Granulocyte colony-stimulating factor (GCSF) fused with Fc Domain produced from E. coli is less effective than Polyethylene Glycol-conjugated GCSF

Human granulocyte colony-stimulating factor (GCSF) is a well-known cytokine for neutropenia treatment. However, daily injections are required due to the short circulating half-life of the protein. To overcome this bottleneck, we fused GCSF with the Fc domain of IgG1 at the C terminus (GCSF-Fc) and with the maltose binding protein (MBP) tag at the N-terminus and expressed it as a soluble protein in the cytoplasm of E. coli. We also conjugated PEG aldehyde to GCSF to make PEG-GCSF. The bioactivities of GCSF-Fc and PEG-GCSF were similar to native GCSF using the mouse M-NFS-60 myelogenous leukemia cell line. The EC50 dose-response curves for GCSF, GCSF-Fc and PEG-GCSF were 37 ± 12 pM, 75 ± 13.5 pM and 46 ± 5.5 pM, respectively. When the proteins were injected into neutropenic rats, the group injected with PEG-GCSF showed the highest and fastest recovery of neutrophils, followed by GCSF-Fc and GCSF. ELISA assay revealed the PEG-GCSF had the longest plasma circulation (>72 h), followed by GCSF-Fc (>48 h) and GCSF (~24 h), which is consistent with the in vivo activities of the proteins. In summary, the GCSF-Fc purified from E. coli was not as efficient as PEG-GCSF in treating neutropenic rats.

back into the bloodstream at physiological pH 14 . Consequently, the Fc region of IgG has been applied to extend the half-life of many target proteins. An example of this is Etanercept that is a fusion of the TNF receptor and Fc region that is approved for autoimmune disease 15 .
Previous studies showed the fusion of GCSF/IgG1-Fc which was produced in mammalian cells has a 5-to 8-fold longer half-life than that of GCSF 8,9 . In this study, we conjugated 20 kDa PEG to the N-terminus of purified GCSF by reductive alkylation method at pH 6. We also generated a fusion of the IgG1 Fc domain to the C-terminus of GCSF and purified the chimeric protein. To evaluate the in vitro bioactivity of PEG-GCSF and GCSF-Fc, the candidate proteins were cultured with mouse myeloblast M-NFS-60 cells. In addition, the pharmacokinetic and granulocytic recovery capabilities of GCSF, PEG-GCSF and GCSF-Fc were also compared in a neutropenic rat model.

Construction of GCSF-Fc plasmid and expression test.
One hundred seventy-five amino acids of the human GCSF was optimized for the expression in E. coli 16 . IgG1 Fc domain was inserted into pDest-HMGWA destination vector 17 at the position right after attB2 sequences using Overlap cloner TM DNA cloning kit. Then, optimized GCSF gene was subcloned into created vector by LR recombination cloning. A tobacco etch virus (TEV) protease recognition site (TEVrs; ENLYFQ V G) was inserted between the tag and GCSF. The sequences of the clones were confirmed by DNA sequencing (Macrogen, Daejeon, Korea).
The expression plasmids were transformed into E. coli Shuffle T7 Express for expression. The transformed cells were grown in Luria Bertani (LB) broth containing 50 µg/mL ampicillin at 37°C in 200 rpm shaking incubator. When OD 600 reached around 0.5, 0.5 mM isopropyl-b-D-thiogalactoside (IPTG) was added and cells were continued to incubate at different temperatures and time, 37°C for 3 h, 30°C for 5 h, 25°C for 8 h and 18°C for 16 h to induce protein expression.
Purification of GCSF-Fc. The culture was scaled up to 1 liter with supplementary IPTG (0.5 mM) at 30°C for 5 h. The collected cells were resuspended into 100 mL buffer containing 20 mM Tris, 5% glycerol (v/v), pH 8. The cells were homogenized by sonication using ultrasonic cell disruptor JY99-IIDN (Ningbo Scientz Biotechnology, Guangdong, China) on ice at 1,000 W for 40 cycles for 10 seconds, followed by intervals of 50 seconds for cooling. After homogenization, the supernatant was collected by centrifugation for 20 min at 27,000 g. Five mL equilibrated protein A resin and TEV protease were added into supernatant. The whole solution was incubated at 4°C for 16 h with gently shaking. The GCSF-Fc was eluted by 0.1 M Glycine pH 3.5 and 1 M Tris pH 9.5 was used to neutralize the eluate. The GCSF-Fc dimer was separated from higher multimers by Superdex 200 26/60 gel filtration column in Tris buffer containing 200 mM NaCl. Based on the chromatogram, the collected GCSF-Fc was analyzed by 10% SDS-PAGE. The endotoxin in final products was removed by Triton TM X-114 and measured by LAL assay.
Mass analysis of GCSF-Fc. The half of the protein was reduced with 10 mM DTT for 30 min at 60 °C and alkylated with 55 mM IAA for 30 min in the dark, following by digested with trypsin. The digested peptides were resuspended in 0.1% TFA and loaded onto Zorbax 300SB-C18 75 μm i.d. × 15 cm column via a trap column (Zorbax 300SB-C18 300 μm i.d. × 5 mm column). Peptides were then separated in an acetonitrile gradient (buffer A -0.1% formic acid; buffer B -100% acetonitrile and 0.1% formic acid) at a flow rate of 200 nl/min with an Agilent 1100 nanoHPLC system (Agilent, USA) and applied on-line to an Q Star XL mass spectrometer (AB Sciex, USA). The gradient was increased from 5% to 40% solution B over 110 min, followed by an increase to 95% B over 1 min, and then 95% B isocratic for 15 min. MS spectra were collected in full scan mode (350-1400 Da) followed by three MS/MS scans of the most intense ions.
Plasmid construction and purification of GCSF from PDIb´a´-GCSF that used enterokinase protease for tag removal. The expression vector containing PDIb´a´-GCSF with enterokinase recognition sites (EKrs), DDDDK between tag and target was constructed following the protocol of previous publication 16 . Then the expression vector was expressed inside E. coli BL21 under the supplementary of 0.5 mM IPTG at 18°C for 16 h.
This GCSF was purified following the preceding report 16 , except using enterokinase protease light chain (EKL) instead of TEV protease to separate tag from GCSF. The EKL was produced using E. coli by our laboratory (manuscript in preparation). The purity of purified protein was confirmed on SDS-PAGE.
SCIenTIfIC RepORTS | 7: 6480 | DOI:10.1038/s41598-017-06726-7 Conjugation and purification of PEG-GCSF. Twenty mg recombinant human GCSF was prepared following the protocol of the previous report 16 . Twenty kDa PEG was conjugated to N-terminus of GCSF by the reductive alkylation method at low pH 18 . Briefly, the buffer of 4 mL GCSF at concentration of 5 mg/mL was changed into 0.1 M sodium phosphate buffer pH 6.0 by dialysis. This solution was added to a vial containing 100 mg 20 kDa mPEG-CHO. When PEG was dissolved completely, 82 µL of 1 M sodium cyanoborohydride was added to the reaction mixture. The reaction mixture was gently stirred in dark at 4°C for 16 h. Then sample was diluted 20 times by 20 mM sodium phosphate buffer pH 4.0 and loaded onto 5 mL HiTrap SP HP cation exchange column pre-equilibrated with the same buffer. After washing with 10 column-volume of the same buffer, a linear gradient elution from 0 to 1 M NaCl of the sodium phosphate buffer pH 4 was applied to elute PEG-GCSF. The fractions were analyzed by SDS-PAGE.
Cell proliferation assay. The M-NFS-60 cells were grown in RPMI-1640 medium containing 10% fetal bovine serum, 1X penicillin and streptomycin, and 0.05 mM β-mercaptoethanol at 37°C in a humidified atmosphere containing 5% CO 2 . The cells (5 × 10 3 cells/well) were seeded into 96-well plate containing growth medium. Simultaneously, different concentrations of GCSF, PEG-GCSF, and GCSF-Fc (0.001, 0.01, 0.1, 1, 10, 100 ng/mL) were added to each well in a final volume of 100 µL. Phosphate buffered saline (PBS) was used as control sample. After 72 h of incubation, 15 µL of 5 mg/mL MTT was added to each well and the cells were kept incubating at 37°C for further 4 h in dark. After draining the solutions, 100 µL of DMSO was added to each well to completely solubilize the formed aggregates. The optical density of the solution was measured at 570 nm using an ELISA reader.
Protein dose-response proliferation was analyzed using following equation and Microsoft Excel software.
Hs 50 The abbreviation is as follows: Re, response of the cells; Bl, baseline at low concentration; Max, the maximum response; conc, concentration of the protein; and Hs, Hill coefficient of stimulation.

In vivo experiment on neutropenic rats. The research protocol was approved by the Institutional Animal
Care and Use Committee of the Asan Institute for Life Science, and mice were maintained in accordance with the Institutional Animal Care and Use Committee guidelines of the Asan Institute for Life Science. All experiments were performed in accordance with relevant guidelines and regulations. Seven weeks 220 g Sprague-Dawley rats were acclimatised for at least 7 days before experiments. Rats were randomly separated into 7 groups with 5 rats in each group. Group 1 was used as control group. Six other groups received an intraperitoneal injection of 100 mg/kg of CPA on day 0 to induce neutropenia. On day 1, these 6 groups received subcutaneous injections of phosphate buffered saline (PBS) (group 2), 100 µg/kg GCSF (group 3), 100 µg/kg PEG-GCSF (group 4), 50 µg/ kg GCSF-Fc (group 5), 100 µg/kg GCSF-Fc (group 6) and 300 µg/kg (group 7). More than 3 mL of blood were collected from day 0 to day 12 with heparinized syringe. 200 µl sample from the collected blood were analysed for complete blood count (CBC) using ADVIA 2120i Hematology System (Siemens Healthineers, Erlangen, Germany). The remaining bloods were centrifuged at 1,500 g for 15 min to collect plasma for ELISA assay. The GCSF serum levels on days 2, 3, 4 from group 3, 4 and 7 were quantitated using Quantikine human GCSF ELISA kits.
Statistical analysis. All data are presented as the mean ± standard error (SE) of n ≥ 3 of 2 independent experiments. Statistical analyses were performed using the SPSS statistical software program (SPSS, version 18.0, Chicago, IL). A Student's t-test was used to determine the statistical significance of group means. All tests were two-sided and p-values less than 0.05 were considered statistically significant. Fig. 1A containing GCSF with an MBP tag at the N-terminus and IgG1 Fc domain at the C-terminus was generated for use. The expression kinetics of MBP-GCSF-Fc and PDIb´a´-GCSF-Fc in E. coli strain BL21 with 0.5 mM IPTG at 18°C for 16 h is shown in Supplementary Fig. S1. During the isolation process the GCSF-Fc after TEV cleavage formed multimers and aggregates hampering further purification (data not shown). Therefore, MBP-GCSF-Fc and PDIb´a´-GCSF-Fc were expressed in Shuffle T7 Express. In our previous study, MBP-GCSF or PDIb´a´-GCSF in BL21(DE3) was expressed well and highly soluble when incubated with 1 mM IPTG at 18°C for 16 h 16 . However, under the conditions of the present study, very few PDIb´a´-GCSF-Fc were expressed (data not shown) and MBP-GCSF-Fc was expressed in low amounts (Fig. 1B). To increase the expression level, several conditions of induction temperatures, 18°C, 25°C, 30°C and 37°C were tested for the expression of MBP-GCSF-Fc. As a result, the expression of protein at 30°C for 5 hrs showed the highest expression and solubility efficiency which were approximately 26.7% and 74.5% respectively (Fig. 1B).

Expression of MBP-GCSF-Fc in E. coli and its purification. The vector shown in
After sonication, MBP-GCSF-Fc was present in the supernatant fraction because this protein was soluble after induction with IPTG. When the fusion protein was mixed with TEV protease and protein A resin beads simultaneously, GCSF-Fc was separated from the fusion protein and bound protein A resin. This target protein was eluted off of the beads by low pH buffer (0.1 M Glycine buffer pH 3.5). To prevent the aggregation of protein at this low pH, 1 M Tris buffer pH 9.5 with at a ratio of 1:10 was used to neutralize the eluate. Despite this, some proteins formed higher multimers and to remove these multimers, a gel filtration column was applied. As a result, approximately 90% of GCSF-Fc multimer was removed from GCSF-Fc dimers (Fig. 2C and D). Based on SDS-PAGE analysis, the GCSF-Fc dimer was highly pure (more than 95%) and endotoxin level in the product was less than 1 EU/µg after endotoxin removal by Triton TM X-114. Finally, approximately 1 mg GCSF-Fc was obtained from 1 L of E. coli cultured cells. PEGylation for GCSF. PEG was conjugated to the N-terminus of GCSF using a reductive alkylation method with 20 kDa mPEG aldehyde at low pH. When the pH was lowered from 8 to 5, GCSF became unstable and formed aggregates (data not shown). To overcome this problem, we lowered the reaction pH to 6 and the protein was stable. After incubation of GCSF with mPEG-CHO and sodium cyanoborohydride, approximately 50% of PEG was conjugated to GCSF (Fig. 3C). The PEG-GCSF was then separated from non-reacted GCSF and mPEG-CHO by a cation exchange column using a sodium phosphate buffer of pH 4, suggesting that the PEGylation increased the stability of GCSF at low pH. Two major peaks were observed (Fig. 3C), and SDS-PAGE analysis of the elution peaks showed that the earlier peak had a single band of approximately 50 kDa that corresponds to PEG-GCSF (Fig. 3B). The non-reacted mPEG-CHO did not bind to the column. Overall, 8 mg of PEG-GCSF was obtained from a starting yield of 20 mg GCSF.
Since the PEGylation efficiency was as low as 50% and the GCSF has one remaining N-terminal glycine residue after TEV cleavage, we purified non-Gly GCSF from PDIb´a´-GCSF cleaved by enterokinase light chain (EKL) protease ( Supplementary Fig. S5). This GCSF isolate was subsequently conjugated with mPEG-CHO using the same method as detailed above. As a result, more than 90% of GCSF was conjugated to PEG based on size exclusion chromatography (SEC)-HPLC ( Supplementary Fig. S6).
In vitro activity of proteins. The biological activity of the purified GCSF-Fc and PEG-GCSF was measured using a proliferation assay with the mouse myelogenous leukemia M-NFS-60 cell line. After incubation of cells with purified GCSF, GCSF-Fc and PEG-GCSF at different concentration for 3 days, the cell number and viability was measured using an MTT assay. These data showed that the number of cells increased dramatically after incubation with all three proteins and followed a sigmoidal dose-response curves (Fig. 4). The EC 50 of GCSF, GCSF-Fc and PEG-GCSF were highly similar with a range of: 37 ± 12 pM, 75 ± 13.5 pM and 46 ± 5.5 pM, respectively. The hill coefficient was approximately 1.5 for all cases, suggesting that fusion with Fc region or conjugation with PEG did not affect the biological activity of GCSF (Fig. 4).

In vivo bioactivity of proteins.
To evaluate the effects of the proteins in an in vivo setting, we injected rats with 100 mg/kg of cyclophosphamide (CPA) reagent on day 0 with follow on GCSF candidate protein injections on day 1. After CPA application, the level of neutrophils and total white blood cells (WBC) in the peripheral blood was sharply reduced and then began to rise after GCSF protein infusion (day 2) (Fig. 5A and B). On day 2, the amounts of neutrophils and WBCs in rats injected with PEG-GCSF and GCSF-Fc were significantly higher than those in GCSF or the PBS treated group (Fig. 5C and D). Animals treated with PEG-GCSF showed the quickest recovery, followed by GCSF-Fc and GCSF, respectively. Of note, on day 6, neutrophils and WBCs of PEG-GCSF and GCSF-Fc treated groups were significantly higher than the corresponding values of GCSF treated group (Fig. 5C and D). The activity of GCSF-Fc compared to GCSF started to diminish on day 7 when the numbers of neutrophil and WBC in GCSF-Fc and GCSF groups equilibrated. The level of WBCs of the GCSF group was similar while the neutrophil levels were relatively higher than that of the PBS group. According to these data, both PEG-GCSF and GCSF-Fc were effective for resolving neutropenia with a single injection, but GCSF-Fc was less effective compared to PEG-GCSF (Fig. 5). Importantly, there was no obvious change in the number of red blood cells and platelets during the course of the experiments (data not shown).
Twenty-four, 48 and 72 hours after protein injection, the duration of proteins in the plasma was measured using a GCSF ELISA kit. Figure 6 shows that the elimination of GCSF was more rapid than GCSF-Fc and PEG-GCSF. Twenty-four hours after injection, GCSF was completely cleared compared to more than 48 h of GCSF-Fc and more than 72 h of PEG-GCSF. These data show that conjugation with PEG or fusion with Fc domain can significantly increase circulating levels of GCSF due to a prolonged half-life.

Discussion
In this study, we described in detail the soluble prokaryotic expression and simple purification of GCSF-Fc by fusion technology with an MBP tag and conventional chromatography. We demonstrate that IgG1 Fc and PEG do not appreciably affect GCSF activity as evidenced by the similarity in in vitro bioactivity. Our in vivo experiments in neutropenic rats revealed that PEG-GCSF was more effective than GCSF-Fc in neutrophil and WBC recovery as well as stability of the protein in plasma.
Recombinant proteins commonly form inclusion bodies when expressed in the cytoplasm of E. coli. Fusion technology with tags such as MBP, N-utilization substance protein A, or protein disulfide bond isomerase has been shown to be effective for enhancing protein expression as well as solubility 16,19 . Because of its relatively small   size, easy purification using MBP chromatography, and efficacy in supporting the expression as well as proper folding of target proteins in the reducing environment of E. coli cytoplasm, MBP is largely the tag of choice. For the expression of MBP-tagged GCSF-Fc, induction by 0.5 mM IPTG at 30°C for 5 h was the optimal condition for expression (Fig. 1B). Nearly 70% of this fusion protein was soluble and this maximized the purification process with only the protein A resin ( Fig. 2A and B). When the Fc domain alone was expressed in E. coli BL21(DE3), multimer and soluble aggregates were formed (data not shown), which may be due to poor disulfide bond formation. To overcome this, MBP-tagged GCSF-Fc was expressed in the Shuffle T7 Express system to support disulfide bond formation. Shuffle T7 Express is an engineered E. coli B strain that creates an oxidizing cytoplasm On day 0, 5 rats in each group but the control group received 100 mg/kg CPA to induce neutropenia. On day 1, neutropenic groups received injections of PBS, GCSF (100 µg/ kg), PEG-GCSF (100 µg/kg), or GCSF-Fc (50 µg/kg, 100 µg/kg and 300 µg/kg). Blood samples from the rat groups over 12 days were sent for complete blood count (CBC) analysis. Data are means ± SE of 5 rats/group. Statistical significance compared to GCSF treatment group: *p < 0.05, **p < 0.01. Figure 6. Protein level in plasma after administration of GCSF-Fc, PEG-GCSF and GCSF. The measurements were performed at 24 h, 48 h and 72 h after protein injection. Three groups of rats were utilized for plasma protein measurement, GCSF (100 µg/kg), PEG-GCSF (100 µg/kg) and GCSF-Fc (300 µg/kg). Data are given as means of ± SE of 3 rats/group. The data points at Day 0 were the amount of supplementary proteins, assuming that all proteins were diluted completely in rats' body plasma. environment by deleting genes of glutaredoxin reductase and thioredoxin reductase combined with a mutation in the peroxiredoxin enzyme. Additionally, this strain continuously expresses the disulfide bond isomerase DsbC in the cytoplasm to promote the correction of mis-oxidized proteins into their correct form 20 . Under these defined conditions, the homodimer of GCSF-Fc protein was successfully purified (Fig. 2D).
PEGylation has several advantages for in vitro protein engineering including: protection the target protein by increasing stability, reducing renal clearance, and mitigating toxicity 21 . To date, numerous strategies for GCSF PEGylation have been applied, including conjugating PEG at the N-terminus, coupling PEG at cysteine 17 or lysine 41, and PEGylation in organic solvent 18,[22][23][24] . Covalent attachment of 20 kDa PEG at the N-terminus residue of GCSF by reductive alkyl conjugation using mPEG-CHO showed the best conjugation efficacy while remaining biologically active 18,25 . These previous studies performed the reaction at pH 5 18, 25 therefore, we also lowered the pH to 5 and found that the purified GCSF was completely aggregated (data not shown). The calculated pI of GCSF was 5.65 and when we reduced pH from 8 to 5, the protein began to aggregate at pH of 5.65. Therefore, we performed the conjugation at pH 6 that allowed the PEGylation reaction to proceed. However, the efficiency was low compared to the high PEGylation yield of previous reports which reached 90% or higher 18,25 . Our GCSF was obtained from MBP-GCSF using TEV protease for tag removal, resulting in a remaining glycine residue at the N-terminus (Gly-GCSF). To test whether the low PEGylation efficacy was caused by this N-terminal glycine, we purified non-Gly GCSF from PDIb´a´-GCSF fusion construct using EKL protease (Supplementary Fig. S5). The conjugation efficiency of this non-Gly GCSF was more than 90% ( Supplementary Fig. S6), indicating that the remaining glycine from TEV protease cleavage at the N-terminus significantly reduced the conjugation efficiency of N-terminal, site-specific, mono-PEGylation of GCSF.
Upon cation exchange chromatography, the PEG-GCSF was eluted earlier than the unreacted GCSF, while neutral unreacted PEG did not bind to the column (Fig. 3C). This may be by cause of the PEG shields the surface charges of GCSF 21 . On SDS-PAGE, PEG-GCSF with high purity appeared at size of approximately 50 kDa which was higher than the reported molecular weight (Fig. 3B), and is likely caused by PEGs capacity to occupy a larger volume in an aqueous environment 25 .
In vitro proliferation experiment demonstrated that the conjugation of PEG at the N-terminus of GCSF did not negatively impact its biological activity (Fig. 4). However, others have reported a decrease of in vitro bioactivity when the protein was conjugated to PEG [26][27][28] . GCSF-Fc showed lower in vitro proliferation activity compared to GCSF, which is consistent with a previous report 8 .
According to previous pharmacokinetics experiments, GCSF-Fc produced in mammalian cells has 5-to 8times longer half-life than GCSF 8,9 . Despite this advantage, the eukaryotic production process can be both complex and time consuming. Our study was designed to produce the therapeutic protein fused with the IgG1 Fc region in E. coli and to test the half-life and the effectiveness in vitro and in vivo. Our results show that GCSF-Fc can confer added benefits (e.g. stability) compared to native GCSF. While GCSF-Fc was more effective than GCSF, it was less effective than PEG-GCSF in our in vivo experiments. Our results are disparate from a previous report 9 that showed GCSF-Fc had similar activity to PEG-GCSF in vivo. The lessened effect of GCSF-Fc produced from E. coli may be due to the reduced stability of non-glycosylated Fc domain produced in E. coli compared to glycosylated Fc domain 29,30 . Or, the non-glycosylated Fc from E. coli might have lower affinity to the FcRn. In general, the fusion of GCSF-Fc was produced effectively in E. coli and it had an improved circulating half-life and hematopoietic properties compared to GCSF. Our study strongly suggests that, in order to be developed as a biodrug produced from a prokaryote in the future, the GCSF-Fc, especially the Fc domain, must be mutated or modified for higher stability or higher affinity to FcRn. Collectively, these data are important for cellular engineering of bioactive GCSF.