lable at ScienceDirect

Biomaterials 117 (2017) 24e31

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

Single chain Fc-dimer-human growth hormone fusion protein forimproved drug delivery

Li Zhou a, Hsuan-YaoWang a, Shanshan Tong a, b, Curtis T. Okamoto a, Wei-Chiang Shen a, *,Jennica L. Zaro a, c, **

a Department of Pharmacology and Pharmaceutical Sciences, University of Southern California School of Pharmacy, 1985 Zonal Ave, Los Angeles, CA 90089-9121, United Statesb Department of Pharmaceutics, Jiangsu University School of Pharmacy, Zhenjiang, Jiangsu, Chinac Department of Pharmaceutical Sciences, West Coast University School of Pharmacy, 590 N. Vermont Ave, Los Angeles, CA 90004, United States

a r t i c l e i n f o

Article history:Received 26 August 2016Received in revised form25 November 2016Accepted 26 November 2016Available online 27 November 2016

Keywords:Fc fusionSingle chain FcProtein deliveryHuman growth hormonePharmacokineticsPharmacodynamics

* Corresponding author. 1985 Zonal Avenue, PSC 49121, United States.** Corresponding author. 590 N. Vermont Ave, LosStates.

E-mail addresses: weishen@usc.edu (W.-C. Shenedu (J.L. Zaro).

http://dx.doi.org/10.1016/j.biomaterials.2016.11.0510142-9612/© 2016 Published by Elsevier Ltd.

a b s t r a c t

Fc fusion protein technology has been successfully used to generate long-acting forms of several proteintherapeutics. In this study, a novel Fc-based drug carrier, single chain Fc-dimer (sc(Fc)2), was designed tocontain two Fc domains recombinantly linked via a flexible linker. Since the Fc dimeric structure ismaintained through the flexible linker, the hinge region was omitted to further stabilize it againstproteolysis and reduce FcgR-related effector functions. The resultant sc(Fc)2 candidate preserved theneonatal Fc receptor (FcRn) binding. sc(Fc)2-mediated delivery was then evaluated using a therapeuticprotein with a short plasma half-life, human growth hormone (hGH), as the protein drug cargo. Thisnovel carrier protein showed a prolonged in vivo half-life and increased hGH-mediated bioactivitycompared to the traditional Fc-based drug carrier. sc(Fc)2 technology has the potential to greatly advanceand expand the use of Fc-technology for improving the pharmacokinetics and bioactivity of proteintherapeutics.

© 2016 Published by Elsevier Ltd.

1. Introduction

The large foundation of knowledge of immunoglobulin G (IgG)molecules has promoted development and optimization of protein-based therapeutics and delivery systems, including monoclonalantibodies (mAbs), immunotoxins, antibody-drug conjugates, andFc-fusion proteins [1]. Fc-fusion proteins, due to binding of the IgGFc domain to the neonatal Fc receptor (FcRn), are involved in therecycling and transcytosis pathways of IgG [2]. The Fc domain of IgGbinds FcRn with high affinity at an acidic pH (<6.5), but withnegligible binding affinity at physiological pH (7.4) [3]. In cells witha slightly acidic extracellular pH, such as the small intestines, IgGbinds to FcRn at the cell surface followed by transcytosis of IgG fromthe apical to the basolateral surface [4e6]. The subsequent

04B, Los Angeles, CA 90089-

Angeles, CA 90004, United

), jZaro@westcoastuniversity.

exposure to the pH of blood, which is approximately 7.4, allows forthe dissociation and release of IgG into circulation. For cells with aneutral extracellular pH, it is generally thought that IgG is inter-nalized by fluid-phase pinocytosis, binds to FcRn in the acidifiedendosome, and is then either recycled or transcytosed [7]. Byinteracting with FcRn through the recycling and/or transcytosispathway, IgG can avoid degradation in lysosomes, resulting in itslong plasma half-life. The strategy of engineering Fc conjugatestakes advantage of this characteristic to prolong the plasma half-lifeof protein drugs [7].

To date, there are nine FDA-approved Fc-fusion proteins, andmany others are at different stages of clinical and preclinicaldevelopment. A majority of Fc-fusion protein drugs consist of aprotein drug linked to the N-terminal of an Fc domain that forms adrug-Fc homodimer (“(drug-Fc)2”) along with drug-Fc monomerimpurities (Fig. 1). In the (drug-Fc)2 homodimer configuration, theprotein drug domains are adjacent to each other, often leading totheir physical instability and/or decreased bioactivity [8,9]. Further,many large protein drugs are not suitable candidates as they cannotbe stably expressed [10]. In order to overcome these disadvantages,“Monomeric” Fc-fusion proteins (Fig. 1), containing a protein drug

Fig. 1. Schematic structures of current Fc-fusion protein technology and their disad-vantages. A majority of FDA-approved Fc-fusion proteins exist as a Fc-homodimer. (A)In this configuration, steric hindrance between the protein domains leads to physicalinstability, decreased bioactivity and/or limitations in the size of protein that can beaccommodated. Monomeric Fc fusion proteins have been recently developed toovercome these limitations. Both Fc-homodimers and Monomeric Fc fusion proteinsare (B) susceptible to instability in the hinge region via protease digestion and disulfidereduction, and (C) generate several impurities during recombinant production.

L. Zhou et al. / Biomaterials 117 (2017) 24e31 25

linked to only one of the two Fc domains (“drug-(Fc)2”) haverecently been tested and clinically approved (eg. Alprolix® andEloctate®). Studies have shown that these Monomeric Fc-fusionproteins have improved half-lives and/or bioactivity compared totheir homodimeric counterparts [8]. However, their main limita-tion is production, which requires dual expression plasmids con-taining the drug-Fc and the Fc sequences. This production protocolgenerates a mixture of multiple fusion products including (drug-Fc)2, drug-(Fc)2 and (Fc)2, creating issues of impurity. Further, for-mation of homodimers (i.e. (drug-Fc)2 and (Fc)2) are favored overthe Monomeric drug-(Fc)2 products, resulting in low productionyields and instability [11]. Due to these limitations, this promisingtechnology cannot be applied to all protein drugs. Current Fc-fusionproteins maintain the hinge region sequence of IgG to link the twoFc domains via disulfide bonds (Fig. 1). Other than linking two Fcdomains, this region is not important for Fc function but introducespotential instability due to disulfide reduction, and also via enzy-matic degradation at several protease cleavage sites present in thehinge region [12]. Additionally, the lower hinge of IgG Fc plays acrucial role in its binding to Fcg receptors (FcgR), initiating effectorfunctions that are out of the designed mechanism of action [13,14].

In this study, we have designed and evaluated a novel type of Fcfusion protein by using a long, flexible glycine-serine (GS) linker[15] to link two Fc chains, with the hinge sequence removed, tocreate a single chain Fc-dimer, sc(Fc)2. A flexible, rather than a rigid,linker was used in this study to allow the two Fc domains to interact

properly with each other. This type of linker is utilized in a varietyof fusion proteins, including the well-established single chain an-tigen binding proteins (i.e. scFv) [15,16]. The use of this novel designallows for the advantages of a Monomeric Fc-fusion protein,without the issues of production impurities and requirement of thehinge region. In this study, human growth hormone (hGH) waslinked to sc(Fc)2 to evaluate the protein drug delivery properties ofthis novel carrier protein. hGH deficiency is associated with severalclinical indications, including short stature, Turner syndrome,chronic kidney disease, HIV-associated wasting and abnormalmetabolism [17]. hGH has a very short plasma half-life of 3.4 h aftersubcutaneous injection, and 0.36 h after intravenous injection [18].Therefore, current treatment of hGH deficiency is limited to needleinjection of hGH several times a week, which is not favored bypatients, especially children and seniors [17]. hGH-sc(Fc)2 fusionprotein was then evaluated for its hGH-mediated bioactivity andpharmacokinetic properties. The evaluation of this novel Fc carrierprotein could also provide insights to mechanistic studies of Fc-FcRn interaction and future application to other therapeutic pep-tides or proteins.

2. Materials and methods

2.1. Cell culture

The human embryonic kidney cell line HEK293 and human co-lon carcinoma cell line T84 were purchased from ATCC (Manassas,VA), and Nb2 cells derived from rat T lymphoma cells were pur-chased from Sigma (St. Louis, MO). Cell culture media were all fromMediatech (Manassas, VA). HEK293 cells were cultured in Dulbec-co's modified Eagle's medium (DMEM) with 2.5 mM L-glutaminesupplemented with 10% fetal bovine serum (FBS) and 50 units ofpenicillin/50 mg streptomycin. T84 cells were cultured in 1:1mixture of Ham's F12 medium and DMEM with 2.5 mM L-gluta-mine, 10% FBS, and 50 units of penicillin/50 mg streptomycin. Nb2cells were cultured in Roswell Park Memorial Institute (RPMI) 1640medium supplemented with 2 mM L-glutamine, 10% FBS, 10% horseserum, 50 units of penicillin/50 mg streptomycin, and 50 mM 2-mercaptoethanol. All cell lines were maintained in a humidifiedincubator at 37 �C with 5% CO2.

2.2. Plasmid construction

2.2.1. pcDNA3.1þ_sc(Fc)2A series of plasmids encoding sc(Fc)2 with different linker

lengths, (Gly-Gly-Gly-Gly-Ser)n (“(G4S)n”) with n ¼ 8e14, wereconstructed using pcDNA3.1þ (Invitrogen, Carlsbad, CA) as thevector and the commercial plasmid pFUSE-hIgG1-Fc2 (Invivogen,San Diego, CA) as the template of Fc sequence (Fig. 2A). The pFUSE-hIgG1-Fc2 contains the DNA sequence of CH2 and CH3 domains ofwild-type hIgG1 with IL2 secretion sequence and a truncated hingesequence at the beginning of the CH2 domain. The truncated hingesequence was removed during sub-cloning.

2.2.2. pcDNA3.1þ_hGH-sc(Fc)2The pcDNA3.1þ_sc(Fc)2 was double digested with HindIII and

EcoRI restriction enzymes (New England Biolabs, Ipswich, MA) toreplace the IL2 secretion signal sequence with the hGH sequence,which has its own secretion signal, allowing for the expressedfusion protein to be collected from the culture medium.

2.2.3. pcDNA3.1þ_hGH-FcThe pcDNA3.1þ expression vector harboring hGH-transferrin

[17] was double digested with XhoI and XbaI restriction enzymes(New England Biolabs) to replace the transferrin fragment with the

Fig. 2. sc(Fc)2 production. (A) Plasmids encoding sc(Fc)2 were constructed inpcDNA3.1 þ vector. (B) Schematic structure of sc(Fc)2. Conditioned medium fromtransiently transfected HEK293 cells was collected and (C) analyzed by non-reducingSDS-PAGE and Western blot probed with goat anti-hIgG (Fc specific) antibody (1:3000dilution), or (D) purified by Protein A-Sepharose® 4B followed by reducing SDS-PAGEwith Coomassie blue staining.

L. Zhou et al. / Biomaterials 117 (2017) 24e3126

Fc fragment.

2.3. Recombinant fusion protein production

Fusion proteins were produced in serum-free CD293 medium(Life Technologies, Grand Island, NY) with 4 mM L-glutamine bytransient transfection of HEK293 cells using polyethylenimine(Polysciences, Warrington, PA). The mediumwas harvested at post-transfection day 4 and day 7. The combinedmedia containing targetfusion proteins were concentrated using a tangential flow filtrationsystem (Millipore, Billerica, MA). Protein A-Sepharose® 4B (Sigma,St. Louis, MO) was used to purify the expressed proteins from theconcentrated media. The combined elution after column purifica-tion was dialyzed (MWCO: 12e14 kD, Spectrum Laboratories,Rancho Dominguez, CA) in freshly prepared PBS, pH 7.4 at 4 �C. Theprotein solutions were filtered through a 0.2 mm membrane, andstored at 4 �C for future analysis. The final purity was determinedby SDS-PAGE with Coomassie blue staining using ChemiDoc™Touch and ImageLab™ software (Bio-Rad, Hercules, CA). Purifiedfusion proteins were identified using SDS-PAGE followed byWestern blot using goat anti-hIgG (Fc specific) antibody (1:3000dilution) (Sigma, St. Louis, MO) and goat anti-hGH specific antibody(1:1000 dilution) (R&D Systems, Minneapolis, MN).

2.4. FcRn binding assay

2.4.1. Cell-based assayT84 cells were seeded in 6-well plates, and cultured for 6 days

post-confluence to allow for differentiation. The iodination ofhIgG1-Fc (ACROBiosystems, Newark, DE) was performed as previ-ously described by our laboratory [17]. Serum-free medium wasprepared by adding 10 mM citric acid/salt into NaHCO3-free RPMImedium and filtered through a 0.2 mm membrane. The cells weredosed with 120 nM of 125I-hIgG1-Fc and ascending concentrationsof unlabeled hIgG1-Fc or sc(Fc)2, and incubated at 37 �C for 15 minin serum-free media adjusted to pH 7.4 for the control group or pH6.0 for the remaining treatment groups. After incubation, the cellswere washed 3 times with cold PBS followed by trypsinization. The

cells were collected in PBS and centrifuged (3000 rpm, 3 min,25 �C), and the radioactivity in the cell pellets wasmeasured using ag-counter (Packard, Downers Grove, IL).

2.4.2. Surface plasmon resonance (SPR)SPR measurements were performed by using a Biacore T100

instrument. shFcRn (ACROBiosystems, Newark, DE) was cross-linked to the dextran surface of a Series S Sensor Chip CM5 (GEHealthcare, Pittsburgh, PA) at pH 4.5e5 by amine coupling using 1-ethyl-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC)and NHS. Residual sites on the dextran were blocked with 1 Methanolamine hydrochloride (pH 8.5). A control flow cell wasblocked with ethanolamine for reference subtraction. Experimentswere performed in running buffer (50 mM phosphate, 100 mMsodium chloride, and 0.01% vol/vol Tween 20, pH 6.0 or 7.4). FcRnwas coated to a final density of ~9000 response units (RU). Dilutionsof recombinant Fc, sc(Fc)2, hGH-Fc and hGH-sc(Fc)2 in runningbuffer were injected over the shFcRn-CM5 chip at 20 mL/min for2 min (except 1 min for Fc group). The proteins were then disso-ciated from the chip for 2.5 min with running buffer. Remainingprotein was removed from the chip with regeneration buffer(10 mM HEPES, 150 mM NaCl, and 0.01% vol/vol Tween 20, pH 7.4)at 30 mL/min for 30 s. Sensorgrams were generated and analyzed byBiacore T100 Evaluation Software (version 2.0.2). The equilibriumRU observed for each injection was plotted against the concentra-tion of protein. The equilibrium Kd values were derived by analysisof the plots using the steady-state affinity model.

2.5. Nb2 cell proliferation assay

hGH bioactivity was measured using a well-established in vitroassay, the Nb2 cell proliferation assay, where suspension cultureshows a dose-dependent proliferation when exposed to exogenoushGH [19]. Nb2 cells growing at log-phasewerewashed 3 times withserum-free RPMI 1640 medium, re-suspended in serum-free assaymedium, and counted using Z1 Coulter particle counter (Beckman,Fullerton, CA). Nb2 cells were then seeded (15,000/well) in 96-wellplates in 200 mL FBS-free assay medium and cultured for 24 h. Thefusion protein samples and commercial recombinant hGH (Soma-tropin, LGC Standard USP1615708, United States PharmacopeiaConvention, Rockville, MD)were diluted into 10 mL PBS, and varyingdoses were added to the serum-starved Nb2 cells. After 4-day in-cubation, cells were incubated overnight with 20 mL of resazurinsolution (Biotium, Hayward, CA). The absorbance was measured at560 nm and 595 nm using a Genios microplate reader (Tecan, SanJose, CA) and normalized against the vehicle control.

2.6. Intravenous pharmacokinetics study

CF1 mice (male, 6 weeks, Charles River Laboratories, Wilming-ton, MA) were utilized for all of the in vivo assays in this study. Allanimal studies were performed in accordance with the NIH guid-ance in “Guide for the Care and Use of Laboratory Animals” andapproved by the University of Southern California InstitutionalAnimal Care and Use Committee. Animals were housed at 12 hlight/12 h dark cycles under standard conditions (room tempera-ture at 22 ± 3 �C and relative humidity at 50 ± 20%) with access toregular rodent chow (Labdiet, St. Louis, MO) and water. PurifiedhGH-sc(Fc)2 and sc(Fc)2 were injected into the tail vein of mice. At5 min, 30 min, 1 h, 4 h, and 8 h post-injection time points, bloodsamples were collected from the saphenous vein. At 72 h after in-jection, blood samples were collected from the heart. Blood sam-ples were immediately mixed with concentrated heparin to avoidcoagulation and then centrifuged at 2000 rpm for 20 min at 4 �C togenerate plasma samples. The plasma samples were tested by

Fig. 3. Short incubation uptake assay of hIgG1-Fc and sc(Fc)2 to determine FcRnbinding. Confluent T84 cells were incubated with 120 nM 125I-hIgG1-Fc and theindicated concentrations of unlabeled hIgG1-Fc or sc(Fc)2, and incubated at 37 �C for15 min in serum-free medium adjusted to pH 6.0. Cells were then washed with PBS,trypsinized, and the radioactivity in the cell pellets was measured. The data wereanalyzed by sigmoidal curve fitting in GraphPad Prism 5, where it was determined thatthe IC50 of hIgG1-Fc was 220.7 nM and of sc(Fc)2 was 344.0 nM. Each data pointrepresents mean ± SD (n ¼ 3).

L. Zhou et al. / Biomaterials 117 (2017) 24e31 27

Western Blot using anti-hIgG (Fc specific) antibody and quantita-tively analyzed by ChemiDoc™ Touch and ImageLab™ software.Purified hGH-sc(Fc)2 and hGH-Fc were injected into the tail vein ofmice. At 5 min, 30 min, 1 h, 4 h, and 8 h post-injection time points,blood samples were collected from the saphenous vein and mixedwith concentrated heparin to avoid coagulation. Blood sampleswere then centrifuged at 2000 rpm for 20 min at 4 �C to generateplasma samples. The plasma samples were analyzed by DoubleAntibody Human Growth Hormone RIA kit (MP Biomedicals, CostaMesa, CA).

2.7. Pharmacodynamics study

Purified hGH-sc(Fc)2, hGH-Fc, commercial recombinant hGHand PBS control were injected subcutaneously into male CF1 mice.The doses were normalized to the hGH portion of each fusionprotein as 1mg/kg hGH. At pre-injection and 1 h, 4 h, 8 h, 16 h, 28 h,48 h post-injection time points, blood samples were collected fromthe saphenous vein. At 72 h after injection, blood samples werecollected from the heart. Blood samples were immediately mixedwith concentrated heparin to avoid coagulation and kept on ice.Next, blood samples were centrifuged at 2000 rpm for 20 min at4 �C to generate plasma samples. The plasma samples wereanalyzed by Mouse/Rat IGF-I Quantikine ELISA Kit (R&D Systems,Minneapolis, MN). The absorbance at 450 nm and 540 nm weremeasured using a Synergy H1 Hybrid Plate Reader (BioTek,Winooski, VT). 4-Parameter logistic regression was used in curvefitting (elisaanalysis.com).

2.8. Statistical analysis

The two-tailed Student's t-test was applied to determine dif-ferences between data sets, where P < 0.05 was considered sta-tistically significant.

3. Results

3.1. sc(Fc)2 expression in mammalian system

Glycosylated Fc conjugates produced by mammalian systemsshowed better thermal stability and can endure exposure to low-pH better than aglycosylated Fc conjugates expressed in E. coli[20], so the human cell line HEK293 was chosen as the expressionplatform. Full human post-translational modifications provided byHEK293 also leaves less concerns on potential immunogenicity.Different lengths (n ¼ 8e14) of a commonly used flexible linker,(G4S)n, were used to link two Fc chains to generate a single chainFc-dimer protein (Fig. 2B). In a small-scale expression test of sc(Fc)2with various linkers, the expression levels increased withincreasing linker length, where the highest expression level wasachieved when n ¼ 12 or 13 (Fig. 2C). The small molecular weightimpurity, which was positive in anti Fc-Western blot, did not in-crease after 7-day incubation with the conditioned medium ofHEK293 cell culture, suggesting it was a product generated duringthe intracellular process of the fusion protein rather than thedegradation in the medium. Based on band intensity, the calculatedpercentage of monomer impurity showed that the (G4S)13 linkerprovided the highest expression purity as well as highest expres-sion level. Therefore, the (G4S)13 linker was selected for further usein all of the subsequent studies for the sc(Fc)2 fusion proteins,including both sc(Fc)2 and hGH-sc(Fc)2 in this report. sc(Fc)2, singlechain-Fc dimer with a (G4S)13 linker, was expressed in large-scaleand purified by Protein A affinity chromatography (Fig. 2D). Thedominant band showed about 80% abundance. Due to the glyco-sylation on the CH2 domains, the apparent molecular weight of

sc(Fc)2 under reducing condition shown on the SDS-PAGE gel washigher than its calculated molecular weight (54 kDa).

3.2. Functional test of sc(Fc)2

To evaluate the biological function of sc(Fc)2, a short time courseuptake assay was carried out to compare the binding affinity ofsc(Fc)2 and hIgG1-Fc to FcRn in T84 cells, which express FcRn in arelatively high levels compared with other commonly used humanintestinal cell lines [4,21]. The cells were dosed with 120 nM of 125I-hIgG1-Fc and ascending concentrations of unlabeled hIgG1-Fc orsc(Fc)2 to compete. The results showed that the internalizationwasspecific for Fc, and was inhibited in a dose-dependent manner byboth hIgG1-Fc and sc(Fc)2 (Fig. 3). The IC50 values for hIgG1-Fc andsc(Fc)2 were 220.7 nM and 344 nM, respectively.

The binding of Fc, sc(Fc)2, hGH-Fc and hGH-sc(Fc)2 to shFcRnwas also studied by SPR based on the experimental conditionsdeveloped by Mezo et al. [22] (Fig. 4). At pH 6.0, Fc showed a Kd of44.9 nM, and sc(Fc)2 showed a weaker Kd of 285 nM. After fusinghGH to the N-terminal of the carrier proteins Fc or sc(Fc)2, the Kddropped to 1.43 mM (hGH-Fc) or 400 nM (hGH-sc(Fc)2). No bindingwas observed at pH 7.4.

3.3. hGH-sc(Fc)2 fusion protein expression in mammalian system

The same expression system using vector pcDNA3.1þ and theHEK293 cell line was adapted to express hGH-sc(Fc)2 as well ashGH-Fc for comparison. Both of the fusion proteins were expressedin large-scale and purified with Protein A affinity chromatography.The purity of hGH-sc(Fc)2, calculated based on Coomassie bluestained gels under reducing condition, was over 80% (Fig. 5). Theidentity of both fusion proteins was determined by the recognitionof anti-hGH and anti-hIgG (Fc specific) antibodies in Western blot(Supplementary Data Fig.1). The fusion proteins were also analyzedfor degradation following storage in different buffer systems atdifferent temperatures. The products were stable for up to2months at 4 �C in PBS, the storage conditionwe used, as measuredby SDS-PAGE/Coomassie blue staining for protein purity and byNb2cell proliferation assay for confirming activity (Supplementary DataFig. 2/Table 1 & Table 2).

Fig. 4. SPR Sensorgams for binding of serial twofold dilutions of (A) Fc (400 nMe6.25 nM), (B) sc(Fc)2 (3200 nMe6.25 nM), (C) hGH- sc(Fc)2 (6400 nMe50 nM), and (D) hGH-Fc(6400 nMe6.25 nM) to immobilized shFcRn at pH 6.0.

Fig. 5. hGH-sc(Fc)2 production. (A) Conditioned media from transiently transfected HEK293 cells was collected and purified by Protein A-Sepharose® 4B followed by reducing ornon-reducing SDS-PAGE with Coomassie blue staining. (B) Schematic structure of hGH-sc(Fc)2.

L. Zhou et al. / Biomaterials 117 (2017) 24e3128

3.4. Nb2 cell proliferation

Following treatment of Nb2 cells with various concentrations ofthe fusion proteins or hGH control, cell growth was stimulated in adose-dependent manner (Fig. 6). The bioactivity of free hGH andhGH-sc(Fc)2 were similar, with EC50 values of 166 pM and 284 pM,respectively. However, the activity of the hGH-Fc fusion proteinwasapproximately 3-fold lower (EC50 ¼ 1064 pM).

3.5. Intravenous pharmacokinetics study

In male CF1 mice, the half-life of hGH-sc(Fc)2 ranged from 3.01(±0.25) to 4.06 (±1.03) h in separate experiments using differentdetection methods, which was much shorter than that of sc(Fc)2

alone (14.57 ± 2.74 h) (Fig. 7). On the other hand, the half-life ofhGH-sc(Fc)2 was approximately 2-fold longer than that of hGH-Fc(1.32 ± 0.23 h) (Fig. 7B), and over 12-fold longer than the previ-ously reported half-life of hGH (less than 15 min) in male CF1 mice[23].

3.6. Pharmacodynamics study

The secretion of GH is pulsatile and can be affected by manyintrinsic and extrinsic factors including sleeping, feeding and so on,therefore direct measurement of GH levels could be misleading.Instead, the circulating IGF-1 level, which is primarily induced byGH stimulation, is often used as an indicator of average GH levels[24e26] and is the most important biomarker for diagnosis and

Fig. 6. Nb2 cell proliferation stimulated by hGH fusion proteins. Nb2 cells were serum-starved for 24 h, and then treated with the indicated concentrations of fusion proteins.Cell proliferation was determined by the resazurin assay after a 4-day incubation. Thedata were collected as duplicates and analyzed by sigmoidal curve fitting in GraphPadPrism 5, where it was determined that the EC50 of hGH, hGH-sc(Fc)2 and hGH-Fc were166 pM, 284 pM and 1064 pM, respectively.

Fig. 7. Pharmacokinetic profiles after i. v. injection. (A) Pharmacokinetic profiles ofhGH-sc(Fc)2 and sc(Fc)2 after i. v. injection. Male CF1 mice were injected via the tailvein with 1.30 mg/kg hGH-sc(Fc)2 or 0.90 mg/kg sc(Fc)2, and blood samples werecollected from the saphenous vein at the indicated time points. The plasma concen-tration (Cp) was determined by quantitative Western blot. Each data point representsmean ± SD (n ¼ 5). (B) Pharmacokinetic profiles of hGH-sc(Fc)2 and hGH-Fc after i. v.injection. Male CF1 mice were injected via the tail vein with 1.30 mg/kg hGH-sc(Fc)2 or0.80 mg/kg hGH-Fc, and blood samples were collected from the saphenous vein at theindicated time points. The Cp was determined by double antibody hGH RIA kit. Eachdata point represents mean ± SD (n ¼ 3).

Fig. 8. IGF-1 plasma levels after subcutaneous injection of hGH-fusion proteins in maleCF1 mice. Mice were injected (s.c.) with 1 mg/kg hGH-fusion proteins (normalized tohGH) and blood samples were collected from the saphenous vein at the indicatedtimepoints. Plasma samples were then isolated and analyzed by Mouse/Rat IGF-1Quantikine ELISA. The data at t ¼ 0 were obtained from the blood samples collected1 h before injection. Each data point represents mean ± SD (n ¼ 4e5). The two-tailstudent t-test was used to compare the data from the hGH-sc(Fc)2 and hGH-Fc groups.

L. Zhou et al. / Biomaterials 117 (2017) 24e31 29

monitoring of GH deficiency [27]. To evaluate in vivo bioactivity,purified hGH-sc(Fc)2, hGH-Fc, commercial hGH standard, and PBScontrol were injected subcutaneously into male CF1 mice. Asshown in Fig. 8, IGF-1 levels in the PBS control group were main-tained at approximately 500 ng/mL throughout the 3-day experi-mental period. The IGF-1 levels of the hGH group showed nosignificant difference from the baseline at this dose, which isconsistent with previous studies [24,25]. On the other hand, the

IGF-1 levels in both hGH-sc(Fc)2 and hGH-Fc fusion protein groupsreached peak levels between 16 h and 28 h. The IGF-1 levels in thehGH-sc(Fc)2 group were significantly higher than that of the hGH-Fc group at the 1e28 h post-injection time points.

4. Discussion

The main complication with protein drugs is their instability,leading to short half-lives and requiring administration via injec-tion at a high dosing frequency. Fc fusion proteins are one type oftechnology used to extend the plasma half-life of protein drugs, andthere are currently 9 FDA-approved Fc-fusion proteins. Most Fc-fusion proteins drugs are in a homodimer-dominant form, whichhave several disadvantages including instability due to the juxta-posed position of the two protein drugs, and limitations in the sizeof the protein drug it can accommodate. Monomeric Fc-fusionproteins, containing a protein drug linked to only one Fc domain,have improved half-lives and/or bioactivity compared to theirhomodimeric counterparts. However, current recombinant pro-duction methods generate a mixture of fusion products (Fig. 1) thatare difficult to separate. Finally, both configurations contain thehinge region, which is susceptible to cleavage by proteases or di-sulfide reduction resulting in destabilization and/or degradation ofthe fusion product. In this study, we explored a single chain form ofthe Fc domain of IgG1, sc(Fc)2, as a novel approach in overcomingthese disadvantages with current Fc-technologies to improve thepharmacokinetics of protein drugs.

Using HEK293 cells that have human glycoforms [28], we haveproduced, purified, and analyzed the stability of the sc(Fc)2 fusionproteins. Although glycosylation is not necessary for FcRn binding,it has been shown to enhance the stability of Fc [29]. As shown inFig. 2C, sc(Fc)2 was designed to contain (G4S)n linkers with differentrepeats from n ¼ 8 to 14, and it was found that the GS linker lengthaffected the production yield and formation of stable dimers.Shorter linkers had a lower production yield and larger percentageof impurities, while the longest repeat tested, n ¼ 14, resulted in arelatively lower production yield. Our preliminary data(Supplementary Data Fig. 3) showed that the majority of hIgG1-Fcin the expressing medium existed as dimers. This result is consis-tent with the general knowledge of the naturally favored dimericstructure of IgG as well as the 2:1 ratio of FcRn-IgG complexesobserved in crystal structures [30,31] and SPR assays [32]. By

*: P < 0.05; **: P < 0.01; ***: P < 0.001.

L. Zhou et al. / Biomaterials 117 (2017) 24e3130

adding a long flexible linker, the dimeric structure of Fc is expectedto be protected, especially in in vivo conditions. Different linkerlengths allow different levels of flexibility in protein folding, andinfluence the production level at the same time. The major impu-rity, which can be recognized by anti-hIgG (Fc specific) antibodyand cannot be purified by Protein A column, may result fromimproper folding near the linker region. The (G4S)13 linker in sc(Fc)2provides the best balance between expression level and purity(Fig. 2C). The binding of these sc(Fc)2 to Protein A indicates theirproper folding at the CH2eCH3 interface that is also the regioninteracting with FcRn [13,31,33]. Purified sc(Fc)2 could specificallycompete with (Fc)2 binding in a dose-dependent manner, similar to(Fc)2 in T84 cells (Fig. 3). Therefore, the Fc domains of sc(Fc)2maintain the original binding affinity to FcRn.

A sc(Fc)2 containing hGH was also produced and analyzed forin vitro/in vivo bioactivity and in vivo pharmacokinetics. In tradi-tional Fc-fusion proteins, therapeutic proteins are fused at the N-terminal of an intact Fc sequence.We tested both N-terminal and C-terminal, however the expression level of hGH-sc(Fc)2 was about 3-fold higher than that of sc(Fc)2-hGH (Supplementary Data Fig. 4).Fusing hGH at the N-terminal promotes the desired folding of thefusion protein. The activity of the hGH-sc(Fc)2 fusion protein wasdetermined by the Nb2 proliferation assay [17,34]. As shown inFig. 6, the hGH-sc(Fc)2 fusion protein maintained 58% biologicalactivity compared to free hGH, and was approximately 3.7-foldhigher than the bioactivity of hGH-Fc. An explanation for thisphenomenon could be that when two hGH-Fc molecules form adimeric structure, the two hGH domains are sterically hindered,reducing receptor binding. Similarly, the steric hindrance alsonegatively influences FcRn binding of the fusion proteins (Fig. 4).The in vivo biological activity of the hGH-fusion proteins was alsodetermined by measuring the increase in IGF-1 secretion followingtreatment in CF1 mice [35e37]. For this assay, CF1 mice weresubcutaneously injected with hGH-sc(Fc)2, hGH-Fc or hGH, and theIGF-1 levels were monitored at various time points using IGF-1specific ELISA. As shown in Fig. 7, the IGF-1 levels were signifi-cantly higher in the hGH-sc(Fc)2 treated group compared to bothhGH-Fc and hGH treatment. Finally, the half-life of hGH-sc(Fc)2 wasabout 2-times longer than hGH-Fc following intravenous injectionin CF1 mice (Fig. 7B). The shorter half-life of hGH-sc(Fc)2 comparedto sc(Fc)2 could be a result of its lower binding to FcRn observed bySPR (Kd ¼ 400 nM vs 285 nM, respectively), as well as GH receptor-mediated degradation as target-mediated disposition for proteindrugs [38,39]. The effect of receptor-binding affinity of each proteindomain on the plasma half-life of a bifunctional fusion protein hasbeen demonstrated in our previous study of hGH-transferrin fusionproteins [23]. The shorter half-life of hGH-Fc compared to hGH-sc(Fc)2 is also likely due to its lower FcRn binding affinity(Kd ¼ 1.43 mM vs 400 nM, respectively). Another possible contrib-uting factor is that in the in vivo situation the dimeric structure ofhGH-Fc could be disrupted forming a higher amount of lowermolecular weight monomers of hGH-Fc. Since the molecularweight of hGH-Fc monomers (~47 kDa) are slightly below theglomerular filtration molecular weight cutoff (50e60 kDa), theshorter half-life of hGH-Fc could be due to a higher kidney elimi-nation of the monomers. Our in vivo data of the two control groups,hGH and hGH-Fc, are consistent with the previously reportedpharmacokinetic/pharmacokinetic studies of hGH-Fc in maleSprague-Dawley rats from another group [40]. In their study, hGH-Fc had a much longer plasma half-life than hGH. After a 14-daydaily injection in hypophysectomized rats, the high-dose hGH-Fcgroup showed a statistically significant increase in weight gaincompared with hGH group.

Taken together, these in vitro and in vivo bioactivity assays showthat the hGH-sc(Fc)2 fusion protein displays a superior bioactive

response compared to the hGH-Fc fusion protein. Further, the longhalf-life obtained for hGH-sc(Fc)2 compared to hGH-Fc is a verypromising result and supports our proposed hypothesis on thesuperior stability of hGH-sc(Fc)2.

5. Conclusion

In this study, we describe the design of a novel single chain Fc-based drug carrier, sc(Fc)2, and evaluated sc(Fc)2-mediated deliveryusing hGH as the protein drug cargo. This novel carrier proteinshowed significant improvements in half-life and bioactivity of theprotein drug cargo comparedwith traditional Fc-based drug carrier.sc(Fc)2 technology has the potential to greatly improve and expandthe use of Fc-technology for improving the pharmacokinetics ofprotein drugs.

Acknowledgement

The authors would like to thank Hsin-Fang Lee and Yuqian Liufor their assistance with carrying out experiments for this study,and Zoe Folchman-Wagner for proof-reading and editing thefinalized version of the manuscript. The authors also thank Dr.Shuxing Li at Center of Excellence in NanoBiophysics, University ofSouthern California for assistance with SPR. ST was supported byJiangsu Government Scholarship for Overseas Studies, JiangsuProvincial Department of Education. This work was supported byJohn A. Biles Professorship (to W.-C. S), University of SouthernCalifornia School of Pharmacy.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.biomaterials.2016.11.051.

References

[1] A. Beck, T. Wurch, C. Bailly, N. Corvaia, Strategies and challenges for the nextgeneration of therapeutic antibodies, Nat. Rev. Immunol. 10 (2010) 345e352.

[2] D.C. Roopenian, S. Akilesh, FcRn: the neonatal Fc receptor comes of age, Nat.Rev. Immunol. 7 (2007) 715e725.

[3] M. Raghavan, V.R. Bonagura, S.L. Morrison, P.J. Bjorkman, Analysis of the pHdependence of the neonatal Fc receptor/immunoglobulin G interaction usingantibody and receptor variants, Biochemistry 34 (1995) 14649e14657.

[4] B.L. Dickinson, K. Badizadegan, Z. Wu, J.C. Ahouse, X. Zhu, N.E. Simister,R.S. Blumberg, W.I. Lencer, Bidirectional FcRn-dependent IgG transport in apolarized human intestinal epithelial cell line, J. Clin. Invest. 104 (1999)903e911.

[5] S.M. Claypool, B.L. Dickinson, J.S. Wagner, F.-E. Johansen, N. Venu,J.A. Borawski, W.I. Lencer, R.S. Blumberg, Bidirectional transepithelial IgGtransport by a strongly polarized basolateral membrane Fcg-receptor, Mol.Biol. Cell 15 (2004) 1746e1759.

[6] S. Tzaban, R.H. Massol, E. Yen, W. Hamman, S.R. Frank, L.A. Lapierre,S.H. Hansen, J.R. Goldenring, R.S. Blumberg, W.I. Lencer, The recycling andtranscytotic pathways for IgG transport by FcRn are distinct and display aninherent polarity, J. Cell Biol. 185 (2009) 673e684.

[7] T.T. Kuo, K. Baker, M. Yoshida, S.W. Qiao, V.G. Aveson, W.I. Lencer,R.S. Blumberg, Neonatal Fc receptor: from immunity to therapeutics, J. Clin.Immunol. 30 (2010) 777e789.

[8] J.A. Dumont, S.C. Low, R.T. Peters, A.J. Bitonti, Monomeric Fc fusions: impact onpharmacokinetic and biological activity of protein therapeutics, Biodrugs 20(2006) 151e160.

[9] J.L. Fast, A.A. Cordes, J.F. Carpenter, T.W. Randolph, Physical instability of atherapeutic Fc fusion protein: domain contributions to conformational andcolloidal stability, Biochemistry 48 (2009) 11724e11736.

[10] R.T. Peters, G. Toby, Q. Lu, T. Liu, J.D. Kulman, S.C. Low, A.J. Bitonti, G.F. Pierce,Biochemical and functional characterization of a recombinant monomericfactor VIII-Fc fusion protein, J. Thromb. Haemost. 11 (2013) 132e141.

[11] P. Carter, Bispecific human IgG by design, J. Immunol. Methods 248 (2001)7e15.

[12] M.H. Ryan, D. Petrone, J.F. Nemeth, E. Barnathan, L. Bjorck, R.E. Jordan, Pro-teolysis of purified IgGs by human and bacterial enzymes in vitro and thedetection of specific proteolytic fragments of endogenous IgG in rheumatoidsynovial fluid, Mol. Immunol. 45 (2008) 1837e1846.

[13] B.D. Wines, M.S. Powell, P.W.H.I. Parren, N. Barnes, P.M. Hogarth, The IgG Fc

L. Zhou et al. / Biomaterials 117 (2017) 24e31 31

contains distinct Fc receptor (FcR) binding sites: the leukocyte receptors Fc RIand Fc RIIa bind to a region in the Fc distinct from that recognized by neonatalFcR and protein A, J. Immunol. 164 (2000) 5313e5318.

[14] R.L. Shields, A.K. Namenuk, K. Hong, Y.G. Meng, J. Rae, J. Briggs, D. Xie, J. Lai,A. Stadlen, B. Li, J.A. Fox, L.G. Presta, High resolution mapping of the bindingsite on human IgG1 for Fc gamma RI, Fc gamma RII, Fc gamma RIII, and FcRnand design of IgG1 variants with improved binding to the Fc gamma R, J. Biol.Chem. 276 (2001) 6591e6604.

[15] X. Chen, J.L. Zaro, W.C. Shen, Fusion protein linkers: property, design andfunctionality, Adv. Drug Deliv. Rev. 65 (2013) 1357e1369.

[16] Z. Shen, H. Yan, Y. Zhang, R.L. Mernaugh, X. Zeng, Engineering peptide linkersfor scFv immunosensors, Anal. Chem. 80 (2008) 1910e1917.

[17] N. Amet, W. Wang, W.C. Shen, Human growth hormone-transferrin fusionprotein for oral delivery in hypophysectomized rats, J. Control Release 141(2010) 177e182.

[18] R. Webster, R. Xie, E. Didier, R. Finn, J. Finnessy, A. Edgington, D. Walker,PEGylation of somatropin (recombinant human growth hormone): impact onits clearance in humans, Xenobiotica 38 (2008) 1340e1351.

[19] T. Tanaka, R.P.C. Shiu, P.W. Gout, C.T. Beer, R.L. Noble, H.G. Friesen, A Newsensitive and specific bioassay for lactogenic hormones: measurement ofprolactin and growth hormone in human serum, J. Clin. Endocrinol. Metab. 51(1980) 1058e1063.

[20] W. Cao, D.M. Piedmonte, M.S. Ricci, P.Y. Yeh, Formulation, drug product, anddelivery: considerations for Fc-fusion proteins, in: Therapeutic Fc-fusionProteins, Wiley-VCH Verlag GmbH & Co. KGaA, 2014, pp. 115e154.

[21] S. Foss, A. Grevys, K.M. Sand, M. Bern, P. Blundell, T.E. Michaelsen, R.J. Pleass,I. Sandlie, J.T. Andersen, Enhanced FcRn-dependent transepithelial delivery ofIgG by Fc-engineering and polymerization, J. Control Release 223 (2015)42e52.

[22] A.R. Mezo, K.A. McDonnell, C.A.T. Hehir, S.C. Low, V.J. Palombella, J.M. Stattel,G.D. Kamphaus, C. Fraley, Y. Zhang, J.A. Dumont, A.J. Bitonti, Reduction of IgGin nonhuman primates by a peptide antagonist of the neonatal Fc receptorFcRn, Proc. Natl. Acad. Sci. 105 (2008) 2337e2342.

[23] X. Chen, H.-F. Lee, J.L. Zaro, W.-C. Shen, Effects of receptor binding on plasmahalf-life of bifunctional transferrin fusion proteins, Mol. Pharm. 8 (2011)457e465.

[24] L. Wu, S. Ji, L. Shen, T. Hu, Phenyl amide linker improves the pharmacokineticsand pharmacodynamics of N-terminally mono-PEGylated human growthhormone, Mol. Pharm. 11 (2014) 3080e3089.

[25] L. Wu, S. Ji, T. Hu, N-Terminal modification with pseudo-bifunctional peg-hexadecane markedly improves the pharmacological profile of human growthhormone, Mol. Pharm. 12 (2015) 1402e1411.

[26] J.Y. Lee, S.K. Kang, H.S. Li, C.Y. Choi, T.E. Park, J.D. Bok, S.H. Lee, C.S. Cho,Y.J. Choi, Production of recombinant human growth hormone conjugated witha transcytotic peptide in pichia pastoris for effective oral protein delivery,Mol. Biotechnol. 57 (2015) 430e438.

[27] A. Mukherjee, S.M. Shalet, The value of IGF1 estimation in adults with GHdeficiency, Eur. J. Endocrinol. 161 (2009) S33eS39.

[28] Biological Drug Products: Development and Strategies, John Wiley and Sons,2013.

[29] L.C. Simmons, D. Reilly, L. Klimowski, T.S. Raju, G. Meng, P. Sims, K. Hong,R.L. Shields, L.A. Damico, P. Rancatore, D.G. Yansura, Expression of full-lengthimmunoglobulins in Escherichia coli: rapid and efficient production of agly-cosylated antibodies, J. Immunol. Methods 263 (2002) 133e147.

[30] S. Krapp, Y. Mimura, R. Jefferis, R. Huber, P. Sondermann, Structural analysis ofhuman IgG-Fc glycoforms reveals a correlation between glycosylation andstructural integrity, J. Mol. Biol. 325 (2003) 979e989.

[31] V. Oganesyan, M.M. Damschroder, K.E. Cook, Q. Li, C. Gao, H. Wu,W.F. Dall'Acqua, Structural insights into neonatal Fc receptor-based recyclingmechanisms, J. Biol. Chem. 289 (2014) 7812e7824.

[32] Y.N. Abdiche, Y.A. Yeung, J. Chaparro-Riggers, I. Barman, P. Strop, S.M. Chin,A. Pham, G. Bolton, D. McDonough, K. Lindquist, J. Pons, A. Rajpal, Theneonatal Fc receptor (FcRn) binds independently to both sites of the IgGhomodimer with identical affinity, MAbs 7 (2015) 331e343.

[33] T. Rath, K. Baker, J.A. Dumont, R.T. Peters, H. Jiang, S.W. Qiao, W.I. Lencer,G.F. Pierce, R.S. Blumberg, Fc-fusion proteins and FcRn: structural insights forlonger-lasting and more effective therapeutics, Crit. Rev. Biotechnol. 35 (2015)235e254.

[34] N. Amet, X. Chen, H.F. Lee, J.L. Zaro, W.C. Shen, Transferrin receptor-mediatedtranscytosis in intestinal epithelial cells for gastrointestinal absorption ofprotein drugs, in: Targeted Delivery of Small and Macromolecular Drugs, CRCPress, 2010, pp. 31e52.

[35] S.M. Donovan, L.C. Atilano, R.L. Hintz, D.M. Wilson, R.G. Rosenfeld, Differentialregulation of the insulin-like growth factors (IGF-I and -II) and IGF bindingproteins during malnutrition in the neonatal rat, Endocrinology 129 (1991)149e157.

[36] X. Zhao, S.M. Donovan, Combined growth hormone (GH) and insulin-likegrowth factor-I (IGF-I) treatment is more effective than GH or IGF-I alone atenhancing recovery from neonatal malnutrition in rats, J. Nutr. 125 (1995)2773e2786.

[37] X. Zhao, T.G. Unterman, S.M. Donovan, Human growth hormone but not hu-man insulin-like growth factor-I enhances recovery from neonatal malnutri-tion in rats, J. Nutr. 125 (1995) 1316e1327.

[38] D.E. Mager, Target-mediated drug disposition and dynamics, Biochem. Phar-macol. 72 (2006) 1e10.

[39] E. Ezan, Pharmacokinetic studies of protein drugs: past, present and future,Adv. Drug Deliv. Rev. 65 (8) (2013) 1065e1073.

[40] S.J. Kim, H.H. Kwak, S.Y. Cho, Y.B. Sohn, S.W. Park, R. Huh, J. Kim, A.R. Ko,D.K. Jin, Pharmacokinetics, pharmacodynamics, and efficacy of a novel long-acting human growth hormone: Fc fusion protein, Mol. Pharm. 12 (2015)3759e3765.