Perifornical UCN3 Neurons Regulate Overeating-Induced Weight Gain

Dear Editor,

Obesity is an increasingly prevalent global metabolic disorder, yet effective long-term interventions remain elusive. Recent studies have highlighted a novel peptide, Urocortin 3 (UCN3), which shows potential for obesity intervention [1]. UCN3 is a member of the corticotropin-releasing factor (CRF) family, which includes CRF, UCN1, and UCN2. Unlike CRF, UCN3 has a very low affinity for CRF receptor 1 (CRFR1) and a high affinity for CRF receptor 2 (CRFR2) [2, 3]. UCN3 and its receptors are abundantly expressed in the brain, pancreas, and gastrointestinal tract, playing important roles in insulin secretion, glucose metabolism, and gastric emptying [4, 5]. Clinical studies have further shown that UCN3 expression is dysregulated in conditions such as obesity and type 2 diabetes [6]. Overexpression of UCN3 in mice enhances carbohydrate metabolism and protects against hyperglycemia and adiposity induced by a high-fat diet [7]. These findings suggest that UCN3 may serve as a crucial mediator in the pathogenesis of obesity.

Overeating is a primary contributor to obesity, and evidence from multiple species has shown that both peripheral and central injections of UCN3 can effectively reduce food intake [8, 9]. Given that UCN3 may not cross the blood-brain barrier [10], its appetite-suppressing effects in the central and peripheral nervous systems may function independently. Within the central nervous system, UCN3 is primarily expressed in specific regions, including the perifornical area (PeFA), median preoptic nucleus, medial amygdala, bed nucleus of the stria terminals, and the superior paraolivary nucleus [2]. Among these areas, the PeFA contains neurons expressing UCN3 (PeFA^UCN3 neurons) that project to the lateral septum (LS), the ventromedial hypothalamic nucleus, and the periaqueductal gray, all of which are involved in regulating feeding behavior [11, 12]. However, to date, no studies have directly tested the role of PeFA^UCN3 neurons in feeding and obesity. Furthermore, the real-time activity of these neurons during feeding behavior remains unknown.

To address this gap, we first selectively silenced these neurons in the PeFA of UCN3-Cre mice by using Cre-dependent AAV vectors to express tetanus neurotoxin (TeNT), a molecular tool for blocking synaptic transmission (Fig. 1A). To confirm the efficacy of this synaptic inactivation, we co-injected AAV-hSyn-DIO-ChR2 and AAV-CAG-DIO-TeNT into the PeFA of UCN3-Cre mice and recorded light-evoked excitatory postsynaptic currents (EPSCs) in the LS, following methods reported in our previous work [12]. A brief light stimulus-evoked robust EPSCs in LS neurons, and this was blocked by Cyanquixaline (CNQX), demonstrating fast glutamatergic transmission. TeNT expression completely blocked the EPSCs induced by optogenetic activation of PeFA^UCN3 terminals, verifying the successful inhibition of neurotransmission from PeFA^UCN3 neurons (Fig. S1).

Fig. 1.

Inactivation of PeFA^UCN3 neurons leads to weight gain, increased feeding behavior, and disrupted circadian rhythms in both feeding and energy metabolism. A Left: schematic of the viral strategy to silence the PeFA^UCN3 neurons with tetanus toxin (TeNT). Right: representative image of TeNT-EGFP expression. Scale bar, 200 μm. B Representative body images of EGFP- and TeNT-expressing mice after 45 days on a standard chow diet. C Left: percentage difference in body weight gain in EGFP (n = 6) and TeNT (n = 7) mice after 45 days on a standard chow diet. Two-way ANOVA: F (1, 22) = 4.631, *P < 0.05, followed by Sidak’s post-hoc test: *P <0.05. Right: differences in weight gain between EGFP (n = 6) and TeNT (n = 7) mice after 45 days on a standard chow diet. Unpaired t test, *P < 0.05. D Quantification of daily standard chow intake in EGFP (n = 5) and TeNT mice (n = 6). Unpaired t test, *P <0.05. E Schematic showing the formula milk self-administration paradigm (“lick to pump”). F Lick raster patterns of individual mice. Gray: EGFP mice, n = 8; green: TeNT mice, n = 8. G Cumulative curves for the number of licks in EGFP and TeNT- mice. Gray: EGFP mice, n = 8; green: TeNT mice, n = 8. H Formula milk consumption in EGFP (n = 8) and TeNT (n = 8) mice. Unpaired t test, **P <0.01. I Dynamics of the respiratory exchange ratio (RER) of EGFP (n = 6) and TeNT (n = 5) mice across 48 h. Each row in the heatmap represents the RER from individual mice. J Average RER in EGFP and TeNT mice during daytime and nighttime. Two-way ANOVA: F (1, 9)_group = 1.780, P >0.05; F (1, 9)_time = 50.80, ****P <0.0001, followed by Sidak’s post-hoc test: *P <0.05, ****P <0.0001. K Dynamics of energy expenditure (EE) of EGFP (n = 6) and TeNT (n = 5) mice across 48 h. Each row in the heatmap represents the EE from an individual mouse. L Average EE in EGFP and TeNT mice during daytime and nighttime. Two-way ANOVA: F (1, 9)_group = 0.059, P >0.05; F (1, 9)_time = 29.33, ***P <0.001, followed by Sidak’s post-hoc test: ***P <0.001. M Food intake per hour of EGFP (n = 6) and TeNT (n = 5) mice across 48 h. N Average food consumption in EGFP and TeNT mice during daytime and nighttime. Two-way ANOVA: F (1, 9)_group = 1.850, P >0.05; F (1, 9)_time = 25.28, ***P <0.001, followed by Sidak’s post-hoc test: ***P <0.001. O Water intake per hour of EGFP (n = 6) and TeNT (n = 5) mice across 48 h. P Average water consumption in EGFP and TeNT mice during daytime and nighttime. Two-way ANOVA: F (1, 9)_group = 0.617, P >0.05; F (1, 9)_time = 19.51, **P <0.01, followed by Sidak’s post-hoc test: **P <0.01. All data are represented as the mean ± SEM. ns., no significant difference.

Body weight was measured on the day of virus injection (day 0) and 45 days post-injection (day 45), during which time animals had ad libitum access to standard chow and water. By day 45, the TeNT group exhibited significantly greater weight gain than the controls (Fig. 1C). Despite this weight gain, silencing PeFA^UCN3 neurons had no effect on glucose tolerance (Fig. S2). In addition, TeNT-expressing mice showed significantly increased food intake over 24-h periods in their home cages (Fig. 1D).

To further investigate this phenotype, we assessed liquid food intake using an operant self-administration task. In this paradigm, each lick at the spout activated a sensor that triggered a pump to deliver a formula milk drop. Compared to the Enhanced Green Fluorescent Protein (EGFP) control group, TeNT-expressing mice exhibited significantly more licks and higher total milk consumption (Fig. 1E–H). These findings suggest that silencing PeFA^UCN3 neurons promotes feeding behavior, likely contributing to the increase in body weight.

Obesity is associated not only with energy intake but may also involve alterations in energy metabolism. To explore this, we conducted 48-h continuous monitoring in metabolic cages for individual mice. Overall, there were no differences between groups in mean oxygen uptake (VO₂), carbon dioxide production (VCO₂), respiratory exchange ratio (RER), or average energy expenditure (EE) (Fig. S3E-H). However, while the control group maintained a clear day-night rhythm—with all indicators (VO₂, VCO₂, RER, and EE) significantly higher at night than during the day—the TeNT group showed no significant differences in these indicators between day and night (Figs. 1J, L and S3B, D), indicating a loss of the circadian rhythm.

This disruption of the circadian rhythm was also reflected in the patterns of food and water intake, with the TeNT group showing similar rhythmic alterations (Fig. 1M, O). In addition, during the daytime, the TeNT group exhibited a significantly higher RER and food intake than the control group, whereas, at night, there were no differences in any metrics between groups (Fig. 1J, N). These findings suggest that silencing PeFA^UCN3 neurons disrupts the circadian rhythm of energy metabolism and feeding.

To investigate the activity changes of PeFA^UCN3 neurons during feeding, we injected AAV-Ef1α-DIO-GCaMP6 into the PeFA region of UCN3-cre mice and implanted an optical fiber. Three weeks post-injection, we recorded the Ca²⁺ signal dynamics of UCN3 neurons during feeding behavior using fiber photometry (Fig. 2A). We found that the onset of feeding, whether with standard chow or palatable high-fat food, led to a rapid decrease in PeFA^UCN3 neuronal activity (Fig. 2B–E). This result suggests that PeFA^UCN3 neurons play an inhibitory role in feeding behavior, becoming less active to potentially facilitate or promote food intake, which aligns with our findings from TeNT treatment.

Fig. 2.

The activity of PeFA^UCN3 neurons is suppressed during food consumption. A Upper: Schematic of the viral strategy of GCaMP6 expression in PeFA^UCN3 neurons. Lower: Example coronal section showing GCaMP6 expression and the location of an optical fiber tip among PeFA^UCN3 neurons. Scale bar. 200 μm. B Average Ca²⁺ response of PeFA^UCN3 neurons elicited when mice first bit the standard chow (n = 6 mice). Vertical line, the moment of the first bite. C The area under the curve (AUC) for GCaMP6 signals in individual mice was assessed within a 40-s window immediately after the first bite. **P <0.01, paired t test. D Average Ca²⁺ response of PeFA^UCN3 neurons elicited when the mice first bit the high-fat food (n = 6 mice). Vertical line, the moment of the first bite. E The AUC for GCaMP6 signals in individual mice was assessed in a 40-s window immediately after the first bite. ***P <0.001, paired t test; the mean ± SEM. Each row in the heatmap represents averaged Ca²⁺ dynamics over bouts from individual mice (ranked by AUC). F Upper: Schematic of the viral strategy of hM3D expression in PeFA^UCN3 neurons. Bottom: Example coronal section showing hM3D-mCherry expression in PeFA^UCN3 neurons. Scale bar, 200 μm. G Representative images of c-Fos expression in mCherry-labeled PeFA^UCN3 neurons in mCherry and hM3D groups 2 h after CNO injection (2 mg/kg). H The proportion of co-labeled cells in mCherry-expressing neurons (n = 8 slices from 2 mice) and hM3D-expressing neurons (n = 8 slices from 2 mice). ****P <0.0001, unpaired t test. I Two-hour standard chow intake in mCherry (n = 9) and hM3D (n = 9) mice after a CNO injection (5 mg/kg). Two-way ANOVA: F(1, 16)_group = 0.2218, P >0.05; F(1, 16)_treatment = 5.233, *P <0.05, followed by Sidak’s post-hoc test: *P <0.05. J Formula milk consumption in mCherry (n = 8) and hM3D (n = 7) mice after CNO or saline injections. Two-way ANOVA: F(1, 13)_group = 0.004, P >0.05; F(1, 13)_treatment = 1.628, P >0.05. All data are represented as the mean ± SEM. ns., no significant difference.

Given that the inhibition of PeFA^UCN3 neurons leads to increased feeding, we hypothesized that activating these neurons might suppress food intake. To test this hypothesis, we specifically expressed the modified human M3 muscarinic receptor hM3D in the UCN3 neurons of the PeFA region (Fig. 2F). The effectiveness of this approach was validated by the significant activation of these neurons upon Clozapine-N-oxide (CNO) injection, as indicated by a marked increase in the proportion of c-Fos-positive UCN3 neurons (Fig. 2G, H). During the peak feeding period from 20:00 to 22:00, CNO injection at 5 mg/kg significantly suppressed feeding behavior in the hM3D group compared to saline injection, while no effect was found in the control group (Fig. 2I). However, 24-h chow intake remained unaffected (Fig. S4). Furthermore, in the operant liquid-food self-administration task, chemogenetic activation did not reduce the total consumption of formula milk (Fig. 2J), indicating a differential effect between palatable liquid food and standard solid chow.

Long-term intervention for obesity remains a global challenge. Previous studies have identified UCN3 as an important peptide involved in appetite regulation and weight gain, offering a promising avenue for obesity intervention [1]. However, how UCN3-expressing neurons in the brain contribute to feeding behavior and weight regulation remains unclear. In this study, we selectively inactivated UCN3 neurons in the PeFA region, which resulted in increased food intake and weight gain, suggesting that these neurons play an inhibitory role in feeding behavior. This finding was further supported by our fiber photometry results, which demonstrated a rapid decrease in neuronal activity upon the initiation of feeding. Building on prior research showing that both peripheral and central administration of UCN3 can reduce food intake [8, 9], our findings suggest that UCN3 may act as a 'brake' on feeding behavior, permitting increased consumption when neuronal activity is suppressed.

Obesity is linked not only to excessive food intake but also to metabolic changes. While evidence connects peripheral UCN3 to metabolic regulation [5], the role of central UCN3 remains relatively unexplored. Our study found no significant difference in total energy expenditure between groups, suggesting that the weight gain may primarily stem from excess energy intake. Notably, silencing PeFA^UCN3 neurons disrupted the circadian rhythms of both feeding and energy metabolism, particularly reflected in increased food intake and respiratory exchange ratio during the biological rest period (daytime for mice). Although the role of PeFA in biological rhythms has not been previously reported, the adjacent paraventricular hypothalamic nucleus (PVN) has been shown to receive projections from the suprachiasmatic nucleus, a key regulator of circadian rhythms, thus influencing feeding behavior and metabolic processes [13]. Injecting UCN3 into the PVN has been demonstrated to reduce the regularity of feeding behavior [14]. Given that alterations in feeding rhythm significantly affect metabolic rhythms [15], the disruptions in metabolic rhythms found in the TeNT group may be a consequence of changed feeding patterns.

Pharmacological studies have shown that central administration of UCN3 or CRFR2 agonists inhibit feeding [9, 14]. Consistent with this, our findings showed that the chemogenetic activation of PeFA^UCN3 neurons reduces food intake, indicating that these neurons suppress feeding through the release of UCN3 peptides. Notably, this inhibitory effect was only apparent during the 2-h nighttime feeding peak and not over 24 h. This could be due to the transient nature of chemogenetic activation or its inherently acute effects, which the organism can naturally compensate for or balance. In addition, CNO injection did not alter formula milk consumption. Given the liquid form and palatability of formula milk, which triggers weaker satiety but greater hedonic satisfaction than solid chow, these results suggest that PeFA^UCN3 neurons might be more sensitive to solid foods. Another potential factor is the timing of the experiments. Our synaptic inactivation results highlighted these neurons' role in regulating feeding rhythms (Fig. 1J, N). Therefore, conducting chemogenetic experiments during the daytime—outside the animals' peak feeding period—may limit its potential inhibitory effects.

In conclusion, our results reveal the important role of PeFA^UCN3 neurons in regulating body weight, feeding behavior, and metabolic rhythms, and identify the suppressed activity changes of these neurons during feeding. These findings support the hypothesis that UCN3 is a potential therapeutic target for obesity and eating disorders, offering valuable insights for future clinical interventions in these conditions.

Supplementary Information

Below is the link to the electronic supplementary material.

Conflict of interests

The authors declare no competing interests.