How useful is optogenetic activation in determining neuronal function within dynamic circuits?

Van Dort et al. (1) used optogenetic activation to demonstrate the capacity of cholinergic neurons in the laterodorsal and pedunculopontine tegmental nuclei (LDT and PPT, respectively) to induce rapid eye movement (REM) sleep. The authors submit that their data clarify the role of LDT/PPT cholinergic neurons in REM sleep initiation, demonstrating that these neurons are “important modulators of REM sleep” (1). However, I would argue that activating cholinergic LDT/PPT neurons does not necessarily support or oppose any measure of endogenous cholinergic involvement in the mechanisms of REM sleep generation.

Before the study by Van Dort et al. (1), the capacity of cholinergic brainstem neurons to modulate REM sleep was demonstrated previously by Huitron-Resendiz et al. (2), who selectively activated PPT cholinergic neurons with Urotensin II, causing REM sleep enhancement. Grace et al. (3) additionally found that the REM sleep-enhancing effects of Urotensin II at the PPT were blocked by simultaneous antagonism of cholinergic receptors in the pontine reticular formation (PRF). The capacity of cholinergic PRF afferents to modulate REM sleep has been established by numerous studies showing the REM sleep-inducing effects of stimulating the PRF with cholinomimetic drugs (see ref. 4 for a summary table). Gain-of-function studies, such as these, determine whether or not neuronal groups have a capacity to fulfill a given function. However, capacity is not necessarily indicative of endogenous function (e.g., see ref. 5).

The function of LDT/PPT cholinergic neurons in REM sleep generation depends on the initial state of the REM sleep control network before LDT/PPT cholinergic input. The importance of initial conditions is made evident by the finding of Van Dort et al. that the capacity of LDT/PPT neurons to induce REM sleep is present only in non-REM sleep (1). The effects of optogenetically activating LDT/PPT neurons only approximate the function of endogenous LDT/PPT activity in the case that LDT/PPT output is a source of inductive drive to the REM sleep control network. However, it is equally possible that LDT/PPT output is noninductive: that is, the activation of REM sleep generating circuits in the PRF may precede the activation of their cholinergic afferents. In this case, the activation of LDT/PPT cholinergic neurons could still have a major capacity to induce REM sleep despite endogenous LDT/PPT activity being insignificant in the natural initiation of the state. In other words, despite the high temporal resolution of optogenetic activation, this approach can still be expected to initiate unphysiological sequences of events in dynamic circuits.

The claim by Van Dort et al. (1) that LDT/PPT neurons are important modulators of REM sleep initiation can only be substantiated by inactivating experiments that they did not perform. Grace et al. (3) found that cholinergic afferent input to the PRF is not needed for the occurrence of REM sleep, because blocking PRF cholinergic afferent input did not affect REM sleep frequency or time (under conditions capable of blocking the effects of stimulating cholinergic PRF afferents). Blocking cholinergic afferent input to the PRF did increase non–REM-to-REM sleep transition duration and failure rate, demonstrating that cholinergic PRF afferents play nonessential accessory role in REM sleep generation: enhancing the reliability of REM sleep switching (3).