Development of a single-chain fragment variable fused-mutant HALT-1 recombinant immunotoxin against G12V mutated KRAS colorectal cancer cells

Background KRAS oncogenes harboring codon G12 and G13 substitutions are considered gatekeeper mutations which drive oncogenesis in many cancers. To date, there are still no target-specific vaccines or drugs available against this genotype, thus reinforcing the need towards the development of targeted therapies such as immunotoxins. Methods This study aims to develop a recombinant anti-mKRAS scFv-fused mutant Hydra actinoporin-like-toxin-1 (mHALT-1) immunotoxin that is capable of recognizing and eradicating codon-12 mutated k-ras antigen abnormal cells. One G13D peptide mimotope (164-D) and one G12V peptide mimotope (68-V) were designed to elicit antigen specific IgG titres against mutated K-ras antigens in immunised Balb/c mice. The RNA was extracted from splenocytes following ELISA confirmation on post-immunized mice sera and was reverse transcribed into cDNA. The scFv combinatorial library was constructed from cDNA repertoire of variable regions of heavy chain (VH) and light chain (VL) fusions connected by a flexible glycine-serine linker, using splicing by overlap extension PCR (SOE-PCR). Anti-mKRAS G12V and G13D scFvs were cloned in pCANTAB5E phagemid and superinfected with helper phage. After few rounds of bio-panning, a specific mKRAS G12V and G13D scFv antibody against G12V and G13D control mimotope was identified and confirmed using ELISA without any cross-reactivity with other mimotopes or controls. Subsequently, the anti-mKRAS scFv was fused to mHALT-1 using SOE-PCR and cloned in pET22b vector. Expressed recombinant immunotoxins were analyzed for their effects on cell proliferation by the MTT assay and targeted specificity by cell-based ELISA on KRAS-positive and KRAS-negative cancer cells. Results The VH and VL genes from spleen RNA of mice immunized with 164-D and 68-V were amplified and randomly linked together, using SOE-PCR producing band sizes about 750 bp. Anti-mKRAS G12V and G13D scFvs were constructed in phagemid pCANTAB5E vectors with a library containing 3.4 × 106 and 2.9 × 106 individual clones, respectively. After three rounds of bio-panning, the anti-mKRAS G12V-34 scFv antibody against G12V control mimotope was identified and confirmed without any cross-reactivity with other controls using ELISA. Anti-mKRAS G12V-34 scFv fragment was fused to mHALT-1 toxin and cloned in pET22b vector with expression as inclusion bodies in E. coli BL21(DE3) (molecular weight of ~46.8 kDa). After successful solubilization and refolding, the mHALT-1-scFv immunotoxin exhibited cytotoxic effects on SW-480 colorectal cancer cells with IC50 of 25.39 μg/mL, with minimal cytotoxicity effect on NHDF cells. Discussion These results suggested that the development of such immunotoxins is potentially useful as an immunotherapeutic application against KRAS-positive malignancies.


INTRODUCTION
Colorectal cancer continues to rise as one of the most frequent causes of cancer-related deaths in both developing and developed countries, especially amongst east Asian nations. While the use of immunotherapies such as anti-epidermal growth factor receptor (EGFR) monoclonal antibodies have found reasonable ground against EGFR positive colorectal cancer patients, the absence of effective targeted immunotherapies against EGFR positive patients harboring a mutated Kirsten rat sarcoma viral oncogene homolog (KRAS) gene have yet to be addressed. This leaves mutated KRAS patients with non-specific antineoplastic drugs as their sole treatment option with a statistically poor 5-year prognosis and a rapid deterioration in the quality of life (National Cancer Registry, National Cancer Institute, Ministry of Health Malaysia, 2018). Past efforts to inhibit mutated KRAS proteins either directly or indirectly have been attempted, but met with little success. Examples of such targets and biologics include farnesyltransferase inhibitors to block KRAS membrane localization, PDEδ, and other effector signaling pathways downstream of KRAS (Ryan & Corcoran, 2018).
Oncogenic KRAS mutations are commonly found in colorectal cancers (40-50%), making it the most prominently mutated proto-oncogenes known to date (Jancik et al., 2010). Of those, about 83% involve codon 12% and 14% involve codon 13, whereas codon 61 accounts for a minor proportion (2%) (Hobbs, Der & Rossman, 2016). Therefore, there is an ongoing dire need to improve on targeted cancer therapies for the management of KRAS-positive colorectal cancer cases.
An approach using third generation recombinant immunotoxins, which are toxin moieties fused to the single-chain variable fragment (scFv) portion of antibodies, have gained much attraction as cancer therapeutics following their ability to specifically target and eradicate cancer cells. The use of scFvs in recombinant immunotoxins provides several advantages in therapeutic applications due to reduced immunogenicity, rapid blood clearance, lower retention times and better tissue penetration (Ahmad et al., 2012).
A comprehensive list of past scFv based immuno-therapeutics tested in both pre-clinical and clinical trial settings was discussed and summarized (Fercher et al., 2018).
Through a phage display scFv library established in this study, sequences capable of binding mutated K-ras antigens were panned out and fused with a mutant Hydra actinoporin-like-toxin-1 (mHALT-1) from Hydra magnipapillata. This small 20.8 kDa pore forming toxin is capable of binding preferentially to the sphingomyelin-associated membrane, and form cation-selective pores which lyses targeted cells due to an imbalance in osmotic pressure (Liew et al., 2015). It acts by forming pores in the cells' lipid membrane which leads to osmotic swelling and ultimately, cell death (Tejuca, Anderluh & Serra, 2009). Such Hydra toxins are extremely potent and can be exploited as cell-directed immunotoxins. Through site-directed mutagenesis, the toxin's affinity towards sphingomyelin was removed, enabling the toxin to specifically target cells solely through the scFv moiety, thus conferring targeted specificity (Liew et al., 2015). Since cell death caused by mHALT-1 does not require cellular internalization to reach a cytosolic target, mHALT-1 works immediately upon contact with the lipid cell membrane, hence, becoming an ideal toxin moiety candidate for immunotoxins.
Currently, several recombinant immunotoxins are undergoing clinical trials against different types of cancers, however, none is currently in the pipeline against KRAS-positive cancers. Therefore, this study reports a proof-of concept approach employing a 3rd generation anti-mKRAS scFv immunotoxin potentially capable of target-specific eradication of KRAS-positive colorectal cancer cells.

Construction of phage-displayed scFv library
Total RNA was extracted from homogenized mice splenocytes immunized with 164-D (YKLVVVGAGDVYKSA) and 68-V (YKLDVVGAVGVGKSA) mimotopes (Ng et al., 2018) using RNeasy Mini kit (Qiagen, Germantown, MD, USA) according to the manufacturer's instructions. Reverse-transcription with the RNA extract was performed with Tetro cDNA synthesis kit (Bioline, London, United Kingdom) following the manufacturer's protocol. Primers for PCR amplification of scFv construction were designed according to Table 1. In the first round of PCR, V H and V L were amplified separately by PCR reactions containing eight forward primers with four reverse primers adapted from Zhou, Fisher & Papas (1994) with modifications. PCR products were then purified using a gel extraction kit (iNtRON, Korea). The assembled scFvs were double digested with SfiI and NotI (NEB, Hitchin, UK) and ligated into pCANTAB5E phagemid vector (a molar vector to insert ratio of 1:3), using electroligase (NEB, Hitchin, UK). The resulting ligation products were electroporated into E. coli TG1 at 1.8 kV, 5 ms. The library was then grown in 1 mL 2x Bacto tryptone (YT)-G medium (2% (v/v) glucose) for 1 h and then plated on SOBAG plates containing 100 mg/mL ampicillin and 2% glucose. Colonies were scraped off into 2x YT medium and pooled to make library glycerol stocks for storage at −80 C. The glycerol stocks of scFv libraries were grown in 2x YT-G media (2% (v/v) glucose) and incubated at 37 C with shaking until it reached mid-log phase (OD 600 : 0.5). A final concentration of 100 mg/mL ampicillin and 4 × 10 10 pfu of M13K07 helper phage were added to the culture and grown at 37 C for 1 h with shaking. The phage-infected E. coli were pelleted by centrifugation and incubated overnight with 2x YT-AK media (100 mg/mL ampicillin and 50 mg/mL kanamycin). Rescued phage particles in the supernatant were precipitated by adding 1/5 volume of 20% 2.5 M PEG/NaCl and incubated on ice for 1 h. The phage pellets were collected by centrifugation at 10,000×g for 20 min at 4 C and resuspended in 1/50 volume of PBS. Following centrifugation at 12,000×g for 15 min at 4 C, the supernatant was filtered through 0.45 mm filter to remove bacterial cells.

Antigen-specific phage selection by biopanning
The PEG-precipitated phage library was subjected to three rounds of selections. Panning selection of the phage was carried out on 96-well immunoplate (SPL, South Korea) immobilized with G12V and G13D control peptide antigens, using 50 mg for the first round, 25 mg for the second round, and 15 mg for the third round diluted in 100 mL of PBS overnight at 4 C. Peptide-coated wells were blocked with 2% (w/v) skim milk powder in PBS at 37 C for 2 h. The pre-blocked phage-scFv was added and incubated at 37 C for 2 h. After washing with 0.05% (v/v) of Tween-20 in PBS (PBST), 100 µL of log-phase E. coli TG1 cells were added and incubated at 37 C for 30 min with shaking. An aliquot of the dilution was plated on 2x YT-AG (100 mg/mL ampicillin and 2% (v/v) glucose) plates and incubated at 30 C overnight.
Rescue and screening of phage antibodies by phage-ELISA Individual phagemid clones were cultured in cluster tubes (Corning, New York, NY, USA) with 400 mL of 2x YT-AG (100 mg/mL ampicillin and 2% (v/v) glucose) media and incubated at 30 C overnight with shaking at 250 rpm. The recombinant phages were rescued by transferring 40 mL overnight culture to 400 mL of 2x YT-AG containing 2.5 × 10 10 pfu of M13K07. Following incubation for 2 h at 37 C with shaking at 150 rpm, the cells were pelleted at 1,500×g and resuspended in 400 mL of 2x YT-AK (100 mg/mL ampicillin and 50 mg/mL kanamycin) media. The culture was incubated overnight at 37 C and the phage-antibody supernatant was used to perform phage-ELISA. MaxiBinding microtiter plates (SPL, South Korea) were coated with 10 mg/mL control peptide antigens (100 mL) diluted in PBS, pH 7 at 4 C overnight. The coated plates were washed two times with PBS and blocked with 1% (w/v) bovine serum albumin (BSA) for 2 h at 37 C. A total of 100 mL of pre-blocked phage antibody supernatant (80 mL + 20 mL 1% (w/v) BSA) was added and incubated for 2 h at 37 C. Unbound antibodies were removed by washing three times with 0.05% (v/v) PBST and PBS, respectively. After the washings, 100 mL of diluted horseradish peroxidase (HRP)-conjugated anti-M13 antibody (Abcam, Cambridge, England, UK) in 1% (w/v) BSA (1:5,000) was added, and the plates were incubated for 1 h at 37 C. The wells were washed as described above. Hundred microliters of 3,3′,5,5′-Tetramethylbenzidine (TMB) substrate was added and the reaction was stopped with 50 mL of 1N sulphuric acid (H 2 SO 4 ) per well. OD 450 was determined using FLUOstar Ò Omega (BMG Labtech, Germany). Positive responses were determined with OD 450 readings, using a threshold of at least two times that of the control (1% BSA) (Singh et al., 2010). BSA-coated wells consist of phage antibodies without antigen coating, while wt M13K07 helper phage controls consist of antigen-coated wells with the addition of wt helper phage. wt M13K07 helper phage controls were used to identify background signals. All samples and controls were assayed in three replicates, and all data were expressed as mean values ± standard deviation indicated by error bars.

scFv DNA sequencing analysis
Positive clone plasmids were isolated with iNtRON Biotechnology DNA-spin TM plasmid DNA purification kit according to the manufacturer's protocol. The scFv DNA region was sequenced using pCANTAB5E sequencing forward primer R1 5′-CCATGATTACGCCAA GCTTTGGAGCC-3′. Sequencing results were aligned using the Clustal Omega Multiple Alignment from BioEdit version 7.1.9 (Tom Hall, Fort Worth, TX, USA). The full-length V H and V L chain region sequences were numbered according to the sequence analysis tool IgBLAST (Ye et al., 2013) for the determination of complementaritydetermining region (CDR) and the germline origins of V regions of scFv clones.

Expression of soluble scFv antibodies
Positive E-tag fusion scFv clones were transformed into non-suppressor E. coli HB2151 cells. The recombinant anti-mKRAS E-tag proteins were expressed and localized to the periplasmic space of E. coli. Briefly, 400 mL of log phase E. coli HB2151 cells were infected with 2 mL of recombinant phage antibodies and incubated at 37 C for 30 min with shaking at 150 rpm. To ensure resulting colonies were true nal r (nalidixic acid resistant) and not carried over from infected E. coli TG1 cells, the culture was streaked on SOBAG-N plate (SOBAG containing 100 mg/mL nalidixic acid). Following overnight incubation at 30 C, a single colony of infected E. coli HB2151 was picked and inoculated in five mL of 2x YT-AG media (100 mg/mL ampicillin and 2% (v/v) glucose) at 30 C overnight. The overnight culture (five mL) was inoculated in 50 mL fresh 2x YT-AG (100 mg/mL ampicillin and 2% (v/v) glucose) and incubated at 30 C for 1 h with shaking at 250 rpm. Following centrifugation at 1,500×g for 20 min, the pellet was resuspended in 50 mL of 2x YT-AI media (100 mg/mL ampicillin and 1 mM IPTG), and incubated at 30 C with shaking at 250 rpm. After 4 h of incubation, the culture was divided into two separate 50 mL tubes and centrifuged as detailed above. The pellets were treated with {[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]amino} ethanesulfonic acid (TES) buffer by mild osmotic shock to liberate the recombinant anti-mKRAS scFv antibodies from the periplasmic space of host E. coli HB2151 cells. Cell pellets were resuspended in 0.5 mL of ice-cold 1X TES after which 0.75 mL of ice-cold 1/5 TES buffer was added, and incubated on ice for 30 min. Following centrifugation at 13,000×g for 10 min, the supernatant containing the soluble periplasmic fraction was used for monoclonal-ELISA. The supernatant was analyzed by SDS-PAGE and western blotting with anti-E tag HRP-conjugated antibodies.

Monoclonal-ELISA
The antigen specificity of soluble scFv antibody was determined by monoclonal-ELISA using an anti-E-tag antibody. MaxiBinding microtiter plates were coated with 10 mg/mL wildtype or control peptide antigens (G12V, G13D) (100 mL) diluted in PBS at 4 C and blocked with 1% (w/v) BSA for 2 h at 37 C. The wells were washed three times with PBS and 100 mL of pre-blocked soluble scFv antibodies (80 mL + 20 mL 1% (w/v) BSA) was added. Plates were incubated for 2 h at 37 C. The wells were washed three times with 0.05% (v/v) PBST and PBS, respectively. HRP-conjugated anti-E tag antibody (Abcam, Cambridge, England, UK) diluted in 1% (w/v) BSA (1:5,000) (100 mL) was added, and the plates were incubated for 1 h at 37 C. The wells were washed as described above. 100 mL of TMB substrate was added and the reaction was stopped by adding 50 mL of 1M H 2 SO 4 per well. The absorbance at 450 nm was measured using FLUOstar Ò Omega (BMG Labtech, Germany). Positive responses were identified with OD 450 readings using a threshold of at least two times that of the control (1% BSA) (Della Cristina et al., 2015). Non-coated wells with sera and blank with antigen-coated wells without sera controls were used to identify background signals. All samples and controls were expressed as mean absorbance values ± standard deviation by subtracting the mean absorbance for non-coated wells with sera and wells without the sera, assuming each antigen was coated consistently between plates. Differences between test and control groups were analyzed using paired Student's t-test. Significance was set at p < 0.05 ( Ã ) and p < 0.01 ( ÃÃ ) for all tests.

Construction of scFv-eGFP
The anti-mKRAS G12V-34 scFv and enhanced green fluorescent protein (eGFP) were amplified separately and were purified using the gel extraction kit (iNtRON, Korea). The PCR products were assembled via SOE-PCR and double digested with XbaI and HindIII (NEB, Hitchin, UK). The digested PCR products were ligated into pET-28a expression vector (a molar vector to insert ratio of 1:3), using electroligase (NEB, Hitchin, UK). The ligation products were electroporated into E.coli BL21(DE3) at 1.8 kV, 5 ms. One milliliter of pre-warmed SOC media was added to the electroporated cells, which were then incubated at 37 C for 1 h with shaking at 150 rpm. The recovered cells were plated out into LB-kan plates and incubated at 37 C overnight. Positive clones were isolated with iNtRON Biotechnology DNA-spin TM plasmid DNA purification kit according to the manufacturer's protocol. The scFv-eGFP DNA region was sequenced using pET22b sequencing forward primer 5′-TAATACGACTCACTATAGGG-3′. Positive scFv-eGFP clones were expressed in E. coli BL21(DE3) in the presence of 1 mM IPTG for 4 h at 37 C with shaking at 200 rpm. The cells were harvested by centrifugation at 7,000×g for 10 min and resuspended in 20 mM Tris-CI buffer (pH 8.0) with the addition of 1 mM phenylmethylsulfonyl fluoride (PMSF) prior to sonication at 130 watt and 20 kHz for a total of 10 min on ice.

Protein expression, isolation, affinity purification and refolding of scFv-eGFP
His-tag fusion scFv-eGFP was expressed in E. coli BL21(DE3) in the presence of 1 mM IPTG for 3 h at 30 C with shaking at 150 rpm. The cells were harvested by centrifugation at 7,000×g for 10 min and resuspended in 20 mM Tris-CI buffer (pH 8.0) with the addition of 1 mM phenylmethylsulfonyl fluoride (PMSF) prior to sonication at 130 watt and 20 kHz for a total of 10 min on ice. The inclusion bodies were resuspended in inclusion body washing buffer (2 M urea; 20 mM Tris-CI; 0.5 M NaCI; 2% Triton X-100; pH 8.0) and solubilized in solubilization buffer (8 M urea; 20 mM sodium phosphate; 50 mM 2-mercaptoethanol; pH 7.8) overnight at RT. The solubilized inclusion bodies were further purified with nickel-nitrilotriacetic acid (Ni-NTA) resin (Qiagen, Hilden, Germany) by using a pH gradient (pH 4.5-7.8). The purified His-tag fusion scFv-eGFP were refolded by stepwise dialysis with phosphate refolding buffer containing L-arginine with decreasing urea concentrations (6 M, 4 M, 2 M and 1 M). Refolded proteins were filtered with a 0.22 µm filter and measured by using Bradford assay (Bio-rad, Irvine, CA, USA) with varying concentrations of BSA as a standard.

Construction of recombinant immunotoxin
Primers for PCR amplification of anti-mKRAS G12V-34 scFv and mHALT-1 were amplified separately using primers sets listed in Table 2. A total of two recombinant immunotoxins with opposing orientations were constructed using SOE-PCR, (1) scFv-mHALT-1; N-terminal anti-mKRAS scFv and C-terminal mHALT-1, and (2) mHALT-1-scFv; N-terminal mHALT-1 and C-terminal anti-mKRAS scFv. Both fragments were amplified separately by PCR. The PCR products were then purified using a gel extraction kit (iNtRON, Korea). For SOE-PCR, purified anti-mKRAS G12V-34 scFv and mHALT-1 amplicons were used. The assembled recombinant immunotoxins were double digested with XhoI and NcoI (NEB, Hitchin, UK) and ligated into pET22b expression vector (a molar vector to insert ratio of 1:3), using electroligase (NEB, Hitchin, UK). The resulting ligation products were electroporated into E. coli BL21(DE3) at 1.8 kV, 5 ms. One milliliter of pre-warmed SOC media was added to the electroporated cells, which were then incubated at 37 C for 1 h with shaking at 200 rpm. The recovered cells were plated out into LB-amp plates and incubated at 37 C overnight. All positive clone plasmids were isolated with iNtRON Biotechnology DNA-spin TM plasmid DNA purification kit according to the manufacturer's protocol. The recombinant immunotoxin DNA region was sequenced using pET22b sequencing forward primer 5′-TAATACGACTCACTAT AGGG-3′. Sequencing results were aligned using the Clustal Omega Multiple Alignment from BioEdit version 7.1.9 (Tom Hall, Fort Worth, TX, USA). The method for protein expression, isolation, affinity purification and refolding of recombinant immunotoxin was carried out according to scFv-eGFP protein as mentioned above.

SDS-PAGE and Western blot
Protein samples were loaded into 12% (w/v) sodium dodecyl sulfate (SDS) polyacrylamide gel and ran at 100 V for 100 min. The gel was stained with Coomassie blue (0.1% Coomassie blue, 45% methanol and 10% glacial acetic acid) and de-stained with 40% methanol and 10% glacial acetic acid solution overnight. The protein was transferred to a polyvinylidene fluoride (PVDF) membrane (Thermo Fisher, Waltham, MA, USA) for western blot analysis using a wet transfer system (Bio-rad, Irvine, CA, USA) for 90 min at 80 V in transfer buffer. The membrane was blocked with 1% (w/v) BSA in PBS for 1 h at RT to block protein-free binding spaces on the PVDF. Membrane was then incubated with HRP-anti-E tag antibody (Abcam, Cambridge, England, UK) (1:1,000) or His-detector nickel-HRP conjugate (KPL, Gaithersburg, MD, USA) (1:10,000) in 1% (w/v)

MTT cell viability assay
MTT assay was performed as described by Ng et al. (2019) with some modifications to determine the cytotoxic effects of recombinant immunotoxins using HCT-116 (ATCC Ò CCL-247 TM ), SW-480 (ATCC Ò CCL-228 TM ), and normal human dermal fibroblast (ATCC Ò PCS-201-012 TM ). Cell lines were prepared into cell suspensions with a density of 10,000 cells/well and incubated for 24 h to allow cell recovery and adherent to the wells. Different concentrations of recombinant immunotoxins (0, 5, 10, 15, 20, 25 and 30 mg/mL), mHALT-1 (0, 5, 10, 15, 20 and 25 mg/mL), anti-mKRAS G12V-34 scFv (0, 5, 10, 15, 20, 25 and 30 mg/mL) and glycerol (1%, 2.5%, 5.0%, 7.0%, 10% and 15%) diluted in DMEM media were added into respective wells and incubated at 37 C for 48 h. The wells with media only without cells served as a negative control, while wells treated with 5% (v/v) dimethyl sulfoxide (DMSO) served as a positive control. After removing the media, 50 mL of 5 mg/mL MTT solution was added into the wells and incubated at 37 C for another 3 h. The resultant purple formazan crystals were dissolved in 200 mL of DMSO. The absorbance intensity was measured by FLUOstar Ò Omega (BMG Labtech, Germany) at 570 nm. All assays were performed in triplicates, and the relative cell viability was expressed as a percentage relative to untreated controls. The IC 50 values were calculated in mg/mL or %, using Microsoft Excel. Differences between HCT116 or SW-480 treated cells against NHDF treated cells were analyzed using paired Student's t-test.

Cell-based ELISA assay
Cell-based ELISA assay was used to determine the binding activity of the refolded recombinant immunotoxins against colorectal cancer cells that expressed G12V, G13D or wt KRAS antigens. SW-480, HCT 116 and NHDF cell lines were seeded in 96-wells plates at 20,000 cells/well and incubated for 24 h. Cells were harvested after treatment with 0, 10 and 20 mg/mL of recombinant immunotoxins for 3 h. To identify background signals, non-coated wells with immunotoxin and blank with cell-coated wells without immunotoxin (untreated cells) were used as negative controls. Following three times of PBS washings, the cells were fixed with 4% (v/v) paraformaldehyde and incubated for 15 min at RT. After washing three times with PBS, the cells were blocked with 2% (w/v) BSA for 1 h to reduce non-specific binding. His-detector nickel-HRP conjugate (KPL, Gaithersburg, MD, USA) diluted at 1/10,000 in 1% (w/v) BSA (100 mL) was added, and the plates were incubated for 1 h at RT. The plates were washed three times with 0.05% (v/v) PBST. A hundred microliters of TMB substrate was added and the reaction was stopped by adding 50 mL of 1M H 2 SO 4 per well. The OD 450 was determined using FLUOstar Ò Omega (BMG Labtech, Germany). All samples and controls were expressed as net absorbance by subtracting the mean absorbance for non-coated wells with immunotoxins and untreated cells. The relative absorbance at 450 nm was presented as mean values ± standard deviation. Paired student's t-test was used to analyze differences between concentrations (0 mg/mL and 10 mg/mL or 20 mg/mL) and cell types (NHDF and HCT 116 or SW-480 cells). Significance was set at p < 0.05 ( Ã ) for all tests.

Immunofluorescences staining assay
Immunofluorescence staining assay was carried out to identify the cellular localization upon binding of scFv using HCT-116, SW-480 and normal human dermal fibroblast cells. Cell lines were prepared into 300 µL of cell suspensions with a density of 50,000 cells/mL and incubated in three separate µ-Slide 8-well chambers (Ibidi, Germany) until confluency of 90% was achieved. The wells were washed three times with PBS to remove the media and different IC 50 of purified scFv-eGFP proteins. After 24 h of incubation at 37 C, cells were washed three times with PBS and fixed with 150 mL of 2% paraformaldehyde in PBS. Permeabilization was performed by incubating the cells with 150 mL of 0.1% (v/v) Triton X-100 for 20 min. The wells were washed three times with PBS, and 100 mL of 300 nM DAPI (4′,6-diamidino-2-phenylindole) stain solution was added. After incubation for 20 min in the dark, the cells were washed three times with PBS. All images were captured and analyzed using a fluorescence microscope Axio Vert A1 (Carl Zeiss, Germany), under 630× magnification and enhanced using ZEISS ZEN software. Both 488 and 405 nm laser lines were used for the excitation of eGFP and DAPI, respectively.

Construction of the anti-mKRAS scFv library
Total RNA was extracted from mice splenocytes immunized with G13D and G12V mutated KRAS mimotopes (termed 164-D and 68-V respectively) developed in our previous study [9] that showed high Ig titer immune responses. As shown in the gel electrophoresis of Fig

Panning of scFv phage display library
The PEG-precipitated anti-mKRAS 164-D and 68-V scFv phages were subjected to three rounds of bio-panning against immobilized G13D and G12V control antigens (KRAS N-terminal protein, position 4-18), respectively. This step was carried out against control antigens instead of mimotopes, because the anti-mKRAS scFv antibodies would be used to recognize G12V/G13D k-ras antigens presented on the surface of cancer cells. For the 164-D library, the input titer was about 10 12 CFU/mL, while the eluted phages titer increased at the second (10 5 ) and third (10 6 ) round of panning. For the 68-V library, the input titer was about 10 13 CFU/mL, while the eluted phages titer increased at the second (10 6 ) and third (10 7 ) round of panning. These results indicated that more specific phage clones were binding to the respective peptide antigens during the third round of panning. As shown in Table 3, the phage recovery rate of second round panning showed about 10-fold increase in enrichment (10 −8 to 10 −7 ), which was increased approximately 60-fold by the third round (10 −8 to 10 −6 ). The overall enrichment was approximately 600-folds, which indicated successful enrichment of both 164-D and 68-V phages  specifically binding to G13D and G12V antigens, respectively. After three rounds of binding selection, positive clones were further identified through phage-ELISAs.

Detection of phage-displayed anti-mKRAS recombinant scFvs
Fifty randomly picked single bacterial clones for each G13D and G12V were tested against its respective peptides. Figure 2 shows a representative of fourteen positive clones, six of which were positive to G13D and eight were positive to G12V peptide. Other bacterial clones gave absorbance values similar to those of the negative control wells coated with BSA (data not shown). Four positive clones of anti-mKRAS G13D scFv phage (G13D-5, G13D-33, G13D-34 and G13D-41) ( Fig. 2A) showed the highest absorbance reading between 1.1 to 1.2, while a lower absorbance reading of 0.6 was obtained for G13D-18 and G13D-21. Background binding of the wt M13K07 negative control was minimal at an absorbance reading of 0.06. Meanwhile, four positive clones of anti-mKRAS G12V Figure 2 Identification of G12V and G13D mKRAS specific clones by phage-ELISA at 450 nm absorbance. After three rounds of panning, 50 individual randomly selected clones were infected with M13K07 helper phage to produce phage displayed scFv antibodies. Phage ELISA data for fourteen positive clones are shown. Clones that exhibited at least two times stronger ELISA signals than the BSA negative control (white bars) were deemed as specific clones. A total of (A) six positive clones of G13D and (B) eight positive clones of G12V were detected. Detection was performed using HRP-conjugated anti-M13 antibody and wt M13K07 helper phages were used as negative phage controls. Data were presented as mean values ± standard deviation indicated by error bars from triplicate experiments. Full-size  DOI: 10.7717/peerj.11063/ fig-2 (G12V-34, G12V-36, G12V-44 and G12V-48) (Fig. 2B) showed the highest absorbance reading between 1.5 to 1.7. A lower absorbance reading of 0.7 was obtained for G12V-2, G12V-7, G12V-28 and G12V-50, but was nevertheless still two times higher than the control. Background binding of the wt M13K07 negative control was once again minimal with an absorbance reading of 0.05. Overall, both G12V and G13D scFv phages bound higher to their respective mutant KRAS antigens compared to BSA, suggesting that the isolated phage clones were selective towards the KRAS antigen. Collectively, the results generated from selection and phage ELISA indicated that high performance phage display libraries were successfully constructed which resulted in fourteen positive clones which were independently propagated and further confirmed by monoclonal ELISA.

Sequence analysis of anti-mKRAS scFv clones
Sequencing of the 14 positive clones revealed that they corresponded to four distinct clones, with two anti-mKRAS G13D antibodies and another two anti-mKRAS G12V antibodies. For G13D clones, G13D-18 and G13D-21 shared identical sequences, while G13D-5, G13D-33, G13D-34, G13D-41 shared the same sequence. For G12V clones, G12V-2, G12V-7, G12V-28, G12V-50 had the same sequence, while G12V-34, G12V-36, G12V-44, G12V-48 shared the same sequence. This reflected the high amount of scFv phages enriched after three rounds of panning, suggesting that the lower affinity scFvs were eliminated after three rounds of panning. Thus, the scFv candidate list was further short-listed to four clones, two from each cluster for G13D and G12V, respectively: G12V-34, G12V-50, G13D-5 and G13D-18. NCBI IgBLAST (http://www.ncbi.nlm.nih.gov/ igblast/) (Lefranc, 2001) was used to determine the number of amino acids, framework (FR), and complementary determining regions (CDRs) of the V H and V L domains. All the deduced amino acid sequences and CDRs are shown in Table 4. Taken together, the sequence homology analysis indicated that four V H germline genes (V H 1-4, V H 1-22, V H 1-56 and V H 1-69) and four V K germline genes (V K 4-59, V K 4-68, V K 6-17 and V K 12-46) were found in these murine immune antibody libraries. There were no sequence data that completely matched the variable regions of these antibodies in the database, suggesting the occurrence of new somatic hypermutations within the hypervariable CDRs.

Expression and detection of soluble anti-mKRAS scFv antibodies by monoclonal-ELISA
Four clones (G12V-34, G12V-50, G13D-5, G13D-18) which were positive in phage-ELISA were selected for scFv expression in E. coli HB2151 cells. Expression of anti-mKRAS scFv proteins by the lac promoter was induced in the presence of 1 mM IPTG for 4 h at 30 C. An anti-mKRAS scFv band at 30 kDa was detected using anti-E tag HRP-conjugated antibodies, further confirmed its expression in the periplasmic fraction (Fig. 3).
No bands were observed in the uninduced periplasmic fraction. The specificity of soluble anti-mKRAS scFvs expressed in the periplasmic space was tested with a monoclonal-ELISA assay using anti-E-tag antibodies. Antigen controls (G13D, G12V) and wildtype K-ras antigens were employed to investigate the specificity of anti-mKRAS scFvs. ELISA results showed that both G13D clones (G13D-5, G13D-18) and G12V clones (G12V-34 and G12V-50) bound specifically to G13D and G12V antigens, respectively. The OD 450 readings of each scFv antibody are as shown in Fig. 4, and the absorbance reading indicated a higher level for G12V-50 at 0.566, followed by G12V-34 at 0.459, G13D-18 at 0.229, and G13D-5 at 0.132. Additionally, cross-reactivity of scFv antibodies towards wildtype and G13D/G12V mutated KRAS antigens were also evaluated. Although G13D clones showed binding affinity towards the G13D antigen, it also showed cross-reactivity towards wildtype (1-fold) and G12V (2-fold), which was undesirable. Similarly, while G12V-50 clone exhibited high specificity towards G12V, it was also found to highly cross-react with wildtype KRAS antigens at the same absorbance reading of 0.582 with G12V. Only the G12V-34 clone showed specific reactivity to G12V peptide, with weak cross-reactivity to wildtype (p = 0.005) and G13D (p = 0.01).

Construction of recombinant immunotoxins
Two recombinant immunotoxin candidates were constructed with the scFv moiety arranged in two opposing orientations (Fig. 5) in order to determine which arrangement would be more effective. Figure 6A shows the amplified anti-mKRAS G12V-34 scFv PCR product with a band size of 801 bp (lane 1) for N-terminal scFv and a band size of  . 6B) with the expected 576 bp fragment (lane 1) for N-terminal mHALT-1, and 558 bp fragment (lane 2) for C-terminal mHALT-1. The fused SOE-PCR products of 1,359 bp are shown in Fig. 6C. Both scFv-mHALT-1 (lane 1) and mHALT-1-scFv (lane 2) were successfully amplified with NcoI and XhoI restriction sites incorporated for subsequent cloning into pET22b vector. Positive transformants in E. coli BL21(DE3) cells were screened by cPCR and confirmed by DNA sequencing (data not shown).

Expression, purification and refolding of recombinant immunotoxins
The 12% SDS-PAGE electrophoresis results (Fig. 7A) indicated that both scFv-mHALT-1 and mHALT-1-scFv was successfully expressed in E. coli BL21(DE3) with an intense band at~44.4 kDa and a moderate intense band at~46.8 kDa, respectively. Variations in size were attributed to the failure of pelB leader sequence (2.4 kDa) to be cleaved off. Since both were expressed as insoluble recombinant proteins, inclusion bodies were isolated by ultrasonication, and the pellet fraction was washed with urea and Triton X-100 to remove proteinaceous contaminants that could affect protein refolding yields, followed by solubilization in chaotropic reagents such as urea. The solubilized inclusion bodies of scFv-mHALT-1 and mHALT-1-scFv were further purified by immobilized affinity chromatography using Ni-NTA resin and eluted by lowering the pH from 7.8 to 4.5. The purified and refolded recombinant immunotoxins scFv-mHALT-1 (Lane 1) and mHALT-1-scFv (Lane 2) of~44.4 kDa and~46.8 kDa respectively, is as shown in Fig. 7B, respectively. The final yield of scFv-mHALT-1 and mHALT-1-scFv immunotoxins were 49 mg/mL and 108 mg/mL with the recovery of 7.0% and 24.0%, respectively. Recombinant immunotoxins exhibit cytotoxic effects against KRAS-positive cancer cells As shown in Fig. 8A, the scFv-mHALT-1 immunotoxin treated HCT 116 and SW-480 cells were killed at an IC 50 of 16.33 mg/mL and 15.60 mg/mL, respectively. As the concentration of scFv-mHALT-1 immunotoxin reached 15 mg/mL, the glycerol solvent percentage Figure 4 Cross-reactivity of soluble anti-mKRAS scFvs with control K-ras (G13D, G12V) and wildtype K-ras antigens in monoclonal-ELISA. Four clones (G13D-5, G13D-18, G12V-34 and G12V-50) were tested for binding specificity towards wt (light grey bars), G13D (dark grey bars) and G12V (black bars) control peptides. Wells were coated with 10 mg/mL of each antigen and bound antibodies were detected with anti E-tag antibodies (1:5,000). All values were presented as mean values ± standard deviation against background signals from plates without the addition of sera and non-coated wells, measuring at 450 nm wavelength. Significant differences were denoted as p-values < 0.05 ( Ã ) and <0.01 ( ÃÃ ) when compared against G12V or G13D antigen controls. was also concurrently increased to 7.6%. Based on glycerol solvent controls, concentrations above 7.6% were shown to affect cell viability, and the absorbance readings recorded from this point onwards were no longer reliable. It is also notable that minimal cytotoxic effects of scFv-mHALT-1 immunotoxin were observed on NHDF cells, which concurred with the fact that NHDF cells do not express the G12V k-ras antigen. In Fig. 8B, mHALT-1-scFv immunotoxin exhibited cytotoxic effects on SW-480 cells and also HCT 116 cells but with higher IC 50 values of 25.39 mg/mL and 31.91 mg/mL respectively compared to the scFv-mHALT-1 immunotoxin. Compared to scFv-mHALT-1, mHALT-1-scFv contains lower glycerol concentrations, because higher concentration of mHALT-1-scFv was obtained following refolding steps. At a 30 mg/mL concentration of mHALT-1-scFv, approximately 6.9% of glycerol content was present. This was a concentration that did not affect cell viability. No cytotoxicity effects were observed for mHALT-1-scFv immunotoxin treated NHDF cells even at concentrations as high as 30 mg/mL, which was a desirable observation. When comparing against mHALT-1 and anti-mKRAS G12V-34 scFv standalone controls (Figs. 8C & 8D), the mHALT-1-scFv immunotoxin was shown to re-exhibit its cytotoxicity effects on KRAS-positive cancer cells with minimal killing of KRAS-negative cells, thus reinforcing the notion that the scFv moiety acted as a targeting tool for the mHALT-1 toxin.
Determinaton of recombinant immunotoxin specificity using cell-based ELISA As shown in Fig. 9, no significant binding for scFv-mHALT-1 (p = 0.473) and mHALT-1-scFv (p = 0.437) immunotoxins were detected towards NHDF cells (wtKRAS-negative), even up to concentrations of 20 mg/mL. This observation was expected with KRAS-  Immunofluorescence localization of scFv-eGFP To study the localization of scFv via eGFP, immunofluorescence staining was performed on NHDF cells (wtKRAS-negative), SW-480 cells (G12V KRAS-positive), and HCT 116 (G13D KRAS-positive) cells as shown in Fig. 10. The cell lines were incubated with different IC 50 of scFv-eGFP for 24 h followed by countered stained with DAPI. As shown in Fig. 10, blue fluorescent cells were observed due to dsDNA's binding to AT regions in all the DAPI-stained nuclei. Green fluorescent was observed in all the permeabilized cells treated with the scFv-eGFP protein with different localization. Of note, green fluorescent was found predominantly on the inner membrane of SW-480 cells, while the scattering of green fluorescent can be observed in the cytoplasm of HCT 116 and NHDF cells. This is suggesting the binding of scFv was specifically tethered to the cell membranes of SW-480 that express G12V k-ras antigen.

DISCUSSION
Phage display antibody technology offers a convenient tool for isolation of high affinity and specific antibody fragments, which can be applied clinically as a potential targeting vehicle for immunotherapy, diagnosis, and drug delivery purposes in cancers. While, KRAS is known to be expressed and anchored to the inner membrane of cells, there were several reports highlighting the MHC-I presentation of mutated KRAS on the cell surface, thus rendering it a plausible and druggable target using immunotherapeutic approaches  (June, 2016;Tran et al., 2016). Therefore, the present study was aimed at identifying anti-mKRAS antibodies for KRAS colorectal cancer by utilizing phage display libraries to screen for antibodies that selectively target G12V and G13D KRAS-positive colorectal cancer cells. ScFv constructs were generated by randomly joined V H and V L through SOE-PCR. Such random pairings generated a high diversity of scFv fragments, yielding antibodies with novel specificities alongside natural combinations in vivo by the murine immune system (Griffiths & Duncan, 1998). The 15-amino acid linker was selected to assemble the V H and V L gene due to several reasons, including the small size of glycine and serine amino acids, which allows the mobility of the connecting functional domains, reduces unfavorable interactions between the linker and protein moieties by the formation of hydrogen bonds with water molecules, and to provide stability at levels comparable to native antibodies (Takkinen et al., 1991;Chen, Zaro & Shen, 2013).
The combinatorial library for mice V H and V L chains were effectively and specifically amplified from 164-D and 68-V immunized mice splenocytes, and as a result, a large pool of scFvs available for affinity selection by bio-panning was constructed with a complexity of 3.4 × 10 6 and 2.9 × 10 6 independent transformants, respectively. The panning protocol involved three rounds of selection against plates coated with G12V and G13D antigens, with an enrichment factor similar to those previously reported (Salvatore et al., 2002;Deng et al., 2005;Gong et al., 2016). Despite a moderate library size, this study demonstrated that utilization of this immunized library in conjunction with an optimized panning protocol was sufficient to identify specific KRAS G12V and G13D anti-mKRAS scFvs.
The successful isolation of specific scFv binders to mKRAS G12V antigen is analogous to a guidance system that is able to carry a very lethal toxin load, thus enabling the selective killing of KRAS-positive colorectal cancer cells while sparing normal cells. Consequently, the anti-mKRAS G12V antibody developed through this study was fused with a mutated pore forming Hydra magnipapillata toxin, giving rise to a total of two recombinant immunotoxins in two orientations (scFv-mHALT-1 and mHALT-1-scFv). Although the soluble expression of immunotoxins is preferable, the high yield of refolded immunotoxins from inclusion bodies is a promising approach to isolate high purity immunotoxins. Previous studies have shown that his-tag recombinant proteins can be purified efficiently under denaturing conditions (Lemercier, Bakalara & Santarelli, 2003;Arnau, Lauritzen & Pedersen, 2006). Alternatively, functional cytoplasmic expression can be attempted using mutant strains with a more oxidized cytoplasm or by fusing to cytoplasmic localization proteins. Also, the use of yeast or mammalian host cells can be an alternative strategy to express secreted and soluble recombinant immunotoxins, thus eliminating the need for in vitro refolding (Della Cristina et al., 2015).
MTT assay results of scFv-mHALT-1 and mHALT-1-scFv recombinants showed that these immunotoxins could elicit cytotoxic effects on KRAS-positive cancer cells without affecting normal fibroblast cells. Three to four HALT subunits are required to form pores that cause cell lysis, thus higher concentrations of immunotoxins are required to kill mutated KRAS cells in this study as compared to other immunotoxin studies showing lower IC 50 values within the ng/mL range (Hashimi et al., 2013;Liew et al., 2015;Xie et al., 2017). Based on these results, cell-based ELISA assay further confirmed the binding specificity of mHALT-1-scFv immunotoxin towards G12V KRAS SW-480 and G13D KRAS HCT 116 cancer cells.
While data obtained from this proof of concept study is promising, several concerns on its application should also be highlighted. This includes vascular leak syndrome and other negative effects of immunotoxins on immune cells that present mutated KRAS antigens such as dendritic cells. Such APCs typically overexpress checkpoint molecules to deter CD8 killing, but would have no protective effects against immunotoxins (Li et al., 2017;Wei, Duffy & Allison, 2018). Meanwhile, the immunogenicity of the HALT can be reduced by deimmunizing the toxin moiety through point mutations on epitope regions (Schmohl et al., 2018).
Therefore, additional validation experiments including testing of immunotoxin stability, serum half-life and pre-clinical in vivo safety assessment are required. Continued efforts to produce soluble and humanized antibody versions from mouse scFv could further enhance its therapeutic potential.

CONCLUSIONS
In conclusion, this work is the first to describe the development of recombinant immunotoxins to target KRAS-positive cancer cells coupled with a potent Hydra toxin. Hence, the development of mHALT-1-scFv immunotoxin is potentially useful as an immunotherapeutic application against KRAS-positive malignancies, thereby reducing dose-related toxicities and undesirable adverse side effects commonly seen in conventional non-targeted chemotherapeutic regimens. Additionally, the anti-mKRAS scFv can also be potentially developed as a predictive tool through histological, cytological or serological based applications for the detection of the G12V mKRAS genotypes during cancer screening or diagnosis. Various toxins, biomarkers or drugs can also be fused to this scFv sequence, thus unleashing vast opportunities for further downstream innovations.