Protective inflammasome activation in AMD

Age-related macular degeneration (AMD) is the most common cause of blindness in the elderly. AMD progression is associated with alterations in inflammatory pathways and the immune system. A new study identifies a protective role for inflammasomes in AMD, suggesting that inflammasome activation might be manipulated as a potential therapeutic strategy for this condition (pages 791–798).

insulin levels) might constitute better surgical criteria than BMI 9 .
How do various operations compare with one another in RCTs powered to see smaller differences? What is the full surgical risk/ benefit ratio in less-obese patients with T2DM, and how low a BMI might we contemplate? Should we consider ethni city because populations such as Asians develop T2DM at lower BMIs? What about long-term effects? Emerging data suggest that rates of diabetes recrudescence after initial post operative remission can be high; however, even a transient disease-free period might confer long-lasting benefits from 'metabolic memory' . RCTs measuring important clinical endpoints, such as cardiovascular events, would be preferable to existing studies of surrogate endpoints like glycated hemoglobin. Such studies require thousands of participants. Given how difficult it is to randomize people into surgical versus nonsurgical interventions, novel methods for recruiting subjects in equipoise are needed, and relevant efforts are under way.
Nevertheless, bariatric surgery is a safe, effective treatment for T2DM, and is likely to benefit people with diabetes and only mild obesity. Several operations activate weightindependent antidiabetes mechanisms 1,2 , rendering benefits related to improved glycemia more than weight loss 9 , and BMI is In Mingrone et al. 6 , there was no corre lation after either operation between the magnitude of changes in body weight and glucose levels, and neither baseline weight nor postoperative weight loss predicted diabetes remission or glucose lowering. Moreover, glycemic improvements preceded substantial weight loss. In Mingrone et al., all surgical subjects achieved glycemic control without diabetes medications within only 15 days after surgery. In Schauer et al. 7 , diabetes medication use plummeted postoperatively, long before maximum weight loss. Thus, the new studies corroborate a growing body of evidence showing that weight-independent mechanisms contribute to diabetes remission after some bariatric operations.
Numerous questions remain. What is the proper timing of surgery relative to other T2DM therapies? Because long diabetes duration is the strongest predictor of postoperative nonremission 2 , surgery could be considered earlier in the disease, rather than as salvage therapy 4 . What are optimal criteria for patient selection? Although BMI is the standard, neither baseline weight nor postsurgical weight loss predict surgical benefits to glycemia 6 or cardio vascular disease 9 . Such benefits seem more related to improved glucose homeostasis than weight loss, and measures of the former (for example, fasting a suboptimal stand-alone surgical criterion. Overall, it is time to view these operations as metabolic, not just bariatric, surgery.

COMPETING FINANCIAL INTERESTS
The author declares competing financial interests: details accompany the full-text HTML version of the paper at http://www.nature.com/naturemedicine/. The appearance of white or yellow lipid deposits beneath the retina, called drusen, marks early AMD. These deposits gradually separate the retinal pigment epithelium (RPE) from the underlying choroidal vascular bed, which provides oxygen and nutrients to photoreceptors. The RPE also helps clear cellular debris such as shed photoreceptor outer segments, which are renewed daily to replenish the enzymes necessary for light activation. The health of these photon-sensitive photoreceptor cells in the retina relies on a functional RPEchoroid interface. As drusen deposits increase in number, size or both, the proper functions of RPE and choroid are disrupted, and the photoreceptors degenerate, resulting in 'dry' or non-neovascular AMD. Abnormal leaky Every year, millions of elderly adults worldwide lose their ability to read, drive or recognize faces as a result of age-related macular degeneration. Fine vision is lost in the central visual field (the macula), resulting in blurred images or a central 'black hole' (scotoma). Although people with AMD may have sufficient peri pheral vision to allow them to perform basic daily activities, many ordinary tasks normally taken for granted, such as reading, can become challenging. As the number of aging people in the world increases, AMD will become an ever-increasing health burden in the next several decades. Thus, understanding the pathogenesis of AMD is crucial to finding ways to treat or prevent the vision loss associated with this disease.
AMD is a complex, multifactorial disease with both environmental and genetic risk factors, many of which are associated with inflammation and immunity. Both smoking and low omega-3 fatty-acid intake are strongly associated with AMD 1,2 , and both are also associated with increased inflammation. Genetic studies have linked innate immunity (complement factor H) with AMD 3 . It is generally believed that increased inflammation contributes to AMD progression, and suppression of inflammation is needed for disease prevention. expression of the proangiogeneic factor vascular endothelial growth factor. Together these data indicate that, through NLRP3 and IL-18, activated inflammasomes have a protective role in AMD progression. It is not known, however, whether depletion of other inflammasome components (such as ASC and caspase-1) will similarly worsen CNV. It will also be interesting to see whether enhanced inflammasome activation or increased levels of IL-18 can suppress CNV. Nevertheless, the protective role of inflammasomes in AMD observed by Doyle et al. 5 challenges the existing dogma that global suppression of inflammatory processes will prevent AMD. This study suggests a paradigm-shifting concept: that controlled inflammatory responses may actually be beneficial in the initial onset of AMD (as in other situations such as infection). Only when the inflammatory response fails to repair tissue damage and becomes dysregulated does it have detrimental effects 9 .
The newly revealed protective role of inflammasomes and their regulation of inflammatory cytokines in AMD also suggest new targets for control of AMD. Much of the current research on AMD pathogenesis has focused on chemokines, such as CCL2 and CX3CR1, that help recruit macrophages to the lesion sites 10-12 . Doyle et al. 5 direct attention to the pathways within macrophages that are activated to influence the progression from dry to wet AMD. Inflammasome-mediated processes are recog nized as important steps in sensing both exo genous (for example, microbial) and endo genous (such as stress and tissue-derived) signals that not only guard against infection but also regulate autoimmune responses and multiple metabolic processes such as obesity and diabetes 6 . Now it seems that inflammasome activation also has a key role in sensing extracellular deposits and cellular damage in AMD to mediate disease progression through IL-18.
This study and future studies on the role of inflammasomes in AMD might lead to new treatments based on manipulation of inflammasome activation and promotion of the protective inflammatory response, as well as direct modulation of IL-18. Currently, there is no standard treatment for dry AMD except dietary supplementation with vitamins, antioxidants and dietary lipids that are associated with decreased risk of progression to wet AMD. For wet AMD, suppression of CNV with vascular endothelial growth factor-specific antibody therapy has been remarkably success ful and is now the first line of treatment. However, preventive and noninvasive therapies for AMD would be preferable. Development of such inflammatory cells, including macrophages, and can be activated by a diverse range of microbial or host damage signals to assemble into oligomers and initiate the apoptotic cascade 6 .
Doyle et al. 5 found that NLRP3 inflammasomes can be activated by drusen isolated from eyes of donors with AMD and by the known drusen components C1q (a protein complex in complement activation) and carboxyethyl-pyrrole (CEP), an oxidative protein modification derived from docosahexaenoate-containing lipid 4 . Mice immunized with CEP had AMD-like lesions, drusen accumulation and other characteristics of dry AMD 7 , and the authors found that NLRP3 and caspase-1 were induced in the macrophages near AMD-like lesions in these mice 5 , suggesting that inflammasome activation occurs near lesion sites.
Upon their activation, NLRP3 inflammasomes can increase the production of inflammatory cytokines such as interleukin-1β (IL-1β) and IL-18 to mediate inflammation and tissue responses 8 . Using a laser-induced CNV mouse model of wet AMD, Doyle et al. 5 showed that lack of NLRP3 or IL-18 exacerbates CNV formation and that this is likely to be mediated through IL-18-regulated blood vessels may also arise from the choroid growing through disrupted RPE into the subretinal space, causing rapid and substantial vision loss in 'wet' or neovascular AMD (Fig. 1). Determining the pathways that regulate progression from the often asymptomatic dry to the sight-threatening wet AMD is of prime importance. Previously, local chronic inflammation was suggested to have a key role in the formation of drusen on the basis of analyses of drusen components 4 . Activation of innate immunity (in particular the complement system) is generally believed to induce cellular damage in the retina, leading to choroidal neovascularization (CNV).
In this issue of Nature Medicine, Doyle et al. 5 provide evidence that some aspects of inflammation may be beneficial and slow the transition from dry to wet AMD through the activation of a protective inflammasome response. Inflammasomes are a group of recently discovered multiprotein complexes composed of an inflammatory component such as NOD-like receptor family, pyrin domaincontaining 3 (NLRP3) and an apoptotic component (such as caspase-1), which are connected through an adaptor protein ASC (also known as PYCARD) 6 (Fig. 1). They are expressed in Inflammasome npg therapies will largely depend on additional fundamental research into AMD to decipher the complex network of inflammatory and other mediators in this devastating blinding disease. approach to compare the cardiac tissue transcriptomes from subjects with DCM with and without RBM20 mutations, as well as those from wild-type and Rbm20-mutant rats. They found that RBM20 mutations dramatically altered not only titin isoform expression but also the splicing of 30 other cardiac proteins, most of which have been previously implicated in cardiomyopathy and heart failure. Importantly, the composition of this network of RBM20-regulated proteins was conserved between humans and rats. The authors largely excluded potential secondary effects on splicing as a result of heart disease.
It remains to be determined which RBM20 substrates are pivotal for the disease phenotype in DCM. Whereas people with an RBM20 mutation often have clinically aggressive DCM 3,6 , heterozygous RBM20-deficient rats show less severe, DCM-like symptoms, such as mildly enlarged hearts, fibrosis and a somewhat increased risk for arrhythmias and sudden cardiac death 2 . The subset of proteins aberrantly spliced in human RBM20-and rat Rbm20mutant hearts includes molecules predicted to be involved in sarcomere structure and function, ion transport and Ca 2+ handling (Fig. 1). The functional impairment of any one protein from this network, or a combination thereof, could cause the DCM phenotype. For example, loss of function of titin alone could be responsible for DCM, as mutations in the titin-spring region can predispose to arrhythmogenic right ventricular cardiomyopathy 7 . Furthermore, the preferential expression of abnormally large titin isoforms in human RBM20-mutant hearts, as documented by Guo et al. 2 , would be expected to make cardiomyocytes less stiff Heart muscle diseases, or cardiomyopathies, rank among the most prevalent causes of premature death in the Western world. A subset of cardiomyopathies is heritable, and causative mutations have been detected in almost 50 different genes so far. Among these primary cardio myopathies, the most common form is dilated cardiomyopathy (DCM), which has been associated with more than 30 known disease genes. In DCM the heart becomes weakened and enlarged and cannot pump blood efficiently. The most frequently mutated gene in human DCM is TTN, which encodes the elastic muscle protein titin, the largest protein in the human body 1 . Notably, titin is also the focus of a new study published in this issue of Nature Medicine by Guo et al. 2 , which provides new insights into the pathology of a hitherto underappreciated form of familial DCM. The authors concentrate on individuals with DCM in whom a component of the cardiac splicing machinery called RNA binding motif protein-20 (RBM20) is mutated 3 . The mutation of RBM20 leads to defects in splicing, triggering the expression of aberrant isoforms of titin, as well as of a network of other myocardial proteins that are collectively regulated by RBM20.
An important component of the study of Guo et al. 2 was the use of a rat strain with a spontaneously occurring mutation, which the same team had previously shown to result in the expression of irregular titin isoforms during heart development 4 . Mammals have a single gene that encodes titin, and the differences between titin isoforms largely arise from alternative splicing of the 'molecular spring region' of the protein (Fig. 1), which is a main determinant of myocardial passive (diastolic) stiffness. Normally, fetal hearts express a large N2BA-type titin isoform of approximately 3,700,000 Da 5 . This isoform is replaced around birth by smaller N2BA variants and the N2B isoform (~3,000,000 Da), which is the predominant titin isoform in the adult heart and is the least compliant of these titin isoforms. This perinatal titin isoform switch leads to the stiffening of the titin springs of adult cardiac sarcomeres. Strikingly, the team had previously shown that the hearts from rats homozygous for the spontaneously occurring mutation express a giant titin isoform called N2BA-G (3,830,000 Da) but no N2B isoform throughout life, making the adult cardiomyocytes from these mutant rats more compliant than normal 4 . Moreover, rats heterozygous for the mutation expressed large N2BA-type titins and low amounts of N2B, even in adulthood 4 . These pathological changes suggested anomalous alternative splicing of titin was occurring in the mutant rats.
Using genetic mapping, Guo et al. 2 have now identified the cause of the aberrant pattern of titin isoforms in the previously identified mutant rats. They showed that the rats have a missense mutation in the gene encoding the splicing regulator RBM20 that removes all of the gene except for the first exon. RBM20 mutations have previously been detected in ~3% of human subjects with DCM 3,6 , and Guo et al. 2 found yet another one (S635A). The authors used an RNA deep-sequencing