Attributes of pheromone-sensitive ORN spiking patterns.

(A) Spikes were identified by their prominence in the raw ORN recording. The inter-spike interval (ISI) was defined as the time from one spike to the next. A train with ≥2 spikes with ISI ≤ 50 ms was considered a burst. The inter-burst interval (IBI) was the ISI between the last spike of the first burst and the first spike of the subsequent burst, omitting any individual spikes in between. Burst duration was the time between the first and last spike of a burst. The inter-event interval (IEI) included the ISI between bursts and individual spikes. The number of action potentials (spikes) per burst, and percentage of total action potentials as part of bursts were computed. The intra-burst interval was the ISI within a burst. Frequency of spikes, bursts, events and intra-bursts were determined as inverse of ISI, IBI, IEI and intra-burst interval. Detaching and reattaching the recording electrode to a long trichoid sensillum (Bi; paired t-test t(9) = −2.184; p = 0.057; N = 10) or adding DMSO to the sensillum lymph ringer (SLR) (Bii; Wilcoxon signed-rank test Z = 0.153; p = 0.922; N = 10) did not affect the mean spiking frequency.

The conductance-based model of a simplified hawkmoth ORN.

(A) Schematic of the model containing a voltage gated sodium channel (Na+), a voltage gated potassium channel (K+), a low voltage activated calcium channel (LVA), a calcium and voltage activated potassium channel (BK), a leak channel, and a cAMP-gated leaky non-specific cation channel (Orco). The Orco channel is represented as a linear conductance whose value oscillates with a sinusoidal shape as a function of the ZT. The circadian rhythm in the Orco conductance is hypothesized to result from a cAMP-dependent increase in channel open-time probability (as found in Drosophila Orco: (Getahun et al., 2013)) via a circadian oscillation of antennal cAMP concentrations (Schendzielorz et al., 2015). This oscillation is driven by the circadian clock and serves as an input; the internal mechanism of the clock is not modeled explicitly. Arrows indicate the flow of ions through the channels. (B) Parametrized circadian oscillation in Orco conductance introduced as an external input to our model, representing the gating of Orco by cAMP whose concentration oscillates on a circadian time scale. The times of minimum and maximum cAMP concentration are indicated by colored arrows. (C) Parametrized IV curve of the Orco channel showing a linear behavior with the slope depending on the cAMP concentration. Line colors correspond to the arrows in B. (D) Markov-chain representing all possible states and transitions for each type of ion channel included in the model. The per capita transition rates (α and β) depend on membrane potential and in the case of the BK channel also on the intracellular Ca2+-concentration. The “N” state vectors contain the population of channels in each state “n”. Directed edges are numbered in red. The only conducting state of each ion channel, representing the condition where all its gates are open, is shown in green.

Spontaneous activity of pheromone-sensitive olfactory receptor neurons (ORNs) of male M. sexta shows Orco-dependent circadian modulation.

(A) Example tip recordings of the spontaneous activity of one long trichoid sensillum during a period of high (top, average frequency ∼ 3 Hz) and low (bottom, average frequency ∼ 0.2 Hz) activity. (Bi) Spontaneous spike frequency of one long-term ORN recording during 17:7 light-dark cycles (same experiment as in A, colored arrows indicate respective Zeitgeber times (ZT)), indicated by the black-white bar at the top, increased during each activity phase and was low during the resting phases. Spike frequencies were averaged for each 1 h bin. (Bii) Mean spontaneous ORN spike frequency across all LD animals in 1 h bins. Data from the individual in Bi is highlighted in red. Recordings had different lengths, therefore the number of recordings used for the element-wise mean for each 1 h bin decreased with the time since the start of the recording, indicated in the top panel. (Biii) Mean spontaneous ORN spike frequency across all LD animals in 1 h bins during the first 48 hours revealed circadian activity. (Ci) The spontaneous spike frequency of one long-term ORN recording in constant darkness (DD, the expected times of lights on are represented by grey bars above) exhibited a circadian pattern. The peak activity shifted in DD due to the animal’s endogenous, free-running circadian period τ = 21.27 h. (Cii) Mean spontaneous ORN spike frequency across all DD animals in 1 h bins. Data from the individual in Ci is highlighted in red. (Ciii) Mean spontaneous ORN spike frequency across all DD animals in 1 h bins during the first 48 hours after aligning the time to the first maximum in spontaneous activity (see Methods) revealed circadian activity. (Di) The spontaneous spike frequency of one long-term ORN recording in constant darkness with infusion of the Orco antagonist OLC15 (orange frame) dissipated the circadian rhythm of spontaneous activity. (Dii) Mean spontaneous ORN spike frequency across all OLC15 animals in 1 h bins. Data from the individual in Di is highlighted in red. In contrast to LD and DD conditions, the spike frequency decreases over time with prolonged exposure to OLC15. (Diii) Mean spontaneous ORN spike frequency across all OLC15 animals in 1 h bins (mean ± SD) during the first 48 hours after aligning the time to the first maximum in spontaneous activity (see Methods). In contrast to LD and DD conditions, the circadian change in spontaneous spiking frequency disappeared. (E) The slopes of the linear fits to the binned spiking activity of each individual animal in the three different conditions (see Methods). Each dot indicates the slope for one experiment. The line indicates the mean. The slope for OLC15 is significantly different from control LD.

Blocking Orco removed circadian modulation of spontaneous ORN spiking patterns.

RAIN analyses demonstrated significant circadian rhythmicity for most animals in most attributes of the spontaneous spiking activity in LD and DD conditions but not in OLC15 (A). Values and background colors indicate the percentage of animals that expressed significant circadian rhythmicity. Circadian differences of attributes were further quantified between the time windows of low vs. high activity (Wilcoxon signed-rank test, α = 0.05; B-H). Significant differences occurred mostly in LD (white boxes) and DD (gray boxes) conditions, but never when Orco was blocked with OLC15 (orange boxes). In both LD and DD, the mean spiking frequency was increased during high activity (Bi, Bii). Only in LD but not in DD, the mean burst frequency (Ci, Cii) increased, whereas the relative number of spikes belonging to a burst (Di, Dii) decreased significantly during the high activity period. Both in LD and DD the mean event frequency (Fi,Fii) increased significantly. Mean burst duration (Ei, Eii), mean number of spikes per bust (Gi,Gii), and mean intra-burst intervals (Hi, Hii) did not differ between low vs. high spiking activity in LD and DD. When blocking Orco, the attributes for low vs. high activity did not differ for any of the attributes tested (Biii-Hiii). In DD and OLC15conditions, activity phases were aligned as described in the Methods; thus, low activity in DD was at subjective ZT 19 and in OLC15 at subjective ZT 13, high activity was in both cases at subjective ZT 24. ZTs for low and high activity in LD were 10 and 0, respectively.

Blocking Orco removed circadian regulation of ultradian rhythms in the spontaneous ORN spiking pattern.

Heat map of the instantaneous frequencies (1/ISI) over two consecutive days of long-term tip recordings of one pheromone-sensitive long trichoid sensillum under different conditions (LD (A), DD (B), OLC15 (C)) in each panel. Representative recordings from one animal in each panel. Pixel color indicates the counts of instantaneous frequencies in that respective bin. The points fall mostly in a band of high frequencies (>10 – ∼100 Hz) and low frequencies (0.01 – ∼10 Hz), with the high-frequency band representing the instantaneous frequencies of spikes within a burst and the low-frequency band the instantaneous frequency between bursts. (A, B) The high frequency band indicates daily and circadian modulation of frequency prevalence. In addition, the low frequency band (<0.01 Hz to ∼ 10 Hz) also displayed circadian modulation of the frequency composition. (C) Infusion of OLC15 deleted the circadian modulation in both frequency bands but not the ultradian rhythm of frequency prevalence.

Blocking Orco affected ultradian and infradian frequencies in the spontaneous ORN firing patterns.

The Fourier analysis of spontaneous spiking activity revealed rhythms with ultradian (<20 h; dark grey), circadian (20-28 h; white), and infradian periods (>28 h; light grey) in LD (Ai), DD (Bi), and OLC15 (Ci). Each line represents one animal where each triangular marker is at the local maximum in the frequency spectra obtained for that animal. The histograms above illustrate how often specific periods of spiking activity rhythms occurred averaged over all animals. Circadian rhythms were detected in LD (N = 5 of 11), DD (N = 7 of 10), and OLC15 (N = 4 of 12). Wavelet analysis of LD (Aii), DD (Bii), and OLC15 (Cii) confirmed the occurrence of multiscale periods. Example plots for one animal each, the same animals as in Figure 5. For each panel, the top plot depicts the mean firing frequency, the middle plot the wavelet power in the same period range as panels i, and the bottom plot the wavelet power for infradian periods up to 7 h on a log scale to highlight infradian time scales. Gray dashed lines indicate cone of uncertainty.

Orco is not under control of the TTFL-based molecular circadian clock.

Relative expression levels of Orco and tim mRNA from male hawkmoth antennae raised under LD conditions. Orco expression levels did not differ significantly between ZTs (ANOVA on ranks, H(5) = 6.91, p = 0.227: N = 4 per ZT). tim expression levels changed significantly throughout the day (1-way ANOVA, F(5, 18) = 6.215, p = 0.002) Red lines indicate mean, tim served as positive control.

The model adequately reproduces the firing pattern of the biological ORN.

(A) The model activity (bottom trace) displays both isolated spikes and bursts of variable length as in the original recording (top trace). The simulated trace shows the membrane potential (an intracellular variable), while the biological data were obtained with extracellular electrodes. (B) Semi-logarithmic plot of the distribution of spikes per burst of an experimental LD recording (top) and simulated model (bottom) of spontaneous action potential activity of pheromone-sensitive neurons. The experimental distribution corresponds to the 1-hour period of highest firing frequency in the recording; the simulated distribution was obtained from a separate 1-hour simulation at fixed high Orco conductance. Both the model and the recordings display exponential decrease of the count’s density as a function of the burst length, shown as a linear decrease in the logarithmic scale. (C) Comparison of experimentally recorded and model-predicted circadian modulation of spiking attributes. Top panels show mean spike frequency (Ci, experimental; Cii, model), whereas bottom panels show the percentage of spikes belonging to bursts (Ciii, experimental; Civ, model). The simulated model reproduced the circadian dynamics observed in the experimental recordings, demonstrating good agreement between model and biological data. (D) Heat map of the simulated model, x-axis shows time (days), y-axis shows instantaneous frequency (Hz), and pixel color indicates count density on a logarithmic scale. The instantaneous frequencies can be roughly divided in a band of high frequencies (top, ∼50 – 80 Hz) and low frequencies (bottom, ∼0.1 – 5 Hz), with the high-frequency band representing the instantaneous frequencies of spikes within a burst and the low-frequency band the instantaneous frequency between bursts. (Di) Simulated spikes with circadian regulation by the Orco channel; the inter-burst frequencies varied widely with circadian rhythmicity while the frequencies within burst remained approximately constant. The simulated results also reproduced the “merging” effect between the two bands when the bottom one approaches the top one. (Dii) Simulated spikes without circadian regulation of Orco had no circadian oscillations.

cAMP modulates open-time probability of Orco in M. sexta.

Perfusion of 8-Br-cAMP-AM significantly increased the spontaneous spiking activity at the end of the activity phase (ZT 1-3; A; Wilcoxon signed-rank test Z = −2.701; p = 0.004). However, activity did not increase when 8-Br-cAMP-AM was added together with the Orco blocker OLC15 (B; paired t-test t(9) = −1.446; p = 0.182). Electrodes were filled with sensillum lymph ringer containing 0.1% and 0.15% DMSO as control in A and B, respectively. Data from individual animals are connected by lines.

A signalosome assembling an endogenous multiscale PTFL membrane clock in the soma of hawkmoth olfactory receptor neurons (ORNs) with 2nd messenger-dependent Orco as central control hub is proposed to mediate membrane potential rhythms.

Schematic of an antennal pheromone-sensitive ORN that is strictly compartmentalized by tight junctions (thick black borders) into three functionally separated membrane compartments: cilium, soma/dendrite, and axon. Soma/dendrite and axon are wrapped by supporting cells (grey background), ensuring common extracellular ionic concentrations for both membrane compartments. Thus, only the cilium is exposed to the sensillum lymph that contains high (∼200 mM) K+ concentration. The ciliary membrane comprises a specialized membrane signalosome devoted to highly sensitive pheromone detection. In contrast, a signalosome in the soma/dendrite controls the membrane potential and intracellular Ca2+ concentration. Sustained depolarization via opening of cation-permeable, homomeric Orco pacemaker channels in the soma are counteracted by delayed voltage-gated hyperpolarizing K+-channels, together generating ultradian oscillations in the membrane potential with interlinked oscillations in intracellular Ca2+. Circadian/daily rhythms in cAMP concentrations mediating Orco conductance, next to Ca2+, superimpose a circadian rhythm on prevalence of ultradian oscillations. Orco is not under direct circadian transcriptional control by the circadian TTFL clock in the nucleus that is entrained to the LD cycle, thus being part of a PTFL membrane clock entrained to other Zeitgebers. PTFL- and TTFL-clocks are coupled by not-yet identified mechanisms. LD: light dark cycle; OR: odor receptor, Orco: olfactory receptor coreceptor, PTFL: posttranslational feedback loop clock; TTFL: transcriptional-translational feedback loop clock; Vm: membrane potential; sine waves: oscillations.

Parameter values for model currents.

Gating variable parameter values.

Membrane and calcium dynamics parameters.

Spontaneous spiking activity of ORNs of pheromone-exposed male M. sexta.

To enhance synchronization, males were exposed to females and pheromones as additional Zeitgeber for 30 mins at ZT 16 before the start of the experiment. Recordings were obtained over multiple days in DD. (A) The spontaneous spike frequency of one long-term ORN recording in constant darkness (DD, the expected times of lights on are represented by grey bars above) (B) Mean spontaneous ORN spike frequency across all DD animals in 1 h bins. Data from the individual in A is highlighted in red. (C) Mean spontaneous ORN spike frequency across all DD animals in 1 h bins during the first 48 hours after aligning the time to the first maximum in spontaneous activity (see Methods). No circadian activity was evident in these pheromone-exposed males.