To determine the role of DMH^LepR neurons in feeding and metabolism, we used a viral loss-of-function approach (Figure 1A). Specifically, synaptic neurotransmission by DMH^LepR neurons was permanently blocked by bilateral microinjection into the DMH of an adeno-associated virus (AAV) encoding Cre-dependent tetanus toxin light chain fused with a GFP reporter (AAV1-CBA-DIO-GFP:TeTx; Kim et al., 2009; Han et al., 2015; Campos et al., 2017). Viral transduction was confirmed by histochemical detection of GFP in the DMH (Figure 1B–C); as expected, GFP was undetected in Cre-negative controls (not shown). Although GFP+ cell bodies were not detected outside of the DMH, abundant GFP+ terminals were detected in the ARC (Figure 1B–C), consistent with previous evidence of an inhibitory DMH^LepR→ ARC^AgRP neurocircuit implicated in feeding control (Garfield et al., 2016; Krashes et al., 2014).

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

Longitudinal measures following DMH-LepR TeTx.

https://cdn.elifesciences.org/articles/63671/elife-63671-fig1-data1-v2.xlsx

Download elife-63671-fig1-data1-v2.xlsx

Whereas previous evidence showed no effect of acute inhibition of DMH^LepR neurons on feeding (Garfield et al., 2016), chronic silencing of DMH^LepR neurons resulted in hyperphagia that was sustained for several days (Figure 1D–E). This transient hyperphagic response was associated with sustained weight gain (Figure 1F) and a modest increase in adipose mass (Figure 1G) that persisted despite daily food intake eventually falling below that of controls (Figure 1E). Consistent with the increased adiposity, we also detected modestly increased plasma leptin levels (Figure 1H) and elevated fasting levels of both blood glucose (control vs. TeTx: 72.0 ± 5.6 vs. 107.1 ± 7.3, t_9.969=3.816; p=0.003) and plasma insulin (control vs. TeTx: 0.49 ± 0.04 vs. 1.24 ± 0.12, t_6.092=5.807; p=0.001), suggestive of insulin resistance. These findings extend and refine previous work implicating a physiological role for DMH^LepR neurons in energy homeostasis (Garfield et al., 2016; Rezai-Zadeh et al., 2014).

Previous work has shown both that leptin treatment depolarizes DMH neurons expressing LepR (Simonds et al., 2014) and that leptin receptor signal transduction in GABAergic DMH neurons is required for the acute anorectic effect of leptin (Xu et al., 2018). To investigate the possibility that leptin’s anorectic effects involve activation of DMH^LepR neurons, we next tested whether DMH^LepR inactivation blunts leptin-mediated anorexia. First, the specificity of GFP:TeTx expression in DMH^LepR neurons was confirmed by establishing that leptin-induced pSTAT3, a marker of LepR signal transduction, colocalizes with virally transduced cells following systemic leptin injection (Figure 2A). Next, control and DMH^LepR-silenced mice were fasted for 24 hr, followed by i.p. injection of either leptin or saline vehicle, after which food was returned. Interestingly, control animals lost more weight during the fast (Figure 2B) and consequently exhibited a greater refeeding response following saline treatment than TeTx mice (Figure 2C; dashed bars). Nonetheless, the effect of leptin to further suppress food intake was readily detected in controls but absent in DMH^LepR-silenced mice (Figure 2C; solid bars). These findings extend previous evidence (Xu et al., 2018) suggesting a key role for DMH^LepR neurons in leptin-mediated suppression of fasting-induced refeeding.

Figure 2

Download asset Open asset

Figure 2—source data 1

Fast-refeeding +/- Leptin.

https://cdn.elifesciences.org/articles/63671/elife-63671-fig2-data1-v2.xlsx

Download elife-63671-fig2-data1-v2.xlsx

We note that since DMH^LepR neurons directly synapse onto and inhibit AgRP neurons (Garfield et al., 2016), TeTx-mediated DMH^LepR silencing is predicted to increase AgRP neuron activity. As AgRP activation not only increases feeding (Aponte et al., 2011) but also promotes de novo lipogenesis and suppresses lipolysis during fasting (Cavalcanti-de-Albuquerque et al., 2019), AgRP disinhibition offers a feasible explanation for the observed effects of DMH^LepR silencing not only to promote hyperphagia and weight gain, but also to preserve body weight during fasting. Future studies are warranted to test this possibility.

To determine whether the observed impairments in energy homeostasis were associated with changes in diurnal rhythmicity in normal (14:10) light–dark cycles, we obtained continuous measures of energy intake, energy expenditure, and locomotor activity (LMA) over several days using indirect calorimetry. We found that, unlike control mice which exhibited typical nocturnal feeding behavior, DMH^LepR-silenced mice exhibited a rapid (i.e., within 1 week; Figure 3—figure supplement 1C–D) and permanent phase advance in daily food intake (Figure 3A), such that dark-cycle food intake was less than light-cycle intake (Figure 3B). Similarly, while control mice displayed the expected increase in dark-cycle LMA, this pattern became undetectable within the first week after viral microinjection in DMH^LepR-silenced mice (Figure 3C–D; Figure 3—figure supplement 1G–H). Rhythms in other metabolic parameters were similarly shifted and blunted by DMH^LepR inactivation. Specifically, we found that heat production in DMH^LepR-silenced mice was reduced selectively in the dark cycle (Figure 3E–F) and respiratory-exchange ratio (RER) was elevated in the light cycle (Figure 3G–H). These metabolic responses likely reflect the shift of a substantial fraction of daily caloric intake from the dark to the light cycle in DMH^LepR-silenced mice (Figure 3A–B). Our finding that the rapid increase of RER following viral microinjection (Figure 3—figure supplement 1E–F) coincides with the timing of excess fat accumulation (Figure 1F; Figure 3—figure supplement 1B) suggests that DMH^LepR inactivation may also increase de novo lipogenesis, as is reported, following AgRP neuron activation (Garfield et al., 2016; Cavalcanti-de-Albuquerque et al., 2019).

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

Female calorimetry.

https://cdn.elifesciences.org/articles/63671/elife-63671-fig3-data1-v2.xlsx

Download elife-63671-fig3-data1-v2.xlsx

Together, these findings suggest that DMH^LepR neuron activity is required for the normal coupling of daily rhythms in feeding, LMA, and associated metabolic parameters to the light–dark cycle. Whether and how DMH^LepR neuron activity influences these parameters in the absence of a normal light–dark cycle (i.e., in constant darkness) or in alternate light–dark schedules (e.g., 12:12 vs. 14:10 employed here) are important unanswered questions.

Although female DMH^LepR-silenced mice did not exhibit the transient hyperphagia observed in males (Figure 1—figure supplement 2A–B), they nonetheless developed a similar degree of obesity (Figure 1—figure supplement 2C–D) that was associated with disrupted diurnal rhythms in food intake (Figure 3—figure supplement 2A–B), LMA (Figure 3—figure supplement 2C–D), heat production (Figure 3—figure supplement 2E–F), and RER (Figure 3—figure supplement 2G–H), similar to responses observed in male DMH^LepR-silenced mice. The key role for DMH^LepR neurons in diurnal behavioral and metabolic control identified in males, therefore, extends to females as well. Given that, compared to male mice (Chao et al., 2011), female mice are protected from both hyperphagia and disrupted circadian rhythms during high-fat diet (HFD) feeding (Palmisano et al., 2017), future studies are warranted to determine both whether sensitivity to HFD requires DMH^LepR activity and, if so, whether these neurons lie downstream of circuits mediating sexually dimorphic responses to HFD.

To investigate both the extent to which diurnal metabolic disruption in DMH^LepR-silenced mice is secondary to the shift in daily patterns of food intake and whether DMH^LepR neurons are required to entrain feeding behavior, we implemented a time-restricted feeding (TRF) paradigm. To minimize baseline differences in satiety status that could influence TRF adaptation, control and DMH^LepR-silenced mice were fasted for 24 hr, from ZT14 (dark-cycle onset) until ZT14’ of the following day. We then restricted food availability to the dark-cycle, active period (ZT14–ZT24) in both groups. After a 5-day TRF acclimation period, both groups were subjected to four additional days of TRF during which measurements were made using indirect calorimetry (for a total of 9 days of TRF), followed by 3 days of ad lib feeding (Figure 4A).

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

Body weight, RER, EE, and LMA during TRF.

https://cdn.elifesciences.org/articles/63671/elife-63671-fig4-data1-v2.xlsx

Download elife-63671-fig4-data1-v2.xlsx

During TRF acclimation, body weight oscillated daily as expected in both groups, being higher after food was available during the dark cycle, and lower after light-cycle fasting. However, whereas control mice were able to maintain their weight during TRF by increasing dark-cycle food intake, DMH^LepR-silenced mice failed to compensate for the imposed light-cycle fast and consequently exhibited a small reduction in body weight (Figure 4—figure supplement 1B). Upon restoration of ad lib feeding, DMH^LepR-silenced mice exhibited rebound hyperphagia sufficient to recover lost weight (Figure 4B–D), but this hyperphagic response was limited to the light cycle, as DMH^LepR-silenced mice rapidly reverted to their mistimed feeding rhythms (Figure 4B–C). Based on these findings, we conclude that DMH^LepR neuron activity is required to entrain feeding behavior during dark-cycle TRF. In addition, the capacity to increase dark-cycle intake in response to weight loss appears to depend upon DMH^LepR neuron activity, since DMH^LepR-silenced mice exhibit rebound hyperphagia following TRF only during the light cycle, a time when normal mice consume little food. Although mechanisms underlying this adaptive response await further study, the capacity to increase intake when food is available for a restricted window each day requires the ability to anticipate when food will be available, in association with a variety of metabolic and neuroendocrine adaptations (Drazen et al., 2006). Future studies are warranted to evaluate the extent to which DMH^LepR neuronal activity is required for adaptation to an equivalent feeding window during the light cycle, and/or narrower periods of food availability.

Our work identifies a crucial physiological role for DMH^LepR neurons in diurnal patterning of feeding behavior, locomotion, and associated metabolic parameters with normal light–dark cycles, as well as the ability to adapt food intake to a restricted feeding paradigm. Given evidence from both humans and rodents that mistimed feeding can predispose to obesity and T2D (Challet, 2019; Huang et al., 2011), these findings have relevance to the pathogenesis of both disorders. An improved understanding of the neural circuits underlying endogenous rhythms of behavior, feeding, and metabolism may facilitate the development of new therapeutic and dietary strategies for the treatment of obesity and related metabolic disorders in humans.

All procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Animal Care Committee at the University of Washington. Following stereotaxic surgery, all studied animals were individually housed with ad lib access to standard chow diet (LabDiet 5053) in a temperature- and humidity-controlled facility with 14:10 light–dark cycles. Adult Lepr^IRES-Cre/+ (LepR-Cre) mice (Jackson Laboratory no. 008320) or Cre-littermate controls were used for all experiments, as described below.

In one cohort of male LepR-Cre (TeTx; n=7) and Cre-littermate (control; n=5) mice, longitudinal measurements of body weight, food intake, and body composition were assessed first (Figure 1B–H and Figure 1—figure supplement 1), followed by indirect calorimetry (data available upon request), and then fast–refeeding studies with leptin administration (Figure 2B–C). A second cohort of male LepR-Cre animals receiving TeTx (n=6) was used to validate the ability of leptin to induce pSTAT3 signaling in DMH^LepR neurons (Figure 2A). In a third cohort of male LepR-Cre mice (TeTx; n = 7) using GFP as a control (control; n=7), the effect of DMH^LepR silencing to induce hyperphagia and weight gain was confirmed (Figure 3—figure supplement 1, panels A and B) and the mice were subjected to indirect calorimetry 48 hr after surgery (Figure 3—figure supplement 1, panels C–H) and again after 4 weeks (Figure 3), which was followed by TRF studies (Figure 4 and Figure 4—figure supplement 1). Finally, a fourth cohort of female LepR-Cre (TeTx; n=8) and Cre-littermate (control; n=8) were used to generate the data shown in Figure 1—figure supplement 2, and Figure 3—figure supplement 2.

The viral vector AAV1-CBA-DIO-GFP:TeTx (TeTx) was generated and validated as described (Han et al., 2015; Chen et al., 2018) and generously provided by Dr. Richard Palmiter and Dr. Larry Zweifel (University of Washington, Seattle, WA). As a control in some studies (see ‘Order of Experiments’), the viral vector AAV5-hSyn-DIO-EGFP (AddGene, Watertown, MA; cat: 50457-AAV5) was used. For viral microinjection, animals were placed in a stereotaxic frame (Kopf 1900; Cartesian Research Inc, Tujunga, CA) under isoflurane anesthesia. The skull was exposed with a small incision, and two small holes were drilled for bilateral 200 nL injection volume of TeTx into the DMH of LepR-Cre or Cre-negative littermate mice based on coordinates from the Mouse Brain Atlas: anterior-posterior (AP) −1.6, dorsal-ventral (DV) −5.6 mm, and lateral 0.40 mm from bregma, identified as the approximate point at which coronal and saggital sutures intersect, and where the bregma–lambda distance approximates 4.21 mm, as previously described (Faber et al., 2020; Franklin and Paxinos, 2008). AAV) was delivered using a Hamilton syringe with a 33-gauge needle at a rate of 50 nL/min (Micro4 controller), followed by a 5 min wait at the injection site and a 1 min wait 0.05 mm dorsal to the injection site before needle withdrawal. Animals received a perioperative subcutaneous injection of buprenorphine hydrochloride (0.05 mg/kg; Reckitt Benckiser, Richmond, VA). Viral expression was verified post hoc in all animals, and any data from animals in which the virus expressed outside the targeted area were excluded from the analysis.

Measurements of body lean and fat mass were determined in live, conscious mice by use of quantitative magnetic resonance spectroscopy (QMR; EchoMRI 3-in-1; Echo MRI, Houston, TX) by the University of Washington Nutrition Obesity Research Center Energy Balance Core.

To validate the ability of leptin to elicit pSTAT3 signaling in DMH^LepR neurons, ad lib fed mice were injected intraperitoneally with leptin (5 mg/kg; Dr. Parlow; National Hormone Peptide Program) and perfused 90 min later, as described below.

To assess the ability of leptin to suppress the compensatory hyperphagia that normally follows a prolonged fast, mice were fasted for 24 hr from ZT2 to ZT2’. On the second day, leptin (3 mg/kg) or vehicle control (PBS, pH 7.9) was injected intraperitoneally in mice 15 min before preweighed food was placed back in the cage, and intake was monitored for the following 24 hr.

Prior to surgery, mice were acclimated to calorimetry cages for at least 7 days, during which time baseline pre-intervention calorimetry values were obtained for each subject. To account for potential confounding effects of stress resulting from rehousing into calorimetry during study, the first 12–16 hr of data (including the first dark cycle) were omitted from all analyses, except for the TRF experiment where continuous intake measurements were required. Energy expenditure measurements were obtained by a computer-controlled indirect calorimeter system (Promethion, Sable Systems, Las Vegas NV) with support from the Energy Balance Core of the NORC at the University of Washington, as previously described (Kaiyala et al., 2015). Oxygen consumption (VO₂) and carbon dioxide production (VCO₂) were sequentially measured for each mouse for 30 s at 5 min intervals, and food and water intakes were measured continuously while mice were housed in a temperature- and humidity-controlled cabinet (Caron Products and Services, Marietta, OH). Ambulatory activity was determined simultaneously and beam breaks in the y-axis were scored as an activity count, and a tally was recorded every 3 min. Data acquisition and instrument control were coordinated by MetaScreen v2.3.15.11, and raw data were processed and binned into 2-hr increments using ExpeData v1.9.14 (Sable Systems, Las Vegas, NV) using an analysis script documenting all aspects of data transformation.

To eliminate the initial effects of varying fed status of animals, 1 day before TRF, animals were placed into clean bedding and fasted for 24 hr from ZT14 (on day 1) to ZT14’ (on day 0) before TRF began. Food was removed each morning at the start of the light cycle (ZT0) and the bedding inspected to ensure no residual food debris remained accessible. Food was returned at the start of the dark cycle (ZT14) each day. Body weight was also measured at both ZT0 and ZT14 daily. Animals were maintained on TRF for a total of 5 days in home cages, then subjected to indirect calorimetry for four additional days of TRF before returning to ad lib feeding for the remaining 3 days of study (Figure 4A).

For brain immunohistochemical analyses, animals were terminally anesthetized with ketamine:xylazine and transcardially perfused with PBS followed by 4% paraformaldehyde in 0.1 mol/L PBS. Brains were removed and postfixed overnight, then transferred into 30% sucrose overnight or until brains sunk in solution. Brains were subsequently sectioned on a freezing-stage microtome (Leica) to obtain 30 μm coronal sections in four series. A single series of sections per animal was used in histological studies, and the remainder stored in −20°C in cryoprotectant. Brain sections were washed in PBS with Tween-20, pH 7.4 (PBST) overnight at 4°C. Sections were then washed at room temperature in PBST (3 x 8 min), followed by a blocking buffer (5% normal donkey serum [NDS], 1% bovine serum albumin [BSA] in PBST with azide) for 60 min with rocking. Sections were then incubated overnight at 4°C in blocking buffer containing primary antiserum (goat anti-GFP, Fitzgerald, 1:1000; rabbit anti-pSTAT3, Sigma-Aldrich, St Louis, MO, 1:1000). Next, sections were washed (3 x 8 min) in PBST before incubating in secondary donkey anti-goat IgG Alexa 488 (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:1000 in blocking buffer. Sections were washed (3 x 8 min) in PBST before incubating with DAPI for 8 min, followed by a final wash (3 x 10 min) in PBS. Sections were mounted to slides and imaged using a Leica SP8X confocal.

Tail blood for plasma hormonal measurement was collected at indicated times. Blood was collected via EDTA-coated capillary tubes and centrifuged at 4°C (7000 rpm, 4 min), and plasma was subsequently removed and stored at −80°C for subsequent assay. Plasma leptin (Crystal Chem, Elk Grove Village, IL; #90030) and plasma insulin (Crystal Chem; #90080) were determined by ELISA.

Prior to analysis, all data were tested for normality by Shapiro–Wilk normality test. All results are presented as means ± SEM. P values for unpaired comparisons were calculated by two-tailed Student’s t test. Time course comparisons between groups were analyzed using a two-way repeated-measures ANOVA with main effects of treatment (control vs. TeTx) and time. All post hoc comparisons were determined using Sidak’s correction for multiple comparisons. All statistical tests indicated were performed using Prism (version 7.4; GraphPad, San Diego, CA) software.

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

We thank R Palmiter, C Campos, L Zweifel, and M Baird for producing the TeTx virus, JT Nelson for assistance with metabolic experiments, and V Damian for maintaining the mouse colony. We also thank R Palmiter for editing the manuscript. We are grateful to N Peters at the University of Washington Keck Imaging Center for technical assistance and the National Institutes of Health (S10-OD-016240) for support to the W.M. Keck Foundation Center for Advanced Studies in Neural Signaling. This work was supported by NIH grants F31-DK113673 (CLF), T32-GM095421 (CLF), DK128802 (ZM), DK089056 and DK124238 (GJM), DK083042 and DK101997 (MWS); the NIDDK-funded Nutrition Obesity Research Center (DK035816) and Diabetes Research Center (DK017047) and the Diabetes, Obesity, and Metabolism (T32 DK007247; CLF) and Nutrition, Obesity, and Atherosclerosis (T32 HL007028; JDD) training grants at the University of Washington; a Department of Defense CDMRP/PRMRP grant W81XWH2010250 (ZM); a Dick and Julia McAbee Endowed Fellowship (JDD); an American Diabetes Association Innovative Basic Science Award (ADA 1–19-IBS-192; GJM); and an American Diabetes Association fellowship grant (ADA 1–19-PDF-103; JDD).

Animal experimentation: All procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Animal Care Committee at the University of Washington. (Jackson Laboratory no. 008320).

© 2021, Faber et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.