0010: A Potential Mechanism for Major Adverse Cardiac Events Associated with JAK Inhibitors: JAK Inhibitor Withdrawal Causes Urokinase Release by Primed STAT Signaling

Sunday, November 12, 2023

9:00 AM – 11:00 AM PT

Location: Poster Hall

Abstract Poster Presenter(s)

Sara McCoy, MD, PhD, RhMSUS

University of Wisconsin
Middleton, WI, United States

Disclosure(s): Bristol-Myers Squibb(BMS): Consultant (Ongoing); Horizon: Consultant (Ongoing); Kiniksa: Consultant (Ongoing); Novartis: Consultant (Ongoing); Otsuka/Visterra: Consultant (Ongoing); Target RWE: Consultant (Ongoing)

Sara McCoy¹, Ilya Gurevic² and jacques Galipeau², ¹University of Wisconsin School of Medicine and Public Health, Middleton, WI, ²University of Wisconsin, Madison, WI

Background/Purpose: Sjögren’s Disease (SjD) has high glandular IFNg levels, associated with disease activity and lymphoma risk. We showed that IFNg-stimulated minor salivary gland (SG)-mesenchymal stromal cells (MSCs) produced CXCL-9, -10, and -11 through JAK2→STAT1 phosphorylation. Chemokine production and STAT1 phosphorylation by SG-MSCs were inhibited by the JAK1/2 inhibitor, ruxolitinib, a potential promising therapy for SjD. Using culture adapted SG MSC as an in vitro rosetta stone of SG-resident parenchymal cells, we investigated the IFNg →JAK/STAT biochemical response of SG-MSCs to FDA approved JAK inhibitors.

Methods: We treated human SG-MSCs with IFNg and JAK1/2 inhibitors. The JAK inhibitors included ruxolitinib and baricitinib, which bind the JAK conformation poised to phosphorylate its STAT substrates (kinase active conformation). We also studied the novel JAK inhibitor, CHZ868, which binds the kinase inactive JAK conformation. We performed western blotting of cell lysates, probing for JAK1/2 or STAT1/3. We performed RNA sequencing and confirmed results with qPCR and ELISA. ANOVA or paired t-tests were used to calculate p-values across multiple variables.

Results: We determined that ruxolitinib treatment increased phospho JAK1 (pJAK1) and phospho JAK2 (pJAK2) of IFNg-stimulated SG-MSCs in a dose-dependent manner (Fig 1a-b). pSTAT1 and pSTAT3 were suppressed by ruxolitinib upon treatment (Fig 1c-d). Discontinuation of ruxolitinib resulted in decreased pJAK1/2 (Fig1e). We found a marked and rapid increase of pSTAT1 upon discontinuation of ruxolitinib and baricitinib, but not CHZ868 (Fig 1f). Next, we wanted to determine the downstream effects of this pSTAT cascade. We performed RNA-seq of three conditions: 1) IFNg throughout; 2) IFNg with ruxolitinib, discontinue ruxolitinib, and collect conditioned media 3 hours later; 3) IFNg with CHZ868, discontinue CHZ868, and collect conditioned media 3 hours later. We found that ruxolitniib withdrawal was associated with increased PLAT and PLAU expression when compared to IFNg treatment or CHZ868 discontinuation (Fig 2a-b). We confirmed that PLAU (plasminogen activator, urokinase [uPA]) increased in response to ruxolitinib discontinuation but not in control or CHZ868 discontinuation by qPCR (Fig 3a-b) and ELISA (Fig 3c).

Conclusion: Our findings suggest that ruxolitinib and baricitinib bind the active phosphorylated form of JAK and lead to a paradoxical cellular accumulation of functionally defective pJAK. Upon inhibitor withdrawal, the primed pJAKs are de-repressed and initiate a pSTAT signaling cascade, ultimately resulting in high levels of uPA. In contrast, CHZ868, which binds the inactive JAK kinase conformation, does not lead to pJAK accumulation, a pSTAT cascade, or uPA production. CHZ868 does not cause this phenomenon. High uPA, associated with human atherosclerotic plaques, might cause proteolysis and plaque rupture. In conclusion, JAK1/2 inhibitors that bind the active pJAK configuration, but not those that bind the inactive JAK configuration, increase STAT signaling on withdrawal. This signaling results in high uPA and might promote plaque rupture (Fig 3d).

Figure 1. Withdrawal of ruxolitinib and baricitinib, but not CHZ868 causes transient increased phospho STAT1. We treated salivary gland-derived MSCs with vehicle, IFNγ (10 ng/mL), ruxolitinib (1000 nM), baricitinib (1000 nM) or CHZ 868 (1000 nM). A)Total JAK1 remains stable with increasing concentrations of ruxolitinib whereas phospho JAK1 increases with increasing ruxolitinib concentrations; B) total JAK2 peaks at 50 nM ruxolitinib, subsequently decreasing, whereas phospho JAK2 increases with ruxolitinib concentration; C) Total STAT1 remains stable with increasing ruxolitinib concentrations, whereas phospho STAT1 is inversely correlated with ruxolitinib dose; D) Total STAT3 decreases with increasing ruxolitinib concentrations, whereas phospho STAT3 is inversely correlated with ruxolitinib dose; E-F) MSCs treated with IFNγ alone served as controls in lanes 1-4. After treatment of MSCs with IFNγ and either ruxolitinib, baricitinib, or CHZ868 for 24 hours, we withdrew each JAKI and harvested cells at 15-180 minutes. Phospho JAK decreases with ruxolitinib withrawal as expected. Total STAT1 remains stable, whereas phospho STAT1 increases with ruxolitinib withdrawal, peaking at 60 minutes post-withdrawal. A similar trend is seen with baricitinib withdrawal, but not CHZ868 withdrawal.

Figure 2. RNA Sequencing identifies PLAT and PLAU and genes differentially regulated upon ruxowithdrawal.
Salivary gland MSCs were treated with IFNγ (10 ng/mL), ruxolitinib (1000 nM), or CHZ 868 (1000 nM) and
then the either ruxolitinib or CHZ868 were withdrawn. Cells were harvested 6 hours later for RNA Sequencing.
A) Compared to continuous IFNγ treatment, ruxolitinib withdrawal causes differential expression of genes associated with keratin sulfate, hemostasis, coagulation, wound healing, T cell activation, response to mechanical stimulus, organ growth, wounding response, mucopolysaccharide metabolic process, and carbohydrate derivative catabolic process. Genes associated with blood coagulation include PLAT and PLAU; B) Ruxo withdrawal compared to CHZ868 withdrawal revealed differential expression of genes related to renal systems, body fluid, organ growth, type I interferon response and signaling pathway, extracellular structure/matrix, and collagen fibril organization. PLAT and PLAU were upregulated in the ruxo withdrawal compared to CHZ868 withdrawal.

Figure 3. Confirmation that PLAU (plasminogen activator, urokinase [uPA]) increases after type I withdrawal. MSCs were treated with IFNγ (10 ng/mL), ruxolitinib (1000 nM), or CHZ 868 (1000 nM). Withdrawal was performed by treating cells with IFNγ plus JAKI, removing the JAKI by washing cells, and adding back media with IFNγ. After three hours, conditioned media were collected. A-B) qPCR of PLAT (plasminogen activator, tissue type) and PLAU show ruxo withdrawal increases expression of these genes compared to controls. This response was more prominent with PLAU than PLAT; C) uPA ELISA of conditioned media from cells treated with IFNγ treatment, ruxo discontinuation, CHZ868 discontinuation, or continued treatment with IFNγ and ruxolitinb

S. McCoy: Bristol-Myers Squibb(BMS), 2, Horizon, 2, Kiniksa, 2, Novartis, 2, Otsuka/Visterra, 2, Target RWE, 2; I. Gurevic: None; j. Galipeau: None.