McVey, John H.

Abstract TFPI is an endothelium associated anticoagulant protein that directly inhibits tissue factor (TF) pro-coagulant activity and prevents disease produced by intravascular thrombosis. TFPI is produced in three alternatively spliced isoforms in mice, α, β and γ, that differ in domain structure and mechanism for association with endothelium. TFPIα has three Kunitz-type serine protease inhibitory domains and a basic C-terminal region. It produces anticoagulant activity by direct inhibition of factor Xa (fXa) via the second Kunitz domain (K2) and, in a fXa dependent manner, inhibition of TF-fVIIa via the first Kunitz domain (K1). The third Kunitz domain (K3) and C-terminal region do not directly inhibit proteolysis. However, alignment of sequences from various vertebrate species demonstrates that K3 has maintained a high degree of sequence conservation for over 430 million years suggesting that K3 maintains a physiological function. TFPIα indirectly associates with endothelium by binding to a glycosylphosphatidyl inositol (GPI) anchored protein and binds non-specifically to glycosaminoglycans. TFPIβ has K1 and K2 but lacks K3 and has a distinct C-terminal region that encodes a GPI-anchor attachment sequence. TFPIγ has K1 and K2 but lacks K3 and has a distinct C-terminal region that results in production of a secreted protein. It is not known if these structurally diverse forms of TFPI with different mechanisms for cell surface association have variable efficacy in their ability to inhibit TF activity in vivo. We characterized TFPI isoform expression in mouse tissues in order to define specific physiological functions for alternative splicing of TFPI and how it may hinder diseases mediated by intravascular TF activity. Message for TFPIα is ~16-fold more abundant than message for TFPIβ or TFPIγ in mouse tissues. In situ hybridization studies of mouse heart and lung found that the isoforms are expressed in the same pattern, primarily in endothelial cells. No cell or vascular bed exclusively producing a single isoform was identified. Western blot analyses revealed the somewhat surprising finding that TFPIβ is the predominant isoform produced by adult mouse tissues. Thus, protein synthesis of TFPI isoforms in mice is regulated at least in part at the level of protein translation. Mouse plasma TFPI levels increase only 10–20% following heparin infusion. These data contrast with the 2- to 4-fold increase in plasma TFPI observed following heparin infusion in humans. The presence of only small amounts of heparin releasable TFPI in mouse plasma is consistent with TFPIβ as the major isoform because TFPIβ is directly GPI-anchored to endothelium and lacks the highly basic C-terminal region that allows TFPIα to associate non-specifically with endothelium glycosaminoglycans. To our knowledge, these are the first data demonstrating TFPIβ production in vivo. In contrast to the adult tissues, western blot analysis of mouse placenta revealed that it produces primarily TFPIα. This finding led us to examine mouse fetal tissue to determine if there is developmental regulation of the production of different TFPI isoforms. These studies demonstrated that TFPIα and TFPIβ are produced in approximately equal amounts in E14.5 embryos. The relative proportion of TFPIα in mouse tissues then decreases during development and very little is produced at 12 weeks of age. The evolutionary conservation of TFPIα and its production in mouse placenta and embryonic tissues suggests that unique features of TFPIα, such as K3, the basic C-terminal region and/or its binding to the endothelial surface through a GPI-anchored co-receptor, perform specific functions in the regulation of intravascular TF activity during vasculogenesis/angiogenesis that are not performed by TFPIβ or TFPIγ.

ABSTRACT Gene transfer vectors containing adenovirus (Ad) serotype 35 (Ad35) fibers have shown promise for cancer and stem cell gene therapy. In this study, we attempted to improve the in vitro and in vivo infection properties of these vectors by increasing their affinity to the Ad35 fiber receptor CD46. We constructed Ad vectors containing either the wild-type Ad35 fiber knob (Ad5/35) or Ad35 knob mutants with 4-fold- and 60-fold-higher affinity to CD46 (Ad5/35+ and Ad5/35++, respectively). In in vitro studies with cell lines, the higher affinities of Ad5/35+ and Ad5/35++ to CD46 did not translate into correspondingly higher transduction efficiencies, regardless of the CD46 receptor density present on cells. However, in vivo, in a mouse model with preestablished CD46 high liver metastases, intravenous injection of Ad5/35++ resulted in more-efficient tumor cell transduction. We conclude that Ad5/35 vectors with increased affinity to CD46 have an advantage in competing with non-CD46-mediated sequestration of vector particles after intravenous injection.