In the current study, using the latest and largest GWAS of insomnia with an integrated analytical pipeline (PWAS, TWAS, MR, and genetic colocalization), we identified four proteins in the brain that had supporting evidence of causality for insomnia. Specifically, the increased protein level of ADO, CAMLG, and ICA1L in the brain was causally associated with a decreased risk for insomnia, while the increased protein level of LXN in the brain was causally associated with an increased risk for insomnia. Moreover, we found that genetically predicted insomnia was causally associated with the risk of developing cardiovascular diseases and depression.

Although the etiology of insomnia is complex, genetics play important roles in the pathogenesis of insomnia [31]. Several GWASs have been performed to explore risk loci for insomnia [5,6,7,8,9], and a study also performed TWAS to explore the transcriptionally regulated genes associated with insomnia [9]. However, to the best of our knowledge, no studies have explored the causal proteins for insomnia.

In the current study, we found that the increased abundance of ADO, CAMLG, and ICA1L in the brain was causally associated with a lower risk of developing insomnia. ADO (2-aminoethanethiol dioxygenase) is a thiol dioxygenase that convert cysteamine to hypotaurine [32]. On the one hand, hypotaurine is an endogenous neurotransmitter that can act on glycine receptors and γ-aminobutyric acid receptor A (GABAAR) [33], both of which are inhibitory receptors and favor sleep [34]. On the other hand, hypotaurine is an important intermediate for taurine, which regulates the production of GABA and melatonin and has been found to increase sleep [35]. In general, these findings could explain that the increased abundance of ADO in the brain was protective against insomnia, and enhancement of the activity of ADO might be a therapeutic target for insomnia.

Calcium-modulating ligand, also known as calcium-modulating cyclophilin ligand, was also found to be protective against insomnia. CAMLG is a cyclophilin B-binding protein, which was found to function in the trafficking of postsynaptic GABAARs to the cell membrane [36]. As mentioned above, GABAergic synapses confer an inhibition effect in the CNS and have sleep-enhancing effects [33]. Therefore, it can be inferred that the upregulated level of CAMLG in the brain could lead to an increased expression of postsynaptic GABAARs, which in turn led to an enhanced GABAergic synaptic function, resulting in better sleep and a lower risk of developing insomnia. However, more functional studies were needed.

ICA1L was also found to be causally associated with the risk of insomnia. ICA1L is enriched in the testes and the brain and functions in sheath formation [37], while its role in the central nervous system (CNS) remains unclear. Moreover, the increased abundance of ICA1L was causally associated with a decreased risk for AD [16, 38] and small vessel disease (SVD) [39, 40]. Together with our results, these findings suggested that ICA1L is crucial in the CNS, while further functional studies were needed.

However, in the MR analysis, we found that CAMLG and ICA1L had inconsistent effect directions (beta) on insomnia between mRNA and protein level, with the increased protein level of CAMLG and ICA1L causally associated with a lower risk of insomnia, while the increased mRNA level of CAMLG and ICA1L causally associated with a higher risk of insomnia (Table 2 and Fig. 2B). These opposite effects between mRNA and protein levels could be attributed to some post-transcriptional procedures such as mRNA splicing and protein degradation [40]. Functional studies were warranted to explore their roles in insomnia.

On the contrary, the increased abundance of LXN (latexin) in the brain was found to be causally associated with a higher risk of insomnia. Latexin, a carboxypeptidase A inhibitor, is a marker of neuronal development [41]. It was also found to be involved in pain sensation. Therefore, it can be inferred that increased latexin could lead to increased pain sensitivity, which in turn leads to insomnia [42]. However, functional studies were needed to ascertain the mechanism.

Insomnia is prevalent in several diseases, while the potential effect of insomnia on these diseases remains uncertain. We found that genetically determined insomnia was leading to a higher risk of developing CAD, HF, and depression, which is consistent with the results using a smaller sample size of insomnia [43,44,45]. Therefore, the treatment of insomnia might be a target for preventing CAD, HF, and depression. However, although insomnia is common in neurodegenerative diseases, we failed to identify the causal effect of insomnia on AD and PD, as well as AD and PD on insomnia, which was also consistent with previous studies [46, 47]. These results indicated that genetic factors may be insufficient to explain the high comorbidity between insomnia and neurodegenerative diseases. More studies are needed to clarify the mechanisms.

Our study had some limitations. First of all, the SNP-based heritability of the brain proteome and transcriptome reference panels was modest because of the relatively small sample size [16]. A larger sample size was needed to generate the proteome and transcriptome reference panels that have larger statistical power. Besides, the mechanism of those candidate genes in insomnia remains unclear, further functional studies were needed to clarify the proteins’ role in the pathogenesis mechanism of insomnia, which could help in the development of protein- or pathway-targeted novel therapies. Moreover, sample overlapping, adjusted parameters and patients’ comorbidities were unavailable for the original GWASs, which might lead to bias for the MR analysis [48]. Future studies with independent samples were needed. Last but not least, the current study was mainly based on European datasets, and confirmatory studies from other ethnicities are needed.