Two drugs can fix the same broken mood and leave completely different nights behind. In mice carrying a depression model braided with insomnia, the insomnia medication lemborexant and the classic antidepressant fluoxetine both lifted depressive and anxious behavior and both quieted the hyperactive stress axis. But under the EEG, the two could hardly have acted more differently: lemborexant pieced fragmented sleep back together, while fluoxetine scattered it further. The gap between them, reported in Neuropharmacology, is a direct challenge to the way depression with sleep problems is treated.
The problem fluoxetine leaves unsolved
Depression and insomnia travel together so often that the pairing has its own clinical shorthand, and the drugs prescribed for the mood side have a well-documented blind spot. SSRIs like fluoxetine relieve the affective symptoms, but they can leave sleep disturbances intact, and in many patients they make sleep architecture worse. The mismatch matters: persistent insomnia during antidepressant treatment is one of the strongest predictors of relapse.
The premise of the new study is that a drug already used for insomnia might cover both bases. Lemborexant is a dual orexin receptor antagonist, or DORA, one of the newer generation of sleep aids that works by dampening the orexin wakefulness signal rather than by sedating the brain wholesale. Orexin neurons are also wired into stress and emotion circuits, which makes the system a plausible handle on the depression side of the comorbidity.
A model built for the overlap
To test that idea, the researchers built a model that tries to recreate the clinical overlap rather than either condition alone. Male ICR mice were run through eight weeks of chronic unpredictable mild stress, the standard behavioral recipe for a depression-like state, and at the second week they received para-chlorophenylalanine (PCPA), a serotonin synthesis inhibitor that strips the brain of the transmitter and produces an insomnia-like phenotype. Starting at week three, groups of mice were given daily oral lemborexant, fluoxetine, or vehicle for six weeks, with behavioral tests and EEG/EMG sleep recordings at the end.
The stressed, vehicle-treated mice looked the part: sluggish grooming with a deteriorating coat, reduced weight gain, longer spells of immobility when suspended by the tail, and an HPA axis stuck in overdrive, where a dexamethasone challenge failed to suppress corticosterone and Fos activation in the hypothalamic paraventricular nucleus. Both drugs reversed most of this. Lemborexant restored body weight and coat condition, fluoxetine improved the coat but not the weight, and both normalized the dexamethasone suppression response at the hormonal and neural level.
The divergence shows up at night
The sleep recordings are where the two treatments part company. The stressed mice slept badly in the light phase: more time awake, less NREM and REM, with long wake bouts and short, shattered sleep episodes. Chronic lemborexant fixed the continuity problem, significantly lengthening NREM and REM sleep bouts and restoring REM sleep toward control levels. Fluoxetine did not. It increased total sleep only modestly and, more tellingly, raised the number of wake, NREM, and REM episodes and the transitions between them, pushing the animals toward exactly the instability that fragmented sleep is defined by.
The acute picture in naive mice was starker still. A single dose of fluoxetine nearly abolished REM sleep, a signature serotonergic effect known from decades of clinical and preclinical work. Lemborexant promoted sleep on first administration. The contrast is pharmacodynamic, not incidental: SSRIs suppress REM as an intrinsic property of boosting serotonin, while orexin antagonism favors natural sleep architecture.
REM continuity tracks recovery
The most suggestive finding is correlational. Across the treated mice, the quality of REM sleep tracked the behavioral recovery: longer REM duration and longer mean REM bouts were associated with less depressive and anxious behavior, and REM latency correlated with affective measures more strongly than NREM latency did. The authors read this through a long-standing theory in mood research, the REM hypothesis, which treats REM dysregulation as a core feature of the disorder rather than a side effect of it. By that reading, lemborexant’s antidepressant-like effect may flow in part through its ability to consolidate REM, and the strong REM-behavior correlation is the fingerprint of that mechanism.
The results also argue against a strictly hierarchical view in which sleep problems are merely symptoms that resolve when mood improves. Here mood improved under both drugs while sleep diverged, which suggests the two domains run on partially overlapping but separable circuits, and that targeting the orexin system can engage both at once: top-down on emotion-related regions like the amygdala, bottom-up on the sleep-wake switch.
Why it matters
The translational hook is direct. Lemborexant is already approved for insomnia in humans, so its antidepressant potential could be tested far more quickly than a new chemical entity. The authors also point out that seltorexant, which selectively blocks the orexin 2 receptor, has demonstrated mood-lifting effects in people with major depressive disorder, and that the paraventricular nucleus, the stress axis’s control room, is dense with that receptor subtype. A dual antagonist like lemborexant may normalize HPA activity both by stabilizing sleep and by acting on orexin signaling directly.
For clinical practice, the study sharpens a known dilemma. First-line antidepressants can leave the sleep half of the comorbidity untreated or worse, and the sleep disturbances that remain predict worse outcomes. A DORA that fixes sleep continuity while mood recovers addresses that gap from the opposite direction.
Limits
The study is preclinical and male-only, so how males and females might differ in their response to DORAs and SSRIs remains untested. The PCPA-augmented model depletes serotonin, which complicates interpretation of fluoxetine’s effects by design. Drugs were dosed at the start of the inactive phase, a standard schedule for sleep-promoting agents but one that may not capture the full ecological profile of an SSRI given in the morning. The precise molecular route from orexin antagonism to mood improvement, including downstream signaling such as BDNF and neuroinflammation, is not yet pinned down, and a DORA plus SSRI combination has not been tested.
Bottom line
The same mood benefit, opposite sleep signatures: fluoxetine improved affect while fragmenting sleep, lemborexant improved affect while consolidating it, and REM continuity correlated with the recovery. If the finding holds in humans, the practical question for prescribers is not whether to treat sleep in depressed patients, but whether the current first-line drugs are the right way to do it.
Source
Lai N, Wang J, Lou J, Miao L, Zhou R, Fan X, Wang Y, Zhang X. Lemborexant produces antidepressant- and anxiolytic-like effects with sleep-wake regulation compared with fluoxetine in male mice with comorbid depression and insomnia. Neuropharmacology. 2026 Jul 30:111127. doi:10.1016/j.neuropharm.2026.111127. PMID 42532180.