IL-18 Immunotherapy for Neovascular AMD: Tolerability and Efficacy in Nonhuman Primates

Abstract

Purpose: Age-related macular degeneration is the most common form of central retinal blindness in the elderly. Of the two end stages of disease, neovascular AMD—although the minority form—is the most severe. Current therapies are highly successful at controlling progression of neovascular lesions; however, a significant number of patients remain refractory to treatment and the development of alternative and additive therapies to anti-VEGFs is essential.

Methods: In order to address the translational potential of interleukin (IL)-18 for use in neovascular AMD, we initiated a nonhuman primate tolerability and efficacy study for the use of intravitreally (IVT) administered clinical grade human IL-18 (SB-485232). Cynomolgus monkeys were injected IVT with increasing doses of human IL-18 (two each at 1000, 3000, and 10,000 ng per eye). In tandem, 21 monkeys were administered nine laser burns in each eye prior to receiving IL-18 as an IVT injection at a range of doses. Fundus fluorescein angiography (FFA) was performed on days 8, 15, and 22 post injection and the development of neovascular lesions was assessed.

Results: We show intravitreal, mature, recombinant human IL-18 is safe and can reduce choroidal neovascular lesion development in cynomolgus monkeys.

Conclusions: Based on our data comparing human IL-18 to current anti-VEGF–based therapy, clinical deployment of IL-18 for neovascular AMD has the potential to lead to a new adjuvant immunotherapy-based treatment for this severe form of central blindness.

It is estimated that 1 in 10 people aged older than 55 years show early signs of AMD. In 10% of patients with AMD, blood vessels sprout from underlying choroidal vasculature disrupting retinal tissue integrity, leading to vision loss. Although less common, the neovascular form of the disease, termed “wet” AMD, is the most severe form and is termed a priority eye disease by the World Health Organization. Wet AMD can be treated acutely with regular intraocular injections of antibodies or fusion proteins directed against VEGF.¹ These treatments have revolutionized AMD therapy, stabilizing the areas of choroidal neovascularization (CNV) and delaying further neovascularization. However, there is no end stage to treatment as patients require regular injections to sustain efficacy and there is a proportion of patients who will remain refractory to treatment. Evidence also suggests that long-term use of “anti-VEGFs” is damaging to the retina due to the role of VEGFs as a neuronal survival factor and long-term progression of underlying AMD continues.² These data illustrate the critical requirement for the development of new therapeutic targets and novel strategies for the treatment and management of AMD. Indeed other mechanistic approaches which do not target VEGF—but used in combination with anti-VEGF therapeutics such as anti-PDGF aptamers (Fovista; Ophthotech Corp., New York, NY, USA)—have proved successful in clinical studies to date,³ and are currently undergoing phase III studies in neovascular AMD. The use of anti-PDGF aptamers, dosed between anti-VEGF doses, sensitizes the infiltrating choroidal neovessels to the anti-VEGF therapeutic resulting in enhanced efficacy and demonstrating effective interaction between two independent mechanisms. Here we provide evidence that interleukin-18 (IL-18) can be considered as a safe and relatively effective intravitreal immunotherapy for the treatment of wet AMD that could now be used alongside current therapeutic approaches. 

Materials and Methods

Mice

JR5558 mice that harbored the Rd8 mutation were bred at GSK and shipped to Trinity College Dublin for experimentation. Mice were housed in a specific pathogen-free environment throughout the course of treatment and were used between the ages of 3 and 8 weeks. Animals were injected intraperitoneally with mouse IL-18. 

Western Blot and Antibodies

We grew retinal microvascular endothelial cells (RMVECs) to confluency and treated them with vehicle or hIL-18 (10, 100, 1000 ng/mL) for 6 or 24 hours as indicated. Cells were then treated with VEGF (0.1, 0.3, 1, 3, 10, 30, 100 ng/mL) for an additional 24 hours. Cells were lysed in RIPA buffer (plus protease inhibitors; Roche, Dublin, Ireland). Protein concentrations were measured by BCA assay and equal amounts of protein were separated by SDS-PAGE gel electrophoresis, transferred to PVDF membrane, and incubated with VEGFR2 (Cell Signaling Technology, Inc., Beverly, MA, USA) or β-actin (Abcam, Cambridge, UK) and visualized by autoradiography. 

Quantitative Polymerase Chain Reaction

We grew RMVECs to confluency and treated them with vehicle or hIL-18 (50 ng/mL) for 4, 8, or 24 hours as indicated. Using a commercial kit (RNeasy Isolation kit; Qiagen, Venlo, Limburg, The Netherlands), RNA was extracted according to the manufacturer's protocol. Relative expression of VEGF-A and VEGFR-1, -2, and -3 was analyzed by quantitative PCR. We prepared cDNA from 20 ng/mL total RNA using an archive kit (High-Capacity cDNA; Applied Biosystems, Waltham, MA, USA). A real-time PCR platform (AB7900FAST; Applied Biosystems) was used for all PCRs in triplicate. Changes in expression were calculated by the change in threshold (ΔΔC_T) method with Gapdh as an endogenous control for gene-expression analysis and were normalized to results obtained with vehicle treated cells. 

Electroretinography

We dark-adapted JR5558 mice overnight and prepared them for ERG under dim red light. Electroretinogram responses were recorded simultaneously from both mouse eyes by means of gold wire electrodes (Roland Consulting GmbH, Havel, Germany) using topical eye gel (Vidisic; Dr. Mann Pharma, Berlin, Germany) as a conducting agent in addition to maintaining corneal hydration. 

Monkey Tolerability Study

Female cynomolgus monkeys (n = 2 per dose) were administered ascending doses (1000, 3000, and 10,000 ng) of IL-18 (SB-485232) and vehicle control (25 mM sodium acetate, 0.1 M EDTA), 6% (wt/vol) sucrose and 0.05% Tween 80 at pH 4) by IVT injection (50 μL per eye) and followed for 15 days. After enucleation, eyes were fixed in 3% glutaraldehyde for a minimum of 24 hours, followed by transfer to 10% neutral buffered formalin before processing and staining with hematoxylin and eosin. Images represent a section of the retina in the region of the macula from the eye of a single monkey administered either vehicle control or IL-18 by IVT injection. Monkeys used were bred in captivity and were of Vietnamese origin. Animals were aged <3 years at time of study. All relevant ethical approval documentation was obtained before study initiation. 

Monkey Efficacy Study

A total of 21 male cynomolgus monkeys (Macaca fascicularis) aged 2 to 4 years and weighing 2 to 5 kg were obtained from Covance Research Products. Animals were fed Certified Primate Diet #2055C (Harlan Laboratories, Inc., Indianapolis, IN, USA) with access to water ad libitum and maintained at 18 to 26°C, with a relative humidity of 30% to 70%, a minimum of 10 air changes/hour, and a 12-hour light/12-hour dark cycle. Prior to IVT injections, the eyes were anesthetized using topical anesthetic (0.5% proparacaine, Bausch & Lomb Pharmaceuticals, Tampa, FL, USA), cleaned with 1% povidone iodine (prepared with sterile saline and 10% povidone iodine, Bausch & Lomb Pharmaceuticals), and rinsed with sterile saline. The periorbital region was cleaned with a dilute, 1% povidone-iodine solution. Doses were administered by intravitreal injection using a 1-mL syringe and a 30-G (injections) or 25-G (implants) needle. Injections were made into the inferior aspect of the eye. A topical antibiotic (Tobrex, 0.3%, Alcon, Fort Worth, TX, USA) was instilled in each eye following the intravitreal injection. For laser photocoagulation, the macula of each eye was targeted with a 532-nm diode green laser (OcuLight GL; IRIDEX Corp, Inc., Mountain View, CA, USA) using a slit lamp delivery system and a Kaufman-Wallow plano fundus contact lens (Ocular Instruments Inc., Bellevue, WA, USA). Animals were anesthetized and nine areas symmetrically placed in the macula of each eye. The laser parameters included a 75-μm spot size and 0.1-second duration. The power setting was assessed by the ability to produce a blister and a small hemorrhage. If hemorrhage was not observed with the first laser treatment, a second laser spot was placed adjacent to the first following the same laser procedure, except at a higher wattage. For areas not adjacent to the fovea, the initial power setting was 500 mW; if a second spot was placed, the power was set to 650 mW. For the area adjacent to the fovea, the power setting was 400 mW for initial treatment and 550 mW for secondary treatment. At the discretion of the vitreoretinal surgeon, power settings were further adjusted based on observations at the time of photocoagulation. Fundus fluorescein angiography was conducted on days 8, 15, 22, and occasionally 29 following laser photocoagulation. Animals fasted for at least 2 hours prior to being anesthetized, and the eyes were dilated with a mydriatic agent. Animals were intubated due to the possibility of emesis following the fluorescein injection. Images were taken at the start and end of the intravenous fluorescein injection. Following fluorescein injection, a rapid series (approximately from dye appearance through 35 seconds) of stereo photographs of the posterior pole were taken of the right eye followed by a stereo pair of the posterior pole of the left eye. Additional stereo pairs were taken of both eyes at approximately 1 to 2 and 5 minutes after fluorescein injection. Between approximately 2 and 5 minutes after fluorescein injection, nonstereoscopic photographs were taken of two midperipheral fields (temporal and nasal) of each eye. Images of FFA were evaluated according to the following grading system which has been described previously⁴: grade I, no hyperfluorescence; grade II, hyperfluorescence without leakage; grade III, early- or midtransit hyperfluorescence and late leakage; grade IV, bright early- or midtransit hyperfluorescence and late leakage outside the borders of the treated area. Since grade IV lesions most closely resemble CNV seen in various human retinal disorders, including AMD, incidence of grade IV lesions was compared between groups. This analysis was performed by a masked observer. All animal procedures were ethically reviewed and approved according to the British Home Office Animals Scientific Procedures Act 1986 and were performed in accordance with European Directive 86/609/EEC, the GSK Policy on the Care, Welfare and Treatment of Animals, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Statistical Methods

The number of grade 4 lesions (0–9) per eye (the response Y) was modeled using a truncated Poisson distribution. The experimental design took advantage of historical controls provided by Covance (Covance database; Covance, Madison, WI, USA). Bayesian simulations were performed to determine group sample sizes that would ensure at least 80% power to detect a decreasing dose-related trend in the response; 1000 datasets from a 4-parameter logistic function with known parameters (Min and Max parameters set to historical positive and negative controls in the Covance database, Hill slope = −1, ED99 = 100) were simulated and a Bayesian (truncated) Poisson linear regression model with informative prior on the Maximum parameter (corresponding to the mean of the historical Vehicle data) was fit to each dataset. The posterior probability (P) of a decreasing dose-related trend (negative slope) was calculated and the number of datasets (n) with P > 80% was recorded. The group sample size was calibrated to keep the power (n/1000) > 80%. 

Final data analysis was conducted using a Bayesian (truncated) Poisson linear regression model with informative prior information from the Covance database. This model was fitted to all the data. The posterior probabilities of a decreasing dose-related trend (negative slope) were calculated as well as the posterior probability of the response at the maximal dose Y_dmax < 1 (efficacy similar to Lucentis). The Bayesian simulation work was carried out using the BRUGS package in R 3.1.1 (available in the public domain at https://cran.r-project.org/web/packages/BRugs/index.html) calling OpenBugs 3.2.3 in the background. The final data analysis was carried out in R. 

Results

IL-18 Prevents Spontaneous Neovascular Lesion Development

We previously reported that the proinflammatory cytokine IL-18 could prevent CNV, in a laser-induced mouse model of the condition.⁵ The laser-induced CNV model was established as a highly reproducible model in nonhuman primates; however, its use in the mouse eye is less straightforward and appears susceptible to variability as highlighted by Poor et al.⁶ in a comprehensive study reported in 2014. Therefore, we assessed the efficacy response of mouse IL-18 (SB-528775; GlaxoSmithKline [GSK], Stevenage, UK) on the formation of neovascular lesions in the central region of the retinas of the recently described JR5558 mice, which develop bilateral spontaneous CNV and retinal angiomatous proliferation (RAP; Figs. 1A, 1B).^7,8 Interleukin 18 was administered by intraperitoneal (IP) injection at doses of 0.1 or 1 mg/kg for 4 days either as lesions developed (between 3–4 weeks old) or when the lesions were fully established (8–10 weeks old), to assess the effect of IL-18 on the progression and regression of established lesions, respectively. Administration of IL-18 prevented both progression (Fig. 1C) and enhanced regression (Fig. 1D) of spontaneous neovascular lesions in the eyes of JR5558 mice as determined by fundus fluorescein angiography (FFA) on day 5. It is notable that in mouse models, anti-VEGF therapeutic approaches are not capable of regressing established lesions and that such effects are only observed when anti-VEGF therapeutics are used in combination with other agents targeting other mechanisms such as PDGF.⁹ Electroretinography measures photoreceptor rod and cone function. No changes in rod or cone function were observed in these mice post IL-18 treatment, as assessed by ERG (Figs. 1E, 1F), indicating no obvious deleterious effects as a consequence of IL-18 treatment. The observation that IL-18 treatment did not improve rod and cone function is consistent with the JR5558 phenotype of microvascular dysfunction and not retinal degeneration in the short term. 

Figure 1

 

Interleukin 18 prevents spontaneous neovascular lesion development. (A) Representative histopathological analysis of JR5558 mouse retinas at 5 weeks old. (B) Progression of neovascular lesion development in single JR5558 mouse from 5 up to and including 8 weeks old. (C) Fundus fluorescein angiography in vehicle control, IL-18 (1000 μg/kg) or (100 μg/kg) injected 3- or (D) 8-week-old JR5558 mice (×4 on consecutive days; FFA performed on day 5). Number of fluorescein enhancing lesions was significantly decreased in IL-18 injected mice (right histogram). *P < 0.05, ANOVA (n = 5 mice per group and data representative of four independent experiments). (E) Rod and (F) Cone photoreceptor electroretinography (ERG) a-wave (left) and b-wave (right) of JR5558 mice injected with IL-18 (100 μg/kg or 1000 μg/kg) for 4 consecutive days (n = 4/5 mice per group). (G) Primary mouse brain microvascular endothelial cells were grown to confluence on the apical chamber of a 0.4-μm Transwell filter. Mouse IL-18 (50 ng/mL) and VEGF (50 ng/mL) were added to the apical and basolateral chambers. Levels of occludin transcript were measured post treatment of cells with VEGF for 6 hours post treatment of cells with IL-18 and VEGF together in the apical chamber, IL-18/VEGF in the basolateral chamber for 6 or 24 hours. Data representative of means ± SEM, one-way ANOVA with Tukey posttest, P < 0.05 representing significance and data representative of two independent experiments. ***P < 0.001.

IL-18 Reduces Expression Levels of VEGFR2

Previously we observed that IL-18 could inhibit the secretion of VEGF from both endothelial and RPE cells.⁴ This appears to be regulated at a posttranscriptional level as IL-18 had no effect on VEGF-A mRNA expression (Fig. 2A) in retinal microvascular endothelial cells (RMVECs). To delve deeper into the effect of IL-18 on the VEGF signaling cascade, we measured mRNA levels of VEGF receptors on RMVECs (Figs. 2B–D). Interleukin 18 treatment inhibited VEGFR-2 expression specifically (Fig. 2C). Consequently, levels of basal and VEGF-induced VEGFR-2 protein decreased with IL-18 pretreatment in a time- and dose-dependent manner (Figs. 2E, 2F). Therefore, in conjunction with lowering levels of secreted VEGF,^4,10 the ability of IL-18 to lower VEGFR-2 expression in parallel indicates that IL-18 would effectively reduce VEGF signal transduction, and hence, the ability of endothelial cells to respond in a proangiogenic manner to any remaining VEGF. 

Figure 2

 

Interleukin 18 inhibits the expression of VEGFR2. (A) Vascular endothelial growth factor-A; (B) VEGFR-1; (C) VEGFR-2; (D) VEGFR-3 transcript levels in confluent RMVECs stimulated with IL-18 (50 ng/mL) for 0, 4, 8 and 24 hours, (n = 3 biological replicates). **P < 0.01, ***P < 0.001, ANOVA. (E) Vascular endothelial growth factor-2 protein expression in confluent RMVECs stimulated with IL-18 for 6 hours prior to VEGF treatment or (F) 24 hours prior to VEGF treatment. These experiments were run alongside each other so the top blot (i.e., VEGF stimulation alone, in the absence of IL-18 treatment) is the matching control for each panel. They have been included as reference blots in each of the figure subsets.

Mature IL-18 Is Safe and Reduces Development of Laser-Induced CNV in Cynomolgus Monkeys

Our data pertaining to a role for IL-18 in regulating neovascularization was supported by independent reports in the literature,^10,11 and strongly indicated that IL-18 could have some therapeutic benefit in neovascular AMD, in contrast to a report by Ambati et al.¹² Human IL-18 (SB-485232, GSK) is a clinically enabled investigational drug that has been injected systemically in >170 cancer patients with an excellent safety profile, with numerous on-going studies.^13,14 We sought to examine the tolerability and efficacy of SB-485232 in a nonhuman primate model of neovascular AMD. Human IL-18 has been reported to have bioactivity in cynomolgus monkeys.¹⁵ The laser-induced CNV model in nonhuman primates is a well-established robust model for screening therapeutic efficacy of drugs directed at neovascular AMD, albeit most of these studies have directly targeted the VEGF pathway.¹⁵ Monkeys were injected intravitreally (IVT) with IL-18 (0.01, 0.2, 1, 2.5, and 10 μg) post induction of laser lesions and FFA was performed on days 8, 15, and 22 (Fig. 3A). Following Bayesian analysis, the estimated probabilities for the existence of a dose-dependent, linear reduction in grade IV lesion counts (those which most likely resemble human neovascular AMD lesions) were 0.77, 0.97 and 0.9 for days 8, 15, and 22, respectively. In short, the data indicate that there is a 77%, 97%, and 90% probability that IL-18 is inducing a dose-dependent reduction in grade IV lesion counts when compared with vehicle control at days 8, 15, and 22 post laser. In Figure 3B, the horizontal red line in the plot depicts the estimated number of grade IV lesions in vehicle injected eyes and the prior historical controls for vehicle-treated eyes at the laboratory conducting these studies, and the horizontal green line depicts worst case scenario (one lesion as grade IV) observed in eyes injected with Lucentis (Fig. 3B). In contrast, the Bayesian assessment of whether the IL-18 data would be equivalent to the worst-case scenario of the Lucentis treatment indicated a low chance for IL-18 being similar to the worst case scenario for Lucentis with a 51%, 16%, and 17% chance for the 8, 15, and 22 days, respectively. Overall, the data demonstrate very clear IL-18 activity in reducing leakage from grade IV lesions, but this activity is not as pronounced as that of Lucentis. An interesting observation is the fact that the 10-μg dose appeared to be less efficacious than the 1- and 2.5-μg/eye doses. Interleukin 18 is approximately 95% homologous with the cynomolgus protein and, hence, is immunogenic in cynomolgus monkeys.¹⁶ The apparent reduced efficacy of the 10-μg dose when compared with lower doses may likely be due to this response which effectively neutralizes the bioactivity of IL-18. In addition, in flatmounts stained for F-actin, RPE cells appeared to maintain their honeycomb structure with no evidence of cellular dysmorphia observed at any dose or any time point post IVT injection of IL-18 (Fig. 4A). Furthermore, postmortem analysis of nonlasered ocular tissues 15 days post injection showed no toxic effect to either the retina or RPE, with no apoptotic (blebbing cells) or necrotic (swollen cells) phenotypes evident in any cell type at any dose (1, 3 or 10 μg IL-18; Figs. 4B–D). Additionally, histopathologic assessment of cynomolgus monkey retinal sections was carried out 15 days post IVT injection of IL-18 (1000, 3000, and 10,000 ng) and with no abnormal findings reported (Fig. 4E). Pharmacokinetic plasma analysis of IL-18 showed detectable and dose-proportional levels of IL-18 up to and including 7 days post IVT injection in all animals injected with the 10-μg dose of IL-18 (Supplementary Figs. S1, S2). 

Figure 3

 

Mature IL-18 reduces development of laser induced choroidal neovascularization in cynomolgus monkeys. (A) Grade IV lesion development post IVT injection of IL-18. Columns 1 and 2: Fundus photographs of monkeys post IVT injection of IL-18 (10, 100, 1000, 2500, and 10,000 ng). Columns 3–5: Fundus fluorescein angiography post IVT injection of 10, 100, 1000, 2500, and 10,000 ng IL-18. (B) Bayesian analysis of grade IV lesion development post IVT injection of IL-18. Vehicle (n = 2); 0.01 μg (n = 2); 0.1 μg (n = 3); 1 μg (n = 4); 2.5 μg (n = 5); 10 μg (n = 3); Lucentis (n = 2).

Figure 4

 

Mature IL-18 does not induce RPE cell death in cynomolgus monkeys. (A) Immunohistochemical analysis of flatmounts stained for isolectin (top row, red) or phalloidin (F-actin, bottom row, green) post IVT injection of vehicle control or IL-18 (10, 100, 1000, 2500, and 10,000 ng) or Lucentis. (B) Histopathological analysis of RPE and photoreceptor outer segments (OS) post intravitreal injection of vehicle control or IL-18 (10,000 ng), (C) vehicle control, and (D) IL-18 (10,000 ng) in high magnification. (E) Histopathological assessment of cynomolgus monkey retinal sections 15 days post IVT injection of IL-18 (1000, 3000, 10,000 ng human IL-18).

Discussion

We live in an aging society, with the US population aged older than 65 years expected to double by 2050. Consistent with this trend, the prevalence of age-related conditions is also increasing rapidly.¹⁷ Age-related macular degeneration is already the most common cause of central retinal vision loss in those aged older than 50 years globally and this figure is predicted to increase by 50% by 2020.¹ In fact, estimates of the global cost of visual impairment due to AMD alone are $343 billion (US), including $255 billion in direct health care costs.¹⁶ These statistics are stark and indicate the very substantial societal burden assumed by this condition, not only in relation to the negative impact on the quality of life of the affected individuals, but also the associated increased costs in health- and home care. Current state-of-the-art “anti-VEGF” therapies can involve monthly intraocular injections, which are a significant burden for patients and caregivers, and it is clear that improvements in treatment duration is a key market driver. Furthermore, there is also a need to provide alternative treatment options to subjects not responding to anti-VEGF therapies.¹⁸ Here we present data that indicates that mouse IL-18 inhibits vascular leakage in a mouse model of spontaneous neovascularization. We also present data indicating that human recombinant IL-18 is safe and has disease modifying efficacy in the nonhuman primate eye, can reduce CNV development in this species compared with vehicle control, and may now have major therapeutic utility as an adjunctive immunotherapy in the first instance in the context of neovascular AMD. 

Limitations of Results and Interpretation

The main limitation of this study has been the small numbers of animals used overall; however, given the nature of the study and the species used, we have had to refine the scope of the work, such that the minimum possible number of nonhuman primates required to show an efficacy response were used. In effect, in conjunction with GSK statisticians, a Bayesian approach was adopted for the design and implementation of the adaptive efficacy study. While it is abundantly clear that IL-18 is not as efficacious when compared with Lucentis in preventing CNV development, it certainly has a significant modifying effect on CNV lesion development when compared with vehicle control–injected eyes. In this regard, it is interesting to note that effective combinatorial approaches currently in development such as anti-PDGF aptamers have absolutely no impact on any preclinical CNV model when used alone. The mechanism of action of IL-18 is vastly different to current proven strategies, yet alone, it can reduce grade IV CNV development, and if used as an adjunctive therapy clinically, could yield important improvements in clinical disease management. However, given the limitations in all animal models of AMD, the only way we will now be able to assess the clinical utility of IL-18 is through robust clinical studies. 

Steps That Need to Be Taken for the Findings to Be Applied in the Clinic

As a small biotherapeutic agonist, IL-18 has the potential to eventually pave the way for a slow release encapsulated immunotherapeutic adjunctive strategy for neovascular AMD. The next step in the translational pipeline to clinic involves enabling IL-18 for IVT injection in human subjects. Currently, this developmental drug is being used in numerous clinical trials as a systemic infusion for cancer patients and has an excellent safety profile; however, it has never been tested as an IVT injection for any ocular condition. We are now working toward obtaining a full good laboratory practice compliant nonhuman primate safety toxicology assessment of IL-18 post IVT injection based on our observations to date. In that regard, we envision clinical application of this developmental drug upon completion of these studies in the short term. 

Acknowledgments

We thank Caroline Woods, Charles Murray, and David Flynn for animal husbandry. We also acknowledge Anran Wang and John Peterson for their help running the simulations for the Bayesian design of the primate study. 

Supported by Enterprise Ireland, GSK, Science Foundation Ireland (12/YI/B2614) and BrightFocus Foundation. The Ocular Genetics Unit at TCD is supported by SFI, the Health Research Board of Ireland, Irish Research Council for Science Engineering and Technology, and the European Research Council. 

Disclosure: S.L. Doyle, P; F.J. López, GlaxoSmithKline (F); L. Celkova, None; K. Brennan, None; K. Mulfaul, None; E. Ozaki, None; P.F. Kenna, GlaxoSmithKline (F); E. Kurali, None; N. Hudson, None; T. Doggett, None; T.A. Ferguson, None; P. Humphries, P; P. Adamson, GlaxoSmithKline (F); M. Campbell, GlaxoSmithKline (F), P