Oligoclonal selection of nanobodies targeting vascular endothelial growth factor.

Abstract While monoclonal antibodies are efficient therapeutics for cancer treatment, nanobodies or variable heavy domain – due to their small size, high stability, and solubility – have many advantages in comparison. Oligoclonal nanobodies are a mixture of nanobodies against different epitopes of an antigen. Specific nanobodies against vascular endothelial growth factor (VEGF, which has an important role in cancer angiogenesis) were selected from an immune camel library using biopanning. Specific binding of the nanobodies to VEGF antigen was assessed by periplasmic extract enzyme-linked immunosorbent assay (ELISA). Bioinformatics analysis and molecular docking were performed on selected nanobodies against VEGF. The in vitro inhibitory effects of each single nanobody, as well as a pool of selected nanobodies (oligoclonal nanobodies), on proliferation and tube formation by/in human umbilical vein endothelial cells (HUVEC) cells was evaluated using MTT and Tube formation assays, respectively. Four nanobodies showed the highest signal intensity in the periplasmic extract ELISA. Sequencing revealed that four unique nanobodies with different CDR3 rejoin were selected. Oligoclonal nanobodies inhibited proliferation and tube formation of the HUVEC cells more potently than did each individual nanobody. Taken together, this data from this study suggests that in vitro use of nanobodies (in an oligoclonal mode) that target distinct epitopes on VEGF could be promising as a novel therapy to treat VEGF-dependent pathologies. However, this needs to be further tested in in vivo studies.


Introduction
There are six important biological mechanisms in the development of cancer. These include proliferation signaling, disabling of growth suppressors, blockage of cell death, immortal replication, angiogenesis, and metastasis (Hanahan and Weinberg 2011). Angiogenesis is an important factor in the cancer process, with many proteins playing key roles in the angiogenic process (Otrock et al. 2007). Some proteins act as regulators, such as fibroblast growth factor (FGF), angiogenin, transforming growth factor (TGF), angiopoietin, interleukin (IL)-8, and tumor necrosis factor (TNF)-a. Other proteins have functional roles, like platelet-derived growth factor (PDGF), IL-1b, stromal cell-derived factor one (SDF-1/CXCL12), cyclooxygenase (COX)-2, vascular endothelial growth factor (VEGF), and VEGF receptor (VEGF-R) (Ceradini et al. 2004).
The VEGF family consists of five members, VEGF-A, -B, -C, and -D, and placental growth factor (PlGF). Of these, VEGF-A seems to have a more critical role than others in angiogenesis (Ferrara et al. 2003). VEGF-A has two receptors, including VEGF-R1 (Flt-1) and VEGF-R2 (Flk-1/KDR), to which all its activity is dependent upon. VEGF-A is a secretion glycoprotein belonging to the platelet-derived growth factor (PDGF) family (Shibuya 2001). Through 2015, more than 350 intact monoclonal antibodies (mAb) or fragments were evaluated in a clinical trial against a variety of pathologies; of these, more than 70 were approved for use by the United States Food and Drug Administration (FDA) (Ecker and Jones 2015). Among these mAb drugs, cetuximab, panitumumab, necitumumab, and nimotuzumab are used as anti-epidermal growth factor receptor (EGFR) agents; trastuzumab, pertuzumab, and ado-trastuzumab emtansine are employed against human epidermal growth factor receptor-2 (HER-2); bevacizumab and ranibizumab are used against VEGF-A, and ramucirumab is useful against VEGF-R2. These antibodies have several mechanisms of actions (MoA) for affecting tumor cells. The most common MoA is to block the antigens/receptors directly or indirectly and in turn suppress or induce cell signaling (Carter and Lazar 2018). One VEGF inhibitor recombinant antibody, i.e. bevacizumab, is able to deactivate all human VEGF isoforms (Hurwitz et al. 2004;Ferrara et al. 2005).
Monoclonal antibodies (mAb) are commonly using extensively; however, considerations about production costs/expensive processes have diminished the willingness of manufacturers to produce them. Further, the routinely large sizes of mAb impact on/restrict abilities to penetrate deeply into tumors/tumor cells (Kolkman and Law 2010). Thus, efforts to find new antibodies with both a smaller size and the effectiveness of mAb have led to increases in the production of single domain antibody fragments (sdAb; 12-14 kDa), single chain variable Fragment (scFv; 26-30 kDa), and nano-antibodies (nanobodies or Nb; 11-15 kDa) (Kolkman and Law 2010;Scott et al. 2012;Spadiut et al. 2014).
As a result of discovering the structures of immunoglobulins (Ig) from Camelidae host like camels and llamas, a promising novel tool was provided for the development of antibodies with a wide range of applications. Camel Ig has two separate heavy chain antibodies (HCAb) that each consist of a variable heavy domain (VHH) and two constant domain (CH2 and CH3).
However, unlike human Ig, camel Ig has neither a light (L) chain nor a CH1 domain (Hamers-Casterman et al. 1993;Ebrahimizadeh et al. 2015). A similar type of antibody lacking an L chain (V-NAR) was found in sharks and ratfishes, but these had structural differences from camel HCAb (Flajnik et al. 2011). Though VHH and various V-NAR have similar molecular dimensions, only VHH is termed as nanobody due to its size (2.5-4.0 nm) Nanobodies have ideal biophysical and pharmacological properties due to both their small size and single domain entity. Furthermore, their high stability, as well as high specificity and affinity (in nM range) for cognate targets make them good candidates in "drug" development (Cortez-Retamozo et al. 2004;Hassanzadeh-Ghassabeh et al. 2013). In this vein, nanobodies can potentially be important diagnostic tools. Because a rapid diagnosis of early stages of cancer and an ability to readily evaluate efficacy of other cancer therapeutics is important, nanobodies would fit this role in that they have high penetrating potentials into tumors/cancer cells, are able to detect very small targets, and they are cleared quickly from the body (Oliveira et al. 2010).
As "anti-tumor" drugs, nanobodies can be divided into three classes. The first is "naked nanobody", i.e. one without an F c domain. The second is a recombinant fusion form attached to an activator domain. The third is a nanobody used to direct a drug to a target tissue (Bannas et al. 2017). Recently, a nanobodybased product was developed to inhibit VEGF and angiopoietin-2 (K€ onning et al. 2017). Another anti-VEGF nanobody is VA12, a product that showed stability at high temperatures and low pHs (Ebrahimizadeh et al. 2015). At this point, a few nanobodybased therapeutics have entered into clinical trials. For example, ALX-0761, ALX-0061, and ozoralizumab are anti-inflammatory nanobodies that have been tested on >700 patients. In no case were there any reported side-effects (Rissiek et al. 2014). ALX-0171 nanobody was tested in clinical trials against respiratory syncytial virus (RSV) infections (Detalle et al. 2017); its affinity was in the sub-nanomolar range and it significantly inhibited viral replication in the early phase.
Some studies have argued that some antigens bind to receptors on cells through at least two binding sites; this hypothesis highlights the importance of using antibodies as poly-or oligoclonal forms (Nowakowski et al. 2002). Oligoclonal nanobodies are mixtures of nanobodies against different epitopes of a given antigen (Carter and Lazar 2018). Until 2016, only %12 oligoclonal nanobodies had been medically tested (Corti and Kearns 2016). In the primary studies, oligoclonal nanobodies were used to inhibit many targets, including HER-2 or tumor-associated glycoprotein (TAG)-72 (Jamnani et al. 2012(Jamnani et al. , 2014Sharifzadeh et al. 2013). The results of those studies showed there was more potency of the antibodies against their own targets when applied in an oligoclonal rather than a monoclonal mode. Recent studies used two antibodies as an oligoclonal drug against breast cancer and melanoma (Larkin et al. 2015;Swain et al. 2015). In the first, a combination of trastuzumab and pertuzumab (anti-HER2 antibodies) was employed; in the second study, a mixture of ipilimumab (anti CTLA4) and nivolumab (anti PD1) was used.
Considering the major role of VEGF in angiogenesis and the potent effect of nanobodies, the present study sought to identify a set of oligoclonal nanobodies targeting various epitopes on VEGF (selected from immune camel VHH library) and examine these agents for potential effects on in vitro human endothelial cell proliferation and tube formation (to reflect impact on VEGF availability/functional use to the cells).

Selection of specific nanobodies
A nanobody library was constructed in a previous study (Kazemi-Lomedasht et al. 2015). Biopanning (four rounds) was performed to enrich a nanobody library against VEGF based on methods used by Kazemi-Lomedasht et al. (2015). In brief, 10 mg VEGF/ml in NaHCO 3 buffer was coated onto wells of a 96-well plate overnight at 4 C. Control wells received only buffer. About 10 12 CFU (colony-forming unit) of phage library was added to each well and the plates were incubated for 1 hr at 37 C. Unbonded phages were removed by washing (10 times using PBS-T (phosphate-buffered saline [PBS, pH 7.4] containing 0.05% [v/v] Tween 20). Bonded phages were then eluted using 10 mM TEA and neutralized with 1 M Tris-HCl (pH 8.0). Eluted phages (output) were then amplified in logarithmic phase (OD 600nm ¼ 0.5-0.7) Escherichia coli TG1 cells. Output phages were rescued using VCSM13 helper phage. To purify released phage particles, the bacterial pellet was collected (5000 Â g, 15 min) and recombinant phage particles in the pellet purified by incubation in PEG/NaCl solution (20% PEG-2.5 M NaCl) for 1 hr at 4 C (Lee et al. 2007;Kazemi-Lomedasht et al. 2015). The phage particles were then isolated by centrifugation of the PEG/ NaCl solution (10000 Â g, 30 min), re-suspended in PBS, and stored (as input) at À20 C. The input phages were in turn used for the next rounds of biopanning.
Polyclonal phage enzyme-linked immunosorbent assay (ELISA) was then performed to evaluate the success of the biopanning. In brief, eluted phages from each round were added to wells coated with VEGF and detection was monitored using an anti-M13-HRP antibody (1:7000 dilution in PBS). Single colonies from the fourth round of biopanning were screened using a Periplasmic Extract ELISA (PE-ELISA). For this, %47 colonies were chosen and cultured at 37 C in LB medium until they attained log phase. Expression of the nanobodies was induced by adding 1 mM IPTG and incubating the plate at 37 C for a further 16 hr. Periplasmic-expressed nanobodies were then extracted by osmotic shock; detection of any nanobodies present was performed using an anti-HA antibody (1:2000 in PBS) (Kazemi-Lomedasht et al. 2015). After washing the wells with PBST (5 times), HRP-anti-mouse IgG (1:5000 in PBS; secondary antibody) was added. Positive wells (OD 3X control well OD) indicated which colonies' associated nanobodies were to be sequenced.

Sub-cloning of nanobodies into expression vectors
Plasmids were extracted from positive colonies and used as templates for amplification of nanobody genes. PCR was performed using A6E (5 0 -GATGTGCAGCTGCAGGAGTCTGG-RGGAGG-3 0 ) and P38 (5 0 -GGACTAGTGCGGCCGCTGGAGACGGT GACCTGGGT-3 0 ) primers (annealing: 55 C for 40 sec). PCR products and pHEN-6C (bacterial expression vector) were then digested with BstEII and PstI restriction enzymes (using manufacturer protocols) and ligated together using T4 DNA ligase at 16 C for 5 hr. The recombinant construct was then transferred into E. coli WK6 cells via heat-shock.

Expression and purification of nanobodies
Expression of nanobodies into the periplasmic space (due to the existence of PelB signal sequence at N-terminal of pHEN-6C plasmid) of the bacteria was induced by treatment with 1 mM IPTG for 16 hr at 28 C. Following osmotic shock, recombinant nanobodies were purified using nickel affinity chromatography, and the nanobody concentrations determined using a BCA Protein Kit according to manufacturer instructions. The purity of the isolated nanobodies was then evaluated by Western blot analysis. In brief, after resolution over a 15% SDA-PAGE gel, protein bands were transferred to a PVDF membrane that, in turn, was blocked with 4% skim milk (in PBS) solution for 2 hr at room temperature (RT). After washing, anti-His antibody (1:500 in PBS) solution was placed on the membrane and the sample incubated for 2 hr at RT. After additional washes with PBST (3 times), signals on the blot were developed by additional of 4-CN substrate (10 mg/ml methanol) for 20 min at RT. The blots were then photographed and band intensities were quantified using Image J (https://imagej.nih.gov).

Affinity calculation
The affinity of the isolated nanobodies for VEGF was calculated using the method of Beatty et al. (1987). Two concentrations of VEGF (10 and 1 mg/ml) were selected using a checkerboard assay; the VEGF solutions were coated onto dedicated wells of 96-well plates by overnight incubation at 4 C. After washing the wells with PBST, test nanobodies (0-10 nM) were then added to dedicated wells. After a 1-hr incubation at RT, the wells received HRP-anti-His antibody (1:500 dilution in PBS). The plates were then incubated for 1 hr at RT before unbound primary antibody was removed by gentle rinsing of the wells with PBST. The wells then received 0.6 mg TMB/ml substrate and color formation was allowed to proceed for 15 min. Wells were then assessed for absorbance at 450 nm in a Epoch plate reader (BioTek, Winooski, VY). The affinity (k aff ) of the ligand to its target was then calculated using a Beatty equation.

MTT assay
A HUVEC (human umbilical vein endothelial cell) primary line generated in a previous study (Kazemi-Lomedasht et al. 2014) was employed to evaluate inhibitory effects of the test nanobodies (in mono-and oligoclonal modes) on cell proliferation. In brief, in each well of a 96-well plate, 10 4 HUVEC cells were cultured in EBM-2 medium containing 2% FBS for 12 hr at 37 C. Each of the following: PBS (negative control), H39Nb (nanobody negative control), bevacizumab (positive control), each nanobody alone, or a mixture of all nanobodies (oligoclonal mode)all at 10 mg/ml, along with 50 ng VEGF/ml were added to dedicated wells. All materials were evaluated in triplicate wells. The plate was then incubated 72 hr at 37 C before 50 ml of 5 mg MTT/ml was added to each wells and the plate incubated a further 4 hr. The medium in all wells were then replaced with 100 ml DMSO and the plate was incubated for 30 min. Absorbance in each well was then measured at 570 nm in the Epoch plate reader. Inhibition of proliferation was evaluated as 1 À (100 Â [proliferation of treated cells/proliferation of control cells]). Data [OD 570 values] for treated cells were normalized against the highest OD 570 seen with the control cells, i.e., the latter taken as 100% proliferation.

Tube formation assay
The inhibitory effect of the test nanobodies on formation of the tubular structures was evaluated using a tube formation assay on Matrigel. In brief, Geltrex TM LDEV was thawed on ice and 50 ml of the material then transferred into wells of 96-well plates. A total of %3 Â 10 4 HUVEC cells in 100 ml EBM-2 medium was then added to each well. Each of the following: PBS (negative control), H39Nb (nanobody negative control), bevacizumab (positive control), each nanobody alone, or a mixture of all nanobodies (oligoclonal mode)all at 10 mg/ml, along with 50 ng VEGF/ml were added to dedicated wells. All materials were evaluated for effects in triplicate wells. The plates were then incubated for 5 hr at 37 C. After removing the medium, cells were stained using a Calcein AM Cell Viability Kit, according to manufacturer instructions. The presence of tubular structures was then detected using an INV100-FL fluorescence inverted biological microscope (BEL, Monza, Italy). The numbers of branched and tubular structures in all cells were evaluated/ counted in three different fields/slide. A total of 1000 cells/slide were counted.

Statistical analysis
All data are reported as means ± SD when proper. A Students ttest was used to assess the significance of any differences between any two given test groups. All analyses were performed using Prism Software v(0).5 (GraphPad, San Diego, CA). A pvalue <0.05 was accepted as statistically significant.

Enrichment and screening
To qualitatively evaluate the biopanning here, titers of eluted phages from each round of biopanning were determined ( Table  1). The highest enrichment was observed in the fourth round of biopanning. Polyclonal phage ELISA was done on eluted phages in each round to quantitatively evaluate the biopanning process (Table 2). Observed signal intensity was increased after each biopanning round. Forty-seven colonies (randomly selected from fourth round of biopanning) were screened for binding to VEGF in a PE-ELISA experiment. Four nanobodies, i.e. Nb25, Nb30, Nb31, and Nb34, showed the greatest binding. Sequencing showed that the four nanobodies with unique CDR3 sequence had been selected. Framework and CDR regions of the four nanobodies were identified (Table 3). The affinity of each nanobody for VEGF was calculated and found to be 1-6 nM using a Beatty analysis/ELISA-based method.

Primary analysis
Physico-chemical parameters of the four nanobodies were calculated. Each had a molecular length of %130 amino acids and molecular weights of %14 kDa (Table 4). The 3D structures of the nanobodies and the sequence of VEGF (UniProt #P15692) were predicted (Figure 1), and a QMEAN score estimated for each predicted structure. The scores of all predicted structures were between À0.38 to À1.02, confirming the reliability of each prediction (Table 4, Row 5).

Docking results
The predicted 3D structures of Nb25, Nb30, Nb31, and Nb34 were prepared for docking analysis, i.e. water molecules were removed, force-field energy was minimized, Hydrogens were add and final structures saved in PDB format. The percentage of residues in favored/outlier regions was estimated using a RAMPAGE web server (Table 4). Total free energy from the interaction between nanobodies (or bevacizumab) as ligand and human VEGF (as receptor) was calculated. Correlation parameters were set on shape only and the final search number was 25. Lower calculated energy (more negative) meant a stronger affinity. The results showed of the tested materials, Nb34 and bevacizumab had the highest binding affinity for VEGF (Table 5). Bevacizumab, in compare with Nb34, showed a weaker binding affinity for VEGF. The amino acids involved in the interaction between VEGF and the nanobodies/bevacizumab, and the number of constructed hydrogen bonds was identified as well. The data indicated that Nb34 and Nb30 both generated seven H-bonds (Table 6).

Purification results
The test nanobodies were purified via nickel affinity chromatography and purification was confirmed by Western blot analyses.
Each blot included molecular weight markers. The results indicated that the weight of the purified nanobodies was %14 kDa each ( Figure 2).

MTT assay
MTT assays were performed to evaluate inhibitory effects of each nanobody, alone and when pooled (oligoclonal). The data indicated that all tested individual nanobodies (Nb25, Nb30, Nb31, and Nb34) were able to inhibit proliferation of HUVEC cells stimulated by VEGF (Figure 3). The results showed that inhibition (as % vs. control HUVEC cell proliferation) was 69% for bevacizumab, 62.5% for both Nb25 and Nb30, 65.5% for both Nb31, and Nb34, and 82% the oligoclonal mode nanobodies. As individual substances, bevacizumab (positive control) and H39Nb (negative control nanobody) led to the greatest and least inbitory effects, respectively.

Tube formation assay
Tube formation was assessed to examine other forms of inhibitory effects of the test nanobodies (individual and oligoclonal mode) in cells. As the results show, each nanobody significantly inhibited tube formation of HUVEC cells in comparison to what was noted in positive control wells (with only VEGF and no nanobody) (Figure 4). Results showed that inhibition (vs. control HUVEC cell levels) was evident in treated cells, with only 19% formed branched structures with bevacizumab, 31% with Nb25, 28% with Nb30, 21% with Nb31, 22% with Nb34; and 9% with the oligoclonal mode nanobodies. H39Nb (negative control nanobody) did not cause significant effects on tube formation. No tube-like structures were seen in cells that only received PBS (in place of VEGF).

Discussion
Treatment of cancer is a major focus of many studies and there is intense competition to achieve this goal (Chabner and Roberts 2005). As angiogenesis is a critical step in cancer progression, much research has focused on ways to inhibit this process. Among all molecules involved in angiogenesis, VEGF and its receptors seem to have important roles in the progression of the process (Ferrara et al. 2003). This is reflected in the fact that increased levels of serum VEGF have been noted in breast,  prostate, and colorectal cancer patients (Duque et al. 1999;George et al. 2000;Ali et al. 2011). For many years, there have been three primary methods to treat cancer, i.e., surgery followed by radiation therapy and/or chemotherapy. In recent years, immune-based treatments using specific antibody against tumor-based targets have been developed (Trapani and Darcy 2017). Bevacizumab is a recombinant mAb developed for treating colorectal cancer. However, there were important limitations (i.e. size resulted in decreased tumor penetrance) that prohibited extensive mAb use in treating this/ other cancers. Nanobodies could fill this gap in that they have a small size that makes them capable of penetrating internal layers of solid tumors. Further, their production in bacterial hosts has facilitated their application potential by expediting the formation/testing phase in product development. Lastly, nanobodies have unique characteristics, including high affinity and solubility (Muyldermans et al. 2009;Arezumand et al. 2017). Accordingly, researchers have used nanobodies for various applications (Kazemi-Lomedasht et al. 2015a, 2015b, 2017. Specifically, Kazemi-Lomedasht et al. developed nanobodies targeting VEGF that potentially inhibited human endothelial cell proliferation and tube formation. The polyclonal antibodies that were generated here recognized several epitopes on the target antigen and elicited stronger responses. Monoclonal antibodies compete for the same epitopes on the surface of an antigen. Importantly, mAb binding affinity can be affected by various parameters including interactions with other proteins, post-translational modification (PTM), local pH, temperature, and ion concentrations, etc. Thus, the amount of epitopes is a limiting factor for efficacy. Accordingly, higher doses of monoclonal antibodies would have to be administered to improve efficacy which, in turn, often leads to unwanted side effects (Haurum 2006). In the case of bevacizumab, side effects such as bleeding, phlebitis, and emboli formation resulted in a stoppage of patient treatment (Ranpura et al. 2011). In comparison, the combined use of antibodies in poly or oligoclonal forms that recognize different epitopes on a target antigen could act more efficiently than a single mAb (Haurum 2006) and so enhance efficacy. Several studies have demonstrated that the combined use of noncompetitive antibodies enhanced their function (Logtenberg 2007). In one study, it was seen that combining two epitope-distinct mAb (i.e. here, against a receptor) enhanced the overall therapeutic activity (Ben-Kasus et al. 2009). Spangler et al. (2010) noted that combinations of anti-EGFR mAb led to inhibition of receptor recycling and decreased levels of EGFR. These findings, and the fact that single domain entity nanobodies make them good options for use in antibody mixtures (Harmsen and De Haard 2007), have led to the performance of several studies using mixtures of nanobodies (oligoclonal mode) to target distinct epitopes on tumor cells. In those studies, improved anti-tumor activity with an oligoclonal mixture of nanobodies was achieved compared to that from each nanobody alone (Jamnani et al. 2012;Sharifzadeh et al. 2012). Along these lines of reasoning, our group has developed nanobodies targeting various epitopes on VEGF in order to inhibit angiogenesis in tumors. This would be in line with the studies by Nikkhoi et al. (2017Nikkhoi et al. ( , 2018 who used oligoclonal nanobodies to inhibit HER2 receptors on MC4L2 HER2 þ cells in ovarian and breast cancers. The current studies showed that indeed, use of oligoclonal modes of nanobodies generated against VEGF (or receptor) imparted the greatest effects on normal dividing epithelial cells (i.e. HUVEC) as compared to effects from any of the individual nanobodies.
The present study also reflected the many techniques for the production and amplification of nanobodies from extended libraries, i.e. phage display, ribosome display, mRNA display, and yeast display. Overall, phage display was our first choice because of a possibility of immunizing the animal host (Camelidae) by multiple antigens at the same time (   results of molecular docking and findings of experimental assays with some of the isolated nanobodies, it was found that docking predictions were well matched, and that results could be useful in subsequent studies on different proteins. Still, it was interesting to note that compared to the various nanobodies, Bevacizumab had the higher inhibitory effect on cell proliferation however, the molecular docking predicted a small amount of higher binding affinity for Nb34. This difference might result from using the Fab fragments (1BJ1) instead of complete Bevacizumab molecule in docking analysis. Bevacizumab has two binding sites that could potentially impart more inhibitory effects on the VEGF molecule (Fuh et al. 2006;Walker et al. 2016).

Conclusions
This study using nanobodies targeting distinct epitopes on VEGF, found for the first time that there was a greater efficacy when using an oligoclonal mode rather than a single nanobody. These outcomes were borne out from analyses of the inhibition of proliferation and ability to undergo tube formation by human endothelial cells in vitro. These results are helpful in that they point toward a potential for using oligoclonal nanobodies as novel agents to target distinct epitopes on VEGF in order to mitigate VEGF-dependent pathologies. However, this needs to be further tested in in vivo model studies.

Disclosure statement
The authors declare no conflicts of interested. The authors alone are responsible for the content of this manuscript.

Funding
This work was supported by the Pasteur Institute of Iran, Tehran, Iran.