SPECIFIC AIMS

Pseudomonas aeruginosa induces the production and release of amyloids, including beta amyloid, from pulmonary microvascular endothelial cells (PMVECs)^1,2. Once released, these pathologic amyloid species disrupt the endothelial barrier and they may propagate through the vasculature to promote secondary end-organ damage, including neurocognitive decline³. The mechanism regulating the production of cytotoxic beta amyloid relative to non-cytotoxic beta amyloid is unknown and is the focus of this proposal.

Beta amyloid is a peptide derived from amyloid precursor protein through consecutive cleavages by β- and -secretases, and has both physiologic and pathologic roles^4,5,6,7. Amyloid precursor protein is a transmembrane protein expressed in many tissue types, and until recently, has only been extensively studied in neurons and brain endothelial cells⁸. In the production of beta amyloid, β-secretase first cleaves amyloid precursor protein, followed by -secretase⁹. -secretase is a multiprotein complex whose activity is due to presenilin-1. Preliminary data from our lab suggests that PMVECs possess all known proteins of the -secretase complex.

-secretase activating protein (GSAP) is a newly discovered component of both β- and -secretases^10,11,12,13. Research indicates GSAP modifies -secretase activity by localizing the enzyme to amyloid precursor protein¹⁴. We observed GSAP expression in endothelial cells using non-biased microarray, RNAseq, methylation and miRNA screens, and confirmed protein expression and function in microvascular endothelium. Studies from our lab show GSAP activity is required for infection-induced production of cytotoxic beta amyloid. GSAP was deleted in endothelial cells using CRISPR-Cas9. In the absence of GSAP, endothelial cells were protected from the primary infection and the infection-elicited beta amyloidspecies were not cytotoxic, but rather had antimicrobial properties² These data suggest the presence of a functional GSAP is essential for bacteria to induce a cytotoxic response.

Here, we test the hypothesis that Pseudomonas aeruginosa infection of PMVECs increases GSAP expression and/or function that is necessary to produce cytotoxic beta amyloid. (Figure 1)

Specific aims test the related hypotheses that GSAP expression:

AIM 1: Is a critical determinant of endothelial injury following infection with Pseudomonas aeruginosa; and,

AIM 2: Is necessary for production of cytotoxic beta amyloid.

RESEARCH STRATEGY

Background: Pseudomonas aeruginosa infection of the pulmonary endothelium causes production and release of amyloids, including beta amyloid (A¹⁵. These amyloids have the potential to propagate through the vasculature and ultimately damage other organs. These amyloids are heat stable and protease resistant and they can be enriched in the 50% ammonium sulfate fraction. They can be detected using Thioflavin T fluorescence and Congo red fluorescence and birefringence. Once they are generated, the cytotoxic amyloids are transmissible among cells and also self-replicating. Thus, endothelial cytotoxic amyloids display characteristics of prions.

Beta amyloid peptides range from 36 to 43 amino acids in length and are involved in the pathology of Alzheimer’s disease¹⁶. The most prominent forms involved in disease are beta amyloid 1-40 and 1-42. They are derived from amyloid precursor protein through consecutive cleavages by β- and -secretases. Amyloid precursor protein is a transmembrane protein expressed in many tissue types, including the lung¹⁷.  β-secretase is the first protease to cleave amyloid precursor protein. It cleaves amyloid precursor protein towards the N-terminus producing a soluble APP-β peptide and a C-terminal fragment. Next,-secretase cleaves the C-terminal fragment into an amyloid intracellular domain (AICD) and the amyloid beta peptide¹⁸.

-secretase is composed of 5 subunits; presenilins 1 or 2 (PSEN1/2), presenilin enhancer 2 (PEN2), anterior pharynx defective 1 (APH1), nicastrin, and gamma secretase activating protein (GSAP)^12,19. Presenilins1/2 contain the catalytic portion of the gamma secretase complex^20,21.  Presenilin enhancer 2 is involved in proteolytic activity and complex formation and stability²². APH1 aids in the stabilization of newly synthesized presenilin facilitating the generation of the active form of presenilin^23,24. Nicastrin functions as a substrate receptor for the complex²⁵. GSAP is a recently described component of -secretase, whose function is to localize -secretase to amyloid precursor protein, and away from other potential enzyme targets, like Notch¹². GSAP has been reportedly producced as a 98kDa protein that is rapidly cleaved to a 16kDa active form¹². However, the findings that GSAP is produced as a 98kDa protein that is rapidly cleaved to a 16kDa protein and that it regulates gamma clevage of APP have been contradicted by a few groups^26,27,28. Recently it has been found that GSAP activation is dependent on caspase-3¹¹. As GSAP modifies the abundance and/or balance of beta amyloid products generated by cleavage of amyloid precursor protein, thus, it may mold the antimicrobial and/or cytotoxic properties of endothelial amyloids^29,30. We have tested this principle by CRISPR-Cas9 deletion of GSAP. Experiments showed substantially reduced cytotoxicity, and increased antimicrobial properties of endothelial amyloids. These data suggest that mechanisms regulating GSAP abundance and/or activity determine the antimicrobial versus cytotoxic balance of endothelial amyloids, specifically A.

Significance: Nosocomial pneumonia is the second most common nosocomial infection in the United States, and is associated with an increased length of stay and an increased cost of care estimated to be more than $40,000 per patient³¹. P. aeruginosa is a major cause of nosocomial pneumonia³². Patients in the intensive care unit (ICU) that contract nosocomial pneumonia suffer increased incidences of morbidity and mortality³³, abrupt cognitive impairment³⁴, and secondary end organ damage post discharge. Nosocomial pneumonia occurs in 5 to 15 patients per 1,000 hospital admissions³⁵ and accounts for 25% of all ICU infections³⁵ and more than 50% of all prescribed antibiotics³⁶. Contracting nosocomial pneumonia increases a patient’s length of stay to 7-9 days³⁶. The estimated attributable mortality of nosocomial pneumonia is between 33% and 50%³¹. Survivors often have a higher rate of healthcare utilization³⁷. In the first year post release from the ICU, cardiovascular and neurocognitive dysfunction are major concerns³⁴. Importantly in hospitalized patients, P. aeruginosa is the most common multidrug-resistant gram-negative pathogen causing pneumonia³⁸.The exact link between nosocomial pneumonia and end organ damage seen in ICU patients post-discharge has not been determined.

Here, we test a novel potential mechanism of action that may account for progressive tissue injury in the aftermath of critical illness. Work from our lab has shows that lung infection elicits cytotoxic amyloid production. These amyloids fulfill the criterion for prion proteins because they are self-replicating and transmissible³⁹. We have shown that these infection-induced amyloids are transmissible in culture, in the absence of bacterial infection, indefinitely. In experiments, the amyloids were transmissible in culture beyond 3 months, without any evidence of decreased potency. Thus, primary lung infection may set into motion a cascade of events that contribute to end organ dysfunction following critical illness.

This concept has been vetted in animal infection models, and also in human ICU patients. Through collaborations with a critical care team at the University of Alabama at Birmingham led by Dr. Jean-Francois Pittet, our lab analyzed bronchoalveolar lavage fluid, plasma, cerebrospinal fluid, and urine from ICU patients with and without nosocomial pneumonia due to Pseudomonas aeruginosa, Klebsiella pneumoniae, or Staphylococcus aureus infection. Patients harboring infection had cytotoxic amyloid species present in all of these biological fluids. Amyloids present in the uninfected patients were not cytotoxic. Ongoing work by Dr. Mike Lin shows that amyloids present in the cerebrospinal fluid of infected patients impair neurological processing, leading to decreased long term potentiation of rodent hippocampus brain slices². These translational studies in humans support the concept that infection-induced endothelial amyloids harm the host and represent a previously unacknowledged mechanism of human disease.

Innovation: Previously, GSAP has only been extensively studied in neurons and brain endothelial cells. There has been no description of the role of GSAP outside of these cell types. This study aims to reveal the first function(s) of the GSAP protein in pulmonary microvascular endothelial cells during primary and secondary P. aeruginosa infections. This study also aims to determine the role of GSAP in the protection of the endothelial barrier during infection with P. aeruginosa and its mutants. The outcome of this proposal could provide insight into a new prevention and/or therapy for end-organ damage and neurocognitive decline seen in ICU patients that acquire nosocomial pneumonia.

Preliminary Studies: Infecting PMVECs with the P. aeruginosa mutant ExoY⁺ (PA103 exoUexoT::Tc/pUCPexoY)causes production and release of Avariants (i.e. A_40/42)  into the supernatant and/or extracellular space¹. These cytotoxic amyloids are heat stable and protease resistant and they can be enriched in the 50% ammonium sulfate fraction. They are detected using Thioflavin T fluorescence and Congo red fluorescence and birefringence. Once generated, the cytotoxic amyloids are transmissible among cells; we have passaged them among naïve, i.e., uninfected, cells for more than 4 consecutive months, demonstrating they are also self-replicating. Thus, endothelial cytotoxic amyloids display characteristics of prions^1,2.

We identified GSAP expression in endothelium in an unbiased series of mRNA profiling, RNAseq, methylation and miR profiling studies, using both pulmonary artery and microvascular endothelial cells (data not shown). Preliminary data indicate PMVECs possess the molecular components of β- and -secretases. For example, our lab’s RNAseq results indicate lung endothelial cells express β- secretases 1 and 2, and multiple components of the -secretase complex. To corroborate the RNAseq results and test protein expression, my preliminary studies resolved constitutive expression of multiple components of the -secretase complex, including presenilin-1, presenilin-2, presenilin enhancer 2, nicastrin, and anterior pharynx defective 1 (Figure 2).

To test the physiological relevance of GSAP, I generated CRISPR-Cas9 knockout cell lines. Pulmonary microvascular endothelial cells underwent guide RNA targeting DNA excision in exon 6 of the gene and 86 clones were expanded. PCR screening revealed 11 cell lines harbored complete GSAP gene deletions; examples of these knockouts are shown in Figure 3. I am currently screening all cell lines to confirm the presence and absence of mRNA and protein in control and knockout cells, respectively. However, our ongoing bioassays demonstrate three notable findings. First, GSAP knockout cells grow slower than their controls (data not shown). Second, GSAP knockout cells are protected from primary P. aeruginosa infection (Figure 4). Third, supernatant collected from GSAP knockout cells is not cytotoxic.

We next tested whether GSAP deficient cells generate cytotoxic or non-cytotoxic supernatant. To test this idea, cytotoxicity assays were performed. Supernatant obtained from PMVECs following infection with virulent P. aeruginosa was highly cytotoxic (data not shown),  and it did not possess any antimicrobial activity.  However, in GSAP deficient cells, P. aeruginosa elicited supernatant was not cytotoxic. Thus, GSAP contributes to the endothelial production of amyloids with cytotoxic properties, and inhibiting GSAP expression enables the endothelium to produce noncytotoxic amyloids. We have tested whether GSAP contributes to the production of Ain the endothelial cell supernatant. Wild-type and GSAP knockout endothelial cells were infected with P. aeruginosa for 7 hours, supernatant was collected, and amyloids were immunoprecipitated using an anti-amyloid antibody. We then immunoblotted specifically for AAs shown in Figure 5, cytotoxic Arecovered in the supernatant is greatly reduced in GSAP deleted cells.

AIM 1: Test the hypothesis that GSAP is a critical determinant of endothelial injury following infection with P. aeruginosa.

Rationale: Our preliminary data suggest that GSAP expression and/or activity is required to produce cytotoxic A. Thus, it may also be involved in the injury to PMVECs seen during P. aeruginosa infections.  Here we will determine the impact that GSAP has on PMVECs during a primary infection with P. aeruginosa. The 16kD form of GSAP is the active form¹², although it remains unclear as to whether expression of the 16kD active form ofGSAP parallels endothelial injury. This issue will be addressed in this aim. At the completion of this aim, we will have determined what impact GSAP has on PMVECs during P. aeruginosa infection and which form (the 98kD or the 16kD) is responsible for what we see.

Approach: We will use PMVECs infected with P. aeruginosa mutant ExoY⁺ and use ExoY^K81M  (PA103 exoUexoT::Tc/pUCPexoY^K81M) and ΔPcrV as controls. ExoY⁺ possesses a functional type III secretion system that introduces only catalytically active ExoY into the host cell cytoplasm⁴⁰. In contrast, the control ExoY^K81M is not catalytically active, but interacts directly with the actin cytoskeleton leading to impaired actin branching⁴⁰. Our second control, ΔPcrV, produces exoenzymes U and T but has the PcrV scaffold protein knocked out, leading to deposition of the exoenzymes in the extracellular space instead of in the host’s cytosol^41,42,43. Primary cultures of PMVECs will be infected with the above-mentioned mutants of P. aeruginosa. These infections will be conducted over 7 hours. Cell lysates and supernatants will be collected hourly throughout and used to measure abundance of the GSAP holoprotein and its 16kD active form by Western blotting. We will confirm association of GSAP with - and/or -secretases by co-immunoprecipitation with -secretase-1 and presenilin-1, respectively, and also confirm that it does not interact with Notch, as previously described^10,11,12,13. We will next assess the endothelial response to infection with measurements of intercellular gaps⁴⁴, macromolecular permeability^45,46,47, and LDH release¹.

Next, we will control for the 98kD and 16kD forms by generating inducible cell lines. We will utilize our GSAP knockout cells (i.e., from CRISPR-Cas9 deletion) for rescue experiments, where GSAP rescue will be engineered using the Tet-On and Tet-Off dual expression system, in which GSAP is controlled by doxycycline inducible gene expression coupled to protein stabilization by a small molecule (Shield1)^48,49,50. In this case, we will rescue expression of the 98kD and 16kD active protein independently. These constructs will be designed with FLAG carboxy-terminal tags so that the full length and truncated proteins can be detected. Studies will be conducted over a doxycycline (with and without Shield1) concentration range and time course. The time course for expression of the 98kD and 16kD proteins will be tracked. Then, the response of these cells to infection will be tested with 25%, 50%, 75% and 100% rescue of GSAP abundance. At each time point during the 7 hour infections, we will measure the abundance of GSAP and its 98kD and 16kD forms, to evaluate whether P. aeruginosa increases cleavage of the 98kD holoprotein into its 16kD active form^10,11,12. Again, we will assess the endothelial response to infection with measurements of intercellular gaps⁴⁴, macromolecular permeability^45,46,47, and LDH release¹.

Anticipated Outcomes: We first expect that infection with P. aeruginosa mutant ExoY⁺ will cause endothelial damage in PMVECs but GSAP knockout cells will be protected from the primary infection. Second, we expect an increase in the abundance of the 16kD form of GSAP paralleling endothelial barrier disruption. Third, we expect that the abundance of the GSAP 98kD form will not increase in parallel with endothelial barrier disruption.

Problems: The technical feasibility of this sub-aim has been established. We have generated cells with genetic deletion of GSAP, and we have demonstrated the ability to rescue protein expression using Tet-On and Tet-Off approaches^{48,49,51,52,53,54,55}. It is possible that it is not the 16kD form that parallels endothelial barrier disruption, and that it is instead the 98kD form that does. In this case, we will have found a previously unknown function of the 98kD GSAP holoprotein. There is a chance that re-expressing the 16kDa or 98kDa forms of GSAP will not rescue the phenotype. I.e. the cells will not begin to gap. If this is the case, we will have to look at other potential injury mechanisms such as other components of the gamma secretase complex or off target effects of the primers we chose. We will address these issues in later studies.

AIM 2: Test the hypothesis that GSAP is necessary for production of cytotoxic A.

Sub-Aim 2a. Determine whether genetic inhibition of GSAP is sufficient to abolish the production of cytotoxic A

Rationale: GSAP was recently suggested to critically target -secretase to amyloid precursor protein necessary for production of cytotoxic A¹². However, controversy has recently emerged in this interpretation, as not all cell types seem to rely on GSAP for production of A_40/42⁵³. Here, we will test whether genetic inhibition of GSAP in PMVECs is sufficient to abolish or reduce the production of cytotoxic A.

Approach: To test this idea, we will use our GSAP knockout cells along with controls and infect them with ExoY⁺, ExoY^K81M, and ΔPcrVbacteria. Over the course of the 7-hour infection we will collect lysates and supernatant hourly, and the abundance of GSAP and A_42 will be measured. We will then perform Western blots using antibodies specific for A₄₀ (Millipore Sigma ABN240) andA₄₂ (Bioss antibodies bsm-0107M) to determine which form is present. Wewill utilize our GSAP knockout cells for rescue experiments, as mentioned above. Sensitivity to infection will be assessed as mentioned in aim 1. The time course for expression of the 98- and 16- kDa proteins will be tracked, and the impact of GSAP on constitutive A production assessed. Then, the response of these cells to infection will be tested with 25%, 50%, 75% and 100% rescue of GSAP abundance. At each time point during the 7 hour infection we will measure the abundance of GSAP and its 16kDa active form, to  evaluate whether ExoY⁺ increases production or function of the 16kDa active form. We will next perform cytotoxicity assays^2,48 in which filter-sterilized supernatant from P. aeruginosa infected GSAP knockout and control cells will be added to naïve PMVECs over 48 hours and sensitivity to supernatant will be assessed with measurements of intercellular gaps⁴⁴, macromolecular permeability^45,46,47, LDH release¹, and apoptosis^56,57. We will also infect wild-type cells with the above-mentioned bacteria and over 7 hours collect the supernatant. Then we will take that supernatant and use an Aspecific antibody to neutralize the supernatant, then elute Aoff the antibody and add it directly to naïve PMVECs as previously shown³. We will do this same thing with GSAP knockout cells and compare the results to determine if the Aproduced is cytotoxic or noncytotoxic. Upon completion of this sub-aim we will have determined whether genetic removal of the GSAP gene is sufficient to abolish or reduce cytotoxic Aproduction in P. aeruginosa infected PMVECs.

Problems: The technical feasibility of this sub-aim has been established. We have generated cells with genetic deletion of GSAP, and we have demonstrated the ability to rescue protein expression using Tet-On and Tet-Off approaches^{48,49,51,52,53,54,55}. As mentioned above, there is a chance that re-expressing the 16kDa and 98kDa forms will not rescue the phenotype as we expect. If this is the case, I plan to look at other components of the GSAP complex to determine if by chance, knocking out GSAP effected the expression or activity of other components of the complex. These issues will be addressed in later studies.

Anticipated Outcome: We expect that the GSAP knockout cells will not produce cytotoxic A If this is true, then our data would suggest a principal role for GSAP in production of cytotoxic A, and further, suggest infection acts within the cell to either activate GSAP, or alternatively, to promote GSAP abundance. We think this latter possibility is most likely.

Sub-Aim 2b. Determine if inhibition of GSAP shifts the production of cytotoxic Ato a non-cytotoxic or antimicrobial form.

Rationale: Infection of PMVECs with P. aeruginosa mutant ExoY⁺causes production and secretion of a cytotoxic form of A^1,2. A_{},however A₄₂ has been found to be antimicrobial as well^2,58.Here we seek to determine whether inhibiting GSAP will shift the production of a cytotoxic form of Ato a noncytotoxic or antimicrobial form.

Approach: Infections will be performed using our GSAP knockout cells and bacterial mutants as mentioned above. Over the course of 7 hours, supernatant will be collected. Supernatant will be filter-sterilized with a .22μm filter and added back to naïve PMVECs for 48 hours. Sensitivity to supernatant will be assessed as mentioned in aims 1 and 2a. Western blots will be conducted on the supernatant and lysates to determine which form of A is present. Again, sensitivity will be measured as mentioned above. Next, we will use standard Kirby-Bauer assays to determine any antimicrobial properties that our supernatants and proteins may have.

Problems: Whereas our studies will provide mechanistic insight into the role of GSAP in production of pulmonary endothelial cell cytotoxic A, at the end of these studies we will not know the mechanism of how GSAP shifts amyloid from a noncytotoxic to a cytotoxic form. Similarly, we will not know the mechanism by which ExoY⁺ promotes this GSAP activity. These issues will be addressed in later studies.

Anticipated Outcome: We expect that inhibition of GSAP will change the form of Auced from a cytotoxic form to a non-cytotoxic or antimicrobial form.