Coupling Therapeutics to Human Erythrocytes Demonstrates Target-Dependent Effects on Red Cell Physiology While Preserving Efficacy

Delivery of bio-therapeutics by red blood cells (RBCs) can greatly enhance pharmacokinetics and pharmacodynamics of the appended or loaded agents, and may even potentiate induction of immunologic tolerance. Our group and others have successfully used fusion proteins, antibodies, and peptides to couple therapeutics to murine, but not human, RBCs. It is known that extracellular ligands have the potential to induce marked, epitope-dependent changes in red cell physiology, including changes in deformability, phosphatidyl-serine (PS) exposure, and reactive oxygen species (ROS) production, particularly for agents targeted to glycophorin A and Band 3, two highly-expressed membrane proteins. To produce clinically translatable strategies for human RBCs, it is critical to identify optimal red cell target epitopes, understand their effects on red cell physiology, and create humanized or human-like ligands to minimize immunogenicity.

We constructed single chain antibodies (scFv) against antigenic determinants on Band 3 protein (Wr^b) and RHCE protein (Rh17/Hr₀) on human erythrocytes using phage display libraries prepared from immunized cynamolgous macaques (Macacafascicularis). Both these antigens are present on essentially 100% of the human population. Unfused scFvs were produced in E.coli while fusions of scFv with the extracellular domain of human thrombomodulin (TM-scFv) were produced in Drosophila S2 cells. Binding of recombinant proteins to human RBCs was measured by radioimmunoassay and flow cytometry. Generation of activated protein (APC) by RBCs loaded with TM-scFv fusions was measured by colorimetric assay. RBCs pre-incubated with varying concentrations of anti-Band3 and anti-RHCE fusions were assessed for osmotic resistance and mechanical integrity by exposure to hypo-osmolar medium and rotation in the presence of glass beads, respectively. PS exposure was measured by annexin V binding, and ROS generation was measured by dihydrorhodamine-associated fluorescence. Effects on RBC rheology were measured by flowing through microfluidic channels under controlled shear rates. Efficacy of TM-scFv fusions in diseased micro-vessels was assessed using a TNF-alpha stimulated, endothelialized microfluidic model.

Single-chain antibody fragments and TM fusion proteins targeted to conserved epitopes on Band 3 protein and RHCE protein bound to human, but not murine or porcine, RBCs with high specificity and affinity (~50 nM), and in numbers consistent with the expected level of target expression (10⁵ and 10⁶ copies/RBC for RHCE and Band3, respectively). Coating RBCs with proteins targeted to Band 3 lessened RBC hypo-osmolar hemolysis (20% reduction) but increased hemolysis (2-fold) under mechanical stress, changes compatible with decreased red cell deformability. Proteins targeted to RHCE did not induce significant changes in hemolysis of RBCs under either osmotic or mechanical stress. Targeting neither Band 3 nor RHCE induced significant exposure of PS or production of ROS. Target-dependent effects on RBC rheology were observed under varying shear stresses in a microfluidic system. Fusion proteins of TM targeted to both epitopes demonstrated dose- and surface-copy-number-dependent generation of APC in the presence of PC and thrombin. Both TM-scFv fusion proteins were efficacious in a microfluidic model of disseminated intravascular coagulation using whole human blood by demonstrating near complete abrogation of fibrin generation in response to endothelial activation with TNF-alpha.

In summary, we designed human RBC-specific non-human primate single chain antibody fragments capable of fusion to therapeutic cargoes. The TM-scFv fusions maintained therapeutic activity when bound to human RBCs and showed effective thromboprophylaxis in a whole-blood model of vasculitic injury. These antibodies and fusion proteins bound to erythroid-specific epitopes, and demonstrated target-dependent effects on several aspects of red cell physiology. The non-human primate origin of the antibodies should minimize their potential immunogenicity and the findings provide a platform to translate red cell targeted drug delivery into the clinical realm.