CAG-targeted brain-permeable therapy tested in biallelic humanized polyQ mouse models

Abstract

In polyglutamine (polyQ) diseases, including Huntington disease (HD) and spinocerebellar ataxia type 3 (SCA3), targeting the mutant CAG tract in mRNA could be a therapeutic strategy for lowering pathogenic protein. We explored the viability of this therapeutic strategy in vivo at the level of the reagent design, toxicity, systemic delivery, brain regions transduction, silencing efficiency, and allele preference. We designed a series of CAG-directed short hairpin RNAs (shRNAs) based on a previous A2 reagent, allele selective in vitro. Humanized HD (Hu^128Q/21Q) and SCA3 (Ki^150Q/21Q) mice with mutant ∼100 CAGs and normal 21 CAGs alleles were used to simulate biallelic conditions occurring in patients. We administered AAV-PHP.eB shRNAs-encoding vectors into the blood as an equivalent of non-invasive CAG-directed brain-targeted therapy crossing the blood-brain barrier. We demonstrate that optimized CAG-targeted A4(P10) and A4(P10,11) shReagents can lower mutant huntingtin and ataxin-3 protein and its aggregates by targeting brain regions selectively and with diminished toxicity compared to other tested shRNAs. The important considerations of the approach are the silencing efficiency depending on the transduction region and careful dose adjustment. Moreover, the CAG approach could be suitable to target somatic expansion. Our work paves the way toward developing the therapy for polyQ diseases, potentially shortening drug development.

Keywords: AAV-PHP.eB; CAG repeats targeting; Huntington disease; MT: Oligonucleotides: Therapies and Applications; SCA3; blood-brain barrier; gene therapy; neurodegenerative disease; shRNA; short hairpin RNA; spinocerebellar ataxia type 3; systemic delivery.

Graphical abstract

Figure 1

Distribution, transduction efficiency, side effects, and lowering HTT protein of AAV-PHP.eB_EGFP_shRNA in polyQ mouse models (A) Illustration of EGFP signal distribution in Hu^128Q/21Q brain structures performed 3 weeks post-retroorbital injection of AAV-PHP.eB_EGFP_shRNA (dose: 1.5 × 10¹³ vg/kg). Magnification 20×, scale bar, 100 μm. (B) The graph shows the relative transduction level as EGFP fluorescence intensity (RawIntDen) per square micrometer (intensity/μm²) in the 8 transduced brain structures 3 weeks post-injection. The dark-green bars show intensely transduced brain structures TH, CB, and BS, while light-green bars correspond to structures CTX, HP, and STR, showing weaker transduction. The data show mean ± the standard error of the mean (SEM) as error bars (n = 3). (C) Evaluation of efficiency in lowering HTT protein in the TH and STR 3 weeks post-injection of 8 shRNA shReagents (doses: 1.5 × 10¹³ and 0.5 × 10¹³ vg/kg for A2). (D) The table presents the tremor score in Hu^128Q/21Q mice at 3 weeks and in Hu^128Q/21Q and Ki^150Q/21Q mice at 15 weeks post-injection AAV-PHP.eB_EGFP_shRNA. In (C)the fold change was calculated from mean and presented ± SEM as error bars (n = 2, 3, or 4). The mean of shScrambled are set as value of 1 fold change vs shReagents. Means (respective shReagents and shScrambled) were analyzed with the Student’s t test; ∗p < 0.05; ∗∗p < 0.01. (E) Assessment of EGFP signals in brain structures of Hu^128Q/21Q (top) and Ki^150Q/21Q mice (bottom) 15 weeks post-injection. Magnification 40×; scale bar, 1000 μm. Brain regions: brain stem (BS), cerebellum (CB), cortex (CTX), hippocampus (HP), midbrain (MB), olfactory bulb (OB), and striatum (STR). Photographs show EGFP (green) and DAPI (blue).

Figure 2

Lowering effect of shRNAs in brain regions of HD mouse model (A) Schematic of A4(P10A) (blue) and A4(P10,11A) (yellow) binding on the CAG tract in exon 1 of the HTT mRNA. (B) Quantification of HTT protein in the TH, CB, BS, CTX, STR, and HP of Hu^128Q/21Q mice injected with AAV-PHP.eB expressing A4(P10A) and A4(P10,11A) and shScrambled (1.5 × 10¹³ vg/kg). Brains with green dots in the corner show transduced brain regions; (top) well-transduced (TH, CB, and BS, dark green dots) and weakly transduced (CTX, STR, and HP, light green dots) brain structures. Vinculin was used as a control protein for normalization. Representative western blot images are shown under graphs; 128Q represents mutant HTT protein and 21Q represents WT HTT protein. (C) Collective lowering of HTT protein in well- (left) and weakly transduced (right) brain structures. (D) HTT aggregates were evaluated by immunohistochemical staining of the BS and CB with the anti-HTT EM48 antibody. The graphs shows data of HTT aggregate surface as a percentage of EGFP⁺ area (transduced cell area) and the data of HTT aggregate density (mean IntDen) in the BS or CB 15 weeks post-injection with A4(P10A) as shReagent and control shScrambled delivered in vector AAV-PHP.eB (n > 5 areas in brain structure in each mouse, n = 3). (E) Relative mRNA level of HTT (fold change) in various brain regions of the HD mouse model post-injection A4(P10A) and A4(P10,11A). Relative HTT transcript levels were measured by qPCR and normalized by the Actinβ gene expression level. The mRNA was extracted from the TH, CB, CTX, STR, and HP of Hu^128Q/21Q. In (B-C) the fold change was calculated from mean and presented ± SEM as error bars (n = 3). In (B-C) the mean of shScrambled are set as value of 1 fold change vs shReagent on the graphs. The (B and C) means were analyzed with a Student’s t test. The data in (D) were analysed with Mann-Whitney test and ploted on graphs as median ±ranks. The data in (E) were analysed with Kruskal-Wallis test for multiple comparisons and subsequently post hoc analysed with Dunn test for pairwise comparison and the parameters were ploted as box plot. The p values in tests were indicated as follows ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Figure 3

Lowering effect of shRNAs in brain regions of SCA3 mouse model (A) Schematic of A4(P10A) (blue) and A4(P10,11A) (yellow) binding on the CAG tract in exon 10 of the ATXN3 mRNA. (B) Quantification of ATXN3 protein in the TH, CB, BS, CTX, STR, and HP of Ki^150Q/21Q mice injected with AAV-PHP.eB expressing A4(P10A) and A4(P10,11A) and shScrambled (1.5 × 10¹³ vg/kg). Brains with green dots in the corner show transduction in the brain region; (top) well-transduced (TH, CB, and BS, dark-green dots) and lower weakly transduced (CTX, STR, and HP, light-green dots) brain structures. Lamina B (LAM) was used as a control protein for normalization. Representative western blot images are shown under graphs; 21Q represents normal ATXN3 and 150Q represents mutant ATXN3 protein. The differences in PAGE mobility apparent on blot pictures are due to CAG expansion/contraction and the resulting differences in the length of the polyQ tract in the mutant ATXN3 protein in individual mice. (C) The collective lowering of ATXN3 protein in well (top) and weakly transduced (bottom) brain structures. (D) ATXN3 aggregates were evaluated by immunohistochemical staining of the BS with the anti-ATXN3 clone 1H9 antibody. The graph shows the data of ATXN3 aggregate surface expressed as a value of percentage of eGFP⁺ area (transfected cell area) and the data of ATXN3 aggregate density expressed as mean IntDen in the BS 15 weeks post-injection, with A4(P10A) as shReagent and control shScrambled delivered in vector AAV-PHP.eB (n > 5 area in brain structure in each mouse, n = 3). (E) Relative mRNA level of ATXN3 (fold change) in various brain regions of the SCA3 mouse model post-injection A4(P10A) and A4(P10,11A). Relative ATXN3 transcript levels were measured by qPCR and normalized by Actinβ gene expression level. The mRNA was extracted from the TH, CB, CTX, and STR Ki^150Q/21Q. In (B and C) the fold change was calculated from mean and presented ± SEM as error bars (n = 3). The mean of shScrambled are set as value of 1 fold change vs shReagents on the graphs. The (B–C) means were analyzed with Student’s t test; The data in (D) were analysed with Mann-Whitney test and ploted on graphs as median ±ranks. The data in (E) were analysed with Kruskal-Wallis test for multiple comparisons and subsequently post hoc analysed with Dunn test for pairwise comparison and the parameters were ploted as box plot. The p values in tests were indicated as follows ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Figure 4

Phenotypic changes in HD and SCA3 mouse models 15 weeks post-injection AAV_PHP.eB_shRNA (A and B) The graph demonstrates the phenotype scoring test (PST) results, where 0 points are the normal condition and 12 points are the most affected mouse condition. The points were scored by animals 15 weeks post-injection and are expressed as average scores. (C and D) The graphs demonstrate differences in body weight gain of (C) HD and (D) SCA3 mice measured at 4, 10, 12, and 15 weeks post-injection with A4(P10A) (n = 8) and A4(P10,11A) (n = 9) and shScrambled (n = 11) shRNAs; the weight is expressed in grams. (E and F) Postmortem organ weights of brain, heart, spleen, and testis were recorded 15 weeks post-injection. The PST and organ weight data represent the mean ± SEM as error bars. Means (shReagent and shScrambled) were analyzed with Student’s t test; ∗p < 0.05 (n = 6). The body weight data were analyzed by two-way ANOVA (p < 0.05) and Fisher’s least significant difference post hoc (∗p < 0.05; ∗∗p < 0.01).

Figure 5

Liver transduction and biochemical parameters evaluation in HD and SCA3 mouse model The EGFP signal in representative liver sections of (A) Hu^128Q/21Q and Ki^150Q/21Q mice. (B–E) (B) The change in AST and ALT parameters, (C) level of total cholesterol (Chol) and triglycerides (TGs), (D) concentrations of magnesium (Mg) and calcium (Ca), and (E) creatine kinase (CK-MB) in the serum of Hu^128Q/21Q and Ki^150Q/21Q mice 15 weeks after AAV-PHP.eB_shRNA injections. The data shown in (B)–(E) are mean ± SEM as error bars (n = 3 or 4), and the analysis compares shReagent- vs shScrambled -treated mice using Student’s t-test; ∗p < 0.05; ∗∗p < 0.01.