TYK2 inhibition reduces type 3 immunity and modifies disease progression in murine spondyloarthritis

Spondyloarthritis (SpA) represents a family of inflammatory diseases of the spine and peripheral joints. Ankylosing spondylitis (AS) is the prototypic form of SpA in which progressive disease can lead to fusion of the spine. Therapeutically, knowledge of type 3 immunity has translated into the development of IL-23– and IL-17A–blocking antibodies for the treatment of SpA. Despite being able to provide symptomatic control, the current biologics do not prevent the fusion of joints in AS patients. Thus, there is an unmet need for disease-modifying drugs. Genetic studies have linked the Janus kinase TYK2 to AS. TYK2 is a mediator of type 3 immunity through intracellular signaling of IL-23. Here, we describe and characterize a potentially novel small-molecule inhibitor of TYK2 that blocked IL-23 signaling in vitro and inhibited disease progression in animal models of SpA. The effect of the inhibitor appears to be TYK2 specific, using TYK2-inactive mice, which further revealed a duality in the induction of IL-17A and IL-22 by IL-23. Specifically, IL-22 production was TYK2/JAK2/STAT3 dependent, while IL-17A was mostly JAK2 dependent. Finally, we examined the effects of AS-associated TYK2 SNPs on TYK2 expression and function and correlated them with AS disease progression. This work provides evidence that TYK2 inhibitors have great potential as an orally delivered therapeutic for SpA.

For PBMC isolation, venous blood was collected in heparinized tubes prior to cell isolation with Ficoll-Paque plus (GE Healthcare). Cells were cryopreserved with 55% FCS and 10% DMSO in RPMI and stored in liquid nitrogen. Thawed PBMC were rested overnight in complete media (RPMI 1640 with L-glut + 10% HI-FCS + penicillin/streptomycin + 50 µM mercaptoethanol) before use.
For whole blood DNA isolation, venous blood was collected in EDTA-treated tubes prior to DNA isolation with a QIAamp blood DNA isolation kit following the manufacturer's protocols (Qiagen). DNA was stored at 4C.
For whole blood RNA isolation, venous blood was collected in Tempus tubes, and RNA isolation with Spin Isolation Kits (Applied Biosystems). RNA was stored at -80C.
Synovial tissue biopsies were obtained from clinically inflamed knee or ankle joints and processed as described previously (4). 6-8 biopsies were pooled and used for RNA isolation. Total RNA was extracted from synovial biopsies by homogenization of biopsies in STAT60 (Tel-Test) according to manufacturer's instruction, treated with DNAse, and purified using RNeasy columns (Qiagen).

Murine disease models, treatment and clinical scoring
SKG and BALB/cAJcl mice were obtained from CLEA Japan and were bred in homozygous harems. Mice were cohoused from different breeding cages upon weaning and only female SKG mice were used for arthritis experiments owing to higher disease severity and penetrance vs. male mice (5,6). Male BALB/cAJcl mice were used for ear dermatitis and γδ in vitro assays. Male B10.RIII mice were obtained from Jackson Laboratories for minicircle experiments. TYK2 K923E mice were obtained from M. Müller (Vienna) (7), and bred in homozygous harems for ear dermatitis experiments. Male C57BL/6NCrl were obtained from Charles River as controls. Il17a cre mice (Il17atm1.1(icre)Stck/J) and ROSA26 dTomato (B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J) were obtained from Jackson and male F1 offspring (IL-17 cre.dTomato ) were used for ear dermatitis experiments.
For in vivo experiments involving NDI-031407 (Nimbus Therapeutics), compound was prepared fresh weekly in autoclave-sterilized vehicle (5 g/L 4000cps methylcellulose [Sigma], 10 g/L NaCl, pH 5 in deionized water). Up to 100 mg/kg NDI-031407 was administered oral BID by gavage. Pharmacokinetic experiments performed by Nimbus revealed this dose to provide adequate serum NDI-031407 saturation over a 12 hour period (Supplementary Figure 14). Tofacinib (MedChemExpress) was similarly prepared and administered.
Disease was induced in SKG mice by IP injection of 3 mg curdlan (C7821, Sigma) in PBS. Mice were weighed and scored weekly for clinical signs of SpA. A composite SpA score was created based on presence or absence of: blepharitis (0.5 points/eye), dermatitis (0.5 for any ear, 1 for tail), swollen/red ankles and paws (0.5 points/limb) and swollen paw/foot digits excluding pollex/hallux (0.1 point/digit). The maximum score possible is 6.1. At the endpoint, lymph nodes were dissected for FACS, one ankle was dissected and skin/toes removed prior to storage in RNA later for qPCR, the other ankle was kept intact and fixed in 10% neutral buffered formalin (HT501128, Sigma). The upper tail, pelvis and distal small intestine were dissected and fixed in formalin.
Disease was induced in B10.RIII mice by hydrodynamic delivery of 6 µg of IL-23 expressing minicircle (System Biosciences) as published (9,10). Mice were scored every 3 days using the aforementioned SpA scoring system. At one week post-minicircle administration, serum was obtained from saphenous vein and IL-23 assessed by ELISA (BioLegend) to ensure treatment groups had comparable average serum IL-23 concentrations. Tissue was harvested as above for respective analyses.
Intradermal IL-23 induced inflammation was performed as previously reported (11). Mice were first anaesthetized with isoflurane prior to dorsal intradermal injection of 400 ng IL-23 in 20 µl PBS using a 31 gauge, 0.3 ml insulin syringe. For assessment of cytokine in whole ear, ears were snap frozen in liquid nitrogen prior to homogenization with protease cocktail (1 mM Benzamidine hydrochloride, 1 mM PMSF, 1 µg/ml pepstatin, 0.5 µg/ml aprotinin, 0.5 µg/ml leupeptin). IL-17A assessed in ear homogenate with the Luminex assay (R&D Systems). For assessment of skin cells by FACS, mice were first treated with 300 µg BFA (Sigma) in 600 µl PBS for 5 hours prior to tissue harvest. Ears were digested in RPMI with penicillin/streptomycin, 0.25 mg/ml Liberase TM (Roche) and 1 µl/ml BFA (BioLegend) for 60 mins at 37C in a shaking incubator. After incubation, digested tissue was titrated with a 21 gauge needle, prior to 35 µm filtration. Washed cells were treated with 0.2 mg/ml DNAse in RPMI for 5 minutes at room temperature to obtain single cell suspension for FACS.

In vitro murine and human assays
For cell-based JAK activity assays, human PBMCs were pre-incubated with NDI-031407 for one hour in RPMI containing 10% fetal bovine serum, at 37 o C in humidified incubator with 5% CO2. The mixture was then stimulated with IL-12 (R&D Systems) for 30 minutes or GM-CSF (R&D Systems) for 10 minutes. The stimulation was terminated by adding cold PBS and pelleting the cells. Meso Scale Discovery (MSD) lysis buffer was then added to the cell pellets and the lysate analyzed for phospho-and total-STAT4 (for IL-12) or STAT5 (for GM-CSF) per manufacturer's instruction and read on MSD Sector Imager 6000. For the NK-92 cell assay, cells cultured in αMEM without ribonucleosides (STEMCELL Technologies) with 2mM Lglutamine, 0.2 mM myo-inositol (Sigma), 0.1 mM 2-mercaptoethanol, 0.02 mM folic acid (Sigma), 12.5 % horse serum (Sigma), 12.5% heat inactivated Fetal Calf Serum (Gibco) and 100 U/ml recombinant IL-2 (R&D System). For assay, cells were IL-2 starved overnight and NDI-031407 was pre-incubated with the cells for one hour following by IL-12 stimulation for 24 hrs. At the end of incubation cells were pelleted and IFN in the supernatant was measured by human IFN MSD kit per manufacturer's instructions.
For human in vitro Th17 assays, PBMC were obtained from blood as mentioned above and used fresh. For supernatant cytokine IL-17A detection by ELISA (BioLegend), whole CD4+ T cells were obtained by magnetic selection (Mojosort, BioLegend). Cells were then cultured in XVIVO15 serum-free media (Lonza) with anti-CD2/CD3/CD28 beads (Miltenyi) at a 1 bead: 2 cell ratio, supplemented with 200 U/ml IL-2 (Genscript) for four days. Where indicated cells were further treated with 20 ng/ml IL-1β, 20 ng/ml IL-6 and/or 20 ng/ml IL-23 (all from BioLegend) for the duration of the experiment. Tofacitinib (Selleck Chem), ruxolitinib (Selleck Chem) and NDI-031407 were prepared in DMSO at the indicated concentrations and added for the duration of the experiment. For proliferation and cell viability assessment, cells were stained with Tag-it Violet (BioLegend) before stimulation, then stained at endpoint with Helix NIR and Annexin V FITC (BioLegend) for FACS analysis. For STAT phosphorylation assays, whole PBMC were activated with anti-CD2/CD3/CD28 beads for four days in complete RPMI. Activated cells were then rested in serum-free RPMI for 3 hours, with fixable live/dead staining added in the final 15 mins. Cells were washed in serum free media and pre-treated with JAKinibs at the indicated concentrations for 30 minutes prior to stimulation with 400 ng/ml IL-23 or IL-6 for 15 minutes. H2O2 activated pervanadate (P0758S, NEB) was used as a positive control for STAT phosphorylation. Cells were fixed by addition of PFA (BioLegend) directly to stimulated cells and stained for FACS.
For murine in vitro assays, popliteal, inguinal, lumbar, brachial and cervical lymph nodes were homogenized and pooled. For cytokine assays, lymphocytes were pretreated with brefeldin A (BFA, BioLegend) and JAKinib for 30 minutes prior to a 4.5 hour treatment with 20 ng/ml IL-1β (BioLegend) and/or 20 ng/ml IL-23 (RnD) in complete RPMI. PMA/ionomycin (BioLegend) was used as a positive control.
For pSTAT assays, lymphocytes were rested in serum free media for 3 hours before treating for 15 minutes with 400 ng/ml IL-23 or activated pervanadate as above.

SNP analysis
TYK2 SNPs were assessed with off-the-shelf Taqman assays and assessed using a 7900HT (ABI) as per the manufacturer's recommendations. SKG genotyping was performed using a custom Taqman SNP assay: Forward primer CAGGTGGAGAAGCTCATTGCT, Reverse primer CTGGCCGGAATAGAGTTTGC, VIC probe (SKG) AATGCCCTGTTATGAC, FAM probe (WT) ATGCCCTGGTATGACA.

Gene expression
RNA was extracted from human whole blood as discussed above. RNA extracted from sorted human immune cells with TRIzol (Invitrogen), following the manufacturer's protocol. RNA extracted from whole ankle in TRIzol, with Dounce homogenizer. All RNA measured by Nanodrop and used if 260/280 >1.8.
For both murine and human gene expression assays, RNA was reverse transcribed with Superscript IV pretreated with DNAse (Invitrogen). qPCR was performed with power SYBR green (ABI) on the 7900HT (ABI). Primers were designed in house with Primer Express 3 (ABI) and ordered from ACGT (Toronto). All primers were tested for specificity in silico using primer BLAST (NCBI) and by melt curve analysis. Efficiency was assessed in serially diluted samples and ΔΔCt equation used when primer efficiency was >95%, or Pfaffl equation (12) used for primers with efficiency <95%. At least 5 housekeepers were compared between treatments/patients and controls and the most stable housekeeper gene selected for the respective experiments. Primer list provided in Supplementary Table 4.

Flow cytometry
For all panels, except for phosphoflow experiments as discussed above, single cell suspensions were first stained with a fixable live dead stain in NIR spectrum (L/D NIR (Invitrogen) or FVS780 (BD)) as directed by the manufacturers. Cells were blocked with FcX (BioLegend), prior to staining with surface antibodies. For PrimeFlow experiments, cells were fixed, permeabilized and mRNA probe hybridization carried out following the manufacturer's protocol (eBiosciences). For experiments in which transcription factors were stained, cells were fixed and permeabilized with True-Nuclear kit (BioLegend) as directed. For experiments in which cytokines were stained, cells were fixed with a PFA buffer and permeabilized with intracellular staining buffer (BioLegend) as indicated. BFA (BioLegend) with or without PMA/ionomycin (BioLegend) was used for in vitro stimulations to detect cytokine. For phosphoflow experiments, cells were fixed in PFA and permeabilized with True-Phos perm buffer (Biolegend) as indicated. Antibodies used are listed in Supplementary Tables 5 and 6. Data acquired on Aria III, Canto II or Fortesa x20 (BD) and analyzed with FlowJo. For cell sorting, cells were sorted into RPMI with 50% FCS on ice. Sorted cells were pelleted and RNA immediately extracted by TRIzol.

In vivo imaging
MRI was performed on isoflurane anaesthetized mice using a Bruker 7T system at 60 µm resolution (STTARR, University Health Network, Toronto). 2D FLASH (4.5 ms echo, 750 ms repetition, 5 averages) was used to acquire T1 weighted scans. 2D rare (factor 8, 48 ms echo, 3600 repetition, 6 averages) was used to acquire T2-weighted scans. 7 slices of 0.5 mm each were acquired in the axial plane, centered on the SIJ and two slices nearest to mid SIJ were scored. Images analyzed with MIPAV.
Bones were dissected at endpoint and fixed in 10% NBF for 3 days prior to storage in PBS. Imaging was performed on Sanco µCT100 with 10 µm resolution (70 kV/114 µA, aluminum filter 0.5 mm and 1200 mgHA/ccm beam hardening) (Sunnybrook Hospital, Toronto). The scans were analyzed with Image J.

Statistics
All statistical analysis performed with GraphPad Prism. Data tested for normality before statistical test selected. All graphs show mean with standard error of mean, unless indicated otherwise. Where applicable, individual points represent separate patient, animal or treated well. Two-tailed statistical tests used, with specific test indicated in the respective figure legends. For all graphs: A P value less than 0.05 was considered significant. Not significant (n.s.), p>0.05; (*) 0.  Figure 9: Extended data for model of local IL-23 inflammation. A) IL-23 and PBS delivered intradermally to contralateral ears in a vehicle/Brefeldin A treated mouse from the experiment depicted in Figure 5. Draining lymph node (CLN) homogenates were tested by flow cytometric analysis for the indicated cytokines in T cells. Experiment repeated in NDI-031407-treated and TYK2 K923E mice with similar results (not shown). B) C57BL/6 (WT) or TYK2 K923E mice injected intradermally with PBS or IL-23 as above. 100mg/kg NDI-031407 or methylcellulose oral BID started one day prior to the first IL-23 injection and continued daily. Ears harvest 5 hr after final IL-23 injection for cytokine analysis by luminex assay as discussed in the methods. Each datapoint is a single mouse. in IL-12 induced IFNγ assay with mouse whole blood. PK study was conducted in C57BL/6 mice dosed by oral gavage with NDI-031407 in 0.5% methylcellulose. Plasma was prepared from blood drawn at indicated time points and analyzed for NDI-031407 concentration using LC/MS. Mouse whole blood assay was performed by incubating the compound with blood for one hour followed by IL-12 stimulation for 24 hours on anti-CD3 antibody coated plate. At the end of the incubation, IFNγ level in the supernatant was quantified by mesoscale discovery assay (not shown).