DSIP: the delta sleep peptide, fifty years on

A peptide found in the blood of sleeping rabbits

DSIP (delta sleep-inducing peptide, emideltide) was isolated by the Monnier and Schoenenberger group in Basel from rabbit cerebral venous blood². According to the FDA briefing document, it was obtained from the blood of rabbits in response to electrical stimulation of the thalamus³. In 1977 they published its sequence: Trp-Ala-Gly-Gly-Asp-Ala-Ser-Gly-Glu, nine amino acids¹.

The characterization experiment was careful for its time: the investigators synthesized the peptide, five possible metabolic products, two analogs with exchanged amino acids and a related tripeptide, and infused them into the cerebral ventricles of rabbits under double-blind conditions. A total of 58 rabbits, including controls, were evaluated, and only synthetic DSIP significantly and specifically enhanced delta and spindle patterns on the electroencephalogram¹. This article goes deeper into what was briefly summarized in Semax, Selank and DSIP.

The mystery: no gene, no precursor, no receptor

Almost five decades later, DSIP is still an anomaly. A 2006 review notes that the link with sleep was never characterized, partly because the gene, the precursor protein and a possible receptor were never isolated; it describes the hypothesis of DSIP as a sleep factor as "extremely poorly documented and still weak", and adds that its structure resembles no known peptide family². The same authors observed slow-wave sleep promoting activity with certain synthetic analogs "but not DSIP itself", and propose that an endogenous DSIP-like peptide is responsible for DSIP-like immunoreactivity².

That immunoreactivity has caused confusion. While searching for the protein, one group isolated from porcine brain a DSIP-immunoreactive leucine zipper protein (DIP), whose human homolog resembles transcriptional regulators of the TSC-22 family, not the nonapeptide⁴. And in healthy humans, circulating DSIP follows a daily rhythm with its peak at 15:00 and its trough at 01:00, correlated with body temperature (r² = 0.66), and is lower during REM and slow-wave sleep than during wakefulness⁵: the opposite of a simple "sleep hormone" pattern.

Sleep in humans: from open studies to controlled trials

The first clinical data were encouraging. Schneider-Helmert summarized two insomnia studies in which single injections of 25 nmol/kg before sleep improved sleep, with normalization of sleep structure after four administrations⁶. In an open study, 7 patients with severe insomnia received series of 10 injections, and in all but one sleep normalized over 3 to 7 months of follow-up⁷. According to the FDA, the first-in-human randomized, double-blind crossover study in 6 healthy volunteers described a 59% increase in median sleep time in the 130 minutes after the infusion, without loss of alertness³. In narcolepsy there is a single case: a 35-year-old man in whom repeated injections reduced sleep attacks and increased daytime alertness⁸.

The controlled trials were more modest. In a double-blind crossover study with 25 nmol/kg over four nights, the differences versus placebo were already present at baseline, and the authors concluded that the improvement was "of little clinical significance"⁹. In another double-blind trial of 16 patients with chronic insomnia, DSIP was associated with higher sleep efficiency and shorter sleep latency, but the effects were weak and partly attributable to an incidental change in the placebo group¹⁰. The FDA summarized the body of evidence as "inconclusive and at best preliminary"³.

Alcohol and opiate withdrawal

The hypothesis came from animal studies in which the effect of DSIP on sleep was reversed by naloxone¹¹. In Geneva, 107 inpatients with alcohol (n = 47) or opiate (n = 60) withdrawal received intravenous DSIP. On clinical assessment by physicians and nursing staff, signs and symptoms disappeared or improved markedly and rapidly in 97% of opiate-dependent patients and 87% of alcoholics evaluated; anxiety was slower to subside, and tolerance was good apart from headache in a few¹¹. In a later open study of 7 opioid-dependent patients, the mean withdrawal score (SOWS) fell from 34.4 to 17.8 after the first injection, although only two completed the planned schedule³.

The FDA's reading is that none of these studies had a control group or blinding³. On safety, it counted doses of 25 to 150 nmol/kg given intravenously to 209 subjects for 1 to 15 days, with no reports in its FAERS system; in the Geneva study, nine patients had mild transient effects and three had serious effects (two with hypotension at the start of the first injection)³.

Stress, the pituitary axis and mitochondria

In a randomized, double-blind crossover trial in 11 healthy volunteers, a single intravenous dose of 25 nmol/kg reduced plasma ACTH for at least 3 hours compared with saline, with no change in cortisol¹². However, another group observed no effect of DSIP on the ACTH and cortisol response to CRH (5 men per condition) or on the post-meal ACTH and cortisol surge (10 men)¹³. In the laboratory, pretreating rats with DSIP completely prevented the hypoxia-induced reduction in brain mitochondrial respiration¹⁴. The FDA also lists rodent effects on temperature, the stress response, oxidative stress, seizures and pain, and describes a partly opioid mechanism: DSIP does not bind opioid receptors but facilitates enkephalin release³.

Half-life, brain entry and analogs

DSIP clears quickly from plasma: in 4 dogs the half-life was 4.0 ± 0.7 minutes, and 2.9 and 2.0 minutes in a monkey and in rats¹⁵. The FDA cites a half-life of 5 to 10 minutes in human serum in vitro, through hydrolysis starting at the N-terminus³. Even so, a radiolabeled analog crosses the rat blood-brain barrier and the dog blood-CSF barrier by a non-competitive mechanism¹⁶. Analogs studied by investigators include DSIP-12, with beta-alanine at position 2, which was more effective than DSIP in a rat seizure model¹⁷, and Deltaran, a product of DSIP with glycine³¹⁴. The product's research sheet is at DSIP.

Where the evidence ends

• The clinical studies date from the 1980s and 1990s, with samples of 1 to 107 subjects, almost always intravenous³⁷⁸¹¹.
• The two placebo-controlled insomnia trials did not confirm a clear clinical benefit⁹¹⁰.
• The withdrawal studies had no control group or blinding³.
• There are no subcutaneous pharmacokinetic studies, and the FDA warned that its action on the opioid system "could lead to development of addiction"³.

In July 2026, the FDA advisory committee voted 6 for, 7 against and one abstention against adding emideltide to the list of substances for compounding¹⁸. One witness summed up the problem: "the last study on this was 30 years old"¹⁸.