Communication using sound, such as human speech, relies on matching the appropriate vocal response to an acoustic signal. You say, 'Hello'; I say 'How are you?' We are interested in how the nervous system produces this match. Understanding this problem requires understanding sound production, motor circuits in the brain, organs of hearing and auditory brain regions. The match also depends on social context: you might ignore my 'hello' if angry with me or respond more effusively if a long conversation is called for. Social context is also reflected in brain activity and is influenced by emotional state conveyed by circulating hormones and neuromodulators.

We use a specific model system to study these issues, a frog species that relies almost entirely on vocalizations for social communication (for example calls, see Fig 1 below). The underwater songs of the South African clawed frog. Xenopus laevis (an NIH model organism) provide rich and well-characterized experimental system for investigating how the neural basis of vocal communication. Much of the neural circuitry involved has been well conserved evolutionarily but is particularly prominent in this species because of its importance in reproduction. Also conserved are the hormones that shape neural circuitry during development, their downstream target genes and the cellular mechanisms that control vocal pattern generation and the auditory processing of communication sounds. The ability to evoke fictive songs from the isolated vocal organ and even the isolated brain (unique thus far among vertebrates) provides particularly powerful experimental approaches. We also use these approaches to determine how vocal signaling arose during evolution, changed as species diverged and comes to be expressed in specific male- and female- vocal patterns.

Vocal behaviors are used to signal sexual receptivity and unreceptivity, dominance and territoriality. Each sex has a distinct and behaviorally powerful repertoire of songs. In X. laevis, specific calls have the ability to vocally suppress other individuals or to excite calling by another frog. The songs of clawed frogs are a series of repeated clicks. Information on the sex and reproductive state of the vocal signaler is conveyed primarily by click rate. We have developed a way to stimulate the auditory system of these frogs (that normally hear under water) in air so that we can study how the nervous system decodes the communicative significance of different click rates and translates that information into the suppression or enhancement of vocal output.

Songs are generated by rapid contractions of intrinsic laryngeal muscles in response to patterned activity in the laryngeal nerves. The isolated larynx can produce actual vocalizations, and we can record from the laryngeal nerve of singing frogs. We can also evoke neural correlates of singing from the isolated brain. These experimentally favorable preparations permit identification of male- and female-specific electrophysiological and muscular characteristics that underlie the different vocal repertoires of the sexes.

Many sexually differentiated characteristics are due to sex differences in hormone secretion during development and adulthood. Which tissues develop in a sexually differentiated form and which do not is determined by the expression of specific steroid receptors, which genes are regulated by those hormones and how alterations in gene expression affect cell/cell interactions. Androgen controls sexually differentiated vocal development by regulating cell numbers and types. Sex differences in the larynx include synaptic efficacy - weak synapses in males versus strong synapses in females- and in muscle fiber type - rapidly contracting fibers in males and slowly contracting fibers in females. Estrogen controls the efficacy of the laryngeal synapse. Muscle contraction rate is tied to expression of an androgen-regulated, embryonic-like myosin, LM. We have used a related frog species, Xenopus tropicalis, with a more tractable genotype to determine the place of LM among the family of myosin heavy chain genes. LM is found within a family of embryonic isoforms but is expressed only in the larynx and under the influence of androgenic steroids. We have developed an in vitro preparation of the developing larynx to study the relation between generating new muscle cells and LM expression. The developmental switch from a female-like, mostly slow twitch complement of muscle fibers to a masculine all fast twitch complement relies on androgen-induced division of muscle stem cells.

We utilize a larynx/brain preparation to examine sexually differentiated motor programs using intracellular recording patch clamp and tracing techniques. These approaches have outlines a hindbrain circuit for vocal production that includes the parabrachial nucleus of the rostral hindbrain, essential for generating different vocal patterns, interneurons and motor neurons in nucleus ambiguous. We are exploring how auditory information gains access to downstream motor elements using vocal suppression and vocal excitation paradigms indentified behaviorally.

In the CNS, all elements of the neural circuit involved in vocal production express androgen receptor. These include a specific nucleus of the inferior colliculus in which neurons are tuned to the temporal features of song. In the forebrain, the amygdala receives auditory input and projects to the parabrachial nucleus. Gonadotropin stimulates advertisement calling, neurons of the amygdala express gonadotropin receptor and this hormone activates gene expression specifically in these neurons.

Each species of Xenopus and of its sister genus Silurana produces a highly specific male advertisement call, a species-specific acoustic signature. A phylogenetic analysis of these calls suggest that simpler call types such as the click and more complex call types such as biphasic trills arose from a call of intermediate complexity in temporal pattern. The lack of phylogenetic signal in call types suggests selection for signaling species identity. We are using multiple approaches (including candidate genes we have identified and QTL-based approaches) to identify the alterations in gene sequences, regulation or patterns of expression that underlie vocal evolution.


Fig 1. Oscillograms (sound intensity vs. time) of the 6 male and 2 female song types of Xenopus laevis.