Dissociating the Effects of Light at Night from Circadian Misalignment in a Neurodevelopmental Disorder Mouse Model Using Ultradian Light–Dark Cycles

A significant proportion of individuals with neurodevelopmental disorders (NDDs) experience disturbances of their daily sleep–wake cycles. Common complaints include delayed bedtimes and frequent nighttime awakenings [1,2,3,4]. These difficulties are associated with challenging daytime behaviors for the impacted individuals [5,6] but also negatively affect the sleep and overall health of their parents and/or caregivers [7,8]. Additionally, altered sleep patterns may lead to increased nighttime exposure to light from electronic screens [9,10,11], which even in young people without NDDs has been shown to delay sleep onset [12,13,14,15,16].

This led us to investigate whether nighttime light exposure could disrupt circadian timing and exacerbate NDD-associated symptoms in a transgenic mouse line lacking the Contactin-associated protein-like 2 (Cntnap2) gene. This gene encodes CASPR2, a member of the neurexin family of transmembrane proteins [17], with key roles in brain development, synapse formation and function. Mutations in the CNTNAP2 gene predispose to NDDs, including autism spectrum disorders (ASD); in particular, individuals with biallelic mutations develop a recognizable syndrome (often called CASPR2 deficiency disorder) with early-onset epilepsy, severe language impairment, intellectual disabilities, and autistic features in a substantial subset [18,19,20,21,22]. Mouse models also show epilepsy and autistic-like deficits [23,24]. Importantly for the present study, prior work has shown that Cntnap2 KO mice show disruptions in circadian locomotor activity [25,26] and EEG-defined sleep patterns [27]. The validity of the Cntnap2 KO mouse model is supported by intervention studies showing that: the FDA-approved drug risperidone reduces repetitive behaviors [23], administration of oxytocin enhances social behaviors [24], treatment with melatonin improves sleep–wake rhythms [26].

We have previously shown that exposing Cntnap2 KO mice to dim light at night (DLaN) exacerbated the already abnormal locomotor activity and altered the neural activity in the suprachiasmatic nucleus (SCN), as well as the rhythms in clock gene expression as measured by PER2::LUC [26]. Notably, DLaN elicited excessive grooming behavior otherwise not observed in these mutants during the night [26,28]. These aberrant effects were mitigated by shifting to a long-wavelength (red) light [28]. An unresolved question is whether the detrimental effects of nighttime light exposure are due to light's direct impact on behaviors or to the circadian disruption driven by DLaN. One approach to investigate this issue is to place the animals in an ultradian light–dark (LD) cycle that falls outside the circadian clock's entrainment range. Prior work by Hattar and colleagues employed an ultradian 3.5 h light/3.5 h dark (T7) cycle to pinpoint the effects of light alone [29,30,31]. They reported that while the T7 cycle lengthens the circadian period, it does not disrupt internal rhythmicity of the SCN or cause arrhythmicity in sleep architecture and body temperature. The T7 cycle, however, did induce mood disturbances without increasing anxiety, suggesting that nighttime light alone can affect mood.

In the present study, we examined the impact of ultradian light exposure on wild-type (WT) and Cntnap2 KO mice housed under T7 conditions (LD 3.5/3.5) in comparison to LD, constant darkness (DD), and DLaN. The impact of the T7 light cycle was assessed on social interactions using the three-chamber test, repetitive grooming behavior, and locomotor rhythms via passive infrared (PIR) detection. Finally, the effects of the T7 lighting on the expression of the immediate early gene cFos were investigated in the basal lateral amygdala (BLA), a region implicated in repetitive behaviors, since two weeks' exposure to DLaN significantly increased the expression of this neural activity marker in this brain region in both the WT and mutants [28].

Social deficits, a key symptom of NDD, have been reported in both juvenile [23] and adult [26,27,28,29] Cntnap2 KO mice, and are exacerbated by exposure to DLaN [26,28]. To untangle the direct effects of exposure to light at night on behavior from those elicited by the circadian disruption, social interactions were assessed in WT and Cntnap2 KO mice held for two weeks under one of four lighting conditions: LD, DLaN, T7, and DD (Figure 1). Mice in LD or DLaN, but not under T7 and DD, are synchronized to the environment, whilst those in DD exhibit a free-running rhythm driven by the endogenous clock.

The animals were tested in the three-chamber social arena during the active phase between Zeitgeber time (ZT) 17 and 19 if in LD or DLaN, or their subjective night (Circadian time (CT) 17–19) if in DD or T7 (Figure 1). The time spent in the chamber with either the novel mouse or the inanimate object was measured (Figure 2A,B) to determine their individual preference, and the resulting data analyzed by two-way ANOVA (Table 1). As previously reported [24,28], both the WT and mutants were significantly impacted by DLaN, with the WT exhibiting a significant reduction in the time spent with the novel mouse (p = 0.006), while neither T7 (p = 0.400) nor DD (p = 0.257) altered their social behavior compared to their counterpart in LD (Figure 2A,B). Similarly, in the Cntnap2 KO mice, DLaN amplified the social deficits by reducing the time spent with the novel mouse (p = 0.002) as compared to the mutants in LD, whilst this effect was not observed in mice held in T7 (p = 0.879) or DD (p = 0.986) (Figure 2A). In comparison to the WT, the Cntnap2 KO mice exhibited fewer social interactions regardless of the lighting conditions (LD: p < 0.001, DLaN: p < 0.001, T7: p = 0.049, and DD: p = 0.033); nonetheless, only DLaN significantly lessened sociability in both genotypes (Figure 2A,B).

Another hallmark symptom of NDDs is repetitive behavior, which is recapitulated in the Cntnap2 mutants in the form of excessive grooming during the day [23] and night [26,28]. The time spent grooming was measured during the night (ZT 16–18) or subjective night (CT 16–18) (Figure 1). Significant effects of genotype and lighting cycle, as well as a significant interaction between the two factors, were revealed by two-way ANOVA (Table 1). In general, the WT mice spent a limited amount of time grooming (Figure 2C), which was not altered by DLaN (p = 0.750) or T7 (p = 0.339). The Cntnap2 KO exposed to DLaN exhibited a dramatic increase in grooming (p < 0.001), but a similar augmentation was not seen under T7 (p = 0.907) (Figure 2C) as compared to mutants in LD. Despite the mutants exhibiting more grooming under each lighting condition (DLaN, p < 0.001; T7, p = 0.001; DD, p < 0.01) as compared to their WT counterparts, the aberrant effects of DLaN were significantly greater in comparison to the other three Cntnap2 KO groups (Figure 2C).

The Cntnap2 KO mouse model exhibits altered diurnal activity rhythms, including reduced nighttime activity [25,26,27,28]. To further characterize the effects of nighttime exposure to light, we assessed locomotor activity rhythms using passive infrared (PIR) sensors under each lighting condition. At least 10 consecutive days of activity data were collected per animal, and diurnal and circadian parameters were derived. As shown by the representative actograms (Figure 3A), both WT and KO mice exhibited rhythms with periods longer than 24 h when held on the T7 cycles (Table 2). While not a focus of the present study, we observed that both genotypes showed longer free-running periods when assessed by PIR rather than by wheel-running (Supplementary Figure S1; see also [32]). Main effects of genotype and/or lighting conditions were identified by two-way ANOVA on total activity, period length, rhythm power, and onset variability, along with a significant genotype × lighting interaction for both period and power (Table 2).

In WT mice, DLaN significantly reduced the power of the rhythms (p = 0.001), with no significant changes in other parameters; the T7 cycles had a broader impact, significantly increasing the period (p < 0.001) and onset variability (p < 0.001) whilst also reducing rhythm power (p < 0.001) relative to those in LD and/or DD (Figure 3B and Table 2). In the Cntnap2 KO mice, DLaN did not alter the power of the rhythms (p = 0.07) but significantly increased the onset variability (p = 0.037; Figure 3B), while T7 illumination significantly lengthened the free-running period (p < 0.001) and increased the variability of the activity onset (p < 0.001; Table 2 and Figure 3B) as compared to their counterparts in LD or DD. Direct comparisons between genotypes revealed significant differences in total activity (p = 0.035), rhythm power (p = 0.015), and onset variability (p = 0.004) with no changes in period (p = 0.644) under LD, as well as in onset variability (p < 0.001) and period (p < 0.001) under DLaN and T7, respectively. The effect of genotype was less dramatic under DD conditions, where a robust reduction in power (p < 0.001) was detected (Figure 3B and Table 2) with no changes in the other parameters.

Finally, sex-divergent effects were a striking feature of this dataset (Figure 4). While this variable did not seem to influence the social and grooming behaviors (Figure 4A), significant effects were observed for total activity, period, power, and onset variation (three-way ANOVA with genotype, lighting cycles, and sex as factors, Figure 4B and Table 3).

Overall, the Cntnap2 KO mice exhibited weaker locomotor activity rhythms, revealing robust genotype-specific effects of T7 illumination. Furthermore, sex-dependent differences emerged as a prominent feature, underscoring the importance of including both sexes in circadian rhythm studies involving ASD models.

Evidence suggests that the BLA, located in the temporal lobe of the cerebral cortex, is a critical mediator of aberrant repetitive behaviors [33,34]; in addition, we have previously shown that DLaN elicits increased expression of cFos, a marker of neural activity, in a population of glutamatergic neurons in this nucleus [28] in both WT and Cntnap2 KO mice, which correlated with the observed behavioral changes. Therefore, we examined its expression in the BLA of mice held in the four lighting conditions (Figure 5). A significant effect of the lighting cycles, but not of genotype, was present (Table 4), with both WT and Cntnap2 KO mice exposed to DLaN displaying a significant increase in the number of cFos-positive cells as compared to their counterparts in LD (p = 0.0002 and p < 0.0001, respectively; Figure 5 and Table 4), whilst neither the T7 cycle nor DD altered the number of cFos positive cells. In general, both WT and mutants held in T7 or in DD presented with a lower number of positive cells as compared to their counterparts in LD, with a significant difference between WT in LD and DD (p = 0.0283 WT in DD vs. WT in LD). Notwithstanding the well-documented effect of LD and DLaN on cFos expression in the BLA in our previous work [28], we included these groups in the present study for a more accurate evaluation of the effects observed, hence the small sample size. These findings are consistent with the behavioral evidence suggesting that nighttime light exposure by itself is not sufficient to activate BLA neurons and elicit the aberrant effects observed.

Patients with NDDs, including ASD, often present with delayed sleep onset, fragmented sleep, and blunted circadian rhythms [1,2,3,4,5,6]. We hypothesize that a circadian disruption not only contributes to the core symptoms of NDDs but also renders affected individuals more vulnerable to its adverse effects. This vulnerability suggests a potential benefit from circadian-based therapeutic interventions.

To investigate this hypothesis, we employed the Cntnap2 KO mouse, a well-established model of ASD and related NDDs [35]. Our prior work has shown that these mutants exhibit abnormal activity and sleep rhythms, altered neural activity in the SCN, and exaggerated sensitivity to environmental circadian disruption. Furthermore, exposure to DLaN triggers repetitive behaviors, otherwise absent in the Cntnap2 KO during the night, aggravates their social deficits [26,28], and worsens the altered SCN neural activity and PER2::LUC-driven rhythms [26]. Notably, administration of melatonin restores rhythmicity and mitigates behavioral abnormalities, with the greatest improvement observed in animals displaying the most robust circadian rhythms.

In the current study, we examined the effects of a non-24 h ultradian lighting schedule—specifically, a T7 cycle (3.5 h light/3.5 h dark)—on behavior and activity rhythms in Cntnap2 KO and WT mice. The T7 cycle exposes the animals to light during their active phase each circadian cycle, and both genotypes exhibited free-running rhythms with periods longer than 24 h. Consistent with previous reports [23,26], the mutation alone triggered reduced social interactions but not abnormal grooming during the night/active phase; strikingly, Cntnap2 KO mice under T7 did not display a worsening of their behavioral deficits, in contrast to the response to DLaN (Figure 2). Prior work by the Hattar group showed that T7 exposure induces depression-like behaviors (e.g., reduced sucrose preference, increased immobility in the forced swim test) and impairs memory performance in tasks such as the Morris water maze and novel object recognition [29,31]. These impairments were mostly assessed during the subjective day; however, recent findings from Fuchs and colleagues [36] suggest that behavioral outcomes under T7 conditions are phase-dependent, with mood and memory impairments emerging only when mice are tested during their subjective night.

Although prior reports indicated that the T7 cycle does not overtly disrupt the central circadian clock—based on SCN gene expression, body temperature rhythms, and sleep architecture [29,37]—our high-resolution locomotor activity analysis revealed subtle but important alterations (Figure 3). After two weeks in T7, both WT and KO mice maintained their rhythmicity, albeit with significant lengthening of the period and increased variability in the activity onset, consistent with prior work [29,37]. Notably, T7 lighting also reduced the power of the rhythms and increased onset variability of the mice relative to DD. Moreover, we detected robust sex differences in these circadian parameters (Figure 4), emphasizing the importance of sex as a biological variable.

Our findings highlight a key dissociation: while T7 alters the activity rhythms, it does not elicit the same behavioral changes seen with DLaN in the Cntnap2 KO mice, suggesting that light at night alone is insufficient to exacerbate autistic-like behaviors; rather, an interaction between light exposure and circadian misalignment may be required (see also [38]). The differential behavioral outcomes in mice under DLaN vs. T7 provide a valuable framework to dissect mechanistic underpinnings of light-induced pathology. Thus, we conclude that while the T7 lighting does produce some disturbances, DLaN exposure is more behaviorally perturbing.

To identify the relevant neural circuits, we build on previous findings that melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs) are necessary for DLaN-induced effects, since Opn4^DTA mice lacking these cells do not show DLaN-induced changes in locomotor activity [37]. Additionally, DLaN increases cFos expression in glutamatergic neurons of the BLA, a region implicated in social behavior and repetitive grooming [28,33,34]. In contrast, T7 exposure did not elicit a cFos response in the BLA (Figure 5), nor did it exacerbate grooming or social deficits. Future studies using chemogenetic silencing or targeted lesions of BLA cell populations will be instrumental in testing its causal role in mediating the behavioral impact of DLaN.

A key innovation of this study is the use of an ultradian T7 lighting schedule to dissociate the direct effects of nighttime light exposure from the circadian disruption it often induces. Prior research has established that DLaN elicits, alters, and/or amplifies behavioral and physiological abnormalities in both WT and Cntnap2 KO mice. However, it has remained unclear whether these adverse outcomes result from the exposure to light per se or from the misalignment of internal circadian rhythms. By implementing the T7 cycle—comprising 3.5 h of light and dark intervals that fall outside the range of circadian entrainment—we provide evidence that the deleterious effects of DLaN are contingent on circadian misalignment. Despite frequent nocturnal light exposure, mutant mice under the T7 cycle maintained their rhythmicity and did not exhibit excessive repetitive behavior, worsening of the pre-existing social deficits, or elevated cFos expression in the amygdala triggered by DLaN. These findings support the "two-hit" hypothesis of NDDs: a genetic predisposition (e.g., Cntnap2 mutation) primes the neural circuits for dysfunctionality, which can be further amplified by an environmental disturbance or stressor. We propose that circadian disruption can serve as the environmental stressor or "second hit" for those who are vulnerable.

Several limitations should be acknowledged. First, we employed 250 lx illumination for the T7 cycles which was consistent with earlier work. It would be interesting to compare the behavioral outcomes driven by T7 using different light intensities (10–250 lx). Second, we measured both the autistic-like behaviors and cFos expression at a fixed phase (ZT/CT 16–18) and did not carry out a time-series analysis. Third, while our activity rhythm analysis had a sufficient sample size to detect sex differences, this was not the case for the autistic-like behavioral measurements, and the lack of statistically significant sex differences in these measures should be viewed cautiously (Figure 4).

Our findings advance the field by refining the mechanistic understanding of how light at night affects behavior and by introducing ultradian lighting as a novel paradigm for dissecting circadian versus non-circadian influences on neural function in a disease model. Most importantly, they have translational implications for individuals with NDDs, including ASD, who are frequently exposed to light at nighttime from electronic devices and/or artificial lighting. Our results demonstrate that the adverse behavioral and neural effects of light at night in a genetically susceptible model are contingent on circadian disruption, rather than exposure to light alone. This distinction emphasizes the importance of maintaining circadian rhythmicity as a protective factor against environmental challenges in individuals at risk because of their genetic make-up. Interventions aimed at supporting circadian health—such as structured light exposure schedules, consistent sleep–wake timing, or the use of circadian-stabilizing agents like melatonin—may help mitigate the impact of nocturnal light exposure in individuals with NDDs. Furthermore, the ultradian lighting paradigm used here provides a mechanistically informative model for testing the efficacy of such circadian-targeted interventions without introducing the confounding effects of arrhythmicity. Ultimately, these findings support the development of chronobiology-informed strategies to improve behavioral outcomes and quality of life in individuals with NDDs.