Modified human glucagon-like peptide-1 (GLP-1) produced in E. coli has a long-acting therapeutic effect in type 2 diabetic mice

Glucagon-like peptide 1 (GLP-1) is a very potent insulinotropic hormone secreted into the blood stream after eating. Thus, it has potential to be used in therapeutic treatment of diabetes. The half-life of GLP-1, however, is very short due to its rapid cleavage by dipeptidyl peptidase IV (DPP-IV). This presents a great challenge if it is to be used as a therapeutic drug. GLP-1, like many other small peptides, is commonly produced through chemical synthesis, but is limited by cost and product quantity. In order to overcome these problems, a sequence encoding a six codon-optimized tandem repeats of modified GLP-1 was constructed and expressed in the E. coli to produce a protease-resistant protein, 6×mGLP-1. The purified recombinant 6×mGLP-1, with a yield of approximately 20 mg/L, could be digested with trypsin to obtain single peptides. The single mGLP-1 peptides significantly stimulated the proliferation of a mouse pancreatic β cell line, MIN6. The recombinant peptide also greatly improved the oral glucose tolerance test of mice, exerted a positive glucoregulatory effect, and most notably had a glucose lowering effect for as long as 16.7 hours in mice altered to create a type 2 diabetic condition and exerted a positive glucoregulatory effect in db/db mice. These results indicate that recombinant 6×mGLP-1 has great potential to be used as an effective and cost-efficient drug for the treatment of type 2 diabetes.


Introduction
Diabetes is a major chronic systemic metabolic disease resulting from the dysfunction of carbohydrate metabolism due to a relative deficiency of insulin [1]. The prevalence of this disease in humans has been estimated to be 415 million worldwide and is predicted to increase to 642 million by 2040 [2]. The majority of diabetic patients have Type 2 Diabetes Mellitus, which is characterized by a combination of interrelated metabolic disorders, including continuous hyperglycemia, peripheral insulin resistance, and decreased β cell function, loss or dedifferentiation [3]. Although various treatments, such as diet, exercise, antidiabetic drugs, and subcutaneous insulin injection are available, no cure is currently available for type 2 diabetes. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 as an attempt to prolong the half-life of the recombinant protein in the bloodstream. The modified DNA construct of GLP-1s (mGLP-1) was constructed to contain six tandem repeats of mGLP-1 genes, plus a His-tag at the C-terminus, and was named 6×mGLP-1 (Fig 1). After optimizing the codon sequence for expression in E. coli, the 6×mGLP-1 gene sequence was synthesized by GenScript Co., Ltd (Nanjing, China) into pUC57-6×mGLP-1 plasmid.
Large scale production and purification of 6×mGLP-1 A single-colony transetta (DE3) strain with pEASY-E2-6×mGLP-1 plasmid was cultured in 500 mL LB culture with 50 μg/mL ampicillin at 37˚C. The cultures were then induced by the addition of 0.3 mmol/L IPTG for 8 hours. The cells were then harvested by centrifugation at 12,000 rpm for 5 min at 4˚C and pelleted cells were disrupted by ultra-sonication after re-suspension in PBS buffer (137 mmol/L NaCl, 2.7 mmol/L KCl, 10 mmol/L Na 2 HPO 4 , 2 mmol/L KH 2 PO 4 , pH 7.4). After centrifugation at 15000 rpm for 30 min at 4˚C, 6×mGLP-1-His fusion protein, in the form of inclusion bodies, was collected from the crude cell lysate, dissolved in 8 mol/L urea solution, and further purified though a His-tagged affinity chromatography column of nickel-nitrilotriacetic acid (Ni-NTA) bind resin (Qiagen, Hilden, Germany). The recombinant protein was then subjected to dialysis for renaturation, and then lyophilized. Samples of the purified 6×mGLP-1-His fusion protein were kept at 4˚C for further use.
Reversed-phase high performance liquid chromatography (HPLC) was used to confirm the purity of 6×mGLP-1, and monitor the digestion of 6×mGLP-1 by trypsin. The HPLC assay was performed on C 18 column (250×4.6 mm, I.D.S-5 μm, 12 nm, YMC-Pack ODS-A, Germany) using LC-2010A/C chromatographic instrument (Shimadzu, Japan), at ambient temperature with a mobile phase of acetonitrile/water (90: 10, v/v) at a flow rate of 1.0 mL/min. Trifluoroacetic acid (TFA, 0.1%) was added into the water. Gradient assay was lasted for 20 min. All samples were detected at 280 nm. The blank, PBS (pH 7.4), trypsin and commercial GLP-1 were used as controls.

Biological activity of the recombinant 6×mGLP-1 in vitro
Based on the method described by Brandsma et al. [18], with minor modifications, a mouse pancreatic tumor β cell line (MIN6) (iCell Bioscience Inc, Shanghai, China) was cultured and maintained in Roswell Park Memorial Institute (RPMI) 1640 medium containing 10% (v/v) fetal bovine serum, 100 U/mL penicillin, 0.1 mg/mL streptomycin and 50 μmol/L 2-mercaptoethanol at 37˚C in an incubator with 5% CO 2 and 95% humidified air. MIN6 cells were cultured to a density of 6×10 4 cells/mL, then seeded into a 96-well flat plate (200 μL per well) and re-cultured for 24 hours for use in determining the effect of recombinant 6×mGLP-1 on β cell proliferation. 10 μg/mL commercial GLP-1 standard or 6×mGLP-1 digested by trypsin (0.25%) were then added to the cell cultures, with 0.25% (w/v) trypsin-treated and Dulbecco's phosphate buffered saline (DPBS) cultures serving as negative controls. The cultures were subsequently incubated for an additional 48 hours. Subsequently, 100 μL 0.5% (w/v) 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) per well was added to each sample and further incubated for another 4 hours on a shaker at 37˚C, after which 100 μL dimethylsulfoxide (DMSO) was added to each well. The absorbance at 490 nm was then measured in each well in a microplate reader. Six biological replicates were used for each sample treatment and the experiment was repeated twice.

Oral glucose tolerance test
The oral glucose tolerance test (OGTT) was performed according to Kong et al. [19] with minor modifications. Sixteen, 8-week-old male KM mice, weighing 46.06 ± 1.66 g, were fasted overnight and divided into control and test groups (n = 8 per group). Each group was administered 2 g/kg body weight glucose into their stomach via oral administration with a proper gavage needle. The test group was simultaneously administrated, via intraperitoneal injection, a 1.0 mL dose of 0.1 mg/mL 6×mGLP-1 dissolved in physiological saline. The control group was administered an equivalent amount of physiological saline. Blood samples were taken from the tail vein of each mouse at 0, 10, 30, 60 and 120 min, and blood glucose levels were measured with a blood-glucose meter (Sinocare, Changsha, China).
Blood glucose-lowering effect of recombinant 6×mGLP-1 on type 2 diabetic mice models Fifteen KM mice were injected intraperitoneally with a 30 mg/kg body weight dose of streptozotocin (STZ, Sigma-Aldrich, St. Louis, MO, United States) dissolved in citrate buffer (0.1 mol/L citric acid, 0.1 mol/L sodium citrate, pH 4.2) for 2~4 consecutive days to selectively damage pancreatic β cells. In addition, they were also administered a high-fat diet (crude protein content !22%, crude fat !4.5%, and about 4000 kcal/kg calories) for one month in order to establish a stable hyperglycemic type 2 diabetic condition. Another fifteen age-, sex-and body weight-matched KM mice were given equivalent amount citrate buffer and fed with standard food as control. Mice were housed in cages with 5 mice/cage and given free access to high-fat food and water (23 ± 2˚C). A 45 ± 5% relative humidity was maintained and the mice were subjected to a 12/12 h light/ dark cycle. After approximately one month, when blood glucose levels in the STZ-induced mice were as high as 20.80 ± 2.21 mmol/L, OGTT was performed as described above (n = 15).
Fourteen STZ-induced type 2 diabetic mice were selected and randomly divided into 2 groups (n = 7 per group). The mice in both groups were fasted until the end of the experiment but given a normal water supply. The test group was injected intraperitoneally in the lower, left portion of the abdominal cavity with 0.1 mg/kg 6×mGLP-1 (1.0 mL) dissolved in physiological saline while the control group was injected with an equivalent dose of physiological saline. Blood samples were obtained from the tail vein of each mouse and glucose levels at various time points were measured with a blood glucose monitor.
In another experiment, twelve db/db mice of 9 week-old bought from Beijing Vital River Laboratory Animal Technology Co., Ltd. were given free access to standard food and water (23 ± 2˚C). A 45 ± 5% relative humidity was maintained and the mice were subjected to a 12/12 h light/dark cycle. One week later, they were divided into 2 groups randomly. In one group, mice were given 0.1 mg/mL 6×mGLP-1 (1.0 mL) via oral gavage, while mice in the other group were given the same volume of physiological saline (n = 6). Mice were all fasted but given free access to water. Blood samples were collected from the tail tip at various time points and measured using a glucose monitor.

Statistical analysis
The presented data are means ± standard deviation (SD). Significant differences between control and test groups were determined by a Student's t-test or one-way analysis of variance (ANOVA) followed by post-hoc hypothesis tests, where appropriate, using SPSS version 19.0 software. In all statistical analyses, a P value < 0.05 was considered to indicate a statistically significant difference.

Results and discussion
Protease resistant design of 6×mGLP-1 In order to overcome the short half-life of GLP-1 in the bloodstream, studies have mainly focused on identifying and developing DPP-IV inhibitors [20,21], modifying the structure of GLP-1, or searching for structural analogues that are more resistant to DPPIV cleavage [22]. Based on these studies, and the functional form of native GLP-1(7-36), a modified human GLP-1 (mGLP-1) was designed and constructed by site-specific mutations that would have resistance to both DPP-IV and trypsin cleavage. This included replacing the enzyme-susceptible Ala at 8 position and Lys at the 26 and 34 position. Ala 8 was mutated to code for Ser 8 or Gly 8 , and Lys 26, 34 were substituted with Gln 26 , Asn 34 or Gsn 26 , Asp 34 . These structural modifications have been confirmed to exert complete resistance to DPP-IV induced cleavage of GLP-1 and trypsin degradation [23][24][25][26]. In the present study, these point-modifications were placed into a GLP-1 construct consisting of six mGLP-1 coding sequences present as a tandem repeat (6×mGLP-1) in order to obtain a longer half-life once the mGLP-1 was present in the bloodstream. Furthermore, two Cys residues were added before the last amino acid, arginine (Arg), in order to enhance the stability of mGLP-1 in the bloodstream by the formation of disulfide bonds (Fig 1A).
Additionally, each amino acid sequence of mGLP-1 started with a His and ended with Arg, and an extra Arg was added at the N-terminal of 6×mGLP-1 (Fig 1B), so that the tandemly repeated six mGLP-1 construct would be cleaved by trypsin into six separate mGLP-1monomers without any amino acids from the expression vector or His-tag. If properly functioning, this design would reduce production costs and potentially overcome the present limitation on the clinical use of GLP-1.

Expression and purification of recombinant 6×mGLP-1
The 6×mGLP-1 codon sequence was optimized for expression in E.coli strain, transetta (DE3). As shown in the SDS-PAGE (Fig 2A) and immunoblotting (western blotting, Fig 2B), the target protein was expressed abundantly in the form of an insoluble inclusion body. The molecular weight (MW) of designed 6×mGLP-1 was slightly more than 20 kDa (Fig 2), which is consistent with its predicted size of approximately 22 kDa. High-quality 6×mGLP-1 recombinant protein was recovered with through a series of proficient steps, including denaturation and refolding in Ni-NTA affinity chromagraphy and during the process of dialysis. SDS-PAGE analysis of the reconstituted sample of 6×mGLP-1 indicated that a large amount of highly pure recombinant 6×mGLP-1 protein was obtained (Fig 2C). The purification result of HPLC analysis also showed that the purity of 6×mGLP-1 was as high as 100%, and the retention time of the peak was 11.393 min (Fig 2F). The estimated yield was 20 mg/L.

Digestion analysis of 6×mGLP-1 by protease in vitro
The purified recombinant 6×mGLP-1 formed dimers (possibly because of additional disulfide bond) that could be cut by trypsin into single stable mGLP-1s with physiological activity as expected was observed to be more stable against to trypsin enzyme degradation. Consistent with Jomori and his colleagues' findings [14], the individual mGLP-1 monomers, however, exhibited strong resistance to trypsin digestion, which was directly verified by conducting an in vitro protease digestion analysis assay. Coomassie-stained SDS-PAGE gels indicated that rapid in vitro cleavage (within 25 min) of the purified recombinant 6×mGLP-1 by trypsin resulted in small fragments (Fig 2D). By 120 min, very little intact 6×mGLP-1 was present ( Fig  2D). These results demonstrate that 6×mGLP-1 can be cleaved by trypsin. Furthermore, although the 6×mGLP-1 polypeptide could be cut into single mGLP-1 fragments by trypsin in vitro, the mGLP-1 monomer (Fig 2E, yellow arrow) was resistant to any further digestion by trypsin. As shown in Fig 2D, 6×mGLP-1 appeared to form a dimer with a molecular weight above 40 kDa (red arrow), which was corresponding to the theoretical molecular weight of 44 kDa for the dimer. The dimer appeared more stable than the monomer form of 6×mGLP-1, because it only began to decrease 45 min after the trypsin digestion. Aberle et al. [27], reported that inter-disulfide bonds represent a means to stabilize peptides or protein. Therefore, we speculated that the formation of the dimer was due to the additional Cys residues that were inserted into 6×mGLP-1. This insertion was expected to prolong the half-life of 6×mGLP-1 in the bloodstream and allow for the glucose lowering effect to persist in the bloodstream for a longer period of time [27,28].
HPLC was also used to monitor the digestion of 6×mGLP-1 by trypsin, as shown in Fig 2F. For purified 6×mGLP-1 only, the retention time of peak was at 11.393 min; however, the retention time of peak of 6×mGLP-1 incubated with trypsin for 2 hours at 11.393 min was missing, and one major peak was at 10.098 min which was consistent with that of GLP-1 commercial standard, 10.402 min. We speculated that 6×mGLP-1 was hydrolyzed to monomer GLP-1s when it was incubated with trypsin, which were more stable against to trypsin enzyme degradation. Trypsin was also activated at 37˚C in PBS (pH 7.4). Dynamic changes occurred both in trypsin solution and the reaction solution of trypsin and 6×mGLP-1, corresponding to the results of HPLC analysis. There were two other major peaks at 5.751 and 8.676 min were observed in the reaction solution. Presumably some other forms of products were produced from 6×mGLP-1 and trypsin incubation.

Promoting β cell proliferation
One of the main functions of native GLP-1 in humans is to promote the proliferation of pancreatic β cells. Therefore, recombinant 6×mGLP-1 was assessed for this ability by using a mouse pancreatic β cell line, MIN6. The 6×mGLP-1 was digested by trypsin into GLP-1 monomers to exhibit activity. Proliferation of pancreatic β cells was promoted by both 6×mGLP-1 and 6×mGLP-1 digested with trypsin as well as the GLP-1 synthesized chemically as standard. Although 0.25% (w/v) trypsin had a significant negative effect (P = 0.002) on cell proliferation compared to DPBS-treated cells, relative to trypsin-treated cells, the mixture containing 0.25% (w/v) trypsin and 10 μg/mL 6×mGLP-1 compensated for the negative impact of the trypsin and actually showed a significant promotion of cell proliferation (P = 0.001). The synthesized native human GLP-1 standard exhibited the same extent promotion effect as 6×mGLP-1 (Fig  3). The results do suggest, however, that 6×mGLP-1 as a whole or digested by trypsin in vitro could have a positive effect on the proliferation of mouse pancreatic β cells. This was consistent with Brandsma and his colleagues' studies [18], in which 10×GLP-1 was proved to increase insulin release from MIN6 cells in the presence of 10 mmol/L glucose, although with reduced 10 μg/mL 6×mGLP-1 digested by 0.25% trypsin; GLP-1 standard: chemically synthesized native human GLP-1; Trypsin + GLP-1: 10 μg/mL GLP-1 digested by 0.25% trypsin. Cell density, measured as optical density at A490, was determined 48 h after the addition of the test compounds. The data represent the mean ± standard deviation (n = 6). The experiment was repeated three times and the results were statistically analyzed using one-way ANOVA method following Tukey post-hoc hypothesis test. The symbols a, b, and c refer to the significance level, and different symbol means statistical significance (at 5%, 1%, and 0.1% level) between treatments, or else, no significance.

Amelioration of glucose tolerance in normal KM mice
It is widely accepted that GLP-1 plays potent roles in glucose homeostasis, and similarly our recombinant 6×mGLP-1 was proved to exert a blood glucose-lowering effect on normal KM mice when administered intraperitoneally. OGTT in mice showed that plasma glucose level of the 6×mGLP-1-treated mice was 10.98 ± 1.04 mmol/L at its peak, which was significantly lower (P = 0.005) than the control group which had a peak plasma glucose level of 14.02 ± 1.44 mmol/L (Fig 4A). The difference between the control and test groups of mice in plasma glucose levels persisted even as the glucose levels began to decrease in both groups. At 120 minutes, when the blood glucose level in the 6×mGLP-1 test group returned to normal (5.50 ± 0.50 mmol/L), the control group still had a glucose level of 6.88 ± 1.23 mmol/L (Fig 4A). The difference between the two groups was statistically significant (P = 0.049). Furthermore, the area under the curve (AUC) of 6×mGLP-1-administrated mice was significantly less (P = 0.002) than the AUC of mice injected with physiological saline (Fig 4B). These results demonstrated that purified recombinant 6×mGLP-1 can significantly improve the oral glucose tolerance of mice.
Blood glucose-lowering activity of 6×mGLP-1 in type 2 diabetic mice models STZ-induced type 2 diabetic mice were used to evaluated the biological activity of 6×mGLP-1. The blood glucose concentration of STZ-induced mice fed with a high fat diet was stably 21.03 ± 1.51 mmol/L five weeks after the STZ-induction began. This level was much higher than the glucose levels (8.10 ± 0.88 mmol/L) in mice that were fed a normal diet. The difference between the glucose levels in the STZ-induced and normal mice was highly significant (P = 4.34E-04). STZ-induced mice also showed impaired glucose tolerance compared to mice in control group (Fig 5A). The area under the blood glucose level curve (AUC) of (A) from 0-120 min. This experiment was repeated twice and the results were statistically analyzed using Student's t-test. * and ** represents significance differences between the two groups at 5% and 1% level, respectively. https://doi.org/10.1371/journal.pone.0181939.g004 Characterization of six tandem modified analogue of human GLP-1 (mGLP-1) in E. coli STZ-induced type 2 diabetic mice were fasted but with normal water supply when separately treated with physiological saline and 6×mGLP-1. The plasma glucose level in the control group of mice remained high for at least 400 min, and then began to decrease, possibly due to hunger (at 1000 min with a blood glucose level of 16.44 ± 2.58 mmol/L, comparable with 15.35 ± 1.82 mmol/L when fasted overnight in OGTT, as shown in Fig 5A); while the blood glucose level in the test group decreased gradually over a period of 60-1,000 min (Fig 5B). These data demonstrated that the effect of 6×mGLP-1 occurring over an extended period of time suggested its ability to gradually lower plasma glucose, without the risk of hypoglycemia (5.87 ± 1.31 mmol/L glucose concentration at 1,000 min, nearly 16.7 hours, much longer than our previous monomer mGLP-1 [29]). A significant difference (P = 0.026) between the two groups in blood glucose levels developed rapidly within 10 min and the separation lasted throughout the entire experiment. Notably, the level of the significant difference between the test and control groups increased with the duration of the experiment (P<0.05 at 0-60 min, . Blood glucose levels were measured over a 1000-min time period. (C) The area under the blood glucose level curve (AUC) of (B) from 0-400 min. (D) Blood glucose-lowering effect of recombinant 6×mGLP-1 in db/db mice (n = 6). 6×mGLP-1 was administrated into mice via oral gavage. All experiments were repeated twice and the results were statistically analyzed using Student's t-test. *, ** and *** represents significant differences between the control and test groups at the 5%, 1%, and 0.1% level, respectively. https://doi.org/10.1371/journal.pone.0181939.g005 Characterization of six tandem modified analogue of human GLP-1 (mGLP-1) in E. coli P <0.01 at 60-120 min, and P<0.001 at 150-1,000 min). These differences may reflect the gradual breakdown of 6×mGLP-1 into mGLP-1 monomers by endogenous trypsin in the mice, allowing for extended activity of the recombinant protein. The AUC of the control group of mice from 0 to 400 minutes was significantly higher (P = 8.5E-06) than the AUC of the test group (Fig 5C).
db/db mice were also used to evaluated the biological activity of 6×mGLP-1. The blood glucose level of oral 6×mGLP-1-treated mice began to decrease at 90 min, and didn't show any statistical difference until 250min (P = 0.017), with glucose concentration 17.30 ± 4.77 mmol/L. Then it began to rise (Fig 5D). Although the route of giving 6×mGLP-1 via oral gavage made it function later possibly resulted from digestion of 6×mGLP-1 into monomers in the gastrointestinal tract at first before absorption into the blood circulation, and last no longer than 4 hours compared with i.p. injection, it put great confidence in chronic treatment of db/db mice via oral administration next.
GLP-1 was confirmed to exert pleiotropic therapeutic effects on diabetes. GLP-1 analogue therapy has been considered a promising therapeutic option for type 2 diabetes mellitus. Preclinical observations suggested its direct renoprotective effects in the setting of hypertension, diabetic and nondiabetic nephropathy, independent of the glucose-lowering effects [30]. As to GLP-1 receptor agonists, attention has been paid on chronic kidney disease primarily caused by hypertension and diabetes. Hence, potential non-glycemia related effects of 6×mGLP-1 will be studied and evaluated in further studies, especially in diabetic renal impairment.
In summary, we modified GLP-1 to prolong the half-life of GLP-1. Six codon-optimized tandem repeats of modified GLP-1, 6×mGLP-1, was successfully expressed in E.coli. Recombinant 6×mGLP-1 that could be cut by trypsin into single stable mGLP-1s with physiological activity as expected was proved to be more stable against to trypsin enzyme degradation. The use of 6×mGLP-1 lowered glucose and might have an antidiabetic effect and this finding could encourage clinical trials using 6×mGLP-1 in type 2 diabetic patients. Utilization of this protein could provide a potential cost-effective method for GLP-1 therapy and warrants additional research in the future.
Supporting information S1 Table. A