A dermo-hypothalamic axis driven by TNF-α mediates sleep disturbances in psoriasis

Abstract

Sleep disturbances are a debilitating feature of psoriasis, but whether they arise from itching or from direct effects of inflammation on the brain remains unclear. Here we show, using clinical data and a mouse model of psoriasis-like inflammation, that affected animals exhibit marked wakefulness and fragmented non-rapid eye movement sleep, even when itching is eliminated. This sleep disruption is linked to overactivity of wake-promoting neurons in a brain region called the anterior hypothalamic area. We find evidence of a local inflammatory response in this region, including activation of microglia and elevated levels of the signaling protein tumor necrosis factor-alpha. Directly delivering an inhibitor of this protein into the anterior hypothalamic area significantly restores normal sleep. These findings reveal a direct pathway from skin inflammation to sleep-regulating brain circuits and identify tumor necrosis factor-alpha as a potential target for treating insomnia in psoriasis.

Clinical data and a mouse model of psoriasis reveal a direct pathway from skin inflammation to sleep-regulating brain circuits, identifying tumor necrosis factor-alpha as a potential therapeutic target for psoriasis-associated insomnia.

Introduction

Chronic inflammatory diseases, long defined by their target organ pathology, are now recognized as systemic disorders whose reach extends deep into the central nervous system, profoundly altering mood, cognition, and behavior^1–3. Psoriasis, an immune-mediated condition affecting 2–3% of the global population, serves as a compelling model of this interplay. Beyond its hallmark cutaneous lesions, psoriasis is inextricably linked to a high burden of comorbidities, among which sleep disturbances are particularly prevalent and debilitating. Affecting up to 80% of patients, these sleep disruptions significantly impair quality of life and are strongly correlated with disease severity, yet their underlying etiology remains an enigma^4–6.

Historically, the connection between psoriasis and poor sleep has been attributed to secondary, symptom-driven factors such as nocturnal pruritus, cutaneous pain, or the psychological weight of chronic illness^7–9. However, this symptom-centric view is increasingly challenged by compelling clinical observations. For instance, the rapid improvement in sleep quality seen in patients treated with biologic therapies often precedes significant cutaneous clearing, and sleep abnormalities persist even in the absence of severe itching or psoriatic arthritis^10–12. This therapeutic and symptomatic discordance strongly suggests the presence of a more fundamental mechanism—namely, one in which the core immunopathology of psoriasis directly interacts with the brain’s sleep-regulating circuits.

The pathogenesis of psoriasis is driven by a characteristic cytokine storm, with tumor necrosis factor-alpha (TNF-α) acting as a master regulator of the inflammatory cascade¹³. TNF-α and other pro-inflammatory cytokines are not confined to the periphery; they are potent neuromodulators that can traverse the blood-brain barrier (BBB) or signal via afferent neural pathways to influence CNS function^14–16. A prime candidate for this neuroimmune integration is the hypothalamus, the master regulator of homeostasis and the sleep-wake cycle. Within this region, a delicate balance between sleep-promoting GABAergic neurons (e.g., in the ventrolateral preoptic nucleus) and wake-promoting glutamatergic neurons (e.g., in the anterior hypothalamic area) dictates sleep architecture^17–19. Systemic inflammation is known to disrupt this balance, yet the specific cell types and molecular pathways through which psoriasis-associated inflammation commandeers hypothalamic circuitry to induce wakefulness have not been elucidated.

Here, we sought to bridge this critical knowledge gap. We integrated clinical data from a psoriatic cohort with a sophisticated, pruritus-controlled imiquimod-induced mouse model. By combining electrophysiological recordings, neuroanatomical mapping, and targeted pharmacological manipulation, we delineated a novel dermo-neuro-immune axis connecting skin inflammation to central sleep dysregulation. Our findings establish that TNF-α-mediated hypothalamic neuronal activation, particularly the hyperactivity of anterior hypothalamic area, central (AHC) glutamatergic neurons, is a key driver of psoriasis-associated insomnia, providing a mechanistic rationale for targeted therapies and revealing how a peripheral immune disorder can hijack central neural circuits to manifest as a systemic disease.

Results

Psoriasis is clinically associated with significant sleep disruption

To first establish the clinical relevance of sleep dysfunction in psoriasis, we conducted a case-control study involving 66 patients and 66 demographically matched healthy controls. The two groups were successfully matched, showing no significant differences in age (patients: 41.5 ± 12.8 years v.s. controls: 40.9 ± 9.6 years; P = 0.764), body mass index (BMI) (24.3 ± 3.6 v.s. 23.9 ± 3.1 kg m⁻²; P = 0.491), or sex distribution (P = 0.293) (Table 1).

Table 1.

Demographic and clinical characteristics

	Psoriasis (n = 66)	Control (n = 66)	P value
Gender (M/F)	48/18	46/20	0.293
Age (years)	41.5 ± 12.8	40.9 ± 9.6	0.764
BMI (kg m−2)	24.3 ± 3.6	23.9 ± 3.1	0.491
Duration	6.3 ± 6.1	–	–
BSA (%)	1.8 ± 0.8	–	–
Itch NRS score	1.7 ± 1.1	–	–
Local treatment (%)	21 (32%)	–	–
Systemic treatment (%)	20 (30%)	–	–
Local and systemic treatment (%)	13 (20%)	–	–

BSA body surface area, BMI body mass index, NRS numeric rating scale.

Results are presented as mean values and standard deviations.

Patients with psoriasis reported markedly poorer sleep quality, as indicated by a significantly higher global Pittsburgh Sleep Quality Index (PSQI) score compared to controls (4.9 ± 2.3 v.s. 3.1 ± 1.6; P < 0.01). This difference was clinically profound: 53.0% of psoriasis patients met the criteria for poor sleep (PSQI > 5), more than double the rate observed in the control group (21.2%; P < 0.01) (Table 2). These findings confirm a strong clinical association between psoriasis and sleep disturbances, providing a human-centric foundation for our subsequent mechanistic investigation in an animal model.

Table 2.

Sleep quality indicators

	Psoriasis (n = 66)	Control (n = 66)	P value
Sleep quality	1.1 ± 0.7	0.5 ± 0.5	<0.01
Sleep latency	0.9 ± 0.7	0.5 ± 0.6	<0.01
Sleep duration	0.7 ± 0.7	0.6 ± 0.5	0.485
Sleep efficiency	0.11 ± 0.4	0.02 ± 0.1	0.052
Sleep disturbances	0.8 ± 0.6	0.6 ± 0.5	<0.05
Use of sleeping medication	0.05 ± 0.3	0.02 ± 0.1	0.413
Daytime disfunction	1.3 ± 0.9	0.8 ± 0.7	<0.05
Global PSQI	4.9 ± 2.3	3.1 ± 1.6	<0.01
PSQI ≤ 5 n (%)	31 (47%)	52 (79%)
PSQI ≥ 6 n (%)	35 (53%)	14 (21%)

PSQI Pittsburgh Sleep Quality Index.

Results are presented as mean values and standard deviations.

IMQ-induced mice develop a robust psoriasis-like phenotype and systemic inflammation independent of pruritus control

To investigate the link between psoriatic inflammation and sleep disturbances, we first established a well-characterized imiquimod (IMQ)-induced mouse model (Fig. 1A). Daily topical application of IMQ for 5 consecutive days induced a robust psoriasis-like phenotype, characterized by pronounced erythema, silvery-white scaling, and significant skin thickening (Fig. 1B). This was quantitatively confirmed by a dramatic increase in the Psoriasis Area and Severity Index (PASI) score beginning on day 3 (Fig. 1D) and a marked epidermal hyperplasia observed in histological sections (Fig. 1B, C).

Fig. 1. IMQ-induced psoriasis-like mouse model and the effect of DH treatment.

A Schematic timeline of the experimental design. 8 to 10-week-old male C57BL/6 mice were randomly assigned into four groups and treated with IMQ, IMQ with DH, DH alone, or vehicle control (CON) for 5 consecutive days. B Representative dorsal skin images and corresponding H&E-stained sections. IMQ and IMQ + DH treatment induced typical psoriasis-like features, including erythema, scaling, skin thickening, and epidermal acanthosis. H&E staining shows marked epidermal hyperplasia, hyperkeratosis, and inflammatory cell infiltration in the dermis. Scale bar: 100 µm. C Quantification of epidermal thickness (n = 5 per group). D PASI scores of dorsal skin during the treatment period (n = 5 per group). E Serum levels of tumor necrosis factor-α (TNF-α) measured by ELISA (n = 5 per group). F Serum levels of interleukin-6 (IL-6) measured by ELISA (n = 5 per group). Data are presented as mean + SEM. Data were assessed by two-way ANOVA followed by the Bonferroni post hoc test for multiple comparisons using GraphPad Prism 8 software. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant.

To isolate the effects of inflammation from those of pruritus (itching), a parallel group of mice was co-treated with the topical anti-pruritic agent dyclonine hydrochloride (DH). Two-way ANOVA revealed a significant main effect of IMQ treatment (F(1, 16) = 177.6, P < 0.0001), whereas neither the main effect of co-treatment nor the treatment × co-treatment interaction reached statistical significance (both P > 0.05). Critically, co-treatment with DH did not ameliorate the cutaneous pathology. Epidermal thickness and PASI scores in the IMQ + DH group were indistinguishable from those in the IMQ-only group, and both were significantly elevated compared to control groups (Bonferroni post hoc test, both P < 0.05; Fig. 1C, D).

We next assessed whether the cutaneous inflammation was associated with a systemic inflammatory response. ELISA analysis of serum revealed that both IMQ and IMQ + DH mice had significantly elevated levels of the pro-inflammatory cytokines TNF-α and IL-6 compared to vehicle-treated controls (Fig. 1E, F; two-way ANOVA: main effect of treatment, F(1, 16) = 79.94 for TNF-α and F(1, 16) = 30.59 for IL-6, both P < 0.0001; no significant main effect of co-treatment or treatment × co-treatment interaction, both P > 0.05; Bonferroni post hoc test, P < 0.05 for IMQ and IMQ + DH v.s. vehicle; n = 5 for each group). Consistent with the dermatological findings, DH co-treatment had no effect on these systemic cytokine levels.

Taken together, these results validate a robust mouse model that successfully recapitulates both the cutaneous and systemic inflammatory hallmarks of psoriasis, providing a powerful platform to investigate the central effects of inflammation independent of pruritus-related sensory input.

Psoriatic inflammation drives a state of hyperarousal and disrupted non-rapid eye movement (NREM) sleep independent of pruritus

Having established a robust model of pruritus-independent inflammation, we next performed continuous electroencephalogram (EEG) and electromyogram (EMG) recordings to dissect the sleep architecture in these mice (Figs. 2A and S3A, B). First, we confirmed the efficacy of our anti-pruritic treatment. As expected, IMQ application induced significant scratching behavior, which was completely abolished by co-treatment with DH, reducing scratching bouts to control levels (Fig. 2B; two-way ANOVA: significant main effect of treatment, F(1, 16) = 26.90, P < 0.0001; significant main effect of co-treatment, F(1, 16) = 14.01, P = 0.0018; significant treatment × co-treatment interaction, F(1, 16) = 13.18, P = 0.0022; Bonferroni post hoc test, P < 0.05 for all pairwise comparisons; n = 5 for each group).

Fig. 2. Effects of psoriasis-like mice on sleep.

A Schematic illustration of the experimental design for EEG/EMG recordings. Eight to 10-week-old male C57BL/6 mice were surgically implanted with EEG and EMG electrodes and then randomly assigned to four groups. Mice were treated with IMQ, IMQ with DH, DH alone, or vehicle control for 5 consecutive days, followed by EEG/EMG recordings. B Number of scratching bouts recorded over a 2-h period. C Percentage of time spent in wakefulness, NREM sleep, and REM sleep across a 24-h light/dark cycle. Data are shown for CON and IMQ groups to illustrate inflammation-driven sleep changes independent of pruritus (see (B) for itch suppression); complete dataset with all four conditions is shown in (D, E). D Total time spent in wakefulness, NREM sleep, and REM sleep. Values represent the cumulative sum across the entire 24-h cycle (combining both light and dark phases) to provide an overall assessment of sleep-wake architecture. E Number of episodes of wakefulness, NREM sleep, and REM sleep. F Mean episode duration of wakefulness, NREM sleep, and REM sleep. Data are presented as mean + SEM (n = 5 per group). Data were assessed by two-way ANOVA followed by the Bonferroni post hoc test for multiple comparisons using GraphPad Prism 8 software. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant.

Despite the effective suppression of itch, both IMQ and IMQ + DH mice exhibited profound and nearly identical sleep disturbances. Over a 24-h period, both groups spent significantly more time in wakefulness and correspondingly less time in NREM sleep compared to controls (Fig. 2D; two-way ANOVA: main effect of treatment, F(1, 16) = 102.0, P < 0.0001; main effect of co-treatment, F(1, 16) = 10.46, P = 0.0052; no significant treatment × co-treatment interaction, P > 0.05; Bonferroni post hoc test, P < 0.05 for IMQ and IMQ + DH v.s. control; n = 5 for each group). This state of hyperarousal was sustained across the entire light–dark cycle (Fig. 2C). In contrast, the total duration of rapid eye movement (REM) sleep remained unchanged across all groups (Fig. 2D).

Analysis of sleep microstructure revealed that this phenotype was driven by a consolidation of wakefulness and a fragmentation of sleep. Psoriatic mice displayed fewer but significantly longer wake episodes, indicating an inability to transition into or maintain sleep (Fig. 2E, F; two-way ANOVA: main effect of treatment, F(1, 16) = 14.11, P = 0.0017; no significant main effect of co-treatment or treatment × co-treatment interaction, both P > 0.05; Bonferroni post hoc test, P < 0.05 for IMQ and IMQ + DH v.s. control; n = 5 for each group). Conversely, when the mice did enter NREM sleep, the bouts were shorter in duration, reflecting poor sleep quality (Fig. 2F; two-way ANOVA: main effect of treatment, F(1, 16) = 25.13, P = 0.0001; main effect of co-treatment, F(1, 16) = 7.656, P = 0.0137; no significant treatment × co-treatment interaction, P > 0.05; Bonferroni post hoc test, P < 0.05 for IMQ and IMQ + DH v.s. control; n = 5 for each group).

Crucially, there were no significant differences in any sleep parameter—including total times, episode counts, or bout durations—between the IMQ and IMQ + DH groups. This provides unequivocal evidence that the observed sleep disruption is a direct consequence of systemic inflammation rather than a secondary effect of peripheral itch sensation. These findings strongly implicate a central mechanism in mediating inflammation-induced hyperarousal.

Psoriatic inflammation selectively activates wake-promoting glutamatergic neurons in the anterior hypothalamic area

Given the profound state of hyperarousal in psoriasis-like mice, we next sought to identify the underlying neural correlates by mapping neuronal activity across the brain. We performed immunostaining for c-Fos, an immediate early gene product widely used as a marker of recent neuronal activation (Fig. 3A). A brain-wide screen revealed significantly increased c-Fos expression in several key sleep-wake regulatory nuclei in IMQ-treated mice, including the dorsomedial hypothalamus (DMH), the ventrolateral periaqueductal gray (VLPAG), and, most prominently, the AHC (Fig. 3B).

Fig. 3. The number of c-Fos-positive neurons and hyperactivity of AHC glutamatergic neurons in psoriasis-like mice.

A Schematic illustration of the experimental design for IMQ-induced psoriasis-like mice followed by brain perfusion. B Representative coronal brain sections showing c-Fos-positive neurons in the AHC, DMH, and VLPAG of CON and IMQ-treated mice. Scale bar: 500 µm. C Quantification of c-Fos-positive neurons in the AHC, DMH, and VLPAG regions (n = 4 per group). D Representative immunofluorescence images showing co-localization of c-Fos (green) with the glutamatergic neuronal marker VGLUT2 (red) and the GABAergic neuronal marker GAD67 (purple) in the AHC (Bregma −2.22 mm). Nuclei were counterstained with DAPI (blue). Scale bar: 50 µm. E Quantification of the proportion of VGLUT2⁺ and GAD67⁺ neurons among c-Fos⁺ cells in the AHC (n = 4 per group). Data are presented as mean + SEM. Statistical significance was determined using an unpaired two-tailed Student’s t test. *P < 0.05, **P < 0.01, ****P < 0.0001.

Due to its robust activation and its established role in promoting wakefulness²⁰, we focused our subsequent analysis on the AHC (Fig. 3C, D). To determine the neurochemical identity of the activated cells, we performed double-label immunofluorescence for c-Fos and specific neuronal markers. Strikingly, the vast majority of c-Fos-positive neurons in the AHC of IMQ-treated mice co-expressed the vesicular glutamate transporter 2 (VGLUT2), a definitive marker for excitatory glutamatergic neurons (Fig. 3E). In contrast, there was minimal co-localization with markers for inhibitory GABAergic neurons. Meanwhile, similar analysis performed in the cortical region revealed no statistically significant difference between the two groups (Fig. S1).

These findings provide a direct neuroanatomical basis for the observed sleep disturbances. They demonstrate that systemic inflammation in the psoriasis-like model does not cause widespread, non-specific neuronal activation, but rather selectively engages an excitatory, wake-promoting glutamatergic circuit within the hypothalamus. This points to a specific cellular target through which peripheral inflammation drives central hyperarousal.

Psoriatic inflammation triggers a TNF-α-dominated neuroinflammatory cascade and microglial activation in the anterior hypothalamic area

To elucidate the molecular mechanisms driving the hyperactivity of AHC neurons, we performed unbiased transcriptomic profiling (RNA-seq) on microdissected anterior hypothalamic area tissue from IMQ- and control-treated mice. Principal component analysis (PCA) revealed a distinct separation between the two groups, indicating a profound shift in the AHC transcriptome (Fig. 4A). We identified 798 upregulated and 388 downregulated genes in the IMQ group (Fig. 4B). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses of these differentially expressed genes showed a striking enrichment of pathways related to immune responses, including “regulation of cytokine production,” “inflammatory response,” and “cytokine-cytokine receptor interaction.” Notably, pathways specifically involving “tumor necrosis factor production” and “interleukin-6 production” were among the most significantly enriched (Fig. 4C, D).

Fig. 4. Psoriatic inflammation induces a TNF-α-dominated neuroinflammatory cascade and microglial activation in the anterior hypothalamic area.

A PCA of RNA-seq data from the anterior hypothalamic area tissue of CON and IMQ-treated mice (n = 3 per group). B Volcano plot showing DEGs between CON and IMQ-treated mice (n = 3 per group). C GO enrichment analysis showing significantly enriched biological processes among DEGs. D Dot plot showing KEGG pathway enrichment analysis of DEGs. E Quantitative RT–PCR analysis showing increased mRNA expression levels of Tnfa, Il1b, and Il6 in the anterior hypothalamic area tissue of IMQ-treated mice (n = 4 per group). F ELISA analysis showing elevated protein levels of TNF-α and IL-6 in the anterior hypothalamic area tissue of IMQ-treated mice (n = 4 per group). G Representative immunofluorescence images showing microglial activation in the AHC of IMQ-treated mice. Scale bar: 20 µm. H Quantitative analysis of microglial morphological parameters, including soma area, filament length, and number of endpoints (n = 4 per group). Data are presented as mean + SEM. Statistical significance was determined using an unpaired two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant.

To validate these transcriptomic findings, we measured cytokine levels directly within the anterior hypothalamic area. Both qRT-PCR and ELISA confirmed that the gene expression and protein levels of TNF-α and IL-6 were significantly elevated in the anterior hypothalamic area of IMQ-treated mice compared to controls (Fig. 4E, F). Given that microglia are the primary innate immune cells of the brain, we next investigated their activation state. Immunofluorescence staining for the microglial marker Iba1 revealed that microglia in the AHC of IMQ mice exhibited a classic activated morphology, characterized by enlarged soma, retracted processes, and increased branching complexity (Fig. 4G, H). Consistent alterations were also observed in the cortical region (Fig. S2E, F).

Crucially, to confirm that this neuroinflammatory signature was a direct consequence of inflammation and not secondary to pruritus, we performed a parallel RNA-seq analysis on AHC tissue from DH- and IMQ + DH-treated mice. This comparison revealed a nearly identical upregulation of inflammatory and cytokine-related pathways in the IMQ + DH group compared with the IMQ group described above (Fig. S2A–D), demonstrating that the AHC neuroinflammatory cascade occurs independently of peripheral itch signaling.

Collectively, these data establish that systemic psoriatic inflammation triggers a robust, TNF-α-centric neuroinflammatory response within the anterior hypothalamic area, characterized by microglial activation and local cytokine production. This provides a direct molecular link between the peripheral disease and the central neuronal hyperactivity driving sleep disruption.

Blockade of TNF-α signaling within the hypothalamic area is sufficient to rescue sleep architecture in psoriasis-like mice

Our findings thus far identified TNF-α-driven neuroinflammation in the anterior hypothalamic area as the likely cause of sleep disruption. To directly test this causal hypothesis, we performed a targeted pharmacological intervention. Psoriasis-like mice received a stereotaxic microinjection of Etanercept, a potent TNF-α inhibitor, or an artificial cerebrospinal fluid (Acsf) vehicle diffusely acting on the hypothalamic region, followed by 24-h sleep recordings (Figs. 5A, B and S3C).

Fig. 5. Local inhibition of TNF-α in the anterior hypothalamic area ameliorates sleep disturbances in psoriasis-like mice.

A Schematic timeline of the experimental protocol. Eight to 10-week-old male C57BL/6 mice were randomly assigned to three groups: CON+Acsf, IMQ+Acsf, and IMQ+etanercept. Mice received bilateral microinjection of Acsf or etanercept into the anterior hypothalamic area. IMQ+Acsf and IMQ+etanercept mice were treated with IMQ for 5 consecutive days, whereas CON+Acsf mice received vehicle treatment. EEG/EMG recordings were subsequently performed. B Representative sagittal brain section showing cannula placement targeting the anterior hypothalamic area. C Percentage of time spent in wakefulness, NREM sleep, and REM sleep across a 24-h light–dark cycle in CON+Acsf, IMQ+Acsf, and IMQ+etanercept groups. D Total time spent in wakefulness, NREM sleep, and REM sleep. E Number of episodes of wakefulness, NREM sleep, and REM sleep. F Mean episode duration of wakefulness, NREM sleep, and REM sleep. G Quantification of c-Fos-positive neurons in the AHC from CON+Acsf, IMQ + Acsf, and IMQ + Etanercept groups. Data are presented as mean + SEM (n = 4 per group). Data were assessed by one-way ANOVA followed by the Tukey post hoc test for multiple comparisons using GraphPad Prism 8 software. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant.

As expected, mice in the IMQ + Acsf group replicated the sleep deficits observed previously, exhibiting prolonged wakefulness and reduced NREM sleep compared to controls. Strikingly, a single bilateral administration of Etanercept into the anterior hypothalamic area dramatically reversed this phenotype. The total time spent in wakefulness was significantly reduced, while NREM sleep duration was restored to levels statistically indistinguishable from those of healthy controls (Fig. 5C, D, F  =  34.35, P  <  0.0001, n  =  4 for each group, one-way ANOVA, Tukey’s multiple comparison post hoc test). This intervention specifically targeted the wake/NREM balance, as REM sleep duration remained unaffected across all groups (Fig. 5D).

Analysis of sleep microstructure confirmed a fundamental restoration of sleep quality. Etanercept administration dismantled the consolidated wakefulness characteristic of the psoriatic state, significantly reducing the duration of wake episodes (Fig. 5E, F  =  9.345, P  =  0.0064, n  =  4 for each group, one-way ANOVA, Tukey’s multiple comparison post hoc test; Fig. 5F, F  =  29.13, P  =  0.0001, n  =  4 for each group, one-way ANOVA, Tukey’s multiple comparison post hoc test). Concurrently, it reversed the fragmentation of NREM sleep, increasing the duration of individual NREM bouts (Fig. 5F, F  =  8.138, P  =  0.0096, n  =  4 for each group, one-way ANOVA, Tukey’s multiple comparison post hoc test).

To further confirm the functional effect of TNF-α blockade in the anterior hypothalamic area, we examined c-Fos expression as a marker of neuronal activation. Our quantitative and statistical analyses consistently demonstrated that IMQ + Acsf mice exhibited a significant increase in the number of c-Fos-positive neurons in the AHC compared to CON + Acsf controls. Notably, this increase was significantly attenuated in IMQ + Etanercept-treated mice (Fig. 5G; F = 45.18, P < 0.0001, n  =  4 for each group, one-way ANOVA, Tukey’s multiple comparison post hoc test).

These results provide direct, compelling evidence that elevated TNF-α signaling specifically within the anterior hypothalamic area is a key functional driver of the sleep disturbances associated with psoriatic inflammation. The finding that a localized central blockade can normalize a peripherally induced behavioral deficit solidifies the dermo-hypothalamic axis as a critical pathway in the pathology of psoriasis-related comorbidities.

Discussion

The profound impact of psoriasis on sleep quality is a well-documented clinical reality, yet its biological basis has remained a critical knowledge gap. Epidemiological studies consistently report that 60–80% of psoriasis patients suffer from significant sleep disturbances, including difficulty initiating and maintaining sleep, which mirrors the phenotype of consolidated wakefulness and fragmented NREM sleep we observed in our animal model^6,21,22. Our study moves beyond this clinical correlation to propose a functional neurobiological connection: a dermo-hypothalamic link where systemic inflammatory signaling triggers alterations in central sleep-wake circuitry. By demonstrating that TNF-α drives hyperactivity of wake-promoting neurons in the anterior hypothalamic area, we fundamentally reframe psoriatic insomnia from a secondary symptom to a direct neurobiological sequela of the disease.

A major conceptual advance of our work is the uncoupling of inflammation from pruritus. While previous studies could not rule out itch as a primary driver, our pruritus-controlled model provides unequivocal evidence that the inflammatory cascade itself is sufficient to disrupt sleep. This is a crucial distinction, as it provides a mechanistic explanation for why systemic therapies like TNF-α inhibitors can rapidly improve patient-reported sleep outcomes, often before maximal improvement in skin lesions is achieved^10,12,23. Our findings suggest these therapies are not just reducing peripheral discomfort but are actively reversing a neuroinflammatory state within the brain.

In our study, we observed significantly elevated levels of TNF-α and IL-6 in the peripheral blood of the imiquimod-treated mice. These findings primarily confirm the successful establishment of the psoriasis model and the presence of a robust systemic inflammatory state. Although we did not directly track the transport of these specific peripheral cytokines into the brain, existing literature provides strong support for the pathway by which systemic inflammation influences the CNS. Firstly, it is well-established that systemic inflammation can impair the integrity of the BBB, thereby facilitating the transport of large molecular proteins, including cytokines, from the periphery into the CNS²⁴. Secondly, animal studies have shown that the imiquimod-induced psoriasis model exhibits increased pro-inflammatory cytokine expression in the brain, supporting the conclusion that peripheral inflammatory signals can propagate to the CNS²⁵. Therefore, it is plausible that the systemic inflammatory milieu in our model facilitates the central upregulation of TNF-α we observed in the AHC.

Our results also contribute to a more nuanced understanding of how inflammation regulates sleep. Systemic inflammation is classically associated with “sickness behavior”, a state characterized by lethargy and increased NREM sleep, which is thought to be an adaptive, energy-conserving response mediated by somnogenic cytokines like IL-1β and TNF-α acting on sleep-promoting nuclei like the ventrolateral preoptic area^16,26,27. Our findings present a compelling contrast. In the context of psoriasis-like chronic inflammation, the net effect is not somnolence but sustained hyperarousal. We propose that this effect results from the specific targeting of the anterior hypothalamic area, a central hub for wakefulness promotion. The selective activation of AHC glutamatergic neurons appears to override or bypass the canonical sleep-promoting sickness pathways, resulting in a distinct insomnia phenotype. This highlights that the ultimate behavioral output of neuroinflammation is highly dependent on the specific circuits and neuronal populations engaged.

The central role of TNF-α as the molecular driver of this axis aligns with its established, albeit complex, function in sleep modulation. While acute, low-dose administration of TNF-α can be sleep-promoting²⁸, chronic elevation in pathological states like rheumatoid arthritis and multiple sclerosis is strongly linked to sleep disruption^{29, 30}. Our study provides critical spatial and cellular resolution to this paradox. We demonstrate that in the anterior hypothalamic area, TNF-α orchestrates a local neuroinflammatory loop involving microglial activation and further cytokine production, creating an environment that favors sustained neuronal excitability. The complete rescue of sleep architecture by a localized TNF-α blockade in the hypothalamic area is the definitive evidence of its causal role in this specific circuit. It should be emphasized that direct administration of etanercept into the anterior hypothalamus area for rescue experiments would be ideal. However, the tissue damage caused by cannula implantation directly into the anterior hypothalamus would obviously have adverse effects on sleep monitoring. Therefore, we adopted the cannula strategy of ‘dorsal-placement and drug diffusion’ above the hypothalamus.

Looking forward, these findings open several avenues for clinical translation, from refining current therapeutic strategies to developing novel diagnostics and targeted treatments. Our work provides a clear neurobiological rationale for prioritizing TNF-α inhibitors in psoriasis patients who report significant fatigue and sleep loss, validating sleep improvement as a direct therapeutic endpoint independent of cutaneous response. Furthermore, the identification of the AHC as a neuroinflammatory hotspot suggests that advanced neuroimaging techniques, such as PET scans using ligands for activated microglia, could emerge as powerful biomarkers to quantify CNS involvement in psoriasis and other inflammatory conditions. Ultimately, our study paves the way for more sophisticated, non-immunosuppressive therapies. Future drug discovery could focus on developing brain-penetrant small molecules that selectively inhibit downstream TNF-α signaling pathways within hypothalamic circuits, offering a novel strategy to normalize sleep without the risks associated with systemic immune blockade.

While our study is robust, some limitations warrant consideration. First, the imiquimod model, though powerful, recapitulates the acute inflammatory phase of psoriasis rather than the chronic, relapsing-remitting nature of the human disease. Future studies using chronic genetic models are needed to explore the long-term plasticity of these hypothalamic circuits. Second, while we identified TNF-α as a key mediator, the precise pathways through which it signals from the periphery to the AHC—whether via transport across the BBB, signaling through the circumventricular organs, or activation of vagal afferents—remain a key area for future investigation. Finally, our work focused on the AHC, but this nucleus is part of a larger network, and future research should explore how its hyperactivity alters communication with other sleep-wake centers to orchestrate the global state of arousal.

In conclusion, our research bridges the disciplines of dermatology, immunology, and neuroscience to solve a long-standing clinical puzzle. We establish a causal link from psoriatic skin inflammation to hypothalamic neuronal hyperactivity and sleep disruption, mediated by a TNF-α-driven neuroinflammatory cascade. This dermo-hypothalamic axis not only provides a new framework for understanding the systemic nature of psoriasis but also highlights central neural circuits as a viable therapeutic target for treating the debilitating comorbidities of this and potentially other chronic inflammatory diseases.

Methods

Human participants and clinical assessment

This case-control study enrolled 66 patients with psoriasis and 66 age- and sex-matched healthy controls at the First Affiliated Hospital with Nanjing Medical University. The study was conducted in accordance with the Declaration of Helsinki and was approved by the local ethics committee (Approval No. 2025-SR-297). A statement confirming that informed consent was obtained. Exclusion criteria were: (1) major systemic comorbidities, including malignancies or renal disease; (2) current or past psychiatric disorders, such as depression or anxiety; and (3) use of neuroactive medications, including antidepressants, psychostimulants, antiepileptics, or chronic antihistamines. Sleep quality was quantitatively evaluated using the Pittsburgh Sleep Quality Index (PSQI), with a global score >5 defined as poor sleep. All ethical regulations relevant to human research participants were followed.

Animals and housing

Male C57BL/6 mice (8–10 weeks old; Vital River Laboratory Animal Technology Co., Beijing, China) were used for all experiments. Animals were housed in a specific-pathogen-free facility under a 12:12 h light/dark cycle (lights on at 07:00) at a controlled ambient temperature (22 ± 1°C) and humidity (50 ± 10%). Autoclaved chow and filtered water were available ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC-2303012) and complied with the Guide for the Care and Use of Laboratory Animals. We have complied with all relevant ethical regulations for animal use.

Imiquimod-induced psoriasis-like mouse model

A psoriasis-like phenotype was induced by the daily topical application of 62.5 mg of 5% imiquimod (IMQ) cream (Aldara; 3 M Health Care Ltd., UK) to the shaved dorsal skin for 5 consecutive days. Disease progression was monitored daily by photographing the skin lesions and calculating the PASI score, which was based on the severity of erythema, scaling, and skin thickness. Erythema and scaling were visually scored on days 1, 2, 3, 4, 5, and 6 on a scale from 0 to 4: 0 for asymptomatic, 1 for minor, 2 for moderate, 3 for nasty, and 4 for extremely nasty manifestations. On day 6, mice were humanely euthanized, and dorsal skin samples were collected for hematoxylin and eosin (H&E) staining. Epidermal thickness was subsequently quantified using ImageJ software. The cumulative PASI score (ranging from 0 to 12) was derived by summing the scores for erythema, scaling, and epidermal thickness, serving as an indicator of inflammation severity.

Histological analysis

Dorsal skin samples were fixed in 10% neutral-buffered formalin for 24 h, processed through a graded ethanol series, and embedded in paraffin. Sections (5 μm thick) were cut and stained with H&E (Sigma-Aldrich, St. Louis, MO, USA) for standard histopathological evaluation. Images were captured using a bright-field microscope (Olympus, Tokyo, Japan).

Surgical implantation for EEG/EMG recording

Mice were anesthetized with isoflurane (2–3% for induction, 1–1.5% for maintenance) and secured in a stereotaxic frame (RWD Life Science, Shenzhen, China). For EEG, two stainless-steel screw electrodes were implanted into the skull over the frontal cortex. For EMG, two stainless-steel wires were inserted into the nuchal muscles. The electrode assembly was fixed to the skull with dental cement. Mice were housed individually and allowed a post-operative recovery period of at least 7 days before recordings began.

EEG/EMG recording and sleep analysis

Following recovery from surgery, continuous 24-h EEG/EMG signals were recorded using a Medusa hardware/software system (Bio-Signal Technologies, McKinney, TX, USA) at a sampling rate of 500 Hz. Recordings were performed in a sound-attenuated, electromagnetically shielded chamber under a standard 12:12 h light/dark cycle. Sirenia Sleep Pro software was used for the initial automated analysis of mouse sleep states, adhering to the following criteria: Wakefulness: Characterized by desynchronized, low-amplitude EEG rhythms and elevated EMG activity often accompanied by phasic bursts. NREM sleep: Defined by synchronized, high-amplitude, and low-frequency (1–4 Hz, delta) EEG activity, coupled with lower EMG activity compared to wakefulness. REM sleep: Identified by consecutive theta (6–9 Hz) oscillations in the EEG and further decreased EMG activity relative to the preceding NREM sleep stage. Hyperarousal was operationally defined as a statistically significant increase in total wake time and a corresponding decrease in NREM sleep duration compared to control conditions.

To complement the EEG/EMG recordings, simultaneous video monitoring of the mice’s activity states was acquired. These video recordings provided critical behavioral correlates—such as periods of sustained rest, distinct movements, and observed eye movements—which significantly assisted in the manual verification and refinement of the automated sleep stage scoring. For example, the direct observation of rapid eye movements was indicative of REM sleep, while sustained behavioral quiescence paired with muscle relaxation further supported the identification of NREM sleep episodes. This comprehensive approach ensured robust and accurate classification of sleep stages.

Stereotaxic microinjection

Mice were anesthetized and placed in a stereotaxic frame as described above, with body temperature maintained at 37 °C using a heating pad. Bilateral stainless-steel guide cannulae (26-gauge, RWD Life Science) were implanted 2.0 mm above the anterior hypothalamic area using the following coordinates from bregma: anteroposterior (AP) −0.7 mm, mediolateral (ML) ± 0.3 mm, and dorsoventral (DV) −4.0 mm. After a 3-day recovery period, bilateral microinjections were performed. Etanercept (Sansheng Guojian Pharmaceutical, Shanghai, China) was dissolved in Acsf. A volume of 0.5 μL per side containing Etanercept (at a specified dose, e.g., 50 ng) or Acsf vehicle was infused into the AHC over 2 min using 33-gauge injection cannulae connected to Hamilton syringes. The injection cannulae were left in place for an additional 5 min to allow for diffusion. After behavioral testing, animals were perfused and brain sections were collected for histological verification of cannula placement. Only animals in which the cannula tip was confirmed to be located adjacent to the hypothalamic region were included in the final analysis.

Immunofluorescence and immunohistochemistry staining

Mice were deeply anesthetized with isoflurane and transcardially perfused with ice-cold phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were post-fixed overnight in 4% PFA and then cryoprotected in 30% sucrose. Coronal sections (35 μm thick) were prepared using a freezing microtome (Leica Microsystems, Wetzlar, Germany). Free-floating sections were permeabilized with 0.1% Triton X-100 in PBS and blocked with 10% normal bovine serum. Sections were incubated overnight at 4 °C with the following primary antibodies: rabbit anti-c-Fos (1:1000; #2250, Cell Signaling Technology, USA), rabbit anti-GAD67 (1:500; Abcam, #ab213697, China), and rabbit anti-VGLUT2 (1:500; #135403, Synaptic Systems, Germany). After washing, sections were incubated with corresponding Alexa Fluor-conjugated secondary antibodies (1:1000; Thermo Fisher Scientific, USA) for 1 h at room temperature. Nuclei were counterstained with DAPI. Images were acquired using a Zeiss LSM 800 confocal microscope.

Image processing and quantification of c-Fos-positive cells

Brain sections were carefully mounted onto slides and subjected to fluorescence microscopic analysis. Initially, we conducted a comprehensive whole brain-wide immunofluorescence analysis, which revealed c-Fos-positive neurons in several regions, including the hypothalamus, ventrolateral periaqueductal gray, and cortex. Given their established relevance to the sleep-wake cycle and our specific research focus, we subsequently concentrated our detailed investigation on the hypothalamic area.

Microscopic analysis was performed using a Nikon Eclipse Ni-e microscope equipped with a camera, capturing microphotographs at 20x magnification, which allowed for clear visualization of the hypothalamic area. Within the hypothalamus, sections were further delineated into specific subregions, primarily the DMH and the anterior hypothalamic area. For quantification, the captured microphotographs were processed using the freely available ImageJ software. This software facilitated image processing and enhancement, enabling clearer visualization of the immunoreactive cells. Specifically, c-Fos-positive cells were manually counted using ImageJ’s “multi-point” tool, identifying robustly stained, spherical nuclei as positive signals within each defined microscopic field.

Enzyme-linked immunosorbent assay (ELISA)

Following euthanasia, blood was collected via cardiac puncture and centrifuged at 2500×g for 15 min at 4 °C to obtain serum. The AHC was rapidly dissected from the brain, weighed, and homogenized in ice-cold 1× PBS containing protease inhibitors. Homogenates were centrifuged at 500×g for 5 min at 4 °C, and the supernatant was collected. Concentrations of IL-6, TNF-α, and IL-1β in serum and AHC lysates were quantified using commercial ELISA kits (MultiSciences, Hangzhou, China; Cat. EK206, EK282, EK201B, respectively) according to the manufacturer’s instructions.

RNA sequencing

Bulk RNA sequencing was performed with a total of twelve samples from 8 to 10 weeks old Male IMQ and CON, IMQ + DH and DH mice with each sample containing the anterior hypothalamic area tissue pooled from three mice. To isolate anterior hypothalamic area tissue, brains were dissected into an RNase-free slush of PBS and cut into 500 μm-thick sections on a vibratome. The anterior hypothalamic area from both hemispheres was then identified and dissected free. AHC tissue from three mice was pooled into each sample and weighed before being flash frozen on dry ice. Total RNA was isolated from the anterior hypothalamic area tissue using the TRIzol reagent. Approximately 1 µg of RNA per sample was sent on dry ice to Genedenovo Biotechnology Co., Ltd (Guangzhou, China) for RNA sequencing with their Poly(A) RNA sequencing service. The extracted mRNA is enriched using mRNA Capture Beads. After purification with beads, the mRNA is fragmented using high temperatures. The fragmented mRNA is then used as a template to synthesize the first strand of cDNA in a reverse transcription enzyme mixture system. While synthesizing the second strand of cDNA, end repair and A-tailing are completed. Next, adapters are ligated, and Hieff NGS® DNA Selection Beads are used for purification to select target fragments. PCR library amplification is then performed, and finally, detection is carried out using the Illumina Novaseq X Plus. For RNA-seq data analysis, high-quality reads were mapped to the reference genome using the alignment tool HISAT2 (v.2.4). For each transcription region, an FPKM (fragment per kilobase of transcript per million mapped reads) value was calculated to quantify its expression abundance using StringTie software (v.10).

Microglial morphology analysis

To analyze microglial morphology at high resolution, representative Iba1-positive microglia from the anterior hypothalamic area region of CON and IMQ mice were imaged using a Zeiss LSM 800 confocal microscope. For each animal, one microglial cell fully contained within the z-stack was selected per quadrant. A maximum intensity projection was generated to create a two-dimensional representation that enhanced the visibility of fine processes while preserving overall morphological integrity. After conversion to an 8-bit black and white image, edges were enhanced using an unsharp mask filter set to a 0.5 pixel sigma radius and a 0.6 mask weight. Impulse noises were removed using the Despeckle function before and after global thresholding. The Close command was run to bridge gaps in adjacent microglial processes that had perhaps eroded or provided weak signal, then dark noise was smoothed by running the Remove Outliers command with a 7 pixel radius and a deviation threshold of 50 against the surrounding median. Measurements of this filament structure were automatically generated after application of smooth processes function to the structure and centre processes function for alignment of the cell body. The data acquired included total the volume of cell body per cell (µm³), the number of branches per cell, and the larger length of branches per cell.

Statistics and reproducibility

All statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, La Jolla, CA, USA). The Shapiro–Wilk test was used to check the normality of variable distribution before analysis. The t test or one-way ANOVA was applied to normally distributed variables. Non-normally distributed variables were analyzed using the Kruskal–Wallis test. Post hoc comparisons were performed using either Tukey’s or Dunnett’s test. Two-way ANOVA was used followed by Bonferroni’s multiple-comparison post hoc tests. Data are presented as mean + standard error of the mean (SEM). A P value < 0.05 was considered statistically significant. Sample sizes and number of replicates of each experiment were stated in the corresponding figure legends.

Reporting summary

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

Author contributions

Dandan Chen: conceptualization, data curation, formal analysis, investigation, visualization, writing—original draft and visualization. Ying Ma: conceptualization, data curation, investigation, validation, visualization, writing—review and editing and funding acquisition. Jiliang Lu: conceptualization, data curation, investigation, and visualization. Xiaoya Xi: investigation, resources, and visualization. Xinyi Dai: investigation and resources. Chenxingyue Zhang: investigation. Ziyue Diao: investigation. Xing Guo: project administration, visualization, methodology, and formal analysis. Zhiqiang Yin: conceptualization, project administration, supervision, writing—review and editing, and funding acquisition.

Peer review

Peer review information

Communications Biology thanks Ayaka Nakai, Hiromasa Funato and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Joao Valente. A peer review file is available.

Funding

Zhiqiang Yin and Ying Ma disclose support for the research of this work from the National Natural Science Foundation of China [No. 82573974 and No. 82201302] and the National Key Research and Development Program of China [No. 2023YFC2509800]. Zhiqiang Yin discloses support for publication of this work from the National Natural Science Foundation of China [No. 82573974]. Dandan Chen, Jiliang Lu, Xiaoya Xi, Xinyi Dai, Chenxingyue Zhang, Ziyue Diao and Xing Guo declare no relevant funding.

Data availability

The source data behind the graphs in the paper are in the Supplementary Data 1. Raw sequencing data have been submitted to the NCBI Sequence Read Archive with accession codes PRJNA1467934 and PRJNA1468657. Processed sequencing data are in the Supplementary Data 2.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10527-y.