Chemical enhancement of torsinA function in cell and animal models of torsion dystonia

Abstract

Movement disorders represent a significant societal burden for which therapeutic options are limited and focused on treating disease symptomality. Early-onset torsion dystonia (EOTD) is one such disorder characterized by sustained and involuntary muscle contractions that frequently cause repetitive movements or abnormal postures. Transmitted in an autosomal dominant manner with reduced penetrance, EOTD is caused in most cases by the deletion of a glutamic acid (DeltaE) in the DYT1 (also known as TOR1A) gene product, torsinA. Although some patients respond well to anticholingerics, therapy is primarily limited to either neurosurgery or chemodenervation. As mutant torsinA (DeltaE) expression results in decreased torsinA function, therapeutic strategies directed toward enhancement of wild-type (WT) torsinA activity in patients who are heterozygous for mutant DYT1 may restore normal cellular functionality. Here, we report results from the first-ever screen for candidate small molecule therapeutics for EOTD, using multiple activity-based readouts for torsinA function in Caenorhabditis elegans, subsequent validation in human DYT1 patient fibroblasts, and behavioral rescue in a mouse model of DYT1 dystonia. We exploited the nematode to rapidly discern chemical effectors of torsinA and identified two classes of antibiotics, quinolones and aminopenicillins, which enhance WT torsinA activity in two separate in vivo assays. Representative molecules were assayed in EOTD patient fibroblasts for improvements in torsinA-dependent secretory function, which was improved significantly by ampicillin. Furthermore, a behavioral defect associated with an EOTD mouse knock-in model was also rescued following administration of ampicillin. These combined data indicate that specific small molecules that enhance torsinA activity represent a promising new approach toward therapeutic development for EOTD, and potentially for other diseases involving the processing of mutant proteins.

Fig. 1.

Assays for torsinA activity used in drug screening procedures. (A,B) Human torsinA (WT) suppresses Q82::GFP (Q82) aggregation in the body wall muscle cells of C. elegans. (A) Representative images of the following worm strains at the L3 stage: P_unc-54::Q82::GFP (Q82), P_unc-54::Q82::GFP + P_unc-54::torsinA (Q82 + WT), P_unc-54::Q82::GFP + P_unc-54::torsinA(ΔE) (Q82 + ΔE), and P_unc-54::Q82::GFP + P_unc-54::torsinA + P_unc-54::torsinA(ΔE) (Q82 + WT/ΔE). Arrows indicate soluble GFP and symbols represent the worm strains depicted in B. (B) The average aggregate number per worm at different developmental stages (larval stages L2–L4 and young adults) of Q82 (triangle), Q82 + WT (diamond), Q82 + ΔE (circle) and Q82 + WT/ΔE (square) worms. Data points are the mean aggregate number of 30 worms for each transgenic line at each time point. The oval indicates the stage used to examine compound exposure of Q82 + WT/ΔE worms. (C,D) torsinA prevents dopamine neuron (DA neuron) neurodegeneration resulting from the overexpression of α-syn in C. elegans as animals age. (C) Representative images of a worm displaying DA neuron degeneration [top image, P_dat-1::GFP + P_dat-1::α-syn (α-syn)], in which only one cephalic (CEP)-class DA neuron is intact (arrow), and a torsinA-rescued worm [bottom image, P_dat-1::GFP + P_dat-1::α-syn + P_dat-1::torsinA (α-syn + WT)], in which all four of the CEP-class neurons (arrows) are protected. (D) Quantitative analysis of the average CEP DA neuron degeneration in isogenic lines of worms expressing α-syn, α-syn + WT, or α-syn + ΔE at the 4-day-old adult stage. Data points are the means of three independent experiments ± the standard error of the mean (S.E.M.); *P<0.05.

Fig. 2.

Analysis of positive drug candidates in different transgenic worm backgrounds. (A–D) Drug efficacy is depicted for color-coded compounds as the standardized enhancement in chaperone activity for torsinA. The values were calculated using the following formula: (A – B)/A, where ‘A’ stands for the average aggregate number in the solvent control and ‘B’ represents the average aggregate number in the treatment. Data points are the mean from three independent experiments (with 30 L3-stage worms/analysis) ± S.E.M.; the statistical significance compared with respective solvent controls in A and with treatment in B was calculated by t test, *P<0.05. The genetic backgrounds analyzed were as follows: (A) Q82 + WT/ΔE, (B) Q82 alone (no torsinA), (C) Q82 + WT and (D) Q82 + ΔE. (E–G) torsinA dependency of the five compounds that were identified in the initial polyglutamine screen for DA neuroprotection. The standardized enhancement in DA neuroprotection for torsinA for each compound was analyzed in 4-day-old adult worms with the following genetic backgrounds: (E) α-syn + WT, (F) α-syn + ΔE, and (G) α-syn alone (no torsinA). Data points are the mean of 35–70 worms for each genetic background from three independent experiments ± S.E.M. Each value was compared with its appropriate vehicle control and statistical significance calculated by t test, *P<0.05.

Fig. 3.

SAR modeling of quinolone and β-lactam antibiotics reveals structural relatedness. (A) The structure of a generalized quinolone showing the bicyclic antibiotic core (red) and surrounding chemical moieties. (B) The structure of an aminopenicillin antibiotic showing the β-lactam nucleus (red) and surrounding chemical moieties. (C) Alignment of structures between a quinolone (norfloxacin; top structure) and an aminopenicillin (cyclacillin; bottom structure) displaying the relative chemical positions of the common components. (D) Alignment of norfloxacin (gray) and cyclacillin (light blue). (E) A pharmacophoric model using alignment of the structures of five quinolones and three aminopenicillins that enhanced WT torsinA activity. The antibiotic core region of the active molecules may function as a scaffold for presenting the alternating H-bond donor/acceptor, which is shown in red (arrow).

Fig. 4.

Ampicillin restores normal torsinA readouts in EOTD patient cells and in a mouse knock-in model for EOTD. (A,B) Control (WT) and DYT1 (WT/ΔE) fibroblasts were infected with a vector encoding Gluc and, 72 hours later, treated with saline as a vehicle control (gray bars) or ampicillin (yellow). Gluc secretion was then measured over the following 48-hour period. Light unit values were standardized to 100% for control cells treated with vehicle only. Both 3 and 6 μg/ml of ampicillin restored DYT1 fibroblast secretion to levels that were not significantly different from control cells treated with vehicle. (C,D) Ampicillin treatment significantly improved motor coordination and balance in Dyt1 ΔGAG knock-in mice (WT/ΔE). Following administration of a second intraperitoneal injection of ampicillin (day 4), ΔE knock-in mice showed significantly fewer slips and an improvement in beam-walking tests. Data were normalized to control mice that were injected with vehicle (saline) solution. *P<0.05. Vertical bars represent mean ± S.E.M.

Fig. 5.

Effect of ampicillin treatment on the levels of torsinA in human DYT1 (WT/ΔE) patient fibroblasts and in Dyt1 ΔGAG (WT/ΔE) knock-in mouse striatum. (A,B) Representative blots from one of the four different experiments carried out using the control fibroblast line (HF71) (A), and from one of the three different experiments using a DYT1 patient fibroblast line (HF49) (B), treated with ampicillin over the course of 72 hours. Western blotting was performed using anti-β-actin and anti-torsinA antibodies. (C) Densitometry was performed to quantitate torsinA and actin bands on all western blots. The average of the ratio of torsinA and actin levels is expressed as the mean ± S.D. for all time points. The asterisk represents a significant difference in torsinA levels in DYT1 (WT/ΔE) fibroblasts compared with the control (WT) fibroblast line after 72 hours of treatment with ampicillin (P<0.0015). (D) Representative images of the western detection for β-tubulin and torsinA in extracts of treated WT or knock-in (WT/ΔE) mouse striatum. Genotypes [control littermate (WT) and knock-in (WT/ΔE)] and the treatments (0 or 100 mg/kg ampicillin) of the samples are shown. (E) Quantified torsinA level in the striatum. TorsinA levels were normalized to β-tubulin and each group was statistically compared with the level in the vehicle-treated WT mice. The vertical bars represent mean ± S.E.M. *P<0.05.