The RLR intrinsic antiviral system is expressed in neural retina and restricts lentiviral transduction of human Mueller cells

Abstract

The retinoic acid-inducible gene I (RIG)-I-like receptor (RLR) family of RNA sensor proteins plays a key role in the innate immune response to viral nucleic acids, including viral gene delivery vectors, but little is known about the expression of RLR proteins in the retina. The purpose of this study was to characterize cell-specific expression patterns of RLR proteins in non-human primate (NHP) neural retina tissue and to examine if RLR pathway signaling restricts viral gene delivery transduction. Since RLR protein signaling converges at the mitochondrial antiviral signaling protein (MAVS), experiments were performed to determine if knockdown of MAVS affected FIVGFP transduction efficiency in the human Mueller cell line MIO-M1. Immunoblotting confirmed expression of RIG-I, melanoma differentiation-associated protein 5 (MDA5), laboratory of genetics and physiology 2 (LGP2), and MAVS proteins in MIO-M1 cells and NHP retina tissue. Double label immunofluorescence (IF) studies revealed RIG-I, LGP2, and MAVS were expressed in Mueller microglial cells in the NHP retina. In addition, LGP2 and MDA5 proteins were detected in cone and retinal ganglion cells (RGC). MDA5 was also present in a subset of calretinin positive amacrine cells, and in nuclei within the inner nuclear layer (INL). Knockdown of MAVS significantly increased the transduction efficiency of the lentiviral vector FIVGFP in MIO-M1 cells, compared to control cells. FIVGFP or AAVGFP challenge did not alter expression of the LGP2, MAVS, MDA5 or RIG-I genes in MIO-M1 cells or NHP retina tissue compared to media treated controls. Our data demonstrate that innate immune response proteins involved in viral RNA sensing, including MDA5, RIG-I, LGP2, and MAVS, are expressed in several cell types within the NHP neural retina. In addition, the MAVS protein restricts non-human lentiviral transduction efficiency in MIO-M1 cells.

1.0. Introduction:

The eye is an ideal target for gene therapy due to its accessibility, immune privilege, and ease of examination (Cheng and Punzo, 2022; Wasnik and Thool, 2022). In the human genome at least 381 genes are known to cause inherited retinal diseases (IRD) including retinitis pigmentosa, Usher syndrome, Leber’s congenital amaurosis (LCA) and retinoschisis (Chan et al., 2021; Chien and Huang, 2022). A variety of approaches are now being employed to combat IRDs including viral gene therapy (Cheng and Punzo, 2022), nonviral delivery of nucleic acids (Chien and Huang, 2022; Gonzalez-Rioja et al., 2023), RNA-base therapies (Girach et al., 2022), gene editing (Altay et al., 2022), stem cell therapy (Voisin et al., 2023), and optogenetic therapy (Sakai et al., 2022). The first FDA approved ocular gene therapy, the Luxterna AAV-RPE65 vector (Askou et al., 2021), has demonstrated sustained clinical results for up to 7.5 years (Leroy et al., 2022). A number of clinical trials to treat IRDs are currently in progress and may result in the approval of more ocular gene therapies (Cheng and Punzo, 2022; Wasnik and Thool, 2022)

The most commonly utilized viral gene delivery vectors are the ssDNA virus Adeno-associated virus (AAV) and the ssRNA lentivirus (LV). Although AAV vectors are more commonly used due to their lack of replication in target cells and decreased risk of insertional mutagenesis, their transgene capacity is limited to 5 kb (Hori et al., 2019; Wasnik and Thool, 2022). The development of concatemerized “dual” AAV vectors, with increased capacity, have shown promising results in animal studies of IRD but have not yet been utilized in human clinical trials (Ferla et al., 2023; Trapani et al., 2014). LVs can deliver transgenes as large as 8 kb and obtain sustained gene expression (Wasnik and Thool, 2022). The use of pseudotyped non-primate lentiviral vectors such as feline immunodeficiency virus (FIV), which do not infect or induce disease in humans, provides a means to develop biologically safe lentiviral gene delivery vectors (Cavalieri et al., 2018; Saenz et al., 2012).

Regardless of the choice of viral vector, evidence is mounting that innate and adaptive host immune responses may limit the safety and/or efficacy of gene therapy (Annoni et al., 2019; Baldrick et al., 2023; Bucher et al., 2021; Coroadinha, 2023; Wiley et al., 2023). The innate immune response plays a key role in anti-viral responses to envelope, nucleic acid, and protein components of viral gene delivery vectors (Annoni et al., 2019; Bucher et al., 2021). One key component of the innate immune response, intrinsic host cell restriction factors (RF), can interfere with vector trafficking and gene expression, thereby limiting transduction of targeted cells (Coroadinha, 2023; Sauter and Kirchhoff, 2021; Sauter and Brandt, 2021). Recognition of a viral gene delivery vector by pattern recognition receptors (PRRs) can also result in unwanted and potentially dangerous ocular inflammation (Annoni et al., 2019; Baldrick et al., 2023; Bucher et al., 2021; Wiley et al., 2023).

PRRs which recognize viral nucleic acids can be divided into two groups based on their subcellular localization (Coroadinha, 2023). The endosomal members of the Toll-like receptor (TLR) family, namely TLR3, TLR7, TLR8, and TLR9, detect several forms of pathogen-associated nucleic acids (Koepke et al., 2021). The cytosolic nucleic acid receptors include the DNA receptors cyclic GMP-AMP synthase (cGAS), interferon gamma inducible protein 16 (IFI16), absent in melanoma 2 (AIM2), DEAD-box helicase 41 (DDX41), and the RNA sensor RLR family (Thoresen et al., 2021). The RLR family consists of the RIG-I, MDA5, and LGP2 proteins (Rehwinkel and Gack, 2020). RIG-I recognizes short-stranded viral RNA, while LGP2 assists MDA5 in the recognition of long-stranded RNA (Chen et al., 2021). The central DExD/H-box helicase and carboxy-terminal domains of RIG-I and MDA5 recognize 5’ tri-phosphate groups on viral RNA (Chen et al., 2021; Rehwinkel and Gack, 2020). The N-terminal caspase activation and recruitment domain (CARD) domains of RIG-I and MDA5 then interact with the CARD domain of MAVS (Ren et al., 2020). This interaction activates downstream nuclear factor kappa B (NF-κB) and interferon regulatory factor 3 or 7 (IRF3/7) pathways leading to the production of type I interferon (IFN) and pro-inflammatory cytokines (Chen et al., 2021; Ren et al., 2020; Thoresen et al., 2021).

Increasing the transduction efficiency of viral vectors, and limiting inflammation, is especially important to the success of ocular gene delivery due to the small volume of vector that can be administered (Aktas et al., 2018). The serotype of AAV vector, and delivery route to the retina, may be optimized to limit vector-induced inflammation (Gehrke et al., 2022; Liu et al., 2020; Wiley et al., 2023). In addition, engineered AAV vectors which incorporate oligonucleotides to antagonize innate immune responses, may reduce ocular inflammation (Chan et al., 2021). LV vector design, including capsid modifications (He et al., 2017) and purification methods, the blocking of complement activation and/or cytokine stimulation, and immunosuppression are also being explored as tools to modulate the host innate immune responses to LV vectors (Annoni et al., 2019).

A previous study indicated that the non-human derived lentiviral vector FIVGFP transduced the human Mueller cell line MIO-M1 at a very low efficiency (Sauter and Brandt, 2021). Knockdown of the restriction factors TRIM5α or TRIM11, however, significantly increased transduction efficiency in this cell type (Sauter and Brandt, 2021). In this study we examined the expression of RLR proteins in MIO-M1 cells and NHP neural retina tissue. By double label IF analysis we determined the retinal cell types expressing these restriction factors. Knockdown of the MAVS protein demonstrated that FIVGFP transduction was restricted by MAVS signaling in MIO-M1 cells.

2.0. Methods:

2.1. Macaque retina tissue

Eyes from euthanized rhesus macaques (Macaca mulatta) were obtained as they became available from the Wisconsin National Primate Research Center of the University of Wisconsin-Madison. Animals were free of infectious agents at the time of sacrifice and no animals were deliberately sacrificed for these studies. Macaque eyes were kept on ice and dissected within one hour of sacrifice. Posterior eye cups were incubated with phosphate-buffered saline (PBS)/1 mM EDTA for 30 minutes at 37°C to loosen neural retina tissue and separate the retina from RPE cells. Neural retina tissue was rinsed in PBS before proceeding with further studies. All experiments adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

2.2. Cell Culture

The human Mueller cell line MIO-M1 was obtained from the laboratory of Astrid Limb (Limb et al., 2002). Cells were maintained in DMEM/F12 (Cellgro, Manassas, VA, 10–090-CV)/10% fetal bovine serum (FBS) supplemented with a 1:100 dilution of L-glutamine-penicillin-streptomycin (SIGMA, St. Louis, MO, G1146) at 37°C in 5% CO2. Cells were utilized at a low passage number (≤ 5) for all experiments.

2.3. shRNA knockdown

MIO-M1 cells were transduced with MAVS shRNA lentiviral particles (Santa Cruz Biotechnology (SCBT), sc-75755-V), which contain a pool of 3 target-specific constructs, or control shRNA lentiviral particles (SCBT, sc-108080), which encode a scrambled shRNA sequence that will not lead to the specific degradation of any known cellular mRNA, at an MOI of 2 in the presence of 5 μg/ml Polybrene (SCBT, sc-134220). Cells were maintained in DMEM/F12/10% FBS with the addition of Primocin (InvivoGen, San Diego, CA, ant-pm-1). Puromycin (SCBT, sc-108071A, 5 μg/ml) was added to the culture media at 3 days post transduction to select for transduced cells. Puromycin resistant cells were cloned by limiting dilution and screened for protein knockdown (KD) by immunoblotting with a MAVS antibody (see Table 1).

Table 1:

Primary Antibodies utilized for Immunoblots and Immunofluorescence

Target	Species	Company	Catalog #	Immunoblot dilution	IF dilution
MDA5	rabbit	Invitrogen	700360	200	100
RIG-I	rabbit	Invitrogen	PA5-111253	250	100
LGP2	rabbit	Invitrogen	PA5-110841	500	100
MAVS	mouse	SCBT	sc-166583	1000	1000
Arrestin-C	goat	SCBT	sc-54355	NA	400
Calretinin	mouse	SCBT	365956	NA	50
Vimentin	rabbit	SCBT	5565	NA	200
Vimentin	mouse	SCBT	373717	NA	200
RBPMS	mouse	Novus	3905	NA	50
β-Actin	mouse	SCBT	sc-47778	2000	NA
TurboGFP	rabbit	Evrogen	AB513	NA	1000

NA= not applicable

2.4. Immunoblotting

Lysates were prepared from macaque neural retina tissue immediately post-sacrifice (Gerhardinger et al., 2001). MIO-M1 cell lysates were prepared in Laemmli’s sample buffer (BIORAD, Hercules, CA, 1610737) with a 1:100 dilution of protease inhibitor cocktail (SIGMA P8340). Protein concentrations were determined by Pierce 660 nm assay (Thermo Scientific, Rockford, IL, 2260). Lysates were electrophoresed on 4–15% Mini-PROTEAN TGX precast gels (BIORAD, 4568084S) with BenchMark Pre-stained Protein Standard (Invitrogen, Carlsbad, CA 10748–010). Proteins were electrophoretically transferred to nitrocellulose prior to blocking with 5% non-fat dry milk in Genius Buffer I (100mM maleic acid, 150mM NaCl, pH 7.5) containing 0.3% v/v Tween-20. LI-COR blots were blocked in 3% bovine serum albumin (BSA)/tris-buffered saline (TBS). The primary antibodies were diluted, as indicated in Table 1, and incubated overnight (ON) at 4°C. Horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5000), or IRdye-conjugated antibodies (1:10,000), were applied for 1 hour at room temperature (RT). The enhanced chemiluminescent (ECL) blots were developed using WesternSure PREMIUM chemiluminescent Substrate (LI-COR, Lincoln, NE, 926–95000). LI-COR blots were scanned on an Odyssey DLx imaging system and quantitation of MAVS and β-actin was performed with Image Studio software. The mean MAVS/β-actin ratio for the Control LV clone was set at 100%.

The secondary antibodies used for immunoblotting included mouse anti-rabbit HRP (SCBT, sc-2357), goat anti-mouse HRP (Novus Biologicals, Centennial, CO, NBP1–75144,), and IRdye 800CW donkey anti-mouse IgG (LICOR, 926–32212).

2.5. Immunofluorescence

Neural retina tissue obtained from euthanized macaques was fixed in 10% neutral buffered formalin (Thermo Fisher, Kalamazoo, MI, 305–510) for 24 hours before paraffin embedding and sectioning. Tissue sections were de-paraffinized and antigen retrieval was performed via heat treatment with citrate buffer (10mM citric acid/ 0.05% Tween, pH 6; 92°C, 20 minutes). Slides were pretreated with TrueBlack Lipofuscin Autofluorescence Quencher (Biotium, 23007) following the manufacturers protocol. Sections were blocked in 10% FBS, 5% bovine serum albumin (BSA), and 1% fish gelatin in PBS for 1 hour at RT before incubation with the primary antibody diluted in 2.5% FBS/PBS ON at 4°C.

MIO-M1 cells were plated on poly-L-lysine (SIGMA, P4707) coated chamber slides (Millipore, PEZGS0896). Cells were fixed in 4% paraformaldehyde/PBS and permeabilized for 10 minutes in 0.1%TritonX-100/PBS. Slides were then blocked with 5% normal donkey serum (NDS) before incubation with the primary antibody diluted in 2.5% NDS/PBS for 1 hour at RT.

The secondary antibodies, diluted 1:400 in PBS containing 2.5% FBS or NDS, were applied for 1 hour at RT. The secondary antibodies included: donkey anti-rabbit Alexa Fluor 488 (Life Technologies, Carlsbad, CA A21206), goat anti-mouse Alexa Fluor 594 (Life Technologies, A11032), goat anti-rabbit Alexa Fluor 488 (Life Technologies, A11008), donkey anti-goat Alexa Fluor 594 (Invitrogen, A11058), goat anti-mouse Alexa Fluor 488 (Invitrogen, A3273), and goat anti-rabbit Alexa Fluor 594 (Invitrogen, A11037). Slides were stained with 1 μg/ml Hoechst (Invitrogen, 33342) to visualize nuclei before mounting with Immu-mount (Thermo Scientific, 9990402).

Retina sections were examined with an Andor Revolution XD Confocal Spinning Disk Imaging System, equipped with four solid state laser sources (405 nm, 488 nm, 561 nm and 640 nm), a Nikon Eclipse Ti inverted microscope and a Yokogawa CSU-X1 confocal spinning disk head. Confocal images were obtained with Nikon 20x Plan Fluor dry objective and Nikon 60x Plan Apo VC oil objective.

2.6. FIVGFP Transduction

Wild type MIO-M1, and MAVS KD clones were transduced ON at 37°C with FIVGFP [packaged as described previously (Aktas et al., 2018)] at a multiplicity of infection (MOI) of 40. The following day, cells were trypsinized and plated onto poly-L-lysine coated glass chamber slides. At three- or seven-days post transduction, cells were stained with an anti-TurboGFP antibody as described in IF methods. DAPI positive nuclei and GFP positive cells were counted in five random fields for each well and data was expressed as mean percent GFP positive cells. p values were determined by two-sample t-test. Sample size (N)=5.

2.7. Gene expression analysis

Triplicate wells of MIO-M1 cells were transduced ON at 37 °C with FIVGFP at an MOI of 40. AAVGFP serotype 2 was produced with the AAV Helper-Free system (Stratagene, #240071) including the pAAV-RC, pHelper, and pAAV-hrGFP (#240074) plasmids as described (Reid and Lipinski, 2018). Triplicate wells of MIO-M1 cells were transduced ON at 37 °C with AAVGFP at an MOI of 6, which is sufficient to produce GFP expression within 24 hours. Twenty-four hours post transduction, RNA was isolated with the RNeasy Mini Kit (Qiagen, #74104), quantitated on a Nanodrop spectrophotometer (Nanodrop Technologies, Wilmington, DE, #ND-1000), and cDNA was synthesized with the RT² First Strand Kit (Qiagen, 3304104).

Neural retina tissue from 21.7-, 27-, and 14.5-year-old female rhesus macaques were incubated ON at 37°C with 6.4 × 10⁸ transforming units (TU) of FIVGFP. Neural retina tissue from two 11-year-old female rhesus macaques were incubated with 2 × 10¹⁰ TU of AAVGFP for 4 hours at 37°C. Tissues were rinsed in PBS prior to homogenization in TRIzol reagent (Ambion/Life Technologies, Grand Island, NY, #15596–026). RNA isolation was performed following the TRIzol Reagent protocol. DNase digestion (Qiagen, Valencia, CA, RNase-Free DNase Set, #79254) was completed prior to RNA cleanup on RNeasy spin columns (Qiagen, RNeasy Mini Kit, #74104). RNA was eluted and quantitated, and cDNA synthesized as described above.

qPCR primers were designed to amplify human or rhesus macaque MDA5, RIG-I, LGP2, MAVS, and the house keeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Table 2). qPCR reactions were performed in duplicate with Bullseye EvaGreen qPCR MasterMix (MIDSCI, #BEQPCR-R) on an ABI Quantstudio 7 Flex thermacycler. The fold change gene expression was calculated using the ΔΔCT method from the Qiagen RT² qPCR Primer Assay Handbook. Standard deviation was determined in Excel. Traditionally, plus or minus fold change gene expression values less than two are not considered significant. The null hypothesis was that viral vector treatment did not affect gene expression, with a fold change at least as small as 0.5 and at least as large as 2. A test of equivalence was performed to estimate if fold-change gene expression was between < 2-fold and > 0.5-fold. p values were calculated by one-sample t-test for equivalence between a factor change of 0.5 and 2 with alpha set at 0.01.

Table 2:

Quantitative PCR primers for RLR gene analysis

Primer	Sequence (5’−3’)
GAPDH For	GTCAAGGCTGAGAACGGGAA
GAPDH Rev	AAATGAGCCCCAGCCTTCTC
hLGP2 For	GGACTTGCTGAAGAAGCTCAT
hLGP2 Rev	CGTGCTCCCTGTGATAGAAATC
hMAVS For	GATGGAGGATATAGGGCGTTTC
hMAVS Rev	CCTGATGGCTCTAGTTCTTGTC
hMDA5 For	GAGGAATCAGCACGAGGAATAA
hMDA5 Rev	TCATTGGTGACGAGACCATAAC
hRIG-I For	GGAAGCAGGCAAGTCTTACA
hRIG-I Rev	CCCACCCAAATCTCATCTTGA
mLGP2 For	GAGCCGGTACCTAGAACTTAAAC
mLGP2 Rev	CTTGGTCCATGAGCTTCTTC
mMAVS For	TTTGAGTCGGAGCTGAGTAAG
mMAVS Rev	AACTTCTGGTCGATGTGTGG
mMDA5 For	CAACCACAGTTCAGCCAAATC
mMDA5 Rev	CTCTTGCTGCCACATTCTCT
mRIG-I For	CTGCCAACAAACACACAGTAATC
mRIG-I Rev	CCAGGGACAGTTTCAAGAAGAG

h=human

m=macaque

3.0. Results:

3.1. Expression of RIG-I-like receptors

To determine whether RNA sensor proteins and the MAVS signaling protein were expressed in MIO-M1 cells and NHP neural retina, we performed immunoblotting on cell or retina lysates. Blotting with a rabbit monoclonal antibody against MDA5 detected a single band in both the MIO-M1 and NHP retina lysates (Fig 1 A). An antibody to RIG-I identified a single band in the MIO-M1 and NHP retina lysates, although the retina band was of a slightly lower MW (molecular weight) (Fig 1B). The LGP2 antibody identified a doublet and two lower bands in the MIO-M1 lysate, and a similar band pattern in the NHP retina lysate (Fig 1C). A monoclonal antibody to MAVS detected a single band in both MIO-M1 and NHP retina lysates of the predicted MW for aggregated MAVS (Fig 1D). These results indicate that the human Mueller cell line MIO-M1 and macaque neural retina tissue both express the RLR proteins MDA5, RIG-I, LGP2, and the MAVS protein.

Figure 1. RLR proteins are detected in NHP neural retina tissue and MIO-M1 cells by immunoblotting.

Immunoblotting of macaque neural retina tissue and MIO-M1 cell lysates was performed with antibodies to A) MDA5, B) RIG-I, C) LGP2, and D) MAVS. Table 1 lists primary antibodies utilized for blots. HRP-conjugated secondary antibodies were applied prior to ECL. Full length immunoblots are shown. Film was aligned with nitrocellulose membranes after development for placement of digitally generated MWM bands.

To explore cell type specific distribution of the RLR and MAVS proteins in NHP neural retina tissue, double label IF with retinal cell marker antibodies was performed on paraffin embedded retina tissue (Sauter et al., 2018). Neural retina tissue from a minimum of three rhesus macaques was stained with each antibody to verify results. A no primary antibody control demonstrated that some autofluorescence remained in the photoreceptor layer even after sections were treated with an autofluorescence eliminator reagent (Fig 2A–C). Staining of the retina sections with the RIG-I antibody identified a filamentous pattern in the ganglion cell layer (GCL) (Fig 2D). IF with the Mueller cell marker antibody vimentin (Fig 2E), identified this filamentous pattern of RIG-I staining as the end feet of Mueller cells (Fig 2F).

Figure 2. RIG-I expression in NHP neural retina tissue.

A) No primary antibody control, DAPI staining. B,C) No primary antibody control with arrows indicating green and red autofluorescence in PR layer. D) Arrows indicate RIG-I staining. E) Arrows indicate vimentin staining of Mueller cells. F) Merged image with arrows indicating localization of RIG-I to Mueller cells. Photoreceptors (PR). Outer nuclear layer (ONL). Inner nuclear layer (INL). Ganglion cell layer (GCL). Panels A-C scale bar= 50 μm. Panels D-F scale bar= 15 μm.

IF with a MDA5 antibody revealed specific staining in several layers of the macaque retina tissue sections. Bright, tear drop-shaped structures above the ONL expressed the MDA5 protein (Fig 3A). Staining with the cone-specific marker antibody arrestin-C, highlighted the entire cone cell structure (Fig 3B), and when this image was merged with MDA5 staining, it was evident that MDA5 was expressed in the inner segment of cones cells (Fig 3C). Large, round cells in the GCL were also stained by the MDA5 antibody (Fig 3D). An antibody to the ganglion cell marker protein RNA-binding protein with multiple splicing (RBPMS) identified these cells as RGCs (Fig 3E) and merging of images indicated that MDA5 was expressed in RGCs (Fig 3F). MDA5 staining was also evident in small, round cells in the INL (Fig 3G). Calretinin positive amacrine cells (Fig 3H) were shown to colocalize with some of these MDA5 positive cells (Fig 3I). We also detected MDA5 positive nuclei in the INL (Fig 3J), whose staining was confirmed by colocalization with DAPI (Fig 3K–L).

Figure 3: MDA5 expression in NHP neural retina tissue.

A) Arrows indicate MDA5 staining above ONL. B) Arrows indicate cone cells labeled with the arrestin-C antibody. C) Arrows in merged image indicate MDA5 staining in cone cells. D) Arrows indicate MDA5 staining in GCL. E) Arrows indicate RBPMS labeling of RGCs. F) Merged image with arrows indicating MDA5 staining in RGCs. G) Arrows indicate MDA5 staining in the INL. H) Arrows indicate calretinin positive amacrine cells. I) Merged image indicates MDA5 is expressed in some calretinin positive amacrine cells. J) Arrow indicates MDA5 staining in INL. K) DAPI staining of INL nuclei. L) Merged image showing MDA5 staining in INL nuclei. Scale bar=15 μm

An antibody to LGP2 revealed teardrop-shaped staining above the at the bottom of the PR layer (Fig 4A). Double label IF with the cone cell marker antibody arrestin-C (Fig 4B) identified this staining to be in the inner segment of cone cells (Fig 4C). LGP2 staining was also detected in a filamentous pattern in the GCL (Fig 4D). This staining co-localized with the Mueller cell marker protein vimentin (Fig 4E), identifying these LGP2 positive cells as Mueller cells (Fig 4F). Large, round cells, identified by staining with the RGC marker antibody RBPMS, were also shown to express the LGP2 protein (Fig 4 G–I).

Figure 4. LGP2 expression in NHP neural retina tissue.

A) Arrows indicate LGP2 staining in PR layer. B) Arrows indicate cone cells labeled with the arrestin-C antibody. C) Arrows in merged image indicate LGP2 staining in cone cells. D) Arrows indicate LGP2 staining in the lower GCL. E) Arrows indicate vimentin staining of Mueller cells F) Merged image with arrows indicating LGP2 expression in Mueller cells. G) Arrows indicate LGP2 staining in GCL. H) Arrows indicate RBPMS antibody staining of RGCs. I) Merged image with arrows indicating LGP2 staining in RGCs. Photoreceptors (PR). Outer nuclear layer (ONL). Inner nuclear layer (INL). Ganglion cell layer (GCL). Scale bar= 15 μm.

The MAVS protein was detected in a filamentous pattern in the retina which suggested Mueller cell localization (Fig 5A, B). Staining with the vimentin antibody identified Mueller cells in the retina (Fig 5C), and when these images were merged, colocalization of MAVS and vimentin staining was detected (Fig 5D).

Figure 5. MAVS expression in NHP neural retina tissue.

A,B) Arrows indicate filamentous MAVS staining. C) Arrows indicate Mueller cells labeled with vimentin antibody. D) Merged image with arrows indicating MAVS expression in Mueller cells. Photoreceptors (PR). Outer nuclear layer (ONL). Inner nuclear layer (INL). Ganglion cell layer (GCL). Panel A scale bar= 50 μm. Panels B-D scale bar= 15 μm.

3.2. MAVS Knockdown in MIO-M1 cells

All RLR receptors signal through the MAVS protein, therefore we decided to determine the effect of MAVS knockdown (KD) on lentiviral transduction of MIO-M1 cells. To knockdown MAVS in MIO-M1 cells we utilized MAVS shRNA lentiviral particles and, as a control for off target effects, control shRNA lentiviral particles. MIO-M1 cells were transduced with lentiviral particles and single cell puromycin resistant clones were produced by limiting dilution. Clones were screened by immunoblotting with the MAVS antibody, using β-actin as a loading control. A representative immunoblot demonstrates a decreased level of MAVS protein in two MAVS KD clones compared to a control LV clone (Fig 6A). Triplicate samples of the control LV clone, MAVS clone C4, and MAVS clone D12 were blotted and LI-COR analysis was utilized for quantitation. MAVS clones C4 and D12 produced 22.2% and 28.3 %, respectively, of MAVS protein compared to the Control LV clone (Fig 6B). We were unable to achieve viable clones with 100% knockout of the MAVS protein, suggesting that this protein is essential for MIO-M1 cell function.

Figure 6. Knockdown of MAVS in MIO-M1 cells.

A) Immunoblotting of a Control LV clone and two MAVS knockdown clones with the MAVS antibody and a β-actin antibody as a loading control. (see Table 1 for antibody information) B) Graph of the quantitation of three separate LI-COR immunoblots of the Control LV, MAVS C4 KD, and MAVS D12 KD clones with MAVS and β-actin antibodies. The mean MAVS/β-actin ratio of the Control LV clones was set at 100%. Error bars represent standard error of the mean (SEM). *p≤ 0.004. N=3.

3.3. MAVS knockdown increased FIVGFP transduction efficiency

To study the effect of MAVS knockout in MIO-M1 cells, we transduced a control LV clone and the MAVS KD clones C4 and D12 with FIVGFP. To determine the efficiency of transduction, slides were stained with an anti-GFP antibody at 3 and 7 days post transduction (Sauter and Brandt, 2021). Three identical experiments were performed and significant increases in FIVGFP transduction efficiency were detected for MAVS KD clones C4 and D12, at both day 3 and day 7 post transduction, compared to the control LV clone, in all experiments. Data from a representative experiment demonstrated that KD of the MAVS protein increased FIVGFP transduction efficiency, up to 18% on day 3 and 12% on day 7, over that of the control LV clone (Fig 7).

Figure 7: Knockdown of MAVS increased transduction efficiency of FIVGFP in MIO-M1 cells.

Control LV MIO-M1 cells and MAVS KD clones C4 and D12 were transduced overnight with FIVGFP. The mean percent GFP positive cells was determined at day 3 and day 7 post transduction by IF with an anti-GFP antibody (see Table 1) in 5 random fields for each cell type. Significance was calculated by two-sample t test (*p ≤ 0.03). N=5. Error bars represent SEM.

3.4. Does viral gene delivery vector transduction alter RLR gene expression?

To determine whether viral gdv transduction of MIO-M1 cells altered RLR gene expression, triplicate wells of MIO-M1 cells were challenged overnight with FIVGFP or AAVGFP. After RNA isolation and cDNA synthesis, qPCR was performed with primers to LGP2, MAVS, MDA5, RIG-I and GAPDH as a housekeeping gene control (Table 2). All fold change gene expression results were less than +/− 2-fold from the media treated control (Table 3). Traditionally, plus or minus fold change gene expression values less than two are not considered significant, and statistical analysis confirmed our results fell within this range (Table 3). Therefore, the LGP2, MAVS, MDA5, and RIG-I genes were not significantly activated or repressed by FIVGFP or AAVGFP challenge of MIO-M1 cells compared to media treated control cells.

Table 3:

Viral gene delivery vector challenge of MIO-M1 cells

Gene	FIVGFP* (+/− sd)	p value	AAVGFP* (+/− sd)	p value
LGP2	1.14 (+/− 0.17)	0.006	1.15 (+/− 0.17)	0.007
MAVS	0.86 (+/− 0.2)	0.005	1.09 (+/− 0.07)	<0.001
MDA5	0.95 (+/− 0.21)	0.007	1.18 (+/− 0.04)	<0.001
RIG-I	0.68 (+/− 0.18)	0.004	0.8 (+/− 0.08)	<0.001

^*Mean fold change gene expression compared to media control

sd= standard deviation, N=3

p values were calculated by one-sample t-test for equivalence between a factor change of 0.5 and 2

To investigate whether viral gdv challenge of NHP neural retina tissue altered RLR gene expression, tissue from female macaques was challenged with FIVGFP, AAVGFP, or media only as a control. After RNA isolation and cDNA synthesis, qPCR was performed with primers specific to rhesus macaque RLR genes (Table 2). All fold change gene expression values were less than +/− 2-fold from the media treated control retina, and statistical analysis confirmed the results were within this range (Table 4). Therefore, FIVGFP transduction of neural retina tissue from 3 macaques did not significantly alter expression of the LGP2, MAVS, MDA5 or RIG-I genes compared to the media treated control tissue (Table 4). In addition, AAVGFP transduction of NHP neural retina tissue from 2 macaques also did not significantly alter expression of the LGP2, MAVS, MDA5, or RIG-I genes compared to the media control (Table 4).

Table 4:

Viral gene delivery vector challenge of rhesus macaque neural retina tissue

FIVGFP challenge*
Gene	Macaque A	Macaque B	Macaque C	Mean (+/− sd)	p value
LGP2	1.02	1.04	1.55	1.2 (+/− 0.25)	0.02
MAVS	1.02	0.65	0.93	0.87 (+/− 0.16)	0.005
MDA5	1.06	1.72	0.84	1.56 (+/− 0.37)	0.05
RIG-I	0.74	1.22	1.62	1.19 (+/− 0.36)	0.04
AAVGFP challenge*
Gene	Macaque A	Macaque B	Mean (+/− sd)	p value
LGP2	0.84	1.09	0.97 (+/− 0.13)	0.04
MAVS	0.81	1.6	1.21 (+/− 0.4)	0.15
MDA5	0.85	0.28	0.57 (+/− 0.29)	0.08
RIG-I	0.81	0.4	0.61 (+/− 0.21)	0.06

^*Fold change gene expression compared to media treated retina

sd= standard deviation

p values were calculated by one-sample t-test for equivalence between a factor change of 0.5 and 2

4.0. Discussion:

The RLR family of RNA sensors plays a key role in the innate immune response to both RNA and DNA viruses. The mitochondrial MAVS protein acts as a switch in this signaling transduction cascade by receiving RLR signaling and activating IRF3/7 and NF-κB pathways leading to expression of type I IFN and proinflammatory cytokines (Ren et al., 2020). RIG-I has been demonstrated to be crucial for retina pigment epithelial cell IFN-generating RNA immune responses (Schustak et al., 2021). Studies have linked RLR proteins to the innate immune responses of hepatocytes to hepatitis A and D viruses (Colasanti et al., 2023) (Gillich et al., 2023), human Sertoli cells to Zika virus (Jiyarom et al., 2022), and intestinal epithelial cells to SARS-CoV-2 (Bojkova et al., 2023; Zhang et al., 2022). Neuronal cells, including microglia, also express RLRs (Peltier et al., 2010) and utilize proteins such as RIG-I to mediate innate immune responses to RNA viruses including Japanese encephalitis virus (JEV) (Jiang et al., 2014; Nazmi et al., 2011). Organoids containing neural-like cells also responded to foreign RNA through RLRs (Katayama et al., 2023). DNA viruses, such as Adenoviruses, and Herpes viruses including Epstein Barr, herpes simplex, varicella zoster, and Kaposi’s sarcoma-associated herpes virus also activate the RLR pathway, often through interaction with noncoding viral RNAs (Crill et al., 2015; Minamitani et al., 2011; Rasmussen et al., 2009; Samanta et al., 2006; West et al., 2014; Zhao and Karijolich, 2019). These data demonstrate that many cell types, including neuronal cells, can respond to virus infection via RLR sensors. Our data demonstrate that several cell types in the NHP neural retina, as well as the human Mueller cell line MIO-M1, express RLR proteins and are primed to participate in antiviral innate immunity.

Viral infection can upregulate expression of the RLR family of genes in diverse cell types as demonstrated by Sendai virus infection of HeLa cells (Takahashi et al., 2020), rhinovirus infection of human Sino nasal epithelial cells (Lee et al., 2023), La Crosse virus infection of primary neurons (Mukherjee et al., 2013), and JEV infection of microglia (Jiang et al., 2014). Upregulation of the RLR pathway genes also occurred during AAV production in HEK293 cells (Wang et al., 2023) and AAV vector subretinal injection in mice (Chandler et al., 2019). We did not detect a significant upregulation of the RIG-I, MDA5, LGP2, or MAVS genes following overnight transduction of MIO-M1 cells with FIVGFP or AAVGFP gdvs. It is possible that longer transduction times would demonstrate upregulated RLR genes, but since innate immune responses are designed to function early in viral infection, we would expect to see changes in gene expression at the time point we analyzed.

Innate immune responses to viral nuclei acids must be considered when designing viral vectors for ocular gene delivery (Coroadinha, 2023), especially considering that these responses can be the primary driver of neuroinflammation (Annoni et al., 2019; Baldrick et al., 2023; Bucher et al., 2021; Shirley et al., 2020; Yang et al., 2023). The introduction of viral antigens during gene therapy has the potential to add to an already heightened state of immunity triggering a proinflammatory state (Yang et al., 2023). Not only can nucleic acid sensing trigger inflammation, but viral vector transduction efficiency can be limited by host anti-viral responses (Annoni et al., 2019; Coroadinha, 2023; Ren et al., 2020; Shao et al., 2018). AAV gene delivery vectors contain a ssDNA genome, but dsRNA replication intermediates can stimulate intracellular RLR RNA sensors (Coroadinha, 2023). For example, AAV transduction of primary human hepatocytes and hepatocytes in a human chimeric mouse model activated the innate immune response by MDA5 sensing and MAVS signaling (Shao et al., 2018). In HEK293 cells, transduction of AAV was linked to transcriptome changes, including stimulation of the RLR pathway, which led to host cellular stress responses (Wang et al., 2023). Lentiviral vectors, which package two strands of genomic ssRNA in their capsids, can also trigger RLRs after uncoating of viral RNA in the cytoplasm (Coroadinha, 2023; Shirley et al., 2020). For example, RIG-I and MDA5 were reported to detect HIV-1 RNAs in macrophages (Coroadinha, 2023). In fact, control of HIV-1 infection in the absence of anti-retroviral therapy in elite controllers has been linked to innate recognition of viral mRNA by RIG-I (Martin-Gayo et al., 2022). In this study we determined that the RIG-I pathway signaling protein MAVS limited lentiviral gdv transduction in human Mueller cells.

The high doses of viral vectors needed for gene delivery increase the risk of nucleic acid or protein contaminants from producer cells, which can also exacerbate immune responses. Thus, strategies to increase vector efficiency, thereby lowering vector dose and reducing innate immune responses, are crucial to the success of ocular gene therapy. Previous studies from our lab have indicated that treatment with proteasome inhibitors can increase lentiviral transduction efficiency in human trabecular meshwork cells, monkey organ-cultured anterior segments, and the human Mueller cell line MIO-M1 (Aktas et al., 2018; Sauter and Brandt, 2021). AAV mediated gene therapy of primate retinal pigment epithelial cells and human retinal explants was enhanced by hydroxychloroquine pretreatment, though the mechanism of action is not known (Chandler et al., 2019). Transient dexamethasone treatment of mice, prior to LV mediated gene transfer, reduced the IFN mediated inflammatory response and increased hepatocyte transduction efficiency (Agudo et al., 2012). Adenovirus-induced innate immune responses were also significantly reduced in dexamethasone treated mice (Seregin et al., 2009). However, dexamethasone can cause steroid-induced glaucoma, a dangerous side effect of extended immunosuppression (Phulke et al., 2017). Our previous studies have demonstrated that lentiviral transduction efficiency was increased in human Mueller cells by knockdown of the intrinsic restriction factors TRIM5α or TRIM11 (Sauter and Brandt, 2021). Decreasing the innate immune sensing of viral nucleic acids could also allow for better vector transduction efficiency. For example, knockdown of MDA5 or MAVS in HeLa cells was demonstrated to increase transgene expression in AAV-transduced cells (Shao et al., 2018). Similarly, our findings demonstrated an increase in LV gdv transgene production following MAVS KD in a human Mueller cell line.

Our previous studies have mapped the expression of TLRs to various cell types in the NHP neural retina including Mueller, RGCs, amacrine, and bipolar cells (Sauter et al., 2018). This current study is the first known report of RNA sensor protein cellular localization in the NHP retina. Double label IF detected RLR proteins in Mueller, cone, ganglion, and amacrine cells within the NHP neural retina. RIG-I, LGP2, and MAVS were all expressed in retinal Mueller cells. Mueller cells are the predominant glial cell type in the retina and span the entire thickness of the retina, forming the retinal margins at the ILM and outer limiting membranes (OLM). Considering their location and abundance, it is likely that Mueller cells will encounter invading microbes before other retinal cell types. Though their primary job is maintenance of retinal health, Mueller cells can function as innate immune cells by producing antimicrobial molecules in response to microbial challenge (Singh et al., 2014), and initiate immune responses via TLR signaling (Kumar et al., 2013; Kumar and Shamsuddin, 2012). Our detection of RLR sensor proteins in NHP retinal Mueller cells suggests this cell type also participates in retinal innate immune responses to viral RNA. Knockdown of MAVS in the human Mueller cell line MIO-M1, increased FIVGFP transduction, suggesting that the RLR pathway interacts with gdv RNA.

The RLR proteins MDA5 and LGP2 were expressed in NHP retinal cone cells, predominantly in the inner segments of this cell type. Very few reports have examined the role of retinal photoreceptors in innate immunity. Photoreceptor proteins have been linked to microglial cell activation and retinal inflammation in a mouse model of retina degeneration (Kohno et al., 2013). A mouse cone photoreceptor cell line was found to constitutively express all known TLR proteins and initiate innate immune responses following bacterial challenge (Singh and Kumar, 2015). The presence of MDA5 and LGP2 in NHP retinal cone cells suggests that photoreceptors may play a role in retinal innate immune sensing of viral RNA.

We detected MDA5 and LGP2 protein expression in NHP RGCs. Our previous studies also detected several TLR proteins in NHP RGCs (Sauter et al., 2018), suggesting that this cell type plays an active role in innate immune responses to invading microbes. Little is known about the expression or role of nucleic acid receptors in RGC, but the cGAS DNA receptor has been linked to axon regeneration in RGCs (Wang et al., 2023). Our results confirm that RGCs express RLR proteins which suggests that this retinal cell type might participate in the innate immune response to viral RNA.

The convergence of RLR signaling on MAVS suggests that a RNA sensor protein and MAVS signaling protein would need to be in the same cell to induce IFN-stimulated genes. It is therefore curious that some retinal cell types expressed RLR sensor proteins but did not express MAVS. For example, we detected RIG-I, LGP2, and MAVS in NHP Mueller microglial cells but could not detect MDA5 in this cell type. It is possible that the steady state MDA5 protein expression was too low to detect by IF in NHP Mueller retinal cells, since immunoblotting of MIO-M1 cells detected the MDA5 protein. Also, NHP retinal cone and GCs expressed MDA5 and LGP2, and calretinin positive amacrine cells expressed MDA5, but MAVS was not detectable by IF in these cell types. It is also possible that viral RNA exposure is necessary to induce expression of the genes involved in RNA sensing in NHP neural retina tissue. However, we found no change in expression of these genes in NHP neural retina tissue transduced with LV or AAV vectors.

In summary, we have demonstrated for the first time that several cell types in NHP retina tissue express RLR RNA sensing proteins and the MAVS protein. This study suggests that the primate retina is primed to participate in the innate immune response to viral RNA. The early inflammatory response, following viral gene delivery administration to the eye, may be partially due to the interaction of vector RNA with these receptors. In addition, the RLR pathway may restrict gdv transduction in retinal cells, as demonstrated by an increase in FIVGFP transduction efficiency following knockdown of MAVS in MIO-M1 cells.

Highlights.

1. RLR proteins and MAVS are expressed in NHP retina tissue and a human Mueller cell line

2. NHP retina microglial, cone, ganglion, and amacrine cells express RLR proteins

3. MAVS knockdown increased lentiviral vector transduction efficiency of Mueller cells

4. Viral vector challenge did not alter RLR gene expression in Mueller cells or NHP retina tissue

Abbreviations

RLR

RIG-I-like receptor

RIG-I

retinoic acid-inducible gene I

NHP

non-human primate

MAVS

mitochondrial antiviral signaling protein

MDA5

melanoma differentiation-associated protein 5

LGP2

laboratory of genetics and physiology 2

IF

immunofluorescence

RGC

retinal ganglion cell

INL

inner nuclear layer

IRD

inherited retinal diseases

LCA

Leber’s congenital amaurosis

AAV

adeno-associated virus

LV

lentivirus

FIV

feline immunodeficiency virus

GFP

green fluorescent protein

RF

restriction factor

PRR

pattern recognition receptor

TLR

Toll-like receptor

cGAS

cyclic GMP-AMP synthase

IFI16

interferon gamma inducible protein 16

AIM2

absent in melanoma 2

DDX41

DEAD-box helicase 41

CARD

caspase activation and recruitment domain

NF-κB

nuclear factor kappa B

IRF

interferon regulatory factor

IFN

interferon

PBS

phosphate-buffered saline

FBS

fetal bovine serum

KD

knockdown

BSA

bovine serum albumin

TBS

tris-buffered saline

HRP

horseradish peroxidase

ON

overnight

RT

room temperature

ECL

enhanced chemiluminescence

NDS

normal donkey serum

MOI

multiplicity of infection

TU

transforming units

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

MW

molecular weight

PR

photoreceptor

ONL

outer nuclear layer

GCL

ganglion cell layer

RBPMS

RNA-binding protein with multiple splicing

GDV

gene delivery vector

OLM

outer limiting membrane

DAPI

4′,6-diamidino-2-phenylindole