Novelty exploration-activated ensemble in the lateral hypothalamus confers analgesic and anxiolytic effects

Abstract

While attention distraction alleviates pain and negative affect, the underlying neural circuits remain unclear. Here, we show that novelty exposure, a means to attract attention, significantly alleviates acute and chronic pain in mice. Using a Fos-driven viral strategy, we identified a lateral hypothalamus (LH) neuronal ensemble activated during novelty exploration. This novelty ensemble also responds to pain- and anxiety-like stimuli. Activation of this ensemble produces analgesic and anxiolytic effects, whereas its inhibition exacerbates pain and anxiety-like behavior in mouse pain models. The LH ensemble comprises both GABAergic and glutamatergic subpopulations, both contributing to pain and anxiety modulation. However, activating their specific projections to the lateral preoptic area, lateral habenula, ventral tegmental area, and lateral periaqueductal gray regulates pain and anxiety in distinct patterns. Together, we define an LH novelty-activated neuronal subpopulation mediating the analgesic and anxiolytic benefits of novelty exposure, revealing circuit-specific targets for relieving pain and anxiety.

Introduction

Non-pharmacological treatments, such as attention distraction, psychological counseling, and regular exercise, have been demonstrated to alleviate pain and emotional distress, including anxiety and depression, and reduce the need for analgesic medications^1,2,3,4. Consequently, multimodal pain therapy, which incorporates these supportive treatments, is increasingly favored for effective management of pain and related affective disorders^2,4.

Attention distraction-induced analgesia and anxiolysis occur when cognitive tasks compete with painful or aversive stimuli for attention, resulting in suppressed pain and diminished negative emotional processing^5,6. Strategies such as watching movies, listening to music, engaging in cognitive or virtual reality tasks, and participating in non-medical conversations have effectively diverted attention from pain and anxiety, raising the threshold for both pain sensation and negative affect in human subjects^{4,7,8,9,10,11}. Similarly, low noise levels or environmental enrichment as well as novelty exploration in animals can induce analgesic and anxiolytic effects^{12,13,14,15,16}. Neuroimaging studies further demonstrate that attentional distraction using various cognitive distraction paradigms alters the neural responses to pain and emotional stimuli across multiple brain regions involved in the ascending and descending pain pathways and emotional processing, including the somatosensory cortex, mid-cingulate cortex, insula, prefrontal cortex, amygdala, and periaqueductal gray (PAG)^8,17,18,19. These findings suggest that attentional networks regulate pain and emotion by orchestrating multiple neural circuits, resulting in both analgesic and anxiolytic effects. Nonetheless, the precise neural mechanisms underlying this coordinated modulation remain incompletely understood. Although the endogenous opioid system has been implicated in reward processing²⁰ and certain forms of stress-induced analgesia²¹, its specific contribution to novelty exploration-induced analgesia remains unclear.

The lateral hypothalamus (LH) is increasingly recognized as an important hub integrating cognitive, emotional and sensory information. Reflecting this role, the LH contains heterogeneous neuronal populations – most notably orexin/hypocretin- and melanin-concentrating hormone (MCH)-producing neurons, as well as GABAergic and glutamatergic neurons - that govern motivated behaviors including novel object exploration, predatory attack, and reward-seeking^22,23,24,25. Furthermore, these specific LH populations also influence pain and affect^26,27,28,29. Notably, classical manipulations (electrical/chemical stimulation or local morphine) regulate nociception and anxiety in experimental animals^30,31, and cell type specific interventions show that LH glutamatergic neurons shape the sensory and affective aspects of neuropathic pain, likely via projections to the lateral habenula and PAG^26,32.

Consistent with its integrative function, the LH receives nociceptive and affective inputs from nuclei across the forebrain, midbrain, and medulla^26,27. However, whether LH neurons specifically mediate attentional distraction-induced analgesia and anxiolysis remains unclear. We therefore hypothesize that a subset of LH neurons is recruited by attentional distraction to relieve pain and negative emotion in chronic pain. To test this, we utilized a novelty exploration paradigm. Although novel environments can occasionally trigger neophobia, rodents exhibit a strong preference for voluntarily exploring novel stimuli when potential threats are minimized³³, making this behavior conceptually aligned with the visual distraction paradigms used in clinical research.

Here, we show that exploring a novel object or environment alleviates pain perception and anxiety-like behaviors in mice with chronic neuropathic pain. Using Fos-based viral vectors to selectively label and manipulate LH neurons activated during this process (the novelty ensemble), we demonstrate the recruitment of specific glutamatergic and GABAergic subpopulations. These distinct neuronal groups produce significant analgesic and anxiolytic effects via their targeted projections to downstream nuclei. Ultimately, these parallel pathways mediate the relief of chronic pain and associated anxiety through attentional distraction.

Results

Novel object and environment exploration alleviates pain

Distraction is a well-known modulator of pain perception, typically reducing subjective pain in humans⁷. Building on this, we first investigated whether exposure to either a novel object or novel environment mitigates nocifensive behaviors in acute inflammatory pain induced by 1% formalin injection into the hind paw (Fig. 1a). As expected, formalin elicited the classical biphasic nocifensive responses, reported in a previous study³⁴: a brief acute phase lasting for 10–15 min, followed by a prolonged inflammatory phase starting around 25 min after injection (Fig. 1b, c). Because the secondary phase nocifensive behavior lasted longer (20–35 min) with a stable plateau (Fig. 1b, c) and was reported to be controlled by supraspinal mechanisms^35,36, we proposed that it may be regulated by higher brain function, such as attention distraction by novelty exploration. Therefore, we measured secondary phase nocifensive behavior to assess the effect of novelty exploration on pain. To minimize stress, mice were habituated to a cylindric arena for 10 min daily over 7 consecutive days (Fig. 1d). On the testing day, all mice received a 20 μl formalin injection into the right hind paw and were returned to their home cages for 20 min. Mice were then divided into 3 groups. One group of mice were placed in the testing arena and their nocifensive behaviors (paw licking and flinching) were recorded between 20 and 40 min after injection, corresponding to the secondary inflammatory phase (Fig. 1d). The other two groups of mice were exposed to either a novel object at the center of the familiar cylindric arena (30 cm in diameter, 40 cm in height) or were placed in a novel cubic environment (30 cm in side length) (Fig. 1d). In both male and female mice, exposure to either novel object or novel environment significantly decreased total licking time and the number of licking bouts in comparison with those exposed to the familiar arena without novel object (Fig. 1e-j). The difference in the effects of novelty exploration on nocifensive behavior between male and female mice was not significant (Fig. 1g, j).

Fig. 1: Novelty exploration alleviates pain-like behaviors and activates LH neurons.

a–c Nocifensive behavior post-formalin (n = 12) or saline (n = 5 mice) injection. Quantified by paw licking time (b, F_(1,15) = 32.42, P < 0.0001) and bouts (c, F_(1,15) = 49.25, P < 0.0001). Pink shading: acute/inflammatory phases. d Design for e-l. e–j Nocifensive behavior 20–40 min post-formalin in familiar/novel environments or with novel object (n = 8 mice/group). Males: licking time (e, F_(6,63) = 3.74, P = 0.003), bouts (f, F_(6,63) = 3.36, P = 0.006), total time (g, F_(2,21) = 10.22, P = 0.0008). Females: licking time (h, F_(2,21) = 11.77, P = 0.0004), bouts (i, F_(6,63) = 2.741, P = 0.02), total time (j, F_(2,21) = 11.60, P = 0.0004). Naltrexone (5 mg/kg, i.p., 30 min pre-formalin) effects on licking bouts (k) and time (l) at 20–40 min during familiar or novel object exposure (saline n = 11, naltrexone n = 9 mice). Object effect: F_(1,18) = 107.6, P < 0.0001 (k) and F_(1,18) = 102.2, P < 0.0001 (l). Drug effect: F_(1,18) = 0.15, P = 0.70 (k) and F_(1,18) = 6.57, P = 0.02 (l). m Design for (n-q). Hot plate PWL post-saline (n = 9 mice/group) or CFA (n = 12 mice/group) with/without novel object. Saline: males (n, t = 5.11, P = 0.0009), females (o, t = 5.13, P = 0.0009). CFA: males (p, t = 4.46, P = 0.001), females (q, t = 4.95, P = 0.0004). r, s c-Fos expression comparing novel object (r) or novel environment (s) vs familiar environment (n = 8 mice/group). Color denotes significance. (t, u) Images and c-Fos summary for familiar environment (n = 8), novel object (n = 9), and novel environment (n = 7 mice) (F_(2,21) = 34.33, P < 0.0001). *P < 0.05, **P < 0.01, n.s, not significant. Two-sided paired (n-q) or unpaired t-test (r, s); one-way (g, j, u), two-way (k, l), or two-way RM (b, c, e, f, h, i) ANOVA with Tukey’s post hoc test. Data are mean ± SEM. Scale bars, 100 μm. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file. ACC Anterior Cingulate Cortex; BLA Basolateral Amygdala; BST Bed Nucleus of the Stria Terminalis; GPe External Globus Pallidus, LH Lateral Hypothalamus, LHb Lateral Habenula, LPB Lateral Parabrachial Nucleus, LSV Lateral Septum Ventral, M1 Primary Motor Cortex, mPFC Medial Prefrontal Cortex, MS Medial Septum, PAG Periaqueductal Gray, PH Posterior Hypothalamic Nucleus, Pir Piriform Cortex, Pn Pontine Nuclei, PVT Paraventricular Thalamic Nucleus, S1 Primary Somatosensory Cortex, V1 Primary Visual Cortex, VLGMC Ventral Lateral Geniculate Nucleus Magnocellular part.

Given that attention-induced analgesia in human is partly mediated by endogenous opioids³⁷, we intraperitoneally administered naltrexone (5 mg/kg), an opioid receptor antagonist, to formalin-injected mice exposed to novel objects during phase 2. Naltrexone did not reverse the reduction in total licking time or the number of licking bouts observed in the novelty exploration group (Fig. 1k, l). These results suggest that the analgesic effect of novelty exploration in the formalin test is resistant to the administered dose of naltrexone.

To further assess the impact of novelty exploration on inflammatory pain, we injected either saline or complete Freund’s adjuvant (CFA) into the hind paw of male and female mice. For seven consecutive days prior to testing, mice were habituated to the hotplate test apparatus for 10 min daily. On the test day, both saline- and CFA-treated mice were placed on the 52 °C hotplate and paw withdrawal latency (PWL) was measured (Fig. 1m). We chose this assay over mechanical testing (e.g., von Frey) because it allows for the simultaneous assessment of voluntary object exploration and nociceptive responses within a consistent, standardized environment, minimizing investigator interference. In both male and female mice, novel object exposure significantly increased PWL in saline and CFA group (Fig. 1n-q), indicating that novel exploration attenuates thermal sensitivity under basal and inflammatory states. Importantly, we calculated the percent change in PWL induced by the novel object relative to the no-object condition and found no significant difference in the magnitude of analgesia between the Saline and CFA groups (Male: Saline 58.81 ± 13.51% vs. CFA 38.94 ± 8.06%, p = 0.22; Female: Saline 39.34 ± 8.77% vs. CFA 35.68 ± 7.92%, p = 0.76; unpaired t-test). This demonstrates that the analgesic efficacy of novelty remains similar under physiological and inflammatory conditions.

Finally, we examined whether the presence of a novel, age- and sex- matched conspecific mouse attenuates pain sensitivity in another mouse in physiological condition. In this set of experiments, we employed a conditioned place aversion (CPA) paradigm in both male and female mice (Supplementary Fig. 1a). Mice that received repetitive suprathreshold von Frey filament (2 g) stimuli in the conditioned chamber for 3 consecutive days (30 min daily) developed CPA (Supplementary Fig. 1a-c). By contrast, when the same stimulation was paired with the presence of a novel conspecific mouse (placed in a cylinder-shaped metal wire cage in the conditioning chamber) during each conditioning session, mice did not develop CPA (Supplementary Fig. 1b, d). These results suggest that exposure to social novelty can alleviate pain and associated aversion induced by suprathreshold mechanical stimulation.

These data demonstrate that novelty exploration acts as a reproducible, opioid-independent suppressor of inflammatory pain.

The LH is activated by novelty exploration and neuropathic pain

To map novelty-activated neurons in brain regions involved in attention and pain processing, we performed c-Fos-immunostaining experiments. Specifically, we compared the numbers of c-Fos-positive neurons across brain regions under two paradigms: (1) mice exploring a familiar arena with versus without a novel object, and (2) mice exploring a novel arena versus a familiar arena without exposure to a novel object (Fig. 1r, s). We then color-coded brain regions according to the statistical significance of these differences. Both novelty paradigms elicited robust increases in c-Fos-expression across several cortical and subcortical regions relative to controls (Fig. 1r, s; Supplementary Fig. 2; Supplementary Table 1, 2). Notably, the lateral hypothalamus (LH), primary motor cortex, anterior cingulate cortex, and primary sensory cortex consistently showed the greatest activation in both comparisons (Fig. 1r, s; Supplementary Fig. 2; Supplementary Table 1, 2). While acknowledging cortical involvement, we prioritized a subcortical hub, the LH, as a target region for causal relationship tests to minimize engagement of a broad, multifunctional cortical network. This choice aligns with previous reports implicating the LH in multiple facets of novelty exploration^38,39. Further analysis revealed that LH neurons activated following either novel object or novel environment exposure were located in the LH at coronal planes ranging from −0.6 to −2.2 mm relative to the bregma, indicating a broad anterior to posterior coverage (Fig. 1t, u; Supplementary Fig. 1e-h). Importantly, we replicated this robust c-Fos induction in an independent cohort of mice (Supplementary Fig. 1o–p), confirming the reproducibility of the LH activation during the transition from a quiet, habituated state to an active, novelty-seeking state.

Moreover, we found that c-Fos expression was similarly upregulated in LH neurons following various pain modalities, including suprathreshold von Frey (2 g) stimulation, formalin injection, and spared nerve injury (SNI)-induced neuropathic pain, compared to their corresponding controls (Supplementary Fig. 1i-n). These data support that the LH is implicated in pain processing, consistent with previous studies^27,40.

LH novelty ensemble is activated by painful stimuli

The above evidence suggests that the LH contains neurons implicated in either novelty exploration, pain sensation, or both. To investigate whether these conditions activate overlapping populations, we employed Fos-based viral TRAP (targeted recombination in active populations) strategy to label neurons activated by one stimulus and examined their responsiveness to the other. We co-injected AAV-cfos-Cre-ERT2 and AAV-EF1α-DIO-eYFP into the LH (Fig. 2a, b) and 3 weeks later, induced labeling. Mice were habituated and then randomly assigned to three groups (Fig. 2b): (1) Novelty + TAM: tamoxifen (15 mg/kg, i.p.) administered 5 h before novel object exposure (10 min/object; 60 min total); (2) Familiar+TAM: tamoxifen administration followed by familiar object exposure for 1 h; and (3) Novelty+Saline: saline followed by novel object exposure (10 min/object; 60 min in total). We observed a robust increase in eYFP-labeled LH neurons exclusively in the Novelty + TAM group (Fig. 2c, d), confirming the specificity and efficiency of the Fos-TRAP approach.

Fig. 2: Fos-TRAP-labeled LH novelty ensemble is activated by pain stimulation.

a Diagram of activity-driven, tamoxifen-enabled Fos-TRAP labeling. b Experimental protocol examining whether Fos-TRAP-eYFP-labeled novelty-activated LH neurons are reactivated by novel object exploration, hind paw mechanical stimulation (von Frey), and formalin injection. c, d Example images and summary showing Fos-TRAP-eYFP selectively labeled LH neurons following tamoxifen administration and novel object exposure (F_(2,9) = 22.40, P = 0.0003; n = 4 mice). Representative images and summary showing the overlap between Fos-TRAP-eYFP-labeled LH neurons (green, 1st session of novel object exploration for 1 h) and c-Fos-immunostained neurons (magenta) induced by: e, f a 2nd session of novel object exploration 1-2 weeks later (n = 7 mice); g, h repetitive hind paw stimulation with 2 g von Frey filament for 1 h (n = 7 mice); i, j hind paw formalin injection (n = 7 mice). White arrows indicate the overlap of eYFP- and c-Fos-neurons. Scale bar, 100μm. **P < 0.01, n.s, not significant. One-way ANOVA with Tukey’s post hoc test (c, d). Data are mean ± SEM. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

We first assessed the reactivation of this ensemble by re-exposing the novelty-tagged mice to novel objects (60 min total) one week later. Approximately 60.4% of eYFP+ LH neurons were co-stained with c-Fos-antibody (Fig. 2e, f), indicating that eYFP+ LH neurons were robustly re-recruited by the second session of novel object exploration. Conversely, ~32.6% of the total c-Fos(+) population expressed eYFP. While the difference in these proportions likely reflects unequal pool sizes and viral labeling efficiency, the data confirm that a substantial subset of LH neurons is consistently recruited by repeated novelty exploration.

To determine whether these novelty-activated neurons also process pain, we subjected eYFP-labeled mice to either suprathreshold von Frey filament (2 g) stimulation on the hind paw or formalin injection into the hind paw (Fig. 2b). Both stimuli robustly induced c-Fos expression in the LH. Notably, we found significant overlap: approximately half of the eYFP+ novelty neurons were activated by pain (45.8% for von Frey, 45.7% for formalin) (Fig. 2g-j). In the reverse direction, ~31.5% and 31.4% of total c-Fos(+) neurons were eYFP-labeled in mice that received von Frey filament stimulation and formalin injection, respectively. These findings indicate a substantial convergence of novelty and acute pain signals onto the same LH population.

We labeled LH neurons activated by formalin injection with the Fos-TRAP technique and assessed their reactivation by novel object exposure one week later (Supplementary Fig. 3a). Consistent with our initial findings, approximately 36% of the pain-tagged (eYFP+) neurons were c-Fos(+) following novel object exploration, and 40.6% of novel object exploration-induced c-Fos(+) neurons were eYFP-labeled (Supplementary Fig. 3b-d).

Collectively, our findings demonstrate that the LH contains a considerable population of neurons capable of integrating both novelty and pain signals.

LH novelty ensemble preferentially encodes nociception

To determine if salient and nociceptive stimuli are specifically encoded by novelty-activated neurons, we labeled the Fos(+) neurons with GCaMP6 using the Fos-TRAP strategy as described (Fig. 3a). To compare these neurons with LH neurons not part of the novelty-activated ensemble, we employed a complementary Cre-off strategy to target the non-activated (Fos-) subset⁴¹. We injected AAV-cFos-CreERT2 into the LH combined with either AAV-EF1α-DIO-GCaMP6f (Cre-ON) to label the Fos(+) population, or AAV-hSyn-DO-GCaMP6f (Cre-OFF) to label the Fos(-) population, and data from control mice injected with either AAV-EF1α-DIO-eYFP or AAV-EF1α-DO-eYFP were pooled for analysis (Fig. 3a, b). Three weeks after tamoxifen-induced labeling, we validated this segregation by re-exposing mice to novel objects. Immunostaining revealed that ~64.1 % of GCaMP6(+) neurons in the Fos(+) group were reactivated (c-Fos-antibody-labeled), whereas only ~11.5 % of GCaMP(-) neurons in the Fos(-) group were labeled by the c-Fos-antibody (Supplementary Fig. 4a, b). These data confirm the reliable separation of the novelty-responsive (Fos(+)) neurons from non-responsive (Fos(-)) neurons.

Fig. 3: GCaMP6 signals in LH novelty-captured (Fos(+)) and non-captured (Fos(-)) neurons in response to nociceptive stimulation in SNI mice.

a Schematic for transfecting GCaMP6 into LH novelty-trapped (Fos(+)) and non-trapped (Fos(-)) neurons. Fiber photometry was performed during novel object exploration, formalin injection, and von Frey/thermal stimulation in SNI mice. Control data (DIO-eYFP and DO-eYFP) were pooled as eYFP neurons. b Representative images of GCaMP6-labeled Fos(+) and Fos(-) (green) neurons (representative of 4 independent mice). Scale bar, 100 μm. c Representative GCaMP6/eYFP signals (top) and distance to novel object (bottom). Blue shaded areas indicate object approaches. Average traces (d) and quantification (e) of signal changes during novel object approach (e, F_(2,57) = 24.76, P < 0.0001; Fos(+) n = 19, Fos(-) n = 19, eYFP n = 22 trials). Heat maps (f), average traces (g, h), and AUC values (i, j) of signals induced by hind paw formalin/saline injections (i, F_(2,55) = 40.98, P < 0.0001; j, F_(2,46) = 5.11, P = 0.0099). Heat maps (k, n, q, t), average traces (l, o, r, u), and AUC values (m, p, s, v) of signals responding to 0.16 g von Frey and 48 °C heating block on ipsilateral (k–p) and contralateral (q–v) paws of SNI mice. AUC quantifications: m F_(2,51) = 28.13, P < 0.0001; p F_(2,58) = 20.58, P < 0.0001; s F_(2,51) = 22.64, P < 0.0001; v F_(2,51) = 6.979, P = 0.0021. *P < 0.05, **P < 0.01, n.s, not significant. One-way ANOVA with Tukey’s post hoc test (e, i, j, m, p, s, v). Data are mean ± SEM. Dashed lines in (d, l, o, r, u) indicate stimulation onset. Data represent trial-based analysis from 6 mice per group. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

During the novel object test, Fos(+) neurons exhibited robust calcium transients time-locked to novel object approaches. In contrast, the Fos(-) population showed negligible activity changes during novel object exploration (Fig. 3c-e), confirming that novel object exploration primarily recruits the Fos(+) ensemble.

We next examined the responsiveness of these two populations to nociceptive stimuli in pain models. In the formalin test, Fos(+) neurons showed a robust increase in calcium activity during formalin-induced paw flinching compared to saline controls, whereas Fos(-) neurons exhibited only weak fluctuations (Fig. 3f-j). Importantly, this activity in Fos(+) neurons was driven by nociception rather than movement, as locomotion-associated paw lifting elicited no significant calcium changes (Supplementary Fig. 4c-e).

Similarly, in the bilateral SNI model, we recorded GCaMP6 signals from LH Fos(+) and Fos(-) neurons when we applied mechanical (0.16 g von Frey filament) and thermal (48°C) stimulation onto both hind paws. These stimuli evoked a sharp increase of GCaMP6 signal in Fos(+) neurons regardless of the stimulation side (Fig. 3k-v). In contrast, GCaMP6 signals in Fos(-) neurons exhibited significantly weaker responses to mechanical and thermal stimuli on hind paws (Fig. 3k-v).

In naïve mice, GCaMP6 signal in Fos(+) neurons increased upon acute suprathreshold stimulation (2 g von Frey filament or 48°C heating block), while Fos(-) neurons showed only marginal alterations in GCaMP6 signal, which was significantly weaker than that in Fos(+) neurons (Supplementary Fig. 4f-q).

Together, these data indicate that the LH novelty ensemble is implicated in processing nociception, independent of pain states.

Finally, we examined non-nociceptive salient stimuli (Supplementary Fig. 5). The Fos(+) ensemble was reliably recruited by diverse events, including air puffs, anxiety (entrance into the open arms in the elevated plus maze (EPM)), and rewards (water or sucrose). In contrast, Fos(-) neurons remained unresponsive to rewards and exhibited only weak activity upon exposure to aversive and anxious stimuli (air puffs or EPM) (Supplementary Fig. 5a–o). These data suggest that LH-novelty ensemble is not limited to pain processing but serves as a generalized hub for encoding high-salience information of both negative and positive valences, whereas LH non-novelty ensemble is not implicated in these behaviors.

LH novelty ensemble modulates pain and anxiety

Although our GCaMP6 data demonstrate that LH-novelty ensemble responds robustly to nociceptive and salient stimuli, their causal role in pain and emotional modulation remains unclear. To address this, we targeted these neurons using the Fos-TRAP strategy by co-injecting AAV-cfos-Cre-ERT2 with Cre-dependent ChR2, NpHR, or eYFP viruses into the LH (Fig. 4a). Following the same induction protocol described in Fig. 2b, viral expression was validated via histology and patch-clamp recordings (Fig. 4b, c; Supplementary Fig. 6a, b, 7a, b).

Fig. 4: Bidirectional optogenetic modulation of LH-novelty ensemble regulates pain and anxiety-like behavior.

a Schematic for expressing ChR2, NpHR, or eYFP in LH novelty ensemble. Representative images and electrophysiological traces confirming ChR2 (b) and NpHR (c) function. Similar results were obtained from 6 cells across 3 mice. Scale bars, 100 µm. Blue light illumination (ChR2 n = 8, eYFP n = 7 mice): (d) Formalin-induced licking time (t = 3.44, P = 0.003) and bouts (t = 3.18, P = 0.005). PWT and PWL 24 h post-CFA (e, PWT: F(_1,12) = 27.11, P = 0.0002; PWL: F_(1,13) = 16.92, P = 0.001) and 2 weeks post-SNI (f, PWT: F_(2,24) = 15.17, P < 0.0001; PWL: F_(2,24) = 35.35, P < 0.0001). g CPP time (eYFP t = 0.03, P = 0.98; ChR2 t = 3.35, P = 0.01). OFT tracks (h), center time (i, t = 2.16, P = 0.048), distance (j, t = 0.50, P = 0.62), and velocity (k, t = 0.50, P = 0.62). EPM tracks (l), open arms time (m, t = 4.18, P = 0.0009), and entries (n, t = 2.76, P = 0.02). Yellow light illumination (NpHR n = 7, eYFP n = 7 mice, unless noted): o Formalin-induced licking time (t = 5.36, P < 0.0001) and bouts (t = 3.35, P = 0.004) (NpHR n = 8 mice). PWT and PWL post-CFA (p, PWT: F_(1,12) = 21.82, P = 0.0005; PWL: F_(1,12) = 5.02, P = 0.04) and post-SNI (q, PWT: F_(2,24) = 5.88, P = 0.008; PWL: F_(2,24) = 14.37, P < 0.0001). r CPA time (eYFP t = 0.83, P = 0.43; NpHR t = 3.88, P = 0.006). OFT tracks (s), center time (t, t = 5.53, P < 0.0001), distance (u, t = 4.19, P = 0.0009), and velocity (v, t = 4.19, P = 0.0009). EPM tracks (w), open arms time (x, t = 2.32, P = 0.04), and entries (y, t = 2.50, P = 0.03). *P < 0.05, **P < 0.01, n.s, not significant. Two-sided unpaired t-test (d, i–k, m–o, t–v, x, y), two-way ANOVA with Tukey’s (e, f, p, q), paired t-test (g, r). Data are mean ± SEM. Blue/yellow dots in (g, r) represent light-paired sides; black dots indicate non-stimulated sides. Dashed squares in (h, s) outline OFT center. Black outlines in (l, w) denote EPM closed arms. PWT: paw withdrawal threshold; PWL: paw withdrawal latency. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

We first examined the effects of optogenetic activation (ChR2) in various pain models. In formalin-, CFA- (24-48 h post-injection), or SNI-treated mice, photostimulation of LH-novelty neurons (470 nm, 20 Hz, 4 mW) robustly reduced paw licking behaviors during the second phase (25-45 min) of the formalin test (Fig. 4d), and significantly increased pain thresholds in both CFA and SNI mice (Fig. 4e, f). Furthermore, in SNI mice with established chronic pain and negative emotions (4–5 weeks after surgery), ChR2 stimulation produced significant conditioned place preference (CPP) (Fig. 4g) and reversed anxiety-like behaviors, as indicated by increased center time in the open field test (OFT) (Fig. 4h-k), and longer duration and increased frequency of entries into the open arms of the EPM (Fig. 4l-n).

Conversely, we assessed the effects of bilateral optogenetic inhibition (NpHR) of LH-novelty ensemble. In contrast to activation of LH-novelty ensemble, inhibition of LH-novelty ensemble enhanced pain-like behaviors, including increased paw licking during the second phase of the formalin test (Fig. 4o), and reduced mechanical and thermal pain thresholds in both CFA and SNI models (Fig. 4p, q). Moreover, inhibiting these neurons in SNI mice induced conditional place aversion (CPA) (Fig. 4r) and exacerbated anxiety-like behaviors in the OFT and EPM (Fig. 4s-y).

We also tested the effects of bidirectional optogenetic manipulations of LH-novelty ensemble in mice under physiological conditions. Activation of LH-novelty ensemble did not alter pain thresholds (unlike in pain models) (Supplementary Fig. 6c-g), but induced CPP (Supplementary Fig. 6h-k) and anxiolytic-like effects in the EPM (Supplementary Fig. 6l-r). Conversely, inhibition decreased mechanical and thermal thresholds (Supplementary Fig. 7a-g), elicited CPA (Supplementary Fig. 7h-k), and enhanced anxiety-like behavior in both OFT and EPM (Supplementary Fig. 7l-r). Together, these results indicate that LH-novelty ensemble are essential for maintaining basal nociceptive thresholds and emotional state.

To verify the specificity of these stimulation-induced effects, we performed parallel experiments on the non-novelty (Fos-) population. We selectively targeted Fos(-) neurons by co-injecting AAV-cfos-Cre-ERT2 with AAV-hSyn-DO-ChR2- eYFP into the LH. After confirming that activation of the transfected ChR2 in Fos(-) neurons effectively evoked time-locked inward currents and neuronal firing (Supplementary Fig. 8a, b), we performed a series of behavioral tests in eYFP and ChR2 mice. We observed that activation of the Fos(-) neurons failed to alter pain thresholds (Supplementary Fig. 8c, d, h, i), place preference (Supplementary Fig. 8e, j), or anxiety-like behaviors (Supplementary Fig. 8f, g, k, l) in both non-injured and SNI mice. These results suggest that the novelty-responsive LH ensemble plays different roles from novelty-unrelated LH neurons in modulation of pain and emotions.

Novelty and pain activate LH GABA and glutamate neurons

The LH contains molecularly and functionally diverse neuron populations²⁷. To characterize LH-novelty ensemble, we co-injected AAV-c-fos-CreERT2 and AAV-EF1α-DIO-eYFP into the LH (Supplementary Fig. 9a), and labeled novelty-captured neurons as described in Fig. 2B. Subsequent immunohistochemical analysis of the eYFP(+) population revealed a heterogeneous composition: 48.79% of neurons were GABAergic (Supplementary Fig. 9b, c) and 25.15% of neurons were CaMKII(+) (Supplementary Fig. 9d, e). Furthermore, we found that specific peptidergic subpopulations were also recruited, with melanin-concentrating hormone (MCH)- and orexin-expressing neurons accounting for approximately 5.96% and 26.25% of the total c-Fos-positive population, respectively (Supplementary Fig. 9f-i). As CaMKII is frequently used as a marker for excitatory neurons^42,43, glutamatergic excitatory neurons may be major components in CaMKII(+) neurons in the LH. Given that orexin and MCH neurons often overlap with glutamatergic and GABAergic classes^44,45, we next sought to dissect the specific functional roles of GABAergic and glutamatergic neurons in LH novelty ensemble.

To label GABAergic and glutamatergic neurons in novelty exploration-activated ensemble, we employed a c-Fos promoter–driven, doxycycline (DOX)-inducible, Cre-dependent viral system in Vgat-Cre and Vglut2-Cre mice. Specifically, AAV-cfos-rtTA and AAV-TRE3G-DIO-GCaMP6f (or eYFP as control) were co-injected into the LH (Fig. 5a, g). Three weeks after injection, mice were given DOX in drinking water (2 mg/ml) for 3 days to prime the system, followed by exposure to a sequence of 6 novel objects for 1 h (10 min each) to induce c-Fos-mediated labeling. Following this novelty-tagging, we performed fiber photometry to record the calcium dynamics of these GCaMP6-labeled GABAergic and glutamatergic neurons during subsequent novelty exploration and pain behaviors (Fig. 5a, g; Supplementary Figs. 10a, 11a).

Fig. 5: LH GABAergic and glutamatergic novelty ensembles are activated in response to pain-like behaviors.

a Schematic diagram of Cre-dependent GCaMP6 or eYFP labeling of novelty-activated LH GABAergic neurons in Vgat-Cre mice. Post-viral injection (3 weeks) and DOX treatment (2 mg/ml, 3 days), mice explored 6 novel objects (10 min each, 1 h total). Right: representative GCaMP6 expression 1 week post-exploration. Scale bar, 100 μm. b Heat maps, traces, and quantification (F_(1,10) = 11.11, P = 0.008) of GCaMP6/eYFP signals responding to hind paw formalin/saline injections (15–16 trials; GCaMP6 n = 8, eYFP n = 4 mice). Signal changes responding to 0.16 g von Frey and 48 °C heating block in SNI mice (GCaMP6 n = 8, eYFP n = 4). AUC quantifications: c t = 3.66, P = 0.004; d t = 5.27, P = 0.0007; e t = 4.62, P = 0.001; f t = 2.36, P = 0.04. g Schematic diagram of GCaMP6/eYFP labeling of novelty-activated LH glutamatergic neurons in Vglut2-Cre mice. Right: representative GCaMP6 expression. h Signals responding to formalin/saline injections (interaction: F_(1,10) = 22.63, P = 0.0008; 21 trials; GCaMP6 n = 8, eYFP n = 4 mice). Signal changes responding to 0.16 g von Frey and 48 °C heating block in SNI mice (GCaMP6: 17–20 trials, n = 8 mice; eYFP: 12–15 trials, n = 4 mice). AUC quantifications: i t = 3.95, P = 0.003; j t = 3.49, P = 0.006; k t = 2.53, P = 0.03; l t = 4.41, P = 0.002. *P < 0.05, **P < 0.01, n.s, not significant. Two-way ANOVA with Tukey’s post hoc test (b, h), two-tailed unpaired t-test (c-f, i-l). Data are mean ± SEM. Dashed lines in (c–f, i–l) indicate stimulation onset. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

For LH^GABA-novelty ensemble, in vivo fiber photometry revealed robust calcium transients during novel object exploration (Supplementary Fig. 10b-e). These neurons were also consistently activated by noxious stimuli, including formalin injection, and suprathreshold mechanical (2 g in naïve, 0.16 g in SNI mice) and thermal stimulation, in mice under both physiological and SNI conditions (Fig. 5b-f; Supplementary Fig. 10f-i). Additionally, GCaMP6 signals responded to diverse emotional states: the signal increased during entry into the open arms of the EPM, while decreased in the closed arms of the EPM (Supplementary Fig. 10j, k), and showed moderate elevations during reward consumption (water/sucrose) and upon exposure to aversive air puffs (Supplementary Fig. 10l, m).

Similarly, LH^Glu-novelty ensemble exhibited increased GCaMP6 signal when mice explored novel objects (Supplementary Fig. 11a-e). Their activity was also elevated during nocifensive behaviors following formalin injection (Fig. 5h), suprathreshold mechanical and thermal stimulation in mice under physiological or SNI conditions (Fig. 5i-l; Supplementary Fig. 11f-i). Furthermore, these neurons responded to a broad range of emotional and motivational stimuli—including anxiety-provoking contexts (EPM exploration), reward (water/sucrose consumption) and aversion (air puff)—with activity patterns that closely mirrored those of the overall LH novelty ensemble (Supplementary Fig. 11j-m).

The strikingly similar response profiles of LH^GABA- and LH^Glu-novelty ensembles to novelty, pain, and emotional stimuli are likely underpinned by their comparable synaptic input patterns from diverse forebrain, midbrain, and brain stem regions, as revealed by rabies virus-based tracing (Supplementary Fig. 12a-e; 13a-e). Given these parallel response profiles, we next asked whether both subtypes also share the ability to regulate pain and anxiety behaviors.

LH^GABA- and LH^Glu-novelty ensemble modulate pain and anxiety

To investigate the functional roles of each neuronal subtype in pain and anxiety-like behavior, we employed the same Cre-dependent strategy used above (Fig. 5a, g), except that either AAV-TRE3G-DIO-ChR2-eYFP or AAV-TRE3G-DIO-Arch-eYFP was injected into the LH to allow for optogenetic activation or inhibition of LH^GABA-novelty ensemble in Vgat-Cre mice and LH^Glu-novelty ensemble in Vglut2-Cre mice (Figs. 6a-c, 7a-c; Supplementary Fig. 14a, 14k, 15a, 15k).

Fig. 6: Optogenetic manipulation of LHGABA-novelty ensemble modulates pain and anxiety-like behaviors.

a Schematic diagram showing DOX-enabled, Cre-dependent expression of ChR2, Arch, or eYFP in LH GABAergic novelty ensemble of Vgat-Cre mice. Validation of ChR2 (b) and Arch (c) function. Representative expression, photocurrent traces, and peak current amplitudes (ChR2 n = 8, Arch n = 7 cells). Scale bar, 100μm. Blue light illumination (ChR2 n = 8, eYFP n = 8 mice): d Formalin-induced licking time (t = 2.29, P = 0.039) and bouts (t = 1.38, P = 0.19) during LH illumination (5 ms, 4 mW, 20 Hz). PWT and PWL 24 h post-CFA (e, PWT: F_(1,14) = 15.52, P = 0.002; PWL: F_(1,14) = 27.60, P = 0.0001) and 2 weeks post-SNI (f, PWT: F_(1,14) = 14.84, P = 0.002; PWL: F_(1,14) = 71.67, P < 0.0001). g CPP time in paired chamber (eYFP t = 0.56, P = 0.60; ChR2 t = 2.92, P = 0.027). OFT tracks (h), center time (i, t = 2.21, P = 0.046), distance (j, t = 1.73, P = 0.11), and velocity (k, t = 1.80, P = 0.095). EPM tracks, open arms time (t = 2.84, P = 0.014), and frequency (t = 2.29, P = 0.038). Yellow light illumination (Arch n = 8, eYFP n = 8 mice): n Formalin-induced licking time (t = 2.50, P = 0.026) and bouts (t = 0.22, P = 0.831) during LH illumination (589 nm, 3 mW). PWT and PWL post-CFA (o, PWT: F_(1,14) = 12.53, P = 0.003; PWL: F_(1,14) = 50.98, P < 0.0001) and post-SNI (p, PWT: F_(1,14) = 7.23, P = 0.018; PWL: F_(1,14) = 28.21, P = 0.0001). q CPA time in paired chamber (Arch t = 2.47, P = 0.048; eYFP t = 1.33, P = 0.22). OFT tracks (r), center time (s, t = 3.48, P = 0.008), distance (t, t = 0.79, P = 0.44), and velocity (u, t = 0.83, P = 0.42). v, w EPM tracks, open arms time (t = 3.41, P = 0.011), and frequency (t = 2.56, P = 0.02). *P < 0.05, **P < 0.01, n.s, not significant. Two-sided unpaired t-test (d, i-k, m, n, s-u, w), two-way ANOVA with Tukey’s post hoc test (e, f, o, p), two-sided paired t-test (g, q). Data are mean ± SEM. Blue/yellow dots in (g, q) represent light-paired sides; black dots indicate non-stimulated sides. Dashed squares in (h, r) outline OFT center zone. Black outlines in (l, v) denote EPM closed arms. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

Fig. 7: Optogenetic manipulation of LH glutamatergic novelty ensemble modulates pain and anxiety-like behavior.

a Viral strategy for DOX-enabled, Cre-dependent expression of ChR2, Arch, or eYFP in the LH glutamatergic novelty ensemble of Vglut2-Cre mice. Validation of ChR2 (b) and Arch (c) showing viral expression, photocurrents, and peak amplitudes (n = 7 cells/group). Scale bar, 100 μm. LH blue light illumination (ChR2). d Formalin-induced licking time (t = 2.71, P = 0.018) and bouts (t = 3.07, P = 0.009). PWT and PWL in CFA (e, PWT: F_{(1, 13)} = 27.11, P = 0.0002; PWL: F_{(1, 13)} = 23.95, P = 0.0003) and SNI (f, PWT: F_{(1, 13)} = 120.6, P < 0.0001; PWL: F_{(1, 13)} = 36.11, P < 0.0001) mice. g Chamber time for eYFP (t = 1.39, P = 0.21, n = 8 mice) and ChR2 (t = 3.00, P = 0.024, n = 7 mice). OFT tracks (h), center time (i, t = 3.27, P = 0.006), distance (j, t = 2.10, P = 0.006) and velocity (k, t = 2.1, P = 0.064). l EPM open arm time (t = 2.19, P = 0.047) and frequency (t = 2.71, P = 0.01). ChR2 n = 7, eYFP n = 8 mice/group for (d–f, h-l). LH yellow light illumination (Arch). m Formalin-induced licking time (t = 2.29, P = 0.039) and bouts (t = 0.75, P = 0.47). PWT and PWL in CFA (n, PWT: F_{(1, 13)} = 7.40, P = 0.02; PWL: F_{(1, 13)} = 4.93, P = 0.045) and SNI (o, PWT: F_{(1, 13)} = 10.82, P = 0.006; PWL: F_{(1, 13)} = 9.15, P = 0.009). p CPP chamber time for Arch (t = 2.79, P = 0.03, n = 7 mice) and eYFP (t = 0.92, P = 0.39, n = 8 mice). OFT tracks (q), center time (r, t = 3.58, P = 0.006), distance (s, t = 0.56, P = 0.58) and velocity (t, t = 0.59, P = 0.56). u EPM open arm time (t = 2.18, P = 0.047) and frequency (t = 5.19, P = 0.0009). Arch n = 7 (m-o, q-t) or 8 (u), eYFP n = 8 mice/group. *P < 0.05, **P < 0.01. Two-way RM ANOVA with Tukey’s post hoc test (e, f, n, o), two-sided paired t-test (g, p), and two-sided unpaired t-test (d, i-m, r–u). Dots in (g, p) mark light-paired (color) and non-stimulated (black) sides. Dashed squares in (h, q) outline OFT center zone. Black outlines in (l, u) denote EPM closed arms. Data are mean ± SEM. Created in BioRender. Zhou, C. (2026) https://BioRender.com/ikcpymj. Source data are provided as a Source Data file.

Optogenetic activation (ChR2) of LH^GABA-novelty ensemble in pain models (formalin, CFA, or SNI) significantly mitigated pain behaviors, indicated by decreased formalin-induced licking, elevated pain thresholds in CFA and SNI mice, and induction of CPP in SNI mice (Fig. 6d-g). ChR2 activation also alleviated anxiety-like behaviors in SNI mice, reflected by increased time spent in the center in the OFT and increased open-arm entries in the EPM (Fig. 6h-m). In contrast, inhibition (Arch) of LH^GABA-novelty ensemble enhanced nocifensive behaviors in mice subjected to formalin injection into the hind paw, decreased pain thresholds and enhanced both aversive and anxiety-like behaviors in SNI mice (Fig. 6n-w). In naïve mice, activation of LH^GABA-novelty ensemble did not change pain thresholds (Supplementary Fig. 14b-d) but induced CPP (Supplementary Fig. 14e) and attenuated anxiety-like behavior in the OFT and the EPM (Supplementary Fig. 14f-j), whereas inhibition of LH^GABA-novelty ensemble reduced pain thresholds (Supplementary Fig. 14l-n), induced CPA (Supplementary Fig. 14o), and increased anxiety-like behavior in the OFT and EPM (Supplementary Fig. 14p-t).

Similarly, optogenetic activation (ChR2) of LH^Glu-novelty ensemble under pain conditions robustly suppressed formalin-induced licking, elevated pain thresholds in both CFA and SNI mice, established CPP, and reduced anxiety-like behavior (Fig. 7d-l). In non-injured mice, activation of these neurons did not change baseline nociceptive thresholds, but induced CPP and reduced anxiety-like behavior in the OFT and EPM (Supplementary Fig. 15b-j). Conversely, inhibition (Arch) of LH^Glu-novelty ensemble exacerbated nocifensive behavior, decreased pain thresholds, induced CPA, and worsened anxiety-like behavior in the OFT and EPM in mouse pain models (Fig. 7m-u). In naive mice, inhibition of these neurons reduced pain thresholds, induced CPA, and caused anxiety-like behavior in the OFT and EPM (Supplementary Fig. 15l-t).

These results suggest that both LH^GABA-novelty and LH^Glu-novelty ensemble have comparable roles in regulating pain and anxiety-like behaviors, similar to LH novelty ensemble.

Divergent LH novelty projections modulate pain and anxiety

Since activation of either inhibitory LH^GABA- or excitatory LH^Glu-novelty ensemble produces similar analgesic and anxiolytic effects, this raises the possibility that these neuronal populations may differ in their efferent profiles. Cre-dependent anterograde tracing in Vgat-Cre (Supplementary Fig. 16a, b) and Vglut2-Cre mice (Supplementary Fig. 17a, b) revealed that both subtypes broadly project to key regions regulating attention, pain and emotion, including the lateral preoptic nucleus (LPO), lateral habenula (LHb), ventral tegmental area (VTA), and lateral periaqueductal gray (LPAG), with extensive overlaps and subtle differences (Supplementary Figs. 16c-e, 17c-e), consistent with previous reports^27,46.

To dissect the circuit mechanisms, we focused on these four downstream targets (LPO, LHb, VTA, and LPAG) due to their dense innervation by both populations (Supplementary Figs. 16, 17). These nuclei are critically involved in pain and affective control and contain heterogeneous populations capable of exerting opposing regulatory effects^{47,48,49,50,51}, making them ideal candidates for testing projection-specific contributions.

We then selectively activated axon terminals of LH^GABA- and LH^Glu-novelty ensemble within the LPO, LHb, VTA and LPAG (Supplementary Fig. 18a, d). AAV-cfos-rtTA and AAV-TRE3G-DIO-ChR2-eYFP (or controls) were injected into the LH of Vgat-Cre and Vglut2-Cre mice, and optical fibers were implanted above each target region to enable terminal stimulation. Whole-cell patch-clamp recordings verified synaptic connectivity of these pathways: in Vgat-Cre mice, blue light-evoked inhibitory postsynaptic currents (IPSCs) were blocked by bicuculline (20 µM; Supplementary Fig. 18b), confirming GABAergic transmission, while in Vglut2-Cre mice, blue light-evoked excitatory postsynaptic currents (EPSCs) were abolished by CNQX (20 µM) plus AP-5 (50 µM; Supplementary Fig. 18e), confirming glutamatergic transmission.

Behaviorally, activating individual projections of LH^GABA-novelty ensemble yielded dissociable outcomes in SNI mice that did not simply mirror somatic activation (Supplementary Fig. 18c, Fig. 6). Stimulating terminals in the LPO (Fig. 8a-f) or VTA (Fig. 8g-l) failed to produce analgesia (VTA stimulation even lowered thresholds), induced CPA, and heightened anxiety, different from the behavioral outcomes following somatic stimulation of LH^GABA-novelty ensemble. In contrast, activating the projections to the LHb elevated pain thresholds, without altering place preference or anxiety-like behaviors (Fig. 8m-r). By comparison, activation of the projection to the LPAG produced no significant effects on pain thresholds or anxiety-like behavior (Fig. 8s-x). Thus, projections to the LHb likely mediate analgesic effects of LH^GABA-novelty ensemble. These results suggest that the anxiolytic effect of LH^GABA-novelty ensemble may be mediated by their outputs other than the examined ones.

Fig. 8: LH GABAergic novelty ensemble projections modulate pain, aversion, and anxiety in SNI mice.

ChR2 or eYFP was expressed in the LH GABAergic ensemble (Fig. 6a), with optical fibers in LPO (a–f), VTA (g–l), LHb (m–r), and LPAG (s-x). LPO photostimulation. PWT (a, F_(1,13) = 2.06, P = 0.18) and PWL (b, F_(1,13) = 0.39, P = 0.54) (ChR2 n = 8, eYFP n = 7 mice). CPP time for ChR2 (c, t = 3.25, P = 0.017) and eYFP (d, t = 0.75, P = 0.48); n = 7 mice/group. Time in open field (OF) center (e, t = 2.79, P = 0.016) and EPM open arms (f, t = 2.46, P = 0.03); n = 7 mice/group. VTA photostimulation. PWT (g, F_(1,13) = 5.77, P = 0.03) and PWL (h, F_(1,13) = 40.77, P < 0.0001) (ChR2 n = 7, eYFP n = 8 mice). CPP time for ChR2 (i, t = 2.76, P = 0.033) and eYFP (j, t = 0.37, P = 0.72); n = 7 mice/group. Time in OF center (k, t = 2.21, P = 0.048) and EPM open arms (l, t = 3.12, P = 0.009); n = 7 mice/group. LHb photostimulation. PWT (m, F_(1,13) = 35.61, P < 0.0001) and PWL (n, F_(1,13) = 99.09, P < 0.0001) (ChR2 n = 7, eYFP n = 8 mice). CPP time for ChR2 (o, t = 0.56, P = 0.60) and eYFP (p, t = 0.42, P = 0.69); n = 7 mice/group. Time in OF center (q, t = 0.54, P = 0.60) and EPM open arms (r, t = 0.45, P = 0.66); n = 7 mice/group. LPAG photostimulation. PWT (s, F_(1,13) = 0.10, P = 0.75) and PWL (t, F_(1,13) = 0.29, P = 0.59) (ChR2 n = 7, eYFP n = 8 mice). CPP time for ChR2 (u, t = 0.42, P = 0.69) and eYFP (v, t = 1.32, P = 0.23); n = 7 mice/group. Time in OF center (w, t = 1.36, P = 0.20) and EPM open arms (x, t = 0.23, P = 0.82); n = 7 mice/group. *P < 0.05; **P < 0.01; n.s, not significant. Two-way ANOVA with Tukey’s post hoc test (a, b, g, h, m, n, s, t), two-sided paired t-test (c, d, i, j, o, p, u, v), and two-sided unpaired t-test (e, f, k, l, q, r, w, x). Data are mean ± SEM. Source data are provided as a Source Data file.

In contrast, several outputs of LH^Glu-novelty ensemble mimicked the ensemble in modulating pain and emotions (Supplementary Fig. 18f). Activation of their projections to the LPO and LPAG both robustly alleviated pain and anxiety-like behaviors, mirroring somatic activation (Fig. 9a-f, s-x; Fig. 7). Activation of the projection to the VTA primarily alleviated pain with minimal impact on anxiety-like behavior (Fig. 9g-l), whereas activation of the projection to the LHb caused hyperalgesia without affecting anxiety-like behavior (Fig. 9m-r).

Fig. 9: LH glutamatergic novelty ensemble projections modulate pain, aversion, and anxiety in SNI mice.

ChR2 or eYFP was expressed in the LH glutamatergic ensemble (Fig. 7a) with optical fibers in LPO (a-f), VTA (g-l), LHb (m-r), and LPAG (s-x). LPO photostimulation. PWT (a, F_(1,14) = 32.96, P < 0.0001) and PWL (b, F_(1,14) = 57.07, P < 0.0001); n = 7 mice/group. CPP time for ChR2 (c, t = 2.80, P = 0.027) and eYFP (d, t = 2.00, P = 0.09); n = 8 mice/group. Time in open field (OF) center (e, t = 2.51, P = 0.02) and EPM open arms (f, t = 2.33, P = 0.036); n = 8 mice/group. VTA photostimulation. PWT (g, F_(1,14) = 15.77, P = 0.001) and PWL (h, F_(1,14) = 23.22, P = 0.0003); n = 8 mice/group. CPP time for ChR2 (i, t = 1.70, P = 0.13) and eYFP (j, t = 0.91, P = 0.39); n = 8 mice/group. Time in OF center (k, t = 1.42, P = 0.18) and EPM open arms (l, t = 0.81, P = 0.43); n = 7 mice/group. LHb photostimulation. PWT (m, F_(1,15) = 10.56, P = 0.005) and PWL (n, F_(1,15) = 25.31, P = 0.0001) (ChR2 n = 9, eYFP n = 8 mice). CPP time for ChR2 (o, t = 0.46, P = 0.60) and eYFP (p, t = 0.61, P = 0.56); n = 8 mice/group. Time in OF center (q, t = 1.19, P = 0.25) and EPM open arms (r, t = 0.42, P = 0.68); n = 8 mice/group. LPAG photostimulation. PWT (s, F_(1,15) = 58.04, P < 0.0001) and PWL (t, F_(1,15) = 58.38, P < 0.0001) (ChR2 n = 9, eYFP n = 8 mice). CPP time for ChR2 (u, t = 4.33, P = 0.003) and eYFP (v, t = 0.05, P = 0.96); n = 8 mice/group. Time in OF center (w, t = 2.27, P = 0.04) and EPM open arms (x, t = 2.22, P = 0.04); n = 8 mice/group. *P < 0.05; **P < 0.01; n.s, not significant. Two-way ANOVA with Tukey’s post hoc test (a, b, g, h, m, n, s, t), two-tailed paired t-test (c, d, i, j, o, p, u, v), and two-tailed unpaired t-test (e, f, k, l, q, r, w, x). Data are mean ± SEM. Source data are provided as a Source Data file.

Together, these data indicate that LH^GABA- and LH^Glu-novelty ensembles engage partially overlapping but functionally divergent downstream circuits that differently control pain and anxiety. The similar net analgesic and anxiolytic effects observed with somatic activation likely arise from integration across these projections. Activating LH^GABA-novelty ensemble appears to cause analgesia via inhibiting the LHb, and activation of LH^Glu-novelty ensemble leads to analgesia and anxiolysis via stimulating LPO/LPAG (with VTA contributing to analgesia). These findings underscore a complex output pattern where distinct projections differentially contribute to the overall behavioral effects of LH novelty ensemble.

Discussion

In this study, we dissected the neural circuits by which novelty exploration modulates pain perception and negative affect. Our results demonstrate that novelty-driven distraction suppresses both the sensory and affective dimensions of pain. Through viral vector-labeling, in vivo calcium imaging, and optogenetic circuit manipulation, we identified a subset of GABAergic and glutamatergic LH neurons (LH-novelty ensemble) that act as a shared substrate for novelty processing and pain/anxiety modulation. Selective activation of their projections to the LPO, LHb, VTA, and LPAG regulates pain and emotion in distinct patterns, highlighting a specific circuit mechanism implementing attention distraction.

Our findings reinforce and extend existing literature demonstrating the analgesic potential of attention- or novelty-based distraction in human studies^1,4,7,9,52. Using two distinct novelty paradigms (novel object and novel environment), we showed that acute novelty exposure reliably attenuated pain behaviors in both male and female mice across acute (formalin) and persistent (CFA, SNI) pain models. While previous studies implicate opioid systems in distraction-induced analgesia^6,17,19, systemic naltrexone failed to block novelty-induced analgesia in formalin-induced inflammatory pain, suggesting the involvement of a non-opioid mechanism in this specific context. Although antagonist efficacy of naltrexone varies with pain models and opioid receptor subtype⁵³, our data suggest that novelty-induced analgesia may recruit distinct circuits, such as the LH pathways identified here, that operate in parallel to opioid systems.

Through comprehensive c-Fos mapping and Fos-TRAP labeling, we identified the LH as an important region integrating novelty and pain signals. Specifically, the LH-novelty ensemble is reactivated by noxious stimuli, and conversely, pain-activated neurons respond to novelty exploration. This bidirectional overlap, confirmed by fiber photometry, suggests that these neurons may function as a general salience-processing hub that integrates multimodal salient stimuli rather than acting as a dedicated detector for novelty alone. Crucially, this functional profile appears specific to the novelty-activated population because novelty-unrelated (Fos(-)) LH neurons exhibited minimal responses to noxious stimuli and rewards. This distinction was further validated causally, as optogenetic activation of the Fos(-) population failed to alter pain thresholds and anxiety-like behaviors, in contrast to the potent modulation of pain and anxiety-like behavior observed when the Fos(+) ensemble was activated.

Regarding cell-type specificity, we found that the LH-novelty ensemble comprises both GABAergic and glutamatergic populations. Both types were activated by novelty exploration and noxious stimulation, which may reflect shared synaptic inputs. Functionally, optogenetic activation of either LH^GABA- or LH^Glu-novelty ensemble attenuated hyperalgesia and anxiety-like behaviors in pain models, while their inhibition exacerbated these symptoms. Previous studies have shown that CaMKIIα(+), orexinergic, and parvalbumin(+) neurons in the LH modulate pain and related emotion^27,32,54. Our study extends these findings by demonstrating that a group of GABAergic and glutamatergic neurons are naturally recruited by novelty and are capable of modulating both pain and negative emotions. While the observed analgesia may partially result from anxiety relief because of the inherent link between pain and emotion⁵⁵, our projection-specific data (see below) reveal that these behaviors can be regulated via mechanistically dissociable pathways.

Despite releasing different neurotransmitters and leading to opposite modulation of activity in downstream neurons, LH^GABA- and LH^Glu-neurons are known to play similar roles in promoting arousal and wakefulness^43,56,57,58. Consistently, we found that activation of somata of either LH^GABA- or LH^Glu-novelty ensembles yielded convergent analgesic and anxiolytic phenotypes. However, activation of their downstream projections exerted distinct effects. We identified that the analgesic effects of the LH^GABA-novelty ensemble were primarily mediated by the LHb projection, whereas those of the LH^Glu-novelty ensemble were mimicked by its projections to the LPO, LPAG, and VTA. This discrepancy suggests that the activation of somata of LH novelty ensemble engages a heterogeneous mixture of collateral projections, the net effect of which may mask the functional diversity of specific circuits^59,60. Thus, examination of the function of specific projections offers more fruitful insights than manipulating cell bodies alone. Moreover, the distinct behavioral outcomes elicited by manipulating different projections demonstrate advantages over electrical stimulation by avoiding the activation of non-specific fibers contributing to local circuits or passing by.

Recent studies highlight the molecular heterogeneity of the LH, most notably the presence of orexin- and MCH-expressing populations^46,61. Our data identified some of these peptidergic neurons more or less as major components within the LH novelty ensemble. Given that orexin and MCH neurons largely overlap with glutamatergic and GABAergic populations, respectively^44,45, our optogenetic manipulations likely recruited these peptidergic neurons. While previous studies indicate that these neuropeptides can modulate pain and emotion^29,62, our current study focused on GABAergic and glutamatergic neurons. Dissecting the specific contributions of these peptidergic subtypes warrants further investigation.

Several limitations of this study should be noted. First, while we focused on LH projections, many downstream targets are themselves interconnected²³, raising the possibility that polysynaptic or reciprocal interactions may contribute to the observed effects, especially during persistent pain states. Second, although we defined LH projections by neurotransmitter phenotype, the cellular diversity of recipient neurons within each downstream target remains unclear. More detailed circuit mapping and cell-type-specific analyses are needed to articulate these pathways.

In summary, we identified a subpopulation of LH neurons linking novelty exploration with relief from pain and anxiety. Both LH^GABA- and LH^Glu-novelty ensembles contribute to these effects by recruiting distinct output pathways. These findings reveal a neural circuit independent of the opioidergic system mediating the influence of attention on pain and anxiety, and suggest potential targets for neuromodulation-based therapies for chronic pain.

Methods

Animals and ethics statement

All procedures for animal care and experimentation (Protocol No. 202207S123) were reviewed and approved by the Institutional Animal Care and Use Committee and the Office of Laboratory Animal Resources at Xuzhou Medical University, in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (1988, China). Mice were maintained in the university’s animal facility in a strictly controlled environment, consisting of a 12-h light/dark cycle, an ambient temperature of 21–22 °C, and a relative humidity of 50 ± 10%, with ad libitum access to food and water. Mice were typically group-housed (5 per cage). The transgenic mouse lines used included Vglut2-Cre (The Jackson Laboratory, stock No. 016963) and Vgat-Cre (The Jackson Laboratory, stock No. 017535), both maintained on a C57BL/6J genetic background. Wild-type C57BL/6J mice were purchased from the animal facility in Xuzhou Medical University. Since no significant sex differences were observed in novelty-related pain modulation experiments, approximately equal numbers of male and female mice were included in neuronal tracing and modulation experiments, and data from both sexes were pooled for these analyses. All surgeries were performed on mice aged 8–12 weeks. All behavioral assessments were performed during the light phase.

Viral constructs

A series of adeno-associated viral (AAV) vectors were obtained from Brain VTA (Wuhan, China), including but not limited to: AAV-cfos-Cre-ERT2, AAV-EF1α-DIO-hChR2(H134R)-eYFP, AAV-EF1α-DIO-NpHR-eYFP, AAV-EF1α-DIO-eYFP, AAV- EF1α-DIO-GCaMP6s, AAV-hSyn-DO-hChR2(H134R)-eGFP, AAV-hSyn-DO-GCaMP6, AAV-hSyn-DO-eYFP, AAV-hSyn-DIO-mGFP-synaptophysin-mRuby, AAV-CaMKII-ChR2-eYFP, AAV-cfos- rtTA, AAV-TRE3G-DIO-GCaMP6m, AAV-TRE3G-DIO-ChR2-eYFP, AAV-TRE3G- DIO-Arch-eYFP, AAV-TRE3G-DIO-eYFP, AAV-EF1α-DIO-TVA-eGFP, AAV-EF1α-DIO-RG, and RV-EnvA-ΔG-dsRed. Titers for AAVs and RV were 2 × 10¹² to 5 × 10¹² vg/ml and 2 × 10⁸, respectively.

Stereotaxic surgery

Stereotaxic injections and optical fiber implantations were performed as previously described^63,64. Mice were anesthetized using isoflurane (3% for induction, 1.5% for maintenance) and stabilized on a stereotaxic frame (RWD Life Science, Shenzhen, China). Their bodies were placed on a heating pad. After the head skin of a mouse was disinfected, its skull was exposed and viral vector was injected into the LH (AP: −1.20; ML: ±1.10; DV − 5.10 mm) using a 10 μl Hamilton syringe at a speed of 50 nL/min controlled by a micro-syringe pump (KD Scientific, Holliston, MA, USA). For optogenetic or fiber photometry experiments, optical fibers (200 µm in diameter, NA, 0.37, Inper, Hangzhou, China) were implanted in individual brain regions at the following coordinates (relative to bregma, in mm), including LH (AP: −1.20; ML: ±1.10; DV: −5.10), LHb (AP: −1.58; ML: ±0.50; DV: −2.80), VLPAG (AP: −4.70; ML: ±0.30; DV − 2.75), LPO (AP: +0.26; ML: ±0.60; DV − 5.20), and VTA (AP: −3.20; ML: ±0.50; DV − 4.50). Postoperative analgesia was achieved by administering meloxicam (4 mg/kg, Aladdin, Shanghai, China) in the drinking water for three days.

Novelty exploration paradigms

Mice were randomly assigned to one of three groups: (1) Control, exposure to a familiar arena; (2) Novel Object, exposure to a novel object placed at the center of a familiar arena; or (3) Novel Environment, exposure to a novel square testing arena. Before testing, mice were place in the familiar environment, a cylindric open-field arena made of 3 mm thick transparent plexiglas (30 cm in diameter, 40 cm in height), to habituate for 10 min per day over seven consecutive days.

On the test day, all mice received 1% formalin (20 μl) injection into the hind paw, followed by a 25 min stay in the home cage, then were placed individually into their testing environments for 20 min examination. For the Novel Environment group, mice were place in a novel environment (a cubic chamber, 30 cm in side length). For the Novel Object group, mice were place in the familiar environment (a cylindric arena) and were sequentially exposed to 6 novel objects (plastic or metal, 3 cm in the longest dimension, cubic or cone shapes, yellow or green) (one object for 10 min, 60 min in total). For the control group, mice were placed in the familiar arena without a novel object.

Activity-dependent neuronal labeling

To label novelty-activated neurons with the Fos-TRAP technique, AAV-cfos-Cre-ERT2 was co-injected into the LH (AP: −1.20; ML: ±1.10; DV: −5.10 mm) with Cre-dependent AAVs (AAV-EF1α-DIO-GCaMP6, AAV-EF1α-DIO-eYFP, AAV-EF1α-DIO-ChR2-eYFP, or AAV-EF1α-DIO-NpHR-eYFP). After 3 weeks’ recovery, tamoxifen (TOM, 15 mg/kg, i.p.) was administered, and 5 h later, mice were exposed to novel objects (6 different objects, 10 min each, total 1 h). This process activated c-fos-expression in novelty-responsive LH neurons, driving Cre-recombinase expression and subsequent reporter/effector expression. Throughout the manuscript, we refer to these cells as novelty-trapped neurons (or simply novelty ensemble). We note that this is an operational definition based on the labeling window and does not imply absolute functional exclusivity for novelty detection. Mice recovered for at least one week before further experiments.

For labeling the complementary population (neurons not captured during novelty exposure), separate groups of mice received co-injection of AAV-cfos-Cre-ERT2 along with DO AAVs (AAV-hSyn-DO-hChR2(H134R)-eGFP, or AAV-hSyn-DO-GCaMP6s) into the LH. Tamoxifen and novel object exposure procedures were identical to the Fos-TRAP group. In this strategy, neurons activated by novelty (expressing c-fos-driven Cre-recombinase) had the DO construct inverted, preventing expression of ChR2-eGFP or GCaMP6. Conversely, neurons that did not undergo recombination (non-novelty ensemble neurons) (Fos(-) neurons) maintained the construct in a functional orientation, resulting in constitutive expression. Mice were allowed to recover until 3 weeks after virus injection.

For activity-dependent neuronal labeling in Vgat-Cre or Vglut2-Cre transgenic mice. Mice received stereotaxic co-injection of AAV-cfos-rtTA with AAV-TRE3G-DIO-GCaMP6m, AAV-TRE3G-DIO-ChR2-eYFP, AAV-TRE3G-DIO-Arch-eYFP, or AAV-TRE3G-DIO-eYFP into the LH (AP: −1.20; ML: ±1.10; DV: −5.10 mm). 3 weeks later, mice were provided with drinking water containing doxycycline (DOX; 2 mg/ml, supplemented with 1% sucrose) for three consecutive days to allow the binding of rtTA with TRE3G if these molecules were expressed. Then, mice were exposed to the novelty exploration paradigm as described above. This approach enables c-Fos-driven expression of rtTA. The binding of rtTA and TRE3G allows Cre-dependent expression of GCaMP6m, ChR2-eYFP, Arch-eYFP, or eYFP.

For both labeling strategies, behavioral testing or brain tissue collection for histological analysis was performed at least one week after novelty exploration.

Spared nerve injury

Neuropathic pain mouse model was established with spared nerve injury (SNI) of the sciatic nerve following a standard protocol^65,66. Under isoflurane anesthesia, the sciatic nerve was exposed and two branches (the common peroneal and tibial nerves) were ligated and transected, leaving the third branch (the sural nerve) intact. Sham controls underwent nerve exposure without damage. Mice recovered on a heating pad after surgery until they woke up, then were placed in a clean cage.

von Frey filament test

Mechanical sensitivity was assessed using von Frey filaments (0.01–2 g) after habituation on a mesh platform at least for 1 h. Filaments were applied to the plantar of the hind paw, and withdrawal thresholds were determined using the up-down method^63,65,66.

Hot plate test

The hot plate test was exclusively used to assess nociceptive escape behaviors (shown in Fig. 1m-q) during novel object exposure. Mice were placed individually on a hot plate apparatus (Zhenghuabiolog, Huaibei, China) maintained at 52 °C. The latency to the first pain response, such as hind paw licking, shaking, or jumping, was recorded as the pain threshold. In novelty conditions, mice were exposed to a novel object on the hot plate. A cutoff time (typically 30 s) was set to prevent tissue damage. The apparatus was cleaned between trials.

Hargreaves’ test

For all other behavioral assessments of thermal sensitivity throughout the study, the Hargreaves test was utilized to measure the paw withdrawal reflex; the two methods were not used interchangeably. After 1 h habituation on a glass surface, thermal sensitivity was measured using the Hargreaves’ apparatus (Boerni, Tianjin, China). Radiant heat was applied to the hind paw and withdrawal latencies were recorded over three trials per paw, and the average was used for analysis^66,67.

Formalin-induced pain model (hind paw injection)

20 μl 1% formalin was injected subcutaneously into the plantar surface of the hind paw. Immediately after injection, mice were placed in a transparent observation chamber, and the nocifensive behaviors (e.g., licking, biting, or shaking of the injected paw) were recorded. Responses were typically analyzed in two phases: phase I (0–5 min, acute response) and phase II (25–45 min, inflammatory response).

Complete Freund’s adjuvant-induced pain model

CFA (20 µl) was injected in the plantar area of the hind paw to induce inflammatory pain.

The open field test

Each mouse was placed in the center of an open field arena (usually 40 cm × 40 cm × 40 cm) and allowed to explore freely for 6 min. Locomotor activity was recorded using the EthoVision XT tracking system (Noldus Information Technology, Wageningen, the Netherlands). Parameters, including total distance traveled, number of entries, and time spent in the center zone, were quantified to assess general activity and anxiety-like behavior. The arena was cleaned with 75% ethanol between tests.

Elevated plus maze (EPM)

The EPM consists of two open arms and two closed arms (each arm typically 30 cm long and 5 cm wide) elevated 50 cm above the floor. Mice were placed at the junction of the arms facing an open arm and allowed to explore for 6 min. The number of entries and time spent in the open arms were recorded and analyzed with EthoVision XT (Noldus Information Technology, Wageningen, the Netherlands). The increase in time in the open arms represents reduced anxiety-like behavior. The maze was cleaned between trials.

Conditioned place preference (CPP) test

CPP experiments were conducted using a two-chamber apparatus with distinct visual and tactile cues. Mice were allowed to freely explore both chambers during 20 min pre-conditioning session to assess baseline chamber preference. Mice showing more than 75% initial preference for either chamber were excluded from further testing.

For the CPA paradigm involving behavioral intervention, mice received repetitive von Frey filament (2 g) stimulation in one chamber. In the social novelty condition, a conspecific stranger was introduced into a metal cylinder-shaped cage in the same chamber during conditioning.

For the optogenetic CPP paradigm, mice received light stimulation in one chamber during conditioning (parameters detailed in the optogenetic stimulation section), with no stimulation in the other chamber.

Twenty-four hours after conditioning sessions, a test session was conducted in which mice had free access to both chambers, and no stimulation or light was delivered. The time spent in the conditioned chamber was automatically recorded and analyzed with EthoVision XT (Noldus Information Technology, Wageningen, the Netherlands). Chamber preference or aversion was determined by comparing the time spent in the stimulus-paired vs. unpaired chamber before and after conditioning.

In vivo optogenetic stimulation

For in vivo optogenetic manipulation, light power at the fiber tip was measured before each experiment. For activation experiments (ChR2), blue light (473 nm, 20 Hz, 5 ms pulse width, ~4 mW) was delivered. For inhibition experiments, yellow light (constant, ~3 mW) was delivered. The specific stimulation paradigms for each behavioral assay were as follows:

Conditioned place preference (CPP): During the 30-min conditioning sessions, light was delivered in the stimulus-paired chamber using a repeating cycle of 120 s ON and 60 s OFF.

Formalin test: Light stimulation (120 s ON, 60 s OFF cycle) was delivered into the LH, and the nociceptive licking behavior was specifically recorded and analyzed during the late phase (between 20 and 40 min) after formalin injection into the hind paw.

Open field test (OFT) and Elevated plus maze (EPM): Light stimulation was delivered continuously throughout the entire duration of the testing sessions.

Pain assessments (von Frey and Hargreaves tests): Light stimulation was initiated 1 min prior to the presentation of the mechanical or thermal stimulus and remained continuous until the pain reflex behavior was completed.

Fiber photometry

To monitor the activity of LH neurons, a fiber photometry instrument (ThinkerTech, Nanjing, China)^65,68 was used. A 470 nm LED (typical <0.05 mW at fiber tip) was used for exciting GCaMP6, while a 405 nm LED served as an isosbestic control to model and remove slow fluctuations primarily caused by photo-bleaching and movement artifacts. The raw fluorescence signals were digitized at a sampling rate of 1000 Hz. To process and analyze the fiber photometry recording data, raw fluorescence traces were first down-sampled offline to 100 Hz for manageable data size and reduced noise. The 405 nm signal was then used to correct GCaMP6 signal for non-neuronal artifacts.

To quantify the activity changes, we averaged GCaMP6 signal (F) during a 3 s baseline period immediately preceding the event to calculate the mean (Mean_baseline) and standard deviation (SD_baseline). The data were then converted into z-scores using the formula: Z = (F-Mean_baseline)/SD_baseline. The magnitude of the response was quantified by calculating the area under the curve (AUC) of the Z-score traces within a defined 5 s post-stimulus window. When comparing the GCaMP6 signals of Fos+ and Fos- neurons in the LH, data from mice injected with either AAV-EF1α-DIO-eYFP or AAV-hSynDO-eYFP were pooled into a single eYFP group as a control, since neither exhibited stimulus-evoked fluorescence changes.

Brain slice electrophysiology

Brain slice electrophysiological recording was conducted with whole-cell patch-clamp recording^63,65,68. Acute coronal slices (280 μm) containing the LH, LPO, LHb, LPAG or VTA were prepared with a Leica VT-1200 vibratome (Nussloch, Germany) in ice-cold, oxygenated sucrose-based cutting solution (in mM) (85 NaCl, 75 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 4.0 MgCl₂, 0.5 CaCl₂, 24 NaHCO₃, and 25 glucose). After 1 h incubation at 32 °C, slices were transferred to standard artificial cerebrospinal fluid (ACSF) at 26 °C for at least 30 min. ACSF contains (in mM) 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 1.2 MgCl₂, 2.4 CaCl₂, 26 NaHCO₃, and 11 glucose.

Whole-cell patch-clamp recordings were performed using an electrophysiological rig equipped with a Nikon FN-1 upright microscope (Tokyo, Japan), an MultiClamp 700B amplifier, a Digidata 1500B, MultiClamp commander and pClamp 10.7 software (Molecular Devices, San Jose, CA, USA). Neurons with holding currents >–50 pA or resting membrane potentials above –40 mV were excluded. Optogenetic stimulation was delivered via an optical fiber (200 µm, NA 0.37) from PlexBright LED-light sources (460 nm/560 nm, 2 mW) (Plexon Inc., Hong Kong, China). Pharmacological properties of blue-light-evoked inhibitory and excitatory postsynaptic currents (IPSCs or EPSCs) were characterized with either 10 µM bicuculline (BIC) or 20 µM CNQX + 50 µM AP-5.

Immunohistochemistry and imaging

Mice were euthanized via CO₂ inhalation and transcardially perfused with PBS followed by 4% paraformaldehyde. Brains were extracted, post-fixed in 4% paraformaldehyde, cryoprotected in 30% sucrose, and sectioned coronally (30 μm) using a Leica CM1950 cryostat. Slices were blocked (10% donkey serum) and permeabilized (0.1% Triton X-100) for 1.5 h at room temperature, then incubated with primary antibodies overnight at 4 °C. Following PBS washes, sections were incubated with Alexa Fluor-conjugated secondary antibodies for 2 h at room temperature. Slices were then washed, dried, mounted with medium (MeilunBio, MA0222), and imaged using an Olympus FV-1000 confocal microscope. Image processing was performed using ImageJ (NIH).

Primary antibodies: c-Fos (Rabbit mAb, Clone 9F6, Cell Signaling Technology 2250, 1:1000); NeuN (Guinea pig polyclonal, Millipore ABN90, 1:2000); CaMKII (Mouse IgG, Cell Signaling Technology 50049, 1:400); GABA (Rabbit polyclonal, ProteinTech 16093-1-AP, 1:200); MCH (Rabbit polyclonal, Abcam ab274415, 1:200); Orexin (Rabbit polyclonal, ABclonal A23383, 1:200).

Secondary antibodies (Jackson ImmunoResearch): Alexa Fluor 488-conjugated Donkey anti-Rabbit IgG (H + L) (711-545-152, 1:500); Alexa Fluor 555 (or Cy3)-conjugated Donkey anti-Guinea pig IgG (H + L) (706-165-148, 1:500); Alexa Fluor 647-conjugated Donkey anti-Mouse IgG (H + L) (715-605-150, 1:500).

Cell counting and image analysis

An investigator blinded to animal groups performed c-Fos cell counting using ImageJ. A uniform fluorescence intensity threshold was applied across all images to objectively isolate intensely activated c-Fos+ neurons from background signals. ROIs were manually delineated for each section based on distinct anatomical boundaries of the LH. For LH identification (approx. Bregma −0.6 to −2.2 mm), boundaries were defined laterally to the fornix (f) and medioventrally to the internal capsule (ic), according to the mouse brain atlas (Paxinos and Franklin). To ensure consistency across mice, we analyzed an equal number of sections per brain at corresponding neuroanatomical levels. In addition, a size filter 40–120 μm² and circularity index (0.5–1.0) were applied to specifically include neuronal somata while excluding small artifacts, non-neuronal debris, or fragments of processes. This filtering accounts for the conservative counts in both control and experimental groups. The center of this mapped LHA novelty ensemble guided subsequent viral injection coordinates.

Chemicals

D-2-Amino-5-phosphonovaleric acid (AP-5) (Cat No. HY-100714A), bicuculline methobromide (Cat No. HY-100783), complete Freund’s adjuvant (CFA) (Cat No. HY-153808), CNQX disodium (FG9065 disodium) (Cat No. HY-15066A), and Naltrexone (Cat No. HY-76711) were purchased from MedChem Express (Monmouth Junction, NJ, USA). Aladdin Scientific (Shanghai, China) provided meloxicam (Cat No. M129228).

Data analysis and Statistics

All analyses were performed with GraphPad Prism 8.0. Electrophysiological and fiber photometry data were processed with Clampfit 10.7 (Molecular Devices, San Jose, CA). Figures were composed using Adobe Illustrator 2020. Data are presented as mean ± SEM. Comparisons between two groups were made with paired or unpaired two-tailed t-tests. Multi-group comparisons with one-way or two-way ANOVAs (with repeated-measures as appropriate), followed by Tukey’s post-hoc test. Mann-Whitney U-test or nonparametric ANOVA was used when variance assumption was invalid. Details of statistical testing, sample size (n), and significance are provided in figure legends. Power analyses (StatMate 2.0) ensured group sizes were adequate for ≥80% statistical power. Schematic diagrams and illustrations were custom-drawn by the authors. Specific elements, such as the brain section, were created with BioRender.com under a paid subscription (Publication Licenses have been provided). No generative AI tools were used in the creation of any text or images in this manuscript.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.