overexpressing EGFR. Cells were seeded in duplicates into 96-well plate at a density of 5000 cells in 0.1 ml medium per well and grown overnight. Cells were then treated with polyIC at the indicated concentrations using the PPEA complexes. PEI-PEG ratio = 1:1; w/w ratio PEI: Poly(IC) =0.78. U138MG cells do not express EGFR; A431 express 2-3x10⁶ and MDA-MB-468 express 2x10⁶ EGFRs/cell.

We examined the ability of the medium from PPEA-polyplex treated cancer cells to stimulate PBMCs. Upon activation, PBMC produce cytotoxic cytokines, such as INF- ɣ and TNF-α, known to contribute to tumor suppression and eradication [70,71]. PBMC were challenged with culture supernatant from PPEA-polyplex treated cells and their response was measured using ELISA. Table 4 illustrates the levels of INF-ɣ and TNF-α expression by the immune cells, detected 24 hrs following the challenge. The data clearly show that the medium originated from polyIC-targeted A431, MDA-MB-231 and SK-BR-3 cells (polyIC, 2 µg/ml) stimulate the production of cytotoxic cytokines as compared to culture supernatant from untreated cells or unchallenged PBMC cells (Table 4). We conclude that the culture supernatants derived from polyIC-targeted EGFR overexpressing cells sustain the capability to selectively stimulate the immune cells whereas cells devoid of EGFR do not.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Evaluation of cytotoxic cytokine levels secreted by activated PBMC cells.

https://doi.org/10.1371/journal.pone.0334584.t004

PBMC-mediated bystander effect

Data from clinical studies has demonstrated a correlation between the presence of tumor infiltrating lymphocytes which produce cytokines such as INF-ɣ and TNF-α, and better outcomes for patients with many different cancers [72,73]. Previously, we reported that expression of INF-ɣ and TNF-α by PBMCs strongly enhanced the bystander killing of untreated tumor cells [27]. To assess the PBMC-mediated bystander effect, MDA-MB-231 and SK-BR-3 cells were first treated with PPEA-polyplex and 24 hrs later, freshly isolated PBMCs were added to the treated cells for co-incubation. Following a further 24 hrs, medium from the treated cells co-cultured with PBMC, was added to newly seeded, untreated cells (Fig 7A). The PBMC-mediated bystander killing was examined 72 hrs later, and compared with the “direct” bystander killing, mediated by medium from PPEA-polyplex-treated MDA-MB-231 and SK-BR-3 cells (Fig 7B and 7C). The medium derived from polyIC-treated cells co-incubated with PBMCs, exerted a strong PBMC-mediated bystander effect, killing up to 80−90% of the nontreated cells (Fig 7C). Medium derived from MDA-MB-231 or SK-BR-3 cells challenged with PPEA-polyplex alone, eliminated only 20 and 35% of untreated tumor cells, respectively (Fig 7B). These results demonstrate that PPEA-polyplex treatment and immune cells display a cooperative cell-killing activity by enhancing bystander killing of EGFR-overexpressing tumors.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. “Direct” versus PBMC-mediated bystander effect.

50,000 of MDA-MB-231 and SK-BR-3 cells were seeded into 24-well plates in duplicates and grown overnight with 1 ml medium per well. Cells were then treated with the PPEA-polyplex at the indicated concentrations. To test the “direct” bystander effect, 48 hrs following treatment 0.1 ml of medium from challenged cells was exchanged for 0.1 ml medium from non-treated cells (“indicator cells) seeded on 96 well plates (4000 cell/well) 24 hrs earlier. To analyze the PBMC-mediated bystander effect, 5x105 of PBMCs were added to treated cells 24 hrs following treatment. After 24 hrs of co-culture, 0.15 ml of medium from the “conditioned medium” was added to untreated cells seeded into 96 well plates (4000 cell/well) 24 hrs earlier. Survival of untreated cells was determined by methylene blue assay, 72 hrs after challenge with the conditioned medium. (A) Shows experiment design. (B) Shows the “direct” bystander effect of medium from polyIC-targeted MDA-MB-231 and SK-BR-3 cells on unchallenged MDA-MB-231 and SK-BR-3, respectively. (C) Shows PBMC-mediated bystander effect of PBMCs co-incubated with polyIC-treated MDA-MB-231 and SK-BR-3 cells on unchallenged MDA-MB-231 and SK-BR-3, respectively.

https://doi.org/10.1371/journal.pone.0334584.g007

PPEA complexes induce bystander effects

“Direct” versus PBMC-mediated bystander effects were also measured by treating MDA-MB-231(human breast cancer) and SK-BR-3 (human breast cancer) cells with the PPEA-polyplex and testing the direct or indirect bystander effects of the medium on PBMCs. Figure S6A shows the experimental design, Figure S6B shows the “direct” bystander effect of medium from polyIC- targeted MDA-MB-231 and SK-BR-3 cells on unchallenged MDA-MB-231 and SK-BR-3, respectively and Figure S6C the PBMC-mediated bystander effect of PBMCs co-incubated with polyIC- treated MDA-MB-231 and SK-BR-3 cells on unchallenged MDA-MB-231 and SK-BR-3, respectively. The supernatants from PPEA-polyplex treated cells activate human immune cells.

Efficacy of PPEA-polyplexes on human A431 xenografts growing in nude mice

The in vivo antitumor activity of the PPEA-polyplex was examined using the A431 subcutaneous xenograft model. Figure 8 shows that the PPEA-polyIC-polyplex inhibited tumor growth strongly. As a control, we treated the animals with a PPEA-poly Inosine (polyI)-polyplex made in the same way as the PPEA-polyIC-polyplex. Poly Inosine is not a double stranded RNA and would not be expected to kill the cells either directly or to induce immune cytokines which amplify the killing of the tumor cells. The tumors treated with the PPEA-poly Inosine (polyI)-polyplex grew at a similar rate as tumors in the untreated group.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. In vivo efficacy of PPEA-polyIC and poly I-polyplexes.

https://doi.org/10.1371/journal.pone.0334584.g008

Female nude mice were injected subcutaneously (s.c.) with A431 cells as described in the Methods section. After tumor establishment, mice were randomly divided into five groups and treated with daily intravenous injections (6 days/week) of the indicated polyplex at either 0.75 mg/kg or 0.1 mg/kg for 10 days, totaling 9 injections. Tumor volumes were measured on days 3, 6 and 10, and the chart presents the average tumor volume for each group. Statistical analysis: Two-tailed paired t test. Data are presented as mean ± SEM. P-values ware for comparison with the tumors growing in untreated (UT) mice; ** p < 0.01; *** p < 0.001.