Kamuvudine-9 Protects Retinal Structure and Function in a Novel Model of Experimental Rhegmatogenous Retinal Detachment

Abstract

Purpose: Rhegmatogenous retinal detachment (RRD) is a vision-threatening event that benefits from surgical intervention. While awaiting surgical reattachment, irreversible hypoxic and inflammatory damage to the retina often occurs. An interim therapy protecting photoreceptors could improve functional outcomes. We sought to determine whether Kamuvudine-9 (K-9), a derivative of nucleoside reverse transcriptase inhibitors (NRTIs) that inhibits inflammasome activation, and the NRTIs lamivudine (3TC) and azidothymidine (AZT) could protect the retina following RRD.

Methods: RRD was induced in mice via subretinal injection (SRI) of 1% carboxymethylcellulose (CMC). To simulate outcomes following the clinical management of RRD, we determined the optimal conditions by which SRI of CMC induced spontaneous retinal reattachment (SRR) occurs over 10 days (RRD/SRR). K-9, 3TC, or AZT was administered via intraperitoneal injection. Inflammasome activation pathways were monitored by abundance of cleaved caspase-1, IL-18, and cleaved caspase-8, and photoreceptor death was assessed by TUNEL staining. Retinal function was assessed by full-field scotopic electroretinography.

Results: RRD induced retinal inflammasome activation and photoreceptor death in mice. Systemic administration of K-9, 3TC, or AZT inhibited retinal inflammasome activation and photoreceptor death. In the RRD/SRR model, K-9 protected retinal electrical function during the time of RRD and induced an improvement following retinal reattachment.

Conclusions: K-9 and NRTIs exhibit anti-inflammatory and neuroprotective activities in experimental RRD. Given its capacity to protect photoreceptor function during the period of RRD and enhance retinal function following reattachment, K-9 shows promise as a retinal neuroprotectant and warrants study in RRD. Further, this novel RRD/SRR model may facilitate experimental evaluation of functional outcomes relevant to RRD.

Rhegmatogenous retinal detachment (RRD) is a serious and relatively common ocular pathology. Incidence estimates of RRD range from 6.3 to 17.9 per 100,000 persons.¹ RRD results from passage of vitreous humor through a retinal tear into the subretinal space, separating the neurosensory retina from the underlying retinal pigmented epithelium. Once detached, photoreceptors are vulnerable to hypoxia and damaging inflammatory signaling, which hasten photoreceptor apoptosis and vision loss.^2–6 In this emergent context, prompt therapeutic action is needed to minimize lasting damage. 

Currently, RRD management hinges on surgical repair. Even with successful reattachment, however, many patients have lasting visual deficits.⁷ In “macula-off” RRD repaired within 9 days, only half of patients may regain 20/50 vision.⁸ Visual outcomes worsen with increasing delay in surgical repair. For example, only one-third of patients who undergo surgery after 10 to 19 days regain 20/50 vision,⁸ and even delays of more than 3 days have been reported to result in worse visual outcomes.⁹ While detached, damage to the retina leaves residual functional deficits even after anatomic repair. Therefore, a targeted therapy might provide protective benefits while patients await RRD surgical repair. 

One candidate pathway is the NLRP3 inflammasome, a multimeric protein complex that recognizes damage-associated molecular patterns released by injured photoreceptors and launches a potent inflammatory response characterized by caspase-1 activation, release of interleukins, and cell death.¹⁰ Previously, we discovered that nucleoside reverse transcriptase inhibitors (NRTIs), which are used to treat HIV and hepatitis B infections, also inhibit inflammasome activation independent of their antiretroviral activity.¹¹ Previous studies by us and others demonstrated that NRTIs block inflammasome activation and its associated damage in a variety of contexts, including models of diabetes,¹² diabetic retinopathy,^13,14 aging,¹⁵ choroidal neovascularization,¹⁶ and atrophic age-related macular degeneration.^11,17−19 Therefore, we hypothesized that the anti-inflammatory effects of NRTIs might also be protective in RRD. In addition to NRTIs, which have numerous adverse effects owing to their off-target effects on host polymerases,²⁰ we also tested Kamuvudine-9 (K-9), an NRTI derivative that has been engineered to avoid undesirable host polymerase inhibition yet retain inflammasome suppression.^11,18,21 

We employed a mouse model of RRD in which a retinal tear is created and a viscous substance, carboxymethylcellulose (CMC), is injected into the subretinal space, thereby simulating human RRD in which the vitreous humor migrates into the subretinal space via the retinal tear created by vitreoretinal traction. In the standard RRD model, the retina remains detached for weeks, often months, during which time electroretinography (ERG) is nearly nonrecordable.²² Thus, the ability of an intervention to protect or improve electrical function cannot be assessed. Therefore, we developed a new mouse model of RRD with spontaneous retinal reattachment (SRR) to better simulate the clinical course of RRD and to enable evaluation of K-9 treatment on functional outcomes following retinal reattachment. 

Materials and Methods

Mice

All experiments involving animals were approved by the University of Virginia Institutional Animal Care and Use Committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Animal subjects were male and female mice aged 6 to 10 weeks. Wild-type C57BL/6J mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). To anesthetize mice for study procedures, 2,2,2,-tribromoethanol, 99% (375 mg/kg; cat. 75-80-9; Alfa Aesar, Tewksbury, MA, USA), ketamine hydrochloride (100 mg/kg; Ft. Dodge Animal Health, Overland Park, KS, USA), and xylazine (10 mg/kg; Phoenix Scientific, St. Joseph, MO, USA) were delivered by intraperitoneal (IP) injection. Pupils were dilated with topical 1% tropicamide and 2.5% phenylephrine (Alcon Laboratories, Ft. Worth, TX, USA). 

Retinal Imaging

Fundus images were obtained using the TRC-50 IX camera (Topcon, Tokyo, Japan), linked to a digital imaging system (Sony, Tokyo, Japan) or Mm IV Retinal Microscope (Phoenix Research Laboratories, Pleasanton, CA, USA). Spectral domain optical coherence tomography (OCT) was performed by attaching an OCT2 scan head to the Mm IV Retinal Microscope (Phoenix Research Laboratories). 

RRD Models

A persistent RRD model in mice was established by subretinal injection (SRI) of 3 µL 1% CMC (Refresh Liquigel; Allergan, Irvine, CA, USA). The final model of RRD/SRR was achieved by SRI of 3 µL 0.03% CMC. In order to establish the RRD/SRR model, OCT images were obtained 1 day post-SRI of 0.03% CMC ranging in volume from 1 to 4 µL. OCT images were also obtained 3 days post-SRI of 3 µL CMC ranging in concentration from 0.01% to 1%. OCT images for the established RRD/SRR model were obtained on days 0, 1, 2, 3, 5, 7, and 10 post-SRI of 3 µL 0.03% CMC for representation as well as to establish spontaneous retinal reattachment over time. SRIs were performed as previously described²³ using a 35-gauge needle (Ito Co., Fuji, Japan). 

NRTI and Kamuvudine-9 Treatment

NRTIs and K-9 were delivered via IP injection in equimolar amounts. Lamivudine (3TC, 50 mg/kg, molecular weight (MW) = 229.26 g/mol) or azidothymidine (AZT, 60 mg/kg, MW = 267.24 g/mol) was injected twice daily (SelleckChem, Houston, TX, USA). K-9 (alkylated derivative of 3TC, MW = 271.34 g/mol) was synthesized as previously described^21,24 and administered either at a low dose (K-9L; 90 mg/kg daily) or a high dose (K-9H; 60 mg/kg twice daily, equimolar to NRTI doses). IP injection of phosphate-buffered saline (PBS) was used as control. 

Immunoblotting

Retinal tissue was extracted and lysed by sonication in radioimmunoprecipitation assay buffer (cat. 89900; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with protease inhibitor (cat. 11697498001; Sigma, St. Louis, MO, USA) and phosphatase inhibitor (cat. 4906845001; Sigma). The Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) was used to determine protein concentrations of interest. Samples were prepared that contained equal concentrations of total protein (10−100 µg). Proteins were denatured by boiling for 10 minutes in Laemmli buffer with β-mercaptoethanol, resolved by SDS-PAGE, and then transferred onto Immobilon-FL PVDF membranes (Millipore, Billerica, MA, USA). Membranes were blocked in Odyssey Blocking Buffer (PBS) or LI-COR blocking buffer (PBS) for 1 hour at room temperature and then incubated overnight with primary antibody at 4 °C. Across study experiments, the following antibodies were used: anti-mouse caspase-1 (1:1000; AG-20B-0042; AdipoGen Life Sciences, San Diego, CA, USA), anti-mouse caspase-8 (1:1000; gift of Dr. R. Hakem, University of Toronto), GAPDH (1:1000; 14C10, 2118; Cell Signaling Technology, Danvers, MA, USA), and anti-human/mouse/rat vinculin (1:1000; clone 7F9; Thermo Fisher Scientific). To visualize immunoreactive bands, species-specific secondary antibodies conjugated with IRDye were used (1:2000; LI-COR Biosciences, Lincoln, NE, USA). Immunoblot images were captured using the Odyssey CLx Imaging System (LI-COR Biosciences) or autoradiography film. 

IL-18 ELISA

Total protein from retinas was quantified as detailed above. Secreted IL-18 in each sample (100 µg protein) was detected by ELISA (Mouse IL-8 ELISA, DY7625; R&D Systems, Minneapolis, MN, USA), according to the manufacturer's instructions. 

TUNEL Assay

Retinal cross cryosections at a thickness of 10 µm were stained with TUNEL using the In Situ Cell Death Detection Kit (Roche Diagnostics, Indianapolis, IN, USA) according to the manufacturer's instructions, as previously described.²⁵ The sections were wet-mounted with ProLong anti-fade reagent with DAPI (P36935; Thermo Fisher Scientific)²⁶ and visualized using an A1R laser confocal microscope (Nikon, Tokyo, Japan). To calculate the number of TUNEL-positive cells in the outer nuclear layer (ONL), the retina analysis toolkit of ImageJ (2.1.0; National Institutes of Health, Bethesda, MD, USA) was used. 

Electroretinography

ERG was recorded at baseline, as well as days 7, 11, 14, and 28 following SRI of CMC. Eyes that developed cataracts were excluded before ERG measurement (less than 5%). Mice were dark adapted overnight, and then ERG was measured using Ganzfeld ERG (Phoenix Research Laboratories). LabScribe Software (version 3.016800; Phoenix Research Laboratories) was used to collect scotopic combined responses with no background illumination (0 cd/m²). Responses were recorded to white-flash stimuli ranging from –1.7 to 1.0 log cd s/m². For each stimulus, three responses were obtained and averaged. 

Statistical Analysis

Quantitative results were reported as the mean ± SEM. The abundance of cleaved caspase-1 over time was tested for increasing monotonicity using the Mann–Kendall trend test applied to the median values at each time point. Comparisons between drug and vehicle treatments for immunoblotting, TUNEL staining, and ELISA were preplanned, and thus adjustment for multiple comparisons is not required. Nevertheless, to be conservative, these data were analyzed using one-way ANOVA followed by Tukey's multiple comparisons test. ERG a-waves and b-waves were analyzed using a linear mixed-effect model, which was fit using treatment and day as fixed effects, along with an eye-specific random effect to account for the longitudinal data. Mixed-effect models were fit using the R package lme4.²⁷ P values were generated using the R package lmerTest.²⁸ P values <0.05 were considered statistically significant. 

Results

Experimental RRD induced a progressive increase in the abundance of cleaved caspase-1 in retinal lysates isolated from mice over 24 to 72 hours (Figs. 1A, 1B). These data indicate in vivo inflammasome activation following RRD, and they are consistent with the prior literature demonstrating that hypoxia inducible factor-1α, which is increased in RRD,²⁹ induces inflammasome activation.^30,31 Three days after inducing RRD, mice treated with systemic K-9 exhibited dose-dependent and complete reduction in retinal cleaved caspase-1, compared with PBS-treated mice (Figs. 1C, 1D). Three days after inducing RRD, mice treated with either of the NRTIs (3TC or AZT) exhibited a marked reduction in retinal caspase-1 levels, compared with PBS-treated mice (Figs. 1E, 1F). These results are compatible with the known inflammasome inhibitory activity of K-9 and NRTIs.^{11,12,17,18,21} 

Figure 1.

 

K-9 and NRTIs inhibit RRD-induced caspase-1 cleavage. (A) Representative immunoblot of caspase-1 in the retina in the mouse RRD model at baseline and 24, 48, and 72 hours post-SRI (3 µL 1% CMC; n = 3). (B) Densitometric quantification of cleaved caspase-1 from A, normalized to GAPDH (P value from Mann–Kendall trend test; n = 3 eyes per group; data shown as mean ± SEM). (C) Representative immunoblot of caspase-1 in the retina on day 3 of the RRD model after treatment with PBS, K-9L (90 mg/kg IP once daily), or K-9H (60 mg/kg IP twice daily), n = 3. (D) Densitometric quantification of cleaved caspase-1 from C, normalized to GAPDH (P values from one-way ANOVA followed by Tukey's test; n = 3 eyes per group; data shown as mean ± SEM). (E) Representative immunoblot of caspase-1 in the retina in RRD model on day 3 after treatment with PBS, 3TC, or AZT (n = 3). (F) Densitometric quantification of cleaved caspase-1 from E normalized to GAPDH (P values from one-way ANOVA followed by Tukey's test; n = 3 eyes per group; data shown as mean ± SEM).

Next, we analyzed retinal neuroprotection by assessing TUNEL staining to measure dead or dying cells. Three days after inducing RRD, we observed a dose-dependent reduction in the number of TUNEL-positive cells in the ONL of the retina following K-9 treatment, compared with PBS treatment (Fig. 2A). Similarly, mice treated with systemic 3TC or AZT displayed significantly fewer TUNEL-positive retinal cells, compared with PBS-treated mice (Fig. 2B). These findings demonstrate that systemic delivery of these three inflammasome inhibitors can protect photoreceptors from damage during periods of RRD. Of note, equimolar doses of K-9H conferred a greater degree of photoreceptor protection than 3TC (56% ± 7% vs. 38% ± 10%, P < 0.05). 

Figure 2.

 

K-9 and NRTIs protect photoreceptors in RRD model. (A) Left panels show representative confocal images of retinal ONL on day 3 of the RRD model after treatment with PBS, K-9L, or K-9H (n = 5–6 eyes per group), stained with TUNEL (top; green) and TUNEL (green)/DAPI (blue) (bottom). Scale bars: 100 µm. Right panel shows quantification of TUNEL-positive ONL cells (P values from one-way ANOVA followed by Tukey's test; n = 5–6 eyes per group; data shown as mean ± SEM). (B) Left panels show representative confocal images of retinal ONL on day 3 of the RRD model after treatment with 3TC, AZT, or PBS (n = 5–6 eyes per group), stained with TUNEL (top; green) or TUNEL (green)/DAPI (blue) (bottom). Scale bars: 100 µm. (B) Right panel shows quantification of TUNEL-positive ONL cells (P values from one-way ANOVA followed by Tukey's test; n = 5–6 eyes per group; data represented as mean ± SEM).

We next sought to understand potential mechanisms of neuroprotection induced by K-9. Consistent with RRD-induced inflammasome activation, we found that levels of IL-18, a key output of inflammasome activation, were elevated in the retina following RRD (Fig. 3A). We also found that K-9 treatment inhibited IL-18 levels in the detached mouse retina compared to PBS treatment (Fig. 3A). Next, we studied caspase-8, a proapoptotic molecule that is elevated in mouse models of RRD³² and also important in inflammasome signaling.²⁶ Following RRD, we observed increased levels of cleaved (active) caspase-8³³ in the mouse retina (Fig. 3B) and that K-9 treatment inhibited caspase-8 cleavage compared to PBS treatment. These findings provide mechanistic insights into neuroprotection conferred by K-9. 

Figure 3.

 

K-9 inhibits IL-18 and caspase-8 activation in RRD model. (A) ELISA analysis of IL-18 in attached mouse retina (RRD–) or detached retina treated with PBS or K-9 (60 mg/kg) IP twice daily. ND, not detected. P values from one-way ANOVA followed by Tukey's test; n = 4 eyes per group; data represented as mean ± SEM. (B) Representative immunoblot of caspase-8 in attached mouse retina (control) or detached retina 1 day after treatment with PBS or K-9. n = 3.

To simulate the clinical course of RRD, wherein most retinas are reattached within a week, we developed a novel model of RRD with spontaneous retinal reattachment (RRD/SRR) by testing a range of lower CMC concentrations (0.01%−1%) and volumes (1−4 µL) and monitoring the temporal evolution of the RRD (Fig. 4). We found that an SRI volume of 3 µL and concentration of 0.03% resulted in SRR over a period of 5 to 10 days (Fig. 5), resembling the usual clinical time course of surgical restoration of retinal apposition to the underlying retinal pigment epithelium and normoxia. 

Figure 4.

 

SRI concentration and volume curves used to develop the RRD/SRR model. (A) Image-guided spectral domain optical coherence tomography (SD-OCT); representative retinal images of mice 3 days post-SRI of 3 µL CMC of various concentrations 0.01% to 1% (n = 3). (B) Image-guided SD-OCT; mice retinas 1 day post-SRI of 0.03% CMC of various volumes 1 to 4 µL (n = 3). Scale bars: 100 µm.

Figure 5.

 

Kinetics of retinal attachment in RRD/SRR model. (A) Image-guided SD-OCT; representative images of the right (OD) and left (OS) retinas of wild-type mice over days 0 to 10 post-SRI of 3 µL 0.03% CMC (n = 4). Scale bars: 100 µm. (B) Incidence of SRR over time (n = 4).

We then tested K-9 in this RRD/SRR model via systemic administration during the first 10 days following RRD (Fig. 6A). We found that PBS-treated mice displayed larger reductions in the amplitudes of both a- and b-waves over days 7 to 28, as measured by full-field scotopic ERG, when compared to mice treated with K-9 (Figs. 6B, 6C). Interestingly, despite anatomic reattachment of the retina, the ERG in PBS-treated animals revealed persistently poor a- and b-wave responses up to 14 days following RRD (Figs. 6B, 6C). In contrast, K-9 preserved retinal electrical function during the period of RRD and reattachment (Figs. 6B, 6C). Notably, although K-9 treatment ceased on day 10, ERG amplitudes continued to improve toward baseline until day 28, although they did not completely recover to baseline levels (Figs. 6B, 6C). 

Figure 6.

 

K-9 improves ERG functional outcomes in RRD/SRR model. (A) Experimental overview: the RRD/SRR model was established on day 0 (SRI of 3 µL 0.03% CMC). K-9 (60 mg/kg) or PBS was administered days 0 to 10 twice daily IP. Scotopic ERG was measured at baseline, as well as days 7, 11, 14, and 28. (B) Amplitudes of a- and b-waves normalized to baseline over time in the K-9 (red; n = 20) and PBS (blue; n = 21) groups at the stimulation intensity of 1.0 log cd sec/m². P values from linear mixed-effect model; data represented as mean ± SEM.

Discussion

In RRD, hypoxic and inflammatory signaling induce progressive damage to the retinal ONL.^2–6 Consequently, even with successful surgical reattachment, RRD often results in photoreceptor death and lasting functional deficits.⁷⁻⁹ An interim therapy to protect photoreceptors prior to retinal reattachment may improve visual outcomes. In the present study, K-9 and NRTIs demonstrate anti-inflammatory and neuroprotective effects that may be beneficial in the setting of clinical RRD. 

The introduction of a mouse model of RRD created by subretinal injection of a viscous substance³⁴ has enabled the study of genetic and pharmacologic interventions in a tractable platform. Our studies reveal that experimental RRD induces inflammasome activation in mice, akin to inflammasome activation in RRD patient samples,⁶ and that three inflammasome inhibitors (K-9, AZT, and 3TC) reduce TUNEL-positive photoreceptor death, a feature that occurs early in the course of retinal detachment in both humans² and mice.³⁴ These data are consistent with the report that Nlrp3^–/– mice display reduced photoreceptor death after RRD⁶ and comport with findings that AZT preserves retinal function in a model of retinal ischemia-reperfusion injury.³⁵ 

In a mouse model of photo-oxidative damage-induced retinal degeneration, however, Nlrp3^–/– mice exhibited partial neuroprotection, whereas mice deficient in caspase-1 and caspase-11 displayed greater photoreceptor survival.³⁶ These differences could be due to mechanistic divergences across animal models, but it is also possible that multiple inflammasomes could be involved in photoreceptor demise. Indeed, we demonstrated the existence of a dual NLRC4–NLRP3 inflammasome that is critical in models of retinal pigmented epithelium degeneration that are responsive to K-9 treatment.³⁷ 

Since existing mouse models of RRD result in retinal detachment that persists for several weeks or longer,³⁴ we developed a novel RRD/SRR model in mice. By injecting a lower than typical concentration of CMC, which resembles the gelatinous texture and concentration of human vitreous humor,³⁸ we achieved a model in which spontaneous retinal reattachment occurs over 5 to 10 days of RRD, which approximates the time course of RRD in the standard-of-care clinical setting.^39,40 By self-reattaching, our model avoids reattachment procedures that would be technically complex and likely trauma inducing in the mouse eye. More important, reattachment of the retina facilitates electrophysiologic evaluation and clinically relevant assessments of functional outcomes in RRD. 

In this RRD/SRR model, we found that K-9, when administered during the period of retinal detachment, reduced the loss of retinal electrical function during detachment and led to an improvement in ERG parameters following retinal reattachment. These data provide good retinal structure–function correlation and suggest that K-9 has the potential to impart improved visual outcomes in the setting of RRD. 

Multiple real-world factors contribute to delayed surgical retinal reattachment, including patient unawareness, difficulty of retinal subspecialty consultation, operating room availability, pandemic-related delays, sociodemographic factors, constraints on physicians’ schedules, and the need to stabilize patient health preoperatively (e.g., blood pressure, glucose levels, and coagulation status).^41–44 In these contexts and more, K-9, which retains the inflammasome inhibitory activity of the parent NRTI class but lacks the off-target mitochondrial toxicity that plagues NRTIs,^11,20 may provide patients with interim, vision-saving neuroprotection. Our findings provide a rationale for testing this hypothesis via clinical trials. 

Acknowledgments

The authors thank R. Hakem and A. Hakem for the kind gift of the anti–caspase-8 antibody. 

Supported by the UVA Strategic Investment Fund, National Institutes of Health (NIH) grants (R01EY028027, R01EY29799, R01EY31039, R01AG078892), DuPont Guerry, III, Professorship, and a gift from Mr. and Mrs. Eli W. Tullis (JA); NIH grants (R01EY028027, R01EY031039, R01EY032512), BrightFocus Foundation, and the Owens Family Foundation (BDG); NIH grant R01DA054992 and the South Carolina Center for Rural and Primary Healthcare (JM); and NIH grant R35GM119751 (CIS). 

Disclosure: P. Huang, (P); C.C. Thomas, None; K. Ambati, University of Virginia (P), University of Kentucky (P); R. Dholkawala, None; A. Nagasaka, None; P. Yerramothu, None; S. Narendran, None; F. Pereira, (P); Y. Nagasaka, None; I. Apicella, None; X. Cai, None; R.D. Makin, None; J. Magagnoli, None; C.I. Stains, None; R. Yin, None; S. Wang, None; B.D. Gelfand, DiceRx (I), University of Virginia (P), University of Kentucky (P); J. Ambati, iVeena Delivery Systems (I, S), Inflammasome Therapeutics (I, S), DiceRx (I), Abbvie (C), Allergan (C), Boehringer-Ingelheim (C), Retinal Solutions (C), and Saksin LifeSciences (C), University of Virginia (P), University of Kentucky (P)