`
`Liu et at BMC Cancer (2015)15:170
`DOI 10.1186/512885-015-1140-1
`
`EXRIBff o2f
`
`K,4A(M COURTPpURrrnG
`
`BMC
`Cancer
`
`RESEARCH ARTICLE
`
`Open Access
`
`A novel engineered VEGF blocker with an
`excellent pharmacokinetic profile and robust
`anti-tumor activity
`
`Lily Liu', Haijia Vu 1, Xin Huartg 2, Hongzhi Tan', Song Li 2, Van Luo 1, Li Zhang3, Sumei Jiang t, Huifeng Jia 3,
`Yao Xiong 3, Ruliang Zhang4, Vi Huang 3, Charles C Ch U5,6, 1 and Wenzhi Tian"
`
`Abstract
`
`Background: Relatively poor penetration and retention in tumor tissue has been documented for large molecule
`drugs including therapeutic antibodies and recombinant immunoglobulin constant region (Fc)-fusion proteins due
`to their large size, positive charge, and strong target binding affinity. Therefore, when designing a large molecular
`drug candidate, smaller size, neutral charge, and optimal affinity should be considered.
`Methods: We engineered a recombinant protein by molecular engineering the second domain of VEGFR1 and a
`few flanking residues fused with the Fc fragment of human IgG 1, which we named HB-002.1. This recombinant
`protein was extensively characterized both in vitro and in vivo for its target-binding and target-blocking activities,
`pharmacokinetic profile, angiogenesis inhibition activity, and anti-tumor therapeutic efficacy.
`Results: HB-002.1 has a molecular weight of —80 kDa, isoelectric point of —6.7, and an optimal target binding
`affinity of <1 nM. The pharmacokinetic profile was excellent with a half-life of 5 days, maximal concentration of
`20.27 pg/mI, and area under the curve of 81.46 pg days/ml. When tested in a transgenic zebrafish embryonic
`angiogenesis model, dramatic inhibition in angiogenesis was exhibited by a markedly reduced number of subintestinal
`vessels. When tested for anti-tumor efficacy, HB-002,1 was confirmed in two xenograft tumor models I and
`Colo-205) to have a robust tumor killing activity, showing a percentage of inhibition over go% at the dose of
`20 mg/kg. Most promisingly, HB-002.1 showed a superior therapeutic efficacy compared to bevacizumab in the
`A54g xenograft model (tumor inhibition: 84.7% for HB-002.1 versus 67.6% for bevacizumab, P <0.0001).
`Conclusions: 1-18-002.1 is a strong angiogenesis inhibitor that has the potential to be a novel promising drug for
`angiogenesis-related diseases such as tumor neoplasms and age-related macular degeneration.
`
`Keywords: VEGF inhibitor, VEGFR1, Recombinant Fc-fusion protein, Anti-tumor therapy, Angiogenesis
`
`Background
`Targeted tumor therapy is the focus of recent intense
`drug development by the pharmaceutical industry with
`the primary interests centered on antibody drugs [1].
`However antibody and/or recombinant protein drugs
`with molecular weights (MWs) of over 100 kDa usually
`have relatively poor tumor penetration and retention
`capacity for which the molecular size, charge, as well as
`target binding affinity play important roles [2]. There are
`
`* Correspondence: tiant10602@huabobio.com
`Department or cell Biology, Huabo Biopharm Cc Ltd., shanghai 201203,
`china
`Full list of author information is available at the end of the article
`
`several barriers to large molecule transport in solid tu-
`mors due to disordered vasculature, tissue structure, as
`well as extracellular matrix (ECM). These factors, which
`impact penetration and retention of large molecule
`drugs, have to be considered when designing new mo-
`lecular constructs.
`Angiogenesis, the process by which the existing vascular
`network expands to form new blood vessels, is mainly me-
`diated by vascular endothelial growth factor (VEGF),
`which upon binding with VEGF receptor (VEGFR), can in-
`duce phosphorylation of the receptors expressed in the
`blood vessel endothelial cells [1], thus leading to prolifera-
`tion of the endothelial cells and the development of the
`
`(ii) BioMed Central
`
`e 2015 Liu et at, licensee BlaMed Central. This is an Open Access article distributed under the terms of the Creative Commons
`Attribution License (httpi/creativecommons org/licenses/by/2.0), which permits unrestricted use, distribution, and
`reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain
`Dedication waiver thrtpi/c,eativecomnlons.org/publicdomairvzero/t.00 applies to the data made available in this article,
`unless otherwise stated
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 1
`
`
`
`Liu et al. BMC Cancer (2015)15:170
`
`Page 2 of 14
`
`vascular system. Under pathological conditions, VEGF-A
`and other members of the VEGF family including placen-
`tal growth factor (PIGF) are upregulated [3-6]. Among the
`factors contributing to angiogenesis, VEGF-A is the main
`ligand driving angiogenesis, making it an important target
`for drug development.
`Several drugs targeting VEGF have been approved for
`use in the treatment of cancer [7] as well as for wet age-
`related macular degeneration (AMD) [8]. Bevacizumab is
`a humanized antibody targeting VEGF-A and was ap-
`proved under the trade name of Avastin in 2004 for the
`treatment of metastatic colon cancer [9-11] as well as sev-
`eral other solid tumors including lung cancers [12,13),
`glioblastoma [14,15], renal cancers [16], and ovarian can-
`cers [17-19]. The main mechanism by which bevacizumab
`exerts anti-tumor activity is by preventing VEGF-A from
`binding with its receptors, thus resulting in inhibition of
`new blood vessel growth in tumor tissues. Bevacizumab is
`a humanized IgGi with over 90% of human and less than
`10% of murine components [20]. The recommended dose
`for bevacizumab is 5 mg/kg every 2 weeks, even though it
`could be detected in serum for 12 weeks [21]. Bevacizu-
`mab is the first VEGF blocker proven to improve survival
`by 30% in patients with metastatic colorectal cancer
`[22]. However due to target limitation (only targeting
`VEGF-A) as well as relatively poor tissue penetration
`because of its large size, the overall impact of bevacizu-
`mab in prolonging survival was very limited [22,23],
`with 5-year survival generally between 5% and 8% [23],
`suggesting that VEGF-A blockade alone may not be
`good enough to completely prevent tumor angiogenesis
`and corresponding tumor growth.
`Aflibercept (originally called VEGF-Trap) was approved
`in August of 2012 under the trade name of Zaltrap for the
`treatment of metastatic colon cancer, and the same mol-
`ecule was approved in November of 2011 under the trade
`name of Eylea for the treatment of AMD. Aflibercept is a
`recombinant fusion protein consisting of the second im-
`munoglobulin (Ig) domain of VEGFR1 and the third Ig
`domain of VEGFR2, fused to the immunoglobulin con-
`stant region (Fc) portion of human IgGi [24]. Unlike bev-
`acizumab, aflibercept exhibits affinity for all isoforms of
`VEGF and P1GF [25] and exerts robust antivascular ef-
`fects by rapid regression of existing tumor vessels [26],
`normalization of surviving mature vessels [27], and in-
`hibition of new tumor vessel growth [28). The anti-tumor
`efficacy of aflibercept has been confirmed in several solid
`tumor models, all demonstrating effective tumor inhib-
`ition [29]. Aflibercept has a MW of 110 kDa and has a
`half-life in plasma of 4-5 days [24]. The clinical benefits
`for aflibercept treatment of metastatic colon cancer pa-
`tients are similar to bevacizumab [30].
`It has been documented that the VEGF-binding affin-
`ity of VEGFR1 is 10 fold higher than that of VEGFR2
`
`[31] and the second Ig domain of VEGFR1 is critical for
`VEGF binding [32]. We reasoned that a recombinant
`protein composed of only the second domain (D2) of
`VEGFR1 might retain sufficient VEGF binding, but also
`have better bioavailability and penetration properties
`due to its smaller size as compared to the previously de-
`scribed current generation of drugs that block VEGF.
`We therefore designed an expression vector that
`expressed a recombinant protein consisting of the D2
`portion of VEGFR1 fused with the Fc portion of human
`IgGi. This protein was extensively characterized for its
`target-binding affinity, angiogenesis inhibition, and phar-
`macokinetic (PK) profile, as well as for its anti-tumor ef-
`ficacy in several xenograft tumor models.
`
`Methods
`Engineering of recombinant proteins
`HB-002.1 is a recombinant protein consisting of two com-
`ponents: one is the D2 domain of human VEGFR1 (Flt!)
`(P134-T226) plus 5 (S129-R133) and 2 (N227, T228)
`amino acids of upstream and downstream flanking se-
`quence respectively, and the second is the Fc fragment of
`human IgG1. To construct the HB-002.1 expression vec-
`tor, 57 nucleotides encoding the signal peptide of mouse
`IgGi heavy chain were added to the 5' end of VEGFR1-
`D2, a Kozak sequence was added to the 5' end of the
`signal peptide sequence, and cloning sites, Hindu! and
`EcoRl, were added to the 5' and 3' ends of the resulting
`sequence, respectively. This designed D2 expression cassette
`sequence was synthesized (GenScript) and subcloned
`into the Hindlll and EcoRI sites of the pHB-Fc vector
`(Generay, ID: X9913T).
`The recombinant Fltl[2]-Fc protein contains the
`VEGFR1-D2 domain (P134-T226) without the addition
`of flanking region amino acids, plus the Fc fragment of
`human IgGi.
`All recombinant proteins were expressed and purified
`from Chinese hamster ovary (CHO) cells (Cat# CCL-61,
`ATCC). 5 jig of each protein were loaded on 10% SDS-
`PAGE gels under reducing as well as non-reducing con-
`ditions. Gels were stained with 0.3% Coomassie Brilliant
`Blue R-250 and destained with 20% methanol.
`
`Western blotting and digestion of proteins with
`N-glycosidase F
`To validate the identity of the purified protein, Western
`blotting analysis was performed [33]. Briefly, different
`amounts of the purified protein (1, 0.5, 0.25 Vg) were sepa-
`rated by electrophoresis in 4-12% Bis-Tris protein gels, and
`then transferred to a polyvinylidene difluoride membrane.
`The membrane was probed using antibodies specific either
`for Fc fragment (horseradish peroxidase (HRP)-conjugated
`rabbit F(ab')2 anti-human lgG, Fc-fragment specific (Immu-
`noResearch Lab) or HRP*Polyclonal Rabbit Anti-Human
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 2
`
`
`
`Liu etal. BMC Cancer (2015)15:170
`
`Page 3 of 14
`
`IgG (Fc) (Cat#C030222, Cellway-Lab, Luoyang, China)), or
`for human VEGFR1 (Cat# 10136-RPO2, Sino Biological Inc)
`followed by incubation with secondary antibody (HRP-con-
`jugated Affinipure F(ab')2 Fragment Goat Anti Rabbit
`IgGi, F(ab')2 Fragment Specific (ImmunoResearch Lab)).
`Specific bands were visualized via the ECL kit according to
`the manufacturer's instructions (Amersham).
`To analyze the impact of glycosylation on protein activ-
`ity, HB-002.1 protein (Lot#20130521, 3.62 mg/ml) diluted
`to 0.5 mg/ml in 100 mM of ammonium bicarbonate was
`incubated with N-glycosidase F (Cat#11365193001, Sigma)
`(5 Unit/10 Vg protein) at 37°C for 18 hours. Digested and
`non-digested proteins were analyzed in 12% SDS-PAGE
`under reducing and non-reducing conditions. In parallel,
`the digested protein was also assayed for target binding ac-
`tivity, which was compared to that of the parental protein.
`
`Target-binding assay
`Target binding affinity of HB-002.1 was measured by
`ELISA in Falcon 96-Well ELISA Micro Plates coated
`overnight at room temperature with VEGF ligands or
`PIGF (R&D Systems) in PBS (100 ng per well). Coated
`plates were blocked with 3% dry fat milk in PBS-T buffer
`(PBS containing 0.05% Tween-20) and then 100 p1 of
`serially diluted I-IB-002.1 or bevacizumab (Lot#:N3526,
`Roche) or hlgG-Fc (Cat#:10702-HNAH, Sino Biological
`Inc) (from 5 nM to 0.0024 nM) were transferred into the
`plates. After incubation at room temperature for 1 hour,
`plates were washed S times with PBS-T solution, and
`then incubated with HRP-conjugated Fc-specific anti-
`body (Cat#C030222, Celiway-Lab, Luoyang, China) at
`room temperature for 1 hour. Plates were washed 5 times
`with PBS-T buffer and then developed with 100 p1 of
`HRP-substrate solution for up to 5 minutes. The reaction
`was stopped with 1 N H2SO4, and the absorbance at 450
`nM was determined in a standard plate reader.
`To determine the kinetic target binding affinity of HB-
`002.1, varying amounts of VEGF-A were mixed with 0.5
`nM of HB-002.1, Flt1[2]-Fc, hlgG-Fc or bevacizumab and
`then incubated for 2 hours at room temperature. The mix-
`tures were transferred to VEGF-A coated plates and in-
`cubated for 1 hour at room temperature, the non-bound
`proteins in solution were washed away, and the amounts
`of I-IB-002.1, Flt1 [21-Fc, hlgG-Fc or bevacizumab bound
`to the plates were measured by HRP-conjugated rabbit
`anti-human IgG-Fc antibody. The kinetic binding affinities
`were analyzed according to the amounts of free VEGF
`blocker in the mixtures.
`
`VEGFR2 phosphorylation assay
`4 ml of human umbilical vein endothelial cells (HUVECs)
`(Cat#HUVEC-004, ALLCELLS) in complete HUVEC-
`adapted medium (Cat#H-004, ALLCELLS) were incubated
`in 6 cm dishes at 37°C, 5% CO2 for 24 hours, cells were
`
`starved for 2 hours and then challenged for 15 minutes
`with either medium alone, or VEGF-A (20 ngiml) only, or
`VEGF-A pre-incubated with varying amounts of HB-
`002.1. Cells were washed twice with cold PBS and then
`dissolved in 200 p1 of lysis buffer (50 mM Tris, pH 7.4, 1%
`sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 1 mM
`EDTA, pH 8.0, 150 mM NaCi). After centrifugation and
`quantitation, equal amounts of supernatant from each
`sample were subjected to Western blotting analysis using
`antibodies specific either for total VEGFR2 (Cat# 2479,
`Cell Signaling Technology) or for VEGFR2 phosphotyro-
`sine (Cat# 3770S, Cell Signaling Technology).
`
`VEGF-Induced HUVEC proliferation and tube formation
`assay
`I-UJVEC proliferation in response to VEGF-A and the
`impact of HB-002.1 on cell proliferation was measured
`using CCK-8 kits (Cat# CKO4-11, DOJINDO Laborator-
`ies) following the manufacturer's instructions. Briefly,
`2000 HUVECs per well were plated in a 96-well plate,
`which was incubated at 37°C for 2 hours. 100 1il of re-
`agent solution containing 20 ng/ml of VEGF-A and
`varying amounts of HB-002.1, bevacizumab or hlgG-Fc
`were transferred to the plate. Cells were cultured for
`72 hours at 37°C, and then CCK-8 was added to these
`cultures, which were incubated for 4 additional hours
`followed by spectrophotometric analysis at 450 nm.
`The VEGF-induced tube formation assay was conducted
`as previously described [34]. Briefly, 50 p1 of HUVECs at 3
`x l05lml in culture medium were mixed with 50 0 of cul-
`ture medium containing 20 ng/ml of VEGF-A plus 1000
`nM HB-002.1 protein, bevacizumab or control human
`IgG. The mixtures were added to 96-well plates contain-
`ing 50 p1 of solidified MatrigeL Plates were incubated in a
`cell culture incubator at 37°C for 24 hours. Tube forma-
`tion was observed using an inverted phase contrast micro-
`scope (Eclipse TS100, Nikon). Images were captured with
`a CCD color camera (KP-D2OAU, Hitachi) attached to the
`microscope using 40x magnification plus 1.5x amplifica-
`tion by the CCD camera. The tube length in three differ-
`ent fields was measured using Image-Pro Plus software
`(Version 6.0, Media Cybernetics).
`
`Angiogenesis analysis
`The impact of HB-002.1 on angiogenesis was investigated
`using a transgenic zebrafish embryonic angiogenesis model
`[35]. Briefly, the tested protein or control drugs were
`microinjected into the common cardinal vein of zebraflsh
`at 48 hours post-fertilization (hpf). The subintestinal vessels
`(SIVs) were visualized under a Multi-Purpose Zoom Micro-
`scope (Nikon AZ100), and the area of the SIVs at 72 hpf
`was measured as mean f]uoresence intensity (MFI) using
`NIS-Elements D imaging software. The percentage of
`angiogenesis inhibition was calculated as (MR of vehicle
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 3
`
`
`
`Liu etal. BMC Cancer (2015)15:170
`
`Page 4 of 14
`
`treated S1Vs - MFI of drug treated SIVs)IMFl of vehicle
`treated SlVs x 100.
`
`Pharmacokinetic analysis
`16 BALBIc mice (female, age of 4-5 weeks, body weight
`of 18-20 g) received a subcutaneous (s.c.) injection of
`50 Vg HB-002.1 protein (-.2.5 mg/kg mouse) and bled at
`1, 2, 4, 6, 24, 48, 72, and 144 hours after injection. Levels
`of J-IB-002.1 in the plasma were measured by ELISA
`assay using human VEGF16S (R&D Systems) as capture
`protein and l-IRP-anti-human Fc (Jackson ImmunoRe-
`search Lab) as the detection antibody.
`
`In vivo efficacy study
`Mouse xenograft tumor models using human Colo-205
`and A549 cancer cells were applied to the investigation
`of the in vivo efficacy of l-(B-002.1. Cells purchased from
`ATCC were resuspended in serum-free medium. BALB/c
`nude mice were ordered from Shanghai SLAC Laboratory
`Animal Co. Ltd. The animals were specific pathogen free
`and approximately 4 - 5 weeks old upon arrival at Phar-
`maLegacy Laboratories. The procedures that were applied
`to animals in this protocol had been approved by Pharma-
`Legacy Laboratories JACUC before the execution of the
`study. Approximately 5 x 10' cells in 200 p1 of serum-free
`medium/matrigel (50:50 v/v) were injected s.c. in the right
`flank of each of the 70 mice for each model under
`anesthesia by 3 - 4% isoflurane. When the average tumor
`volume reached 100 - 200 mm3, 50 mice bearing tumors
`of suitable size were randomized into 5 groups (10 mice
`per group) according to tumor volume and body weight.
`Mice were treated with two different doses (5 mg/kg,
`20 mg/kg) of HB 002.1 or control drugs by intraperitoneal
`(i.p.) injections twice weekly for four weeks except for
`doxorubicin which was given only in one injection. Tumor
`volume and body weight were measured twice a week
`until the termination of the study. Tumor growth inhib-
`ition (TGI%) = (1-(change in mean treated tumor volume/
`change in mean control untreated tumor volume)) x 100.
`Tumor weight measured at time of mice sacrifice.
`
`Histology analysis
`Tumors were harvested and sectioned at the end of the ex-
`periments. Tumor sections were subsequently dewaxed
`and rehydrated. After quenching endogenous peroxidase
`activity, sections were immunohistochemically stained with
`respective antibody. Stained sections were dehydrated in
`alcohol and xylene, and then mounted. The procedure for
`hematoxylin and eosin (H&E) staining of tumor sections
`was as follows: dewaxing in xylene, gradient ethanol dehy-
`dration, hematoxylin staining, rinsing with tap water, coun-
`terstaining with eosin, rinsing with ethanol, gradient
`ethanol dehydration, and vitrification with xylene. Immu-
`nohistochemical staining was performed using antibodies
`
`specific for CD31 (Cat#: ab9498, Abeam) followed by goat
`anti-mouse secondary antibody (Cat#: K1T5002, Fuzhou
`Maixim) and goat anti-rabbit secondary antibody (Cat#:
`K1T5005, Fuzhou Maixim), respectively. The microvessel
`density was quantified by the visual approximation tech-
`nique, which involved manual counting vessels in three
`different microscope fields at IN magnification. The hist-
`ology results were analyzed by a pathologist on a single-
`blind basis. For tumor necrosis evaluation on H&E stained
`slides, homogenous staining in pink or pale color without
`cellular profiles/outline were considered necrotic cells,
`while cellular profiles/outlines with dark blue nuclei were
`considered healthy cells.
`
`Statistics
`Statistical software used for data analysis and presenta-
`tion was SAS 9.3 (SAS Institute), Prism 5 (GraphPad
`Software), and Excel 11 (Microsoft). Binding curves were
`calculated and presented using Prism 5 nonlinear reg-
`ression least squares fit sigmoidal dose-response variable
`slope (also known as four-parameter dose-response)
`curves. Comparisons between different treatment groups
`in HUVEC proliferation was performed using a two-way
`analysis of variance (ANOVA), which included the main
`effects of treatment group and loglO concentration, as
`well as the treatment group x loglO concentration inter-
`action. Upon finding a significant interaction effect, sep-
`arate one-way ANOVA comparisons were carried out at
`each concentration. If a significant difference was found,
`then Tukey's multiple comparisons were used. Compari-
`sons between different treatment groups in tube forma-
`tion by one-way ANOVA provided a F-test with a small
`P value (P = 0.0015) supporting subsequent Tukey's mul-
`tiple comparison test. Comparisons between control (ve-
`hicle-treated) and different treatment groups for inhibition
`of zebrafish angiogenesis were made by Dunnett's multiple
`comparison test. In vivo tumor volumes and weights were
`expressed as mean ± standard error of the mean or geo-
`metric mean with 95% confidence interval. Comparisons
`between different in vivo treatments and control PBS
`treated mice for changes in tumor weights were made by
`Mann-Whitney two-tailed test. For tumor volume, re-
`peated measures (RM) ANOVA with a mixed models ap-
`proach was used to determine if the treatment groups
`behaved differently across time (Le. the "group x time"
`interaction). A logio transformation of tumor volume was
`used to satisfy the required underlying assumptions of this
`statistical model. Since graphical analysis and theoretical
`considerations suggest that tumor volume grows logarith-
`mically, such that its rate of growth decreases over time, a
`log10transformation was applied to day (specifically, log 10of
`Day +1), and included as a linear main effect, as well as in
`the interaction term with group. The model contained one
`repeated "within subjects" factor of time, a "between
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 4
`
`
`
`Liu of at BMC Cancer (2015) 15:170
`
`Pages of 14
`
`animals' factor of treatment group, and the group x time
`interaction. Both group and time were considered fixed ef-
`fects in each of the RM ANOVA models, as necessarily
`was, the group x time interaction. Upon finding a signifi-
`cant difference, interest only focused on the comparison of
`the treatment groups to control (PBS), but not amongst
`each other. To calculate the statistical significance of treat-
`ments on TGI%, we calculated the ratio of tumor volume
`at Day 35 relative to Day 0 for each mouse, followed by a
`log transformation of this ratio to achieve normality (log
`Day35/Dayo), which is analytically equivalent to looking at
`percent change in tumor volume, but is more suited to
`conventional analysis. ANOVA was then used to compare
`the mean log ratios with the Student-Newman-Keuls test
`to make multiple comparisons. P < 0.05 was considered
`significant. For CD31 staining of tumor sections, only
`group descriptive statistics were calculated. No inferential
`statistical comparisons were performed since the sample
`size was so small (n = 3).
`
`Results
`
`Engineering and production of HB-002.1
`It has been documented that the second Ig-like domain
`(D2) of human VEGFR1 (Fltl[2]) is critical to VEGF
`binding [32], however the purified Fltl[2] fused with Pc
`
`did not bind to VEGF at all, and neither did truncated
`protein containing the first 2 domains (Flt1[1,2]) or that
`containing domains 2 and 3 (rltl[2,3]) [32]. Only protein
`containing domains 1-3 had fill VEGF binding activity
`comparable to that of the whole extracellular portion of
`wild type VEGFR1. This phenomenon was confirmed as
`well by Barleon et al [36), revealing the requirement of
`VEGF binding for the first three Ig-like domains, Based on
`these studies, we designed the HB-002.1 protein in which S
`flanking amino acids (S129-R133) at the N-terminal and 2
`amino acids (N227, T228) at the C-terminal of the D2 do-
`main were included with D2 (Figure 1A). The D2 domain-
`only (Fltl[2]-Fc) was also expressed as a control for VEGF
`binding assay.
`The 1-13-002.1 and Flt1[2)-Fc proteins were produced
`in CHO cells upon transfection with the corresponding
`construct. The secreted proteins were purified and
`resolved in 10% SDS-PAGE gels showing MWs of HB-
`002.1 and Flt1[2]-Fc at -110 kDa in non-reducing
`conditions, and -45 kDa in reducing conditions (Figure 1B),
`both relatively larger than the calculated MW, which
`is most likely due to glycosylation since there are
`two N-linked glycosylation sites in the D2 domain.
`Bevacizumab resolved in the correct MW positions
`(Figure 1B).
`
`A.
`
`Elti (2)-Fc
`
`HBOO2.1
`
`B.
`
`LB
`
`VEGFR1-02
`
`hlgol-Fc
`
`N-Ftank (saa)
`
`OFrank (2aa)
`it
`
`C.
`
`1
`
`D.
`
`NR
`
`P
`
`I 2 3 i,oa4 5 6
`
`rrtgo-Fc
`
`hr gO 'Fc
`
`P
`
`NR
`
`Load (Ito):
`
`0.5 0.25 aDa i 0.5 0.25
`
`htgG-Fc
`
`htgS-Fe
`
`Load (eg)
`
`P 0.2
`
`NF
`I 0.5 5 kDa 1 0.5 0.25
`
`Blotting with Fe-specific Ab
`
`Blotting with VEOFR1-specific Ab
`
`4
`
`I. 4: HB-002.l;
`2, 5: Bevacizumab
`3 6: Flt1(2)-Fc
`
`Ito
`
`=5—
`
`10
`05
`43
`
`34
`25
`
`Figure 1 Engineering and production of HB-002.1. (A) Diagram of HB-002,1 engineered structure, Illustration on top represents Fiti 121-Fc
`consisting of the 02 domain only fused with the Fc portion of human gG1. HB-002.1 contains the 02 domain plus sand 2 amino adds at the
`Sand 3' flanking region respectively. Signal peptide derived from the heavy chain of mouse IgGi ([5) was included for both constructs. (B)
`SOS-PAGE gel analysis. Three proteins were included in the analysis: HB-002.1 (Lane 1,4); Bevacizumab (Lane 2,5); Flt1l2l-Fc (Lane 3, 6). 5 pg of
`each protein were loaded under reducing and non-reducing conditions. (C-D), Western blot analysis. 0.25, 0.5, and 1 pg of HE 002.1 protein and
`1 pg of hlgG-Fc were resolved on 10% SOS-PAGE gels under reducing (RI or non reducing (NB) conditions, transferred to a polinylidene difluoride
`membrane, and probed with Fc-specific (C) or VEGFRI-speciftc (0) antibody (Ab). For comparison, protein MW size markers are shown in kDa.
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 5
`
`
`
`Mu et at. OMC Cancer (2015)15:170
`
`Pages of 14
`
`To confirm the identity of the proteins, Western blotting
`was performed using antibodies specific for Fc (Figure 1C)
`or VEGFR1 (Figure 1D), showing specific bands for each
`specified portion of the protein at different protein loading
`amounts.
`
`HB-002.l has strong binding affinity to VEGF-A
`HB-002.1 was first analyzed for its binding affinity to
`VEGF-A and compared with that of Fltl[2]-Fc and beva-
`cizumab. The data showed that HB-002.1 had a high af-
`finity with a half maximal effective concentration (ECSO)
`
`of 24 pM, which was 3-fold higher than that of bevacizu-
`mab (EC5O=72 pM) (Figure 2A). As expected, Flt1[2]-Fc
`only had a minimal binding activity to VEGF-A, con-
`firming a binding requirement for the flanking sequence.
`Binding activity of 1-18-002.1 to VEGF-B and PIGF was
`also investigated by ELISA, showing a modest binding to
`VEGF-B (Figure 28) but low binding to PIGF (Figure 2C).
`To determine the target-binding kinetics of HB-002.1,
`equilibrium binding assays were performed in which
`varying amounts of VEGF-A were mixed with 0.5 nM of
`l-18-002.1 or bevacizumab, and the unbound HB-002.1
`
`A.
`
`VEGF-A
`
`-
`
`• FIt1(2)-Fc
`1-15-002.1
`Bevacizurnab
`hlgG-Fc
`
`A
`
`D.
`
`0.6
`
`• 1-16-002.1
`Bevacizurnab
`
`A
`
`3-
`
`0
`
`In n
`0
`0
`
`0
`0.001
`
`0.01
`
`'n- "r rn,.,
`0.1
`1
`10
`conc. nM
`
`0.0
`0.01
`
`0.1
`
`I
`VEGE-165 (nM)
`
`10
`
`100
`
`B.
`
`VEGF-B
`
`E.
`
`Non-reducing
`
`Reducing
`
`• rhVEGFR1-Fc
`• HB-002.1
`hlgG-Fc
`
`11 00
`10
`1
`Protein cone. (nM)
`
`1000
`
`C.
`
`PIGF
`
`0.4
`
`0.31
`
`0.2
`
`0.1 -j
`
`0
`0
`
`• rhvEGFR1-Fc
`• HB-002.1
`htgG-Fe
`
`F.
`
`0
`In
`
`0
`
`D ND
`
`M 7 4:MT
`
`D ND
`
`- w
`
`S
`
`a
`
`-t Digested
`-.- Non-cgested
`-.- hlgG-Fc
`
`01
`
`1000
`
`0.01
`
`0.1
`
`100
`10
`1
`1 10
`Protein conc. (nM)
`Protein conc.(nM)
`Figure 2 Target binding activity of HB-002.1.Target binding activity of intact as well as deglycosylated HB-002.1 was analyzed by ELISA (A) Binding
`to VEGF-A was compared to bevacizurnab and Fft1(2). hlgG-Fc was used as negative control. (8-C) Binding to VEGF-B (8) and PIGF (C) was compared
`with rhVEGFR1-Fc (0) Kinetic binding affinity was measured by equilibrium binding assays that measures unbound HB-t2.1 or bevacizumab after
`incubation of 0.5 nM of HB-002.1 or bevacizumab with varying amounts of VEGF-165. (E) HB-002.1 deglycosylated by treatment with N-glycosidase F
`(D) or not deglycosylated (ND) was separated by SDS-PAGE under reducing or non-reducing conditions and visualized by staining with coomassie
`Brilliant Blue. (F) VEGF-A binding affinity was compared between intact (non-digested) and deglycosylated (digested) HB-oo2.1.
`
`100
`
`iobo
`
`Mylan Exhibit 1121
`Mylan v. Regeneron, IPR2021-00881
`Page 6
`
`
`
`Liu et aL BMC Cancer (2015) 15:170
`
`Page 7 of 14
`
`or bevacizumab was measured by ELISA using VEGF-A
`coated plate, revealing that HB-002.1 displays an equilib-
`rium dissociation constant (KD) of 180 pM, whereas bev-
`acizumab has a KD of 890 pM (Figure 2D).
`Since two different-sized bands were observed both in
`SDS-PAGE gels and in Western blots, we wondered if
`this was due to Winked glycosylation and if this gly-
`cosylation might have an impact on VEGF binding. To
`address these questions, HB-002.1 protein was digested
`with N-glycosidase F and then resolved in 10% SDS-
`PAGE gels, which showed a single band under reducing
`conditions and a smaller size single band under non-
`reducing conditions when compared to that of non-
`digested parental protein (Figure 2E). This confirmed
`our hypothesis that the doublet bands were due to N-
`linked glycosylation. The digested protein retained simi-
`lar VEGF-binding activity to that of parental protein
`(Figure 2F), indicating glycosylation is not essential for
`high affinity binding, which is consistent with the report
`by Barleon et al [36].
`
`HB-002.1 dose-dependently inhibited VEGF-induced
`VEGFR2 phosphorylation, HUVEC proliferation and tube
`formation
`Due to the strong VEGF binding affinity, we anticipated
`that FIB-002.1 must also have strong blocking activity
`against VEGF-induced VEGFR2 phosphorylation as well
`as the resulting cell proliferation and tube formation. As
`shown in Figure 3A, while strong phosphorylation was
`observed with VEGF addition and VEGF plus hlgG, the
`induced phosphorylation was sequentially diminished
`following addition of sequentially increasing amounts of
`HB-002.1, which is comparable to that of bevacizumab
`showing a dose-dependent inhibition of VEGFR2 phos-
`phorylation (Tyr951/1175) in 1-EUVECs [37].
`Comparisons between different treatment groups in
`HUVEC proliferation (Figure 3B) by two-way analysis of
`variance (ANOVA) provided a F-test with a small P value
`(P <0.0004) supporting subsequent evaluation for differ-
`ences among treatment groups. Significant inhibition, as
`compared to hlgG-Fc, in VEGF-induced HUVEC proli-
`feration was observed in a dose-dependent manner for
`I-EB-002.1 (P < 0.05 at all except lowest dose), which was
`comparable to that of bevacizumab (P <0.05 at all doses)
`(Figure 3B). The same phenomenon was also observed for
`VEGF-induced tube formation (Figure 3C, D), for which
`HB-002.1 had a significant and comparable inhibition to
`that of bevacizumab (P <0.05) as compared to hlgG, sug-
`gesting a strong blocking activity of HB-002.1 in VEGF-
`mediated cell biological activity.
`
`HB-002.1 dose-dependently Inhibited in vivo angiogenesis
`Using a transgenic zebrafish embryonic angiogenesis model
`[35], the impact of HB-002.1 on in viva angiogenesis was
`
`investigated and showed a dramatic reduction in number
`of SIVs. While 7-8 SIVs were usually observed in zebrafish
`at 72 hpf (Figure 4A), a decreased number of SIVs was ob-
`served when treated with HB..002.1 (Figure 43). The level
`of inhibition versus vehicle group reached 7.5 (±3.5) %
`(P>0.05), 15.2 (±3.3) % (P<0.01), and 21.4 (±2.4) %
`(P <0.001) for 1-13-002.1 at the doses of 4.4, 14.7, 44 ng. re-
`spectively. Because bevacizumab specifically binds human
`VEGF and its activity against zebrafish VEGF is not known,
`a broad spectrum angiogenesis inhibitor, endostatin,
`known to inhibit angiogenesis in this model [38], was used
`as a positive control showing a 9.7 (±2.8) % and 20.1 (±2.6)
`% inhibition at the dose of 44 and 100 ng respectively
`(Figure 4B). These results suggest a strong angiogenesis
`inhibition activity for HB-002.1.
`
`HB-002.1 has an excellent pharmacokinetic profile
`HB-002.1 has a much smaller molecular mass than
`current VEGF inhibitors, bevacizumab and aflibercept
`(-80 vs. -160 and -110 kDa, respectively), thus it might
`have a shorter half-life and worse PK profile compared
`to these drugs. To address these questions, 25 mg/kg of
`HB-002.1 were injected s.c. into mice (n = 16) and plasma
`taken at different time points post-injection were analyzed
`for 1-13-002.1 levels by ELISA. The results indicated that
`HB-002.1