Light at night disrupts diel patterns of cytokine gene expression and endocrine profiles in zebra finch (Taeniopygia guttata)

Nearly all vertebrate species on earth synchronize their daily and annual activities based on information from the sun¹. This daily alignment of behavior and physiology to photic cues is regulated by the response of circadian machinery to light:dark (LD) cycles. In mammals, the master circadian clock resides in the suprachiasmatic nucleus (SCN) of the hypothalamus². Among birds, a more complex circadian system of three pacemakers viz. the avian equivalent of the SCN in hypothalamus, pineal gland, and the retina regulate daily timekeeping³.

The worldwide increase in artificial lighting in the past century had led to alterations in circadian- and seasonally-controlled behavior and physiology of many vertebrates, including birds^4–6. For example, in songbirds, artificial light at night (ALAN) and constant exposure to light (LL) have been shown to impact singing, sleep, cognitive performance, personality, and reproduction^7–11. More specifically, great tits, Parus major, when exposed to dim light at night (DLAN), sleep less and spend less time at nest boxes⁷, while Indian house crows, Corvus splendens, exposed to LL spent significantly more time to correctly perform memory tasks¹⁰. Timing of circannual gonadal recrudescence in spotted munia, Lonchura punctulate, is also altered in LL and under aberrant LD cycles of 3.5 h light: 3.5 h darkness¹¹. Alterations in avian circadian clocks are also reported in response to ALAN¹² and LL¹³.

An emerging finding in the mammalian literature is that the immune system is also influenced by circadian function and thus susceptible to ALAN¹⁴. Similarly, several avian studies have demonstrated susceptibility of immune responses to circadian dysfunction and sleep loss^15,16. For example, while ALAN increased oxidative stress in great tit nestlings¹⁶, exposure to extended hours of light and LL suppressed several measures of adaptive immunity (humoral and cell-mediated immunity) in broiler chickens, Gallus gallus¹⁷ and quail, Coturnix japonica¹⁸. Other studies have shown effects of circadian disruption upon innate immune function, namely inflammatory markers, as measured by pro-inflammatory cytokine gene expression in brain and peripheral tissues. More specifically, sleep loss increased pro-inflammatory cytokine gene expression in brain, adipose tissue, and spleen of zebra finches, Taeneopygia guttata¹⁹. Cytokine gene expression in the avian spleen, a commonly used indicator of immune responses²⁰, was significantly altered under conditions of LL and melatonin supplementation²¹. Furthermore, the avian spleen displayed rhythmic oscillations of cytokine mRNA abundance over a 24 h cycle as well as a daily rhythm in response to lipopolysaccharide (LPS)-induced immune challenge when applied at midday vs. midnight²¹. An upregulation of proinflammatory cytokines in adipose tissue from sleep disturbances or LD cycle perturbations may be attributed to the infiltration of classical proinflammatory M1 macrophages into adipose tissue²². In brain, light-induced changes in expression levels of cytokines can affect several aspects of normal central nervous system functioning, including sleep, synaptic pruning, and regulation of a variety of neuroendocrine functions in addition to neuroinflammation²³. In general, brain pro-inflammatory cytokines approach maximum levels at night during periods of increased slow-wave sleep²⁴. Further, Kupffer cells in liver are also able to synthesize a variety of cytokines and an upregulation of hepatic pro-inflammatory response can have systemic effects on other organs or can alter hepatic metabolic function through paracrine action²⁵.

The effects of circadian disruption upon immunity could be mediated through several endocrine pathways that include melatonin, thyroid hormones, testosterone, glucocorticoids, and their interactions²⁶. Melatonin is an indolamine primarily secreted by the pineal gland at night and suppressed by light during the day²⁷. Melatonin is typically immunoenhancing in birds, as indicated by studies in Japanese quail²⁸, jungle bush quail (Perdicula asiatica^29,30) and chicken³¹. Partially-abolished diurnal rhythms in IL-6 and IL-18 mRNA expression under LL were restored by melatonin supplementation in chickens³¹. On the other hand, stimulatory effects of melatonin on cellular and humoral immune response in quail are dependent on opioids¹⁸, while modulation of seasonal immunity in Indian tropical birds, Perdicula asiatica, involves an interaction between melatonin and sex steroids with melatonin enhancing and sex steroids suppressing immunity³². In addition, plasma corticosterone (cort) concentration follows a daily rhythm with the peak close to sunrise in diurnal species, such as Gambel’s white crowned sparrows (Zonotrichia leucophrys gambelii)³³. In contrast, nocturnal bats exhibit peak plasma cort concentrations around sunset³⁴. Free-living great tits breeding in LAN had higher cort levels than those breeding under control sites³⁵. Further, effects of LAN on glucocorticoid responses can be light temperature-dependent. Zebra finches exposed to 5000 K LAN had increased corticosterone levels compared to finches exposed to 3000 K³⁶. Exposure to LL or LAN may alter the immune response through activation of the hypothalamic-pituitary-adrenal (HPA) axis²⁶. Acute elevation in cort levels in response to short-term stress can enhance immune function, while chronic activation of the HPA axis may have opposite effects³⁷.

While many mammalian studies report diurnal rhythms in immune responses, including expression profile of cytokines in brain and peripheral tissues^37,38, less is known in birds. Recent studies involving LAN protocols offer clues to circadian regulation of immune response in birds; however, the effects of circadian disruption on daily oscillations in immune function are largely unknown. The aims of this study are to (a) elucidate the diurnal variation of peripheral and neural expression of cytokine genes over the 24-hour cycle, and (b) to test whether 10-days of exposure to constant light or DLAN affects rhythmicity of cytokine gene expression, and/or the profiles of key endocrine regulators of immunity. Given the varied functions of cytokines across tissues^22,23,25, we hypothesize that diurnal variation in cytokines will follow a tissue-specific expression pattern and that these diurnal waveforms will be susceptible to light-dark cycle perturbations. We also hypothesize that cytokine gene expression will peak mostly during the day in various tissues to coincide with increased activity and higher disease risk, although some cytokines in brain could peak at night to modulate sleep. We also predict that endocrine profiles will be affected by different light treatments and could potentially modulate rhythmicity of cytokine gene expression. For example, if constant light and DLAN treatments suppress melatonin rhythms and/or alter the amplitude of cort rhythms on a chronic level, then we would predict disruption of constitutive rhythms of cytokine gene expression. Given previous studies, we also predicted that DLAN would disrupt some (but not all) of the diel rhythms in immune function as opposed to complete abolishment of rhythms from constant light.

There is accumulating evidence that prolonged exposure to light-at-night (LAN) leads to negative health and fitness consequences^6–11. In this study, we tested the hypothesis that exposure to LAN alters the daily patterns of cytokine gene expression in addition to key hormones implicated in the regulation of immunity. Our results demonstrated that LAN influences the diel patterns of cytokine gene expression and blood levels of melatonin and corticosterone, along with locomotor activity (cf. Suppl. Fig. 1) in a diurnal avian species. Taken together, our results suggest a more severe effect of bright LAN than dim LAN on tissue-specific daily rhythms of immunity in zebra finches.

The night hormone melatonin accurately encoded the duration of dark (and hence light) with higher night levels than day in zebra finches under 12 L:12D. A peak of melatonin rhythm at ZT 19 is consistent with a mid-night peak in captive zebra finches^3,39. Furthermore, a significant reduction in night-time melatonin levels to near-day time levels along with an arrhythmic profile under LLbright was expected given the light sensitivity of the melatonin biosynthesis pathway²⁷. Thus, an absence of a daily rhythm in melatonin under LLbright was reflective of a lack of physiological differentiation between day and night and hence a state of circadian arrhythmicity^39,40. Interestingly, the DLAN-induced suppression of melatonin was not as profound as under LLbright. A reduction of melatonin levels at light/dark transition times has also been reported in European blackbirds (Turdus merula) exposed to DLAN⁴¹. We speculate that similar mid-night melatonin levels under pitch-dark and DLAN, and reduced early- and late-night levels under LLbright and DLAN, perhaps indicate that the synchronization of daily rhythms depends on the absolute light intensity as well as the relative interpretation of the phases of illumination based on the photophase contrast^9,40.

On another level, LAN affects the circadian system through modulation of the hypothalamic-pituitary-adrenal (HPA) axis. In this study, a peak in baseline plasma cort levels towards the end of the 24 h day under 12 L:12D was similar to those reported in other avian species, including Gambel’s white-crowned sparrows³³, starlings (Starnus vulgaris⁴²) and tropical Nazca boobies (Sula granti⁴³). This end-of-night cort peak possibly anticipates higher energy demands and alertness associated with the upcoming active phase for diurnal finches^44,45. However, end-of-night cort levels in DLAN were significantly lower than those under pitch dark, which was contrary to our expectation of finding elevated cort levels, perhaps owing to circadian disruption, sleep disturbances or restlessness under LAN^7,9,35. This absence of a night-time elevation in cort under DLAN is consistent with those reported in mammals⁴⁶, but not in free-living birds³⁵. This discrepancy could be attributed to a number of factors that include variation in light intensity, species differences, and/or comparing birds from captive and field studies. However, under aperiodic cycles, zebra finches, much like great tits, showed an elevated baseline cort response³⁵, along with a non-normal, arrhythmic diel pattern as reported in mice⁴⁷. Thus, 10 days of exposure to ALAN, both bright and dim in intensity, could significantly disrupt diel patterns and alter baseline levels of two important hormones that regulate immunity.

Similar to the endocrine axis, key parameters of the immune system, including cytokines have been shown to exhibit circadian rhythms in mammals^48,49 and to a lesser degree in birds²¹. Zebra finches entrained to 12 L:12D exhibited significant daily rhythms in transcripts encoding IL-1β and IL-6, the pro-inflammatory cytokines which function as key regulators of the early immune response to infection⁵⁰ and for IL-10, an anti-inflammatory cytokine which dampens inflammatory responses⁵¹ (cf. Figures 2 and 3). Peak expression of pro-inflammatory cytokines during the light phase or in alignment with the dark-to-light transition (eg. IL-1β in fat) in peripheral tissues, and a parallel reduction in anti-inflammatory cytokines (except in fat) is conceivable due to the increased demand of immunity during light hours when birds have a higher risk of exposure to infections when feeding, exploring, and interacting with other individuals^21,52. However, microglia, the primary innate immune cells of the brain, exhibit diurnal rhythms in inflammatory potential with a peak response generally occurring during the inactive phase of the organism^38,53. Similar to previous reports, pro- and anti-inflammatory cytokine gene expression in the brain mostly peaked during the dark phase under 12 L:12D with an early night peak of IL-1β in hippocampus, a late night peak of IL-6 and IL-10 in nidopallium and hypothalamus, and an end-of-the-night peak of IL-6 in hippocampus (cf. Table 1). Although not established in the current study, this may suggest a sleep regulatory function of neural cytokines⁵⁴ in zebra finches. Further, unlike in rats, where IL-1β mRNA peaked just after lights–on in cortex, hippocampus and hypothalamus⁵⁵, we did not find a significant rhythm in IL-1β expression in nidopallium and hypothalamus, similar to the absence of day-night variation recently reported in zebra finches¹⁹. However, we would like to caution that there is a possibility of expression levels in a small nucleus was masked by other nuclei in the brain region, as the mRNA levels were tested in whole hypothalamus, hippocampus and nidopallium.

Given the increasing evidence that neuroendocrine systems regulate immunity and vice versa⁵⁶, we expected an abolishment of diel rhythms in hormone profiles would also induce aperiodic gene expression of cytokines. This was indeed the case except for significant oscillations of IL-6 in hypothalamus and liver under LLbright. In fact, diurnal variations in microglial inflammatory responses are shown to be entrained by rhythmic expression of glucocorticoids in mammals⁵³. Hence, a concurrent aperiodicity in immune response and its endocrine mediators in the present study were expected. However, the physiological significance of a rhythmic hypothalamic and hepatic IL-6 expression under LLbright remains to be ascertained in future studies. While peripheral metabolic rhythms are known to be entrained independent of the master circadian clock⁵⁷, it is unlikely that metabolic entrainment caused significant hepatic immune rhythm persistence under LLbright since food was available ad libitum and food replenishment times were randomized across the 24 hour day (cf. methods section). In comparison to pro-inflammatory cytokine gene expression under LD, elevated night levels of IL-6 in hypothalamus and liver under LLbright, and perhaps the elevation of IL-6 in hippocampus as well as IL-1β and IL-6 in fat under DLAN were indicative of a night light-induced immune response⁵⁸. Sleep loss has been shown to increase microglial activity and pro-inflammatory cytokine release in the hippocampus of rats⁵⁹. Such sleep loss induced elevation in neuroinflammation is associated with neurological damage and cognitive decline⁶⁰. However, it is interesting to note that elevated proinflammatory response (IL-6) to LLbright and DLAN was evidenced in hypothalamus and hippocampus, but not in cortex, suggesting effects of environmental perturbations on neuroinflammation varied across brain regions. Further studies are warranted to understand how LAN affects neuroinflammation across different brain regions. Also, an upregulation of proinflammatory cytokines in fat and liver in response to LLbright may be attributed to the infiltration of classical proinflammatory M1 macrophages into adipose tissues²² and an increased activity of Kupffer cells, the resident tissue macrophages of the liver²⁵. Mammalian studies suggest an association of a heightened state of inflammation in adipose tissue with an increased macrophage infiltration in fat of obese individuals. Macrophage infiltration has been shown to alter insulin sensitivity locally in the adipose tissues and systemically⁶¹. If a proinflammatory state of adipose tissue alters its metabolic and endocrine sensitivity in birds, similar to those reported in mammals, then the light pollution may severely affect fitness in birds, especially species heavily relying on lipid catabolism as a primary energy source during long migratory flights⁶². Further, a chronic pro-inflammatory hepatic response is associated with a variety of physiological and pathophysiological conditions, including hepatic infections, fatty liver disease, liver injury, and fibrosis in mammals²⁵. The spleen is involved in humoral and cellular immune responses through its role in generation, maturation and storage of lymphocytes. Cytokine gene expression in the avian spleen is commonly used as an indicator of immune response²⁰. Consistent with a previous study reporting increased splenic inflammation under aperiodic conditions⁶³, the night time expression of pro-inflammatory cytokines in spleen were elevated under LLbright and DLAN. We would also not rule out a circadian disruption in splenic response to a stimulus (e.g., lipopolysaccharide injection) as reported in other studies²¹.

Studies in diurnal mammals⁶⁴ and domesticated birds⁵⁸ suggest that perturbations in LD cycle tend to alter pro-inflammatory responses in brain and peripheral tissues, perhaps through modulation of melatonin release¹⁵. Immununoenhancing properties of melatonin are well documented¹⁵. Melatonin administration rescues the immunosuppressive effects of cortisol in humans⁶⁵. Thus, significant downregulation of cytokine expression in nidopallium and hippocampus under LLbright may be attributed to dampened melatonin and elevated cort responses. However, a lack of similar response in other tissues, especially peripheral tissues, exemplifies the complexities of neuroendocrine-immune (NEI) interactions⁵⁶. It is probable that inflammatory responses from an aperiodic light environment are not solely mediated by melatonin and/or cort but rather by their interaction with other hormones. For example, stimulatory effects of melatonin on cellular and humoral immune response in quail are dependent on opioids¹⁸, while modulation of seasonal immunity in Indian tropical birds, Perdicula asiatica, involves an interaction between melatonin and sex steroids³².

Furthermore, since the magnitude and duration of the nocturnal increase in melatonin acts as an entrainment cue for biological functions^27,40, we hypothesized that the effects of LAN intensity on melatonin would also be reflected in cytokine gene expressions of zebra finches. Indeed, in comparison with LLbright, DLAN-induced effects on diurnal rhythms of cytokines were less apparent. Pro-inflammatory cytokines in liver and fat, IL-6 in nidopallium and hippocampus, and IL-10 in liver and hippocampus were significantly rhythmic under DLAN, although there were significant changes in waveform parameters. DLAN is known to alter metabolic state through a shift in timing of food intake⁶⁶. Perhaps, persistence of diurnal variation of the three interleukins in liver, a key metabolic organ, with shifts in waveforms may be attributed to a change in daily activity and feeding patterns. Zebra finches exposed to DLAN were more active during the dark phase than birds in pitch dark in the present study (Suppl Fig. 1). However, an association of behavioral rhythms with immune system need to be ascertained in future studies. Such DLAN induced changes in phase relationships between immune rhythms and other physiological and behavioral rhythms could be speculated to have adverse fitness consequences in the wild, but further study is warranted.

To our knowledge, this is the first study examining tissue-specific alterations in daily oscillations in immune responses in an avian species exposed to LAN. While further study is required to identify a cause-and-effect relationship between endocrine and immune axes, we document concurrent alterations in rhythms of cytokines and their key endocrine regulators that are dependent upon night-light intensity. The alterations of endocrine and immune daily rhythms in the present study were in response to a relatively brighter LAN (3 lux) intensity. We speculate that the diurnal rhythms would show LAN-intensity and -wavelength dependent alterations. A dose dependent advance in activity and suppression of melatonin daily rhythm to LAN has been shown in great tits, Parus major⁹ and histological and molecular analyses on testes of great tits indicated a dose-dependent reproductive response to ALAN, with the higher effect on spermatogenesis under 5 lux than under 0.5 and 1.5 lux⁶⁷. Further, zebra finches exposed to 5000 K LAN had increased corticosterone levels compared to finches exposed to 3000 K³⁶. It should be noted that we did not test for the ‘free’ endogenous circadian rhythms of cytokines under constant dim light (LL_dim). However DLAN-induced waveform alterations and LL-induced rhythm disruption suggest diel changes in inflammatory cytokine production may contribute to circadian control of immunity in this diurnal avian species. Future studies are needed to determine whether cytokine rhythm perturbations from LAN are mediated through direct effects of light by a central circadian clock or via its effects on local clock machinery of immune cells. In addition, the results of this study provide an impetus for researchers to evaluate other components of immune function (cellular and humoral responses) that may be differently affected by light perturbations. Lastly, elucidating the adaptive function of these rhythms will increase our understanding of effects upon host fitness and response to disease in a world that is increasingly being affected by light pollution.

These experiments were funded by the National Science Foundation (IOS-1557882) and the National Institutes of Health (R15GM117534) to NTA. We thank Nicholas Wheeler, Keelee Pullum, Sasha Sairajeev, Anna Strunjas and Kristen Eads for assistance with sampling and qPCR analysis.

I.M. and N.T.A. designed the study; I.M. performed the experiment; I.M., N.T.A. and M.M.R. did sampling; I.M., R.M.K., W.I.P., A.A.S. acquired the data and carried out the analysis; I.M. and N.T.A. wrote the manuscript. All authors read and approved the final manuscript.

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

The authors declare no competing interests.

is available for this paper at 10.1038/s41598-019-51791-9.