Strategies for optimizing the serum persistence of engineered human arginase I for cancer therapy

Abstract

Systemic L-arginine depletion following intravenous administration of l-arginine hydrolyzing enzymes has been shown to selectively impact tumors displaying urea cycle defects including a large fraction of hepatocellular carcinomas, metastatic melanomas and small cell lung carcinomas. However, the human arginases display poor serum stability (t(1/2)=4.8h) whereas a bacterial arginine deiminase evaluated in phase II clinical trials was reported to be immunogenic, eliciting strong neutralizing antibody responses. Recently, we showed that substitution of the Mn(2+) metal center in human Arginase I with Co(2+) (Co-hArgI) results in an enzyme that displays 10-fold higher catalytic efficiency for L-Arg hydrolysis, 12-15 fold reduction in the IC(50) towards a variety of malignant cell lines and, importantly a t(1/2)=22h in serum. To investigate the utility of Co-hArgI for L-Arg depletion therapy in cancer we systematically investigated three strategies for enhancing the persistence of the enzyme in circulation: (i) site specific conjugation of Co-hArgI engineered with an accessible N-terminal Cys residue to 20kDa PEG-maleimide (Co-hArgI-C(PEG-20K)); (ii) engineering of the homotrimeric Co-hArgI into a linked, monomeric 110kDa polypeptide (Co-hArgI x3) and (iii) lysyl conjugation of 5kDa PEG-N-hydroxysuccinimide (NHS) ester (Co-hArgI-K(PEG-5K)). Surprisingly, even though all three formulations resulted in proteins with a predicted hydrodynamic radius larger than the cut-off for renal filtration, only Co-hArgI amine conjugated to 5kDa PEG remained in circulation for sufficiently long durations. Using Co-hArgI-K(PEG-5K) labeled with an end-terminal fluorescein for easy detection, we demonstrated that following intraperitoneal administration at 6mg/kg weight, a well tolerated dose, the circulation t(1/2) of the protein in Balb/c mice is 63±10h. Very low levels of serum L-Arg (<5μM) could be sustained for over 75h after injection, representing a 9-fold increase in pharmacodynamic efficacy relative to similarly prepared Mn(2+)-containing hArgI conjugated to 5kDa PEG-NHS ester (Mn-hArgI-K(PEG-5K)). The favorable pharmacokinetic and pharmacodynamic properties of Co-hArgI-K(PEG-5K) reported here, coupled with its human origin which should reduce the likelihood of adverse immune responses, make it a promising candidate for cancer therapy.

Figure 1

Schematic representation of the pharmacological formulations of hArgI: A) hArgI ×3: three polypeptides of arginase linked N-terminal to C-terminal with a flexible amino acid linker. B) hArgI-C^PEG-20K: a Cys residue was engineered into the N-terminal for conjugation with PEG-20K maleimide. C) hArgI-K^PEG-5K or Co-hArgI-K^PEG-10K: the lysyl residues of hArgI were conjugated with PEG 5 or 10K succinimidyl-ester.

Figure 2

Analytical SEC of hArgI Formulations: (A) The Co-hArgI ×3 fusion protein with an apparent MW of 100 KDa. (B) The singly PEGylated Co-hArgI-C^PEG-20K with an apparent mass of 520 KDa. (C) The multiply PEGylated Co-hArgI-K^PEG-5K with an apparent MW of 750 KDa.

Figure 3

in vivo serum l-Arg levels in mice following IP injection at a 6 mg/kg dose each of: A) Co-hArgI-K^PEG-5K (●), B) Mn-hArgI-K^PEG-5K (◯), C) Co-hArgI-C^PEG-20K (◆) and D) Co-hArgI ×3 (▵). Co-hArgI-K^PEG-5K (●) yielded a calculated serum L-Arg depletion time of 78 ± 10 hrs. B) Mn-hArgI-K^PEG-5K (◯) with a calculated serum L-Arg depletion time of 9 ± 3 hrs. C) Co-hArgI-C^PEG-20K with an estimated serum L-Arg depletion time of ~ 0.5 hrs and Co-hArgI ×3 (▵)showing no apparent effect upon serum L-Arg levels. (Figures C & D: the fitting lines were extrapolated to 200 hrs to provide a sense of scale with Figures A & B)

Figure 4

Pharmacokinetics in balb/c mice following IP administration at 6 mg/kg. Each sample contained a portion of protein PEGylated with FITC terminating polyethylene glycol and the concentration of protein was monitored by in-gel fluorescence: of (●) Co-hArgI-K^PEG-5K (n=5). The absorption rate (K_a) was too fast to estimate, but the elimination rate (K_el) was well fit by equation 1, yielding a serum t_½ life of 63 ± 10 hrs. (◻) Co-hArgI-K^PEG-10K Absorption rate K_a = 0.13 ± 0.05 hr⁻¹; t_½ = 69 ± 20 hrs.