Thiol Complexes of Gadolinium for Imaging Therapy-Induced Oxidative Stress in Pre-Clinical Tumors

Introduction: Oxygen is required as the terminal electron acceptor in mitochondrial respiration, and under conditions of hypoxia there can be paradoxically increased levels of reactive oxygen species (ROS). In turn, high levels of ROS can lead to compensatory production of antioxidants via pathways mediated by the Nrf2 transcription factor and the Antioxidant Response Element (ARE). ARE-driven responses can include the increased production of reduced thioredoxin, reduced glutathione, and transfer of reduced cysteine equivalents between intracellular and extracellular pools [1]. Hyperglycemia and exposure to 2-deoxygluxose (2DG) can also lead to an oxidative stress response [2]. Increased thiol/disulfide ratio in the extracellular milieu is associated with increased tumor resistance to radiotherapy and platinum-based drugs which act by inducing oxidative stress. We present the development of DOTAand DO3A-based thiol complexes of gadolinium that are responsive to the extracellular thiol/disulfide ratio. These complexes covalently bind human serum albumin (HSA) at the conserved Cys position under normal oxidizing conditions, but are designed to be competed off by endogenous thiols in highly reducing microenvironments. We propose a method to exploit the differential r1 relaxivities of the bound vs. unbound forms to detect therapy-mediated changes in tumor redox status in a pre-clinical model. Methods: Two previously-described DOTA-based, and 3 novel DO3A-based, thiol complexes of gadolinium (fig. 1) have been synthesized and characterized in these studies. Relaxivity measurements were made at 37°C and pH 7.4 in buffered saline in the absence (r1,free, table 1) or presence of 0.66 mM HSA. The r1 relaxivity of the bound fraction of each molecule (r1,bound, table 1) was calculated from a knowledge of the corresponding r1,free and the apparent equilibrium association constant of each molecule with HSA (KA, table 2). Equilibrium binding was measured by means of rapid filtration through 30 kDa MWCO centrifugal filters [3]. While covalent binding to HSA at Cys is expected, the binding to commercial HSA in solution exhibited equilibrium characteristics that were consistent with the commercial HSA being partially bound at the Cys position to endogenous small molecule thiols such as cysteine or homocysteine. The KA values reported in table 2 are internally consistent for a given lot of HSA (Sigma-Aldrich). Sensitivity to solution thiol content was demonstrated by measuring KA in the presence of varying amounts of either homocysteine or the corresponding thiol complex of Eu (table 3). Quantitative dynamic contrast-enhanced MRI (DCE-MRI) requires knowledge of the arterial input function [4]. Additionally, quantitation with “responsive” Gd agents such as Gd-3c-SH and Gd-4b-SH may require dual boluses – of a non-responsive Gd agent followed by the responsive agent [5]. Since these Gd thiol molecules spontaneously bind circulating plasma albumin, we have sought to identify a “saturating” dose of each Gd thiol such that at relatively long times post-injection, the change in tissue T1 (∆T1) is only a weak function of the injected Gd dose. DCE-MRI experiments were performed on MIA-PaCa-2 tumor xenograft bearing SCID mice using a GRE sequence (TR=72ms, TE=2.5ms, α=75°, matrix=128x128). Using the identified Gd dose, DCE-MRI T1 measurements were made on untreated mice and mice treated 24 h previously with 2DG (500 mg/Kg, i.p.) (fig. 2).