Abstract
Neuroinflammation can positively influence axon regeneration following injury in the central nervous system. Inflammation promotes the release of neurotrophic molecules and stimulates intrinsic proregenerative molecular machinery in neurons, but the detailed mechanisms driving this effect are not fully understood. We evaluated how microRNAs are regulated in retinal neurons in response to intraocular inflammation to identify their potential role in axon regeneration. We found that miR-383-5p is downregulated in retinal ganglion cells in response to zymosan-induced intraocular inflammation. MiR-383-5p downregulation in neurons is sufficient to promote axon growth in vitro, and the intravitreal injection of a miR-383-5p inhibitor into the eye promotes axon regeneration following optic nerve crush. MiR-383-5p directly targets ciliary neurotrophic factor (CNTF) receptor components, and miR-383-5p inhibition sensitizes adult retinal neurons to the outgrowth-promoting effects of CNTF. Interestingly, we also demonstrate that CNTF treatment is sufficient to reduce miR-383-5p levels in neurons, constituting a positive-feedback module, whereby initial CNTF treatment reduces miR-383-5p levels, which then disinhibits CNTF receptor components to sensitize neurons to the ligand. Additionally, miR-383-5p inhibition derepresses the mitochondrial antioxidant protein peroxiredoxin-3 (PRDX3) which was required for the proregenerative effects associated with miR-383-5p loss-of-function in vitro. We have thus identified a positive-feedback mechanism that facilitates neuronal CNTF sensitivity in neurons and a new molecular signaling module that promotes inflammation-induced axon regeneration.
Significance Statement
Inflammation can both positively and negatively influence the neuronal response to injury. Identifying molecular signaling pathways that mimic proregenerative effects of inflammation while bypassing cytotoxic effects is important for our understanding of the precise functions of inflammation in central nervous system injury and repair. We demonstrate that miR-383-5p is suppressed in neurons in response to inflammatory stimuli and regulates members of the ciliary neurotrophic factor receptor complex, as well as the expression of an antioxidant protein to improve axon regeneration in an optic nerve crush model. These findings identify a new molecular signaling module that promotes axon regeneration and that may bypass detrimental effects of inflammation.
Introduction
Following injury to the adult mammalian central nervous system (CNS), spontaneous axon regeneration is limited through an age-dependent decrease in intrinsic regenerative capacity (Schwab and Bartholdi, 1996; Filbin, 2006; He and Jin, 2016; A. P. Tran et al., 2018). Regulating the intrinsic growth state of neurons through modulating transcription factors or tumor suppressors can promote axon regeneration in experimental models of CNS injury (Park et al., 2008; Smith et al., 2009; Venkatesh and Blackmore, 2017; Fawcett and Verhaagen, 2018; Y. Cheng et al., 2022). Interestingly, inflammation can also influence neuronal viability and regeneration following injury (Benowitz and Popovich, 2011). Lens injury or intravitreal application of ß/γ-crystallins or zymosan produces intraocular inflammation and activate macrophages, neutrophils, retinal astrocytes, and Müller cells (Leon et al., 2000; Müller et al., 2007; Fischer et al., 2008; Leibinger et al., 2009; Hauk et al., 2010; Kurimoto et al., 2013). The inflammation-dependent release of oncomodulin and interleukin-type cytokines—including ciliary neurotrophic factor (CNTF), leukemia inhibitor factor (LIF), and interleukin 6 (IL6)—promote the survival and regeneration of retinal ganglion cells (RGCs) following optic nerve crush (ONC; Müller et al., 2007; Leibinger et al., 2009; Yin et al., 2009; Kurimoto et al., 2013). IL6 family cytokines bind to a heteromeric receptor complex to activate GP130 (Ip et al., 1992; Hirano et al., 1994), leading to downstream signaling via MAPK/ERK (Boulton et al., 1994) and JAK/STAT (Bonni et al., 1993; Müller et al., 2009), which can promote neuronal survival and RGC axon regeneration (Leaver et al., 2006; Y. Hu et al., 2007; Müller et al., 2009). However, intraocular inflammation also enhances the intrinsic growth state of RGCs (Hauk et al., 2010), and inflammation likely facilitates axon regeneration via a combination of intrinsic and extrinsic factors.
Scavenging reactive oxygen species (ROS) and maintaining mitochondrial homeostasis within damaged axons are also known to promote regeneration by alleviating the energetic deficits of axotomy (Noro et al., 2015; B. Zhou et al., 2016; Cartoni et al., 2016; X-T. Cheng and Sheng, 2021). Peroxiredoxins are six highly abundant proteins (PRDX1–6) that function as potent scavengers of hydrogen peroxide and peroxynitrate in cells (Mitsumoto et al., 2001). PRDX3 is downregulated following axotomy in vitro and preventing this downregulation reduces post-lesion neuronal apoptosis (W. Hu et al., 2018). The successful regeneration of CNS axons likely requires coordination between prosurvival and proregenerative signaling networks.
Complex proregenerative gene programs can be regulated in neurons via the expression of microRNAs (miRNAs). MiRNAs are small noncoding RNAs that bind to the 3′-untranslated region (3′UTR) of target messenger RNA (mRNA), reducing target gene expression by mRNA deadenylation, degradation, or translational suppression (Ambros, 2004; Valinezhad Orang et al., 2014). Since miRNAs simultaneously target numerous mRNAs, they regulate complex polygenetic traits in the CNS, such as the neuronal stress response, and axonal injury and repair (Sun et al., 2018; Juźwik et al., 2019; Wang et al., 2020). Several miRNAs have been identified to function cell-autonomously to regulate neurite growth in cultured primary neurons (Vo et al., 2005; S. Zhou et al., 2012; Wu and Murashov, 2013; Hancock et al., 2014; F. Han et al., 2015; Lu et al., 2015; Nampoothiri and Rajanikant, 2019). Other miRNAs regulate axon regeneration in vivo including miR-133b in a zebrafish axotomy model, miR-26a in mammalian sensory neurons, and miR-135 in mammalian optic nerve injury (Yu et al., 2011; J-J. Jiang et al., 2015; van Battum et al., 2018).
The purpose of the current study was to identify miRNAs that coordinate the proregenerative effects of zymosan-induced intraocular inflammation. We identified miR-383-5p as an important downregulated miRNA that promotes intrinsic axon regeneration. We found that miR-383-5p is downregulated in response to CNTF, which derepresses CNTF receptor components further sensitizing neurons to this neurotrophin. Additionally, miR-383-5p downregulation disinhibits expression of the mitochondrial antioxidant protein PRDX3 to enable axon regeneration. These results define a new proregenerative signaling module that is activated by inflammation and identify miR-383-5p inhibition as a potential therapeutic strategy to simultaneously modulate CNTF sensitivity and prevent postaxotomy energetic failure without inflammation.
Materials and Methods
In vivo ONC and intravitreal eye injection
All animal procedures were approved by the Montreal Neurological Institute Animal Care and Use Committee, following the Canadian Council on Animal Care guidelines. For all studies involving animals, an equal number of both male and female mice were utilized. For ONC surgeries, adult male and female C57Bl/6 mice 2–3 months old were anesthetized with isoflurane/oxygen. The left optic nerve was exposed and crushed 0.5–1.0 mm from the optic disc with fine forceps (Dumont #5) for 10 s. Care was taken to avoid damaging the ophthalmic artery. An examination of the fundus was made after each surgery to verify the vascular integrity of the retina. The right eye and optic nerve were left intact.
To assess miRNA expression in RGCs, we injected zymosan A (12.5 μg/μl; Sigma-Aldrich) or PBS control in the vitreous chamber of the left eye using a custom-made glass microneedle (Wiretrol II capillary, Drummond Scientific) at the time of the ONC injury. Under general anesthesia, the sclera was exposed, and the tip of the needle was inserted into the superior ocular quadrant at a 45° angle through the sclera and retina into the vitreous space. At 3 d postinjury (DPI), mice were perfused transcardially with cold PBS (0.1 M), and the eyes were dissected and embedded in Tissue-Tek optimal cutting temperature (OCT) compound. Preparation of slides for laser capture microdissection (LCM), LCM procedure, and extraction of RNA from the LCM-isolated RGC layer were done as described previously (Juźwik et al., 2018).
For drug delivery, 2 μl of miRCURY LNA miRNA inhibitor of miR-383-5p (LNA-383) and nontargeting negative control (LNA-NT) were intravitreally injected into the left eyes of mice at 1.5 nmol/μl at the time of the crush injury (Day 0) and again at 7 DPI. To label regenerating axons, we intravitreally injected fluorescently conjugated cholera toxin beta (Ctβ) at 5 μg/μl (Invitrogen) 2–3 d prior to euthanasia. At 14 DPI, animals were perfused transcardially with cold PBS (0.1 M) followed by 4% paraformaldehyde (PFA) in PBS. For optic nerve regeneration assessment after LNA injection, optic nerves were dissected for further processing by whole-mount clearing, and dissected eyes were processed for immunohistochemistry. For all experiments, the lenses of dissected eyes were examined for any signs of lens injury occurring due to intravitreal injection. For miR-383-5p overexpression experiments, 2 μl of AAV2-GFP or AAV2-383-5p (Applied Biological Materials) was intravitreally injected into the left eyes of adult mice 14 d prior to ONC. Zymosan A was intravitreally injected at 12.5 μg/eye at the time of ONC as before, and animals were killed as above at 14 DPI. For regeneration assessment after AAV2 and zymosan combinatorial experiments, optic nerves were embedded in OCT and cryosectioned at 12 μm/section, followed by immunohistochemistry.
Whole-mount optic nerve clearing and axon regeneration quantification
For immunolabeling-enabled three–dimensional imaging of solvent-cleared organs, optic nerves were dissected, embedded in agar blocks (1%; Thermo Fisher Scientific), and cleared using increasing concentration of tetrahydrofuran (50, 80 and 100%; Sigma-Aldrich), followed by immersion in dibenzyl ether (Sigma-Aldrich). Transparent blocks were imaged with a LaVision light sheet microscope (4× objective) and further processed with the InspectorPro software. 3D whole mounts were analyzed with the ImageJ/FIJI software. Whole optic nerve images were straightened prior to the assessment of axonal regeneration. Ctβ-positive regenerating axons were quantified at 1,000, 1,500, and 2,000 μm from the lesion site, throughout the thickness of the optic nerve using ImageJ/FIJI. Specifically, every 500 μm stacked optic nerve image was visualized in two separate windows with one person counting Ctβ-positive axons in the longitudinal plane and a second person counting Ctβ-positive axons in the transverse plane to correctly confirm that each longitudinal axon occurs in the transverse plane and vice versa. Both persons were blinded to the experimental condition during axon counts.
Retinal cross section, whole-mount staining, and optic nerve immunohistochemistry
For retinal cross sections, eye cups were dissected and cryoprotected in 30% sucrose at 4°C overnight. Eye cups were further embedded in Tissue-Tek OCT compound, and retinal cross sections of 12 μm thickness were cut using a Leica cryostat CM3050S. Retinal cross sections and optic nerve sections were blocked with 3% bovine serum albumin (BSA) in 0.3% Tween/PBS for 60 min and then probed with primary antibody overnight at 4°C. The following antibodies were used: anti-PRDX3 (Jeyaraju et al., 2006), anti-BRN3A (sc-31984, Santa Cruz Biotechnology), anti-LIFR (ab101228, Abcam), anti-CD130 for GP130 (ab202850, Abcam), and anti-βIII-tubulin (801202, BioLegend). Optic nerve cryosections were stained with anti-GAP43 (ab16053, Abcam). Secondary staining was conducted using Alexa Fluor 488- or 568-conjugated antibodies (Thermo Fisher Scientific) and Hoechst 33342 dye as a nuclear counterstain in 3% BSA in 0.3% Tween/PBS for 2 h at room temperature. Retinal cross section images were acquired using a Carl Zeiss Axio Imager M1 with Eclipse software (Empix Imaging). Fluorescence quantification was conducted by raters blinded to the experimental condition using the ImageJ software.
For retinal whole mounts, fixed eyes were dissected and washed three times with 0.5% Triton X-100 in PBS for 10 min. Whole-mount retinas were freeze-permeabilized in 2% Triton X-100/PBS for 15 min at −80°C, followed by blocking overnight at 4°C in 2% Triton X-100, 5% BSA/PBS. Primary antibody (anti-BRN3A) was incubated for 3 d at 4°C in blocking solution, followed by three washes with 0.5% Triton X-100/PBS as before. Secondary antibodies were incubated in blocking solution for 2 h at room temperature, followed by washing and mounting with Fluoromount-G mounting media. Whole-mount eyes were visualized at 20× using a Zeiss LSM 900 confocal microscope (Leica). For each retina, six images were taken, for which BRN3A-positive cells were quantified using the Analyze Particles plugin in ImageJ. Optic nerve sections were imaged using a 20× objective on a Zeiss LSM 900 confocal microscope (Leica). Optic nerve regeneration assessment of cryosections (2–4 sections/nerve) was conducted by a blinded rater quantifying the number of GAP43 axons present at 250 μm intervals from the ONC site using the ImageJ software. The number of regenerating fibers was normalized to the width of the nerve at each distance from the crush site assessed.
Quantitative RT-PCR (qRT-PCR)
Total RNA extraction was done using the miRNeasy Mini Kit (Qiagen), according to the manufacturer's instructions. With exception to Figure 4A, miRNA expression was assessed using multiplex qRT-PCR with Taqman miRNA Assays (Thermo Fisher Scientific) and snoRNA202 as the endogenous control, as previously described (C. S. Moore et al., 2013; Juźwik et al., 2018). For Figure 4A, qPCR was conducted using miRCURY LNA miRNA qPCR Assays (Qiagen) against miR-383-5p and snoRNA202 as the endogenous control, according to manufacturer’s instruction. For all qPCR experiments, fold change ratio (FCR) calculations for miRNA expression were performed using the 2−ΔΔCT method (Livak and Schmittgen, 2001).
Neuronal cultures
Cortical neurons from embryonic day 16 (E16) WT C57Bl/6 mice were prepared as previously described (Paré et al., 2018). Cortical neurons were seeded in poly-L lysine (PLL)-coated 96–well plates at 10,000 cells/well and cultured in Neurobasal (Invitrogen) supplemented with 2% B27, 1% N2, and 1% L-glutamine at 37°C, 5% CO2. Retinal neuron cultures from postnatal day 5–7 (P5–7) CD1 mice were dissociated using papain (20 U/ml) and cultured on PLL- and laminin-coated 48–well plates at 30,000 cells/well in Neurobasal supplemented with 2% B27, 1% N2, and 1% L-glutamine at 37°C, 5% CO2. Adult neuronal culture protocol was adapted from Müller et al. (2009). Briefly, adult female C57Bl/6 mice were killed by CO2 asphyxiation, and eyes were immediately enucleated and placed in cold L-15 medium (Invitrogen). Retinas were rapidly dissected and digested with papain (20 U/ml) for 30 min at 37°C, 5% CO2. Each retina was washed and resuspended with 5 ml of DMEM (Invitrogen) and pelleted at 800 RCF for 5 min at 22°C. The retinal pellet was resuspended in 5 ml of DMEM with 2% B27, 10 μM forskolin, and 50 ng/ml brain-derived neurotrophic factor. In 48-well PLL- and laminin-coated plates, 240 μl of cell suspension was plated per well. The number of βIII-tubulin-positive neurons in culture was examined between groups to ensure there were no differences in the cell density between replicates.
Neurons were fixed with 4% (PFA)/20% sucrose and stained with anti-βIII-tubulin antibody (801202, BioLegend), Hoechst 33342 dye (Sigma-Aldrich) nuclear counterstain, and Alexa Fluor 488-conjugated antibody (Thermo Fisher Scientific). Automated image acquisition was performed using ImageXpress. For embryonic cortical neurons and early postnatal retinal neurons, image analysis was conducted using the Neurite Outgrowth module of MetaXpress (Molecular Devices) to quantify neurite outgrowth normalized to the number of Hoechst/βIII-tubulin double-positive cells. Adult retinal neurons were traced manually by a blinded rater as previously described, using the Simple Neurite Tracer plugin on ImageJ (Müller et al., 2009; Arshadi et al., 2021).
For overexpression and loss-of-function assays in embryonic cortical and early postnatal retinal neurons, miRVana miRNA mimics miR-383-5p (M383), and Negative Control #1 (Thermo Fisher Scientific); miRCURY LNA miRNA inhibitor miR-383-5p and Negative Control A (Qiagen); and ON-TARGET plus SMART pool mouse Prdx3 or ON-TARGET plus nontargeting siRNA number 1 (Dharmacon) were used. Mimics were transfected at 20 nM final concentration, and LNA and siRNA were transfected at 15 nM. Neurons were cultured without antibiotics and transfected using Lipofectamine RNAiMax (Thermo Fisher Scientific), according to manufacturer's instructions. For mimic experiments, media were removed at 3 d in vitro (DIV), and cells were treated with 0.25% Trypsin-EDTA (Invitrogen) for 7 min at 37°C, 5% CO2, to retract all neuronal processes. Trypsin was quenched with DMEM and 10% FBS, and cells were spun at 1,000 RPM for 3 min at 22°C. Pelleted cells were resuspended in DMEM and 10% FBS and then reseeded in 96-well plates at 10,000 cells/well. For conditioned media outgrowth experiments, reseeded neurons were allowed to grow in supplemented mouse astrocyte-conditioned media (ACM; adapted from Bohlen et al., 2017) or control media for 24 h at 37°C, 5% CO2. Enzyme-linked immunosorbent assay (ELISA) for the quantification of CNTF was conducted using a colorimetric mouse CNTF sandwich-ELISA kit (imscntfkt, Innovative Research) according to manufacturer's instructions. For siRNA experiments, siRNA was transfected at 1 DIV; neurons reseeded at 3 DIV, transfected with LNA inhibitors 2 h postreseeding, and fixed 48 h later. For adult retinal neuron experiments, LNAs were added to resuspended cells at 25 nM immediately prior to plating and were maintained for 6 DIV prior to fixation, staining, and analysis.
Luciferase assays
HEK293 cells were seeded in 96-well plates and transfected with mmu-miR-383-5p mimic (Thermo Fisher Scientific) and 3′UTR dual-luciferase constructs (GeneCopoeia). Transfection was performed with Lipofectamine 2000 (Invitrogen) as per manufacturer's instructions. Cell lysate was collected 24 h post-transfection for analysis using the Luc-Pair Duo-Luciferase HS assay kit (GeneCopoeia), and luminescence was detected using an Enspire microplate reader (PerkinElmer). Measures of firefly luciferase activity were normalized to renilla luciferase. See Extended Data Table 5-1 for the 3′UTR sequences of putative miR-383-5p mRNA targets.
Western blot
MiRNA mimics or siRNA were transfected at 1 DIV and neurons were lysed 72 h later with HEPES-based RIPA buffer supplemented with protease inhibitors (complete protease inhibitor cocktail, Roche) and phosphatase inhibitors (10 mM sodium fluoride and 1 mM sodium orthovanadate). Protein lysates were quantified by DC protein assay (Bio-Rad Laboratories), boiled in a sample buffer, separated by SDS–PAGE, and transferred to PVDF membranes. Membranes were blocked with 5% skim milk powder in TBS-T for 1 h, probed with primary antibody at 4°C overnight, and probed with the appropriate HRP-conjugated secondary antibody for 1 h at room temperature. The following antibodies were used: antibodies were the same as those used for immunohistochemistry and anti-pyruvate dehydrogenase (PDH) E2/E3BP (ab110333, Abcam) and anti-α-tubulin (T9026, Sigma-Aldrich). HRP-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories or Pierce.
In silico assessment of predicated miR-383-5p targets
Putative or experimentally validated target genes for miR-383-5p were predicted by searching through the seven following databases: Diana-microT (Reczko et al., 2012; Paraskevopoulou et al., 2013), microRNA.org (Betel et al., 2008), miRDB (Wong and Wang, 2015), miRTarBase (Chou et al., 2016; Huang et al., 2022), RNA22 (Miranda et al., 2006), TargetScan (Agarwal et al., 2015), and TarBase (Vlachos et al., 2015), as well as the miRGate database (Andrés-León et al., 2015). Gene enrichment analysis on gene sets from in silico miRNA target predictions was conducted using the g:Profiler web interface (Raudvere et al., 2019) or STRING.db (Szklarczyk et al., 2023) where indicated.
Experimental design and statistical analysis
Detailed experimental design is described throughout the results section for each experiment, and sample sizes are indicated in figure legends. Statistical analyses were performed using GraphPad Prism 6. As indicated in figure legends, the following statistical tests were used: two-tailed student's t test; one-way analysis of variance (ANOVA); and two-way ANOVA. Post hoc statistical tests to elucidate the locus of effects include Dunnett's, Sidak's, Tukey's multiple-comparison tests, and Bonferroni-corrected paired t tests, where indicated. Normality of data was assessed for all statistical comparisons, and nonparametric tests were utilized when normality of data is violated. For repeated-measure ANOVA in which sphericity is violated, the Greenhouse–Geiser correction on the degrees of freedom was utilized. Significance was defined as *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001.
Results
MiR-383-5p is downregulated in response to zymosan-dependent intraocular inflammation
To identify miRNAs that may mediate inflammation-dependent regeneration, the effect of intravitreal zymosan injections on the expression of candidate miRNAs in RGCs subjected to ONC was assessed. To focus on miRNAs that may impact axon regeneration in multiple models of regeneration, we identified candidate miRNAs from a previous study identifying miRNAs regulated in the proregenerative dorsal root ganglion (DRG) conditioning lesion model (Neumann and Woolf, 1999; Strickland et al., 2011). Zymosan or PBS was intravitreally injected into the left eyes of C57Bl/6 mice at the time of ONC (Fig. 1A). At 3 DPI, animals were killed, and the ganglion cell layer (GCL) was isolated from retinal sections by LCM (Fig. 1B). Analyzing miRNA expression by qRT-PCR revealed that miR-383-5p was significantly downregulated in RGCs subjected to ONC with zymosan injection relative to ONC with PBS injection (t(4) = 2.561; p = 0.0387), whereas miR-18a-5p, miR-21-5p, and miR-223-3p were upregulated by ∼5-, ∼17-, and ∼20-fold, respectively (Fig. 1C). These miRNAs were regulated in the same direction in both RGCs subjected to ONC and proregenerative zymosan intraocular inflammation and the proregenerative DRG conditioning lesion model, while other miRNAs previously identified in DRGs were not regulated in the retina (Fig. 1C,D; Strickland et al., 2011). MiR-21-5p is upregulated in axotomized sensory neurons and promotes the growth of sensory neurons in culture, whereas miR-223-3p has been described for its neuroprotective properties in the experimental autoimmune encephalomyelitis animal model (Strickland et al., 2011; H-J. Li et al., 2018; Morquette et al., 2019). MiR-383-5p is known to regulate the neuronal response to oxidative stress. For example, the overexpression of miR-383 in the ARPE-19 cell line enhances apoptosis by reducing expression of the antioxidant protein PRDX3 (Y. Jiang et al., 2017). Throughout neuronal development, the intrinsic growth capacity of CNS projecting neurons is dramatically reduced, and other groups have identified proregenerative strategies by counteracting some of these developmental changes in gene expression (Rheaume et al., 2023; Xing et al., 2023; Lukomska et al., 2024). Indeed, the expression of miR-383-5p in the CNS increases with mouse development, inversely correlating with intrinsic regenerative potential (Rahmanian et al., 2019). However, the potential role of miR-383-5p inhibition in promoting axon regeneration has not yet been investigated in the literature. To fill this gap, we examined the potential for miR-383-5p inhibition to promote the regeneration of CNS projection neurons via the disinhibition of multiple regeneration promoting genes.
miR-383-5p is downregulated in proregenerative contexts. A, Schematic of experimental design depicting intravitreal injection of zymosan or PBS vehicle control at the time of ONC injury. B, Representative retinal cross sections stained with HistoGene for LCM (ROI, region of interest; GCL, retinal ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer); scale bar, 50 μm. C, qRT-PCR expression of candidate miRNAs in the GCL at 3 DPI (n = 3–4). FCR was calculated based on the 2−ΔΔCT method. D, qRT-PCR miRNA expression summary data for RGCs in response to zymosan intraocular inflammation. Upregulated or downregulated miRNAs in DRG neurons following a sciatic nerve lesion as reported in Strickland et al. (2011). Error bars represent standard errors of the mean (SEM). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.
MiR-383-5p inhibition is sufficient to promote axon growth and regeneration
We first investigated if downregulating miR-383-5p is sufficient to promote neurite outgrowth. Dissociated retinal neurons prepared from P5 to P7 mice were transfected with a locked nucleic acid targeting miR-383-5p (LNA-383) to sequester endogenous miR-383-5p or a nontargeting LNA control (LNA-NT) for 72 h. Neurons were then trypsinized and reseeded in new plates to restart axon outgrowth allowing for the assessment of neurite regeneration. LNA-383 transfection significantly increased neurite outgrowth in retinal neurons (t(6) = 2.687; p = 0.0362; Fig. 2A,B). We also asked if the effect of miR-383-5p inhibition on neurite growth is conserved in other types of neurons. Dissociated E16 mouse cortical neurons were treated with LNA-383, reseeded, and fixed after 4 DIV to assess neurite outgrowth. Treatment of cortical neurons with LNA-383 also promoted significant axon growth compared with that with LNA-NT (t(7) = 6.676; p = 0.0003), demonstrating that the effect of miR-383-5p inhibition on neurite regeneration in vitro is conserved between these neuronal subtypes (Fig. 2C,D).
MiR-383-5p inhibition promotes neurite outgrowth and optic nerve regeneration. A, C, βIII-tubulin-stained (Tuj1) mouse retinal (A) or cortical (C) neurons treated with LNA-383 or LNA-NT (scale bar, 100 μm). B, D, Quantification of normalized mean neurite outgrowth in retinal (B) or cortical (D) neurons treated with LNA-NT or LNA-383 (n = 4–5; Student's t test). E, Retinal cross sections demonstrating uptake of FAM-labeled LNA into RGCs. GCL, retinal ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer. Scale bar, 50 μm. Red box inlet is a zoomed region to demonstrate colocalization of BRN3A-positive RGCs and FAM-labeled LNA. F, Quantification of BRN3A-positive RGCs in retinal whole mounts 7 d following LNA-NT/383 left intravitreal eye injection, normalized to the uninjected contralateral eye (n = 3). G, Experimental design of in vivo LNA-383 regeneration assessment. LNA-NT or LNA-383 was intravitreally injected into the left eye (vitreous chamber) of adult mice at the time of ONC. An additional injection of LNA was conducted 7 d later. At 12 DPI, fluorescently conjugated Ctβ was intravitreally injected, and animals were killed at 14 DPI. H, Maximal projection of z-stacks for whole-mount imaged optic nerves at 14 DPI, cleared and regenerating fibers visualized by Ctβ. Arrowheads mark crush site (scale bar, 300 μm). I, Quantification of axon regeneration (n = 4–7; two-way ANOVA; Šídák's multiple-comparison test). Error bars represent SEM. *p < 0.05; **p < 0.01; ***p < 0.001.
To determine if miR-383-5p inhibition could mimic the growth-promoting effects of intraocular inflammation, we applied LNA-383 in the in vivo ONC model to assess its effects on regeneration. We conducted an intravitreal injection of fluorescently labeled LNA-383 (FAM) at 1.5 nmol/μl of uninjured mice and confirmed that LNA-383 is successfully taken up by BRN3A-positive RGCs (Fig. 2E). Furthermore, we verified that an intravitreal injection of unlabeled LNA-383 or LNA-NT did show any signs of toxicity with RGC survival remaining unaffected 7 DPI (Fig. 2F). To assess the impact of LNA-383 on optic nerve regeneration, we injected mice with LNA at the time of ONC and again 7 d later to maintain LNA levels throughout the 2 week duration of the experiment (Fig. 2G). The anterograde tracer Ctβ was intravitreally injected 2 d prior to sacrifice to label regenerating axons, and optic nerve regeneration was assessed at 14 DPI in whole-mount cleared optic nerves. Quantification of axons projecting past the lesion site revealed that the inhibition of miR-383-5p with LNA-383 significantly promotes axon regeneration (two-way ANOVA; F(1,17) = 5.011; p = 0.0336). This effect was driven by differences at short distances compared with LNA-NT with an average of ∼20 regenerating axon fibers at 1,000 μm past the injury site (t(27) = 3.376; p = 0.0067; Fig. 2H,I).
To determine if miR-383-5p downregulation is required for zymosan-dependent axon regeneration, we transduced RGCs with an AAV2 expressing either GFP or miR-383-5p 14 d prior to ONC and zymosan injection (Fig. 3A,B). Consistent with other investigations conducted with zymosan intraocular inflammation, we observed CD68/EB1-positive macrophages within the retina of zymosan-injected eyes 14 d after inflammation induction (Fig. 3C). We stained cryosections of optic nerves for GAP43 to avoid an additional Ctβ injection into an already inflamed eye. We examined axon regeneration and found that miR-383-5p overexpression significantly reduced the number of regenerating axons (F(1,25) = 35.37; p < 0.0001) up to 1,000 μm past the injury site (t(25) = 3.067; p = 0.0296; Fig. 3D,E). Together, these data demonstrate that miR-383-5p inhibition is sufficient to promote neurite outgrowth and regeneration and contributes to zymosan-dependent axon regeneration.
MiR-383-5p overexpression reduces zymosan-induced regeneration. A, Experimental design of in vivo miR-383-5p overexpression and zymosan regeneration assessment. AAV2-GFP or AAV2-383-5p was intravitreally injected into the left eyes of mice 14 d prior to zymosan (Z) injection and ONC. At 14 DPI animals were killed. B, The whole-mount retina expressing AAV2-GFP (scale bar, 500 μm). C, The evaluation of CD68-positive macrophages in whole-mount retinas of mice 2 weeks following PBS or zymosan injection. CD68-positive macrophages are observed in zymosan-injected eyes (white arrows). D, Representative images of regeneration assessment in optic nerve sections stained against GAP43 (scale bar, 250 μm). Red asterisk indicates the site of ONC. E, Quantification of regenerating fibers (n = 3–4; two-way ANOVA; Šídák's multiple-comparison test). Error bars represent SEM. *p < 0.05; **p < 0.01; ***p < 0.001.
Astrocytic CNTF secretions reduces neuronal miR-383-5p levels
To better understand how miR-383-5p expression could be regulated by zymosan, we considered how factors known to be released by cells in the context of proregenerative inflammation may impact its expression levels in neurons. Zymosan-induced regeneration is believed to be driven in part through astrocytic release of CNTF and LIF (Leaver et al., 2006; Müller et al., 2007; Leibinger et al., 2009). To examine if these factors affect the expression of miR-383-5p, we stimulated cortical neurons with ACM or with recombinant CNTF or LIF. Cortical neurons were used for these assays due to their ability to be produced in high yields for biochemical assays and due to the neuronal purity of cortical neuron cultures compared with mixed retinal cultures. The treatment of cortical neurons with ACM, which contains ∼400 pg/ml CNTF, significantly reduced the expression of miR-383-5p (t(6) = 4.749; p = 0.0032; Fig. 4A,B). To examine if CNTF or LIF alone were sufficient to regulate miR-383-5p levels, we treated neurons with dose curves of recombinant CNTF or LIF and found that CNTF but not LIF could significantly reduce miR-383-5p levels (F(4,10) = 4.438; p = 0.0255; Fig. 4C; F(5,12) = 2.258; p = 0.115; Fig. 4D). Therefore, astrocytic CNTF can lead to the reduction of miR-383-5p levels in neurons.
MiR-383-5p is regulated by CNTF. A, qRT-PCR expression of miR-383-5p in mouse cortical neurons treated with ACM (n = 4 per condition). B, ELISA demonstrating CNTF concentration in control and ACM (n = 2–3/condition). C, D, MiR-383-5p level in response to mouse cortical neuron treatment with increasing concentrations of CNTF or LIF (D) continuously until 8 DIV (n = 4; one-way ANOVA; Šídák's multiple-comparison test). FCR was calculated based on the 2−ΔΔCT method (n = 3; one-way ANOVA; Dunnet's multiple comparison). *p < 0.05; **p < 0.01.
MiR-383-5p targets CNTF receptor components and regulates CNTF sensitivity in neurons
We next sought to identify target genes of miR-383-5p that may impact its effect on axon regeneration. We conducted an in silico analysis of miR-383-5p targets by compiling a list of predicted miR-383-5p-binding mRNAs and filtering based on predictive score and cellular context. We utilized miRGate to identify high-probability miR-383-5p targets from multiple algorithms using a common reference genome (Andrés-León et al., 2015). We supplemented this list with experimentally validated targets from the miRTarBase database (Huang et al., 2022), as well as those identified in CLEAR-CLIP-derived mouse total brain miRNome-wide chimera RNAs (M. J. Moore et al., 2015). We then filtered these mRNA targets to include only those mRNAs identified in RNA sequencing datasets from fluorescence-activated cell sorting (FACS)-isolated RGCs and E16 mouse cortical neurons (Fig. 5A). A network of the final list of 1,295 proteins was highly interconnected (network not shown), with a significant PPI enrichment score (p = 10−16). We conducted statistical enrichment analysis for Reactome Pathways, identifying 64 enriched pathways (Fig. 5B). Our in silico analysis identified a number of miR-383-5p targets that have been described in the literature, including Prdx3 (KK-W. Li et al., 2013; Y. Jiang et al., 2017; X. Zhang et al., 2019), Wnt2 (Liu et al., 2021), and Bcl2 (Tao et al., 2021). Intriguingly, our in silico analysis revealed that the CNTF receptor components Lifr and Gp130 are predicted miR-383-5p targets in at least two databases (Fig. 5C, Extended Data Fig. 1-1A).
MiR-383-5p targets CNTF receptor components to sensitize neurons to CNTF. A, MiR-383-5p target prediction flowchart. Numerous databases for in silico miRNA target prediction were utilized. Validated targets were added to this list from miRTarBase as well as CLEAR-CLIP whole mouse brain miRNA–mRNA chimeras from M. J. Moore et al. (2015). MiRNA targets were then filtered to include those also expressed in FACS RGCs and E16 cortical neurons. B, Reactome analysis of curated list of miR-383-5p target genes. C, Databases that predict CNTF receptor components targeted by miR-383-5p (Extended Data Table 5-1). D, Quantification of luciferase activity in HEK293 cells cotransfected with M383 and luciferase vectors with 3′UTR of predicted targets (n = 3–4; two-way ANOVA; Šídák's multiple-comparison test). E–H, Retinal cross section presenting the RGC layer of mice intravitreally injected with LNA-NT or LNA-383 and stained with antibodies to LIFR (E) or GP130 (G; scale bar, 25 μm). Quantification of fluorescence intensity of LIFR (F) and GP130 (H) in BRN3A-positive cells (n = 4; two-tailed Student's t test). I, Representative images of adult retinal neuron outgrowth in response to transfection with LNA and culturing in 0 or 200 ng/ml CNTF for 6 DIV (scale bar, 50 μm). J, Quantification of CNTF-dependent neurite outgrowth from adult mouse retinal neurons treated with LNA-NT or LNA-383 and cultured in CNTF (n = 3; two-way ANOVA; Šídák's multiple-comparison test). K, Quantification of retinal neuron survival in culture with LNA-NT/383 and 0/200 ng/ml CNTF. Error bars represent SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.
Table 5-1
Excel file presenting miR-383 targets identified by in silico databases and PANTHER analysis. See figure 5 for more details (Sheet A) In silico miR-383-5p mRNA targets predicted using Diana- microT, microRNA.org, miRDPB, miR-TarBase, RNA22, TargetScan, and TarBase. CNTF receptor components analyzed (Lifr and gp130/Il6st) are highlighted in green. (Sheet B) MiR-383-5p targets predicted in silico by five or more of the seven databases. Experimentally validated miR-383-5p targets were extracted from the miRTarBase database searching for hsa-miR-383-5p. Overlapping genes between these two lists are highlighted in orange. (Sheet C) PANTHER Reactome analysis conducted for targets predicted by two or more of the seven databases to determine statistically overrepresented pathways. Download Table 5-1, XLSX file.
While it is likely that the proregenerative effect of miR-383-5p inhibition is distributed over numerous targets, we first explored the effects on the CNTF signaling axis because of the important potential implications for sensitizing neurons to proregenerative interventions. We conducted a luciferase assay to confirm that miR-383-5p targets the 3′UTR encoding Lifr and Gp130. HEK293 cells were cotransfected with a miR-383-5p mimic (M383) and a vector containing the luciferase gene flanked by the target 3′UTR or empty flanking control for 24 h. M383 significantly suppressed luciferase gene expression when regulated by the 3′UTR of Lifr (t(14) = 4.522; p = 0.0014) or Gp130 (t(14) = 4.960; p = 0.0006) confirming that these two genes are miR-383-5p targets (Fig. 5D). Furthermore, we demonstrated that LIFR and GP130 protein levels are elevated in RGCs transfected with LNA-383 compared with those with LNA-NT (Fig. 5E–H). Intravitreal injection of LNA-383 resulted in significant 28 and 32% increases in LIFR (t(6) = 2.544; p = 0.0438) and GP130 (t(6) = 2.949; p = 0.0256) protein expression, respectively, in BRN3A-positive RGCs by immunofluorescence analysis.
These findings raised the possibility that CNTF-dependent downregulation of miR-383-5p may elevate the expression of CNTF receptor components to further sensitize cells to the proregenerative activity of CNTF. To test this possibility, we transfected the dissociated adult retinal neurons with LNA-383 or LNA-NT and have grown them in the presence or absence of CNTF for 6 DIV. As in early postnatal mouse retinal neurons and E16 cortical neurons, LNA-383 promoted neurite outgrowth of adult dissociated retinal neurons in the absence of exogenous CNTF (t(657) = 3.658; p = 0.0016). The addition of 200 ng/ml CNTF further enhanced outgrowth in LNA-383-transfected adult retinal neurons (t(657) = 4.656; p < 0.0001; Fig. 5I,J). The outgrowth-promoting effects of LNA-383 and CNTF stimulation were independent of any effects on RGC survival in the cultures (Fig. 5K). Therefore, reductions in miR-383-5p that occur in response to zymosan intraocular inflammation or CNTF produce a baseline outgrowth phenotype and can further sensitize adult retinal neurons to the progrowth effects of CNTF.
MiR-383-5p inhibition derepresses PRDX3 to promote intrinsic axon regeneration
Our previous results demonstrated that miR-383-5p inhibition is sufficient for increasing optic nerve regeneration in vivo and that miR-383-5p regulates the sensitivity of retinal neurons to the proregenerative effects of CNTF. However, miR-383-5p inhibition also promoted neurite outgrowth in the absence of exogenous CNTF (Figs. 2A–D, 5I,J), suggesting that additional mechanisms of action are likely driving the neuron-intrinsic outgrowth phenotypes observed with LNA-383 treatment. Given that miRNAs function by simultaneously regulating multiple genes, we were interested in examining other miR-383-5p targets that might play a role in regeneration via different mechanisms. To identify high-probability miR-383-5p targets, we subanalyzed miR-383-5p targets predicted by five or more of the seven databases examined, leaving 41 genes. We cross-referenced this list of high-probability miR-383-5p targets with a list of experimentally validated targets from the miRTarBase database, revealing Prdx3 and Npat as two targets meeting both criteria (Fig. 6A, Extended Data Fig. 1-1B). We focused our attention on the antioxidant protein PRDX3, as this miR-383-5p target has been experimentally validated via luciferase assay by others (KK-W. Li et al., 2013). Furthermore, PRDX3 is expressed in the retina and optic nerve including in RGC cell bodies and axons, whereas Npat expression in RGCs is comparatively low (Fig. 6B; Chidlow et al., 2016; N. M. Tran et al., 2019). PRDX3 is a member of the peroxiredoxin family of proteins that primarily function to scavenge ROS and protect cells from apoptosis (L. Li and Yu, 2015). In an ARPE-19 retinal pigment epithelial cell line miR-383-5p downregulation enhances PRDX3 expression reducing high-glucose ROS production and apoptosis (Y. Jiang et al., 2017).
Prdx3 is a target of miR-383-5p and is necessary for LNA-383-mediated regeneration. A, Selection of Prdx3 as a prioritized target of miR-383-5p. Filtering of miR-383-5p targets was based on overlap of in silico predicted targets from five or more databases and miRTarBase experimentally validated targets for hsa-miR-383-5p (Extended Data Fig. 1-1B). B, Mean expression values and percentage of RGCs expressing transcripts for Prdx3 and Npat averaged across 46 RGC subtypes. Data were derived from the online RGC expression atlas provided by N. M. Tran et al. (2019). C, Quantification of luciferase activity in HEK293 cells cotransfected with M383 and luciferase vectors with 3′UTR of Prdx3 (n = 3; two-way ANOVA; Šídák's multiple-comparison test). D, E, Western blot and quantification for PRDX3 expression, with mitochondrial marker PDH (n = 4; two-tailed student's t test). F, Retinal cross section focused on the RGC layer of mice intravitreally injected with LNA-NT or LNA-383 and stained to assess PRDX3 expression (scale bar, 25 μm). G, Quantification of fluorescence intensity of PRDX3 in BRN3A-positive cells following injection with LNA-383 or LNA-383 (n = 4; two-tailed Student's t test). H, Western blot analysis of PRDX3 expression from mouse cortical neurons transfected with an siRNA targeting Prdx3 (siPrdx3) or negative control (siNC) for 72 h. I, Neuronal viability in MTT assay in mouse cortical neurons cotransfected with siPrdx and LNA-383 and treated with 10 μM sodium arsenite (n = 4; two-way ANOVA; Dunnett's multiple-comparison test). J, Schematic of experimental workflow for siRNA and LNA combinatorial experiments. K, βIII-tubulin-stained (Tuj1) cortical neurons that were transfected with siRNA 1 DIV, reseeded 72 h later to retract all neuronal processes and then transfected with LNA-NT or LNA-383 for 48 h (scale bar, 100 μm). L, Quantification of neurite outgrowth in cortical neurons treated with siRNA targeting Prdx3 and LNA-383 (n = 7; two-way ANOVA; p < 0.05; Šídák's multiple-comparison test). Error bars represent SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.
We conducted a luciferase assay to confirm that miR-383-5p targets the 3′UTR encoding PRDX3. M383 significantly suppressed luciferase gene expression when regulated by the 3′UTR of Prdx3 (t(8) = 96.720; p = 0.0003; Fig. 6C). Furthermore, transfection of cortical neurons with M383 produced a significant downregulation of PRDX3 protein expression as assessed by Western blotting (t(3) = 4.418; p = 0.0108; Fig. 6D,E). Expression of PRDX3 was also upregulated by ∼50% in RGCs following an intravitreal injection of LNA-383 (t(6) = 3.721; p = 0.0098; Fig. 6F,G). To determine if LNA-383 could protect cells from oxidative stress by derepressing PRDX3, we examined the response of cortical neurons, which have much higher transfection efficiency than retinal neurons, to arsenite-induced oxidative stress while manipulating miR-383-5p expression. Cortical neurons were transfected with siPrdx3 or a control siRNA to knockdown PRDX3 expression (Fig. 6H), and 72 h post-transfection neurons were transfected with LNA-NT or LNA-383 for an additional 2 d prior to treatment with sodium arsenite for 3 h. Sodium arsenite stress resulted in an ∼25% reduction in cell viability when assessed by MTT assay. Neurons expressing LNA-383 were resilient to arsenite stress, but this protection was abolished with PRDX3 knockdown (t(16) = 2.644; p = 0.0453; Fig. 6I). Together this supports a model whereby miR-383-5p suppression in vitro and in vivo derepresses Prdx3 resulting in enhanced PRDX3 protein expression and antioxidant function.
We reasoned that PRDX3 could also function to promote neuronal repair by preventing the accumulation of ROS and maintaining mitochondrial homeostasis within damaged axons, two interventions that are known to promote axon regeneration by alleviating the energetic deficits associated with axotomy (Noro et al., 2015; B. Zhou et al., 2016; Cartoni et al., 2016; X-T. Cheng and Sheng, 2021). To determine if the intrinsic proregenerative activity of miR-383-5p loss-of-function depends on its regulation of PRDX3, we performed combinatorial experiments in mouse cortical neurons with both LNA-383 and an siRNA directed to Prdx3. 72 h post-siPrdx3 transfection, neurons were trypsinized, reseeded and transfected with LNA-NT or LNA-383, and cultured for 48 h (Fig. 6J). LNA-383 and siNC cotransfection significantly promoted neurite outgrowth by ∼30% (t(24) = 4.247; p = 0.0017). However, siPrdx3-transfected neurons were insensitive to the outgrowth-promoting effects of LNA-383 (t(24) = 1.475; p = 0.6312; Fig. 6K,L). The intrinsic outgrowth-promoting effects of LNA-383 are thus dependent on PRDX3 expression.
Discussion
Inflammation can promote axon survival and regeneration in some contexts, but it can also have harmful effects. The identification of molecular signals that mimic the positive effects of inflammation and bypass detrimental effects is important to our basic understanding of CNS injury and repair. Moreover, miRNAs, which individually regulate the expression of numerous genes, are attractive molecular targets to investigate as potential therapeutic tool compounds as they modulate complex phenotypes in cells (Valinezhad Orang et al., 2014; Sun et al., 2018; Wang et al., 2020). Here, we demonstrate that proregenerative intraocular inflammation results in a reduction in the level of miR-383-5p in RGCs. An inhibitor against miR-383-5p is sufficient to promote modest axon growth in multiple types of neurons in vitro and promotes optic nerve regeneration in vivo. We identified two mechanisms, whereby miR-383-5p inhibition contributes to regeneration. Firstly, CNTF exposure to neurons in vitro reduces miR-383-5p levels in cells, and knockdown of miR-383-5p increases levels of CNTF receptor components, which sensitizes neurons to CNTF signaling to promote axon growth. Our data support a model, whereby inflammation-induced release of CNTF from astrocytes leads to decrease in miR-383-5p levels which further sensitizes neurons to CNTF. Secondly, intrinsic regeneration with miR-383-5p inhibition is associated with the disinhibited expression of the antioxidant protein PRDX3, which was required for regeneration in vitro. Thus, these data define a new molecular signaling module that promotes axon regeneration via the simultaneous regulation of the intrinsic growth state of neurons by modulating PRDX3 expression while facilitating sensitivity to outgrowth-promoting neurotrophins (Fig. 7). Inhibiting miR-383-5p thus recapitulates some of the proregenerative effects of zymosan intraocular inflammation while bypassing the detrimental effects of inflammation.
Schematic of miR-383-5p regulation in response to an inflammatory stimulus. Injection of zymosan at the time of ONC stimulates astrocytic release of CNTF. CNTF in turn downregulates miR-383-5p, releasing its break on CNTF receptor components LIFR and GP130 and the mitochondrial antioxidant PRDX3. Inhibition of miR-383-5p relieves multiple proregenerative pathway components to elicit a distributed axon regenerative response following injury.
The mechanism by which miRNAs are regulated in cells is a topic of extensive inquiry. The upregulation of miRNAs in response to injury, neurotrophin stimulation, or intraocular inflammation can be explained most simply by increases in transcription of the miRNA gene (Krol et al., 2010). However, once mature miRNAs are loaded into AGO, they are highly stable and can be maintained in cells for days despite transcriptional blockade (Winter and Diederichs, 2011). Therefore, the rapid downregulation of miRNAs is more likely to occur via a degradative process, such as target-directed miRNA decay (TDMD). TDMD is a process whereby a trigger RNA atypically binds to the target miRNA within the miRISC, leading to a conformational shift in this complex rendering it susceptible to proteolytic degradation (Sheu-Gruttadauria et al., 2019; J. Han et al., 2020). This process was originally described in the context of the HSUR1 viral RNA which facilitates host cell miR-27 degradation (Cazalla et al., 2010), but recent investigations have identified numerous endogenous TDMD substrates which function in human cancers and throughout development (Simeone et al., 2022; B. T. Jones et al., 2023). It is possible that the rapid reduction of miR-383-5p in response to CNTF stimulation occurs via TDMD simultaneous with an increase in the abundance of a CNTF-responsive trigger RNA. Although determining the precise coding or noncoding trigger RNA transcript(s) which may facilitate the degradation of miR-383-5p was beyond the scope of this study, the complex mechanism(s) by which miR-383-5p and other proregenerative or prosurvival miRNAs are regulated should be explored in future studies.
CNTF application through a variety of routes including viral delivery or application of recombinant protein has been shown to promote survival and regeneration in several animal models of retinal injury, including ONC and ischemic optic neuropathy (Fudalej et al., 2021). These findings have led to the initiation of numerous clinical trials to evaluate the efficacy of CNTF in ophthalmic diseases including glaucoma and ischemic optic neuropathy (Fudalej et al., 2021). The introduction of miR-383-5p inhibitors to RGCs through viral-mediated delivery could be considered as an alternative sustained mode of treatment in these cases. It is also worth considering that attempts to transplant primary RGCs or stem cell-derived RGCs could be facilitated through genetic manipulation of donor cells to suppress miR-383-5p to promote their regenerative potential (K. Y. Zhang et al., 2021). Furthermore, the identification of GP130 as a miR-383-5p target suggests that it could have a wider role in regulating cellular responses to IL6 family cytokines, which regulate a vast array of cellular functions (S. A. Jones and Jenkins, 2018). However, in all cases, it is important to note that while intravitreal injection is the preferred method of drug delivery for retinal diseases, they can also gain access to the systemic circulation or cause systemic inflammatory responses (Avery et al., 2014; del Amo et al., 2017; Ail et al., 2022). While it is unlikely that systemic effects were predominant in our study, such effects could not be ruled out without detailed pharmacokinetic evaluation.
We and others have shown that PRDX3 is an important downstream target of miR-383-5p, which is known to regulate neuronal responses to oxidative stress (KK-W. Li et al., 2013; Y. Jiang et al., 2017). PRDX3 is specifically localized to the mitochondrial matrix where it is an important enzyme in eliminating mitochondrial H2O2; in turn, PRDX3 is expressed most highly in mitochondria-rich areas, such as the retina and optic nerve head (De Simoni et al., 2008; Kil et al., 2012; Chidlow et al., 2016). In rodent models of hippocampal excitotoxicity, traumatic brain injury, and cerebral ischemia, PRDX3 overexpression protects against neuronal damage by limiting both oxidative stress and preserving mitochondrial function (Hattori et al., 2003; Hwang et al., 2010; Y. Jiang et al., 2017; W. Hu et al., 2018). Indeed, we found that LNA-383 led to elevation in PRDX3 in neurons that conferred resistance to arsenite oxidative stress. However, the role of PRDX3 specifically in neuronal regeneration has not been described. Manipulations to enhance axonal transport of mitochondria are known to promote regeneration of RGC axons; thus, it is possible that enhanced PRDX3 expression contributes to optic nerve regeneration by supporting healthy mitochondria in injured RGC axons (Cartoni et al., 2017). Furthermore, intravitreal injection of the free radical scavenger spermidine promotes RGC regeneration in an optic nerve injury model, thus supporting the idea that PRDX3 upregulation would contribute to enhanced regeneration (Noro et al., 2015). The finding that LNA-383's effects on axon growth in culture require the expression of PRDX3 implies that the expression of this target contributes to miR-383-5p outgrowth effects in neurons.
MiRNAs are powerful posttranscriptional regulators of gene expression due to the large number of mRNAs that they can individually regulate. For our molecular profiling, we focused on those targets predicted to regulate CNTF signaling—since this pathway is already implicated in intraocular inflammation-induced optic nerve regeneration—and a well-established miR-383-5p target highly expressed in RGCs (Prdx3). In silico prediction algorithms identify numerous additional potential miR-383-5p targets, and it is likely that the effects of miR-383-5p on axon regeneration are distributed over multiple pathways. An analysis of predicted targets through PANTHER Reactome revealed 80 diverse statistically overrepresented pathways, including some associated with axon growth, regeneration, and guidance. Both MAPK activation and the MAPK family signaling cascades may be enhanced by miR-383-5p knockdown. The integrity of signaling through the MAPK cascade is essential for axonal regeneration in numerous experimental models (Hammarlund et al., 2009; Itoh et al., 2009; Valakh et al., 2013). Indeed, multiple miR-383-5p targets within the MAPK signaling cascade pose essential roles in axon growth, including the cyclic AMP-responsive element–binding protein 1, the nuclear factor kappa B-1, and C-Jun, among others (Herdegen et al., 1997; Jiao et al., 2005; Haenold et al., 2014; Yao et al., 2019). The enrichment of these miR-383-5p targets suggests that other pathways may also contribute to the outgrowth-promoting effects of miR-383-5p inhibition warranting further study on its molecular mechanism of action.
Footnotes
A.E.F. is funded by the Canadian Institutes for Health Research and the Multiple Sclerosis Society of Canada (MSSOC). J.P. is funded by EpilepsieNL (WAR 18-05). C.A.J. was supported by a fellowship from the MSSOC and a Vanier Canada Graduate Scholarship (VCGS). M.A.H. and S.D. are supported by a VCGS. B.M. was funded by the MSSOC and Fonds de recherché du Quebec (FRQS). J.Z. was funded by a MSSOC Master’s studentship.
↵*M.A.H., C.A.J., and B.M. are co-contributing authors.
The authors declare no competing financial interests.
- Correspondence should be addressed to Alyson E. Fournier at alyson.fournier{at}mcgill.ca.
