photo of man with finch on fingertip

Team leader Todd Roberts, UT Southwestern Medical Center, USA, conferring with one of his test subjects. [Image: UTSW]

The techniques of optogenetics, which allow neuronal activity to be tracked and controlled by targeted application of light, have found wide use in studies mapping neural circuits in the brain, and have even been eyed for long-term therapeutic promise in treating neural disorders, heart disease, blindness and more. A team of U.S. scientists has now demonstrated something else that optogenetics can do: teach a bird how to sing (Science, doi: 10.1126/science.aaw4226).

The research team, based at the UT Southwestern Medical Center, Dallas, Texas, used “opto-tutoring” experiments to implant memories of song patterns in young zebra finches that had not yet been tutored by a parent or other older finch. In so doing, the researchers were able to map out some of the complex circuitry that underlies the transition between vocal learning and its physical/motor expression. That mapping could, the investigators believe, ultimately cast light on the considerably more intricate circuitry in the human brain that governs the learning of speech—and how that circuitry can go awry.

Imitation, trial and error

The UT Southwestern team, led by Todd Roberts, focused on zebra finches for their study because the way that the birds learn their songs has some clear parallels to human speech learning.

Young male zebra finches are known to hear the sounds of their fathers’ songs early in life, during a “developmentally sensitive” period. These songs (the details of which differ subtly from bird to bird) become encoded in the young bird’s memory, and it starts to practice the song on its own, apparently comparing its own primitive version with the encoded memory to refine its tune. By the time the young bird has reached adulthood—and after thousands of repetitions and practice runs—it has learned to reproduce the song.

That simple statement of how these birds learn their songs, however, masks a complex, multi-stage dance of neural development. The bird needs to encode the raw sensory pattern of the adult tutor’s song in the brain into an accessible memory, and somehow translate that memory, through trial and error, into the complex set of motor controls that will actually shape the notes and their ordering.

Optogenetic illumination

To get at the neural pathways that govern that transition, Roberts’ team zeroed in on a particular region of the zebra finch brain called the nucleus interfacialis of the nidopallium (NIf)—a sort of neural junction box between sensory and motor processing in the brain network governing auditory and vocal control. The researchers dug into the details of that zone via the relatively new but rapidly maturing technique of optogenetics.

photomicrograph of optogenetically lit up NIf area

The UT Southwestern researchers used optogenetics in the NIf region of the zebra finch brain (shown here) to artificially encode memories of songs the birds had not yet learned. The birds subsequently used these memories to learn syllables of their songs. [Image: UTSW]

In a typical optogenetic experiment, the techniques of gene therapy and molecular biology are used to implant genes encoding a light-sensitive protein (usually in the rhodopsin family) at strategic places in the DNA of populations of neurons. Depending on the specifics, the light-sensitive protein will cause the neurons literally to light up when transmitting a neuronal signal, or will allow neurons to be selectively fired when stimulated by closely targeted pulses of light from a laser or LED. Those capabilities have quickly made optogenetics a method of choice for stimulus–response studies and for detailed mapping of neural networks in studies of live animals, with far more specificity than the alternative of electrical stimulation.

Blue light pulses

For the zebra finch work, the UT Southwestern team began by injecting a virus, genetically modified to express the light-sensitive protein channelrhodopsin-2, into the NIf region of young zebra finches that hadn’t yet been exposed to adult songs. The virus then implanted the gene into the gene sequence of the NIf neurons, causing them to express the protein.

Channelrhodopsin-2 is sensitive in particular to blue light. So, in experiments lasting two days during the birds’ developmentally sensitive period, the researchers delivered 470-nm blue light pulses from an LED diode via a fiber optic interface through the birds’ skulls, to attempt to implant synthetic memories of birdsong at the crucial NIf junction.

The researchers found that, on reaching adulthood, birds that had been trained early with short optical pulses produced songs with noticeably shorter elements than those trained with long optical pulses. What’s more, it turned out that these artificial memories, directly implanted by light signals in the key NIf region, overrode learning from subsequent tutoring by live birds. And if the team introduced injuries to disrupt the synapse connecting the NIf region to a downstream center responsible for motor control—an area called HVC—it would disrupt song development only if the injury was introduced before opto-tutoring, not afterward.

Analogs to speech learning?

All of this suggested that the artificial stimulation of the NIf region through optogenetics introduced a transient, synthetic memory in the birds. Moreover, that memory apparently is stored somewhere else in the brain after the learning happens, and can still be used by the bird for testing and refining its song even if the link between the NIf and HVC regions is subsequently broken. The NIf–HVC pathway is necessary for forming the memory in the early, developmentally sensitive learning period—but once the memory’s formed, storage and recall happen in another place.

Just where that other place is, the researchers acknowledge in the paper, remains something a mystery. The new work, however, does open the door to a more detailed study of the networks governing the learning of vocalizations. And the team hopes that this knowledge could someday illuminate and help map the more complex interactions in human speech learning—and to target and possibly treat genes responsible for some speech pathologies.

“The human brain and the pathways associated with speech and language are immensely more complicated than the songbird’s circuitry,” Roberts said in a press release accompanying the work. “But our research is providing strong clues of where to look for more insight on neurodevelopmental disorders.”