Selective Pharmaceutical Inhibition of PARP14 Mitigates Allergen-Induced IgE and Mucus Overproduction in a Mouse Model of Pulmonary Allergic Response

During allergic responses, IL-4 and IL-13 production increase due to expansion of the population of allergen-specific Th2 lymphocytes as well as an expansion of group 2 innate lymphoid cells. These pleiotropic cytokines act via IL-4Rα receptor complexes on cell targets that include B lymphocytes, alternatively activated macrophages, and airway cells in the lung (1–8). The transcription factor STAT6, induced by IL-4 or IL-13 binding to IL-4Rα, plays a critical role in IgE production, T cell differentiation to functional subsets, chemokine secretion by lung epithelial cells, and in promoting pathological levels of mucus production (6, 8). The choice of Ig H chain C region class during class switching in B cells is directed by IL-4, which is essential for generation of IgE, can promote IgG1, and suppresses selection of the Cγ2a isotype for IgG2a production (9). The switch to IgE is directed by activation of the promoter for a noncoding germ line transcript termed Iε, which is inducible by IL-4 and initiated by binding of STAT6 to the Iɛ promoter (10, 11). Although normally present at very low levels, IgE is substantially increased by atopy and in allergic inflammation, during which it functions by cross-linking FcɛRI on mast cells and basophils (12).

A screen for transcriptional cofactors that modulate the function of STAT6 in these diverse processes identified poly(ADP-ribose) polymerase (PARP)14 (alternatively termed ARTD8) as an interaction partner that binds to the activation domain of STAT6 but not the IFN-γ–induced transcription factor STAT1 (13). Analyses of mice lacking detectable expression of PARP14 after gene inactivation provided evidence that allergic lung inflammation provoked by the model Ag OVA and OVA-specific IgE responses are decreased in the absence of PARP14 (14, 15). Moreover, the domain of PARP14 homologous to mammalian PARP1 and to ADP-ribosyltransferases (ARTs) encoded by bacterial exotoxins is enzymatically active (16). Unlike the members of the PARP gene family that function as processive polymerases, PARP14 is thought to transfer only one or a few ADP-ribose (ADPr) moieties onto protein acceptor sites (17). In addition to the ART catalytic domain, however, PARP14 contains several other functional modules (13, 18, 19). As such, whether there are functional consequences of the ADP-ribosylation executed in vivo by this protein, or if so what are they, remain unanswered questions.

One approach to enable identification of physiological functions of PARP14 in the intact animal would involve administration of a highly selective inhibitor of its ADP-ribosylation activity. This approach was not realized for many years due to challenges inherent in generating such a compound (20–22). The posttranslational modification of proteins by ADP-ribosylation involves the transfer of ADPr from NAD⁺ to an amino acid acceptor in a target protein. Mammals express extracellular as well as intracellular ARTs, and they also encode a variety of enzymatic activities that remove ADPr from modified proteins (23–25). Intracellular ARTs in mammals include those that can form branching polymers by iterative additions of ADPr to ADPr, that is, polymerases or poly-PARPs (e.g., PARP1 and PARP2), and those that do not polymerize, that is, mono-PARPs or mono-ARTs such as PARP14 (23–25). Among the 17 intracellular PARP family proteins, the NAD-binding pockets are structurally similar. Attempts to use existing PARP inhibitors as surrogates suffer from the drawbacks that their IC₅₀ for PARP14 requires the use of concentrations that both are more inhibitory for poly-PARP enzymes such as PARP1 and PARP2 and from cellular toxicities at the concentrations affecting PARP14 (14, 26). Moreover, Parp1 gene inactivation studies as well as use of PARP1/PARP2-specific inhibition using olaparib suggest that OVA-induced allergic lung inflammation depends on the polymerase PARP1 (27–31). This confounding variable makes interpretations of results with broad-spectrum inhibitors such as PJ-34, which only weakly inhibits PARP14, problematic because the coverage of PARP1 will exceed that of PARP14. Recently, however, RBN012759 was identified as a gastroenterically absorbable small molecule (i.e., an “oral absorbable agent”) that selectively inhibits the catalytic activity of PARP14 with workable pharmacological properties such as absorption, in vivo distribution, tissue penetration, metabolism, and excretion (32, 33). Accordingly, we tested whether administration of this compound could alter any functional aspects of the immune response to an aeroallergen.

To do so, we selected a mouse model that entails administration of an asthma-relevant aeroallergen. Airway inflammation is elicited with a known aeroallergen, Alternaria alternata, in one class of such models. This fungal species elicits an IgE-mediated respiratory disease and is a trigger for flares of asthma and allergic disease in patients (34, 35). The primary question of this study was whether aspects of the allergic process in vivo would require the ART activity of PARP14. Nonetheless, therapeutic potential of a pharmaceutical in patients usually hinges on its effect after initial immune experience. Accordingly, we tested how the highly selective PARP14 inhibitor (PARP14i) influenced IgE levels, mucus formation, and other features of allergic response elicited by inhaled rechallenge with Alternaria when administered after completion of a priming phase.

Male and female BALB/c-J mice (The Jackson Laboratory) were housed in ventilated microisolators under specified pathogen-free conditions in a Vanderbilt University Medical Center mouse facility and used at 6–8 wk of age following protocols approved by the Institutional Animal Care and Use Committee. In a follow-up replication study to test for independent validation of the main conclusions, contract studies sponsored by Ribon Therapeutics were conducted at Pharmidex (Hatfield, U.K.) facilities at the European Knowledge Centre under United Kingdom Home Office project license P8AE03703. For this later work, BALB/c mice (Charles River Laboratories, Margate, U.K.; 20–30 g on arrival) were used after a week of cohousing to acclimatize, but experiments were limited to males. All animals throughout the study were held in a biological support unit at Pharmidex facilities, which operates under the United Kingdom Home Office animal scientific procedures act and has a valid establishment license (X07D13023) allowing regulated animal procedures to be carried out in the facility. Mice had access to water and standard chow ad libitum. No emergency care was needed during the study, and no animals had to be removed on welfare grounds during this study.

To prime mice for recall allergic response, Alternaria extract (lot no. 338869, Greer Laboratories, Lenoir, NC; 16.72 EU/ml endotoxin [in 3 mg (protein)/ml of Alt], 13.4% protein in lyophilized cake, with protease activity of 1700 relative fluorescence units at a 1:40 dilution of 3 mg (protein)/ml] was administered in PBS by intranasal instillation (50 µl) once daily for 5 d. Because lots vary in their characteristics, initial testing identified 5 µg as a dose that elicited sufficient type 2 inflammation (data not shown). Two weeks later, mice were rechallenged once daily for 3 consecutive days with the same dose and route of administration. RBN012759 (PARP14i) (33), a compound generated by Ribon Therapeutics (Cambridge, MA), was delivered in a vehicle of 0.5% methylcellulose, 0.2% Tween 80 dissolved in sterile water. Four days prior to Alternaria rechallenge, mice began receiving PARP14i (500 mg/kg) by oral gavage twice daily, a dose determined experimentally and reported previously (33). Peak and terminal trough plasma samples were collected 2 h after the penultimate gavage and the day following the final intranasal challenge and gavage, respectively, by using 15% (w/v) potassium EDTA. Frozen portions of these samples were used for Ab assays and pharmacokinetic analysis. Levels of RBN012759 compound in plasma samples at these times were analyzed (Charles River Laboratories) using liquid chromatography and mass spectrometry. The left mainstem bronchus of each harvested mouse was clamped to isolate the left lobes, followed by bronchoalveolar lavage (BAL) performed on the right lung lobes using sterile PBS (700 µl). The right lung lobes were used to prepare a cell suspension by digestion with collagenase (1.5 mg/ml) and hyaluronidase (1.0 mg/ml) as described (14), followed by flow cytometry and culture. Left upper lobes were placed in formalin overnight and then embedded in paraffin. The remaining tissue was snap-frozen and stored at −80°C for later isolation of RNA. Immunization and pharmaceutical treatment in the validation studies used A. alternata extract (Stallergenes Greer USA) diluted in sterile PBS to 5 µg of Alternaria extract per 40 µl. Mice under isoflurane anesthesia were challenged by intranasal instillation (40 µl per mouse per day, alternating microdrops between each nostril) on days 1–5 and 23–25. RBN012759 (provided by Ribon Therapeutics, Cambridge, MA) and 1-aminobenzotriazole (25 mg/kg, Sigma-Aldrich UK, catalog no. A3940) were formulated in a vehicle of 0.5% methylcellulose (Sigma- Aldrich, catalog no. M0262) and 0.2% Tween 80 (Sigma-Aldrich UK, catalog no. P1754) in distilled water. On day 19 the first doses of 1-aminobenzotriazole (10 ml/kg, twice daily) were administered 2 h before PARP14i or vehicle control. All subsequent doses (days 20–26) were formulated together and administered in combination with the final dose 2 h before mice were harvested on day 26 (as diagrammed in Supplemental Fig. 1A). All dosing (twice daily) was based on a 12-h cycle with the schedules calculated so as to ensure that treatments were to be administered 1 h before the final Alternaria challenge on day 25, after which (day 26) mice were harvested for analyses.

Serial sections (5-µm thickness, positioning two sections per slide) of formalin-fixed, paraffin-embedded lung tissue were deparaffinized with xylene and then stained using periodic acid–Schiff (PAS; 9162B, Newcomer Supply, Middleton, WI). To score airway mucus semiquantitatively, three fields were selected for each sample in a manner masked as to sample identity. Images of all fields were then scored independently by three individuals masked as to sample identity. A scale quantified as 0–3 (0, no excess mucus; 1, marginal and occasional hyperplasia and mucus; 2, substantial and moderate mucus cell hyperplasia and some airway mucus; 3, severe hypersecretion and airway plugging) was used, and average scores for a given sample were generated from the three individual scores, which most often were perfectly concordant and at most differed among each other by 1. The average score for each subject was then used to calculate a mucus hypersecretion index. H&E-stained slides were analyzed by a board-certified veterinary pathologist masked to the experimental manipulation. The following parameters were scored: peribronchiolar inflammation and extent as well as perivascular inflammation and extent. (Interstitial infiltrates were not scored, as the lungs were not insufflated prior to fixation.) Inflammation parameters were scored on a 0–3 scale with 0 indicating absence of lesions, 1 being mild, scattered inflammation, 2 being moderate, multifocal inflammation, and 3 being severe, marked, coalescing inflammation. The extent of each type of inflammation was based on how much of the lung section was affected, with 0 being not at all, 1 being <25% of the section, 2 being 25–50% of the section, and 3 being >50% of the section affected. Differential counts of BAL fluid (BALF) samples, performed by a reader masked as to sample identity, were analyzed after cytospin deposition on microscope slides and Richard-Allan Scientific three-step staining (no. 3300, Thermo Fisher Scientific) (36). For histology, BALF, and cytometry in the validation studies by Pharmidex, mice were euthanized and each trachea was then isolated by a midline incision in the neck and separation of the muscle layers immediately after collecting terminal blood samples. A small incision was made into the trachea and a plastic cannula was inserted and secured in place with a suture. The airway was then lavaged to collect BALF by flushing out the lungs using 0.5 ml of PBS. This procedure was repeated until the recovered volume was 1.6 ml. The BALF was then centrifuged (1500 rpm for 10 min at 4°C) and the supernatants were aliquoted (400 µl) at −80°C for cytokine analyses. The cell pellets were resuspended in 1.6 ml of PBS, after which the BAL cells were analyzed for total and differential numbers using an XT-2000iV analyzer (Sysmex). Results were expressed as cells/ml, with cell types differentially classified as neutrophils, eosinophils, lymphocytes, or macrophages. Following BALF collection the thoracic cavity was opened to expose the lungs, which were dissected free of the animal. The right and left lung lobes were placed into separate sterile containers containing 10% neutral buffered formalin for 48 h before being transferred to 70% ethanol for tissue processing and mucus scoring. Following fixation of the left lung lobe tissue and processing in paraffin wax, sections (5 µM) were transversely cut, mounted on slides, and stained with PAS before being analyzed using digital imaging. Airway mucus production was quantified from the ratio of mucus-positive epithelium to total epithelium as the percentage of the surface covered in mucus using an area quantification algorithm (Halo image analysis software).

Secreted IL-4 was measured using a matching Ab pair (capture, Tonbo Biosciences, no. 70-7041-U500; detection, eBioscience, no. 13-7042-85) with color generated using streptavidin-HRP (R&D Systems, no. Dx998) followed by ultra tetramethylbenzidine reagent (Pierce; Thermo Scientific, no. 34028). Both the purified and biotinylated Abs were used at a concentration of 0.5 µg/well. Supernatants were collected from 1.0 × 10⁶ lung cells/ml media restimulated overnight with plate-bound anti-CD3 (adsorbed to wells at 1.0 µg/ml in PBS) and soluble anti-CD28 (1.0 µg/ml) (Tonbo Biosciences), as previously described (14, 37). Dilution points for the ELISAs were selected based on the linear range of standard curves with purified recombinant cytokine. Cytokine levels in restimulated lung suspension also were measured by Th1/Th2/Th17 cytometric bead array (BD Biosciences, no. 560485). Relative concentrations of circulating Abs were determined using the anticoagulated plasma collected at the time of mouse harvest. High-affinity binding plates were coated with purified Ab (0.5 µg/well) directed against Ig H+L (SouthernBiotech) for IgG classes or IgE (BD Pharmingen, no. 553413) or with Alternaria extract (0.1 µg/well). Quality control testing indicated that only anti-Alternaria IgG1 yielded a reliably specific signal. In the validation studies by Pharmidex, terminal blood samples were collected from the lateral tail vein and placed into a serum tube on day 26, 1 d after the final Alternaria challenge. Each serum sample was kept at room temperature for 45 min to allow coagulation, before being centrifuged (2000 × g, 15 min at 4°C), from which the resulting supernatant was extracted, aliquoted, and stored at −80°C until analyzed for IgE. IgE concentrations in sera were measured using an ELISA kit (Invitrogen, Paisley, U.K., no. EMIGHEX5) per the manufacturer’s instructions. OD was measured at 450 nM using a microplate reader (SpectraMax 340PC). Concentrations of IgE were determined using SoftMax Pro v6.4 (Molecular Devices). Data were reported as mean (±SEM) IgE concentrations (ng/ml). IL-4, IL-5, and IL-13 concentrations in BALF supernatants and lung homogenates (all groups) were measured using magnetic multiplex assays (Bio-Techne, Abingdon, U.K., no. LXSAMSM) per the manufacturer’s instructions. Levels were measured using a Magpix system (Luminex), with data reported as means (±SEM) cytokine concentrations (pg/ml),

Total RNA was extracted from frozen lung tissue using TRIzol (Invitrogen) and a Mini-BeadBeater 96 (BioSpec). RNA concentration and purity were measured using a NanoDrop. cDNA was synthesized from RNA (4 µg) using AMV (avian myeloblastosis virus) reverse transcriptase (Promega), as described (14, 37). Gene expression was quantified using PowerUp SYBR Green master mix (Qiagen, Valencia, CA) via quantitative real-time PCR. Mucus-producing hyperplasia was estimated using Muc5ac primers with gene expression normalized to β-actin.

Protein was extracted from splenocyte suspensions using modified 10 mM Tris (pH 7.4), 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 1% sodium deoxycholate, and 0.1% SDS, supplemented fresh with 0.1 M master protease inhibitor (Sigma-Aldrich, no. P8340). A full-length PARP14 construct encoding the wild-type mouse protein was described previously (36). Lysates of ΦNX cells (also known as Phoenix cells, a derivative line from 293 human embryonic kidney cells) (14) transfected with wild-type of pcDNA3-FLAG-PARP14, as described (14), were either analyzed without further processing or used for immune precipitations with monoclonal anti-FLAG (M2) (Sigma-Aldrich). The resulting immune complexes, collected using protein G beads (Santa Cruz Biotechnology, Santa Cruz, CA), were rinsed, and eluted proteins were analyzed by immunoblotting. Filters were blocked with 5% milk in 0.05% Tween 20 and incubated (2 h at 20°C) with primary Abs directed against the FLAG epitope tag, mouse actin, mouse PARP14, and poly-ADPr (PAR)/mono-ADPr (MAR) (Cell Signaling Technologies, no. 83732S) and rinsed. After incubation with secondary Abs (IRD800 or IRD680) and rinsing, indirect immune fluorescent bands were visualized and quantitated using an Odyssey imaging system (LI-COR Biosciences, Lincoln, NE) as described (37).

B cells were purified (90–95%) by negative selection with single-cell splenocyte suspensions by using biotinylated anti-Thy1.2 mAb followed by streptavidin-conjugated microbeads (BD Pharmingen). In brief, 0.75 × 10⁶ cells/ml were cultured in 3 ml of RPMI 1640 medium (Life Technologies, no. 23400-021) in a six-well cell culture plate (Peak Serum, no. TR5000), supplemented with recombinant mouse BAFF (10 ng/ml). B cells were activated and plasma cell differentiation was stimulated using anti-CD40 (BD Biosciences, no. 553788) (1 µg/ml) or LPS (Sigma-Aldrich, L2630) (1 µg/ml), supplemented with IL-4 (PeproTech, no. 214-14) (10 ng/ml) and IL-5 (PeproTech, no. 215-15) (10 ng/ml) as indicated. RBN012579 was dissolved in 100% DMSO for stock solutions (1 mM; 368 µg/ml). Based on experimental data later reported in Schenkel et al. (33) (e.g., the section “Biochemical and pharmacokinetic characterization of RBN012759,” the Methods section, and legends to figures 4 and 5), cultures were treated with either DMSO or PARP14i (1 or 0.33 µM RBN012759), added once daily. Cultures were harvested and counted on day 5; supernatants were collected for ELISA to quantify relative levels of Ab (IgG1, IgG2a, and IgE) as described (14).

Graphs and statistics were generated using Prism software (GraphPad Software, San Diego, CA). Averages were generated from results in five independent experiments (three with male mice and two with female mice) with no exclusions of subjects; dropout was only for spontaneous mortality in the course of the experiment, the rate for which was not affected by treatment with active compound as compared with vehicle. Two-way ANOVAs were performed on the datasets across full dilution curves, followed at specific points on the curves by either a Student t test or Welch’s unpaired t test, selected according to the variance between the two populations for which the null hypothesis was tested. In independent statistical analyses of the data from the validation study, intergroup deviations were statistically analyzed by a one-way ANOVA. In the case of a significant difference in the mean values among the different levels of treatment, further comparison of RBN012759-treated subjects versus the vehicle group were carried out using Student t testing and the Dunnett’s test, with p < 0.05 considered statistically significant evidence in favor of rejecting the null hypothesis.

A model of allergy induction and adaptive immune recall was used in which mice first were primed by five daily inhalations of Alternaria Ags in a sensitization phase (Fig. 1A). After 12 d, the subject animals started to receive either vehicle or PARP14i (RBN012759) followed after 4 d by three consecutive daily challenges with inhaled Ag while continuing twice-daily gavages with inhibitor or vehicle. Prior analyses of the pharmacokinetics and pharmacodynamics of this agent, as well as of its in vivo target coverage, provided evidence of an effect on PARP14 levels with a dose of 500 mg/kg twice daily, whereas the effectiveness of a 300 mg/kg dose was not established (33). Accordingly, we used the 500 mg/kg dose regimen and, because there is the potential for either inflammation or a pharmaceutical agent to influence its metabolism, we assessed the circulating concentrations of drug at the end of treatment. Plasma samples were collected at 2 h after a dose and at trough concentration just prior to morning gavage (14 h samples). Quantitation of the compound by liquid chromatography–mass spectrometry at 2 h postdose revealed mean concentrations of ∼3.3 µM (∼1227 ng/ml), albeit with significant variance (range, 497–1680 ng/ml) (Fig. 1B). Mean trough concentrations were 237 ng/ml (∼0.64 µM; range, <50–1030 ng/ml), with more than half of the subjects having a concentration ≥0.33 µM (Fig. 1B). No specific target of PARP14 in vivo has been established. Moreover, it is not clear whether the ADP-ribosylation of any downstream target is exclusively catalyzed by PARP14 because several other ADP-ribosyl monotransferases and polymerases are expressed concurrently (17, 21–25). Accordingly, we first validated the inhibitory capacity of RBN012759 by transfecting cells with PARP14 expression vector (14) or a control, culturing in inhibitor or vehicle, and immunoblotting extracted proteins with Ab recognizing mono- as well as poly-ADPr adducts on proteins (MAR and PAR, respectively) (Fig. 1C). This analysis showed that inhibitor treatment substantially reduced the automodification of overexpressed PARP14. In light of the observed inhibition, we tested whether the treatment of mice with enterically delivered RBN012759 decreased the intensity of anti-MAR/PAR bands in splenocytes of subject mice. This analysis revealed multiple bands reproducibly decreased in samples from inhibitor-treated subjects compared with controls (Fig. 1D). PARP14 expression is increased by a variety of proinflammatory stimuli that include TLR4, IFNs, and TNF-α (19). This suggested a biological impact of inhibiting ADP-ribosylation in Alternaria-exposed lung tissue might lead to reduced PARP14, which was observed (Fig. 1E). Taken together, we conclude that this regimen of peroral (gavage) dosing of mice with the tool compound RBN012759 achieved concentrations compatible with target engagement and likely had an impact in vivo, with less ADP-ribosylation of target proteins in tissues of the treated mice.

Mucus hyperproduction and plugging are major factors in the airflow obstruction in human asthma (38, 39), so we selected this parameter in advance as the primary endpoint in analyses of the allergen-rechallenged mice. Because of sex differences in responses of mice to allergens such as A. alternata, we separately analyzed male versus female mice. The histologic results showed that, as expected, allergen sensitization was essential for positive scores (i.e., when comparing Alternaria-challenged mice that received vehicle by gavage) (Fig. 2A–(D). When compared with vehicle controls, pretreatment with the inhibitor specific for PARP14 catalysis caused a reduced amount of PAS-positive cells in the airways as well as mitigating mucus in the airways of sensitized and challenged mice (Fig. 2A, representative samples; (Fig. 2B, aggregate data for all mice). Male (Fig. 2C) and female (Fig. 2D) mice exhibited quantitatively different responses to Alternaria. RBN012759 reduced mucus in both males and females, and in a few cases no hypersecretion was detected, but the median effect did not reduce scores to the level of nonsensitized controls (Fig. 2B, 2C). Similarly, when Muc5a mRNA levels that encode a protein component of mucus were measured by quantitative PCR with cDNA prepared from the RNA of one lung lobe, these were lower in Alternaria-challenged mice treated with active compound when compared with vehicle-treated controls (Fig. 2E). Based on these results, an independent analysis was then arranged to test the effects of RBN012759 using BALB/c male mice and a different batch of Alternaria extract (see Materials and Methods; Supplemental Fig. 1A). These experiments also showed a substantial reduction in airway mucus (Supplemental Fig. 1B). We conclude that the mucus overproduction after challenge with Alternaria Ags was reduced by selective in vivo inhibition of the capacity of PARP14 to execute ADP-ribosylation.

Mouse models such as these often also have increased immune cell infiltration in the parenchyma. To assess such inflammatory effects, we analyzed immune cell populations by counting and flow cytometric phenotyping of enzymatic dispersions of entire portions of lung. Numbers of cells in enzymatic dispersions of lung tissue from male and female mice in these experiments showed that PARP14i blunted increases elicited by allergic sensitization followed by inhaled rechallenge only modestly (with no p < 0.05). Perhaps due to the endotoxin content of the Alternaria extract, by flow cytometric markers neutrophils appeared to have been elicited far more than eosinophils (Table I). Modest reductions were observed for the numbers of B cells and neutrophils (polymorphonuclear neutrophils) in both males and females, with a modest reduction in CD4⁺ T cells in females, although none of these changes achieved p values <0.05 (Table I). These data suggested that inhibition of PARP14 at the recall stage of allergic lung inflammation in vivo may have a modest effect on recruitment or retention of several types of immune cell after the initial phase of priming by inhalation of Alternaria extract, but at most the effect did not achieve statistical significance. However, the magnitude of decrease in mucus overproduction was more substantial and statistically significant than were indices of inflammatory cell recruitment. Review of a sampling of H&E-stained lung sections by a veterinary pathologist yielded results consistent with these flow cytometry data, that is, similar increases in cellular infiltrates in lungs of inhibitor-treated versus vehicle-treated mice subjected to Alternaria challenge (data not shown).

In analyses of BAL of patients with active disease, atopic asthmatics often have increased eosinophils in their airways (40, 41); a similar effect is typical in mouse models of allergic lung inflammation. We analyzed differential counts of the cells recovered by unilateral BAL after clamping one mainstem bronchus. As expected, the fractions of recovered cells that were eosinophils were substantially greater in Alternaria-sensitized and rechallenged mice (Fig. 3). The type 2 inflammatory cytokines IL-4 and IL-13 are major drivers of mucus overproduction, whereas eosinophil survival is promoted by the type 2 cytokine IL-5 (8, 42–45). Interestingly, however, no substantial effect of PARP14i treatment in the postsensitization phase was observed for the BAL samples (Fig. 3) despite the decrease in mucus production in the same mice. Some evidence suggests that inhibition of the drivers of type 2 inflammation may increase Th17 responses or the neutrophil recruitment that is enhanced by IL-17 (46). Despite the decreases in lung tissue as a whole (Table I), RBN012759 treatment in these experiments did not change frequencies of neutrophils in the BALFs recovered from Alternaria-challenged mice (Fig. 3). Of note, the follow-up study for independent replication found striking reductions in the recoveries of inflammatory cells by BAL and the prevalence of eosinophils among them (Supplemental Fig. 1C) in addition to decreased mucus scores in RBN012759-treated mice. Thus, the split phenotype (greater impact on mucus than inflammatory cell populations) may be due either to batch characteristics of Alternaria extract, differences in the degree of target engagement, or both. However, in each case, mucus was reduced.

One potential explanation for the capacity of a PARP14-selective inhibitor to reduce Alternaria-induced mucus accumulation would be that although the extent of infiltration was similar, the amount of type 2 cytokine delivery capacity would be diminished. To test this possibility, equal numbers of cells in the lung suspensions were stimulated via the TCR to gauge T cell–specific effects. Activation by a combination of phorbol ester (PMA) and calcium ionophore was also performed because these stimuli can also elicit cytokines from mast cells, basophils, and innate lymphoid cells as potential sources of IL-4 or IL-13. The results show that IL-4 production was dramatically increased by allergic sensitization, and this effect was blunted by the treatment with PARP14i (Fig. 4A). Surprisingly, however, this reduction was only observed with male subjects (Fig. 4B, 4C). Consistent with the IL-4 data, secreted type 2 cytokines IL-5 and IL-13 also were reduced after restimulation of lung cells from male but not female mice (Fig. 4D, 4E). In the replication study, reduced concentrations of these type 2 cytokines (IL-4, IL-5, and IL-13) in both BALF and lung homogenates were observed for RBN012759-treated mice (Supplemental Fig. 2A and 2B, respectively). There is evidence that PARP14, by an unknown mechanism, can promote Th17 differentiation of naive T cells (14). IL-17 production by the same cell suspensions was also reduced with lung cells from male subjects treated with PARP14i after either T cell–specific or broadly acting stimuli (Fig. 4F and 4G, respectively). Nonetheless, IL-17 production after stimulation of lung cells from RBN012759-treated females was indistinguishable from the controls (data not shown). Moreover, production of the T cell–specific cytokine IL-2 and the proinflammatory cytokine TNF-α were each lower in the inhibitor-treated males and undetectable in the absence of Alternaria sensitization (Fig. 4F, 4G). These were cytokine-specific effects and not a cellular toxicity in that cytokines such as IL-10 were similar in the samples from treated, vehicle control, and nonsensitized mice (Fig. 4F, 4G). We conclude that systemic PARP14 inhibition by treatment of males with drug prior to and during rechallenge attenuates type 2 cytokines (IL-4, IL-5, and IL-13).

The induction of IgE is one of the first responses to IL-4 to be identified, and increased total IgE levels in humans are a hallmark of atopy. In parallel with its essential role in isotype switching to IgH Cε (9, 10), this cytokine inhibits switching to Cγ2a/c and levels of IgG2a/c (9). Accordingly, we analyzed the relative concentrations of Abs of these isotypes in Alternaria-challenged mice and nonimmunized controls. IgE levels in sera of both male and female mice increased dramatically with allergic sensitization (Fig. 5A–C). The allergen-induced increase was substantially blunted for female as well as male mice pretreated with RBN012759 when compared with vehicle controls (Fig. 5A–C). This impact of the PARP14i was observed again in the independent replication study (Supplemental Fig. 1D). A reciprocal effect was observed for IgG2a, whose circulating levels were reduced by allergic sensitization in vehicle-treated mice, whereas the mice subjected to PARP14 inhibition exhibited levels of IgG2a indistinguishable from nonsensitized controls (Fig. 5D). Unlike IgE, the IgG1 isotype is robustly induced even with short-term primary exposure to Ag, and circulating IgG has a far longer half-life than IgE due to selective FcRn-mediated reuptake of IgG via FcRn. Moreover, although promoted by IL-4, IgG1 induction can be IL-4–independent. Of note, IgG1 levels in sera of inhibitor-treated mice and vehicle controls did not differ (Fig. 5E). Moreover, an ELISA that allowed reliable detection of Alternaria-specific IgG1 detected no effect of RBN012759 on Ag-specific Ab (Fig. 5F). We conclude that specific inhibition of the ADP-ribosyl monotransferase activity in PARP14 impacted the Ab response in vivo. Thus, for both males and females RBN012759 interfered with the capacity to generate the STAT6- and IL-4–dependent IgE isotype upon sensitization and challenge and with the repression of IgG2a, whereas long-lived IgG1 was unaffected.

In addition to reduced IgE responses to OVA in PARP14-deficient mice, in vitro evidence supported both B cell–intrinsic and cell-extrinsic mechanisms (14). To test whether the finding of reduced IgE depended on ADP-ribosylation by PARP14, we activated B cells with either LPS or by cross-linking CD40, cultured them in IL-4 with addition either of vehicle of inhibitor, and measured the secreted IgE. RBN012759, at concentrations chosen based on work now published (33), reduced IgE production to an extent almost as substantial as the overall effect in males in vivo (Fig. 6A, 6B), including at 0.33 µM (Fig. 6E), the concentration that was the median trough level in vivo (Fig. 1B). IgG1 concentrations in the same supernatants were not reduced by the presence of inhibitor in the cultures (Fig. 6C, 6D). Although IL-4 can promote B cell proliferation and survival in a PARP14-dependent manner (47), this effect was not likely to be the basis for reduced IgE caused by RBN012759, as the compound did not substantially reduce the cell numbers of these cultures (Table II).

Using an extract of the clinically relevant aeroallergen A. alternata in a mouse model of allergic lung inflammation, we have shown that an inhibitor of catalytic function highly selective for the ART of PARP14 (alternatively designated ARTD8) is biologically active. Of note, even after having first sensitized subjects by repeated Ag inhalations, treatment with RBN012759 mitigated both the allergen-induced increases of bronchial mucus and IgE. These findings provide evidence indicating that several processes biochemically downstream from IL-4Rα and the transcription factor STAT6 depend on the ADP-ribosylation catalyzed by PARP14. Moreover, the evidence suggests that the approach of inhibiting the catalytic function of PARP14 may have therapeutic potential in patients, in that the drug was effective in a recall model in mice that were already sensitized to the aeroallergen.

Prior work had indicated that OVA-induced allergic lung responses were reduced when mice lack PARP14 (14, 15). However, PARP14 has several functional domains other than the conserved catalytic domain toward its C terminus (16, 18, 19, 23, 24). Three modules termed “macro domains,” based on similarity to the extension of a large histone variant termed macro-H2a (24, 47–49), are situated N-terminal to the catalytic domain. Crystal structures of PARP14 indicate that its macro domains can bind ADPr (18, 24). In addition, the amino terminal half of PARP14 contains RNA binding and helicase functions (19). Although these domains may contribute to the spectrum of phenotypes observed for PARP14-deficient mice or specific immune cells from them (13, 19), the findings in the present study provide direct support for substantial effects of protein ADP-monoribosylation under physiological conditions in recall allergic lung inflammation.

Treatment of OVA-sensitized Parp14^−/− mice with the NAD⁺ analog PJ34 further reduced a subset of cytokine readouts but did not affect some other cytokine levels (15). The published interpretation of this result is at best complex and fraught, as the IC₅₀ of PJ34 for PARP14 (1–10 µM) is ∼100-fold less favorable than its IC₅₀ for PARP1 and PARP2 (26). Moreover, inactivation of PARP1 either by genetic or a more specific pharmacological agent blunted OVA-induced allergic lung inflammation and type 2 cytokines (27–31). Of note, assays of RBN012759 indicated that there is a >1000-fold higher IC₅₀ for any affect on PARP1 or any other poly-PARP, and >300-fold selectivity relative to the other enzymes that catalyze ADPr modifications (33). Although the data show that peak blood concentrations of the agent in this study (∼1.3–4 µM) might transiently exceed the IC₅₀ of some other PARPs, the trough concentrations were well below the IC₅₀ of all of the ARTs other than PARP14. Moreover, the functional data, in the present study and previously (33), strongly suggest that tissue penetration of the compound is well below the blood concentration. Together with the lack of any heterozygote phenotype (S.H. Cho and M.R. Boothby, unpublished observations), these points make it very unlikely that the in vivo target of RBN012759 is either PARP1 or any of the other (ADP-ribosyl)polymerases.

The issue of which ART and PARP enzymes are subject to inhibition of catalysis bears on the risk-benefit balance therapeutic potential of a given agent in the context of specific disease targets. PARP1/2 inhibitors are approved for indications in human cancer but not without dose-limiting toxicities (50–52), for which the duration of treatment will be shorter and the natural history of untreated disease more dire than in allergic diseases or asthma. Thus, what likely are on-target toxicities or side effects do not undermine a favorable risk-benefit relationship in cancer yet would be unsuitable for asthma. The need for sufficient specificity even among mono-ARTs (also termed mono-PARPs) is underscored by work in which a PARP7-specific inhibitor enhances type I IFN responses for potential use in cancer (53). Conversely, PARP14 in marrow-derived macrophages may promote Ifnb1 gene expression and suites of type I IFN–regulated genes (54). Accordingly, a less selective mono-ART inhibitor would carry risks of enhancing innate inflammatory cytokines (55) in conditions in which type II (IL-4/5/13–driven) inflammation is the main target.

Although a large percentage of human populations has been shown to be reactive to A. alternata, patients previously diagnosed with asthma are at higher risk of hospital admission or death upon exposure to Alternaria (34, 35). Asthma, including allergen-induced forms of reversible obstructive airways disease, is a highly prevalent condition (56–59). For example, ∼25 million people in the United States (5.5 million children and 19.2 million adults) exhibited evidence of this disease in a self-report survey (56, 59). The standard of care in asthma, which includes inhaled corticosteroids that act on airway inflammation through glucocorticoid nuclear receptors, commonly does not suffice to prevent severe asthma attacks upon re-exposure to environmental allergens (60–62). Strict adherence to regimens of inhaled corticosteroids has remained problematic: a meta-analysis of research from 1985 to 2012 showed no significant improvements in overall clinical success (60–64). Although the rates of complete resistance to corticosteroids are low, even partial resistance can undermine long-term health for asthmatic patients, making identification of additional treatments desirable (65). Thus, despite a number of therapeutic advances and ongoing introduction of new agents to complement inhaled glucocorticoids (reviewed in Ref. 65), a large burden of morbidity remains in which manifestations of the disease are inadequately controlled and thousands of patients still die of uncontrolled asthma each year (59, 65, 66).

Much remains to be determined about the origins of allergic asthma and the pathophysiology of established disease. Although each state can exist independent of the other, there is substantial overlap of asthma with prior atopy (67–70). Atopic or T2-endotype asthma patients typically exhibit an increase in circulating IgE, which points to a contribution of the Th2 of CD4⁺ T cells that secrete type 2 cytokines such as IL-4, IL-5, and IL-13 inasmuch as IL-4 receptor stimulation is crucial for generation of IgE (42–44, 70). Within human asthma and animal models of the disease pathogenesis, eosinophilic (Th2-like) and/or neutrophilic (Th17-like along with Th2) patterns of allergic lung inflammation have been identified in patients and mouse models along with mixed endo-phenotypes such as Th2/Th1 and Th2/Th17 (71–76). Of these, the eosinophilic-pattern disease has seen therapeutic advances with parenteral administration of biologic drugs (e.g., anti–IL-5), whereas the neutrophilic pattern is more associated with severe or glucocorticoid-resistant disease (75, 76) and has seen less progress. Among unmet needs, progress on oral absorbable agents that could mitigate allergic disease or asthma remains important: biologics that require parenteral delivery can raise access and cost issues.

Of note, neither PARP14-deficient nor RBN012759-treated mice have exhibited long- or short-term signals of abnormal health. Moreover, although antiviral IgA was reduced postinfection of Parp14^−/− mice with mouse-adapted human metapneumovirus, a major cause of viral respiratory disease, illness was not worse (14). Taken together, these points suggest that chronic treatment with the selective catalytic inhibitor may prove well tolerated in humans as well as effective in the setting of type II inflammation. The findings in the present study suggest that highly selective catalytic inhibitors of this type will also be helpful in research outside the field of allergic disease research. For instance, several separable steps in the replication of diverse viral pathogens require their encoded activity of macro domains for binding to ADPr, or removal of mono-ADPr from unknown protein targets, or both (77–89). Although one line of work provides evidence that PARP14 mediates increased type I IFN by both ART–dependent and independent mechanisms (89) in a coronavirus model, more selective, potent, and in vivo–acting inhibitors such as RBN012759 should aid in elucidation of complex functions of PARP14 in infection.

The present body of work both raises intriguing new questions and has inherent limitations. Among these are mechanistic issues, most of all with regard to a sexual disparity (90). In favor of likely efficacy, positive data were obtained for the primary endpoints, that is, mucus scoring and IgE, with both male and female mice, and the effect magnitudes for these findings were similar in the two sexes. An encouraging point in support of the findings in the present study is that reductions in BAL and lung inflammation as well as type II cytokines have now been observed in fully independent preliminary replication work with different experimentalists, source of BALB/c male mice (Charles River Laboratories rather than The Jackson Laboratory), mouse colony, and batch of Alternaria extract. Nonetheless, later studies are needed to test whether therapeutic efficacy will extend to inbred strains of mice with different patterns of immunity (e.g., C57BL/6, A/J), aeroallergens other than Alternaria, or to IgE-mediated or allergic disease processes of skin (e.g., eczema) or gut (e.g., food allergy). In addition, the primary endpoint effects in female mice were not accompanied by decreased cytokine release after ex vivo (re)activation of lung cell suspensions with T cell–specific or nonspecific stimuli. This finding leaves open questions as to whether the mechanisms are the same in males and females, and what are the cellular and molecular targets in which PARP14 promotes allergic processes. With regard to cellular targets, in vitro data presented in this study and elsewhere suggest that apart from cytokine production by T cells, PARP14 inhibition changes the pattern of response for both B cells and macrophages (14, 91). However, although the risk of a type 2 statistical error is present, we did not observe a major difference between treatment and control groups with respect to the frequencies of MHC class II⁺ or CD206⁺ macrophages (indices of M1 versus M2) in these experiments on allergic lung inflammation (data not shown). As to molecular targets, a fundamental technical barrier has been the difficulty in identifying and mapping mono-ADPr modifications on endogenous proteins in physiological settings with full-length PARP14. Thus, cell-free in vitro approaches have provided a plethora of potential candidates (92; J. Lim, S.H. Cho, and M.R. Boothby, unpublished observations), but few of these have been validated. Some evidence favors a model in which PARP14-mediated ADP-ribosylation of STAT1 changes its transcriptional effects on monocyte/macrophage-lineage cells (91). In vitro and in vivo, however, major effects of RBN012759 on classical M1-like (M-IFN-γ) macrophages may be indirect and due more to inhibition of the M2-like (M-IL-4) population. Accordingly, despite unimpressive results with surface staining for MHC and II and CD206, it is possible that PARP14 inhibition changes the products of alternatively activated macrophages that are key intermediaries in allergic lung inflammation (93). Unfortunately, this avenue of research is unlikely to be fundable via the National Institutes of Health, limiting the potential for further answers.

We are grateful to Prof. L. Wu and the department for flow cytometry capabilities, the cores of VUMC and Vanderbilt University that were used in this work (Flow Cytometry, Translational Pathology, and Molecular Biology reagents), to S. Toki for helpful technical discussions pertaining to the use of Alternaria extract, and to D. Nichols for teaching A.M.E. histologic techniques.

K.K., L.B.S., K.K.S., J.R.M., M.N., and H.K. are full-time employees of, and hold equity interests in, Ribon Therapeutics but were recused from analyses and interpretation of the primary data and from manuscript preparation. M.R.B. holds equity in Regeneron, Inc., which markets a biologic agent used in treatment of allergic diseases and asthma. The other authors have no financial conflicts of interest.