, 2008a and Mallet et al., 2008b) (Figures 1A–1D). We examined whether the “antiphase” firing of GP-TI and GP-TA neurons was preserved across these two extreme brain states. For this purpose, some of the GP-TI and GP-TA neurons were also recorded during the activated brain state with its characteristic beta oscillations (Figures 1A and 1B). Dichotomous spike timings of GP-TI and GP-TA neurons during slow oscillations (Figures 1E and PD0332991 nmr 1F) were indeed maintained during cortical beta oscillations (Figures 1G and 1H). GP-TI neurons (n = 14) were, on average, most likely to fire at 44.3° ± 18.4° (mean ± SEM; range

of preferred angles 348°–134°, p < 0.05, Rayleigh tests) with respect to the peaks of cortical beta oscillations at 0°/360°. However,

GP-TA neurons (n = 23) fired at a significantly different phase (p < 0.05, Watson-Williams F test) of 275.8° ± 7.4° (range, 208°–358° p < 0.05, Rayleigh tests). Average firing phases of these identified GP-TI and GP-TA neurons were similar to those of several hundred GPe units recorded with multielectrode arrays ( Mallet et al., 2008a). Different spike-firing patterns of identified neurons were mirrored by inversely-related firing rates, irrespective of brain state. During SWA, GP-TI neurons fired much faster than GP-TA neurons, but during beta oscillations, the situation was reversed and GP-TA neurons fired faster than GP-TI neurons (both p < 0.05, Mann-Whitney tests). Furthermore, most GP-TI neurons (93%) decreased their firing rates during Linifanib (ABT-869) transition from SWA to the activated brain state, whereas most GP-TA neurons (97%) increased firing (Figures 1I and 1J). Regularity of firing was SB431542 cost also different, with GP-TI neurons firing more regularly than GP-TA neurons during SWA but less regularly during beta oscillations (both p < 0.05, Mann-Whitney tests). Both GP-TI and GP-TA neurons fired more regularly during beta oscillations as compared to SWA (Figures 1I and 1J). Importantly, a small sample of

identified GPe neurons (n = 7) fired so infrequently during SWA (mean firing rates of 0.002–0.2 Hz) that we could not statistically classify them as GP-TI or GP-TA neurons. However, we established that these very slow-firing neurons had some other key properties of GP-TA neurons. First, like GP-TA neurons, they all strongly increased their firing rate upon transition from SWA to the activated brain state (0.06 ± 0.02 Hz and 15.3 ± 2.4 Hz, respectively). Second, their firing was significantly modulated in time with cortical beta oscillations (p < 0.05, Rayleigh tests) and they were most likely to fire at phases (283.5° ± 10.5°; range, 246°–318°) that were similar to those preferred by identified GP-TA neurons but not by GP-TI neurons (p > 0.05 and p < 0.05, respectively, both Watson-Williams F tests). Thus, all of these slow-firing GPe cells were classified as GP-TA neurons for group analyses (the molecular profiles of these slow cells were also identical to GP-TA neurons; see below).