1. Neuroscience
Download icon

Distinct ipRGC subpopulations mediate light’s acute and circadian effects on body temperature and sleep

  1. Alan C Rupp
  2. Michelle Ren
  3. Cara M Altimus
  4. Diego C Fernandez
  5. Melissa Richardson
  6. Fred Turek
  7. Samer Hattar
  8. Tiffany M Schmidt  Is a corresponding author
  1. Johns Hopkins University, United States
  2. Northwestern University, United States
Research Article
  • Cited 3
  • Views 1,295
  • Annotations
Cite this article as: eLife 2019;8:e44358 doi: 10.7554/eLife.44358

Abstract

The light environment greatly impacts human alertness, mood, and cognition by both acute regulation of physiology and indirect alignment of circadian rhythms. These processes require the melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs), but the relevant downstream brain areas involved remain elusive. ipRGCs project widely in the brain, including to the central circadian pacemaker, the suprachiasmatic nucleus (SCN). Here we show that body temperature and sleep responses to acute light exposure are absent after genetic ablation of all ipRGCs except a subpopulation that projects to the SCN. Furthermore, by chemogenetic activation of the ipRGCs that avoid the SCN, we show that these cells are sufficient for acute changes in body temperature. Our results challenge the idea that the SCN is a major relay for the acute effects of light on non-image forming behaviors and identify the sensory cells that initiate light’s profound effects on body temperature and sleep.

https://doi.org/10.7554/eLife.44358.001

eLife digest

Light, whether natural or artificial, affects our everyday lives in several ways. Exposure to light impacts on our health and well-being. It plays a crucial but indirect role in helping to align our internal body clock with the 24-hour cycle of day and night, and a burst of bright light in the middle of the night can wake us up from sleep.

Decades of research have revealed the circuitry that controls the indirect effects of light on the body's internal clock. A tiny set of cells in the base of the brain called the suprachiasmatic nucleus (SCN for short) generates the body’s daily or “circadian” rhythm. A small group of nerve cells in the retina of the eye called intrinsically photosensitive retinal ganglion cells (ipRGCs) connect with the SCN. These ipRGCs relay information about light to the SCN to ensure that daily rhythms happen at the appropriate times of day. But scientists do not yet know if the same brain circuits regulate the direct effects of light on alertness.

Mice are often used in studies of circadian rhythms but, unlike humans, mice are normally active at night and sleep throughout the day. This means that a burst of bright light in the middle of the night causes mice to become less alert.

Now, in experiments with mice, Rupp et al. show there are two separate circuits from the retina to the brain that influence wakefulness. In the experiments, some mice were genetically engineered to only have ipRGCs that connect with the SCN and to lack those that connect with other brain areas. These mice lived in cages with a normal day/night cycle and their body temperature and sleep-related brain activity were monitored as Rupp et al. sporadically exposed them to bright light at night. These mice continued their normal routines and were unaffected by the bursts of light. In a second set of experiments, ipRGCs that do not connect with the SCN were activated in other mice. This caused an immediate and sustained drop in the body temperature of the mice, which is linked to them becoming less alert.

The experiments suggest that the circuit that connects ipRGCs to the SCN to align the body’s circadian rhythm with light does not control the direct effect of light on wakefulness. Instead, a separate circuit that extends from ipRGCs to an unknown part of the brain area influences wakefulness. Better understanding this second circuit could allow scientists to develop ways to keep people like emergency personnel or overnight shift workers awake and alert at night while avoiding harmful disruptions to their circadian rhythms.

https://doi.org/10.7554/eLife.44358.002

Introduction

Many essential functions are influenced by light both indirectly through alignment of circadian rhythms (photoentrainment) and acutely by a direct mechanism (sometimes referred to as ‘masking’) (Mrosovsky et al., 1999; Altimus et al., 2008; Lupi et al., 2008; Tsai et al., 2009; LeGates et al., 2012). Dysregulation of the circadian system by abnormal lighting conditions has many negative consequences, which has motivated decades of work to identify the mechanisms of circadian photoentrainment (Golombek and Rosenstein, 2010). In contrast, it has only recently become apparent that light exposure can also acutely influence human alertness, cognition, and physiology (Chellappa et al., 2011). As a result, there is a developing awareness of light quality in everyday life (Lucas et al., 2014). It is therefore essential to human health and society to elucidate the circuitry and coding mechanisms underlying light’s acute effects.

Intriguingly, a single population of retinal projection neurons—intrinsically photosensitive retinal ganglion cells (ipRGCs)—have been implicated in the circadian and acute effects of light on many functions, including activity, sleep, and mood (Göz et al., 2008; Güler et al., 2008; Hatori et al., 2008; LeGates et al., 2012; Fernandez et al., 2018). ipRGCs integrate light information from rods, cones, and their endogenous melanopsin phototransduction cascade (Schmidt et al., 2011), and relay that light information to over a dozen central targets (Hattar et al., 2006; Ecker et al., 2010). However, the circuit mechanisms mediating ipRGC-dependent functions are largely unknown.

One notable exception is the control of circadian photoentrainment. It is accepted that ipRGCs mediate photoentrainment by direct innervation of the master circadian pacemaker, the suprachiasmatic nucleus (SCN) of the hypothalamus (Göz et al., 2008; Güler et al., 2008; Hatori et al., 2008; Jones et al., 2015). This is supported by studies demonstrating that genetic ablation of ipRGCs results in mice with normal circadian rhythms that ‘free-run’ with their endogenous rhythm, independent of the light/dark cycle (Göz et al., 2008; Güler et al., 2008; Hatori et al., 2008). Further, mice with genetic ablation of all ipRGCs except those that project to the SCN and intergeniculate leaflet (IGL) display normal circadian photoentrainment (Chen et al., 2011), suggesting that ipRGC projections to the SCN/IGL are sufficient for photoentrainment.

In comparison, the mechanisms by which ipRGCs mediate acute light responses remain largely a mystery. Genetic ablation of ipRGCs or their melanopsin phototransduction cascade blocks or attenuates the acute effects of light on sleep (Altimus et al., 2008; Lupi et al., 2008; Tsai et al., 2009), wheel-running activity (Mrosovsky and Hattar, 2003; Güler et al., 2008), and mood (LeGates et al., 2012; Fernandez et al., 2018). This dual role of ipRGCs in circadian and acute light responses suggests they may share a common circuit mechanism. However, whether the circuit basis for ipRGCs in the acute effects of light and circadian functions is through common or divergent pathways has yet to be determined. ipRGCs project broadly in the brain beyond the SCN (Hattar et al., 2002; Hattar et al., 2006; Gooley et al., 2003; Baver et al., 2008). Additionally, ipRGCs are comprised of multiple subpopulations with distinct genetic, morphological, and electrophysiological signatures (Baver et al., 2008; Schmidt and Kofuji, 2009; Ecker et al., 2010; Schmidt et al., 2011) and distinct functions (Chen et al., 2011; Schmidt et al., 2014). Though there are rare exceptions (Chen et al., 2011; Schmidt et al., 2014), the unique roles played by each ipRGC subsystem remain largely unknown.

It is currently unknown whether distinct ipRGC subpopulations mediate both the acute and circadian effects of light, and two major possibilities exist for how this occurs: (1) ipRGCs mediate both acute and circadian light responses through their innervation of the SCN or (2) ipRGCs mediate circadian photoentrainment through the SCN, but send collateral projections elsewhere in the brain to mediate acute light responses. To date, the predominant understanding has centered on a role for the SCN in both acute and circadian responses to light (Muindi et al., 2014; Morin, 2015; Bedont et al., 2017). However, this model has been controversial due to complications associated with SCN lesions (Redlin and Mrosovsky, 1999) and alternative models proposing a role for direct ipRGC input to other central targets (Redlin and Mrosovsky, 1999; Lupi et al., 2008; Tsai et al., 2009; Hubbard et al., 2013; Muindi et al., 2014). Here, we sought to address the question of how environmental light information—through ipRGCs—mediates both the circadian and acute regulation of physiology. To do so, we investigated the ipRGC subpopulations and coding mechanisms that mediate body temperature and sleep regulation by light. We find that a molecularly distinct subset of ipRGCs is required for the acute, but not circadian, effects of light on thermoregulation and sleep. These findings suggest that, contrary to expectations, functional input to the SCN is not sufficient to drive the acute effects of light on these behaviors. These findings provide new insight into the circuits through which light regulates behavior and physiology.

Results

Brn3b-positive ipRGCs are required for light’s acute effects on thermoregulation

To identify mechanisms of acute thermoregulation, we maintained mice on a 12 hr/12 hr light/dark cycle and then presented a 3 hr light pulse two hours into the night (Zeitgeber time 14, ZT14) while measuring core body temperature (Figure 1A). The nocturnal light pulse paradigm is well-established for studying acute regulation of sleep and wheel-running activity (Mrosovsky et al., 1999; Mrosovsky and Hattar, 2003; Altimus et al., 2008; Lupi et al., 2008). We focused first on body temperature because of its critical role in cognition and alertness (Wright et al., 2002; Darwent et al., 2010), sleep induction and quality (Kräuchi et al., 1999), metabolic control (Kooijman et al., 2015), and circadian resetting (Buhr et al., 2010).

Figure 1 with 2 supplements see all
Melanopsin mediates the acute effects of light on body temperature.

(A) Paradigm to measure body temperature continuously in a 12:12 light dark cycle with a 3 hr light pulse at ZT14. (B) 48 hr of continuous body temperature monitoring in wildtype male mice (n = 13) (C) Relative body temperature in WT during light pulse, compared to baseline (ZT14). p<0.001, paired t-test of mean temperature compared to previous night. (D) Melanopsin-only mice (Gnat1-/-; Gnat2-/-, n = 11) and (E) melanopsin knockout (Opn4-/-, n = 6) 48 hr diurnal body temperature. (F) Diurnal body temperature amplitude in the three groups. p>0.347 for effect of group by one-way ANOVA. (G) Body temperature in melanopsin-only during light pulse, relative to baseline (ZT14). (H) Paired comparison of mean body temperature during light pulse compared to previous night. p<0.001 by paired t-test. (I) Body temperature in melanopsin knockout during light pulse, relative to baseline (ZT14). (J) Paired comparison of mean body temperature during light pulse compared to previous night. All summarized data are mean ± standard deviation.

https://doi.org/10.7554/eLife.44358.003

Body temperature photoentrains to the light/dark cycle with peaks during the night and troughs during the day (Figure 1B). Both rodents and humans utilize ocular light detection to acutely adjust body temperature in response to a nocturnal light pulse (Dijk et al., 1991; Cajochen et al., 2005), though how this body temperature change is initiated by the retina and relayed to the brain is unknown. When we presented wildtype mice with a nocturnal light pulse, we observed a decrease in both body temperature and general activity compared to the previous night (Figure 1C). The decrease in body temperature and activity was sustained for the entire 3 hr stimulus, with moderate rundown (Figure 1C).

We observed that acute body temperature regulation only occurred at relatively bright light intensities (>100 lux) (Figure 1—figure supplement 1). This, in combination with previous reports that body temperature regulation is most sensitive to short-wavelength light (Cajochen et al., 2005), suggested that it might be mediated by the insensitive and blue-shifted melanopsin phototransduction (Lucas et al., 2001; Do et al., 2009). To test this, we measured body temperature in mice lacking either functional rods and cones (melanopsin-only: Gnat1-/-; Gnat2-/-) or lacking melanopsin (melanopsin KO: Opn4-/-). Both genotypes photoentrained their body temperature (Figure 1D,E), with an amplitude indistinguishable from wildtype (Figure 1F). However, we found that acute body temperature decrease to a nocturnal light pulse was present in melanopsin-only mice (Gnat1-/-; Gnat2-/-) (Figure 1G,H and Figure 1—figure supplement 2), but absent from melanopsin knockout mice (Opn4-/-) (Figure 1I,J and Figure 1—figure supplement 2). This indicates that melanopsin is critical for light’s ability to drive acute body temperature decreases, as it is for acute sleep induction (Altimus et al., 2008; Lupi et al., 2008; Tsai et al., 2009). These results suggest that ipRGCs are the only retinal cells that are necessary and sufficient for acute thermoregulation by light.

ipRGCs comprise multiple subtypes (M1-M6) with distinct gene expression profiles, light responses, and central projections (Schmidt et al., 2011; Quattrochi et al., 2019), prompting us to ask which subtypes mediate acute thermoregulation. ipRGCs can be molecularly subdivided based on whether they express the transcription factor Brn3b. Brn3b(+) ipRGCs project to many structures including the olivary pretectal nucleus (OPN) and dorsal lateral geniculate nucleus (dLGN), but largely avoid the SCN (Chen et al., 2011; Li and Schmidt, 2018). In contrast, Brn3b(–) ipRGCs project extensively to the SCN and intergeniculate leaflet (IGL), while avoiding the OPN and dLGN (Chen et al., 2011). Non-M1 (i.e. M2-M6) ipRGC subtypes express Brn3b, along with the majority of M1 ipRGCs. Interestingly, just 200 (out of 700–800) M1 ipRGCs lack any Brn3b expression (Chen et al., 2011). Ablation of Brn3b(+) ipRGCs using melanopsin-Cre and a Cre-dependent diphtheria toxin knocked into the Brn3b locus (Brn3b-DTA: Opn4Cre/+;Brn3bzDTA/+) removes virtually all ipRGC input to brain areas aside from the SCN and IGL (Chen et al., 2011; Li and Schmidt, 2018), and these mice lack a pupillary light reflex and show deficits in contrast sensitivity, but retain circadian photoentrainment of wheel-running activity (Chen et al., 2011; Schmidt et al., 2014).

When we measured body temperature in Brn3b-DTA mice, we found that their body temperature was photoentrained with a similar amplitude to controls (Figure 2A–C). However, despite the presence of melanopsin in the Brn3b(-) ipRGCs of Brn3b-DTA mice (Opn4Cre/+;Brn3bzDTA/+), they did not acutely decrease body temperature in response to a nocturnal light pulse (Figure 2F,G). Importantly, melanopsin heterozygous littermate controls (Opn4Cre/+) displayed normal acute thermoregulation by light (Figure 2D,E), indicating that halving melanopsin gene dosage is not the cause of the impaired body temperature decrease in Brn3b-DTA mice. Additionally, when we compared the change in body temperature of Control to Brn3b-DTA mice during that light pulse, we found that Control mice showed a significantly larger decrease in body temperature (Figure 2—figure supplement 1). These results demonstrate that Brn3b(+) ipRGCs are required for acute thermoregulation by light but not photoentrainment of body temperature and reveal that light information to the SCN is sufficient for circadian photoentrainment of body temperature, but not its acute regulation.

Figure 2 with 1 supplement see all
Brn3b-negative ipRGCs are insufficient for acute body temperature regulation via the SCN.

(A) Diurnal body temperature in control (Opn4Cre/+, n = 9) and (B) Brn3b-DTA (Opn4Cre/+;Brn3bDTA/+, n = 7). (C) Diurnal body temperature amplitude in the two groups. p=0.223 by t-test. (D) Body temperature in control during light pulse, relative to baseline (ZT14). (E) Paired comparison of mean body temperature during light pulse compared to previous night. p=0.001 by paired t-test. (F) Body temperature in Brn3b-DTA during light pulse, relative to baseline (ZT14). (G) Paired comparison of mean body temperature during light pulse compared to previous night. p=0.287 by paired t-test. All summarized data are mean ± standard deviation.

https://doi.org/10.7554/eLife.44358.008

Brn3b-positive ipRGCs are sufficient for acute thermoregulation

Our data thus far suggest that there are two functionally distinct populations of ipRGCs that regulate thermoregulation: (1) Brn3b(–) ipRGCs that project to the SCN to mediate circadian photoentrainment of body temperature and (2) Brn3b(+) ipRGCs that project elsewhere in the brain and are necessary to mediate acute thermoregulation. If Brn3b(+) ipRGCs are not just necessary, but also sufficient, for acute thermoregulation, then activation of this population at ZT14 should result in a body temperature decrease. To test if Brn3b(+) ipRGCs are sufficient for acute thermoregulation, we expressed a chemogenetic activator in Brn3b(+) RGCs (Figure 3A, Brn3bCre/+ with intravitreal AAV2-hSyn-DIO-hM3Dq-mCherry, we refer to these mice as Brn3b-hM3Dq). As a control, we also injected this virus into Control (Brn3b+/+) littermates. We then injected both genotypes first with PBS at ZT14 on the first night, and CNO at ZT14 on the second night. This technique allowed for statistical within animal comparisons of body temperature changes in response to PBS versus CNO injection. Importantly, CNO did not cause a significant decrease in body temperature in the absence of hM3Dq (Figure 3—figure supplement 1). This technique allowed us to acutely activate the Brn3b(+) RGCs with the DREADD agonist clozapine N-oxide (CNO) (Armbruster et al., 2007). We found that after intravitreal viral delivery, many RGCs were infected, including melanopsin-expressing ipRGCs (Figure 3A and Figure 3—figure supplement 1).

Figure 3 with 2 supplements see all
Activation of Brn3b-positive RGCs is sufficient to drive sustained body temperature decreases.

(A) Diagram of intravitreal delivery of AAV2-hSyn-DIO-hM3Dq-mCherry to Brn3bCre/+ mice, and confirmation of infection of ipRGCs. (B) 54 hr continuous diurnal body temperature recordings in Brn3b-hM3Dq mice, with injections of PBS then CNO on consecutive nights at ZT14. (C) Change in body temperature after PBS injection, relative to baseline (time of injection). (D) Change in body temperature after CNO injection, relative to baseline (time of injection). (E) Paired comparison of the change in body temperature with either PBS or CNO injection, compared to temperature at injection time. p=0.002 by paired t-test. All summarized data are mean ± standard deviation.

https://doi.org/10.7554/eLife.44358.011

The body temperature of Brn3b-hM3Dq mice photoentrained to a normal light/dark cycle (Figure 3B). Following CNO administration to Brn3b-hM3Dq mice at ZT14 to depolarize Brn3b(+) RGCs, we observed a robust decrease in body temperature that lasted at least 6 hr (Figure 3D). Importantly, PBS administration in Brn3b-hM3Dq mice (Figure 3C) and nocturnal CNO administration in wildtype control mice (Figure 3—figure supplement 2) had no measurable effect on body temperature, while CNO administration significantly decreased body temperature in Brn3b-hM3Dq compared to pre-injection temperature (Figure 3—figure supplement 2). Together, these results demonstrate that Brn3b(+) ipRGCs mediate the acute effects of light on body temperature though extra-SCN projection(s), while Brn3b(–) ipRGCs mediate circadian photoentrainment of body temperature by projections to the SCN and/or IGL.

Brn3b-positive ipRGCs are required for light’s acute effects on sleep

We next examined the contribution of Brn3b(+) and Brn3b(-) ipRGCs to sleep. To do this, we used EEG and EMG recordings to compare the sleep behavior of Control (Opn4Cre/+) and Brn3b-DTA mice. We first analyzed the daily sleep patterns and proportion of rapid eye movement (REM) and non-REM (NREM) sleep in Control and littermate Brn3b-DTA animals. We found that Brn3b-DTA mice show normal photoentrainment of sleep and similar percent time of sleep across the 24 hour day, with only one 30 min bin at ZT12 (light offset) showing a significant difference between Control and Brn3b-DTA animals (Figure 4A,B). This is consistent with previous reports of normal circadian photoentrainment of daily activity rhythms in Brn3b-DTA mice (Chen et al., 2011). Control and Brn3b-DTA mice also showed similar total percent time awake or asleep across an entire day (Figure 4C), though Brn3b-DTA mice showed a small, but significant, increase in the proportion of total sleep that was classified as NREM and decrease in the proportion of total sleep that was classified as REM (Figure 4—figure supplement 1A).

Figure 4 with 3 supplements see all
Brn3b-positive ipRGCs are not required for circadian photoentrainment of sleep, but are required for its acute induction by light.

(A–C) Percent time spent asleep in 1 hr bins across the 24 hr day for (A) Control (black) mice (n = 14) and (B) Brn3b-DTA (blue) mice (n = 13) lacking Brn3b-positive ipRGCs. Both lines showed normal photoentrainment of sleep, with no main effect of genotype compared to Control by repeated-measures two-way ANOVA (F (1, 25)=1.108, p=0.303). Brn3b-DTA mice showed a significant reduction in sleep only at lights off (ZT 12) by Sidak’s multiple comparisons test (p=0.029). (C) Percent time spent awake and asleep in Control (black) and Brn3b-DTA mice (blue). No differences were observed between genotypes by t-test (p=0.316). (D–G) Percent time spent asleep for (D) Control mice (black) and (F) Brn3b-DTA mice (blue) at baseline (dark line) and during the three hour light pulse (light line). Significant difference from baseline determined by repeated measures two-way ANOVA. Significant effect of treatment for Controls (F (1, 13)=38.09, p<0.001), but not for Brn3b-DTA (F (1, 12)=0.8496, p=0.375). (E) Control mice show significantly more sleep and less wake during a light pulse (paired t-test) while (G) Brn3b-DTA mice showed no change in percent sleep or wake during the same period. Data are mean for ZT14–17. All summarized data are mean ± SEM.

https://doi.org/10.7554/eLife.44358.016

We hypothesized that this small difference in sleep at lights-off in Brn3b-DTA mice could be due to a defect in their acute response to light for sleep modulation. To test this, we subjected mice to a 3 hr light pulse from ZT14–17 (Altimus et al., 2008), when the homeostatic drive for sleep is low and Control and Brn3b-DTA animals display similar amounts of sleep (Figure 4A,B). We found that in Control mice, a light pulse decreased time awake and increased time asleep relative to baseline (previous day) (Figure 4C,D), while in Brn3b-DTA mice a light pulse caused no change in total percent time asleep or awake (Figure 4F,G), but moderately increased sleep in the first 30 min bin (Figure 4F). Importantly, when we compared the time spent asleep during the light pulse between control and Brn3b-DTA animals, the control mice slept significantly more (Figure 4—figure supplement 2). Neither Control nor Brn3b-DTA animals showed any change in proportion of non-REM or REM sleep in response to the light pulse (Figure 4—figure supplement 1B,C). These data show that Brn3b(+) ipRGCs are necessary for the acute light induction of sleep. Consistent with our body temperature data, although Brn3b-DTA mice have apparently normal input to the SCN and show normal circadian photoentrainment of wheel-running activity (Chen et al., 2011), body temperature (Figure 2), and sleep (Figure 4), this ipRGC innervation of the SCN is not sufficient to drive the normal light induction of sleep. These disruptions in light’s acute effects on thermoregulation and sleep are circuit specific effects because Brn3b-DTA mice showed robust inhibition of wheel running behavior to a 3 hr light pulse delivered from ZT14-17 (Figure 4—figure supplement 3).

Discussion

We show here that for the same physiological outcome, the acute effects of light are relayed through distinct circuitry from that of circadian photoentrainment, despite both processes requiring ipRGCs. Our results suggest that for thermoregulation and sleep, ipRGCs can be genetically and functionally segregated into Brn3b(+) ‘acute’ cells, and Brn3b(–) ‘circadian’ cells. Because Brn3b(+) cells largely avoid the SCN, and Brn3b(–) cells preferentially target the SCN, our results discount a critical role for the SCN in acute light responses in these two behaviors, and instead implicate direct ipRGC projections to other brain areas (Gooley et al., 2003; Hattar et al., 2006). Surprisingly, Brn3b(-) cells are sufficient to drive the acute and circadian effects of light on wheel running activity, demonstrating further divergence in the circuits mediating the acute effects of light on behavior, and suggesting that, unlike for thermoregulation and sleep, acute and circadian regulation of activity is driven via the SCN.

Our results indicate that activation of Brn3b(+) RGCs at ZT14 using the Brn3bCre line in combination with Gq-DREADDs is sufficient to induce a body temperature decrease. Because other (non-ipRGC) RGC types express Brn3b (Badea et al., 2009), this manipulation likely also activated multiple non-ipRGCs in addition to Brn3b(+) ipRGCs. However, our data indicate that melanopsin signaling (Figure 1), and therefore ipRGCs, are required for the acute effects of light on thermoregulation. Moreover, when we ablate Brn3b(+) ipRGCs, this acute effect of light on thermoregulation is also lost (Figure 2), again arguing for a necessity of ipRGCs for this behavior. Therefore, though we are unable to specifically activate only Brn3b(+) ipRGCs using available genetic tools, we think it highly likely that the temperature changes driven by the activation of all Brn3b(+) RGCs is occurring through ipRGCs.

The specific Brnb(+) ipRGC subtypes that mediate the light’s acute effects on body temperature and sleep remain a mystery. A majority of all known ipRGC subtypes (M1–M6) are lost in Brn3b-DTA mice (Chen et al., 2011), with the exception of a subset of ~200 M1 ipRGCs. In agreement with this, ipRGC projections to all minor hypothalamic targets are lost in Brn3b-DTA mice, while innervation of the SCN and part of the IGL remains intact (Chen et al., 2011; Li and Schmidt, 2018). This suggests that all non-M1 subtypes and a portion of M1 ipRGCs are Brn3b(+). Each subtype has a distinct reliance on melanopsin versus rod/cone phototransduction for light detection (Schmidt and Kofuji, 2009). The necessity and sufficiency of melanopsin in mediating acute effects of light on body temperature (Figure 1) and sleep (Altimus et al., 2008; Lupi et al., 2008; Tsai et al., 2009) suggests that a subtype with strong melanopsin, but weak rod/cone photodetection is responsible – possibly either M1 or M2 cells. However, experiments to tease this apart will require novel methods to specifically manipulate ipRGC subtypes that are currently unavailable.

The brain areas that mediate the acute effects of light on physiology are essentially unknown. There are many candidate areas that both receive direct ipRGC innervation and have been documented to be involved in light’s acute effects on physiology, including the preoptic areas (Muindi et al., 2014), the ventral subparaventricular zone (Kramer et al., 2001), and the pretectum/superior colliculus (Miller et al., 1998). Aside from the SCN, ipRGC projections to the median (MPO) and ventrolateral preoptic (VLPO) areas have been the most widely supported. The preoptic areas are involved in sleep and body temperature regulation (Szymusiak and McGinty, 2008; Nakamura, 2011) and are activated by an acute light pulse (Lupi et al., 2008; Tsai et al., 2009). In support of our behavioral findings, ipRGC projections to each of these areas is lost in Brn3b-DTA animals (Li and Schmidt, 2018). However, ipRGC projections to these areas are sparse (Gooley et al., 2003; Hattar et al., 2006), suggesting their activation by light may be indirect.

In contrast, the superior colliculus (SC) and pretectum receive robust innervation from ipRGCs (Hattar et al., 2002; Hattar et al., 2006; Gooley et al., 2003; Ecker et al., 2010), their lesioning blocks light’s acute effects on sleep (Miller et al., 1998), and melanopsin knockout mice lose light-induced cFOS expression in the SC (Lupi et al., 2008). However, the SC and pretectum receive robust innervation from many conventional RGCs, making the requirement for melanopsin and ipRGCs in acute sleep and body temperature regulation difficult to reconcile. It is also possible (and perhaps probable), that multiple ipRGC target regions are involved, with distinct areas mediating distinct physiological responses. Future studies silencing each retinorecipient target while activating Brn3b(+) ipRGCs will be necessary to tease apart the downstream circuits mediating light’s acute effects on physiology.

Alternatively, it remains possible that direct ipRGC control of body temperature is the primary and critical step for many acute responses to light that are mediated by ipRGCs. In support of this possibility, changes in body temperature and heat loss can directly influence sleep induction (Kräuchi et al., 1999). This change in sleep is in turn presumably causative of at least some of light’s effects on wheel-running and general activity (Mrosovsky et al., 1999). Further, core body temperature can acutely regulate cognition and alertness (Wright et al., 2002; Darwent et al., 2010). It is therefore possible that ipRGCs can have widespread influence on an animal’s basic physiology and cognitive function simply by regulating body temperature.

Together, our identification of the photopigment and the retinal circuits mediating acute body temperature and sleep induction will facilitate better methods to promote or avoid human alertness and cognition at appropriate times of day (Chellappa et al., 2011). Our results support many recent efforts to capitalize on the specific light-detection properties of melanopsin (Lucas et al., 2014), such as its insensitivity and short-wavelength preference, to promote or avoid its activation at different times of day. However, this approach is problematic because acute activation of melanopsin to promote alertness has the unintended effect of shifting the circadian clock (Provencio et al., 1994), thereby making subsequent sleep difficult. Our identification that the Brn3b(+) ipRGCs specifically mediate light’s acute effects on body temperature provides a cellular basis for developing targeted methods for promoting acute alertness, while minimizing circadian misalignment.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
Identifiers
Genetic reagent (Mus musculus)Opn4tauLacZHattar et al., 2002Jax: 021153
RRID:MGI:5520170
Genetic reagent (Mus musculus)Gnat1-/-PMID: 11095744
Genetic reagent (Mus musculus)Gnat2Cpfl3PMID: 17065522Jax: 006795
Genetic reagent (Mus musculus)Opn4CreEcker et al., 2010RRID:MGI:5285910
Genetic reagent (Mus musculus)Brn3bzDTAChen et al., 2011RRID:MGI:5285910
Genetic reagent (Mus musculus)Brn3bCrePMID: 24608965RRID:IMSR_JAX:030357
Antibodyanti-OPN4 (rabbit polyclonal)Advanced Targeting SystemsAB-N38 (1:1000)
RRID:AB_1608077
AntibodyAlexaFluor 488, anti-rabbit (goat polyclonal)Life TechnologiesA-11008 (1:1000)
RRID:AB_143165
Viral reagentAAV2-hSyn-DIO-hM3Dq-mCherryUNC Vector Core
Chemical compound, drugClozpine-N-oxideSigma
SoftwareR 3.5.2https://cran.r-project.org/
SoftwareGraphpad Prism 7.0https://www.graphpad.com/scientific-software/prism/

Animals (body temperature)

Request a detailed protocol

All procedures were conducted in accordance with NIH guidelines and approved by the Institutional Animal Care and Use Committee of Johns Hopkins University. All mice were maintained on a mixed C57Bl/6J; 129Sv/J background and kept on ad libitum food and water under a 12 hr/12 hr light/dark cycle in group housing until experimentation, with temperature and humidity control. Male and female mice between the ages of 2 and 6 months were used for analysis.

Body temperature recordings

Request a detailed protocol

Each mouse was single-housed at the time of experiment. Surgery was conducted under tribromoethanol (Avertin) anestheshia and a telemetric probe (Starr G2 E-Mitter) was implanted in the peritoneal cavity to monitor core body temperature and general activity. Data were collected in continuous 1- or 2 min bins using VitalVIEW software and analyzed in Microsoft Excel. All experiments were conducted at least 10 days after surgery. Lights were controlled by a programmable timer and all lights were 6500K CFL bulbs illuminated each cage at ~500 lux. Light intensity (Figure 1—figure supplement 1) was modulated using neutral density filters (Roscolux).

Brn3bCre/+ or Control littermate mice were anesthetized with tribromoethanol (Avertin) and 0.5–1 µl AAV2-hSyn-DIO-hM3Dq-mCherry (UNC Vector Core) was injected intravitreally in one eye using a picospritzer and pulled pipet. At least one week later, animals underwent surgery for implantation of telemetric probes (as above). All experiments were conducted at least 10 days after telemetric probe implantation and at least three weeks after viral injection. After behavior, the eyes of each animal were inspected to ensure that >50% infection had been achieved (assessed by fluorescence detectable across more than half of the retina). We routinely saw >80% of the retinas were infected as we have described previously (Keenan et al., 2016).

Diurnal amplitude was measured by subtracting the mean body temperature for the light cycle (ZT0-12) from the mean body temperature for the dark cycle (ZT12-24). Mean body temperature during testing used all data from ZT14-17. Comparisons were performed in one of two ways. First, we compared the mean body temperature during this period on the control (dark) night to that on the night where the light pulse was given. Additionally, we compared the change in body temperature between ZT14 (which served as a baseline) and the mean body temperature from ZT14-17 between the control night and the night where the light pulse was given. For CNO experiments, injections were carried out near ZT14, but specific times were recorded for each mouse to align the data to the time of injection. Comparisons of mean body temperature after PBS or CNO utilized the 6 hr following injection.

Clozapine-N-oxide (Sigma) was prepared as a 0.1 mg/ml solution in PBS and injected at 1 mg/kg intraperitoneally at ZT14.

Animals (Sleep)

Request a detailed protocol

All procedures were conducted in accordance with NIH guidelines and approved by the Institutional Animal Care and Use Committee of Northwestern University. Opn4Cre and Brn3bz-dta were maintained on a mixed C57Bl/6J; 129Sv/J background (Hattar et al., 2002; Hattar et al., 2006; Mu et al., 2005). Male and female littermate Opn4Cre/+ and Opn4Cre/+; Brn3bz-dta/+ animals between the ages of 2 and 3 months were used for sleep analysis.

Sleep recording

Request a detailed protocol

Male and female littermate Opn4Cre/+ and Opn4Cre/+; Brn3bz-dta/+ mice were used for sleep recordings. Electroencephalogram (EEG) and electromyogram (EMG) electrode implantation was performed simultaneously at 8 weeks of age. Mice were anesthetized with a ketamine/xylazine (98 and10 mg/kg respectively) and a 2-channel EEG and 1-channel EMG implant (Pinnacle Technology) was affixed to the skull. Mice were transferred to the sleep-recording cage 6 days after surgery, tethered with a preamplifier, and allowed 3 days to acclimate to the new cage and tether. Mice were housed in 12:12 light/dark conditions before and after EEG implantation. EEG and EMG recording began simultaneously at the end of the habituation period, which were displayed on a monitor and stored in a computer for analysis of sleep states. The high pass filter setting for both EEG channels was set at 0.5 Hz and low pass filtering was set at 100 Hz. EMG signals were high pass filtered at 10 Hz and subjected to a 100 Hz low pass cutoff. EEG and EMG recordings were collected in PAL 8200 sleep recording software (Pinnacle Technology) and scored, using a previously described, multiple classifier, automatic sleep scoring system, into 10 s epochs as wakefulness, NREM sleep, or REM sleep on the basis of rodent sleep criteria (Gao et al., 2016). Light source for all sleep experiments was a 3000 Kelvin light source at 500 lux.

Wheel-running activity and masking experiment

Request a detailed protocol

Mice were placed in cages with a 4.5-inch running wheel, and their activity was monitored with VitalView software (MiniMitter). Analyses of wheel running activity were calculated with ClockLab (Actimetrics). We used 500 lux light intensity. Mice were initially placed under 12:12 LD masking experiments. Mice were exposed, in their home cage, to a timer-controlled 3 hr light pulse at ZT14-ZT17. Percent activity for each mouse was normalized to its own activity at ZT13 (1 hr before light pulse), and analyzed in 30 min bins.

Tissue staining and imaging

Request a detailed protocol

Animals were anesthetized with Avertin and euthanized prior to fresh dissection of retinas in PBS. Retinas were fixed in 4% paraformaldehyde (Sigma) for at least 1 hr on ice. Retinas were then washed in PBS before staining overnight in anti-OPN4 antibody (1:1000, Advanced Targeting Systems) and then washed prior to 2 hr in secondary antibody (1:1000 goat anti-rabbit AlexaFluor 488, Life Technologies). Retinas were then flat-mounted on slides and imaged on a Zeiss LSM 710 confocal microscope.

Statistics

Request a detailed protocol

All statistical tests were performed in Graphpad Prism or R 3.4.4. Specific tests are listed in the text and figure legends. Linear mixed models were performed with the R packages lme4 1.1–21 and emmeans 1.3.4.

Data availability

Request a detailed protocol

All raw data are linked to this manuscript and available online.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
    A path to sleep is through the eye
    1. LP Morin
    (2015)
    eNeuro, 2, 10.1523/ENEURO.0069-14.2015, 26464977.
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
    Central circuitries for body temperature regulation and fever
    1. K Nakamura
    (2011)
    American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 301:R1207–R1228.
    https://doi.org/10.1152/ajpregu.00109.2011
  41. 41
  42. 42
  43. 43
    Masking by light in hamsters with SCN lesions
    1. U Redlin
    2. N Mrosovsky
    (1999)
    Journal of Comparative Physiology A: Sensory, Neural, and Behavioral Physiology 184:439–448.
    https://doi.org/10.1007/s003590050343
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
    Relationship between alertness, performance, and body temperature in humans
    1. KP Wright
    2. JT Hull
    3. CA Czeisler
    (2002)
    American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 283:R1370–R1377.
    https://doi.org/10.1152/ajpregu.00205.2002

Decision letter

  1. Stephen Liberles
    Reviewing Editor; Harvard Medical School, United States
  2. Catherine Dulac
    Senior Editor; Harvard University, United States
  3. Stephen Liberles
    Reviewer; Harvard Medical School, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Distinct ipRGC subpopulations mediate light's acute and circadian effects on body temperature and sleep" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Stephen Liberles as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Catherine Dulac as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

The reviewers were generally positive about the submitted work, but also raised important caveats that would need to be addressed prior to publication. In particular, all three reviewers raised concerns about the specificity of gain-of-function experiments involving DREADDs. Furthermore, reviewer #2 raised an important statistical concern questioning the validity of results involving ablation of Brn3b(+) ipRGCs. Some other information about Brn3b expression, and model validation is also requested. Finally, two of the reviewers thought that additional positive insights into brain regions that may mediate observed effects would broaden the impact and appeal of the paper, but after discussion, it was recognized that this would add significant time for additional experiments, so we leave it only as a suggestion to you for improving your paper rather than as an essential revision. I attached the full reviews below in case they are helpful, and also indicate which requests were considered essential.

Essential revisions:

1) Address questions about the gain-of-function DREADD model used. (See reviewer #1 comment #2, reviewer #2 comment #2, and reviewer #3 comment #5).

2) Address statistical concerns (see reviewer #2, comment #1 and reviewer #3, comment #4).

3) Provide additional information about Brn3b(+) ipRGCs. (See reviewer #1 comment #3)

Reviewer #1:

This nice study provides evidence that the acute effects of melanopsin on body temperature and sleep regulation are mediated by ipRGCs that do not target the SCN. The topic would be interesting to the broad readership of eLife, and the data appear generally convincing with experiments well performed. This manuscript relies predominantly on loss-of-function and gain-of-function manipulations of ipRGC subpopulations containing Brn3b, and some further characterization of mouse models is needed to support conclusions. Also, the Brn3b population still seems quite broad, so positive identification of brain regions relevant for observed effects would strengthen the paper.

1) Brn3b(+) ipRGCs project to multiple brain regions; evidence in this paper is largely negative for SCN-projecting ipRGCs, but does not provide positive evidence for other brain regions. It seems tractable to address this using the DREADD-based approach presented in Figure 3. For example, Cre-dependent DREADD-encoding AAVs could be injected into various recipient brain regions of Opn4Cre mice. (One caveat of such an experiment is if individual ipRGCs project to multiple target nuclei, but it would help strengthen the case against SCN involvement).

2) Some validation of the mice used in Figure 3 would be helpful; after AAV injection in Brn3b-Cre mice, are DREADDs expressed in other retinal cell types? Are DREADDs expressed centrally in the brain?

3) Some background information about Brn3b expression would be helpful in the Introduction. What% of all ipRGCs express Brn3b, and what% of those that target the SCN and other key nuclei express Brn3b?

Reviewer #2:

In this study the authors use intersectional transgenic to determine that ipRGCs contribute to acute changes in body temperature and sleep that are triggered by light. Furthermore, they determine that the population of ipRGCs that mediates these acute change in temperature and sleep is distinct from the population that mediates photoentrainment of circadian rhythms-circadian rhythms instead account for the modulation of body temperature and sleep on a circadian (24 hour) timescale. The evidence to support this conclusion is threefold. First, Opn4ko mice lack acute, light-induced changes in body temperature, while mice lacking rod- or cone-mediated phototransduction retain these acute light-induced changes. This indicates that ipRGCs whose light response is dominated by melanopsin are responsible for mediating this effect. Second, mice that lack the ipRGCs that project to non-SCN targets in the brain also lack acute-light-induced changes in body temperature. Note, in this case, the ipRGCs are killed by diptheriatoxin. Third, activation of all Brn3b(+) RGCs (which is most of them), including Brn3b(+) ipRGCs, via DREADDs induced a drop in body temperature. Finally, the authors show that killing Brn3b(+) ipRGCs prevents acute light-induced sleep, though it is not clear whether this is via the same pathway as acute light-induced changes in temperature.

The observation that different ipRGC-subtypes mediate photoentrainment of circadian rhythm vs. light-induced acute changes in sleep and body temperature through distinct circuits is significant. The data are clearly presented. However, I have a major concern about the primary effect as described in comment #1 below. Comment #2 is also important but more easily addressed.

1) Figure 2G: The authors argue that there is no acute change in body temperature in mice lacking Brn3b(+) ipRGCs. However, 5/7 mice still have a temperature decrease that is comparable in size to the temperature decrease in control mice. Therefore, the correct comparison to make to test their hypothesis is between the change in body temperature in the Opn4Cre/+ Brn3bDTA/+ mice vs. the Opn4Cre/+ (control) mice. This is a major comment that needs to be addressed by the authors.

2) Figure 3: For the DREADD experiment, hM3Dq is expressed in all RGCs that are Brn3b(+). Hence, the induction of temperature changes could be via an entirely different pathway, potentially involving the dLGN. For this reason, it would be important for the authors to state exactly how much transfection they saw and what percent of ipRGCs express the DREAAD. For example, the authors should explain whether their 50% transfection criterion is 50% of ipRGCs or of all RGCs? (subsection “Body temperature recordings”, second paragraph). The authors need to include this caveat in the discussion of this result. In addition, the example image is not large enough for the reader to draw any conclusion about the expression of hM3Dq, except that it is widespread. A larger image and annotations where overlap exists between anti-Opn4 antibody and mCherry would help the interpretation of this figure.

Reviewer #3:

The authors investigate the contribution of melanopsin expressing intrinsically photosensitive ganglion cells to two aspects of photo-regulated behavior- temperature and sleep. They conduct genetic and functional experiments in an attempt to demonstrate that Brn3b(+)melanopsin RGCs are responsible for 'acute' responses to a light pulse. There is much to like about this paper, including the compelling nature of the question. The paper is also well written. However, there are several issues with the manuscript that result in a somewhat lukewarm view of the paper in its current form for eLife. These include the following:

1) Scientific claims of the paper appear overstated. In the Abstract the authors write that "body temperature and sleep responses to light are absent after genetic ablation of all ipRGC." Given this, one would expect dramatically reduced responses to light. Instead, nearly all light induced temperature and sleep responses are intact in each of the mutant genotypes analyzed. The difference that are observed are relatively small and restricted to turning on the lights for 3 hours at night. The title also claims that "distinct ipRGC subpopulations" regulate light's acute affect. The Brn3b-DTA model they use ablates nearly all melanopsin (likely upwards of ~90%, Discussion). To call this a distinct subset seems a bit problematic. This leads to issue two:

2) Somewhat incremental advance in the field. The primary news of the manuscript seems to be that melanopsin is not required for the majority of body temperature regulation by light (leaving one wondering what is) but only for 'acute' responses when light is turned on at a specific point in the dark cycle. This finding, while interesting, does not dramatically move the field forward.

3) Sparse data. Overall, the depth of the experiments and statistical rigor leaves some room for improvement. There are many more experiments the authors themselves mention that would greatly improve the quality of the manuscript. Chief among these are data on true melanopsin ipRGC subtypes and their role in these responses. In parallel the authors do not address the role of particular brain regions in these behaviors. While these studies are difficult, they would greatly elevate this paper. But several other simpler experiments would also help. These include: (1) repeating the studies using only wavelengths of light to which only melanopsin responds to rule out effects from other opsins (479nm or blue light); (2) performing tracing experiments to the brain from single ipRGCs; and (3) a more complete study of the 'acute' light responses which could include examining the effects of a shorter light burst.

4) Lack of appropriate statistical comparisons. The key conclusions of the central figures rely on differences between changes in body temperature between control and experimental groups. While the authors statistically compare light and dark temperatures within each group, they do not appear to compare between them in Figures 1, 2, or 3. This would be required to validate the central claims of the figures and includes statistically comparisons of data in Figure 1G and I, Figure 2D and F, and Figure 3D and F.

5) Problematic gain of function experiments. In Figure 3, the authors attempt to show that activation of Brn3b(+) melanopsin+ RGCs induces temperature decreases. However, from the image shown, less than ~10% of OPN4+ ipRGCs are transduced, and even more problematic, ~90% of the transduced RGCs are not OPN4+ ipRGCs. Thus, the main conclusion we can draw is that RGC activation broadly induces a temperature decrease. This experiment would need to be repeated using the OPN4Cre driver that is featured in the other figures.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for submitting your article "Distinct ipRGC subpopulations mediate light's acute and circadian effects on body temperature and sleep" for consideration by eLife. Your article has been reviewed by one peer reviewer (prior reviewer #2), and the evaluation has been overseen by a Reviewing Editor and Catherine Dulac as the Senior Editor. The reviewer has opted to remain anonymous.

The Reviewing Editor has drafted this decision to help you prepare a revised submission.

A major concern of reviewer #2 persists (please see comment #1 below). Perhaps repeating the experiment with a new cohort of mice would clarify. Also, the new supplementary figure (Figure 3—figure supplement 1D) seems to indicate that CNO exerts an effect in control mice (comment #3 below), an issue which needs to be addressed. Finally, please note other comments related to statistical comparisons (points #4,5).

Reviewer #2:

1) The primary concern regarding the data in Figure 2 is not fully addressed. The authors present the data as before and it still appears that 5/7 Brn3b-DTA mice had a temperature decrease comparable to that of control mice. Also, two control mice have a temperature difference that is less than that of the Brn3b-DTA mice.

The authors did add a new supplementary figure (Figure 2—figure supplement 1) that would directly compare the temperature changes across genotypes and resolve our concern. However, if these are the same mice that are plotted in Figure 2, then some mistake has been made or the data are analyzed in a different way that is not clearly explained. What the reader expects is for the changes in body temperature from the dark to light condition portrayed in Figure 2E (Control mice) to be compared to those from Figure 2G (Brn3b-DTA mice) but the body temperature changes in Figure 2—figure supplement 1 appear to originate from a different data set. For example, Figure 2—figure supplement 1 portrays two Control mice with a temperature decrease of more than 1 degree, but there is no Control mouse in Figure 2E that exhibited a -1 degree change in temperature, let alone a -1.5 degree change. The authors need to look at their data closely – either what is plotted in Figure 2 or what is plotted in Figure 2—figure supplement 1 is incorrect.

2) The authors have provided a quantification of the RGCs that express the DREAADs for the gain-of-function experiment in Figure 3. This is important to provide, and it appears that 50% of all RGCs are expressing the DREAAD. The authors make the argument that the DREAADs expressed in ipRGCs are likely the only impactful ones based on their finding that melanopsin expression by ipRGCs is required for the change in body temperature evoked by a light pulse during subjective night and that ipRGCs are on their own sufficient to evoke the change in body temperature in the absence of rod and cone inputs to the retina. The authors also provide a good discussion of the potential caveats of this approach. Though the experiment is not perfect, making it perfect would require development of new technology. I am satisfied with the current description of results.

3) It appears in Figure 3—figure supplement 1 that CNO evokes a small decrease in body temperature (~0.5 degrees) in mice that do not express the receptor hM3D(Gq) (control mice). This effect is in the same direction and of similar magnitude to the effect of CNO injection in mice that do express the receptor (Brn3b-Cre mice). The authors need to compare these two populations directly rather than compare each population to the PBS control. This will rule out an effect on body temperature coming from off-target effects of CNO.

4) The figure legends should state whether the error shading around the body temperature traces is SEM or SD, starting with Figure 1.

5) In several places, linear mixed models are used in the context of performing statistical comparisons (Figure 1—figure supplement 2, Figure 2—figure supplement 1, etc.). Some details should be provided in the Materials and methods section to describe how these linear models work.

Note, these last two points are both getting at the same point. The authors indicate in their response that the effects are big and reliable. Indeed many of the example traces indicate that is the case. Yet somehow this large, reliable effect is not translating into their summary data, which appears to be highly variable. In addition, there appears to be a distinct temperature change during the light pulse in subjective night in the Opn4 KO mouse (Figure 1), yet the authors describe it as being "absent". Either is fine – large and reliable, or highly variable, but the description of the effect needs to be consistent.

https://doi.org/10.7554/eLife.44358.023

Author response

Essential revisions:

1) Address questions about the gain-of-function DREADD model used. (See reviewer #1 comment #2, reviewer #2 comment #2, and reviewer #3 comment #5).

Before directly addressing the DREADD experiments and associated revisions, we would like to highlight the evidence (in this and other papers) demonstrating that ipRGCs are required for acute thermoregulation and sleep induction by light. First, we find that lack of melanopsin results in loss of changes in body temperature to a light pulse at ZT14-16 (Figure 1), as does ablation of Brn3b(+) ipRGCs (Figure 2). Second, previous studies have demonstrated that both loss of melanopsin expression and ablation of ipRGCs results in a loss of acute light induction of sleep, demonstrating that ipRGCs are necessary for this behavior (Altimus et al., 2008). Thus, these data indicate that ipRGCs are necessary for the acute light-induced changes in body temperature and sleep, making it likely that activation of Brn3b(+) RGCs (Figure 3) is driving effects on body temperature through ipRGCs. Nonetheless, as discussed below, we do provide an expanded discussion of the important caveats of this manipulation.

2) Address statistical concerns (see reviewer #2, comment #1 and reviewer #3, comment #4).

3) Provide additional information about Brn3b+ ipRGCs. (See reviewer #1 comment #3)

Reviewer #1:

[…]

1) Brn3b(+) ipRGCs project to multiple brain regions; evidence in this paper is largely negative for SCN-projecting ipRGCs, but does not provide positive evidence for other brain regions. It seems tractable to address this using the DREADD-based approach presented in Figure 3. For example, Cre-dependent DREADD-encoding AAVs could be injected into various recipient brain regions of Opn4Cre mice. (One caveat of such an experiment is if individual ipRGCs project to multiple target nuclei, but it would help strengthen the case against SCN involvement).

We agree that understanding which brain regions drive these behaviors is an incredibly important question. There are multiple potential candidates that we have outlined in the Discussion, and we are actively following up on these questions in separate studies. We agree with the consensus reached by the reviewers after discussion that this would be beyond the scope of the current study.

2) Some validation of the mice used in Figure 3 would be helpful; after AAV injection in Brn3b-Cre mice, are DREADDs expressed in other retinal cell types? Are DREADDs expressed centrally in the brain?

We have performed additional quantification of the overlap between the AAV reporter and Opn4+ cells. We have also quantified the fraction of Opn4+ cells that are transduced. Overall, the proportion of Opn4+ ipRGCs that are show detectable mCherry fluorescence is ~40%, which is well in line with what we would expect: M1 and M2 ipRGCs represent the vast majority of Opn4+ immunopositive RGCs. M2 ipRGCs represent up to 50% of Opn4+ RGCs and are all Brn3b(+) (Chen et al., 2011). M1 ipRGCs represent at least 50% of Opn4+ RGCs, but just 15% of adult M1 ipRGCs retain Brn3b expression beyond development (Chen et al., 2011). Thus, the maximum possible expression levels we would expect is ~60% of Opn4+ RGCs that label with the reporter. Coupled with the unavoidable fact that some cells will simply not be infected with the AAV, these expression patterns are aligned with our expectations. Additionally, the vast majority of DREADD-expressing RGCs in the retina are not Opn4+. This is expected given that 80% of all RGCs are Brn3b(+). However, as mentioned above, we think it unlikely (though of course not impossible) that these non-ipRGCs drive the associated temperature changes. These data are now included as Figure 3—figure supplement 1.

DREADD expression is confined to the eye because we injected the AAVs directly into the eye. We have never seen evidence of AAV infection in the brain using this experimental paradigm (Takuma Sonoda, Unpublished Data) as many other laboratories who carry out such experiments. Therefore, we are confident that the expression is restricted to the eye.

3) Some background information about Brn3b expression would be helpful in the Introduction. What% of all ipRGCs express Brn3b, and what% of those that target the SCN and other key nuclei express Brn3b?

We thank the reviewer for suggesting additional information be added about Brn3b(+) ipRGCs. We have added additional, and more explicit, explanation to the Results. This enhanced description now better describes the Brn3b(+) and Brn3b(–) ipRGC populations, the subtypes that fall into each category, and their known behavioral functions. We chose to add it in the Results rather than the Introduction because that is where we begin manipulating ipRGCs based on their expression of Brn3b.

Reviewer #2:

[…] The observation that different ipRGC-subtypes mediate photoentrainment of circadian rhythm vs. light-induced acute changes in sleep and body temperature through distinct circuits is significant. The data are clearly presented. However, I have a major concern about the primary effect as described in comment #1 below. Comment #2 is also important but more easily addressed.

1) Figure 2G: The authors argue that there is no acute change in body temperature in mice lacking Brn3b(+) ipRGCs. However, 5/7 mice still have a temperature decrease that is comparable in size to the temperature decrease in control mice. Therefore, the correct comparison to make to test their hypothesis is between the change in body temperature in the Opn4Cre/+ Brn3bDTA mice vs. the Opn4Cre/+ (control) mice. This is a major comment that needs to be addressed by the authors.

We thank both reviewer #2 and reviewer #3 for raising this important point. Most importantly for the central conclusions of this paper: we have now compared the change in body temperature in darkness versus light pulse both within Brn3bDTA/+ animals and between Control and Brn3bDTA/+ animals. In support of our initial conclusions, we find no significant change in body temperature in Brn3bDTA/+ relative to the dark condition. We also find that the Cre/+ shows a significantly larger change in body temperature in response to a light pulse compared to Brn3bDTA/+. We have plotted these changes as well as the change from baseline in both conditions in Figure 2—figure supplement 1.

As requested, we performed similar comparisons for all genotypes used for thermoregulation and sleep experiments in the paper, and are happy to report that the results are each consistent with our initial interpretations. The results can be found in the following figures: Figure 1—figure supplement 2, Figure 2—figure supplement 1, Figure 3—figure supplement 1, and Figure 4—figure supplement 2. We thank the reviewers for suggesting these analyses as they greatly strengthen our conclusions.

2) Figure 3: For the DREADD experiment, hM3Dq is expressed in all RGCs that are Brn3b(+). Hence, the induction of temperature changes could be via an entirely different pathway, potentially involving the dLGN. For this reason, it would be important for the authors to state exactly how much transfection they saw and what percent of ipRGCs express the DREAAD.

We thank the reviewer for highlighting this important point. Though we think it unlikely that a distinct pathway is involved (given the two points mentioned above regarding the necessity of ipRGCs for acute thermoregulation by light), we now quantify the number of cells that are Brn3b(+), Opn4(–); Brn3b(+), Opn4(+), and Brn3b(–), Opn4(+) as well as the proportion of Opn4 immunopositive cells that express Gq-DREADDs. Though ipRGCs were sometimes more faintly labeled than other RGCs, we were still able detect some mCherry expression in about 40% of Opn4+ cells. These are now identified in Figure 3—figure supplement 1 with arrows.

For example, the authors should explain whether their 50% transfection criterion is 50% of ipRGCs or of all RGCs? (subsection “Body temperature recordings”, second paragraph).

We have added a clarification to this statement (50% of all RGCs assessed by fluorescence detectable across more than half of the retina).

The authors need to include this caveat in the discussion of this result.

This is an important point to highlight for the general readership of eLife, and we have added additional discussion of the caveats of this experiment, and thank the reviewer for raising these important points.

As mentioned earlier, we find that lack of melanopsin results in loss of changes in body temperature to a light pulse at ZT14-16 (Figure 1), as does ablation of Brn3b(+) ipRGCs (Figure 2). Moreover, previous studies have demonstrated that both loss of melanopsin expression and ablation of ipRGCs results in a loss of acute light induction of sleep, demonstrating that ipRGCs are necessary for this behavior (Altimus et al., 2008). Thus, these data indicate that ipRGCs are necessary for the acute light-induced changes in body temperature. We therefore chose to next activate Brn3b(+) RGCs because if ipRGCs are in fact required for acute thermoregulation by a light pulse at ZT14-16, then we would expect activating these same cells to mimic the effects of light and cause a subsequent decrease in body temperature. Our finding that activation of Brn3b(+) RGCs causes a reduction in body temperature at ZT14 are therefore consistent with our model. However, we are mindful of the fact that we are activating many non-ipRGCs with this manipulation. When taken in conjunction with our results from Figures 1 and 2 showing ipRGCs are necessary for this behavior, we think it unlikely that a separate pathway is driving the temperature decrease in Figure 3.

In addition, the example image is not large enough for the reader to draw any conclusion about the expression of hM3Dq, except that it is widespread. A larger image and annotations where overlap exists between anti-Opn4 antibody and mCherry would help the interpretation of this figure.

We have added additional examples with images in Figure 3—figure supplement 1.

Reviewer #3:

[…] There are several issues with the manuscript that result in a somewhat lukewarm view of the paper in its current form for eLife. These include the following:

1) Scientific claims of the paper appear overstated. In the Abstract the authors write that "body temperature and sleep responses to light are absent after genetic ablation of all ipRGC." Given this, one would expect dramatically reduced responses to light. Instead, nearly all light induced temperature and sleep responses are intact in each of the mutant genotypes analyzed. The difference that are observed are relatively small and restricted to turning on the lights for 3 hours at night. The title also claims that "distinct ipRGC subpopulations" regulate light's acute affect. The Brn3b-DTA model they use ablates nearly all melanopsin (likely upwards of ~90%, Discussion). To call this a distinct subset seems a bit problematic.

We agree with the reviewer that we are ablating a large number of ipRGCs in the Brn3bDTA animal. However, it is the combination of identifying a subset of ipRGCs that is required for these behaviors in combination with the fact that the light input for these acute, light-evoked behaviors does not appear to be routed through the SCN as previously believed. This is a significant departure from the current models.

This leads to issue two:

2) Somewhat incremental advance in the field. The primary news of the manuscript seems to be that melanopsin is not required for the majority of body temperature regulation by light (leaving one wondering what is) but only for 'acute' responses when light is turned on at a specific point in the dark cycle. This finding, while interesting, does not dramatically move the field forward.

By acute, we mean that the light pulse is relatively short term and not recurring in a cyclic manner. Relative to the 12 hour light/dark cycle, the single, 3 hour stimulus is acute. This type of stimulus is meant to mimic unexpected encounters with environmental light out of sync with the environmental light/dark cycle. As mentioned above, it was previously believed that light input from the retina to the SCN drove both the circadian and acute responses to environmental light. Our data show that this is not the case. Thus, we have shown that light input to the SCN alone is not sufficient for light’s acute effects, and attributed these effects to a specific subset of ipRGCs. These findings have important implications for understanding how light information is integrated over different timescales, and identifies a subset of RGCs to target for future studies.

3) Sparse data. Overall, the depth of the experiments and statistical rigor leaves some room for improvement. There are many more experiments the authors themselves mention that would greatly improve the quality of the manuscript. Chief among these are data on true melanopsin ipRGC subtypes and their role in these responses. In parallel the authors do not address the role of particular brain regions in these behaviors. While these studies are difficult, they would greatly elevate this paper. But several other simpler experiments would also help. These include: (1) repeating the studies using only wavelengths of light to which only melanopsin responds to rule out effects from other opsins (479nm or blue light); (2) performing tracing experiments to the brain from single ipRGCs; and (3) a more complete study of the 'acute' light responses which could include examining the effects of a shorter light burst.

1) Regarding the influence of other options: Our data clearly show that absence of melanopsin (Figure 1) leads to a severe deficit in light-evoked temperature decreases to a 3 hour light pulse at ZT14-16, ruling out a sufficiency of other opsins in driving this behavior. Thus, we feel that the use of additional wavelengths would be somewhat redundant.

2) We agree with the author that understanding the connectivity of the circuit will be very important in the future, but we do not yet have a means to label single RGC axons arising from a defined population, and have not identified the pertinent brain regions for thermoregulation and sleep by light as of yet.

3) We chose this three hour light pulse given its standard use as a “masking” pulse in previous studies. This allowed us to compare more directly with previous studies of circadian photoentrainment, negative masking, and sleep induction by light (Mrosovsky et al., 1999; Mrosovsky and Hattar, 2003; Altimus et al., 2008; Lupi et al., 2008). We agree that there are many interesting experiments that could be done to test the sensitivity of the system to light stimuli of varying lengths and intensities.

4) Lack of appropriate statistical comparisons. The key conclusions of the central figures rely on differences between changes in body temperature between control and experimental groups. While the authors statistically compare light and dark temperatures within each group, they do not appear to compare between them in Figures 1, 2, or 3. This would be required to validate the central claims of the figures and includes statistically comparisons of data in Figure 1G and I, Figure 2D and F, and Figure 3D and F.

We thank the reviewer for this suggestion. Please see our response to reviewer #2, comment #1.

5) Problematic gain of function experiments. In Figure 3, the authors attempt to show that activation of Brn3b+ melanopsin+ RGCs induces temperature decreases. However, from the image shown, less than ~10% of OPN4+ ipRGCs are transduced, and even more problematic, ~90% of the transduced RGCs are not OPN4+ ipRGCs. Thus, the main conclusion we can draw is that RGC activation broadly induces a temperature decrease. This experiment would need to be repeated using the OPN4cre driver that is featured in the other figures.

The reviewer raises important points. As mentioned in our responses to reviewer #1, comment 2 and reviewer #2, comment #2, we have now provided additional quantification of the infection rate as well as added additional discussion of the caveats of this experiment. We have chosen not to use Opn4Cre as a driver because it does not get at the central question of this paper, which is whether there are separate circuits for acute versus circadian effects of light. Activation of all ipRGCs would not allow us to differentiate between the circuits underlying these two effects because we would also be activating Brn3b-negative ipRGCs, which we know are sufficient for circadian photoentrainment of body temperature to light/dark cycles. Therefore, it would be difficult to interpret these experiments, regardless of the outcome.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Reviewer #2:

1) The primary concern regarding the data in Figure 2 is not fully addressed. The authors present the data as before and it still appears that 5/7 Brn3b-DTA mice had a temperature decrease comparable to that of control mice. Also, two control mice have a temperature difference that is less than that of the Brn3b-DTA mice.

The authors did add a new supplementary figure (Figure 2—figure supplement 1) that would directly compare the temperature changes across genotypes and resolve our concern. However, if these are the same mice that are plotted in Figure 2, then some mistake has been made or the data are analyzed in a different way that is not clearly explained. What the reader expects is for the changes in body temperature from the dark to light condition portrayed in Figure 2E (Control mice) to be compared to those from Figure 2G (Brn3b-DTA mice) but the body temperature changes in Figure 2—figure supplement 1 appear to originate from a different data set. For example, Figure 2—figure supplement 1 portrays two Control mice with a temperature decrease of more than 1 degree, but there is no Control mouse in Figure 2E that exhibited a -1 degree change in temperature, let alone a -1.5 degree change. The authors need to look at their data closely-either what is plotted in Figure 2 or what is plotted in Figure 2—figure supplement 1 is incorrect.

We now realize that we did not clearly explain our new analysis, leading to confusion about the data and analyses. Throughout the paper we have performed two distinct analyses on the same dataset. In the original submission and in the data plotted in Figure 1H, J and Figure 2E, G, we have compared the mean of the absolute body temperature on the “Dark” (control) night from ZT14-17 to the mean of the absolutebody temperature on the “Light” night from ZT14-17 (see Author response image 1). If light induces a decrease in body temperature, then we should see significantly lower temperatures in the “Light” condition than in the “Dark” condition for each genotype. We did not see a significant decrease in melanopsin null animals (Figure 1J) or in Brn3b-DTA animals (Figure 2G, also shown in Author response image 1). The reviewers noted that the mean in the “Light” compared to the mean in the “Dark” across the two nights is lower for some Brn3b-DTA animals, but this did not reach statistical significance for the population.

An excellent suggestion from the reviewers for the second submission to ensure that there was no actual light evokedchange in body temperature given the trends mentioned above, was to more directly measure temperature changes in response to the light pulse instead of comparing the same time frame across days. Therefore, we performed new analyses of all of the data, which are represented in Figure 1—figure supplement 2, Figure 2—figure supplement 1, and Figure 3—figure supplement 1. In these analyses, we performed a completely distinct set of comparisons where we compared the change in body temperature of mice on the “Control” (Dark night) from ZT14 (which served as a baseline temperature) to the mean body temp during the entire time period of ZT14-17. We then compared that change to the change in body temperature on the “Light” night between the ZT14 baseline and the time period of ZT14-17 during which the mice were exposed to a light pulse (see Author response image 1). This allowed us to directly measure changes in body temperature from the baseline temperature in response to the light pulse. In this case, if light causes a decrease in body temperature then we would expect the change in body temperature in “Light” to be significantly different from the change in the “Ctrl” condition within a genotype. This analysis allows us to compare how the temperature actually changed within the same animal, on the same night, in response to the light pulse itself. This method accounts for any light-independent changes in body temperature due to, for example, daily variation. These new analyses supported the same conclusions as those from the first submission: there was no change in body temperature in either melanopsin null or Brn3b-DTA animals. As an example, we have included the plots for the Control versus Brn3b-DTA genotypes from Figure 2—figure supplement 1 in Author response image 1. Here you can see that for the Brn3b-DTA animals, many animals showed a similar decrease in body temperature compared to their temperature at ZT14 on both the Ctrl and Light nights. In other words, there was no further decrease in body temperature in the presence of light, arguing against a light-induced change in body temperature, and supporting our original conclusions for not only Brn3b-DTA animals (Figure 2—figure supplement 1), but also melanopsin null animals (Figure 1—figure supplement 2). Because of the different comparisons, the data points from the two analyses we performed (Mean Body Temperature versus Mean Change in Body Temperature) cannot be directly compared and will not give identical amplitudes, which we believe was the source of confusion for the reviewer. We apologize for the confusion and have attempted to clarify this in the figure legends. Importantly, these data for not only Brn3b-DTA, but also each of our other mutant lines, are in complete support of our original analyses, and we thank the reviewers for suggesting these additional comparisons. We have provided a schematic of the comparisons in Author response image 1.

Author response image 1
Schematic of analyses throughout the manuscript with example graphs from Brn3bDTA analyses.

Regarding the reviewer’s observations of temperature decreases in Brn3bDTA animals: we hope that our new explanation clarifies the confusion. Brn3b-DTA mice did not display body temperature reductions similar to Controls because Figure 2—figure supplement 1 shows that the drop in body temperature in Brn3b-DTA mice on the “Light” night is completely similar in magnitude to their normal change in body temperature relative to ZT14 on the “Control” night where no light pulse was given (and also match the normal changes observed on the “Control” night in control animals). Additionally, no Brn3bDTA animals reach the average value of the control animal body temperature drop during the “Light.” We thank the reviewers for suggesting this new analysis because it allowed us to more directly compare natural temperature variations to those induced by light on the same night, and more strongly supports our original conclusions. We feel that the additional clarifications for the measurement methods in the figure captions will help clarify this for readers.

Importantly, we have included all raw data with the submission of this manuscript, so interested readers will be able to investigate changes in individual animals if they should wish.

2) The authors have provided a quantification of the RGCs that express the DREAADs for the gain-of-function experiment in Figure 3. This is important to provide, and it appears that 50% of all RGCs are expressing the DREAAD. The authors make the argument that the DREAADs expressed in ipRGCs are likely the only impactful ones based on their finding that melanopsin expression by ipRGCs is required for the change in body temperature evoked by a light pulse during subjective night and that ipRGCs are on their own sufficient to evoke the change in body temperature in the absence of rod and cone inputs to the retina. The authors also provide a good discussion of the potential caveats of this approach. Though the experiment is not perfect, making it perfect would require development of new technology. I am satisfied with the current description of results.

We are happy that the reviewer is satisfied with the additional quantification and discussion of this approach.

3) It appears in Figure 3—figure supplement 1 that CNO evokes a small decrease in body temperature (~0.5 degrees) in mice that do not express the receptor hM3D(Gq) (control mice). This effect is in the same direction and of similar magnitude to the effect of CNO injection in mice that do express the receptor (Brn3b-Cre mice). The authors need to compare these two populations directly rather than compare each population to the PBS control. This will rule out an effect on body temperature coming from off-target effects of CNO.

In order to test for significant effects of CNO injection in the absence of DREADDs, we feel that the appropriate comparison is to perform a within animal comparison of the effects of PBS injection in an animal versus CNO injection within that same animal, which we have done for 9 animals in the Control group (i.e. each animal received an injection of PBS and an injection of CNO on separate occasions). This allows us to compare the effect of these two different compounds on the same animal. When we perform this experiment, we see no significant change in body temperature, though we see the trend that the reviewer is referring to. Importantly, when we perform these same within animal comparisons in Brn3bCre-GqDREADD mice, we do see a statistical difference in CNO versus PBS injection across 8 animals.

Of note regarding statistical power to compare CNO across the two genotypes: Our inability to detect statistically different effect in Brn3b + CNO and Control + CNO is due to this lack of pairing; our power analyses indicate that we would need to run > 40 mice per group to detect a difference given the magnitude and variance of this dataset.

Therefore, while the statistical model cannot rule out the possibility that CNO has a small effect on body temperature, we believe reasonable interpretation suggests an effect of Brn3b(+) RGC activation on body temperature. CNO injection in Brn3b-GqDREADD mice drives a robust reduction in body temperature that remains ~1°C lower than at injection time for ≥ 6 hours (Figure 3D). There does seem to be a drop in body temperature following CNO injection in Control mice, but it is roughly comparable to that seen with PBS injection and is within ~0.2°C of baseline within a few hours (Figure 3—figure supplement 2C). The difference in magnitude of the effect is borne out in the summary data as well, with approximate doubling of the average body temperature reduction for the 6 hours after injection in Brn3b-Cre vs. Control with CNO (Figure 3—figure supplement 1D).

4) The figure legends should state whether the error shading around the body temperature traces is SEM or SD, starting with Figure 1.

We apologize for this oversight. The figures have now been labeled with the appropriate variance measurement.

5) In several places, linear mixed models are used in the context of performing statistical comparisons (Figure 1—figure supplement 2, Figure 2—figure supplement 1, etc.). Some details should be provided in the Materials and methods section to describe how these linear models work.

We apologize that the details of the linear mixed models were not included in the Materials and methods. Linear mixed models are a statistical alternative to ANOVA models. They are preferable in instances of repeated measures because they are capable of dealing with missing values and of incorporating continuous covariates (such as time, light intensity, etc). All linear mixed models were carried out using the R packages lme4 1.1-21 and emmeans 1.3.4. These packages are now explicitly listed in the Materials and methods.

Note, these last two points are both getting at the same point. The authors indicate in their response that the effects are big and reliable. Indeed many of the example traces indicate that is the case. Yet somehow this large, reliable effect is not translating into their summary data, which appears to be highly variable. In addition, there appears to be a distinct temperature change during the light pulse in subjective night in the Opn4 KO mouse (Figure 1), yet the authors describe it as being "absent". Either is fine – large and reliable, or highly variable, but the description of the effect needs to be consistent.

https://doi.org/10.7554/eLife.44358.024

Article and author information

Author details

  1. Alan C Rupp

    Department of Biology, Johns Hopkins University, Baltimore, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5363-4494
  2. Michelle Ren

    Department of Neurobiology, Northwestern University, Evanston, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Cara M Altimus

    Department of Biology, Johns Hopkins University, Baltimore, United States
    Contribution
    Conceptualization, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Diego C Fernandez

    Department of Biology, Johns Hopkins University, Baltimore, United States
    Present address
    National Institute of Mental Health, Bethesda, United States
    Contribution
    Conceptualization, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Melissa Richardson

    Department of Biology, Johns Hopkins University, Baltimore, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Fred Turek

    Department of Neurobiology, Northwestern University, Evanston, United States
    Contribution
    Conceptualization, Supervision, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Samer Hattar

    1. Department of Biology, Johns Hopkins University, Baltimore, United States
    2. Department of Neuroscience, Johns Hopkins University, Baltimore, United States
    Present address
    National Institute of Mental Health, Bethesda, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3124-9525
  8. Tiffany M Schmidt

    Department of Neurobiology, Northwestern University, Evanston, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    tiffany.schmidt@northwestern.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4791-6775

Funding

Klingenstein-Simons Fellowship in the Neurosciences (Fellowship in the Neurosciences)

  • Tiffany M. Schmidt

Sloan Research Fellowship in Neuroscience (Research Fellowship in Neuroscience)

  • Tiffany M. Schmidt

National Institutes of Health (1DP2EY027983)

  • Tiffany M. Schmidt

National Institutes of Health (GM076430)

  • Samer Hattar

National Institutes of Health (EY024452)

  • Samer Hattar

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Ethics

Animal experimentation: This study was performed in accordance with the Institutional Care and Use Committees of Johns Hopkins and Northwestern Universities (IS00003845, IS00000887, IS00000745).

Senior Editor

  1. Catherine Dulac, Harvard University, United States

Reviewing Editor

  1. Stephen Liberles, Harvard Medical School, United States

Reviewer

  1. Stephen Liberles, Harvard Medical School, United States

Publication history

  1. Received: December 13, 2018
  2. Accepted: June 18, 2019
  3. Version of Record published: July 23, 2019 (version 1)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 1,295
    Page views
  • 232
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)