Nucleoside reverse transcriptase inhibitors and Kamuvudines inhibit amyloid-β induced retinal pigmented epithelium degeneration

Nonfibrillar amyloid-β oligomers (AβOs) are a major component of drusen, the sub-retinal pigmented epithelium (RPE) extracellular deposits characteristic of age-related macular degeneration (AMD), a common cause of global blindness. We report that AβOs induce RPE degeneration, a clinical hallmark of geographic atrophy (GA), a vision-threatening late stage of AMD that is currently untreatable. We demonstrate that AβOs induce activation of the NLRP3 inflammasome in the mouse RPE in vivo and that RPE expression of the purinergic ATP receptor P2RX7, an upstream mediator of NLRP3 inflammasome activation, is required for AβO-induced RPE degeneration. Two classes of small molecule inflammasome inhibitors—nucleoside reverse transcriptase inhibitors (NRTIs) and their antiretrovirally inert modified analog Kamuvudines—both inhibit AβOs-induced RPE degeneration. These findings crystallize the importance of P2RX7 and NLRP3 in a disease-relevant model of AMD and identify inflammasome inhibitors as potential treatments for GA.


INTRODUCTION
Geographic atrophy (GA), an irreversible and untreatable form of dry age-related macular degeneration (AMD), causes blindness in millions of individuals worldwide. 1 Extracellular deposit external to the retinal pigmented epithelium (RPE) termed drusen are hallmark pathological features of AMD. Pro-inflammatory components present in drusen are thought to drive inflammation and disease progression in GA. 2 Amyloid-β oligomers (AβOs) within drusen are linked to RPE degeneration in GA. [3][4][5] Inhibiting AβOsinduced neuroinflammation has been a therapeutic strategy pursued amyloid-driven neurodegenerative diseases. 6 Thus, identifying molecular sensors and inflammatory pathways mediating AβOs-induced RPE degeneration could provide insights to aid in developing GA therapies.
Inflammasomes are multimeric cytosolic protein complexes that recognize unique pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) and trigger innate immune responses by activating caspase-1dependent cytokine production and cell death. 7 AβOs induce activation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome in neurodegenerative disease models. [8][9][10] inflammasome activation is an indispensable driver of RPE degeneration in other models of GA. [11][12][13] Therefore, inhibition of NLRP3 activation is a promising approach to halt or delay RPE degeneration and disease progression in GA.
NLRP3 inflammasome activation generally comprises a two-step process in which both the initial priming and concomitant activating signals are required to produce a functional inflammasome. 14 Transcriptional upregulation of NLRP3, a predominant facet of priming, is an essential step, especially in non-immune cells such as the RPE because basal expression levels are considered insufficient to initiate inflammasome assembly. NLRP3 inflammasome activation also requires the assembly of a multiprotein complex that recruits pro-caspase-1 via the adapter protein ASC (an apoptosis-associated speck-like protein containing a caspase recruitment domain). ASC polymerization, which is necessary for NLRP3 inflammasome activation, 15 is characterized by the formation of large intracellular aggregates termed ASC specks, which are a signature of inflammasome assembly. Upon assembly, pro-caspase-1 is cleaved to caspase-1, which in turn cleaves the cytokine precursors pro-IL-1β and pro-IL-18 into mature IL-1β and IL-18, which are cytotoxic to the RPE. 14 The purinergic P2X7 receptor (P2RX7) is an essential mediator necessary for NLRP3 activation in several systems. 16 P2RX7 has been identified as a putative drug target in several models of GA. [17][18][19][20] AβOs-induced NLRP3 activation requires P2RX7 expression in microglial cells. 21 AβOs-induced RPE cytotoxicity has been recently reported to be driven by mitochondrial dysfunction and reactive oxygen species (ROS), 22 well-characterized inducers of NLRP3 inflammasome activation. 23 Despite multiple studies reporting AβOs-induced NLRP3 activation in RPE, [24][25][26] the role of P2RX7 remains unclear. Therefore, we tested if P2RX7 expression is necessary for AβOs-induced RPE degeneration.
Nucleoside reverse-transcriptase inhibitors (NRTIs) possess intrinsic anti-inflammatory activity independent of their antiretroviral function by virtue of their ability to inhibit P2RX7 and Fig. 1 AβOs promote NLRP3 inflammasome priming and assembly in the RPE. a RPE flat mounts, stained for zonula occludens-1 (ZO-1; red), of Nlrp3-GFP knock-in mice injected subretinally with AβOs show increased GFP expression (green) compared to RPE flat mounts of salineinjected mice, n = 6. b Subretinal injection of AβOs, but not PBS, induced increased ASC speck formation in ASC-Citrine Flox /Best1-Cre+ mice. ASC-Citrine fusion proteins are detected as a green signal. Subretinal injection of AβOs did not induce ASC speck formation in ASC-Citrine Flox mice, n = 6. c Higher magnification images of the observed ASC speck in RPE flat mounts of ASC-Citrine Flox /Best1-Cre+ mice injected with subretinal AβOs demonstrate the size of the specks. Selected areas of interest highlighted by the yellow squares depict the magnified regions. Scale bars (50 μm) subsequent NLRP3 activation. 17 Repurposing NRTIs has been suggested as a possible treatment strategy for several chronic diseases. 17,27,28 However, the toxicities associated with systemic NRTI use, which is attributed to their off-target inhibition of cellular polymerases, reduces enthusiasm for such therapeutic ventures. 29,30 Interestingly, the anti-inflammatory function of NRTIs is independent of their ability to inhibit reverse transcriptase. Modified NRTIs known as Kamuvudines, which retain the ability to inhibit inflammasome activation but lack the ability to inhibit reverse transcriptase and hence also lack the attendant toxicities, represent better candidates for treating P2RX7-NLRP3 mediated diseases. 17,27 Therefore, we investigated the efficacy of NRTIs and Kamuvudines to inhibit AβOs-induced RPE degeneration.
Here, we demonstrate that P2RX7 is an indispensable component necessary for AβOs-induced RPE degeneration and its inhibition by NRTIs and Kamuvudines blocks AβOs-induced RPE degeneration.

RESULTS
AβOs induce NLRP3 inflammasome assembly in the RPE We monitored NLRP3 inflammasome priming in the RPE in vivo following subretinal injection of AβOs by using Nlrp3-GFP mice, a Degeneration outlined by white arrowheads. Binary (Healthy %) and morphometric (PM, polymegethism (mean (SEM)) quantification of RPE degeneration is shown (Fisher's exact test for binary; two-tailed t-test for morphometry; *P < 0.001). Loss of regular hexagonal cellular boundaries in ZO-1 stained flat mounts is indicative of degenerated RPE. Scale bars (50 μm) reporter mouse line in which transcription of the fluorescent reporter is controlled by endogenous Nlrp3 regulatory elements. 31 Insertion of GFP in the NLRP3 locus renders these mice functionally deficient in NLRP3. Notwithstanding this disruption in NLRP3 protein expression, we also treated these mice with an intravitreous injection of Ac-YVAD-fmk, a caspase-1 inhibitor, to eliminate any residual inflammasome due to potential leakiness. This enabled us to visualize GFP signals free of distortions arising from potential degenerating cells. Following subretinal injection of AβOs in Nlrp3-GFP mice, confocal microscopic analysis of RPE flat mounts revealed increased GFP expression in AβOs-injected eyes compared to control-injected eyes (Fig. 1a), providing in situ Tissue was collected 7 days after injection. a P2rx7 −/− mice are protected from AβOs-induced RPE degeneration, n = 8. b Lower magnification (left panel) and higher magnification (right panel) of RPE flat mounts of P2rx7 hP2RX7Flox and P2rx7 hP2RX7Flox /Best1-Cre+ mice stained with phalloidin (white) and P2RX7 (yellow) demonstrating reduction of P2RX7 signal in the RPE of P2rx7 hP2RX7Flox /Best1-Cre+ mice compared to P2rx7 hP2RX7Flox mice. Black arrowhead points to the optic nerve of P2rx7 hP2RX7Flox /Best1-Cre+ mice, where expression of P2RX persists in non-RPE tissue. c AβOs induced degeneration in P2rx7 hP2RX7Flox (n = 6) but not in P2rx7 hP2RX7Flox /Best1-Cre+ mice (n = 8) (d). Representative images are shown. Fundus photographs, top row; Flat mounts stained for zonula occludens-1 (ZO-1; red), bottom row. Degeneration outlined by white arrowheads. Binary (Healthy %) and morphometric (PM, polymegethism (mean (SEM)) quantification of RPE degeneration is shown (Fisher's exact test for binary; two-tailed t-test for morphometry; *P < 0.01, **P < 0.001). Loss of regular hexagonal cellular boundaries in ZO-1 stained flat mounts is indicative of degenerated RPE. Scale bars for lower magnification (100 μm) and higher magnification (50 μm) evidence of NLRP3 priming in the RPE since GFP expression in these transgenic mice expression correlates with increased NLRP3 mRNA and protein expression. 32,33 Next, we monitored inflammasome assembly in the RPE in vivo by using ASC-Citrine Flox /Best1-Cre+ mice, which we generated by interbreeding ASC-Citrine Flox mice, a reporter mouse model that displays fluorescent ASC specks signifying inflammasome assembly, 34 with Best1-Cre+ mice, which express Cre specifically in the RPE. Following subretinal injection of AβOs, ASC-Citrine Flox /Best1-Cre+ mice displayed increased formation of 1-2 μm ASC specks in the RPE compared to control-treated mice (Fig. 1b, c). The ASC specks were restricted to RPE cells expressing Cre ( Supplementary Fig. 1).

DISCUSSION
Our studies demonstrate that AβOs induce NLRP3 inflammasome priming, assembly, and activation and that AβOs-induced RPE degeneration is dependent on NLRP3. Further, we show that AβOsinduced RPE degeneration requires the expression of P2RX7 in the RPE. Finally, we identify two FDA-approved NRTIs, as well as their less-toxic Kamuvudine derivatives as inhibitors of AβOs-induced RPE degeneration. Coupled with our earlier demonstration that SINE RNA-induced RPE degeneration also requires NLRP3 11 and P2RX7, 18 our data argue that the P2RX7-NLRP3 signaling axis is a common checkpoint in multiple GA-relevant models of RPE degeneration. In addition, the ability of NRTIs and Kamuvudines to inhibit RPE degeneration by both SINE RNAs 17,38 and AβOs enhances the rationale to test them in GA clinical trials.
GA is an example of a chronic, polygenic disease in which complex interplay between multiple genetic variants, environmental and lifestyle factors determine disease susceptibility, progression, and severity. 2,39 As with several other age-related diseases, the development of successful therapeutic strategies for GA remains challenging and has been a futile endeavor thus far. The predominant reason for this challenge is the multi-factorial nature of the disease associated with multiple intrinsic and extrinsic stimuli or stressors. 39 Despite varying initiating events, the invariant loss of RPE in GA supports the notion that this disease shares a common pathobiological course. 2 Recently, there has been considerable emphasis on identifying commonly shared pathways driving disease progression in several age-related diseases. [40][41][42] Chronic inflammation has been recognized as one of the tentative hallmarks that represent the common denominators of aging. 43 Among the several mediators of inflammation, the NLRP3 inflammasome consistently takes a central position in aging as it is activated by a vast variety of aging-associated danger patterns. 44,45 P2RX7-mediated NLRP3 inflammasome activation has been identified as a critical pathway necessary for RPE degeneration in several well-characterized models of GA. [18][19][20] Here, we show that P2RX7-NLRP3 inflammasome activation is essential for RPE degeneration induced by AβOs, another well-recognized stressor implicated in the pathogenesis of GA. Dysregulated activation of the complement system is also implicated as a potential stressor in the pathogenesis of GA. [46][47][48] Inflammasome genes are upregulated in human eyes with the complement factor H Y402H polymorphism, 49 which is associated with increased risk of AMD. 50,51 In addition, RPE cell culture models and animal models have also implicated inflammasome activation as an important driver of complement-induced RPE cytotoxicity. 52,53 Interestingly C3a and C5a, critical components of the complement system that are elevated in human AMD eyes 54 and putative drug targets for GA, 55 activate the NLRP3 inflammasome, 53,54,56 at least in part via P2RX7. 56 Collectively, it is tempting to speculate that P2RX7-mediated NLRP3 inflammasome activation constitutes a unifying molecular cornerstone across diverse pathological pathways in GA.
NRTIs inhibit P2RX7 and block NLRP3 inflammasome activation independent of their ability to block reverse transcriptase. 17 NRTIs are reported to have therapeutic potential against both forms of AMD, GA, and choroidal neovascularization, in animal models. 17,38,57 In addition, clinical evidence has recently emerged that the use of NRTIs reduces the development of type 2 diabetes, which is inflammasome-driven. 58 The efficacy of NRTIs and their non-toxic modifications, Kamuvudines, has been established in SINE RNA-induced RPE degeneration, a wellcharacterized model of GA. 17 This study, by reporting the efficacy of NRTIs and Kamuvudines in blocking AβOs-induced NLRP3 activation and subsequent RPE degeneration provides a strong impetus to explore these modified NRTIs as possible drug candidates for treating GA, particularly as they have a high therapeutic index.

Mice
All animal experiments were approved by the University of Virginia Institutional Animal Care and Use Committees and were performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Visual Research. Both male and female mice between 6 and 10 weeks of age were used in the study. Wild-type C57BL/6J, Best1-Cre, R26-CAG-LSL-ASC-citrine, and P2rx7 −/− mice were obtained from The Jackson Laboratory. Gsdmd −/− and Pycard −/− mice 61,62 described earlier were a generous gift from V.M Dixit (Genentech). Nlrp3 −/− and Casp1 −/− mice described earlier 63 were a generous gift from G. Nunez (University of Michigan). P2rx7 hP2RX7Flox mice have been previously described 37 (Supplementary Fig. 9 and Supplementary Table 1). Nlrp3-GFP mice described earlier 31,32 were a generous gift from F. Martinon and P. Schneider. Conditional RPEspecific knockout mice were generated by interbreeding the "floxed mice" with Best1-Cre mice. For all procedures, anesthesia was achieved by intraperitoneal injection of 100 mg/kg ketamine hydrochloride (Ft. Dodge Animal Health) and 10 mg/kg xylazine (Phoenix Scientific), and pupils were dilated with topical 1% tropicamide and 2.5% phenylephrine (Alcon Laboratories).
Fundus photography Fundus imaging of dilated mouse eyes was performed using a TRC-50 IX camera (Topcon) linked to a digital imaging system (Sony).
Assessment of RPE degeneration Seven days after SRI, RPE health was assessed by fundus photography and immunofluorescence staining of zonula occludens-1 (ZO-1) on RPE flat mounts (whole mount of posterior eye cup containing RPE and choroidal layers). Mouse RPE and choroidal flat mounts were fixed with 2% paraformaldehyde, stained with rabbit polyclonal antibodies against mouse ZO-1 (1:100, Invitrogen), and visualized with Alexa-594 (Invitrogen). All images were obtained by confocal microscopy (model A1R Nikon confocal microscope system, Nikon). The injection site was identified by the characteristic stellate pattern of the RPE at the injection site (Supplementary Fig. 10). The injection site and the surrounding area are first identified using lower magnification, followed by the acquisition of images using higher magnification. Imaging was performed by an operator blinded to the group assignments. RPE degeneration was quantified based on zonula occludens (ZO)-1-stained flat-mount images using two strategies: (1) Binary Assignment: RPE health was assessed based on the presence or absence of morphological disruption in RPE flat mounts by two independent raters who were masked to the group assignments. 13 Both raters deemed 100% of images as gradeable. (inter-rater agreement = 100%; Pearson r 2 = 1, P < 0.0001). (2) Semi-automated cellular morphometry analysis for hexagonally packed cells was performed by three masked graders as previously described. For this analysis, microscopy images of the RPE were captured and transmitted in deidentified fashion to the Doheny Image Reading and Research Lab (DIRRL). All images were rescaled to 304 × 446 pixels to permit importation into the Konan CellCheck software (Ver. 4.0.1), a commercial U.S. FDAcleared software that has been used for clinical trials. RPE cell metrics were generated by three certified reading center graders in an independent, masked fashion. Once the cell centers were defined, the software automatically generated the polymegethism values.
Immunostaining for NLRP3-GFP and ASC specks For NLRP3-GFP visualization, AβOs were injected into the subretinal space in Nlrp3-GFP reporter mice followed by immediate intravitreous administration (0.5 μl) of the Caspase-1 inhibitor (Ac-YVAD-fmk, InvivoGen) to prevent distortion of RPE cellular architecture without influencing inflammasome assembly. Forty-eight hours after SRI, RPE flat mounts were prepared and fixed as described above. Fixed RPE flat mounts were stained with Alexa-594 conjugated rabbit polyclonal antibodies against mouse ZO-1 (1:100, Invitrogen) and conjugated anti-GFP polyclonal Alexa-488 antibodies (1:100, Invitrogen). For ASC speck visualization, AβOs were injected into the subretinal space in ASC-Citrine Flox /Best1-Cre+ mice followed by immediate intravitreous administration (0.5 μl) of the Caspase-1 inhibitor (Ac-YVAD-fmk, InvivoGen) to prevent distortion of RPE cellular architecture without influencing inflammasome assembly. At 48 h after SRI, RPE flat mounts were prepared and fixed as described above. The fixed RPE flat mounts were stained with Alexa-594 conjugated rabbit polyclonal antibodies against mouse ZO-1 (1:100, Invitrogen) and rabbit polyclonal anti-Cre recombinase antibody (1:100, Abcam) followed by a goat antirabbit Alexa-647 antibody (1:200, Invitrogen). All images were obtained by confocal microscopy (model A1R Nikon confocal microscope system, Nikon). Imaging was performed by an operator blinded to the group assignments.

Histology
For hematoxylin and eosin staining, fresh, unfixed mouse eyes were embedded in Optimal Cutting Temperature Compound (Fisher), frozen in isopentane precooled by liquid nitrogen, and cryosectioned at 10 μm.
Electroretinography (ERG) ERG was performed 4 weeks after the subretinal injection. Mice (n = 4-6) were dark-adapted overnight before the experiments and anesthetized as described above. The pupils were dilated with tropicamide (1%) and phenylephrine (2.5%) eye drops. ERG was recorded using a Ganzfeld ERG (Phoenix laboratories). Scotopic combined responses were obtained using the LabScribe software (Phoenix Laboratories) under dark-adapted conditions (no background illumination, 0 cd/m 2 ) in response to white-flash stimuli ranging from −1.7 to 1.0 log cd s/m 2 with twenty responses averaged for each stimulus.

Statistics
The binary readouts of RPE degeneration (i.e., presence or absence of RPE degeneration on fundus and ZO-1-stained flat-mount images) were analyzed using Fisher's exact test. Cell morphometry data were assessed using a Student t-test. P values < 0.05 were deemed statistically significant.