Reduction of retinal ganglion cell death in mouse models of familial dysautonomia using AAV-mediated gene therapy and splicing modulators

Familial dysautonomia (FD) is a rare neurodevelopmental and neurodegenerative disease caused by a splicing mutation in the Elongator Acetyltransferase Complex Subunit 1 (ELP1) gene. The reduction in ELP1 mRNA and protein leads to the death of retinal ganglion cells (RGCs) and visual impairment in all FD patients. Currently patient symptoms are managed, but there is no treatment for the disease. We sought to test the hypothesis that restoring levels of Elp1 would thwart the death of RGCs in FD. To this end, we tested the effectiveness of two therapeutic strategies for rescuing RGCs. Here we provide proof-of-concept data that gene replacement therapy and small molecule splicing modifiers effectively reduce the death of RGCs in mouse models for FD and provide pre-clinical foundational data for translation to FD patients.

www.nature.com/scientificreports/we present pre-clinical data demonstrating the effectiveness of gene therapy and a novel splicing compound in preventing RGC death in FD mouse models.
To test whether a gene replacement strategy could prevent the death of RGCs in our previously established Pax6-cre;Elp1 loxp/loxp mouse model 21 , we generated two AAV2 vectors, one expressing the murine Elp1 gene (mElp1) and the other expressing eGFP, both driven by the U1a promoter (see Methods).We first determined that the AAV2-U1a-mElp1 could drive the expression of Elp1 in the mouse retina and increase Elp1 protein expression in vitro (Supplemental Fig. 1; Supplemental Fig. 3A-C).
Next, cohorts of Pax6cre;Elp1 loxp/loxp mice received intravitreal injections of AAV2-U1a-mElp1 or control (eGFP) vector at postnatal day 21 (P21) before any significant loss of RGCs is observed 21 .Three months following injection, RGCs were quantified by flat-mount analysis (Fig. 2A).This time point was selected since prior work established that, by this age, 40-60% of RGCs have died 21 .We observed a significant response with both the higher AAV2-U1a-mElp1 vector dose (× 10 9 vg/eye), showing a 137% increase in the average number of RGCs (5696 ± 252 (SEM), p < 0.0001) and the lower AAV2-U1a-mElp1 vector dose (× 10 8 vg/eye), showing a 131% increase in the average number of RGCs (5563 ± 172 (SEM), p < 0.0001) in an area of 4mm 2 in Pax6cre; Elp1 loxp/loxp retinae compared to untreated Pax6cre; Elp1 floxp/loxp retinae with an average number of 2404 ± 405 (SEM) RGCs 4mm 2 (Fig. 2B).Interestingly, eyes injected with the control AAV2-U1a-eGFP showed an 82% increase in the average number of RGCs (4386 ± 534 (SEM)) compared to the untreated Pax6cre; Elp1 loxp/loxp (Fig. 2B).However, the mElp1-virus significantly increased RGC numbers by 30% (p = 0.04) above the eGFP-injected eyes.This suggests that (1) injection of a virus in a dose-dependent manner alone induces a protective response in the retina, most likely by stimulating a wound response [22][23][24] , but (2) that the Elp1 virus increased RGCs significantly above the virus-alone-induced protective response.These data provide promising evidence that gene replacement therapy, with further optimization, could be used to ameliorate RGC death in FD.Based on these data, we generated an AAV2 vector driving the expression of the human ELP1 gene (hELP1).Again, we confirmed that the AAV2-U1a-hELP1 could increase ELP1 protein in vitro (Supplemental Fig. 2A,B; Supplemental Fig. 3D) and drive ELP1 mRNA expression in wild-type mouse retinas in vivo (Supplemental Fig. 2C).Subsequently, Pax6cre;Elp1 loxp/loxp and littermate controls were intravitreally injected with three different viral titers of the human ELP1 virus.All mice received bilateral injections at P21, and RGCs were quantified three months later (Fig. 3A).There was a significant increase in RGC survival in the Pax6cre;Elp1 loxp/loxp retinae that received the highest vector dose (5.4 × 10 8 vg/eye), showing a 39% increase in the average number of RGCs (4925 ± 439 (SEM), p = 0.02) (Fig. 3B).There was no significant difference between the uninjected retina and those injected with the AAV2-U1a-eGFP construct, which was used at a lower dose than in the mouse Elp1 virus experiments.To visualize the AAV2-U1a-hELP1 infected RGCs, we double immunolabeled retinae with antibodies to RGCs and human ELP1.In the retinae infected with AAV2-U1a-hELP1, approximately 26% of RGCs expressed the human ELP1 protein (Fig. 3C), visually confirming that driving ELP1 expression can efficiently produce human ELP1 protein.
Another validated approach for treating genetic disease is using orally administered small molecules that modify splicing 25,26 .To test the efficacy of a novel small splicing compound PTC680, an analog to the highly www.nature.com/scientificreports/specific ELP1 splicing modulator PTC258 16 , we generated a new hybrid mouse model in which we introduced the human FD ELP1 original founder splice mutation (TgFD9) transgene into the Pax6cre;Elp1 loxp/loxp conditional knockout mouse model.This new model recapitulates the FD retinal phenotype while recreating the tissuespecific mis-splicing observed in FD patients.To assess the therapeutic effect of PTC680, we administered an oral treatment formulated to dose each TgFD9;Pax6cre;Elp1 loxp/loxp mouse 1.2 mg/kg/day.Treatment started at three different time points, (a) Embryonic day 18.5 (E18.5),(b) P21, and (c) P90, to test the therapeutic efficacy throughout the lifespan and disease progression (Fig. 1B).In all cases, mice consumed either vehicle or PTC680 special chow for 3 months.RGCs were quantified at the end of each 3-month treatment regimen using retinal flat-mount analysis (Fig. 4A).PTC680 treatment increased RGC number by up to 57% depending on the age at which treatment began (Fig. 4B).The most significant increase in RGCs occurred when treatment began at the youngest age (E18.5).Notably, a significant rescue of RGCs occurred even when treatment started at 3 months old.Thus, daily consumption of PTC680 prevents significant RGC death in both mature and immature retinas.
To assess whether RGC rescue was associated with increased splicing efficiency of the mutant FD ELP1 gene, we analyzed ELP1 splicing in retinal tissue from vehicle-and PTC680-treated FD mice.As demonstrated with PTC258 16 , PTC680 significantly increases ELP1 exon 20 inclusion in the FD retina (Fig. 4C; Supplemental Figs. 4,  5, 6), providing evidence that this novel splicing modulator mitigates a primary FD phenotype by increasing human ELP1 in the retina.

Discussion
We demonstrate that both an intravitreally-delivered RGC-specific gene transfer and the systemic delivery of a novel small molecule splicing modulator can prevent the progressive death of RGCs in models of FD.This is the first gene therapy study in an FD model and lays the foundation for a translational path for mitigating RGC loss in FD patients.Additionally, we show that in a new mouse model, the oral administration of the novel PTC680 compound significantly improved RGC survival throughout multiple stages of disease progression.These data are encouraging as most FD patients do not report vision loss until they are teenagers.
Our data show that approximately 30% of RGCs were infected with the human ELP1 virus and expressed the human ELP1 protein, which is in the same range as the percentage of RGCs rescued (39%).Moreover, we provide the first evidence that morphology is not disrupted in RGCs expressing virally transduced ELP1, suggesting that overexpression of human ELP1 is not harmful to retinal neurons.Future studies will be required to test the extent of RGC rescue, i.e., physiological function.One concern is whether the reintroduction of Elp1 protein in RGCs is sufficient to support circuit maintenance with RGC targets in the brain; a second concern, given the broad expression of Elp1 in the brain, is whether visual processing and perception will be improved if RGC inputs are restored.
These data justify the further optimization of an ELP1-vector for translation to a clinical setting.Potential paths forward include optimizing the AAV2 capsid properties via chemical modifications or assembling mosaic capsids consisting of different serotypes could improve targeting specificity and tropism to achieve higher transduction efficiency [27][28][29] .Alternatively, concatamerization of dual AAV vectors [30][31][32][33][34] can be considered as the ELP1 transgene is large (4 kb), making packaging challenging.This would provide more flexibility in promoter choice to maximize transduction efficiency in RGCs.
FD is a complex developmental and degenerative disease, yet because it is caused by a single splicing-associated point mutation, it is an excellent candidate disease for both gene therapy and small molecule splicing modulators., which were a gift from Drs Peter Gruss and Ruth Ashery-Padan.This is a useful mouse model to study the FD optic neuropathy due to the restricted deletion of Elp1 only in retinal neurons with a time course of RGC loss consistent with patients while maintaining a phenotypically healthy mouse.

Animals
A new hybrid mouse model, the TgFD9;Pax6cre;Elp1 loxp/loxp was generated to analyze the splicing compound in a mouse model that expressed the FD founder mutation on a mouse background that was null for mouse Elp1 in the retina.To this end, we crossed our original Pax6-cre;Elp1 loxp/loxp mice, with FD9/FD9; Elp1 loxp/loxp transgenic line that expresses nine copies of the human FD ELP1 mutation 36 .To ensure that crossing in the FD9 transgene did not, by itself, rescue the loss of RGCs in the Pax6cre;Elp1 loxp/loxp model, we quantified RGCs in both models at 3 months and found a ~ 55% reduction, indicating that the low levels of ELP1 transcribed from the human gene were not sufficient to prevent the death of RGCs.
The sample size for all experiments was based on an A priori power analysis of 90% with an effect size of 2 and α of 0.05 to detect a statistical difference between the experimental and control groups [21; Ueki and Lefcort, unpublished].Both male and female mice were used at the ages indicated (E18.5 up to 6 months), and all controls were littermate Pax6cre -;Elp1 loxp/loxp .Mice were randomly assigned to a treatment group based on sex and genotype.All mice were housed in a 12-h light-dark cycle and were fed chow and water ad libitum.

AAV vector design and production
All AAV viruses were made at the Vector Core in the Gene Therapy Center at the University of Massachusetts Medical School.AAV2 was selected based on its reported tropism for mouse retinal ganglion cells and its use and tolerance in the human eye [37][38][39] and preliminary experiments by Ueki and Lefcort (unpublished) revealing www.nature.com/scientificreports/its higher expression efficiency in retinal ganglion cells in our mouse line compared to AAV9.The AAV2 was used at vector doses ranging from 8 × 10 8 to 1.7 × 10 9 vector genomes per eye (vg/eye) per mouse for the mouse Elp1, while the human ELP1 virus was used at doses ranging from 2.7 × 10 7 to 5.4 × 10 8 vg/eye.Vector doses were based on preliminary experiments showing potential viral toxicity at higher doses and evidence for cell rescue at lower doses (data not shown).Each virus expressed the full-length sequence for either mouse Elp1 (Origene MC202501, NM 026079) or human ELP1 (Origene RC2076868), ELP1 (NM_003640), and both were driven by a murine small nuclear RNA promoter (U1a).The U1a promoter was selected based on its small size (~ 250 bp) and ability to effectively transduce cells in the central nervous system 40 .

RT-qPCR
Real-time PCR (RT-qPCR) was used to quantify mRNA collected from snap-frozen retinal tissue.Total RNA was isolated and purified using the Direct-zol RNA mini-prep kit (Zymo Research, #R2050).The RNA concentration and purity was measured using BioTek Epoch 2 microplate spectrophotometer.25 ng of purified RNA was reverse transcribed using SuperScript IV VILO mastermix (Invitrogen, #11756050).qPCR was performed with cDNA equivalent of 25 ng of transcribed RNA using SYBR Green master mix (Bio-Rad, #336501) using the 7500 Fast Real-Time PCR system (Fisher Scientific, Applied Biosystems, #4351106).The cycling condition was performed per the manufacturer's protocol (Bio-Rad).All primers (Supplemental file 1) were designed using NIH Primer-Blast tool.The relative expression analysis was performed using the delta delta CT method (ΔΔCT) and expressed as fold change normalized to the average of the two housekeeping genes.

RT-PCR analysis of ELP1 transcripts
Retinae were snap-frozen in a dry-ice bath with 95% EtOH upon collection.Tissues were homogenized in icecold TRI reagent (Molecular Research Center, Inc., Cincinnati, OH, USA), using a TissueLyser (Qiagen).Total RNA was extracted using the TRI reagent procedure provided by the manufacturer.RNA quality and yield was determined using a Nanodrop ND-1000 spectrophotometer.Reverse transcription was performed with 500 ng of total RNA, random primers, and Superscript™III reverse transcriptase (Invitrogen).PCR was performed for the splicing analysis with the 3 µl of cDNA in a total volume of 20 µl with GoTaq Polymerase 2X (Promega) and 32 amplification cycles.We used human-specific ELP1 primers-forward, 5′-CCT GAG CAG CAA TCA TGTG-3′; reverse, 5′-TAC ATG G TCT TCG TGA CAT C-3′ to amplify human ELP1 expressed from the transgene.The PCR products were separated on 1.5% agarose gels for 2.5 h at 90 V.The relative amounts of WT and mutant (Δ20) ELP1 spliced isoforms in a single PCR were determined using ImageJ and the integrated density value for each band as previously described 41,42 .The relative proportion of the WT isoform detected in a sample was calculated as a percentage.

Manual Hamilton syringe
Using forceps, the eyelids were pushed back to expose the cornea.With a 32-G, a small hole was made at the margin of the cornea and sclera.This was necessary because the A 34-G needle (#W1690678) had a blunt tip.The 5 µl calibrated Hamilton #65 needle was inserted at the margin of the sclera and cornea and into the vitreous cavity and 1 µl of the viral solution was injected slowly into the cavity.The needle was left in position for another 15 s and removed slowly.

WPI-UMP3 with Mirco2T Injector (#0916C)
Using forceps, the eyelids were pushed back to expose the cornea.A 36-G beveled needle (NF36BV) attached to a 10 µl Hamilton NANOFIL syringe (WPI, lot# 08C) was inserted at the margin of the sclera and cornea and into the vitreous cavity.Each eye was injected with 600 nL of AAV preparation bilaterally with a 200 nL/sec delivery rate.The needle was left in position for another 10 s to allow the pressure in the injection system to equilibrate with the pressure in the eye and removed slowly.After all surgeries, Puralube® Vet Ointment (Dechra Veterinary Products) was placed on the eye per IACUC guidelines.Mice were monitored and kept on a 37 °C heating pad until awake and recovered and then transferred to their home cage and monitored for 1 to 2 h.Mice remained in their home cage with cage mates until analysis, 3 months after treatment.

Oral administration of splicing compound using a formulated diet
The newly hybrid transgenic mice carrying the human ELP1 transgene with FD mutation TgFD9;Pax6cre;Elp1 flox/flox were fed a special diet containing 0.0008% of PTC-409680 (Purina Prolab 3000 with PTC-409680, #21032505A2i) corresponding to a daily consumption of 1.2 mg/kg/day.

Retinal histology following ocular gene therapy with AAV2-U1a-mElp1
Retina flat mount dissection and histology were performed as previously described 43,44 .In brief, Mice were euthanized with CO 2 and the temporal surface of the eye was marked with a green tattoo dye (Ketchum Manufacturing) prior to enucleation.After the eyes were enucleated, a hole was made at the sclera-cornea margin and fixed overnight at 4 °C in 4% PFA.Before antibody staining, the neural retinal layer was detached from the retinal pigment epithelium (RPE)/choroid layer.Rabbit anti-Brn3a (1:300; Synaptic Systems, Cat#411 003) was used as the primary antibody.Brn3 was chosen for quantification because it is a well-established nuclear marker of RGCs, and while it only labels ~ 80% of RGCs, it allowed for more consistent and reliable quantification than the antibody to the RNA-binding protein with multiple splicing (RBPMS), which is expressed in the cytoplasm.Cy3 conjugated anti-rabbit (1:500, donkey) secondary antibody was purchased from Jackson Immuno Research.All antibodies were diluted in 1X PBS with 0.3% Triton X-100 and 5% bovine serum albumin (BSA, Cell Signaling Technology).

Retinal histology following ocular gene therapy with AAV2-U1a-hELP1
Mice were euthanized as stated above.Eyes were fixed at room temperature (RT) for 1 h in 4% PFA.Retinae were then removed, with the temporal region marked by a small cut, and incubated in 4% PFA for an additional 15 min at RT. Tissue was permeabilized in 1X PBS containing 0.5% Triton X-100 for 30 min at RT. Non-specific binding was blocked by incubation with 5% BSA containing 0.5% Triton X-100 for 3 h at RT.All retinae were incubated for 2 nights at 4 °C in a primary antibody solution containing 2% BSA, 0.2% Triton X-100, an anti-Brn3 antibody (1:200, Santa Cruz, #sc-6026), and an anti-ELP1(IKAP) antibody (2 µm/ml, Invitrogen, #PA5-111296, RRID AB_2856706).After overnight incubation at 4 °C with secondary antibodies (Donkey anti-Goat IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647 and Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 568 from Thermo Fisher Scientific, catalog # A-21447, RRID AB_2535864 and catalog # A10042, RRID AB_2534017), four identical cuts were made to flatten the tissue into four lobes, and the tissue was mounted with Prolong Gold (Invitrogen).

RGC quantification
Flat mounts were imaged and tiled using a Leica DM6 Thunder microscope at a 10X magnification.For the AAV2-hELP1 injected retinae, flat mount images were obtained with a Leica DMi8 THUNDER imaging system using the navigator spiral scan at a 10X magnification.IMARIS software (Oxford Instruments) was used to quantify the number of Brn3a + cells.To identify Brn3a + cells, the spots module was used.The detected diameter

Figure 1 .
Figure 1.Experimental workflow.(A) Mice were intravitreally injected with an AAV2-vector in both eyes at P21, and the retinal ganglion cell (RGC) number was quantified 3 months later.An experimental or control virus was injected into the vitreous humor of the mouse eye, where it can diffuse into the retina and target RGCs.(B) FD mutant mice were divided randomly into two groups: control and experimental.The control group was fed a standard chow diet, and the experimental group with a specially formulated chow diet containing a novel splicing modulator compound, PTC680.Groups were started on treatment at three different ages: E18.5, P21 or P90.After 3 months on either the control or experimental diet, retinas were analyzed and quantified for exon 20 inclusion and RGC numbers, respectively.Created with BioRender.com.

Figure 2 .
Figure 2. Viral transduction of murine Elp1 rescues RGCs in a mouse model of FD. (A) Representative retinal whole mount (top) and peripheral retina (bottom) 3 months after intravitreal injection with AAV2-U1a-mElp1.RGCs were immunolabeled using an antibody against Brn3a.Scale bar, 500 µm.(B) Quantification of Brn3a + RGCs from control mice (dark gray) compared to FD mice (lighter gray gradient) receiving injections at P21 with either the eGFP (1.6 × 10 9 vg/eye) or experimental mElp1 virus (1.6 × 10 9 vg/eye, 1.7 × 10 9 vg/eye, 8.4 × 10 8 vg/eye).Brn3a + cells were counted in a 1mm 2 area in each quadrant of the peripheral retina.The adjusted p-values are displayed.ns p = 0.2, 0.99, *p = 0.02, 0.04, **** p < 0.0001, one-way ANOVA with Tukey's multiple comparisons follow-up test.Data are shown as average ± SEM, and each data point represents an individual retina.The sample size for individual groups is represented by n within each bar.

Figure 3 .
Figure 3.In vivo expression of human ELP1 prevents loss of RGCs in a mouse model of FD. (A) Representative retinal whole mount (top) and peripheral retina (bottom) 3 months after intravitreal injections with hELP1.RGCs were immunolabeled using an antibody against Brn3a.Scale bar, 250 µm.(B) Quantification of Brn3a + RGCs from control littermate mice (dark gray) compared to FD mice (lighter gray gradient) receiving injections at P21 with either the eGFP (8.0 × 10 8 vg/eye) or experimental ELP1 (5.4 × 10 8 vg/eye, 2.7 × 10 8 vg/eye, 2.7 × 10 7 vg/eye) virus.Brn3a + cells were counted in a 1mm 2 area in each quadrant of the peripheral retina.(C) Representative images of Brn3a + RGCs (gray, yellow asterisks represent lower Brn3a intensities) and the human ELP1 protein expression (red, yellow arrows) 3 months after intravitreal injections into mutant mice.Scale bar, 25 µm.The adjusted p-value is displayed.*p = 0.02, ns = 0.818, ns = 0.057; unpaired, two-sided Mann-Whitney U test.There is a statistically significant difference between the untreated FD mice and FD mice receiving the higher dose of AAV2-hELP1 5.4 × 10 8 vg/eye.Data are shown as average ± SEM, and each data point represents an individual retina.The sample size for individual groups is represented by n within each bar.
All mice were housed in the AALAC-accredited Animal Resource Center at Montana State University.All animal use protocols were approved by the Montana State University Institutional Animal Care and Use Committee (IACUC) (protocol No. 2020-15-IA; Bozeman, MT).The study fulfilled the ARRIVE guidelines, and all experiments complied with relevant guidelines and regulations.The generation of the Pax6cre + ;Elp1 flox/flox CKO is described in Ueki et al., 2018.Briefly, retina-specific Elp1 CKO mice were generated by crossing Elp1 floxed (International Knockout Mouse Consortium, Wellcome Sanger Institute, UK) and αPax6 promoter-driven Cre (Pax6-Cre) mice https://doi.org/10.1038/s41598-023-45376-wwww.nature.com/scientificreports/

Figure 4 .
Figure 4. Daily PTC680 consumption protects retinal ganglion cells from cell death and increases full-length ELP1 transcript in a new retinal FD mouse model.(A) Representative retinal whole mount (top) and peripheral retina (bottom) after daily consumption of PTC680 for 3 months.Scale bar, 500 µm.(B) Quantification of Brn3a + cells from control littermate mice (dark gray) and FD mice (light gray) on the vehicle or experimental chow.All mice consumed chow for 3 months, but each experiment started when the mice were at a different age: E18.5 (**p = 0.004, ***p < 0.001), P21 (*p = 0.01, ***p = 0.0000) or P90 (*p = 0.02, *** = 0.0000), to test the efficacy of starting treatment at different time points within disease progression.The adjusted p-values are displayed, one-way ANOVA with Tukey's multiple comparisons follow-up test (C) representative splicing analysis of human ELP1 transcripts and percent of exon 20 inclusion in retinae from vehicle-treated (dark gray) and PTC 0.0008%-treated (light gray) FD mice at the different experimental start times.This figure only demonstrates exon 20 inclusion in FD mice receiving either the PTC or vehicle treatment alone.All full-length gels can be found in supplemental Figs. 4, 5 and 6.The adjusted p-values are displayed.****p < 0.0001, two-tailed unpaired Student's t-test.Data are shown as average ± SEM, and each data point represents an individual retina.The sample size for individual groups is represented by n within each bar.This experiment was conducted three times.Conditional knockout (CKO) = Pax6cre + ;Elp1 loxp/loxp .
HEK293 cells were seeded onto a 12-well plate at a density of 2.5 × 10 4 cells/well for 24 h.Two viral solutions were diluted (1) AAV2-U1a-hELP1 stock to 2.7 × 10 9 VG/ml in DMEM and (2) AAV2-U1a-eGFP stock to 8.8 × 10 9 VG/ml in DMEM.The ELP1 solution was serially diluted to three different multiplicities of infection (m.o.i.) at 20 K, 100 K, and 400 K. M.O.I. was calculated by dividing the initial seeding density by the total viral particles.At 24 h, the complete medium was removed and supplemented with the viral medium.Cells proliferated for 6 h in the viral medium.At 30 h, the viral medium was removed and replaced by the complete medium.Cells were harvested 4 days post-transduction (PTD).