Circadian rhythm defects in Prader-Willi syndrome neurons

Summary

Prader-Willi syndrome (PWS) is a neurodevelopmental disorder characterized by a spectrum of symptoms, including developmental delay, intellectual disability, and increased risk of autism. PWS is an imprinting disorder caused by the loss of paternal expression of critical genes in the 15q11.2-q13 region, including MAGEL2, SNRPN/SNURF, and SNORD116. PWS patients often suffer from various sleep disorders, including sleep-disordered breathing and central hypersomnolence. Mouse models of PWS also exhibit disruptions in circadian rhythms and sleep. In cultured cells, Magel2 was shown to regulate the expression of Bmal1 and Per2, two core clock genes involved in the circadian rhythm regulatory process. Here, we investigated the circadian clock function in neurons derived from dental pulp stem cells (DPSCs) of PWS patients and neurotypical controls. To study the circadian rhythms of PWS patients in vitro, we introduced the Per2 promoter-driven luciferase reporter (Per2:luc) to these DPSC cell lines to assess their circadian rhythm by bioluminescence. These Per2:luc cells were differentiated for 4 weeks to mature neuronal reporter cell lines, followed by kinetic measurements of luciferase activity over several days. We observed significant differences in circadian period length between PWS neurons and controls. Moreover, treatment with the small molecule longdaysin effectively lengthened the period length of PWS neurons with a shorter period length, as anticipated based on the mechanism of action of this compound. This work lays the foundation for a deeper understanding of PWS pathophysiology and represents a critical first step toward developing high-throughput assays for drug discovery targeting circadian and sleep dysfunction in PWS.

Prader-Willi syndrome (PWS), a neurodevelopmental disorder resulting from the loss of expression of several genes on paternal chromosome 15, is characterized by developmental delay, hypotonia, and excessive daytime sleepiness. Here, we developed an assay to identify, characterize, and rescue circadian rhythm defects in neurons from individuals with PWS.

Introduction

Prader-Willi syndrome (PWS) is a neurodevelopmental disorder with a constellation of symptoms, including hypotonia, failure to thrive, childhood-onset obesity, and developmental delay.¹ PWS is caused by the loss of expression of critical genes in the 15q11.2-q13.1 region on the 15^th chromosome (Prader-Willi/Angelman syndrome critical region). Several of these genes are maternally imprinted and only expressed from the paternal allele, including MAGEL2, SNRPN/SNURF, and NDN.¹ Most cases of PWS result from inheriting a paternal chromosome deleted for this critical region. Other cases are caused by inheriting two maternal copies of chromosome 15 or a mutation in the imprinting center that disrupts imprinted expression of these critical genes. PWS can cause a variety of sleep abnormalities, including sleep-disordered breathing (SDB), excessive daytime sleepiness (EDS), and narcolepsy.² Obstructive sleep apnea (OSA) and alveolar hypoventilation are the most common SDB phenotypes reported in PWS. Childhood-onset obesity is a prominent PWS phenotype along with hypotonia and facial dysmorphia (micrognathia and small naso- and/or oropharynx), all contributing to the SDB issues found in PWS.^2,3 Although SDB may be causing EDS, many studies have failed to confirm that EDS is secondary to SDB.^4,5 In fact, even when SDB is corrected, EDS persists in PWS patients, proving that central hypersomnolence is often a primary clinical phenotype.^2,6

PWS is described as a hypothalamic insufficiency disorder, and hypothalamic dysfunction has been implicated in EDS. The hypothalamus regulates rapid eye movement (REM) and non-REM sleep cycles and harbors the suprachiasmatic nucleus (SCN), which regulates circadian rhythms.⁷ The chromosomal region that causes PWS is a critical region that encompasses several genes that may be involved in disrupting circadian rhythms and hypothalamic function. At least three imprinted genes within this critical region (NDN, SNORD116, and MAGEL2) are linked to circadian function either through evidence from mouse models or in vitro studies.^8,9,10,11

Altered circadian rhythm gene expression was found in Ndn-deficient mice.⁸ Lassi et al. found that in heterozygous Snord116 mutant mice, REM sleep was altered, and there were significant differences in body temperature throughout sleep that they attributed to differing metabolic demands caused by circadian pathway issues.¹⁰ In the same mouse model, another group found significant differences in diurnal rhythmic DNA methylation in PWS mice versus their wild-type littermates. They also found that the rhythmic changes in DNA methylation regulated metabolic pathways disrupted at different times in the circadian cycle.¹¹

Kozlov et al. created a Magel2-null mouse that exhibited altered feeding and breeding behaviors and decreased running wheel activity, a measure of daytime wakefulness.⁹ In vitro studies have shown that MAGEL2 interacts with and regulates critical circadian rhythm proteins, including BMAL1, PER2, and CRY1.^12,13 MAGEL2 is both rhythmic and highly expressed in the SCN.^9,14 Schaaf-Yang syndrome (SYS), a disorder with overlapping phenotypes to PWS, including hypotonia, developmental delay, and obesity, results from truncating mutations in the MAGEL2 gene.¹⁵ SYS patients also have a high prevalence of sleep disorders, including OSA (70%–80%), abnormal sleep cycles, and EDS.^16,17 Understanding the molecular mechanisms leading to sleep abnormalities and EDS in PWS is vital to finding treatment options for sleep problems in both disorders, alleviating caregiver burden and enhancing the quality of life for individuals with PWS.

In this study, we investigated whether the dental pulp stem cells (DPSC) that are derived from individuals with PWS exhibit defective cellular circadian rhythms. To measure circadian rhythmicity, we introduced a Per2 promoter-driven luciferase reporter (Per2:luc) into DPSC cell lines and assessed the circadian bioluminescent rhythms over several days. DPSC are multipotent stem cells that can differentiate into various cell types.¹⁸ We have used DPSC cortical-like cultures extensively to investigate disease phenotypes in PWS, Angelman, and Duplication 15q syndromes.^{19,20,21,22,23} Our lab has established a large collection of DPSC lines from the deciduous teeth of PWS subjects (currently, 46 lines). To investigate the in vitro cellular circadian rhythms in neurons from PWS and control subjects, we established DPSC cell lines harboring the Per2:luc reporter for the in vitro assessment of cellular circadian rhythms. We observed a distinct period length phenotype in neurons from a subset of PWS subjects versus neurotypical control subjects. Also, treatment with a small molecule, longdaysin, lengthened the period of the bioluminescent rhythms in PWS neurons, establishing the utility of this system for future therapeutic discovery. The experiments presented here pave the way for cellular and molecular assessment of circadian clock involvement in PWS pathophysiology and are a critical step in developing the cellular model for drug discovery that targets circadian rhythm defects in PWS.

Material and methods

Generation of DPSC lines

Teeth from children with PWS were collected remotely by the parents or caregivers of subjects after confirmation of the underlying genetic diagnosis. Subjects or their proxy provided informed consent for tooth collection and were given instructions for shipping exfoliated teeth to L.T.R.’s lab under a protocol approved by the University of Tennessee Health Science Center (UTHSC) institutional review board. Neurotypical control teeth were obtained from the Department of Pediatric Dentistry and Community Oral Health at UTHSC. As previously described, tooth pulp was collected and cultured to create DPSC lines.²⁴ Passage 2–4 DPSC lines were then stored in a dedicated DPSC repository for future studies. The UTHSC institutional review board approved the collection of cells in the DPSC repository and subsequent molecular studies in the laboratory prior to the start of experiments.

DPSCs were isolated and cultured according to our previously published protocol.²⁴ Briefly, dental pulp was removed from the pulp cavity. The pulp was minced both mechanically and enzymatically using 3 mg/mL dispase II and 4 mg/mL collagenase I. Cells were then seeded with DMEM/F12 1:1, 10% fetal bovine serum (FBS), 10% newborn calf serum (NCS), and 100 U/mL penicillin and 100 μg/mL streptomycin (Pen/Strep) (Fisher Scientific, Waltham, MA) on 12-well plates coated with poly-d-lysine. Once cultures reached ∼80% confluency, they were passaged using TrypLE Express (Fisher Scientific) for expansion.

Creation of Per2:luc rporter lnes

Lentiviral luciferase reporters under the control of Per2 or Bmal1 promoters were produced in A.C.L.’s laboratory, as previously described.²⁵ Concentrated viral particles (1 × 10⁷ plaque-forming units [PFU]/mL) were used to infect DPSCs with an efficiency of ∼70%. Briefly, DPSCs were seeded at 20,000 cells/cm² on a 12-well poly-d-lysine-coated plate. Once wells were ∼50% confluent, concentrated viral particles (1 × 10⁷ PFU/mL) and 5 μg/mL polybrene were added to the cells overnight. After 24 h, viral supernatant was removed and cells were allowed to recover overnight in DPSC media. After recovery, transduced cells were selected for 7 days using the selection marker blasticidin (1 μg/mL). Once transduced cells reached a confluency of ∼80%, they were passaged for neuronal differentiation and banked for circadian studies.

Neuronal differentiation

Stably transduced DPSCs harboring the Per2:luc or the Bmal1:luc construct were seeded on 35-mm poly-d-lysine and Matrigel-coated plates with DMEM/F12 1:1, 10% FBS, 10% NCS, with 100 U/mL penicillin and 100 μg/mL streptomycin (Pen/Strep). Once cells reached confluency (∼80%), neuronal differentiation was performed as previously described but with an increased maturation phase of 4 weeks.^24,26 Briefly, epigenetic reprogramming was performed for 2 days using 10 μM 5-azacytidine (Acros Scientific, Geel, Belgium) in DMEM/F12 containing 2.5% FBS and 10 ng/mL basic fibroblast growth factor (bFGF) (Fisher Scientific) for 48 h. Then, neuronal differentiation was induced by exposing the cells to 250 μM 3-isobutyl-1-methylxanthine (IBMX), 50 μM forskolin, 200 nM 12-O-tetradecanoylphorbol 13-acetate, 1 mM dibutyryl cyclic AMP (db-cAMP) (Santa Cruz, Dallas, TX), 10 ng/mL bFGF (Invitrogen, Carlsbad, CA), 10 ng/mL nerve growth factor (Invitrogen, Carlsbad, CA), 30 ng/mL NT-3 (Peprotech, Rocky Hill, NJ), and 1% insulin-transferrin-sodium selenite premix (Fisher Scientific) in DMEM/F12 for 3 days. At the end of the neural induction period, neuronal maturation was performed in Neurobasal A media (Fisher Scientific) with 1 mM db-cAMP, 2% B27, 1% N-2 supplement, 30 ng/mL NT-3, and 1× GlutaMAX (Fisher Scientific) for 28 days prior to the start of the circadian experiments.

Bioluminescence recording

Before bioluminescence recordings, the neuronal cultures were synchronized using forskolin (50 μM) and IBMX (250 μM) for 2 h.²⁷ After synchronization, neuronal cultures were washed with Dulbecco’s PBS, and 2 mL recording media was added. Recording media is composed of 1 mM db-cAMP, 2% B27, 1% N-2 supplement, 30 ng/mL NT-3, 1× GlutaMAX (Fisher Scientific), and 1 mM luciferin (Biosynth, Staad, Switzerland) in DMEM. For longdaysin treatment, longdaysin was added to the recording media at either 0.5 or 1.0 μM. Cells were then housed in a Biotek BioSpa 8 Automated Incubator (Agilent, Santa Clara, CA) attached to an H1 Synergy (Agilent) plate reader that recorded bioluminescence rhythms at 2-h increments for 5 days.

Data analysis

LumiCycle Analysis software (Actimetrics, version 2.701) was used to determine the period length and create normalized Per2 promoter-driven bioluminescence traces (Figure S2). For circadian parameter analysis, the first 29 h after beginning the LumiCycle recording, a time frame when cells are still synchronizing, were excluded. For analysis-only hours, 30–110 were used to determine the period length. The data were fit to a sine wave (dampened), and the period length was calculated using the software, with goodness of fit >80%. The background-subtracted data were exported to Prism 10.0 (GraphPad) to create bioluminescence trace graphs and bar graphs and to determine significance. For the longdaysin rescue experiments using BioSpa 8, relative luminescence units were background subtracted and analyzed in Prism 10.0 (GraphPad). The data were analyzed using non-linear fit to a sine wave with non-zero baseline with 95% confidence interval to determine the wavelength of each individual. A smoothing function was performed to resolve the “dampened” sine wave of the Per2:luc traces over time.

Results

Rhythmic circadian cycling revealed after neuronal differentiation

The bioluminescence rhythms from undifferentiated DPSC Per2:luc lines were measured using a LumiCycle device to assess the expression of Per2:luc rhythms over several days.²⁵ A diagram depicting the assay for both the LumiCycle and BioSpa is shown in Figure 1. Per2:luc control DPSC lines were grown to confluency, synchronized using forskolin (50 μM) and IBMX (250 μM) as previously described,^27,28 and placed in the LumiCycle to monitor the expression of luciferase rhythms. In undifferentiated DPSC lines, the bioluminescence expression across time was not rhythmic. CON190, CON238, and CON302 samples showed low levels of non-rhythmic bioluminescence expression (Figure 2A). The same Per2:luc expressing DPSC lines were then differentiated into neuronal cultures and matured for 4 weeks before LumiCycle recording. After neuronal differentiation, maturation, and synchronization, all lines showed stable, robust circadian bioluminescence rhythms over the 5 days of the recording period (Figure 2B). These results suggest that the neuronal cell lines exhibit cell-autonomous circadian oscillation of bioluminescence rhythms, which are present in neuronally differentiated stem cells but not in undifferentiated DPSC.

Figure 1.

Schematic for measuring Per2:luc cycling in neuronal samples

(A) Neurons are differentiated on 35-mm or 96-well plates over 4 weeks. Neurons are chemically synced for 2 h with forskolin and IBMX. The neurons are then treated with longdaysin prior to transfer to the LumiCycle or BioTek BioSpa 8 systems. Measurements are taken every 15 min in the LumiCycle and every 2 h for the BioSpa 8 for 5 days.

(B) Representative traces of bioluminescence readouts for Per2:luc cycling in untreated (black) and longdaysin (blue)-treated neurons. (Created with BioRender.com.)

Figure 2.

Rhythmic Per2:luc circadian cycling after neuronal differentiation

(A) Undifferentiated DPSCs do not show bioluminescent rhythms.

(B) After neuronal differentiation and maturation, DPSC-derived neurons display rhythmic expression of Per2-promoter driven luciferase. Each colored trace represents a neurotypical control subject’s individual DPSC-derived cell line.

The first 30 h are not considered in the analysis due to the acute effects of the IBMX/forskolin treatment on circadian rhythms. Line graphs were generated in GraphPad Prism 10.0 using the background subtracted counts/second generated within the LumiCycle software.

Neurotypical controls and PWS neurons display expected anti-phasic expression for Per2- and Bmal1-promoter driven luciferase

Per2 and Bmal1 negatively regulate each other’s transcription in the circadian cycle.²⁹ To confirm that these genes are regulated appropriately in both the PWS disease state and control neurons, in addition to the Per2:luc DPSC lines, we generated Bmal1:luc stable DPSC lines to measure the cyclic expression of Bmal1-promoter driven luciferase rhythms over several days. The rhythmic expressions of luciferase rhythms are anti-phasic in both control (Figure 3A) and PWS neuronal samples (Figure 3B). When Per2:luc expression is at its peak, Bmal1:luc is at its lowest expression level during the circadian cycle.³⁰ These results support that the circadian transcriptional negative feedback mechanism functions properly in both PWS and control DPSC-derived neurons and underlie the cellular bioluminescence rhythms in these cells. These results also suggest that we can use either Per2:luc or Bmal1:luc to measure circadian oscillations from these cells.

Figure 3.

Neurotypical control and PWS neurons show anti-phasic Per2 and Bmal1-promoter-driven luciferase expression

Per2:luc (blue trace) and Bmal1:Luc (red trace) are clock-controlled circadian genes that negatively regulate each other. Per2-and Bmal1-promoter-driven luciferase shows the expected anti-phasic expression across time in neurotypical control subjects (A) and PWS subjects (B).

The first 30 h are not considered in the analysis due to the acute effects of the IBMX/forskolin treatment on circadian rhythms. Line graphs were generated in GraphPad Prism 10.0 using the background subtracted counts/second generated within the LumiCycle software.

PWS neurons exhibit shortened period length

Having established Per2-and Bmal1-promoter driven luciferase cycling in both PWS and control samples (Figure 3) with expected phase relationships, we chose to characterize circadian rhythms using only Per2:luc stably expressing cell lines for the remainder of the study. Table 1 details the demographics and average period length for all subjects. For these studies, five unrelated control and six PWS Per2:luc reporter lines were differentiated into neurons and matured for 4 weeks. After maturation, IBMX (250 μM) and forskolin (50 μM) were used to synchronize circadian cycling, and the neurons were placed into the LumiCycle to monitor Per2-promoter driven luciferase expression for 5 days. Based on the period lengths and bimodal distribution of the PWS samples (Figure S1), we were able to classify PWS lines into two subsets: those with no significant period length defect (PWS-no defect) versus neurotypical control lines and PWS lines with a significantly shortened period length versus control. Figure 4 shows the average cycling traces for these individuals. The trace depicted for each individual subject represents the average of at least three technical replicates. Each individual trace (normalized and smoothed within the LumiCycle software) is shown in Figure S2. The period length calculations and bioluminescence traces did not consider data collected before 30 h due to the high levels of acute Per2:luc expression in all cell lines caused by synchronizing agents prior to the start of the recording (Figure S2). In the study presented here, it was not possible to assess amplitude differences between the groups due to variability in cell numbers following the neuronal differentiation process. Four control samples (CON189, CON195, CON238, and CON302) (Figure 4A) have overlapping Per2 cycling, while one control line (CON190) deviated in the timing of peak Per2:luc expression (∼5 h) while still maintaining a similar period length to other control lines (Table 1). The mean period length for the control group was 27.4 h. The PWS subjects display either a shortened period length (Figure 4B) or a period length that is not significantly different from control subjects (Figure 4C). Table 1 lists the average period length across the replicates for all lines used in the study. PWS148, PWS152, and PWS297 lines showed a shorter period length phenotype, with an average of 26.5 h. PWS104, PWS192, and PWS198 showed a longer period length, with an average period length of 27.9 h. Figure 4D depicts the average bioluminescence rhythm in one representative subject per group to better illustrate the differences between Per2:luc expression timing within each phenotype. In addition to the differences in average period length, PWS lines and their replicates showed wide variation in the timing of their Per2:luc peaks and troughs regardless of period length classification. When quantified, using unpaired t-tests, the PWS-short phenotype versus control data was statistically significant (p < 0.0001) (Figure 4D). While some PWS subjects had, on average, longer period lengths than control individuals, this subset did not reach statistical significance (p = 0.1794). Additionally, the PWS-short phenotype group has a significantly shorter period length than the PWS-no defect group. These results confirm, however, that the circadian cycle in PWS neurons is disrupted compared to control neurons and that in our cohort, PWS individuals can be classified into two groups, with the shorter period length phenotype being significantly different between both the control and PWS-no defect individuals.

Table 1.

Characteristics and identifiers of subjects

Identifier	Sex	Age, years	Genotype	Average period length ± SEM
PWS104	female	6.4	PWS deletiona	27.4 ± 0.6
PWS148	male	5.8	PWS deletiona	26.8 ± 0.6
PWS152	male	5.4	PWS deletiona	26.3 ± 0.1
PWS192	female	13.3	PWS deletion (type II)	28.7 ± 1.3
PWS198	male	8.0	PWS deletion (type II)	27.9 ± 0.1
PWS297	male	5.6	PWS deletion (type II)	26.7 ± 0.1
CON189	female	8.4	neurotypical control	28.2 ± 0.7
CON190	male	2.6	neurotypical control	27.7 ± 0.2
CON195	male	4.4	neurotypical control	27.0 ± 0.1
CON238	male	8.3	neurotypical control	27.2 ± 0.1
CON302	male	9.9	neurotypical control	27.2 ± 0.2

^aGenetics report does not specify type I or type II deletion.

Figure 4.

PWS neurons show two distinct Per2-promoter-driven luciferase period length phenotypes

(A–C) Per2:luc period length across time was measured in neurotypical control (A) and PWS (B and C) neurons. PWS neurons display two distinct period length phenotypes: short (B) and close to control (no defect) (C).

(D) A representative subject per group depicts the variance in period length timing across the three groups.

(E) Period length for each group was quantified in LumiCycle Analysis software. Each trace depicted represents the average count/second for ≥3 replicate traces per individual. Each data point shown in (E) is represented within the average bioluminescence trace diagram shown in (A)–(C).

The first 30 h are not considered in the analysis due to the acute effects of the IBMX/forskolin treatment on circadian rhythms. Line graphs were generated in GraphPad Prism 10.0 using the background subtracted counts/second generated within the LumiCycle software. Unpaired t tests were used to determine significance.

Longdaysin modulates the circadian period length of PWS neurons

To correct the circadian rhythm defect in a PWS short subject (PWS148), we treated the cell lines with longdaysin to modulate the period length to control levels (Figure 5). Longdaysin is a purine-based casein kinase I (CKI) inhibitor³¹ known to extend the circadian period. In the circadian rhythm transcription-translation feedback loop, CKI phosphorylates the PER protein, targeting it for degradation. Inhibition of CKI phosphorylation by longdaysin has been shown to lengthen period length as a result of PER stabilization.³² Longdaysin was tested at two concentrations (0.5 and 1.0 μM) and added to the recording media for treatment throughout the recording process in the BioSpa. We used a BioSpa plate-reader assay for the rescue experiment as opposed to a LumiCycle because the BioSpa allows the neurons to remain in an incubator with gas exchange until the plate is being read, while the LumiCycle requires the culture dish to be sealed with no gas exchange for the duration of the 5-day assay, leading to greater cell stress and death. Figures 5A and 5C depict the average bioluminescence traces for untreated and 0.5-μM longdaysin treatment in both control and PWS148. Each trace represents the average of three technical replicates. In control neurons, treatment with 1.0 μM longdaysin significantly increased period length from an average of 24.7–27.5 h (Figure 5A). Longdaysin treatment at both doses resulted in significantly increased period lengths in PWS148 neurons (Figure 5B). In the BioSpa assay, the average period length for untreated PWS148 was 21.0 h. When treated with longdaysin at both concentrations, PWS148 period length was significantly increased to 24.1 and 24.6 h, respectively (Figure 5D). With longdaysin treatment, we were able to rescue PWS148 period length to control levels, with no significant difference between PWS148 treated with longdaysin and untreated control neurons. Here, we demonstrate that the PWS short-period length phenotype can be rescued with small molecules like longdaysin and that this assay can be optimized as a drug-screening platform, not only for PWS but also other syndromes where circadian rhythms are affected.

Figure 5.

Longdaysin modulates period length in control and PWS neurons

(A) Average relative bioluminescence for each treatment group for neurotypical control (CON189). Each trace represents the average of three replicates (n = 3).

(B) The average period length for untreated and longdaysin-treated neurons quantified in using nonlinear curve fitting (GraphPad). The analysis used one-way ANOVA with multiple comparisons to determine significance.

(C) Average relative bioluminescence for each treatment group for PWS148 and untreated CON189. Each trace represents the average three replicates (n = 3).

(D) The average period length for untreated and longdaysin -treated PWS and untreated control neurons quantified using nonlinear curve fitting (GraphPad). The analysis used one-way ANOVA with multiple comparisons to determine significance.

Discussion

PWS is a multi-gene disorder resulting in cognitive and behavioral phenotypes, including developmental delay and disordered sleep. Many PWS individuals have SDB, including OSA. In addition to SDB, PWS patients experience EDS and narcolepsy-like symptoms, which do not resolve when SDB is corrected through ventilatory support. EDS leads to decreased quality of life for both PWS subjects and caregivers. Several genes are typically deleted in PWS, which could disrupt circadian rhythm function. However, MAGEL2 is the only gene within the PWS critical region that is directly controlled by the CLOCK/BMAL1 complex (referred to as clock regulated) and shows rhythmic circadian expression.⁹ While Magel2-null mice entrain to light cycles, they exhibit decreased activity amplitude and increased daytime activity compared to wild-type mice.⁹ In addition, SYS, a syndrome with many phenotypes in common with PWS, including sleep dysfunction, EDS, and SDB, results specifically from truncating mutations in the MAGEL2 gene.^16,17 Here, we used our DPSC-neuron system to assay the bioluminescent rhythms at the cellular level across time as an output measure for aberrant circadian rhythm in PWS. Identifying molecular defects in the circadian rhythm cycling of PWS neurons is the first step in developing therapeutics to treat the symptoms of defective circadian cycling (i.e., EDS and SDB) in PWS and other syndromes with circadian rhythm defects.

We show that in Per2:luc expressing undifferentiated DPSC, circadian cycling is not rhythmic across 5-day recordings and does not overlap between control individuals. However, when these lines are differentiated into neurons, DPSC-derived neurons from the same individuals show precise circadian cycling (Figure 2). Our results are consistent with the work of others showing that different synchronizing methods (forskolin, dexamethasone, or temperature) do not result in rhythmic circadian gene expression in induced pluripotent stem cell cultures³³ or embryonic stem cells until terminally differentiated.³⁴ In our system, following neuronal differentiation and maturation, the circadian transcription-translation feedback mechanism is enabled, leading to rhythmic oscillations of Per2-promoter driven luciferase expression. Additionally, we show the anti-phasic expression of both Per2:luc and Bmal1:luc in PWS and neurotypical control individuals (Figure 3). This confirms that PER2 and BMAL1 in both control and PWS DPSC neurons are functioning as expected, establishing the validity of this DPSC-neuron system for the measurement of circadian rhythm and circadian feedback.

Many lines of evidence point to a role for MAGEL2 in this circadian rhythm transcription-translation feedback loop. One group showed that mouse Magel2 represses Clock:Bmal activity, causing a decrease in Per2 expression, and that Magel2 interacts with both Per2 and Bmal1.¹² Although Cry1 and Per2 are known to inhibit the Clock:Bmal1 complex differently, their expression and localization are tightly coupled with mouse Cry1 translocating Per2 from the cytoplasm to the nucleus, where Cry1 represses Clock and Bmal1 expression.³⁵ MAGE proteins, like MAGEL2, regulate ubiquitination through their interaction with RING E3 ubiquitin ligases and deubiquitinases. Modifying ubiquitin signals acts as a rheostat in maintaining the delicate balance and rapid timing of rhythmic protein expression within the circadian pathway.³⁶ One of the interacting partners of MAGEL2, USP7, stabilizes CRY1 expression through deubiquitination.^13,37 While direct interaction with CRY1 has not yet been established, presumably, MAGEL2 exerts this effect on CRY1 indirectly through its interactions with USP7, finely tuning the cyclical expression of essential clock-regulated genes. Figure 6 illustrates a model we propose for the role of MAGEL2 in regulating the circadian rhythm cycle. Circadian rhythms alter sleep/wake cycles and impact cellular metabolism, making any defect in the circadian pathway essential to overall health.^38,39

Figure 6.

MAGEL2 fine-tunes CRY1 stabilization/degradation in the circadian rhythm pathway

In the transcription-translation feedback loop, BMAL1 and CLOCK bind to promoter elements on CLOCK-regulated genes, including MAGEL2, PER2, and CRY1, to promote their expression. PER2 and CRY1 form a heterodimer. Once in the cytoplasm, these CLOCK-regulated proteins accumulate, and post-translational modifications (purple circles) occur through different mechanisms such as ubiquitination and phosphorylation. The RBX1/SCF complex ubiquitinates CRY1, targeting it for degradation, while USP7 deubiquitinates CRY1, stabilizing the protein. MAGEL2 interacts with USP7, likely acting in an opposing manner to destabilize CRY1. Once translocated back to the nucleus, CRY1 and PER2 inhibit the expression of BMAL1/CLOCK, effectively repressing their own transcription. Created with BioRender.com. Modified from Carias et al.¹³

In patient-derived PWS neurons, we found a subset of PWS subjects with a significantly shortened period length (PWS-short) versus neurotypical control and other PWS lines tested (PWS-no defect). In addition to the difference in period length, there is a noticeable difference in the timing and overlap of Per2:luc expression peaks and troughs in the PWS individuals versus neurotypical control subjects. The variance among the replicates (both technical and biological) (Figure S2) in the PWS individuals points to a discordance in the establishment of synchronous circadian periods. This suggests that the loss of MAGEL2 due to PWS causes the circadian pathway to destabilize and potentially lose the buffering capabilities of the control neurons. Testing circadian patterns in SYS samples, where MAGEL2 is mutated but still translated, could shed some light on how MAGEL2 regulates the circadian cycle as well.¹⁵ Additional investigations into the role of MAGEL2 in this transcription-translation feedback loop will be necessary to determine how these MAGEL2-mediated stabilization mechanisms work.

As MAGEL2 has been shown to alter turnover rate, we propose that in PWS, loss of MAGEL2 dysregulates the period length and Per2:luc expression timing, causing the differences we found. One explanation for the shorter period length of the PWS group is that MAGEL2 may affect the RBX1 complex, which ubiquitinates CRY1 and targets it for degradation. The loss of MAGEL2 expression in PWS neurons may diminish the ability of the cell, therefore, to fine-tune the transcription-translation feedback loop in the circadian pathway, destabilizing this pathway and leading to aberrant period lengths. A model for the role of MAGEL2 in this process is presented in Figure 6.

Due to the significant difference in the PWS-short phenotype individuals, longdaysin, a CKI inhibitor, was chosen given its established period-lengthening effects.³¹ After treating PWS-short phenotype cultures with different concentrations of longdaysin, it was evident that period length, in this assay, can be rescued with this drug and that longdaysin increases circadian period length to control levels (Figure 5). Ultimately, using this assay to screen US Food and Drug Administration-approved drugs currently being used to treat EDS, such as pitolisant,⁴⁰ will provide important information regarding the efficacy of these treatments and how these drugs modulate the circadian transcription-translation feedback loop.

Although observing circadian function in primary neurons derived from PWS individuals provides a novel and valuable tool for therapeutic discovery, this work has some limitations. A larger-scale study using more PWS and control individuals will be necessary to determine the frequency and subtle characteristics of the PWS period length phenotypes. Additionally, in both the PWS and control groups, the average period length of females was longer than the males tested. A larger-scale experiment would help clarify any gender-specific differences. Observing circadian rhythms in SYS individuals, who have truncating mutations in MAGEL2, will provide insight into the role of MAGEL2 in period length modulation and may reveal specific domains involved in CRY1 stability. Similarly, restoring MAGEL2 in PWS neurons and abolishing MAGEL2 in control neurons before circadian analysis will reveal the contribution of MAGEL2 to the circadian rhythm transcription-translation feedback loop.

Although currently not within the scope of the studies described here, correlating the period length bioluminescent trace data with clinical findings, sleep studies, and reported sleep issues from individual PWS subjects will be essential to connect the phenotypes described here to actual clinical measures of EDS in PWS. Using both the LumiCycle and the BioSpa, the average period length of neurotypical control subjects is longer than the expected period length in human cells such as cardiomyocytes and fibroblasts.³⁴ This difference may arise from the developmental immaturity of these cultures. In fact, in the first weeks of life, newborns do not display rhythmic sleep patterns until 5 weeks of age, when they establish a circadian rhythm of ∼25 h.⁴¹ The finding presented here that DPSC do not display rhythmic circadian cycling until terminally differentiated lends credence to the hypothesis that circadian cycling could change as the neurons become more mature. Repeating these cycling experiments at time points during differentiation and at more mature stages will be important to answer this question. Notably, the average period length in both control and PWS lines was shorter when performing the BioSpa plate-reader assay than with the LumiCycle assay. This is likely due to differences in the recording increments (15-min increments versus 2-h increments) and stress on the cells. In the LumiCycle assay, the culture dish is sealed with no gas exchange, leading to homeostatic stress, while the BioSpa is an automated incubator that allows the neurons to stay in their preferred culture conditions until the plate is being read. Moving forward, the BioSpa plate-reader assay will allow for a high-throughput and consistent method to measure circadian rhythms in our system.

Conclusion

In this study, using Per2:luc reporter DPSC lines differentiated to neuronal cultures, we assayed Per2:luc expression across time as an output of circadian rhythm function. We identified, for the first time, defects in circadian cycling in primary neurons from individuals with PWS. We also found that neuronal differentiation is required for rhythmic Per2:luc cycling to be established in both controls and PWS neurons. In both PWS and neurotypical control neurons, the transcription-translation feedback loop regulating Bmal1:luc and Per2:luc in the circadian pathway oscillates as expected. Importantly, we found two distinct period-length subsets in PWS neurons and determined that the shorter PWS phenotype significantly differs from both neurotypical control and PWS-no defect individuals. Finally, we were able to modulate the circadian rhythms observed in this novel assay using the small-molecule drug longdaysin, establishing the utility of this assay for therapeutic discovery. We hypothesize that the loss of MAGEL2 in PWS and mutations in SYS will affect the ability of the cell to fine-tune the turnover of key circadian rhythm proteins, particularly CRY1. Establishing this assay and observing circadian rhythm defects in PWS neurons paves the way for future mechanistic studies and drug discovery to rescue circadian cycling in PWS, and possibly SYS, improving the quality of life for patients and caregivers.

Acknowledgments

The authors thank the patients and their families for contributing teeth to these studies. We thank the Foundation for Prader-Willi Research for supporting this work through their pilot grant program and the Shainberg Neuroscience Fund.

Author contributions

Conception and design, A.K.V., C.R., A.C.L., and L.T.R. Collection and assembly of data, A.K.V. and T.H. Provision of study material, C.R., Y.S., and A.C.L. Collection and assembly of data, T.H. Data analysis and interpretation, A.K.V., T.H., and L.T.R. Manuscript writing, A.K.V., T.H., C.R., Y.S., A.C.L., and L.T.R. Final approval of manuscript, A.K.V., T.H., C.R., A.C.L., and L.T.R. Financial support, A.C.L. and L.T.R.

Declaration of interests

The authors declare no competing interests.

Published: March 1, 2025