Small molecule targeting of the STAT5/6 Src homology 2 (SH2) domains to inhibit allergic airway disease

Asthma is a chronic inflammatory disease of the lungs and airways and one of the most burdensome of all chronic maladies. Previous studies have established that expression of experimental and human asthma requires the IL-4/IL-13/IL-4 receptor α (IL-4Rα) signaling pathway, which activates the transcription factor STAT6. However, no small molecules targeting this important pathway are currently in clinical development. To this end, using a preclinical asthma model, we sought to develop and test a small-molecule inhibitor of the Src homology 2 domains in mouse and human STAT6. We previously developed multiple peptidomimetic compounds on the basis of blocking the docking site of STAT6 to IL-4Rα and phosphorylation of Tyr641 in STAT6. Here, we expanded the scope of our initial in vitro structure–activity relationship studies to include central and C-terminal analogs of these peptides to develop a lead compound, PM-43I. Conducting initial dose range, toxicity, and pharmacokinetic experiments with PM-43I, we found that it potently inhibits both STAT5- and STAT6-dependent allergic airway disease in mice. Moreover, PM-43I reversed preexisting allergic airway disease in mice with a minimum ED50 of 0.25 μg/kg. Of note, PM-43I was efficiently cleared through the kidneys with no long-term toxicity. We conclude that PM-43I represents the first of a class of small molecules that may be suitable for further clinical development against asthma.

Asthma is a chronic inflammatory disease of the lungs and airways that is marked by exaggerated airway-constrictive responses against diverse provocative challenges (airway hyperresponsiveness, AHR 3 ); accumulation of mucus and fibrin plugs in the airway that further compromise airflow (plastic bronchitis); airway and systemic eosinophilia; T helper type 2 (T H 2) cells and innate lymphoid cells type 2 (ILC2) that secrete the cytokines interleukin-4 (IL-4), IL-5, and IL-13; and vigorous IgE antibody responses (1)(2)(3). Affecting up to 30 million Americans, with 3 million experiencing severe therapyresistant disease, asthma is one of the most common and burdensome of all chronic afflictions (4,5).
Current asthma therapy provides symptomatic relief for many patients, but the inability of these agents to cure disease ensures their long-term use and attendant risk of inducing serious, even life-threatening side effects. We reasoned that a safer and more effective means of treating asthma could be developed through an improved understanding of the critical immune effector pathways that drive disease expression. The development of asthma is complex and is initiated in part by airway epithelial cells that recognize exogenous antigen or pathogens through pattern recognition receptors. Airway epithelial cells then secrete the inflammatory cytokines IL-25, IL-33, and TSLP (1,2) to promote ILC2 recruitment and enhance dendritic cell luminal sampling, migration to lymph nodes, and antigen presentation to initiate B and T cell activation (1, 6 -8).
Adaptive immunity is critical for maintaining the chronic inflammation associated with asthma that is mediated by T H 2 and T H 17 cells that migrate to the lungs (3). When in the lung, T H 2 cells work cooperatively with ILC2s to coordinate production of the cytokines IL-4 and IL-13 (8). These cytokines signal through distinct multimeric receptor complexes that share the common IL-4R␣ subunit (9). Cytokine binding to IL-4R␣ induces recruitment of the tyrosine kinases Jak1 and Jak3 (Jak1 and Tyk2 for IL-13) that then phosphorylate Tyr 631 within the Tyr-Lys-Pro-Phe docking site for the latent cytoplasmic transcription factor signal transducer and activator of transcription 6 (STAT6) (9). Upon binding to this motif via its Src homology 2 (SH2) domain, STAT6 is phosphorylated by Jak kinases, becoming competent to homodimerize, translocate to the nucleus, and promote STAT6-dependent gene expression (10). The absence of IL-4/IL-13/STAT6 responses in STAT6 Ϫ/Ϫ mice results in complete disease resistance in murine models of asthma, highlighting the central role of STAT6 in the development of asthma (11,12).
In addition to STAT6, STAT5 also contributes to the development of allergic airway disease. STAT5 together with STAT6 is required for optimal T H 2 development (13)(14)(15). STAT5 is also essential for ILC2 development and is the major transcription factor activated by IL-5, which promotes eosinophil development, recruitment, and survival (5,16). STAT5 and STAT6 are both activated by TSLP (17,18), underscoring the fact that multiple diverse ligands drive the activation of a limited number of STAT factors that are crucial to the expression of asthma. Thus, although targeting select cytokines is likely to be beneficial in asthma (19,20), a more effective approach is likely to be targeting the key STAT factors that represent the final common allergic disease mediators (21).
Our laboratories have focused on the design of small-molecule phosphopeptidomimetic inhibitors of the SH2 domain of STAT6 that prevent recruitment to the critical docking site on IL-4R␣ and phosphorylation of Tyr 641 . Early-generation compounds were shown to be potent ligands of STAT6 in vitro, blocking STAT6-dependent gene expression in a variety of human airway epithelial cells, breast cancer cells, and primary splenocytes (22,23). Furthermore, the cell-permeable, phosphatase-stable prodrug PM-242H was effective at preventing and/or reversing allergic lung disease in a murine model of asthma (11). Here we expand the scope of our initial in vitro structure-activity relationship studies to include novel central and C-terminal analogs.

Novel central scaffold and C-terminal analog affinities
The phosphopeptide PM-301 was shown previously to have a high affinity (IC 50 ϭ 117 nM) for STAT6 (23). The cell-permeable, phosphatase-stable PM-30I prodrug PM-37I displayed high cellular potency, completely blocking IL-4 -stimulated STAT6 phosphorylation at ϳ100 nM. PM-37I also blocked EGF-stimulated pSTAT5 in MDA-MD-468 breast cancer cells and IL-4 -dependent STAT6 activation in primary murine splenocytes (23, 24). Here we considered the effects of conformationally constrained modifications of PM-37I on the central dipeptide region and a survey of cyclic amides on the C terminus on STAT inhibition.
To examine the importance of the central (Tle-Pro) scaffold on STAT6 affinity, we synthesized a series of conformationally constrained dipeptides (Fig. 1a). In PM-28I, linkage of the C␣ of pTyrϩ1 to the N␣ of pTyrϩ2 with a Freidinger lactam resulted in a 5-fold decrease in affinity (IC 50 of 630 nM), as assessed by fluorescence polarization. Lactamization of the pTyrϩ2 and the C-terminal amide had a detrimental effect on affinity, as the IC 50 of PM-60I was 2.33 M, a 20-fold decrease in affinity. In contrast to the Freidinger lactams, replacing the central Tle-Pro of PM-301H with the tricyclic dipeptide mimic Haic (PM-34I), was well tolerated, with only a 2-fold loss in affinity compared with PM-301H. These data show a slight but significant decrease in affinity with increasing rigidity in the central scaffold.
The C-terminal methylanilide of PM-301H was replaced with a series of cyclic amides (Fig. 1a). In the piperidinamide series (PM-67I-(25)), morpholine (PM-67I-B) was ϳ2-fold higher in affinity than piperidine (PM67I-A) and was equal to the starting methyl, phenyl amide (PM-301H). Interestingly, the positively charged 4-methylpiperazine analog (PM-67I-C) bound the STAT6 SH2 domain with a 20-fold lower affinity than the lead PM-301H. Fused alky-aryl amides with varying alkyl rings were also probed. Increasing the ring size from n ϭ 5 to n ϭ 7 had no effect on affinity, with all three tested compounds (n ϭ 5, PM-59I; n ϭ 6, PM-87I; n ϭ 7, PM-71I-A) displaying IC 50 values of 230 -260 nM. Interestingly, the tetrahydroisoquinolinyl amide (PM-71I-B), which moves the benzene ring slightly farther from the main chain, was very avid, with an IC 50 of 50 nM. Taken together, these data show that STAT6 affinity, as defined by fluorescence polarization, is only mildly impacted by changes in ring conformation.

Cellular activity screen
The phosphate-containing inhibitors shown in Fig. 1a were converted to a series of cell-permeable, phosphatase-stable prodrugs by addition of phosphate-blocking POM groups (24) and screened for the ability to inhibit IL-4 -stimulated STAT6 inhibition (data not shown). Of this series, PM-43I, PM-63I, PM-74I, PM-80I, PM-81I, and PM-86I were the most potent (STAT6 inhibition Ͼ90% at 5 M) and selected for more detailed analyses. Titration of the inhibitors in Beas-2B cells indicated EC 50 values of 100 -500 nM, as judged by pSTAT6 inhibition (Fig. 1b). The exception was PM-43I, which required between 1 and 2 M to completely inhibit STAT6 phosphorylation.
The six leads were also evaluated for their potential for inhibition against additional SH2 domain-containing proteins, including other STAT factors, in MDA-MB-468 cells stimulated with epidermal growth factor (EGF) for STAT3, STAT5, AKT (via the SH2 domain-containing p85 subunit of phosphatidylinositol 3-kinase) and FAK activation (via SH2 containing Src kinase activity), or IFN-␥ for STAT1 activation (Fig. 1c) (26,27). Considerable variability in cross-reactivity was observed between the screened compounds, with PM-86I showing the highest specificity for STAT6 and no cross-reactivity to any of the additional targets at the highest dose tested. At 5 M, PM-43I and PM-63I showed significant cross-reactivity to STAT5, and PM-43I showed slight inhibition of STAT3. PM-74I was cross-reactive with STAT1, STAT3, STAT5, and AKT (p85/phosphatidylinositol 3-ki-Targeting STAT5/6 SH2 domains in allergic airway disease nase). PM-80I was the only compound to show cross-reactivity to STAT5 below the 5 M maximum dose. PM-81I did not cross-react with STAT5 but did moderately inhibit FAK and STAT1.
Prodrug cytotoxicity after 72-h exposure was assessed in Beas-2B (Fig. 1d) and MDA-MB-486 (Fig. 1e) cells by the MTT oxidation assay. With near-complete viability at Յ 1 M and IC 50 values between 8 M and 10 M, the apparent reduction in STAT6 phosphorylation cannot be attributed to significant cytotoxicity of drug at the tested concentrations. A moderate reduction in cellular activity was consistently observed in cells treated with 5 M PM-63I, suggesting slight toxicity with this species.
Although this is not a complete survey of all of the SH2 domains that could be encountered in a cell, these results show that it is indeed possible to develop inhibitors selective for the conserved STAT family. Furthermore, the surveyed proteins allowed for the identification of potential inhibition of proteins associated with regulation of cell cycle (AKT) and nonallergic immune responses (STAT1). Because of their favorable combination of biochemical and toxicity features, PM-43I and PM-86I were selected for additional in vivo studies.

Targeting STAT5/6 SH2 domains in allergic airway disease In vivo allergic lung disease screen
To determine the in vivo activity of these selected compounds, we assessed the impact of PM-43I and PM-86I on the expression of IL-13-STAT5/6-dependent allergic airway disease using a fungal infectious murine model (Fig. 2a) (28). C57BL/6 mice exposed to inhaled drug vehicle (DLPC) and challenged with Aspergillus niger (AN) developed robust airway hyperresponsiveness, as induced by increasing doses of acetylcholine chloride. In contrast, fungus-challenged mice treated with PM-43I or PM-86I had significantly reduced maximal increases in respiratory system resistance (R RS ; a measure of AHR) (Fig. 2, b and c). Although both drugs significantly inhibited development of AHR, the lowest dose of PM-43I (25 g/kg) produced the most consistent inhibition (Fig. 2c).
Lung inflammatory responses were also quantified. Compared with fungus-challenged mice treated with vehicle, drugtreated mice had significantly fewer cells recovered from bronchoalveolar lavage fluid (Fig. 2d) as well as marked reductions in eosinophilia (Fig. 2e). Similarly, mice treated with high and low doses of PM-43I had significantly reduced numbers of total lung cells secreting IL-4 (Fig. 2f). IL-4 -secreting cells were also significantly reduced in the lungs of mice treated with 250 g/kg PM-86I, but again, the 25 g/kg dose of PM-43I yielded remarkably greater and more consistent suppression of eosinophils and other indices of allergic inflammation.

Antigen-specific immunity
STAT6 signals in both immune (i.e. T cells) and nonimmune (i.e. airway epithelial) cells to coordinate the expression of allergic airway disease. Induction of STAT6-dependent T H 2 cells is further believed to occur in secondary lymphoid organs such as the spleen even when allergen challenge occurs remotely from the spleen (e.g. the airway). Thus, we questioned whether the reduced allergic disease illustrated in Fig. 2 was due to systemic or local suppression of STAT6. To address this, we assessed antigen-specific cytokine recall responses from splenocytes of mice exposed to inhibitors given through distinct routes. Mice were sensitized to ovalbumin through intraperitoneal sensiti- Targeting STAT5/6 SH2 domains in allergic airway disease zation while receiving either PM-43I or PM-86I systemically (i.p., 5,000 g/kg) or locally (i.n., 250 g/kg) (Fig. 2g). Antigenspecific cytokine recall responses of splenocytes were then determined using an enzyme-linked immunocell spot assay (ELISpot) format. Mice receiving either drug i.p. had significantly reduced ovalbumin-specific IL-4 -secreting cells compared with vehicle (DLPC)-treated mice (Fig. 2h). No difference was observed in the total numbers of IFN-␥-or IL-17secreting cells between the systemically treated mice, showing that both PM-43I and PM-86I can specifically inhibit STAT6dependent adaptive T H 2 immune responses when given systemically. In contrast, local administration of PM-43I or PM-86I to the lungs did not inhibit the development of splenic cytokine responses (Fig. 2i). These observations indicate that the reduced allergic airway disease observed could not be attributed to impaired T H 2 responses but, rather, was a result of local inhibition of STAT6 in airway cells.

Dose ranging analysis
Our data so far suggested that, although similar in their antiinflammatory activity, PM-43I was overall more consistently effective compared with PM-86I. We therefore next sought to identify the lowest efficacious in vivo dose of drug in the allergic airway disease model, focusing on PM-43I (Fig. 3a). To improve our ability to detect subtle differences, BALB/c mice were used in these studies, given their exaggerated T H 2 responses and AHR in this model (29). Relative to vehicle-treated animals, mice treated with PM-43I over a wide dose range (0.025-25 g/kg) showed progressive reductions in AHR. Paradoxically, the efficacy of PM-43I increased inversely with respect to the dose given, down to a maximally effective dose of 0.25 g/kg; at lower doses (0.025 g/kg), the drug began to again lose the ability to control AHR (Fig. 3, a and b). Similarly, airway inflammatory cells and cytokine-producing cells were also significantly reduced, most effectively in the lungs of mice receiving 0.25 and 0.025 g/kg doses (Fig. 3, c and d). Thus, PM-43I demonstrates therapeutic control at very low doses (0.25 g/kg).

Phosphatase-stable prodrug cross-reactivity
Although the initial drug screen with the fluorescent polarization assay consisted of drugs containing a naturally occurring phosphotyrosine functional group ( Fig. 1a; PM-28I), PM-43I has a phosphatase-stable (-CF 2 PO 3 Ϫ2 ) motif that may influence binding efficiency and specificity for the highly conserved SH2 domains of STAT family members. Initial in vitro functional screens showed potential cross-reactivity to STAT5 and, to a much lesser extent, STAT3 (Fig. 1c). Sequence analysis of STAT6 (P42226.1) and the STAT6 SH2 domain showed considerably more sequence similarity to STAT5B (P51692.2; 41.55% and 49.22%, respectively) than STAT3 (P40763.2; 28.04% and 36.15%, respectively) (Fig. S1a). Further analysis of the conserved protein SH2 domain family showed that three of the four amino acids that form the SH2 domain phosphoty- Targeting STAT5/6 SH2 domains in allergic airway disease rosine binding pocket of STAT6 are conserved in STAT5B but only two are conserved in STAT3. The higher conservation in STAT family sequences and binding pocket amino acids makes it more likely that an inhibitor would be cross-reactive.
To better characterize the potential impact of the phosphostable PM-43I on binding to the phosphotyrosine binding pocket of STAT6, STAT5B, and STAT3, we sought to determine the cell-free affinity of PM-43I to STAT6, STAT5B, and STAT3 (Fig. S1, b-d). PM-43I bound recombinant STAT6 with an IC 50 of 1.8 M, which reflects the in-cell STAT6 inhibitory concentration of 1-2.5 M (Fig. S1b and Fig. 1b). PM-43I bound recombinant STAT5B with slightly lower affinity (IC 50 ϭ 3.8 M), similar to the observed cell-based inhibition ( Fig. S1c and Fig. 1c). In contrast, the affinity for STAT3 was found to be much lower (IC 50 ϭ 29.9 M), consistent with the limited STAT3 inhibition observed at the maximum dose tested in the initial cross-reactivity screen ( Fig. S1d and Fig. 1c). Taken together, the affinity analysis appears to reflect the degree of sequence similarity among the tested STAT proteins and supports the observed STAT5/6 in vitro selectivity and in vivo attenuation of STAT5/6-dependent immunity.

In vivo structure-function analysis
To identify the structural features that make PM-43I a more potent inhibitor than PM-86I in vivo, we tested PM-43I/PM-86I hybrid prodrugs in the allergic airway disease model. PM-43I and PM-86I differ structurally at their central scaffold regions and C termini. We created two phospho-stable structural hybrids, PM-37I and PM-205I, where these regions have been swapped (Fig. 4a). Using the lowest efficacious dose identified in the dose ranging experiment (0.25 g/kg), a direct comparison of the hybrid compounds was performed. As expected from previous experiments, mice treated with PM-43I, but not PM-86I, had significantly reduced airway reactivity compared with vehicle-treated, fungus-challenged mice (Fig. 4, b and c). PM-37I, the prodrug of the first high-affinity peptide mimetic (PM-301; Fig. 1a) containing the PM-43I C terminus and PM-86I central scaffold, failed to inhibit AHR (Fig. 4, b and c). In contrast, PM-205I, containing the PM-43I central scaffold and PM-86I C terminus, significantly inhibited AHR.
As expected, PM-43I reduced total BALF inflammatory cells, but, interestingly, no other inhibitor affected this quantity (Fig.  4d). All inhibitors reduced BALF eosinophils, with PM-43I showing the greatest reduction. Lung IL-4 -secreting (Fig. 4e) and IL-17-secreting (Fig. 4f) cells were also significantly reduced by PM-43I treatment, consistent with a general reduction in lung inflammation by STAT6 blockade in the lungs. Similarly, PM-205I-treated mice also had significantly fewer lung IL-4 -secreting cells but only moderately reduced IL-17 (Fig. 4f). PM-37I had no inhibitory effect on cytokine-producing cells and, in fact, caused marked up-regulation of lung IFN-␥-secreting cells (Fig. 4, e-g). Taken together, these data show that the central scaffold Freidinger lactam of PM-43I is the critical structural feature contributing to the superior in vivo efficacy of PM-43I and PM-205I. As evident by the moderately enhanced potency of PM-43I over PM-205I (Fig. 4f), variation in the C terminus did not appear to significantly impact drug efficacy beyond IL-17 production.

Efficacy of aerosolized drug
To more accurately simulate the delivery of a peptidomimetic therapeutic agent clinically, we next sought to assess the efficacy of a PM-43I aerosol. PM-43I packaged in DLPC was aerosolized under standard conditions, quantified by HPLC-MS (Fig. S2), and found to deliver the PM-43I prodrug at sufficient quantities for in vivo studies without significant degradation (data not shown). Particle size analysis showed that Ͼ70% of aerosol particles containing PM-43I were of the appropriate size (0.5-3 m) to deposit in the lower airways (Fig.  5a). These data show that the PM-43I prodrug can be stably delivered to the lower airways by aerosolization.
We next delivered PM-43I in a reversal model of pre-existing allergic airway disease, comparing aerosol versus intranasal delivery of the same dose (0.25 g/kg; Fig. 5b). Both PM-43Itreated groups showed significantly reduced AHR after 1 week of therapy compared with the DLPC control (Fig. 5c). However, mice that received aerosolized PM-43I had dramatically lower, and effectively abrogated, AHR compared with i.n.-treated mice, suggesting that aerosolization is a more efficient method of drug delivery to the lungs. Total lung cellularity was also significantly reduced in drugtreated mice, but no significant differences were observed in lung eosinophilia (Fig. 5d).
Lung cytokine production was also assessed, with both PM-43I-treated groups showing significantly reduced lung IL-4 -secreting cells (Fig. 5e) but with no difference found in the number of IL-17-secreting (Fig. 5f) or IFN-␥-secreting (Fig.  5g) cells. Importantly, the total number of fungal colony-forming units recovered from the lungs of challenged mice was not enhanced by drug therapy, showing that PM-43I did not impair anti-fungal immunity. These data show that PM-43I can reverse and potentially serve as an effective therapy for established allergic airway disease without compromising immunity against T H 2-polarizing fungi that are likely causes of asthma and related allergic airway diseases of humans (30).

Antiviral immunity is maintained
Given the chronic nature of asthma, affected subjects will experience intercurrent infections with a variety of pathogens, including viruses. We therefore further assessed the effect of PM-43I on lung immunity against the pulmonary viral pathogen influenza, clearance of which requires IFN-␥ responses (31,32). Mice receiving a daily inhaled dose of PM-43I (75 g/kg) were infected with influenza and monitored for changes in weight and survival over the course of the infection (Fig. S2a). Infected mice, as expected, showed significant weight loss and mortality, but mice treated with a relatively high dose of PM-43I were not adversely affected in either regard and, in fact, showed improved survival compared with the vehicle control (Fig. S2, b and c). This experiment indicates that lung antiviral immunity is maintained in the context of STAT inhibition by PM-43I.

PM-43I is renally excreted and eliminated from the lung within 48 h
We next evaluated the pharmacokinetics of PM-43I by assessing its distribution and clearance in naïve mice using a highly sensitive HPLC-MS method (Fig. S3, a-d) for identification of a prodrug containing two phosphate-blocking groups (Fig. S3a), one blocking group (Fig. S3b), or active drug (Fig.  S3c). Mice treated i.n. with 250 g/kg PM-43I were assessed for drug distribution in the lungs (Fig. 6a), liver (Fig. 6b), kidney (Fig. 6c), and urine (Fig. 6d) over 48 h. All three drug forms were identified at time 0 in the lungs, which likely reflects the time lapse between treatment of tissue stabilization/freezing (Fig.  6a). The bis-POM species was the least abundant species at time 0 and was not detectable beyond 5 min, showing rapid in vivo removal of at least one phosphate-protecting group. The mono-POM species was the most abundant at time 0 in the lungs but was also quickly converted to the bioactive non-POM form or cleared within 30 min. In contrast, bioactive non-POM drug was detectable in the lungs beyond 24 h after i.n. treatment and was not fully cleared until 48 h.  . Aerosol characterization and allergic lung disease reversal. a, aerosol particle size and predicted lung distribution. b, BALB/c mice were challenged intranasally with AN for 2 weeks, and airway hyperresponsiveness was determined. Daily therapy of 0.25 g/kg PM-43I was then given intranasally or by aerosol, and fungal challenges were continued for an additional 2 weeks. c, airway hyperresponsiveness was assessed weekly. d-g, total BALF inflammatory cells (d), total lung IL-4 -secreting (e), IL-17A-secreting (f), and IFN-␥-secreting (g) cells, and total recovered lung fungal colony-forming units (h) were quantified at the end of the experiment. *, p Ͻ 0.05; n Ն 6 mice/treatment group. Data are representative of two or more independent experiments. Eos, eosinophils; Mono, monocytes; Neut, neutrophils; Lym, lymphocytes.

Targeting STAT5/6 SH2 domains in allergic airway disease
In the liver, only the non-POM species was identified, with peak abundance at 5 min (Fig. 6b). Trace quantities of the mono-POM species were identified in the kidneys at time 0, but non-POM was the predominant species identified and cleared by 30 min (Fig. 6c). Only the non-POM species was detected in the urine, with levels peaking at 1 h (Fig. 6d). Together, these data indicate that the PM-43I prodrug is rapidly taken up intracellularly at the site of delivery, converted to active drug rapidly, and then retained within the lung for more than 24 h. Systemic distribution of drug appears to be limited to the cell-impermeable, non-POM species and is likely due to the high blood volume of these tissues. Active non-POM drug is excreted primarily in the urine and not retained in either the kidney or liver.

Impact of PM-43I on long-term immunity and toxicity
To determine whether PM-43I elicits any significant toxicity, we evaluated body weight, blood chemistry, complete blood count, lung function, splenic immune cellularity, and humoral immunity in naïve mice that received PM-43I three times a week at doses 1,000-fold more than the effective dose for up to 8 months while receiving antigen once a week (Fig. 7a). Longterm exposure to PM-43I had no significant impact on airway reactivity compared with the vehicle control (Fig. 7b). Further, PM-43I did not impair weight gain (Fig. 7c). Similarly, blood cellularity (Fig. S4a), blood chemistry (Fig. S4b), and splenic cellularity (Fig. S4, c and d) were not different between vehicle and PM-43I-treated mice after 8 months of therapy. Serum Ova-specific antibody responses revealed that the ratio of anti-gen-specific IgG 2a (T H 1-associated) to IgE (T H 2-associated) antibody isotypes was significantly increased after 8 months of PM-43I treatment in mice that were simultaneously immunized with ovalbumin, suggesting that long-term therapy may suppress previously established allergic memory responses (Fig. 7d). Overall, these data suggest that long-term treatment of mice with PM-43I is nontoxic but does influence antibody isotype profiles, favoring type 1 over type 2 antibodies.

Discussion
Based on seminal work conducted in the mouse (33,34) and verified through human clinical trials (19,20), the IL-4/IL-13-IL-4R␣-STAT5/6 signaling pathway is now recognized as critical for determining the expression of asthma and related allergic diseases. However, because of redundant and alternative pathways for activation of STAT6 (e.g. through TSLP (17), longacting beta agonists (11), and IL-4 (35)), therapeutic agents that target the extracellular limb of this pathway (including cytokines and receptors) will likely fail to completely suppress STAT6 activation. Additional redundant cytokine pathways support STAT5 activation, underscoring this concern (36). Thus, existing and future asthma therapeutic agents that target STAT5/6 ligands and receptors are destined to be incompletely effective in blocking unwanted allergic responses. Moreover, the agents that have been developed in this regard are all monoclonal antibodies, which are prohibitively expensive, especially for patients with mild to moderate disease. Additionally, monoclonal antibodies must be delivered systemically, increasing the Mimetic peptides represent a considerably less expensive alternative drug strategy that has proven effective at targeting extracellular and intracellular proteins (37)(38)(39)(40). The high specificity of monoclonal antibodies can be achieved through the intelligent design of mimetic peptides around known proteinprotein interactions. Several groups have embraced this strategy and developed large peptides ( 10 amino acids) to modulate BH3-mediated apoptosis, suppressor of cytokine signaling modulation of inflammation, and high-density lipoproteindriven cardiovascular disease (37,38,40). We have previously reported our work developing small peptidomimetics (4-mers) targeting the SH2-domain of STAT3 and the suppression of tumor growth (24,26,39). Here we report the development of a novel phosphopeptidomimetic small molecule that is based on the cytoplasmic docking site on IL-4R␣ for STAT6, PM-43I.
Targeting the SH2 domain of STAT6, PM-43I is also a potent inhibitor of STAT5. As STAT5/6 require activation through their SH2 domains in all contexts and share a significant degree of sequence similarity, PM-43I is theoretically capable of completely inactivating STAT5 and STAT6 regardless of the activating stimulus. Indeed, the highly effective nature of PM-43I in our experimental asthma model suggests that we are achieving inhibition of most of the latent STAT5/6 within mouse airway cells at the observed therapeutic doses. As a small molecule, PM-43I is relatively easily synthesized in large quantities and to high purity. Because of the low dose required (0.25-25 g/kg) and the need for only topical delivery to the airway to be highly effective, PM-43I should compare economically favorably to mAb-based asthma therapies. Moreover, PM-43I appears in our preliminary studies to be well tolerated and highly effective. Together, our findings indicate that PM-43I is suitable for entry into clinical development for asthma and related afflictions.
Phosphopeptidomimetics represent a new class of pharmacologic agents that have been developed successfully to target the SH2 domains of a variety of signaling proteins (33,26,39) but have not previously been evaluated for the blockade of multiple STAT factors or for the treatment of asthma. Previously, we developed the phosphopeptidomimetic prodrug PM-242H to target STAT6 selectively and demonstrated that this novel agent was effective at both preventing and reversing asthmalike allergic airway disease in mice (11). Although PM-242H proved the validity of this therapeutic strategy, this agent has an unfavorably high minimum effective dose of 2.5 mg/kg. The iterative redesign of PM-242H as presented here was undertaken largely to improve compound potency but also to determine the efficacy of more broadly cross-reactive STAT inhibitors.
Our in vitro observations were critically important in informing our choice of compounds selected for subsequent in vivo testing. Specifically, based on in vitro testing, we selected PM-43I as one compound that inhibited STAT5 and STAT6 and PM-86I as a compound that was highly selective for inhibiting STAT6. Although PM-43I inhibited STAT6 less effectively than PM-86I in vitro, PM-43I performed significantly

Targeting STAT5/6 SH2 domains in allergic airway disease
better in vivo regarding inhibition of multiple parameters of allergic airway disease (Fig. 2). We presume that PM-43I was more effective because of its ability to inhibit STAT5 in addition to STAT6. The superior in vivo response may be due to the inhibition of the Stat5-dependent, ILC2-driven innate immune responses that promote Stat6-depandant T H 2 adaptive immunity, both of which contribute to the expression of this disease phenotype (12)(13)(14)(15)(16)(17)(18)(41)(42)(43)(44). Another compound, PM-63I, also inhibited STAT5 and was at least 10 times more effective than PM-43I at inhibiting STAT6 (Fig. 1, a-c). However, we did not pursue in vivo studies with this compound because of the enhanced cellular toxicity exhibited against two cell lines at the lower tested doses (Fig. 1, d and e). Nonetheless, the in vivo potency of PM-43I, which potentially extends to the entire family of molecules we have synthesized, suggests that such in vitro toxicity might be irrelevant in vivo. Therefore, further studies with PM-63I using the same mouse model to determine the therapeutic potential of this compound may be warranted.
The chronic, incurable nature of asthma mandates that approved treatments and pharmaceuticals be selected so that they minimize long-term side effects. However, standard asthma care includes the use of pharmaceutical agents with relatively poor long-term safety records. The original long-acting beta agonists approved for use in the United States, formoterol and salmeterol, lead to loss of disease control and excess asthma-related mortality in a small subset of patients that is in part, but not completely (45), alleviated by the co-formulation with inhaled corticosteroids, as now mandated by the Food and Drug Administration. However, inhaled and especially oral corticosteroids have their own notorious, long-term safety concerns that include laryngitis, diabetes, weight gain, hypertension, cataracts, etc. Serious fungal infectious complications of chronic corticosteroid use (46 -49) are particularly concerning because asthma and related conditions, such as chronic rhinosinusitis, are increasingly recognized as manifestations of airway mycosis (airway fungal infection) (30). A primary benefit of agents such as PM-43I that are highly selective in their inhibition of immune pathways is that they are less likely to compromise protective immunity compared with broad-spectrum immunosuppressants like corticosteroids. Indeed, our preliminary toxicity studies suggest that PM-43I did not impact airway antifungal or antiviral immunity even when given in excessive doses (Fig. 5h and Fig. S2).
An important component of the drug delivery strategy presented here is the direct delivery of prodrug to the lungs and deposition on the target cells of the airway, where it is then converted, extracellularly and intracellularly, to active drug by removal of the POM groups by reactive oxygen species, ions, and/or esterases. Delivery of the drug by aerosol bypasses the caustic gastric and first-pass metabolism to allow immediate cellular absorption and intracellular activation of PM-43I in the lungs, where the drug persists for more than 24 h (Fig. 6). It is important to note that extracellular activation, by removal of the protective POM groups, abrogates membrane permeability and systemic drug activity (Fig. 2, h and i), allowing efficient renal clearance of nonabsorbed drug within hours of airway delivery (Fig. 6). We cannot be certain which airway cells primarily take up PM-43I but presume that airway epithelial and smooth muscle cells are main PM-43I targets, given their proximity to the airway lumen. Other airway cells are likely to be targeted also, and the more they are affected by PM-43I, the stronger the suppressive effect on allergic disease expression should be.
The efficient renal clearance of non-POM drug is a critical observation and supports the apparent lack of toxicity observed in the 8-month toxicity study (Fig. 7). The potential toxicity of the inactive POM group byproducts, formaldehyde and pivalic acid, could be a concerning factor, but it is well known that both formaldehyde and pivalic acid are readily metabolized and excreted at the therapeutic range of PM-43I (50 -54). Additionally, numerous drugs containing POM groups (adefovir, pivampicillin, cefditoren, valproate, etc.) are now Food and Drug Administration-approved and regularly used in the clinic.
Numerous drugs targeting immune molecules have been evaluated in asthma, many of which have proven successful (e.g. drugs targeting IL-5 and IgE), but not all. PM-43I offers several distinct advantages over all of these agents. First, as a small, easily synthesized molecule, it avoids the complexities and expense of biological agents such as monoclonal antibodies. Second, although biologic agents that individually target STAT6-activating cytokines have been developed, none individually blocks STAT6 completely and none block both STAT5 and STAT6 as does PM-43I. PM-43I is therefore predicted to be both more effective and less expensive than existing agents.
The direct delivery by aerosolization and activation of PM-43I in the lungs is an additional novel aspect of this drug that minimizes possible systemic toxic effects that are unavoidable with parenterally administered biologics. Assuming immediate and complete POM group breakdown and prodrug molar equivalence for the observed PM-43I therapeutic range (0.25-25 g/kg), a 15-min aerosol treatment would produce a formaldehyde concentration (0.006 -0.6 ppm) below the Occupational Safety and Health Administration (OSHA)-defined 15-min permissible exposure limit of 2 ppm and the 8-h permissible exposure limit of 0.75 ppm (53). Furthermore, the Environmental Protection Agency estimated average indoor formaldehyde concentration of 0.03 ppm is well above the optimal PM-43I therapeutic dose (2.5 g/kg, 0.006 ppm). Together, these data suggest that concerns regarding PM-43I-related toxicities related to the POM groups are negligible.
Although PM-43I demonstrated dose-dependent efficacy in vivo, the inverse in vivo relationship between dose and efficacy is unusual; the drug begins to lose efficacy at doses higher than the maximally effective dose of 0.25 g/kg. This issue highlights the complexities of targeting the intracellular compartment where many distinct potential targets exist, including more than 140 SH2 domain-containing proteins that are potentially cross-reactive targets. Although PM-43I did not cross-react at highly effective pharmacologic doses with other STAT proteins of considerable sequence homology (Fig. S1a) and other critical immune signaling proteins, such as AKT and FAK (Fig. 1C), potential cross-reactivity with the SH2 domains of suppressors of cytokine signaling 1 and 3 (SOCS1 and SOCS3) could block their ability to inhibit STAT6 activity, in effect extending the activity of STAT6 that was not inactivated by drug (55)(56)(57). Similarly, protein tyrosine phosphatase nonreceptor type 6 Targeting STAT5/6 SH2 domains in allergic airway disease (PTPN6, SHP-1) is an SH2 domain-containing phosphatase that dephosphorylates active STAT factors, especially STAT5 and STAT6 (58,59). Inhibition of this protein via its SH2 domain with PM-43I could potentially produce paradoxically enhanced STAT5/6 activity at supratherapeutic doses of PM-43I. If additional testing suggests that PM-43I possesses an excessively narrow therapeutic window related to such potential cross-reactivity, then PM-63I could be explored further as a therapeutic alternative, or PM-43I could be further chemically modified to reduce such cross-reactivity while preserving specificity for STAT5/6.

Concluding remarks
We have developed a potent inhibitor of STAT5 and STAT6 that can be stably delivered to the lungs and inhibit the induction of or reverse pre-existing allergic lung disease in a murine model of allergic lung disease. The lead compound does not appear to impair immunity or present any apparent toxicity but can skew circulating humoral immunity toward a less dominant T H 2 profile. Our studies confirm the feasibility to target the SH2 domain therapeutically and indicate that PM-43I is suitable for entry into clinical development.

Drug/inhibitor
1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC, 770335) was acquired from Avanti Polar Lipids (Alabaster, AL) and used as a vehicle at a 1:5 (drug:DLPC) ratio. In brief, drug and DLPC were solubilized in t-butyl alcohol at the identified mass ratio, frozen, and lyophilized. Individual treatments were suspended in PBS and sonicated for 30 s before use.

Aerosol characterization
Drug (PM-43I) or DLPC was aerosolized using an Aerotech II nebulizer with a flow rate of 10 liters/min. Aerosols were captured using an all-glass impinger under a vacuum. Aerosol particle size was determined with an Andersen cascade impactor. Captured drug was quantified by HPLC-MS and used to calculate the drug dose for animal studies as reported previously (60 -63).

Cell culture
A549, BEAS-2B, and MD468 cells were acquired from the American Type Culture Collection (Manassas, VA). Cells were maintained in 50% Dulbecco's modified Eagle's medium and 50% F-12 complete medium until sufficient confluency and switched to 2% FBS 24 h prior to stimulation. Cells were pre-treated up to 2 h with identified drugs at various concentrations, stimulated with IL-4 (2 ng/ml, R&D Systems), EGF (100 ng/ml, R&D Systems), or IFN-␥ (25 ng/ml, R&D Systems) for identified times and harvested for protein analysis.

Mice
4-to 8-week-old female BALB/c, C57BL/6, and CR1 mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed in the American Association for Accreditation of Laboratory Animal Care-accredited Transgenic Mouse Facility at Baylor College of Medicine under specific pathogenfree conditions. All experimental protocols were approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine and followed federal guidelines.

Infectious allergic lung disease/reversal models
Mice were challenged intranasally every other day with 4 ϫ 10 5 A. niger conidia for a minimum of six challenges and treated with the identified STAT6 inhibitors at identified doses and/or vehicle (DLPC) as described (infectious allergic lung disease model, Fig. 3a; reversal model, Fig. 7a) for each experiment. Following the final challenge, mice rested for a day, and allergic airway disease was assessed.

Allergic airway disease analysis
Allergic airway disease was assessed in accordance with previous reports (28). Increases in R RS because of intravenous acetylcholine, bronchoalveolar lavage (BAL) fluid differential counts, and lung IL-4 -, IL-17A-, and IFN-␥ secreting cells were quantified as described previously (25,28).

Targeting STAT5/6 SH2 domains in allergic airway disease Ovalbumin restimulation
Splenocytes from Ova-sensitized mice were assessed for antigen-specific recall by ELISpot for relevant cytokines. In brief, cells were cultured in the presence of medium or whole ovalbumin (1 mg/ml) overnight, and plates were developed for IL-4, INF-␥, and IL-17 as described previously.

Histology
Lungs were perfused with 4% paraformaldehyde and embedded in paraffin. Lung sections were cut and stained for mucus with the periodic acid-Schiff kit (395B) from Sigma-Aldrich (St. Louis, MO) or cells with hematoxylin and eosin. Lung sections were immunohistologically stained for p STAT6 and pSTAT5.

Fungal burden
Fungal burden in fungus-challenged mice was determined with serial dilutions on Sabouraud agar plates (84088, Sigma-Aldrich) of lung homogenates in the presence of 100 mg/ml chloramphenicol (C0378, Sigma-Aldrich). After overnight incubation at 37°C, plates were assessed for fungal colonyforming units per lung from dilution.

Influenza infection
Mice were placed on PM-43I (1.5 g) or DLPC aerosol therapy prior to challenge with an LD 30 of aerosolized influenza (H3N2), as detailed in Fig. 6a. Changes in weight and survival were monitored throughout the experiment.

Toxicity
Mice were sensitized to Ova-alum i.p. (once per week twice) and treated with PM-43I (5 g) or DLPC for 8 months, as described in Fig. 1a. In brief, sensitized mice were treated intranasally every other day with ovalbumin plus PM-43I or DLPC and monitored for changes in weight. Airway hyperreactivity and Ova-specific antibodies were assessed as described. Blood and spleens were harvested at the end of the experiment for further analysis. Blood samples were submitted to the pathology core at the University of Texas MD Anderson Cancer Center for complete blood counts and blood chemistry. Detailed cellular analysis of splenocytes was performed by the flow cytometry core at Baylor College of Medicine.

Flow cytometric analysis
Spleens were disassociated using an Octo dissociator (Miltenyi Biotech) in RPMI 1640 containing (Gibco) 10% heat-inactivated FBS (HyClone), 2 mM EDTA (Gibco), 0.2 mg/ml DNase I (Sigma), and 1 mg/ml collagenase II (Gibco). The suspension was incubated at room temperature for 20 min. Digestion was stopped by addition of EDTA to a 12 mM concentration. Cells were filtered through a 40-m mesh. Cells were collected, and red blood cells were lysed in RBC lysis buffer (eBioscience) for 5 min at room temperature. Cells were washed and resuspended in flow buffer (DPBS without Ca 2ϩ /Mg 2ϩ (Gibco) supplemented with 2% fetal calf serum, 25 mM HEPES, and 2 mM EDTA). A portion of the cells was used for enumeration and trypan blue viability assessment using a ViCell (Beckman Coulter). Fc receptors were blocked on ice for 10 min using flow buffer containing a 1:250 dilution of anti-mouse CD16/32 (BD Biosciences). Cells were incubated on ice for 30 min with one of two panels of antibodies, panel 1 or panel 2. Panel 1 included anti-mouse CD5, CD4, CD44, CD8, CD25, CD161, and CD62L (BD Biosciences). Panel 2 included anti-mouse CD5, Ly6G CD19, Ly6C, CD21/35, CD11b, CD11c, CD161, CD23 (BD Biosciences), and anti-mouse major histocompatibility complex II (Biolegend). Both groups were washed and resuspended in flow buffer with 1 drop/ml Nuc Blue Fixed DAPI Stain (Life Technologies). Single-stained DAPI control and fluorescence minus one controls (FMOs) were prepared from the remaining pooled spleen cells. Cells were run on a LSRII flow cytometer (BD Biosciences) and analyzed using FlowJo (Tree Star) according to immunophenotyping gating strategies outlined by the International Mouse Phenotyping Consortium.

Ova-specific ELISA
Serum samples were serially diluted and incubated in 96-well plates coated with ovalbumin. Captured antibodies were detected with biotinylated anti-IgE and anti-IgG 2a and developed with streptavidin-alkaline phosphatase. Substrate was added, and the signal was detected by optical density at 405 nm.

Pharmacokinetic analysis
Mice were treated intranasally with 5 g of PM-43I, and tissue samples were harvested at identified time points. Drug was quantified in tissue samples by HPLC-MS.

Analytical method
Tissue samples were homogenized in sample buffer (50% acetonitrile, 50% 0.1% formic acid). Serum and captured aerosol samples were mixed with acetonitrile at a 1:1 ratio. All samples and standards were separated on a C18-column (Phenomenex) and detected by MS-MS (Quattro Premier XE).