We humans spend a third of our life­time asleep. We sleep more as babies, and less in old age. Other ani­mals also sleep. In fact, for ani­mals who can’t afford to sleep com­plete­ly because of envi­ron­men­tal dan­gers, they use this very inge­nious strat­e­gy of sleep­ing with one eye open, and the two sides of the brain actu­al­ly take turns to sleep.

So, sleep is clear­ly very impor­tant. Not sleep­ing enough affects our abil­i­ty to work, to learn, or even to dri­ve safe­ly. Sleep dis­tur­bances are linked to mul­ti­ple psy­chi­atric dis­or­ders, includ­ing depres­sion. And long-term sleep depri­va­tion caus­es oth­er health prob­lems such as obe­si­ty and car­dio­vas­cu­lar diseases.

We used to think that sleep is a pas­sive process caused by reduced sen­so­ry stim­u­la­tion so that our nor­mal men­tal and phys­i­cal activ­i­ties can shut down. We held this belief since the time of Aristotle, and per­haps even before that. But now we know that this idea is com­plete­ly wrong.

About six­ty years ago in the 1950s, sci­en­tists dis­cov­ered two dis­tinct types of sleep, Rapid Eye Movement or REM sleep, dur­ing which we have vivid dreams, and non-REM sleep which is a deep slum­ber. And each night we cycle became REM and non-REM sleep. We now know that both REM and non-REM are con­trolled by ded­i­cat­ed, precisely-wired cir­cuits in the brain, and when the cir­cus are dam­aged due to dis­ease or aging, we devel­op all kinds of sleep-related prob­lems. So the goal of my research team is to fig­ure out the sleep cir­cuits, so that we can improve and per­haps even con­trol sleep.

The good news is that we already have a rough idea of where to look. We know that the sleep cir­cuits are like­ly in the hypo­thal­a­mus and brain­stem. But the bad news is that these brain regions are not just involved in sleep con­trol. In fact, the neur­al cir­cuit is con­trol­ling feed­ing, body tem­per­a­ture, heart rate, parental behav­iors, and var­i­ous emo­tions. They’re all very tight­ly packed togeth­er, even spa­tial­ly inter­min­gled with the sleep circuit. 

And to make life even hard­er, we have a hun­dred bil­lion neu­rons in the brain, and each sin­gle neu­ron has many long and thin process­es for con­nect­ing with oth­er neu­rons. So instead of an elec­tron­ic cir­cuit neat­ly designed by engi­neers, the brain is more like a bowl of spaghet­ti, very dif­fi­cult to untangle.

Fortunately, the neu­rons serv­ing dif­fer­ent func­tions often turn on dif­fer­ent genes and pro­duce from pro­teins in order to do their jobs. In oth­er words, they con­tain dis­tinct cell mark­ers, and if we can fig­ure out these cell mark­ers, we can actu­al­ly use them to sin­gle out the real cul­prit from the oth­er sus­pects. And once we iden­ti­fy the sleep neu­rons, we can label them with dif­fer­ent col­ors so we can look at them, which is a lot of fun. I like to do that. But more impor­tant­ly, we can actu­al­ly intro­duce light-sensitive pro­teins specif­i­cal­ly into the sleep­ing neu­rons so that we can use light to either acti­vate them or shut them down.

We know that neu­rons func­tion by fir­ing these lit­tle elec­tri­cal impuls­es called spikes, and it’s the prop­a­ga­tion of spikes in var­i­ous brain net­works that con­trols behav­ior. So, if we can use light to con­trol spik­ing of the sleep neu­rons, we can con­trol whether the ani­mal goes into sleep or wakes up.

A mouse with a small glowing light attached to its head, with a wire running from it

Our cur­rent work is most­ly done in the mouse. For exam­ple recent­ly we stud­ied a group of neu­rons in the medul­la, which is part of that brain stem. What we found is that when we turn on light to acti­vate these neu­rons, we can reli­ably con­vert non-REM sleep to REM sleep with­in a few sec­onds after we flip the light switch. So we think we can con­trol pre­cise­ly when the mouse starts dream­ing, and in addi­tion to con­trol­ling the spik­ing activ­i­ty, we can also observe the activ­i­ty. And because the sleep neu­rons are buried very deep in the brain, we insert a microen­do­scope into that brain region and attach a tiny cam­era to the head of the mouse. 

So, here you’re see­ing flick­er­ing of the neu­rons, which reflects their spik­ing activ­i­ty. And once we iden­ti­fy some of the sleep neu­rons, we also have the tools to map out the inputs and out­puts of this very com­plex net­work. In oth­er words, we can use the iden­ti­fied sleep neu­rons as a start­ing point to trace out this big net­work. This would allow us to fig­ure out what goes wrong when some­one devel­ops a sleep dis­or­der so we can come up with effec­tive strate­gies to repair the sleep circuit.

Identifying the key ele­ments of the sleep con­trol mech­a­nism might also allow us to apply exter­nal con­trol of these ele­ments, to nor­mal­ize sleep when the nat­ur­al cir­cuit mal­func­tions. At the moment, our tech­nique for con­trol­ling mouse sleep is not yet applic­a­ble to humans because it involves injec­tion of virus and inser­tion optic fibers into the brain. But when safer tech­niques are devel­oped for humans, we could poten­tial­ly turn sleep on and off at will. So my ques­tion for the audi­ence is, if that becomes pos­si­ble, do we want sleep on demand? Thank you.