AUTHORS
Laura M Haynes, PhD, Matthew L Holding, PhD2, Mary Migliorini, MS3, Hannah L DiGiovanni, BS4, Meilyn Ward, BS5, David Siemieniak, BS6, Cory Emal, PhD7, Dudley Strickland, PhD8, Daniel A Lawrence, PhD9, David Ginsburg, MD10
AFFILIATIONS
1 University of Michigan, Ann Arbor, Michigan, United States
2 University of Michigan, Ann Arbor, Michigan, United States
3 University of Maryland School of Medicine, Baltimore, Maryland, United States
4 University of Michigan, Ann Arbor, Michigan, United States
5 University of Michigan, Ann Arbor, Michigan, United States
6 University of Michigan, Ann Arbor, Michigan, United States
7 Eastern Michigan University, Ypsilanti, United States
8 University of Maryland School of Medicine, Baltimore, Maryland, United States
9 University of Michigan, Ann Arbor, Michigan, United States
10 University of Michigan, Ann Arbor, Michigan, United States
ABSTRACT
Plasminogen activator inhibitor-1 (PAI-1; 379 amino acids) is the key physiologic inhibitor of tissue- and urokinase-type plasminogen activators (tPA and uPA), regulating the fibrinolytic process. As an acute phase reactant, PAI-1 has also been implicated in multiple processes including inflammation, infection, and fibrosis.
PAI-1 is a member of the serine protease inhibitor (serpin) protein superfamily, which is structurally characterized by a reactive center loop (RCL) that serves as bait for its target proteases and extends from the globular core of the protein. Before the target protease can fully cleave the RCL, the RCL inserts into the serpin’s central β-sheet, trapping the serine protease-inhibitor complex in the acyl intermediary conformation. In contrast to other serpins, PAI-1 undergoes a relatively rapid latency transition (functional half-life of approximately 1-2 h) in which the RCL spontaneously inserts into the central β-sheet in the absence of a proteolytic challenge, leading to a loss of inhibitory function. Virtually all PAI-1 in vivo, however, is found in complex with its protein cofactor vitronectin (VTN), which modestly prolongs its functional half-life. However, the allosteric mechanism by which VTN achieves this has not yet been fully characterized. To address this knowledge gap, we applied phage display-based deep mutational scanning to simultaneously measure the effects of all possible single amino acid substitutions on PAI-1’s ability to both inhibit uPA and bind VTN to identify amino acid residues in PAI-1, both at and distal to the VTN binding interface, that mediate PAI-1’s affinity for VTN.
Briefly, a library of all possible PAI-1 single-amino-acid substitution variants was expressed as a fusion protein with a truncated p3 protein of M13 filamentous phage. Variants in the library were selected for their ability to inhibit uPA and bind VTN using immunoprecipitation and subsequently identified by high-throughput DNA sequencing. PAI-1 variants that inhibit uPA (indicating that they are properly folded and functional) but have reduced affinity for VTN were interpreted as being critical amino acid residues for binding VTN. These variants were found sequentially and spatially distributed at 65 sites across the PAI-1 protein, with residues concentrated at known VTN binding sites.
To biochemically validate and extend these findings, we next recombinantly expressed thirteen PAI-1 variants with single amino acid substitutions, both at and distal to the known VTN binding site, and measured their affinity for VTN by surface plasmon resonance. All variants exhibited decreased affinity for VTN compared to wild-type (Kd = 0.5 nM), with the most significant effects on decreased VTN binding resulting from Q123K (Kd = 22 nM, p < 0.0001) and Q204G (Kd = 31 nM, p < 0.0001). While Q123 is known to be a critical residue for mediating VTN binding to PAI-1, Q204 is located approximately 40 Å away from the VTN binding site, in a pocket in which a small molecule inhibitor of PAI-1, CDE-096, is known to bind—reciprocally and allosterically inhibiting VTN binding.
Given the strong effects on VTN binding of PAI-1 amino acid substitutions both at the known VTN interface, as well as a distal yet allosterically linked site, we sought to determine how allosteric signals are communicated between these sites. A network analysis in which VTN-affecting residues in PAI-1 within 8 Å of each other were clustered revealed that, although these allosterically linked residues are spatially closer and make more contacts with other amino acids than would be expected at random, they do not define a single linear network of residues that connects the VTN binding site to the Q204 binding pocket. Further analyses demonstrated that the residues involved in the VTN allosteric signaling network are not immutable, as they are not more likely to be evolutionarily conserved than other residues in PAI-1 but rather are enriched in coevolving residues. Together, these findings suggest that the VTN-induced allosteric network in PAI-1 is dispersed and dynamic rather than a single fixed pathway.
DISCLOSURES
Laura M Haynes, PhD: No relevant conflicts of interest to declare.