Human serpins belong to a superfamily of serine protease inhibitors involved in the regulation of essential physiological processes, including coagulation via thrombin inhibition by AT. Inhibitory serpins undergo a remarkable folding mechanism to a thermodynamically unstable (i.e. metastable) conformation, highlighted by a long and flexible RCL, prior to secretion as soluble proteins into circulation. The serpin alpha-1-proteinase inhibitor (API) normally protects tissues from proteases released from inflammatory cells (e.g., neutrophil elastase). Importantly, a variant of API (i.e. API-Pittsburgh or API-M358R) was reported to be the cause of a fatal bleeding disorder in a young patient in the late 1970`s; the point mutation M358R at P1 of the RCL resulted in a dramatic shift in function toward thrombin inhibition. This dissertation summarizes the results from experiments performed with serpins expressed as membrane proteins tethered to the surface of mammalian cells. Serpins API-M358R, AT, HCII, and the non-inhibitory double mutant API-M358R/T345R were anchored to 293 and COS cell plasma membranes with N-terminal non-cleavable protein sequences derived from either the human asialoglycoprotein (AR) or transferrin (TF) receptors. Sub-cellular fractionation (with or without monolayer exposure to thrombin) immunoblots confirmed serpin localization to the integral membrane fraction using either anchoring approach or cell type. The blots also revealed that tethered API-Pittsburgh in particular, and AT to a lesser extent formed serpin-enzyme complex (SEC) with thrombin, while HCII and API-M358R/T345R (as expected) did not. While tethered API-M358R maintained inhibitory function, kinetic studies revealed that the rate of SEC formation was less rapid compared to its soluble counterpart. Additional testing by immunofluorescence microscopy, and flow cytometry confirmed the status of tethered API-M358R as a robust inhibitor of thrombin. That tethered serpins maintained the ability to inhibit thrombin provided the underlying rationale for the thesis hypotheses: surface displayed serpins can be used in gene therapy to temper thrombogenicity associated with certain diseased tissues (i.e. cancer), and tethered serpins can be used as a platform for screening RCL mutant libraries to identify “better” protease (i.e. thrombin) inhibitors. The potential gene therapy scheme was tested by expressing tethered API-Pittsburgh on the surface of T24/83 cancer cells constitutively co-expressing tissue factor (TF), and then measuring endogenous thrombin generation in the presence of re-calcified, and defibrinated human plasma by either discontinuous or continuous fluorescence-based thrombin generation assays (TGA). Unexpectedly, the displayed API-Pittsburgh did not appear to reduce discontinuous TGA thrombin suggesting that the difference may have been too low for accurate measurement by this method. Moreover, the results were identical when the same reaction was continuously monitored by fluorescence-based TGA, indicating that the levels of API-Pittsburgh expression were simply insufficient to effectively counter thrombin generation. The high levels were confirmed when up to 1 µM of hirudin variant 3, or soluble API-M358R, were required to completely abolish the thrombin profile. With this in mind, a measureable reduction in TGA was achieved when 293 cells were co-transfected with DNA ratios of API-M358R: TF adjusted to 9:1. Mammalian cell display, in combination with FACS/flow cytometry, has previously been employed to successfully develop improved monoclonal antibodies. However, there was never any certainty that the technique was applicable to the screening of displayed serpin mutant RCL libraries. The method was tested with a modest library degenerate only at P1; the rationale was that successful sorting would generate the expected wild-type (WT) P1 Arg (i.e. API-Pittsburgh). Unfortunately, repeated attempts did not result in enrichment, and flow cytometry was abandoned. An alternate protocol based on bacterial expression, and previously developed in our lab, was implemented in order to perform the library screens. This technique involved incubating lysates, containing soluble serpin RCL mutant candidates, with immobilized thrombin. Encouragingly, the P1 library screen identified the WT candidate at the expected frequency (5 in 150 lysates) as well as the more rare P1 mutant Lys (1 in 150 lysates). A kinetic comparison between mutant proteins containing the three basic residues revealed that P1 Arg (k2 ~ 105 M-1sec-1) was approximately two orders of magnitude more efficient than either Lys or His (both with k2 ~ 103 M-1sec-1) at inhibiting thrombin. The bacterial expression technique was then enhanced through kinetic optimization in order to facilitate the screening of more complex libraries. Analysis of the P7 to P1 (always Arg) library returned a selection of anticipated non-polar residues (5 in 100 lysates; 2 x Pro, 2 x Leu and Met) at P2. Extensive screening (~ 1300 colonies) of a second expanded library with the repeated VNN nucleotide sequence (i.e. no stop codons to eliminate truncated proteins) at P7 to P2 and P1 Arg, identified 7 x Pro mutants at P2 further confirming the results from the original library screen. Importantly, the assay also identified the novel mutant TLSATPR which registered the largest kinetic response (k2 ~ 5 x 105 M-1sec-1), and even exceeding API- M358R by a factor of ~ 3 (k2 ~ 1.5 x 105 M-1 sec-1). / Thesis / Doctor of Philosophy (Medical Science)

Human serpins belong to a superfamily of serine protease inhibitors involved in the regulation of essential physiological processes, including coagulation via thrombin inhibition by AT. Inhibitory serpins undergo a remarkable folding mechanism to a thermodynamically unstable (i.e. metastable) conformation, highlighted by a long and flexible RCL, prior to secretion as soluble proteins into circulation. The serpin alpha-1-proteinase inhibitor (API) normally protects tissues from proteases released from inflammatory cells (e.g., neutrophil elastase). Importantly, a variant of API (i.e. API-Pittsburgh or API-M358R) was reported to be the cause of a fatal bleeding disorder in a young patient in the late 1970`s; the point mutation M358R at P1 of the RCL resulted in a dramatic shift in function toward thrombin inhibition. This dissertation summarizes the results from experiments performed with serpins expressed as membrane proteins tethered to the surface of mammalian cells. Serpins API-M358R, AT, HCII, and the non-inhibitory double mutant API-M358R/T345R were anchored to 293 and COS cell plasma membranes with N-terminal non-cleavable protein sequences derived from either the human asialoglycoprotein (AR) or transferrin (TF) receptors. Sub-cellular fractionation (with or without monolayer exposure to thrombin) immunoblots confirmed serpin localization to the integral membrane fraction using either anchoring approach or cell type. The blots also revealed that tethered API-Pittsburgh in particular, and AT to a lesser extent formed serpin-enzyme complex (SEC) with thrombin, while HCII and API-M358R/T345R (as expected) did not. While tethered API-M358R maintained inhibitory function, kinetic studies revealed that the rate of SEC formation was less rapid compared to its soluble counterpart. Additional testing by immunofluorescence microscopy, and flow cytometry confirmed the status of tethered API-M358R as a robust inhibitor of thrombin. That tethered serpins maintained the ability to inhibit thrombin provided the underlying rationale for the thesis hypotheses: surface displayed serpins can be used in gene therapy to temper thrombogenicity associated with certain diseased tissues (i.e. cancer), and tethered serpins can be used as a platform for screening RCL mutant libraries to identify “better” protease (i.e. thrombin) inhibitors. The potential gene therapy scheme was tested by expressing tethered API-Pittsburgh on the surface of T24/83 cancer cells constitutively co-expressing tissue factor (TF), and then measuring endogenous thrombin generation in the presence of re-calcified, and defibrinated human plasma by either discontinuous or continuous fluorescence-based thrombin generation assays (TGA). Unexpectedly, the displayed API-Pittsburgh did not appear to reduce discontinuous TGA thrombin suggesting that the difference may have been too low for accurate measurement by this method. Moreover, the results were identical when the same reaction was continuously monitored by fluorescence-based TGA, indicating that the levels of API-Pittsburgh expression were simply insufficient to effectively counter thrombin generation. The high levels were confirmed when up to 1 µM of hirudin variant 3, or soluble API-M358R, were required to completely abolish the thrombin profile. With this in mind, a measureable reduction in TGA was achieved when 293 cells were co-transfected with DNA ratios of API-M358R: TF adjusted to 9:1. Mammalian cell display, in combination with FACS/flow cytometry, has previously been employed to successfully develop improved monoclonal antibodies. However, there was never any certainty that the technique was applicable to the screening of displayed serpin mutant RCL libraries. The method was tested with a modest library degenerate only at P1; the rationale was that successful sorting would generate the expected wild-type (WT) P1 Arg (i.e. API-Pittsburgh). Unfortunately, repeated attempts did not result in enrichment, and flow cytometry was abandoned. An alternate protocol based on bacterial expression, and previously developed in our lab, was implemented in order to perform the library screens. This technique involved incubating lysates, containing soluble serpin RCL mutant candidates, with immobilized thrombin. Encouragingly, the P1 library screen identified the WT candidate at the expected frequency (5 in 150 lysates) as well as the more rare P1 mutant Lys (1 in 150 lysates). A kinetic comparison between mutant proteins containing the three basic residues revealed that P1 Arg (k2 ~ 105 M-1sec-1) was approximately two orders of magnitude more efficient than either Lys or His (both with k2 ~ 103 M-1sec-1) at inhibiting thrombin. The bacterial expression technique was then enhanced through kinetic optimization in order to facilitate the screening of more complex libraries. Analysis of the P7 to P1 (always Arg) library returned a selection of anticipated non-polar residues (5 in 100 lysates; 2 x Pro, 2 x Leu and Met) at P2. Extensive screening (~ 1300 colonies) of a second expanded library with the repeated VNN nucleotide sequence (i.e. no stop codons to eliminate truncated proteins) at P7 to P2 and P1 Arg, identified 7 x Pro mutants at P2 further confirming the results from the original library screen. Importantly, the assay also identified the novel mutant TLSATPR which registered the largest kinetic response (k2 ~ 5 x 105 M-1sec-1), and even exceeding API- M358R by a factor of ~ 3 (k2 ~ 1.5 x 105 M-1 sec-1). / Dissertation / Doctor of Philosophy (PhD)