1 NMDAR - dependent control of call duration in Xenopus laevis

43 44 Many rhythmic behaviors, such as locomotion and vocalization, involve 45 temporally dynamic patterns. How does the brain generate temporal complexity? Here, 46 we use the vocal central pattern generator (CPG) of Xenopus laevis to address this 47 question. Isolated brains can elicit fictive vocalizations allowing us to study the CPG in 48 vitro. The X. laevis advertisement call is temporally-modulated; calls consist of rhythmic 49 click trills that alternate between fast (~60 Hz) and slow (~30 Hz) rates. We investigated 50 the role of two CPG nuclei—the laryngeal motor nucleus (n.IX-X), and the dorsal 51 tegmental area of the medulla (DTAM)—in setting rhythm frequency and call durations. 52 We discovered a local field potential (LFP) wave in DTAM that coincides with fictive 53 fast trills and phasic activity that coincides with fictive clicks. After disrupting n.IX-X 54 connections, the wave persists whereas phasic activity disappears. Wave duration was 55 temperature-dependent and correlated with fictive fast trills. This correlation persisted 56 when wave duration was modified by temperature manipulations. Selectively cooling 57 DTAM, but not n.IX-X, lengthened fictive call and fast trill durations, whereas cooling 58 either nucleus decelerated fictive click rate. The NMDAR antagonist APV blocked 59 waves and fictive fast trills, suggesting that the wave controls fast trill activation and, 60 consequently, call duration. We conclude that two functionally distinct CPG circuits 61 exist—a pattern generator in DTAM that determines call duration, and a rhythm 62 generator (spanning DTAM and n.IX-X) that determines click rates. The newly 63 identified DTAM pattern generator provides an excellent model for understanding 64 NDMAR-dependent rhythmic circuits. 65 66 INTRODUCTION 67 68 Many rhythmic motor behaviors consist of multiple simple rhythms woven into 69 temporally and/or spatially intricate patterns. We wish to understand the neural 70 mechanisms by which discrete rhythms are temporally organized. A major obstacle to 71 understanding temporal patterning lies in the complexity of many behaviors. For example, 72 the control of behaviors such as birdsong or vertebrate locomotion involves the 73 coordination of many muscle groups in elaborate patterns of activation. 74 In this study, we investigated the neural basis of temporal patterning of calling in 75 the frog, Xenopus laevis. Xenopus vocalizations are generated by a simple mechanism of 76 sound production. Calls are produced independent of respiratory movements (unlike 77 most other vertebrate vocal mechanisms) by a single pair of laryngeal muscles. Despite 78 this mechanistic simplicity, the most common male vocalization—advertisement call—is 79 temporally complex, allowing us to explore how a tractable neuronal circuit generates 80 elaborate temporal patterns. 81 Each advertisement call consists of two click trills—fast (~60 Hz) followed by 82 slow (~30 Hz)—occasionally preceded by an introductory phase (~20 – 40 Hz, Fig. 1; 83 Tobias et al., 1998, 2004; Yamaguchi et al., 2008). Fast and slow trills last ~200 ms and 84 ~800 ms, respectively, and introductory phases are variable in duration. Previously, we 85 demonstrated that advertisement calls are generated by a central pattern generator (CPG; 86 Rhodes et al., 2007). The vocal CPG functions on two distinct timescales. On a shorter 87 (millisecond) timescale, the CPG produces distinct rhythms, ~60 Hz, ~30 Hz, and ~20 – 88 40Hz. On a longer timescale, the CPG controls the temporal pattern of trill delivery, with 89 each trill lasting hundreds of milliseconds, and calls repeating about once per second. 90 The Xenopus vocal CPG can be readily studied in vitro. A compound action 91 potential (CAP) on the motor nerve precedes each click produced by the larynx 92 (Yamaguchi and Kelley, 2000); thus nerve activity provides a direct readout of behavior. 93 An isolated brain preparation generates fictive vocalizations (with bath-applied serotonin, 94 5-HT) that are similar to in vivo advertisement calls (Rhodes et al., 2007). The vocal 95 CPG contains two reciprocally connected nuclei—the laryngeal motor nucleus (n.IX-X) 96 and the dorsal tegmental area of medulla (DTAM; Kelley, 1980; Wetzel et al., 1985; 97 Brahic and Kelley, 2003; also called pre-trigeminal nucleus by Schmidt, 1992). In this 98 study, we examined the roles played by the two CPG nuclei in regulating rhythms 99 (shorter timescale) and patterns (longer timescale). 100 We discovered a local field potential (LFP) wave endogenous to DTAM that 101 coincided with fictive fast trills; this activity resembled slow-wave activity originally 102 identified in the DTAM homologue of the leopard frog (Schmidt, 1992). We used 103 temperature and NMDAR perturbations to investigate the involvement of this LFP wave 104 in determining the temporal structure of motor output. Furthermore, we discovered 105 phasic activity in DTAM LFP that correlates with fictive clicks and requires intact 106 connections between DTAM and n.IX-X. Our findings indicate that the vocal CPG 107 includes two functionally distinct, though anatomically overlapping, neural circuits. One 108 circuit, including neurons in DTAM and n.IX-X (with required connections between the 109 two nuclei), functions as a rhythm generator that determines the click rates of fast and 110 slow trills. The second circuit—a pattern generator—is endogenous to DTAM and 111 produces the NMDAR-dependent slow wave, which in turn acts upon the rhythm 112 generator to determine the timing of fast trill onset and offset. 113 114 METHODS 115 116 Animals 117 118 We obtained adult male Xenopus laevis from Nasco (Fort Atkinson, WI). Ten 119 animals (weight, 56.3 ± 13.6 g; length, 7.5 ± 0.8 cm) were used for in vivo vocal 120 recording experiments; 61 frogs (weight, 39.1 ± 5.9 g; length, 6.7 ± 0.4 cm) were used for 121 in vitro experiments. Animals were kept in glass aquaria, and maintained on a 12:12 122 hour light:dark cycle. All procedures were approved by the Institutional Animal Care 123 and Use Committee at Boston University, and complied with National Institutes of 124 Health guidelines. 125 126 Vocal recordings 127 128 A previous study showed that in vivo click and in vitro CAP rates are dependent 129 on temperature (Yamaguchi et al., 2008). Here, we extended our previous study by 130 examining how temperature affected call and trill durations in vivo. The results of this 131 experiment allowed us to quantitatively characterize the thermal sensitivity of vocal 132 rhythms and patterns in vivo, and compare it to data obtained in vitro (described below). 133 Ten adult males were administered human chorionic gonadotropin (hCG; 600 – 1000 IU; 134 Sigma, St. Louis, MO) injected subcutaneously to elicit advertisement calling (Wetzel 135 and Kelley, 1983; Yamaguchi et al., 2008). Animals were placed in a 38 l aquarium; 136 vocalizations were recorded with a hydrophone (H2; Aquarium Audio Products, 137 Anacortes, WA) and a sound-activated recording system (Syrinx software, 138 www.syrinxpc.com; John Burt). Each male was recorded at two temperatures, 18° C and 139 26° C. Tank warming was achieved with an aquarium heater (VISI Therm type VTH 14