Open Access

Construction and application of a human scFv phage display library based on Cre‑LoxP recombination for anti‑PCSK9 antibody selection

  • Authors:
    • Yuan Dong
    • Fanwei Meng
    • Zhiheng Wang
    • Tianyi Yu
    • An Chen
    • Song Xu
    • Jianming Wang
    • Moli Yin
    • Lu Tang
    • Chuanmin Hu
    • Huiyan Wang
    • Jianhui Cai
  • View Affiliations

  • Published online on: December 16, 2020     https://doi.org/10.3892/ijmm.2020.4822
  • Pages: 708-718
  • Copyright: © Dong et al. This is an open access article distributed under the terms of Creative Commons Attribution License.

Metrics: Total Views: 0 (Spandidos Publications: | PMC Statistics: )
Total PDF Downloads: 0 (Spandidos Publications: | PMC Statistics: )


Abstract

A large human natural single‑chain fragment variable (scFv) phage library was constructed based on Cre‑LoxP recombination, and used to successfully identify antibodies against proprotein convertase subtilisin/kexin type 9 (PCSK9). The library was derived from 400 blood samples, 30 bone marrow samples, and 10 cord blood samples from healthy donors. Lymphocytes were isolated from each sample and cDNA was synthesized using reverse transcription‑quantitative PCR. Two‑step overlap PCR was then used for scFv synthesis using a LoxP peptide as the linker. The scFv gene was inserted into the phagemid vector pDF by enzymatic digestion and ligation, and then transformed into Escherichia coli (E. coli) SS320 to establish a primary antibody library in the form of scFvs. A primary antibody library consisting of 5x107 peripheral blood and umbilical cord blood sources, as well as a primary antibody library of 5x107 bone marrow samples were obtained. By optimizing the recombination conditions, the primary phage library was used to infect E. coli BS1365 strain (which expresses the Cre enzyme), and a human scFv recombinant library with a size of 1x1011 was obtained through Cre‑LoxP enzyme‑mediated heavy and light chain replacement and recombination. This constructed recombinant library was employed to screen for antibodies against recombinant PCSK9. After four rounds of selection, a fully human antibody (3D2) was identified with a binding affinity of 1.96±1.56ⅹ10‑10 M towards PCSK9. In vitro, the PCSK9/low‑density lipoprotein receptor (LDLR) pathway of Hep‑G2 cells was inhibited by 3D2 treatment, thereby increasing LDL uptake in these cells. In addition, combination treatment with 3D2 and statin was more effective at increasing LDLR levels than treatment with 3D2 or statin alone. Furthermore, 3D2 resulted in a 3‑fold increase in hepatic LDLR levels, and lowered total serum cholesterol by up to 61.5% in vivo. Taken together, these results suggest that the constructed human Cre‑LoxP scFv phage display library can be used to screen fully human scFv, and that 3D2 may serve as a candidate hypolipidemic therapy.

Introduction

For several decades, phage display technology has been widely used for the selection of fully human antibodies. At present, the most advanced phage antibody library is the Cre-LoxP recombinant antibody library, which was constructed based on the single vector pDF (1). The Cre-LoxP system mediates site-specific DNA recombination, and is thereby the core to gene-targeting technology. Furthermore, the phagemid vector pDF contains two non-homologous LoxP sites on both ends of the antibody gene, which can be recognized by Cre recombinant enzyme. With the application of Cre-LoxP recombination in the construction of a phage display antibody library, different heavy (VH) and light (VL) chain variable regions can be freely recombined, allowing for the indefinite repetition of gene recombination, so as to realize the construction of antibody libraries with large repertoire size and diversity (2).

Proprotein convertase subtilisin/kexin 9 (PCSK9), a member of the subtilisin-like proprotein convertase family, serves a role in low-density lipoprotein receptor (LDLR)-lysosome pathways, as well as cholesterol and fatty acid metabolism (3,4). Cohen et al (5) reported that genetical alterations in the PCSK9 gene, such as gain-of-function mutations, could elevate serum LDL-C levels (>300 mg/dl), and by contrast, that loss-of-function mutations may lower serum LDL-C (<100 mg/dl), resulting in cardiovascular disease (6). Various approaches to inhibiting PCSK9 have been reported for hypercholesterolemia treatment, including the use of antisense oligonucleotides, lipidoid nanoparticle formulated short interfering RNAs directed against PCSK9 mRNA (7,8), antibodies and LDLR subfragments (9).

Monoclonal antibodies are the primary type of commer-cial PCSK9 inhibitors. As a first generation antibody, the murine monoclonal antibody possesses desirable immunogenicity (10), though fully-humanized monoclonal antibodies exhibit superior properties, overcoming the shortcomings of murine monoclonal and chimeric monoclonal antibodies. As such, fully-humanized antibodies have become integral to the development of antibody-based therapeutics (11).

In the present study, three 5×107 primary antibody libraries were constructed using 400 blood samples, 30 bone marrow samples and 10 cord blood samples, respectively, all from healthy donors. By optimizing the recombination conditions, a fully human recombinant single-chain fragment variable (scFv) phage library (size, 1×1011) was obtained. Sequencing 60 random scFvs showed that the diversity of antibody genes in the library was consistent with the antibody gene distribution of the human embryo line. The PCSK9 antigen was used to screen the fully human scFv library, and five different scFv sequences were obtained that specifically targeted to PCSK9. After constructing the expression vectors of five selected scFvs, transient expression revealed that anti-body expression levels ranged from 50-200 mg/l; 3D2 was found to have a coordination effect when used with statins, which promoted the absorption of LDLR and LDL in hepatic Hep-G2 cells. Animal experiments showed that 3D2 increased the expression of LDLR and significantly reduced total serum cholesterol. Therefore, the results of the present study indicate that the constructed large fully human scFv phage display library can be used to screen for therapeutic antibodies, and that 3D2 has great potential for use as a candidate PCSK9 inhibitor.

Materials and methods

Construction of a fully human scFv phage display antibody library RNA extraction

The blood of 400 healthy donors, the bone marrow of 30 healthy donors and cord blood samples from 10 healthy donors were kindly provided by the Affiliated Hospital of Jilin Medical University (Jilin, China) between January 2016 and June 2016. The study was approved by the ethics committee of Jilin Medical University (2016-LW012) prior to the start of the study. All donors were healthy volunteers aged 18-60, and of both sexes. The lymphocytes from these samples were isolated using a human blood lymphocyte isolation kit (cat. no. P8610; Solarbio Science and Technology Co., Ltd.) Total RNA was extracted from the lymphocytes using TRIzol® reagent (SinoGene Scientific Co., Ltd.), and the RNA samples were analysed by 1% agarose gel electrophoresis.

cDNA synthesis

First chain cDNA was synthesized with a reverse transcription kit (Shanghai Hifun Biotechnology Co., Ltd.) according to the manufacturer's instructions, using total RNA as the template.

PCR amplification of scFvs

Stepwise PCR (Fig. 1) was used to synthesize the scFvs for library construction, using a total of 22 different and degenerate primers as described in Table I. Firstly, VH and VL (Vκ/Vλ) chain variable regions were amplified using cDNA as the template. The PCR reactions (50 µl each) contained 2 µl cDNA (40 ng/µl), 1 µl forward primer (10 µM), 1 µl reverse primer (10 µM), 0.5 µl Q5 DNA polymerase, 1 µl dNTP mixture (10 µM) and 34.5 µl ddH2O. The PCR thermocycling conditions were as follows: 1 Cycle at 95°C for 3 min; 30 cycles at 95°C for 30 sec, 57°C for 30 sec and 72°C for 45 sec; and a final incubation at 72°C for 5 min. Then, the PCR products (ranging around 400 bp) were recovered and further amplified separately for VH- and VL-linkers with a BssHII/NheI recognition site and LoxP511 sequence. Finally, the VH- and VL-linkers were joined by overlap PCR. The PCR program comprised two steps: 1 Cycle at 95°C for 3 min; and 2 cycles at 95°C for 1 min, 60°C for 30 sec, 55°C for 50 sec and 72°C for 1 min. Then the HVKEXfor/PVHback primers were added according to previous reports (1). The subsequent PCR conditions were as follows: 5 Cycles of 95°C for 1 min, 60°C for 30 sec, 55°C for 50 sec and 72°C for 1 min; and 23 cycles at 95°C for 1 min, 60°C for 45 sec, and 72°C for 1 min. The amplicons (ranging around 800 bp) were recovered and stored at −80°C.

Table I

Primers used for gene amplification.

Table I

Primers used for gene amplification.

Forward primers for VH
PVHfor1TAT CCT CGA GCG GTA CCS AGG TSC AGC TGG TRC AGT CTG G
PVHfor2TAT CCT CGA GCG GTA CCS AGG TGC AGC TGK TGG AG
PVHfor3TAT CCT CGA GCG GTA CCC AGR TCA CCT TGA AGG AGT CT
PVHfor4TAT CCT CGA GCG GTA CCC AGS TGC AGCT RCA GSA GT

Reverse primer for VH

PVHbackGGC GGA TGC GCT AGC TGA RGA GAC RGT GAC C

Forward primers for Vκ

HVK1TAAT GCGCGC ATG CCG MCA TCC RGW TGA CCC AGT CTC C
HVK2TAAT GCGCGC ATG CCG ATA TTG TGA TGA CYC AGW CTC C
HVK3TAAT GCGCGC ATG CCG AAA TTG TGW TGA CRC AGT CTC C
HVK4TAAT GCGCGC ATG CCG AAA CGA CAC TCA CGC AGT CT

Reverse primers for Vκ

PVK1backGGT CGA CCC TCC GGA ACG TTT GAT HTC CAS CTT GGT
PVK2backGGT CGA CCC TCC GGA ACG TTT AAT CTC CAG TGC TGT

Forward primers for Vλ

HVL1TAAT GCGCGC ATG CCC AGT CTG TGY TGA CKC AGC C
HVL2TAAT GCGCGC ATG CCC AGT CTG CCC TGA CTC AGC C
HVL3TAAT GCGCGC ATG CCT CCT MTG AGC TGA CWC AG
HVL4TAAT GCGCGC ATG CCC AGT YTG TGC TGA CTC AAT C
HVL5TAAT GCGCGC ATG CCC AGR CTG TGG TGA CYC AGG AG
HVL6TAAT GCGCGC ATG CCA ATT TTA TGC TGA CTC AG
HVL7TAAT GCGCGC ATG CCC AGK MTG RGC TGA YGC AGC CAC CCT C

Reverse primers for Vλ

PVLbackGGT CGA CCC TCC GGA ACC TAG GAC GGT SAG CTT GGT CCC

Extended reverse primer for VH, used for overlap PCR

PVHLCCA TAA CTT CGT ATAATG TAT ACT ATA CGAAGT TAT_ CC_T CGA GCG GTA CC
Extended forward primers for Vκ and Vλ
HVKEXforGAG GAG GAG ATA ATG CGC GCA TGC C
Extended reverse primers for VL, used for overlap PCR
PVLEXGGA TAA CTT CGT ATA GTA TAC ATT ATA CGAAGT TAT__ GGT CGA CCC TCC GGAAC

[i] Sequences with single underline were recognition sites (BssHII recognition site GCGCGC; NheI recognition site GCTAGC); sequences with double underline were overlap sequence (LoxP511 gene). The degenerate codons were bolded, where R=purine (A or G), Y=pyrimidine (C or T), M=A or C, K=G or T, S=G or C, H=A, T or C and W=A or T.

Preparation of M13KO7-infected electrocompetent E. Coli SS320 cells

According to previously described methods (12), 500 µl of E. ColiSS320 (Tet+) at exponential phase (OD600=0.8) was mixed with 6×108 pfu/ml of M13K07 helper phage (Kana+) and 4 ml of 2YT top agar. The mixture was poured onto LB agar plate (supplemented with 5 µg/ml tetracycline) and incubated for 20 h at 37°C. A single plaque was selected and inoculated into 1 ml of 2YT/kana/tet medium (10 g/l yeast extract, 16 g/l tryptone, 5 g/l NaCl, 15 g/l granulated agar, 25 µg/ml kanamycin and 5 µg/ml tetracycline; pH to 7.0) and incubated for 8 h at 37°C. The culture was then transferred to 250 ml of 2YT/kana medium and incubated with shaking at 200 × g at 37°C overnight. 5 ml of the overnight culture was inoculated into 900 ml of superbroth/tet/kana medium (12 g/l tryptone, 24 g/l yeast extract, 5 ml/l glycerol, 0.17 M KH2PO4, 0.72 M K2HPO4, 5 µg/ml tetracycline and 25 µg/ml kanamycin) and incubated with shaking at 200 × g at 37°C until the absorbance had reached 0.8 (OD600). The culture was centrifuged at 5,000 × g for 15 min at 4°C. The supernatant was discarded and the pellet was washed with cold 1 mM HEPES (pH 7.5) 3 times. By centrifuging at 5,000 × g for 15 min at 4°C, the precipitate was first resuspended with 20 ml of 1 mM HEPES containing 10% (v/v) glycerol and finally resuspended with 1 ml of 10% (v/v) cold glycerol. Then the cells were snap frozen with liquid nitrogen and stored at −80°C.

Transformation and library construction

According to previously described methods (12), the scFv fragments (1 µg) were digested using BssHII and NheI (1 µg each) and inserted into a pDF vector (also digested with BssHII and NheI) (Fig. 4; Amp+, 1.78 µg) using T4 ligase at 16°C. The constructs were purified using the QIAquick PCR purification kit (cat. no. 28104; Qiagen, Dusseldorf) and electroporated into 25 µl M13KO7-infected SS320 electrocompetent cells at 2,500 V, 200 Ω, 25 µF and 4 msec. Then, 25 ml SOC media was added (5 g/l bactoyeast extract, 20 g/l bactotryptone, 0.5 g/l NaCl, 0.2 g/l KCl, 0.01 M MgCl2 and 0.02 M glucose; pH 7.0) and the cultures were incubated at 37°C for 40 min with agitation (200 × g). Subsequently, 10 µl of the culture was diluted (1:108) and plated onto 2YT/Amp agar medium (2YT supplemented with 50 µg/ml ampicillin) for overnight incubation at 37°C, which was used to evaluate and calculate the size of the scFv phage display library. The rest of the culture was inoculated into 1 l 2YT/Amp/kan/IPTG media (supplemented with 50 µg/ml ampicillin, 25 µg/ml kanamycin and 25 mM IPTG) and incubated overnight at 37°C with gentle shaking. The culture was then centrifuged at 10,000 × g for 15 min at 4°C. The supernatant was removed, mixed with 20% PEG 8000-2.5M NaCl solution, and left to stand on ice for 20 min. After centrifuging at 10,000 × g for 20 min at 4°C, the phage precipitate was resuspended with 20 ml 1X PBS to generate the primary library. The recombinant phage antibody library was constructed by infecting the mid-log-phage-BS1365 strain (Cre+, OD600 0.8) with 250 µl of the primary library (1×1012 pfu/ml) at a multiplicity of infection (MOI) of 200:1. After standing at 37°C for 1 h, 400 ml 2YT/Amp media was added and the culture was incubated at 37°C (200 × g) until the absorbance had reached 0.5 (OD600). Then, 5 ml helper phage (M13KO7; 5×1010 pfu/ml) was added at a ratio of 3000:1, and the culture was gently shaken. After standing at 37°C for 1 h, the phage library was collected and used to infect 1 liter of the mid-log-phase-OMNImax2 strain (Tet+, OD600 0.8) at an MOI ≤1; 10 ml of M13KO7 (5×1010 pfu/ml) was then added and shaken at 30°C for 1 h. The resulting recombinant human scFv phage library was stored in glycerol at −80°C. The library size was calculated as follows: Library size (cells/ml)=number of colonies ×103 × dilution ratio.

Human recombinant PCSK9 protein preparation

Human PCSK9 (Genbank AX207686) was amplified from Hep-G2 cells by reverse transcription PCR, and its sequence was confirmed by sequencing. The Hep-G2 cells (Shanghai Biowing Applied Biotechnology Co., Ltd.) were authenticated as a hepatocellular carcinoma cell line using short tandem repeat profiling, which was also conducted by Shanghai Biowing Applied Biotechnology Co., Ltd. The cells were maintained in RPMI-1640 (Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS, Gibco) and anti-biotics (100 IU/ml penicillin and 100 mg/ml streptomycin) at 37°C and 5% CO2. Then, the gene was cloned into the pcDNA3.1-hIgG1FC vector along with a 6X His tag, between the HindIII and XhoI sites The recombinant plasmid (30 µg) was then used to transfect 5×107 CHO-K1 cells using Lipofectamine® 6000 transfection reagent (cat. no. C0528; Beyotime Biotechnology Co., Ltd.). Recombinant PCSK9 protein was purified from the cell supernatant using PrepEase® His-tag protein purification resin (Biomart Co., Ltd.). The CHO-K1 cells were obtained from the American Type Culture Collection (ATCC® CCL-61™) and maintained in RPMI-1640 (Gibco) supplemented with 10% (v/v) FBS (Gibco) and antibiotics (100 IU/ml penicillin and 100 mg/ml streptomycin) at 37°C and 5% CO2.

Selection of antibodies against human PCSK9

The recombinant human scFv phage library was screened against PCSK9 as previously reported (13,14). Immunotubes were coated with detection antigen (purified PCSK9) diluted to 100 µg/ml (2 ml/immunotube) and incubated at 4°C overnight. After washing 3 times with PBS, the immunotubes were blocked with 2% (w/v) BSA in PBS. The diluted recombinant phage particles (1011 pfu/ml) were subsequently added into the coated immunotubes and incubated at 37°C for 2 h. The immunotubes were then washed 10 times with PBST (PBS containing 0.1% Tween-20) and 10 times with PBS only, to remove unbound phage. Bound phage was eluted with 100 mM triethylamine (Sigma-Aldrich; Merck KGaA) and neutralized with 500 µl of 1 M Tris-HCl (pH 7.5). Mid-log phase E. coli TG1 cells were infected with the eluted phage, plated onto several 2YT-agar plates and cultured until individual colonies had formed. Each clone was used to transform mid-log phase E. coli TG1 cells (using M13KO7 phages), which were screened using a competitive phage-ELISA. The screening process was repeated for 4 rounds.

Expression and purification of antibodies against PCSK9

The selected human PCSK9 scFvs in the phages were amplified by PCR and inserted into a modified mammalian expression vector (pFUSE2-CLIg-hk and pFUSE-CHIg-hG1; InvivoGen). The plasmids were used to transfect 293 cells using the ExpiFectamine™ 293 Transfection Kit (Gibco; Thermo Fisher Scientific, Inc.), and the culture supernatant was collected after 8 days. The antibodies were separated and purified from the supernatant using a Protein A Sepharose column (GE Healthcare Bio-Sciences). The integrity of the antibodies was determined by ELISA and SDS-PAGE.

Biolayer interferometry (BLI) analysis

The affinity measurements were analysed by BLI technology using a Pall ForteBio's BLI-based Octet RED96e system (ForteBio; Sartorius AG) combined with Octet version 13.1.0 software, according to the manufacturer's recommendation. The purified anti-PCSK9 antibodies were biotinylated using the Biotin Labelling Kit-NH2 (Dojindo Molecular Technologies, Inc.). Briefly, 1 mg anti-PCSK9 antibody was mixed in a filtration tube with 100 µl NH2-reactive biotin (10 mM), and incubated at 37°C for 10 min. The tube was then centrifuged at 6,000 × g for 30 min at 4°C to remove uncombined biotin reagent. The biosensors in the Pall ForteBio's BLI-based Octet® RED96e system were hydrated as follows: 200 µl 1X Kinetics Buffer was added to a each well of a round-bottom microplate (the hydration plate), which was placed into the biosensor tray of the system. The biosensors were dipped into the wells for 10 min; 200 µl 50, 25, 5, 2.5 or 0.5 µg/ml biotinylated anti-PCSK9 antibody was added to the wells of a black 96-well polypropylene microplate (the sample plate), followed by 200 µl PCSK9 antigen (200 nM). Then, 200 µl biotinylated anti-PCSK9 antibody (50 µg/ml) was added to the wells, and biotinylated anti-PCSK9 antibody without PCSK9 antigen served as the negative control. The plate temperature was set to 30°C in the Octet® software, the sample plate was placed on the sample plate stage inside the Octet® system, and the biosensor hydration assembly was placed on the biosensor stage. The settings for the affinity assay were as follows: Equilibration for 60 sec, loading for 400 sec, baseline for 600 sec, association for 400 sec and dissociation for 3,600 sec at a flow rate of 1,000 × g. After the assay, the shapes of the individual binding curves were observed and the binding association (Kon) and dissociation rate constants (Koff) are calculated from a global fitting analysis assuming a 1:1 binding model using the Octet Data Analysis software. The affinity constant (KD) was determined from the ratio of the rate constants: KD=Koff/Kon.

Western blot analysis

Hep-G2 cells or liver tissues were lysed to determine the expression levels of LDLR. Cells were treated with radio-immunoprecipitation assay buffer and protease inhibitor cocktail (both from Millipore Sigma) and incubated for 30 min to stop proteolysis. The lysate was centrifuged at 12,000 × g for 30 min (4°C), and 60 µg total protein was electrophoresed on a 10% polyacrylamide gel; the proteins were transferred to a PVDF membrane for 2 h at 300 mV, and then washed 5 times with TBST [50 mM Tris-Cl (pH 7.5), 100 mM NaCl and 0.5% Tween-20]. The membranes were blocked with 5% BSA for 2 h at 37°C, and then incubated with diluted primary antibodies (0.25 µg/ml goat anti-mouse LDLR, rabbit anti-human LDLR or mouse anti-GAPDH antibodies) (Abcam, Cambridge) at 4°C overnight, followed by HRP-labelled goat anti-rabbit IgG (1:1,000) for 1 h at 37°C. The protein bands were detected using an ECL kit.

Competitive phage ELISA screening and indirect ELISA for selected clones

ELISA plates (96-well) were coated with 5 µg/ml PCSK9 protein (100 µl/well) at 4°C overnight. After incubation with blocking buffer (2% BSA/PBS) for 1 h, 100 µl phage antibody (containing supernatants or purified anti-PCSK9 antibodies) was added to each well, and the plates were incubated at 37°C for 1 h. The plates were then washed three times with PBST and incubated with a 1:5,000 dilution of HRP-labelled mouse anti-M13 monoclonal antibody (cat. no. ab50370; Abcam) or HRP-IgG at 37°C for 1 h. Then, 100 µl/well tetramethylbenzidine peroxidase (TMB) substrate solution was added to initiate the colour reaction, which was then terminated using 2 M sulfuric acid (100 µl/well). The absorbance at 450 nm was measured using a Elx800 micro-plate reader (BioTeK Instruments, Inc.), with 2% BSA as the negative control and PBS as the blank control. All experiments were repeated three times.

PCSK9 inhibition in Hep-G2 cells using purified anti-PCSK9 antibody

Hep-G2 cells (5×105 cells/ml) were cultured in a 6-well plate overnight at 37°C, to generate single-cell mono-layers. The supernatant was discarded and replaced with DMEM containing 10% lipoprotein-deficient serum. After, 24 h, the cells were treated with 1, 5 or 10 µg/ml purified antibody or a combination of 0.5 or 1 µg/ml mevinolin, and incubated for 24 h sequentially. After collection, the LDLR levels of theHep-G2 cells were determined by western blotting.

Inhibitory effect of the anti-PCSK9 antibody on PCSK9-LDLR binding

A microtiter plate was coated with 5 µg/ml rLDLR (recombinant human LDLR; Shanghai Xinfan Biological Technology Co., Ltd.). The biotinylated PCSK9 protein was then mixed with different concentrations of anti-PCSK9 anti-body, which were then added to the plate. Bound biotinylated PCSK9 was detected using avidin-HRP and TMB. The reactions were terminated using 50 µl H2SO4 (2 M) and absorbance was measured at 450 nm.

Measurement for serum cholesterol liver LDLR in mice

All animal experiments were approved by the Ethics Committee of Jilin Medical University. Male C57BL/6J mice (weight, 18-22 g) were purchased from the experimental animal centre of Jilin University (Changchun, China), and housed with food and water ad libitum under a constant temperature (23±1°C) with a 12-h light/dark cycle. The mice were randomly divided into three groups and intravenously (i.v.) administered purified anti-PCSK9 antibody (n=84), anti-keyhole limpet hemocyanin antibody (negative control, n=21) or PBS (blank control, n=21).

A total of 21 mice were randomly selected from each group and received a single i.v. injection of anti-PCSK9 antibody (10 mg/kg). The serum was collected at different time points post-injection (24, 48 and 72 h; 7 mice/group). The remaining 63 mice in the anti-PCSK9 antibody group were injected with different concentrations of anti-PCSK9 antibody, and the serum was collected at 3, 6 and 9 days after injection. The total cholesterol (TC) levels were determined using a TC detection kit (cat. no. A111-1-1; Nanjing Jiancheng Bioengineering Institute).

The mice were sacrificed by cervical dislocation, and protein was extracted from the livers (24, 48 and 72 h post-injection groups) using a Membrane and Cytosol Protein Extraction Kit (cat. no. G4422-2; GBCBIO Technologies Inc.). Western blotting was conducted using LDLR and GAPDH antibodies (both Abcam) to determine the levels of LDLR following PCSK9 treatment.

Statistical analysis

Statistical analyses were conducted using Originpro version 9.1 (OriginLab). All of the results conformed to a normal distribution and are expressed as the mean ± SEM. The t-test was used to compare two groups. P<0.05 was considered to indicate a statistically significant difference.

Results

Construction of fully human scFv phage display library

Stepwise PCR was used to synthesize fully human scFvs for library construction (Fig. 1). The concentration of total RNA was 2,164 ng/µl with a A260/A280 ratio of 1.935 (data not shown). Following amplification of the VH, Vκ and Vλ genes (Fig. 2A-C), amplicons ranging around 400 bp were obtained and used as the templates for overlap PCR. Ultimately, the amplicons ranging around 800 bp of the scFv (Fig. 2D) were recovered and cloned into pDF vectors. After recombination of the library in the BS1365 strain, the library size was estimated to be ~1×1011 cells/ml (Fig. 3A). Tests for library accuracy and diversity, based on 60 colony PCR and scFv gene sequencing results, are shown in Fig. 3B, which confirm >75% sequence variation. Sequence blast results revealed that VL and VH were consistent with the human antibody sequences (data not shown).

Generation of human PCSK9 recombinant protein

The human PCSK9 gene is transcribed into a 3,637 bp linear mRNA, and is primarily expressed in the liver. Thus, PCSK9 was amplified from hepatic Hep-G2 cells using reverse transcription PCR (15). The recombinant vector, pcDNA3.1-hIgG1FC-PCSK9, was successfully constructed and expressed in Chinese hamster ovary (CHO)-K1 cells. After purification, human PCSK9 in the cell supernatant was analysed by SDS-PAGE (Fig. 4A). Ultimately, two distinct bands were observed at 70 and 17 kDa, which correspond to the catalytic C-terminal domains and the prodomain, respectively (16).

Selection of human scFv antibodies against PCSK9 using a phage display library

The human scFv phage library was based on Cre-LoxP511 system in vivo recombination for VH and VL generation, and was constructed using a pDF vector (Fig. 4B). The library comprised 1×1011 clones, which were screened against recombinant PCSK9 protein. A total of 4 rounds of selection were performed on immobilized PCSK9, detected by phage ELISA. As shown in Fig. 4C, a significant enrichment of 80% was reached after the fourth selection. Ultimately, 24 colonies recovered from the fourth round showed high binding affinity for PCSK9 (Fig. 4D). All clones exhibited the characteristics of the human antibody gene (Table II). A total of five different scFv sequences were identified; the light chains of 2C7 and 3D2 were κ type, while 3B9, 7E4 and 12D6 were λ type. For comparison, an NCBI online constraint-based multiple alignment tool showed that there were differences in CDR region among the five clones, in which both the VL κ types of CDR3 and the VL-CDRs of λ type, as well as the CDRs of variable heavy chain regions, were significantly different between the five clones.

Table II

Five sequences of scFvs obtained after four rounds of screening, among which the complementarity-determining regions (CDRs) were significantly different.

Table II

Five sequences of scFvs obtained after four rounds of screening, among which the complementarity-determining regions (CDRs) were significantly different.

Five sequences of scFvs
CDR1 CDR2
12D6 DIVMTQTPS-VSVAPGKTARITCGGKDIGSKSVHWYQQKPGQAPVLVIYYDSDRPSGIPE
3B9 DIVLTQSPS-VSVAPGKTARITCGGNDIGSKSVHWYQQKPGQAPVLVIYYDSDRPSGIPE
7F4 DIVLTQAPS-VSVAPGKTARITCEPQEIGSKSVHWYQQKPGQAPVLVIYYDSDRPSGIPE
3D2 EIVLTQSPVTLSVSPGERATLSCRASQSVIGNLAWYQQKPGQAPRLLIYGASARAAGIPD
2C7 EIVLTQSPATLSVSPGERATLSCRASQSVLVNLAWYQQKPGQAPRLLIYGASRVESGIPD
CDR3
12D6 RFSGSNSGNTATLTISRVEAGDEADYYCAAWDDSLNGPVFGGGTKVEIKRSGGSTITSNN
3B9 RFSGSNSGNTATLTISRVEAGDEADYYCHDWDESQRRPVFGGGTKGEIKRSGGSTITSNN
7F4 RFSGSNSGNTATLTISRVEAGDEADYYCAEQDRRLHSQVFGGGTKVEIKRSGGSTITSNN
3D2 RFSGSGSGTDFTLTISSLKSEDFAVYYCQQYGSS-PYTFGQGTKLEIKRSGGSTITSNN
2C7 RFSGSGSGTEFTLTISSLKSEDFAVYYCQQYTLA-SGTFGQGTKLEIKRSGGSTITSNN
CDR1
12D6 VYLRKLSSSGTQVQLVQSGAEVKKPGASVKVSCKVSGYSFDTLGIHWVRRAPGKGLEWMG
3B9 VYLRKLSSSGTQVQLVQSGAEVKKPGASVKVSCKSSGYGLEALGIQWVRWAPGTGLEWMG
7F4 VYLRKLSSSGTQVQLVQSGAEVKKPGASVKVSCQASGVTDHGYDIHWVRQATGQGLEWMG
3D2 VYLRKLSSSGTQVQLVQSGAEVKKPGSSVKVSCQASGGTFSGYDIHWVRQATGQGLEWMG
2C7 VYLRKLSSSGTQVQLVQSGAEVKKPGASVKVSSKVSASRCVTLSIHWVRLAPGKGLEWMG
CDR2 CDR3
12D6WISPYNGNTDYAQNLQDRVSMTTDTSTSTAYMELRSLRSDDTAVYYCARVYSYSMDYWGQ
3B9WISAYNGNTDYAQHFEDRVSMTTDTSTSTAYLELRSLRSDDTAVYYCVHGYSYRVGYWGQ
7F4WMNPLVGADRYAQKFQGRVTMTRDTSTATAYMELTSLTSDDTAVYYCAR-YAHTSLEWGQ
3D2WMNPHSGNAGYAQKFQGRVTMTRDTSTATAYMELTSLTSDDTAVYYCAR-YAQLSLEWGQ
2C7WISRSTDSRVYAQNLHVTVFMTTDTSTSTAYMELRSLRSDDTAVYYCARVYRDRVDYWGQ
241
12D6GTLVTVSS
3B9GTLVTVSS
7F4GTLVTVSS
3D2GTLVTVSS
2C7GTLVTVSS

[i] The letters in bold and underline were the CDRs of the five scFvs.

Construction and characterization of fully human PCSK9 fragments

To generate fully human scFv antibodies, the VH and VL regions of the five selected scFv clones were separately subcloned into pcDNA3.1-CL and pcDNA3.1-CH expression vectors (Fig. 5A). Antibodies were expressed in 293 cells and purified using protein A affinity chromatography. The yields of the five antibodies ranged between 50 and 200 mg/l per culture. SDS-PAGE analysis of the 3D2 antibody revealed that it consisted of a 25 kDa VL and a 50 kDa VH region under reducing conditions, and formed a 170 kDa homodimer under non-reducing conditions, which are typical patterns of human IgG (Fig. 5B). Western blotting and indirect ELISA were used to assess the specificity of the five selected antibodies against PCSK9. The data indicated that all of the antibodies specifically bound the PCSK9 protein in a dose-dependent manner (Fig. 5C). Moreover, antibody affinity was determined using BLI technology, revealing KD values ranging from 8.77ⅹ10−9 to 3.28ⅹ10−10 M (Fig. 5D and T able III). These data suggest that the five fully human scFv fragments were specific for PCSK9.

Human 3D2 antibody inhibits PCSK9 function in vitro

Functional properties of the five PCSK9 antibodies were characterized using in vitro assays. Of these five antibodies, only 3D2 was able to bind PCSK9, thereby blocking the PCSK9-LDLR interaction with an IC50 of 2.25±1.23 nM (n=3) (Fig. 6A). After 24-48 h of incubation with 1, 3 or 10 µg/ml 3D2, the LDLR levels in the Hep-G2 cells increased 1.3-2.5 fold compared with those in the untreated cells (Fig. 6B). In addition, previous reports have indicated that statins induce the expression of both LDLR and PCSK9 (17); therefore, in the present study, the effects of combining 3D2 antibody and statin administration on Hep-G2 cell LDLR levels were investigated. Hep-G2 cells were treated with the statin mevinolin and/or 3D2, and the levels of LDLR protein in cell lysates were determined using western blotting (Fig. 6C). The results indicate that combined administration was more effective than that of mevinolin or 3D2 alone; 1 µg/ml mevinolin combined with 10 µg/ml 3D2 increased the levels of LDLR protein up to 9.0 fold compared with untreated cells. These data suggest that the 3D2 antibody is a potential auxiliary treatment for decreasing cellular lipid levels.

Human 3D2 antibody reduces serum cholesterol by increasing hepatic LDLR in mice

Experiments were performed to verify whether human 3D2 antibody could also increase hepatic LDLR levels in vivo, and subsequently reduce serum TC. C57BL/6J mice (n=7 per group) were treated with a single i.v. injection of 3D2 (10 mg/kg). The results indicate that 3D2 significantly reduced serum TC levels at 24, 48 and 72 h post-injection by 33.5±0.54% (P<0.05), 53.9±0.85% (P<0.05), and 61.5±1.2% (P<0.01), respectively (Fig. 6D). However, the phenomenon was reversible, and the effect was duration-dependent (Fig. 6F). By day 9, the TC levels in all treatment groups were not significantly different from those in the control group. Furthermore, hepatic LDLR protein levels were increased 2.0-fold (Fig. 6E), which indicates that the increase in LDLR level was positively associated with a decrease in the lipid profile, and that the 3D2 antibody reduces lipid levels in vivo.

Discussion

The fully human scFv phage display library holds great value for the development of antibody-based treatments. The construction of a human scFv library is theoretically simple, but a high-quality library is difficult to obtain. There are three main indexes used to evaluate the quality of a phage display library, including gene diversity, library size and screening ability. In the present study, these three key indexes were used to optimize the construction of the human scFv library. A total of 400 blood, 30 bone marrow and 10 cord blood samples were collected from healthy donors. Donors were recruited via different sources to avoid the bias of individual antibody genes, and to ensure the diversity of the scFv gene. The scFv gene was amplified in strict accordance with the proportion of antibody genes of different human germlines, and imitated the human antibody gene repertoire to the maximal extent. The results of randomly sequencing 60 scFvs revealed that the diversity of the constructed phage display library was consistent with the antibody gene distribution of the human germline. Table III

Table III

Binding affinities of antibodies with recombinant PCSK9 protein.a

Table III

Binding affinities of antibodies with recombinant PCSK9 protein.a

AntibodyKon (106 M−1S−1)Koff (10−4S−1)KD (10−9 M)
2C73.16±1.116.74±2.2621.4±7.21
3B92.49±1.218.15±3.2232.8±13.1
3D22.59±1.585.08±4.0219.6±15.6
7F45.27±1.604.62±2.188.77±4.92
12D62.94±2.418.36±3.7028.5±12.8

a Kinetic constants were determined by biolayer interferometry using immobilized PCSK9 protein. The values are the mean ± SEM of at least three independent experiments.

The simplest way to evaluate the quality of an antibody library is by size. Theoretically, the larger the capacity of the library, the higher the affinity of the screened antibodies. The key factors that affect antibody library size are primarily the efficiencies of gene splicing and electrotransformation. Splicing of antibody light and heavy chain genes is generally conducted by overlap PCR. After a long period of investigation, the optimal protocol and conditions for overlap PCR amplification were determined. A primary human scFv library (5×107 in size) was constructed from peripheral and cord blood samples, and a primary human scFv library (5×107 in size) was constructed using bone marrow samples. By optimizing the recombinant conditions, the primary library was used to infect the BS1365 E. coli strain, which expresses the Cre enzyme. Using Cre-LoxP enzyme-mediated heavy and light chain replacement recombination, a recombinant human scFv phage display library with a library size of 1×1011 was obtained. The most important parameter for antibody library quality is the ability to screen antibodies. The scFv library constructed in the present study can be used to screen antibodies with an affinity of 3.28×10−10 M through the screening and verification of those specific to PCSK9, which confirms the optimal quality of the scFv phage display library.

Various studies have indicated that anti-PCSK9 antibodies are powerful inhibitors of PCSK9 (18,19), thus the generation of a novel and less immunogenic human antibody with greater in vivo efficacy is of great clinical significance. In the current study, recombinant human PCSK9 protein was used to screen human scFv antibodies. After 4 rounds of selection, an scFv antibody with high affinity for PCSK9, 3D2, was obtained. Further experimentation revealed that 3D2 increased hepatic LDLR levels in vivo and vitro, which in turn reduced serum TC.

Due to low circulating LDL-C levels in their blood, wild-type C57BL/6J mice are generally not suitable for the study of cholesterol-lowering drugs (20). However, some in vivo studies of PCSK9-targeted compounds revealed enhanced levels of PCSK9 in the serum of mice. Rashid et al (21) found that TC levels were elevated in the serum, and the LDLR levels in hepatocytes were reduced in mice receiving an i.v. injection of PCSK9. To investigate its neutralizing activity, Schroeder et al(22) treated mice with an i.v. injection of an adeno-associated virus (AAV) vector carrying the PCSK9 gene, followed by injection of the candidate antibody. Furthermore, Barale et al (23) obtained a human anti-PCSK9 antibody (mAb1) using the hybridoma technique, which had cross-reactivity with murine PCSK9 and significantly decreased serum cholesterol in WT C57BL/6J mice. In the present study, the 3D2 human anti-PCSK9 antibody, which also cross-reacted with murine PCSK9 (data not shown), effectively reduced serum cholesterol by increasing hepatic LDLR in mice.

In summary, the present study outlines the construction of a human scFv phage antibody library, with large library size and diversity, using Cre-LoxP in vitro recombination. By screening with recombinant human PCSK9 antigen, the diversity and screening ability of the constructed antibody library was further confirmed, and it can be used for screening other target antibodies. The fully human anti-PCSK9 antibody, 3D2, was found to neutralize PCSK9 activity in vitro and in vivo, demonstrating its potential for future clinical development.

Funding

This study was supported by the following financial programs: 'Research and Development of Industrial Technology' Program of Jilin Province, PR China (grant nos. 20170204005YY and 20180623045TC), Program of Jilin Science and Technology bureau, P.R. China (grant nos. 2019001179 and 20200104093), Health Commission Program of Jilin Province (no. 2020Q029) and National Training Program of Innovation and Entrepreneurship for Undergraduates (nos. 201913706005, 201913706039 and 202013706008).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate

The present study was approved by the ethics committee of Jilin Medical University (approval no. 2016-LW012) prior to the start of the study. Informed consents for the collection of peripheral blood, bone marrow and cord were obtained from all subjects.

Authors' contributions

YD and FM constructed the scFv phage display library. ZW and SX prepared the antigen. TY was responsible for cell culture. AC and JW screened the antibody. MY analyzed the data regarding the characterization of antibody. LT and CH purified the protein. HW determined the hypolipidemic effects of antibody. JC analyzed all the data and wrote the manuscript. All authors read and approved the final manuscript.

Patient consent for publication

Written informed consent for publication was obtained from all participants.

Competing interests

The authors declare that they have no competing interests.

Acknowledgments

Not applicable.

References

1 

Zhang Y, Wang W, Lv M, Lin Z, Geng J, Li Y, Shen B, Ma Y, Li Y, Qiao C and Feng J: A single-chain antibody using LoxP511 as the linker enables large-content phage library construction via Cre/LoxP recombination. J Biomol Screen. 19:839–846. 2014. View Article : Google Scholar : PubMed/NCBI

2 

Zhang GM, Chen YP, Guan YZ, Wang Y and An YQ: Modification and identification of a vector for making a large phage antibody library. Chin Med J (Engl). 120:2011–2016. 2007. View Article : Google Scholar

3 

Baragetti A, Grejtakova D, Casula M, Olmastroni E, Jotti GS, Norata GD, Catapano AL and Bellosta S: Proprotein convertase subtilisin-Kexin type-9 (PCSK9) and triglyceride-rich lipo-protein metabolism: Facts and gaps. Pharmacol Res. 130:1–11. 2018. View Article : Google Scholar : PubMed/NCBI

4 

Abifadel M, Varret M, Rabès JP, Allard D, Ouguerram K, Devillers M, Cruaud C, Benjannet S, Wickham L, Erlich D, et al: Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat Genet. 34:154–156. 2003. View Article : Google Scholar : PubMed/NCBI

5 

Cohen J, Pertsemlidis A, Kotowski IK, Graham R, Garcia CK and Hobbs HH: Low LDL cholesterol in individuals of African descent resulting from frequent nonsense mutations in PCSK9. Nat Genet. 37:161–165. 2005. View Article : Google Scholar : PubMed/NCBI

6 

Huang CC, Fornage M, Lloyd-Jones DM, Wei GS, Boerwinkle E and Liu K: Longitudinal association of PCSK9 sequence variations with low-density lipoprotein cholesterol levels: The Coronary Artery Risk Development in Young Adults Study. Circ Cardiovasc Genet. 2:354–361. 2009. View Article : Google Scholar : PubMed/NCBI

7 

Frank-Kamenetsky M, Grefhorst A, Anderson NN, Racie TS, Bramlage B, Akinc A, Butler D, Charisse K, Dorkin R, Fan Y, et al: Therapeutic RNAi targeting PCSK9 acutely lowers plasma cholesterol in rodents and LDL cholesterol in nonhuman primates. Proc Natl Acad Sci USA. 105:11915–11920. 2008. View Article : Google Scholar : PubMed/NCBI

8 

Ray KK, Stoekenbroek RM, Kallend D, Leiter LA, Landmesser U, Wright RS, Wijngaard P and Kastelein JJP: Effect of an siRNA therapeutic targeting PCSK9 on atherogenic lipoproteins: Prespecified secondary end points in ORION 1. Circulation. 138:1304–1316. 2018. View Article : Google Scholar : PubMed/NCBI

9 

Duff CJ, Scott MJ, Kirby IT, Hutchinson SE, Martin SL and Hooper NM: Antibody-mediated disruption of the interaction between PCSK9 and the low-density lipoprotein receptor. Biochem J. 419:577–584. 2009. View Article : Google Scholar : PubMed/NCBI

10 

Dubuc G, Chamberland A, Wassef H, Davignon J, Seidah NG, Bernier L and Prat A: Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis-regulated convertase-1 implicated in familial hypercholesterolemia. Arterioscler Thromb Vasc Biol. 24:1454–1459. 2004. View Article : Google Scholar : PubMed/NCBI

11 

Careskey HE, Davis RA, Alborn WE, Troutt JS, Cao G and Konrad RJ: Atorvastatin increases human serum levels of proprotein convertase subtilisin/kexin type 9. J Lipid Res. 49:394–398. 2008. View Article : Google Scholar

12 

Tonikian R, Zhang Y, Boone C and Sidhu SS: Identifying specificity profiles for peptide recognition modules from phage-displayed peptide libraries. Nat Protoc. 2:1368–1386. 2007. View Article : Google Scholar : PubMed/NCBI

13 

Lakzaei M, Rasaee MJ, Fazaeli AA and Aminian M: A comparison of three strategies for biopanning of phage-scFv library against diphtheria toxin. J Cell Physiol. 234:9486–9494. 2019. View Article : Google Scholar

14 

Tohidkia MR, Sepehri M, Khajeh S, Barar J and Omidi Y: Improved soluble ScFv ELISA screening approach for antibody discovery using phage display technology. SLAS Discov. 22:1026–1034. 2017.PubMed/NCBI

15 

Grozdanov PN, Petkov PM, Karagyozov LK and Dabeva MD: Expression and localization of PCSK9 in rat hepatic cells. Biochem Cell Biol. 84:80–92. 2006. View Article : Google Scholar : PubMed/NCBI

16 

Lipari MT, Li W, Moran P, Kong-Beltran M, Sai T, Lai J, Lin SJ, Kolumam G, Zavala-Solorio J, Izrael-Tomasevic A, et al: Furin-cleaved proprotein convertase subtilisin/kexin type 9 (PCSK9) is active and modulates low density lipoprotein receptor and serum cholesterol levels. J Biol Chem. 287:43482–43491. 2012. View Article : Google Scholar : PubMed/NCBI

17 

Rodriguez F and Harrington RA: Cholesterol, cardiovascular risk, statins, PCSK9 inhibitors, and the future of LDL-C lowering. JAMA. 316:1967–1968. 2016. View Article : Google Scholar : PubMed/NCBI

18 

Harada LM, Carrilho AJ, Oliveira HC, Nakandakare ER and Quintão EC: Regulation of hepatic cholesterol metabolism in CETP/LDLr mice by cholesterol feeding and by drugs (chole-styramine and lovastatin) that lower plasma cholesterol. Clin Exp Pharmacol Physiol. 33:1209–1215. 2006. View Article : Google Scholar : PubMed/NCBI

19 

Cao Y, Yang H, Zhou X, Mao H, Gao T, Hu Z, He L, Pan F and Guo Z: Selection and characterization of human PCSK9 antibody from phage displayed antibody library. Biochem Bioph Res Commun. 463:712–718. 2015. View Article : Google Scholar

20 

Lie J, de Crom R, van Gent T, van Haperen R, Scheek L, Sadeghi-Niaraki F and van Tol A: Elevation of plasma phospholipid transfer protein increases the risk of atherosclerosis despite lower apolipoprotein B-containing lipoproteins. J Lipid Res. 45:805–811. 2004. View Article : Google Scholar : PubMed/NCBI

21 

Rashid S, Curtis DE, Garuti R, Anderson NN, Bashmakov Y, Ho YK, Hammer RE, Moon YA and Horton JD: Decreased plasma cholesterol and hypersensitivity to statins in mice lacking Pcsk9. Proc Natl Acad Sci USA. 102:5374–5379. 2005. View Article : Google Scholar : PubMed/NCBI

22 

Schroeder KM, Beyer TP, Hansen RJ, Han B, Pickard RT, Wroblewski VJ, Kowala MC and Eacho PI: Proteolytic cleavage of antigen extends the durability of an anti-PCSK9 monoclonal antibody. J Lipid Res. 56:2124–2132. 2015. View Article : Google Scholar : PubMed/NCBI

23 

Barale C, Bonomo K, Frascaroli C, Morotti A, Guerrasio A, Cavalot F and Russo I: Platelet function and activation markers in primary hypercholesterolemia treated with anti-PCSK9 mono-clonal antibody: A 12-month follow-up. Nutr Metab Cardiovasc Dis. 30:282–291. 2020. View Article : Google Scholar

Related Articles

Journal Cover

February-2021
Volume 47 Issue 2

Print ISSN: 1107-3756
Online ISSN:1791-244X

Sign up for eToc alerts

Recommend to Library

Copy and paste a formatted citation
x
Spandidos Publications style
Dong Y, Meng F, Wang Z, Yu T, Chen A, Xu S, Wang J, Yin M, Tang L, Hu C, Hu C, et al: Construction and application of a human scFv phage display library based on Cre‑LoxP recombination for anti‑PCSK9 antibody selection. Int J Mol Med 47: 708-718, 2021
APA
Dong, Y., Meng, F., Wang, Z., Yu, T., Chen, A., Xu, S. ... Cai, J. (2021). Construction and application of a human scFv phage display library based on Cre‑LoxP recombination for anti‑PCSK9 antibody selection. International Journal of Molecular Medicine, 47, 708-718. https://doi.org/10.3892/ijmm.2020.4822
MLA
Dong, Y., Meng, F., Wang, Z., Yu, T., Chen, A., Xu, S., Wang, J., Yin, M., Tang, L., Hu, C., Wang, H., Cai, J."Construction and application of a human scFv phage display library based on Cre‑LoxP recombination for anti‑PCSK9 antibody selection". International Journal of Molecular Medicine 47.2 (2021): 708-718.
Chicago
Dong, Y., Meng, F., Wang, Z., Yu, T., Chen, A., Xu, S., Wang, J., Yin, M., Tang, L., Hu, C., Wang, H., Cai, J."Construction and application of a human scFv phage display library based on Cre‑LoxP recombination for anti‑PCSK9 antibody selection". International Journal of Molecular Medicine 47, no. 2 (2021): 708-718. https://doi.org/10.3892/ijmm.2020.4822