Rational design of a fibroblast growth factor 21-based clinical candidate, LY2405319

Abstract

Fibroblast growth factor 21 is a novel hormonal regulator with the potential to treat a broad variety of metabolic abnormalities, such as type 2 diabetes, obesity, hepatic steatosis, and cardiovascular disease. Human recombinant wild type FGF21 (FGF21) has been shown to ameliorate metabolic disorders in rodents and non-human primates. However, development of FGF21 as a drug is challenging and requires re-engineering of its amino acid sequence to improve protein expression and formulation stability. Here we report the design and characterization of a novel FGF21 variant, LY2405319. To enable the development of a potential drug product with a once-daily dosing profile, in a preserved, multi-use formulation, an additional disulfide bond was introduced in FGF21 through Leu118Cys and Ala134Cys mutations. FGF21 was further optimized by deleting the four N-terminal amino acids, His-Pro-Ile-Pro (HPIP), which was subject to proteolytic cleavage. In addition, to eliminate an O-linked glycosylation site in yeast a Ser167Ala mutation was introduced, thus allowing large-scale, homogenous protein production in Pichia pastoris. Altogether re-engineering of FGF21 led to significant improvements in its biopharmaceutical properties. The impact of these changes was assessed in a panel of in vitro and in vivo assays, which confirmed that biological properties of LY2405319 were essentially identical to FGF21. Specifically, subcutaneous administration of LY2405319 in ob/ob and diet-induced obese (DIO) mice over 7-14 days resulted in a 25-50% lowering of plasma glucose coupled with a 10-30% reduction in body weight. Thus, LY2405319 exhibited all the biopharmaceutical and biological properties required for initiation of a clinical program designed to test the hypothesis that administration of exogenous FGF21 would result in effects on disease-related metabolic parameters in humans.

Figure 1. Dynamic light scattering assessment of FGF21 and LY2405319.

Formulation stability of FGF21 (squares) and LY2405319 (diamonds) was assessed via dynamic light scattering at 37°C at 10 mg/mL in phosphate buffered saline at pH 7.4 containing 30 mM m-cresol. Protein concentration was 10 mg/ml.

Figure 2. Thermal stability profile of FGF21 and LY2405319.

Differential Scanning Calorimetry (DSC) was used to assess thermal stability of FGF21 (A) and LY2405319 (B); protein concentration was 1.3–1.6 mg/ml in PBS. Scan rate was 60°C/h.

Figure 3. FGF21 homology model.

An engineered disulfide bond in LY2405319 achieved loop stabilization at the C-terminus. The FGF21 homology model includes residues 18–140 of the human sequence. The native disulfide at C75-C93 and the new engineered disulfide bond at Leu118Cys-Ala134Cys are highlighted in yellow. The intervening loop between Leu118Cys and Ala134Cys is highlighted in green.

Figure 4. Expression levels of FGF21 and LY2405319 in Pichia pastoris.

FGF21 (black) and LY2405319 (red) were expressed in Pichia pastoris at shake-flask scale. Protein amounts were quantitated from crude media using an FGF21 ELISA.

Figure 5. LY2405319 bioactivity is βKlotho dependent.

Klb-dependence was evaluated by treating 3T3-L1/Klb (A), 3T3-L1/Kl (B), and 3T3-L1 (C) fibroblasts for 5 min with various concentrations of FGF21 (black squares) and LY2405319 (red triangles). Cells were also treated for 5 min with 10 nM FGF1 (blue diamonds) to assess klotho cofactor-independent activity and 10 nM FGF23 (green inverted triangles) to assess Kl-specific activity. After treatment, cells were lysed, and ERK1/2 phosphorylation was measured using AlphaScreen technology. Values are presented as mean ± SEM in relative fluorescent units.

Figure 6. LY2405319 bioactivity is comparable to wild type FGF21.

LY2405319 and FGF21 bioactivities were compared in mouse 3T3-L1/Klb fibroblasts (A) and differentiated mouse 3T3-L1 adipocytes (B) by monitoring 2-Deoxy-D-[U-¹⁴C]-glucose accumulation and in human HepG2 hepatoma cells by examining relative expression of GLUT1 mRNA (C) after incubation for 3 h or 1 h, respectively, with various concentrations of FGF21 (black squares) or LY2405319 (red triangles). Mean values ± SEM are presented as fold relative to treatment with vehicle alone.

Figure 7. LY2405319 is not mitogenic in human HepG2 hepatoma cells.

HepG2 human hepatoma cells were treated with 0.4nM FGF2 or various concentrations of FGF21, LY2405319, or insulin for 16 h and pulsed with [methyl-³H] thymidine for 4 h prior to DNA collection and measurement of ³H incorporation.

Figure 8. Treatment of ob/ob mice with either FGF21 or LY2405319 improves metabolic dysfunction.

Male ob/ob mice were treated with FGF21 or LY2405319 at various concentrations, as indicated, by constant infusion for 7 days. Blood glucose is reported as daily levels (A, B) or cumulatively for the treatment period (C). Plasma insulin was measured on the final day of treatment (D). Body weights were measured daily and are presented as body weight change (E) over the course of treatment. Plasma FGF21 and LY2405319 levels on the final day of treatment were measured by ELISA (F). Values are reported as mean ± SEM. (*), (**), and (***), p<0.01, p<0.001, and p<0.001, respectively, as compared to vehicle.

Figure 9. Treatment of DIO mice with either FGF21 or LY2405319 improves metabolic dysfunction.

Male DIO mice were treated with FGF21 or LY2405319 at various concentrations, as indicated, by constant infusion for 14 days. Changes in body weight (A), fat mass (B) cumulative food intake (C), and blood glucose (D) were determined for the duration of treatment. Values are reported as mean ± SEM. (*), p<0.01 as compared to vehicle.