Topoisomerase inhibitors unsilence the dormant allele of Ube3a in neurons

Abstract

Angelman syndrome is a severe neurodevelopmental disorder caused by deletion or mutation of the maternal allele of the ubiquitin protein ligase E3A (Ube3a)^1–3. In neurons, the paternal allele of Ube3a is intact but epigenetically silenced^4–6, raising the possibility that Angelman syndrome could be treated by activating this silenced allele to restore functional UBE3A protein^7,8. Using an unbiased, high-content screen in primary cortical neurons from mice, we identified twelve topoisomerase I inhibitors and four topoisomerase II inhibitors that unsilence the paternal Ube3a allele. These drugs included topotecan, irinotecan, etoposide, and dexrazoxane (ICRF-187). At nanomolar concentrations, topotecan upregulated catalytically active UBE3A in neurons from maternal Ube3a-null mice. Topotecan concomitantly downregulated expression of the Ube3a antisense transcript that overlaps the paternal copy of Ube3a^9–11. These results suggest that topotecan unsilences Ube3a in cis by reducing transcription of an imprinted antisense RNA. When administered in vivo, topotecan unsilenced the paternal Ube3a allele in several regions of the nervous system, including neurons in the hippocampus, neocortex, striatum, cerebellum and spinal cord. Paternal expression of Ube3a remained elevated in a subset of spinal cord neurons for at least twelve weeks after cessation of topotecan treatment, suggesting transient topoisomerase inhibition can have enduring effects on gene expression. While potential off-target effects remain to be investigated, our findings suggest a therapeutic strategy for reactivating the functional but dormant allele of Ube3a in patients with Angelman syndrome.

No effective therapies exist for Angelman syndrome (AS)—an imprinting disorder caused by mutations or deletions in the maternal allele of Ube3a^1–3. Ube3a is biallelically expressed in most tissues of the body; however, in rodents and humans, most neurons express Ube3a only from the maternally-inherited allele^4,12–14. This unique epigenetic pattern of regulation suggested that it might be possible to unsilence the dormant paternal Ube3a allele in neurons^7,8.

To test this possibility, we developed a 384-well high-content screen using primary mouse cortical neurons from Ube3a-Yellow Fluorescent Protein (Ube3a-YFP) knockin mice¹⁵, and searched for drug-like molecules that could unsilence the paternal Ube3a-YFP allele (Fig. 1a). This screen was based on our observation that the imprinting of Ube3a-YFP was maintained in vitro in cultured embryonic cortical neurons. Notably, Ube3a-YFP expression was undetectable (silenced) in cultured neurons when paternally inherited (Ube3a^m+/pYFP), but was expressed when maternally inherited (Ube3a^mYFP/p+) (Fig. 1b), with expression increasing from 4 to 10 days in vitro (DIV) (Fig. 1c). This significant difference between maternal and paternal UBE3A-YFP protein levels provided a large screening window and a Z’-factor score of 0.58 (determined by statistically comparing antibody-enhanced fluorescence intensities and variations between maternal and paternal UBE3A-YFP signals at DIV10), making our high-content platform suitable for unbiased screening.

Figure 1. A small-molecule screen identifies a topoisomerase inhibitor that unsilences the paternal allele of Ube3a in neurons.

a, High-content screen flowchart. E15.5 cortical neurons with a paternally inherited Ube3a-YFP allele were cultured in 384-well plates and treated with small molecules from DIV7–DIV10. Active compounds that unsilence the paternal Ube3a-YFP allele were detected with antibody-enhanced fluorescence and high-content imaging. b, High-content imaging of DIV7 neurons that inherited Ube3a-YFP maternally (m^YFP/p⁺) or paternally (m⁺/p^YFP). Nuclei were stained with DAPI. Scale bar = 50 µm. c, Mean ± s.e.m. levels of UBE3A-YFP fluorescence in neurons cultured from maternal Ube3a-YFP (m^YFP/p⁺) or wild-type (m⁺/p⁺) mice, normalized to levels in age-matched neurons cultured from paternal Ube3a-YFP mice (m⁺/p^YFP). Two-way ANOVA revealed main effects of genotype, duration, and a genotype-duration interaction (P<0.001); Bonferroni post hoc test examined comparisons between maternal and paternal Ube3a-YFP mice from DIV4 to DIV10, *P<0.001; n=2–6 culture wells/day. d, Pie chart depicting categories of the 2,306 screened compounds and graph summarizing presumptive UBE3A-YFP in arbitrary fluorescence units (A.F.U.) after small molecule treatments. Small molecules that were subsequently found to be autofluorescent (Supplementary Fig. 2) are depicted in gray. The initial screen identified one active compound, irinotecan (red). e, High magnification view of wells treated with vehicle (0.2% DMSO) or 10 µM irinotecan for 72 hr. Neuron density and health is similar in vehicle- and irinotecan-treated cells as evidenced by counterstaining with the nuclear marker DAPI. Scale bar = 100 µm. f,g Western blots of cultured cortical neurons probed with GFP or UBE3A antibodies, respectively, ± irinotecan (10 µM for 72h). f, *P<0.05; two-tailed t-test, n=3/group. g, *P<0.001, one-way ANOVA with Bonferroni post hoc, n=7–8/group. All data are presented as means ± s.e.m.

To perform the screen, we cultured Ube3a^m+/pYFP neurons for 7 days and then treated these neurons with compounds (10 µM for 72 hours) from multiple small molecule libraries (Fig. 1d, Supplementary Fig. 1). In total, we screened 2,306 small molecules in quadruplicate, normalizing values to vehicle-treatment (0.2% DMSO) (for full methods see Supplementary Methods; for list of all compounds tested see Supplementary Table 1). While methylation and other epigenetic marks are thought to control imprinting of Ube3a^9,16–18, to our surprise, none of the commonly used compounds that target the epigenome, including chromatin remodeling drugs and DNA methyltransferase inhibitors, unsilenced the paternal Ube3a-YFP allele. A number of compounds were identified as false positives (gray compounds in Fig. 1d) due to their intrinsic fluorescence (Supplementary Fig. 2). Our initial screen identified one compound—irinotecan, an FDA-approved camptothecin-based topoisomerase type I inhibitor. Irinotecan lacked intrinsic fluorescence and upregulated UBE3A-YFP fluorescence (Fig. 1d,e and Supplementary Fig. 3). Irinotecan (10 µM) also upregulated paternal UBE3A-YFP protein (Fig. 1f) and endogenous UBE3A protein (Fig. 1g) in neuronal cultures from Ube3a^m+/pYFP and Ube3a^m–/p+ mice (AS model mice¹³), respectively.

Many topoisomerase I inhibitors, including irinotecan and the related FDA-approved drug topotecan, are derived from the natural product camptothecin (CPT)¹⁹. To explore structure activity relationships, we tested CPT analogs and other topoisomerase inhibitors (Fig. 2a; Supplementary Figs. 4–10), all of which lack inherent fluorescence (Supplementary Fig. 3). We found that irinotecan and topotecan upregulated paternal UBE3A-YFP in a dose-and time-dependent manner in cultured neurons, with topotecan being 20× more potent than irinotecan (Fig. 2a,b; Supplementary Fig. 11). In contrast, an inactive analog of CPT (lactam E-ring-CPT) that does not inhibit topoisomerases²⁰ failed to unsilence the paternal Ube3a-YFP allele (Fig. 2a; Supplementary Fig. 4). Ten additional topoisomerase I inhibitors unsilenced Ube3a-YFP in a dose-dependent manner, including CPT analogs and structurally distinct indenoisoquinolines (Table 1 and Supplementary Figs. 4–7). Furthermore, four structurally distinct topoisomerase II inhibitors (etoposide, dexrazoxane, ICRF-193, and amsacrine) also unsilenced the paternal Ube3a-YFP allele (Table 1 and Supplementary Figs. 8–10). Thus, our data with 16 topoisomerase inhibitors and one inactive analog strongly suggest that inhibition of topoisomerase I or II can unsilence the paternal Ube3a allele.

Figure 2. Topotecan unsilences the paternal allele of Ube3a and the unsilenced protein is catalytically active.

a, Dose-response curves for unsilencing paternal Ube3a-YFP. Inactive = lactam E ring-camptothecin. n=4/data point. b, UBE3A-YFP levels in neurons from Ube3a^m+/pYFP mice increase with duration of topotecan (300 nM) or irinotecan (1 µM) treatment. *P<0.05, oneway ANOVA with Bonferroni post hoc tests relative to day zero, n=4–8/group. A.F.U.=arbitrary fluorescence units. c, Western blots and quantification of UBE3A and the loading control actin. *P<0.05, one-way ANOVA with Bonferroni post hoc tests, n=4/group. d, Quantification of unbound TOP1 and representative western blots. β-tubulin was used as a loading control. One-way ANOVA with Bonferroni post hoc tests, *P<0.05; n=3/group. e, Western blot from vehicle- and topotecan-treated neurons from wild-type (m⁺/p⁺) and maternal Ube3a-deficient (m^–/p⁺) mice. f, Western blots examining UBE3A ubiquitin-thioester formation following immunoprecipitation with an anti-UBE3A antibody and in vitro ubiquitination in the presence or absence of the ubiquitin conjugating enzyme (E2), UBCH7. All data are presented as means ± s.e.m. g, Schematic demonstrating location of 4 primer sets used to probe mRNA expression shown in h. h, Normalized mRNA levels in cultured Ube3a^m–/p+ neurons following vehicle or 300 nM topotecan. Expression is given as a ratio of expression in drug treated cells to vehicle treated cells, normalized to the housekeeping gene RPL22. *P<0.05 compared to 0 hr, Kruskal-Wallis one-way ANOVA followed by post hoc tests, n=4–5 cultures/data point. i, Schematic summarizing methylation status of the Snrpn promoter region on the maternal and paternal chromosome following treatment with vehicle or 300 nM topotecan (see complete primer 1 data set in Supplementary Fig.12). Average methylation status is indicated using a grayscale.

Table 1.

Efficacies and potencies of topoisomerase inhibitors for unsilencing the paternal allele of Ube3a-YFP in cultured neurons.

Compound	Potency EC50 (nM)	Efficacy Emax (fold over vehicle)
Camptothecin derivatives
7-Ethyl-Camptothecin (7-Ethyl-C PT)	7.2 ± 2.3	1.70 ± 0.04
7-Ethyl-10-H ydroxy-CPT	11 ± 3.2	1.99 ± 0.06
10-H ydroxy-CPT	14 ± 5.7	1.82 ± 0.08
Belotecan (CKD 602)	19 ± 4.4	1.88 ± 0.05
Camptothecin (CPT)	21 ± 3.8	2.11 ± 0.05
Topotecan*	54 ± 3.4	2.25 ± 0.05
Rubitecan (9-Nitro-CPT)	62 ± 18	2.09 ± 0.09
Irinotecan*	994 ± 13	2.17 ± 0.05
Silatecan (DB67)	2,244 ± 171	1.65 ± 0.05
Lactam E ring-CPT (inactive)	inactive	inactive
Indenoisoquinoline derivatives
NSC725776	10 ± 1.6	1.76 ± 0.03
NSC706744	11 ± 3.2	1.84 ± 0.07
NSC724998	14 ± 2.2	1.69 ± 0.03
Podophyllotoxin derivative
Etoposide*	1,600 ± 980	1.68 ± 0.04
Bis-dioxopiperazine derivatives
ICRF-193	205 ± 70	2.21 ± 0.09
Dexrazoxane (ICRF-187)*	2,0470 ± 1,450	1.82 ± 0.05
Aminoacridine derivative
Amsacrine	27 ± 5.2	1.74 ± 0.06

^*FDA-approved compounds

We focused our remaining studies on the topoisomerase I inhibitor topotecan because it is approved for use in humans, it unsilenced Ube3a in the low nanomolar range, and topotecan (300 nM, 72 h) restored UBE3A protein to wild-type levels in cultured neurons from Ube3a^m–/p+ mice (Fig. 2c).

Many topoisomerase inhibitors, including topotecan, covalently link topoisomerases to DNA, forming stable DNA-enzyme complexes that are separable from free topoisomerase enzymes¹⁹. Since topotecan inhibits topoisomerase I (TOP1) and Top1 is expressed at high levels in the developing and adult brain^19,21, we focused our subsequent analysis on this enzyme. We found that topotecan (300 nM, 72 h) significantly reduced the amount of free TOP1 (Fig. 2d) in cultured neurons, indicating that topotecan engages its known molecular target at doses that unsilence the paternal Ube3a allele.

UBE3A is a HECT (homology to E6 carboxyl terminus) domain E3 ligase that forms a thioester-ubiquitin intermediate in the presence of E1 and E2 enzymes²². This thioester-ubiquitin intermediate is required for HECT domain E3 ligases to mono- and polyubiquitinate their substrates²³. Interestingly, we noticed that topotecan (300 nM, 72 hr) upregulated UBE3A protein in Ube3a^m–/p+ cultures along with a higher molecular weight form (resolved after running gels for longer times; Fig. 2e). This high molecular weight band was also seen in wild-type (Ube3a^m+/p+) cultures and was lost upon addition of the reducing agent dithiothreitol (DTT) (Fig. 2e). These data suggest that the unsilenced paternal copy of UBE3A is catalytically active and competent to form a thioester-ubiquitin intermediate, just like wild-type, maternal-derived UBE3A²³.

To further demonstrate that UBE3A was catalytically active, we immunoprecipitated UBE3A from cultured wild-type and Ube3a^m–/p+ neurons (+/− topotecan), then tested these samples for a gel-mobility-shift in the presence or absence of the ubiquitin E2 UBCH7²⁴. Both wild-type (maternal-derived) and topotecan-unsilenced (paternal-derived) UBE3A underwent mobility shifts in the presence of UBCH7 plus free ubiquitin that were abolished by DTT (Fig. 2f). This observation indicates the mobility shift was due to addition of covalent ubiquitin, and demonstrates that topotecan can upregulate a functional UBE3A enzyme.

Ube3a is repressed in cis by a large antisense transcript (Ube3a-ATS) that overlaps the paternal allele of Ube3a (Fig. 2g)^9,10. Ube3a-ATS is expressed exclusively from the paternal allele as a result of allele-specific methylation of an imprinting center that overlaps the Ube3a-ATS and Snurf/Snrpn transcription start site²⁵. We next sought to determine if topotecan regulated Ube3a expression through changes in Ube3a-ATS expression or altered methylation at the imprinting center. We found that topotecan upregulated expression of Ube3a in cultured neurons from Ube3a^m–/p+ mice while concomitantly downregulating expression of Ube3a-ATS and Snrpn (Fig. 2h). However, topotecan did not alter methylation at the imprinting center (Fig. 2i, Supplementary Fig. 12). Taken together, these data suggest that topotecan unsilences paternal Ube3a by reducing transcription of a regulatory antisense RNA without appreciably affecting genomic methylation at the imprinting center.

We then sought to determine if topotecan could unsilence the paternal Ube3a allele in vivo. We first identified a dose that was well tolerated, meaning there were no significant decreases in body weight between the beginning and end of the drug treatments (Supplementary Fig. 13). We then administered topotecan (3.74 µg/h) unilaterally into the lateral ventricle of Ube3a^m+/pYFP or Ube3a^m-/p+ mice by intracerebroventricular (i.c.v.) infusion for two weeks and then sacrificed the mice either immediately or 5 hr after drug cessation. Strikingly, topotecan unsilenced paternal Ube3a in the hippocampus, striatum, and cerebral cortex of the infused hemisphere, but had only a modest effect on the contralateral (non-infused) hemisphere with no effect in the cerebellum (Fig. 3a–e, Supplementary Figs. 14–15). Pharmacokinetic analyses demonstrated that a significant amount of topotecan was detectable in the infused hemisphere immediately following treatment whereas low levels were present in the contralateral (non-infused) hemisphere and in cerebellum (Fig. 3a, Supplementary Fig. 14). However, a higher dose of topotecan (21.6 µg/h for five days) did unsilence the paternal allele of Ube3a in Purkinje neurons of the cerebellum (Supplementary Fig. 16). No significant difference in topotecan levels was detected in blood between drug- and vehicle-treated mice (data not shown). Topotecan concentrations significantly declined five hours after cessation of i.c.v. drug delivery (Fig. 3a), indicating that topotecan does not persist and is rapidly removed/metabolized in the brain. Taken together, these pharmacokinetic and pharmacodynamic data suggest that the degree to which topotecan unsilences the paternal Ube3a allele is directly correlated with drug concentrations in the brain. Moreover, our data indicate topotecan has the potential to unsilence the paternal Ube3a allele throughout the nervous system.

Figure 3. Topotecan enduringly unsilences the paternal allele of Ube3a in vivo.

a, Unilateral delivery of topotecan (i.c.v.) using a mini-osmotic pump into the lateral ventricle of Ube3a^m+/pYFP mice in vivo. Two weeks of topotecan infusion (3.74 µg/h) unsilenced the paternal Ube3a-YFP allele in the hippocampus of the infused hemisphere near the site of drug delivery, while only modestly unsilencing Ube3a-YFP in the contralateral (non-infused) hemisphere. Scale bar = 500 µm. Pharmacokinetic analyses measuring topotecan levels in the infused and non-infused hemisphere immediately (t=0) or five hours (t=5) after cessation of drug delivery. *P<0.05, one-way ANOVA with Bonferroni post hoc test, n=5–9/group. b, Representative sections and c, quantification of optical intensity of UBE3A-YFP in hippocampal regions (CA1, CA2/3, and dentate gyrus=DG) of the topotecan-infused hemisphere or the hemisphere of vehicle-treated mice. *P<0.05, Mann-Whitney rank sum test, n=5/group. d, Representative sections and e, quantification of paternal UBE3A-YFP in the striatum following i.c.v. infusion of topotecan. *P<0.05, two-tailed t test, n=4/group. f, Schematic depicting schedule for i.t. delivery of topotecan (50 nmol/day for 10 of 14 days) and endpoints (arrows) immediately, 4 weeks, and 12 weeks after cessation of drug. g,h Topotecan (i.t.) increased the number of UBE3A-YFP-positive spinal neurons compared to vehicle, and the unsilencing of Ube3a-YFP was maintained for at least 12 weeks. *P<0.001, one-way ANOVA with Bonferroni post hoc test, n=7–8/group.

Genomic imprinting is thought to be established only during prescribed germline and embryonic periods of development and imprinted genes typically remain epigenetically regulated throughout life²⁶. Thus, we next sought to determine if topotecan had transient or long-lasting effects on paternal Ube3a expression. To test this possibility, we turned to an intrathecal (i.t.) delivery protocol (Fig. 3f) because topotecan (Fig. 3g) and irinotecan (not shown) unsilenced paternal Ube3a in a sparse population of lumbar spinal neurons, allowing us to quantify all UBE3A-YFP-positive neurons. Moreover, i.t. delivery has been used to deliver topotecan to the brain in humans²⁷. We found that topotecan (50 nmol/5 µL i.t. once daily, for 10 of 14 days) was well tolerated (Supplementary Fig. 13) and significantly increased the number of paternal UBE3A-YFP-positive cells in the lumbar spinal cord of mice (Fig. 3g,h, Supplementary Fig. 17a). The vast majority (>93%) of these UBE3A-YFP-positive cells were NeuN⁺ neurons (Supplementary Fig. 17a,b), indicating topotecan unsilences Ube3a primarily in neurons in vivo. Moreover, the unsilenced paternal UBE3A-YFP protein was expressed at levels comparable to maternal UBE3A-YFP controls (Supplementary Fig. 17c). Remarkably, the number of UBE3A-YFP-positive spinal cord neurons remained elevated 12 weeks following cessation of drug treatment (Fig. 3g,h), much longer than the elimination of topotecan from tissue (Fig. 3a). These results indicate that topotecan can enduringly unsilence paternal Ube3a in a subset of spinal neurons and suggest that a single course of drug treatment has the capacity to permanently modify expression of Ube3a.

In conclusion, we found that topoisomerase inhibitors can unsilence the paternal allele of Ube3a and the paternally-derived protein is functional. These findings suggest that topoisomerase inhibitors have the potential to rescue molecular, cellular, and behavioral deficits associated with loss of UBE3A^7,13,28. Ube3a expression is modestly upregulated in the brain of Top2b knockout mice²⁹, providing genetic support that topoisomerases regulate Ube3a expression. We also found that topotecan reduced expression of paternal Snrpn and Ipw, a genomic region whose deletion is associated with Prader-Willi syndrome^7,17. To what extent topotecan regulates expression of other genes in the brain, including Igf2r, Kcnq1, and Gnas imprinted gene clusters, and whether long-term treatments with topoisomerase inhibitors produce a Prader-Willi-like condition are unknown. However, topotecan and irinotecan are approved for use in patients with cancer, penetrate into the central nervous system, and are well-tolerated when administered chronically to both adult and pediatric patients^27,30. Importantly, patients treated with topoisomerase inhibitors do not display symptoms associated with Prader-Willi syndrome. Ultimately, our study indicates that topoisomerase inhibitors regulate gene expression through a transcriptional mechanism and could be used to treat symptoms associated with AS.

Methods Summary

All animal procedures were approved by the University of North Carolina at Chapel Hill Animal Care and Use Committee. We used male and female Ube3a-YFP knockin mice¹⁵, Ube3a^m–/p+ mice¹³, and their age-matched, wild-type controls. High-content screening was performed on a BD Pathway 855 system. UBE3A-YFP was detected for drug screening with an anti-GFP antibody (Novus Biologicals, NB600-308; 1:1000) because intrinsic YFP fluorescence levels were low in cultured neurons and tissue sections. All data are presented as mean ± s.e.m., with sample sizes (n) shown in figures or stated in text. Statistical analyses were performed using SigmaPlot 11 (Systat Software). Normality tests (Shapiro-Wilk) and equal variance tests were run and passed (P>0.05) before parametric statistical analyses were run.

Full methods accompany this paper as supplementary material.