Molecular control of temporal integration matches decision-making to motivational state

Abstract

Motivations bias our responses to stimuli, producing behavioral outcomes that match our needs and goals. We describe a mechanism behind this phenomenon: adjusting the time over which stimulus-derived information is permitted to accumulate toward a decision. As a Drosophila copulation progresses, the male becomes less likely to continue mating through challenges. We show that a set of Copulation Decision Neurons (CDNs) flexibly integrates information about competing drives to mediate this decision. Early in mating, dopamine signaling restricts CDN integration time by potentiating CaMKII activation in response to stimulatory inputs, imposing a high threshold for changing behaviors. Later into mating, the timescale over which the CDNs integrate termination-promoting information expands, increasing the likelihood of switching behaviors. We suggest scalable windows of temporal integration at dedicated circuit nodes as a key but underappreciated variable in state-based decision-making.

Figure 1:. Copulation Decision Neuron (CDN) activity controls the real-time decision to end matings

(a) Male flies decide whether to stop mating when threatened. The male (red eyes) maintains a stereotyped posture while mating. When challenged (in this case with a 41°C threat) he may decide to detach himself from the female (yellow eyes) and end the mating. (b) The CDNs (labeled by NP2719-Gal4) reside in the abdominal ganglion of the ventral nervous system (neuropil at bottom of image). Scale bar is 20 μm. (c) Acute silencing of the CDNs using the light-gated chloride channel GtACR1 prevents termination in response to a one-minute-long 39°C heat threat. Error bars for proportions here and in all other figures (unless otherwise stated) represent 67% credible intervals, chosen to resemble the standard error of the mean. For the number of samples in each experiment, see Supplementary Table 3. For statistical tests, see Supplementary Table 4. (d) Electrical activity in the CDNs is only necessary at the time of mating termination. Tonic silencing of the CDNs results in extended mating duration (fifth column, mean: 101 minutes), but silencing from the beginning until near the natural end of mating does not affect copulation duration (third column). Matings in which the CDNs are silenced through the normal ~23-minute termination time end seconds after the light is turned off (fourth column). Green rectangles represent the time during which the neurons were silenced. Error bars for copulation duration here and throughout represent standard error of the mean. (e) Acute optogenetic stimulation of the CDNs causes termination. Two seconds of stimulation is sufficient to terminate copulation regardless of how far the mating has progressed. “No ret.” refers to flies that were not fed retinal, the obligate chromophore for CsChrimson’s light sensitivity, showing that light on its own does not cause termination of mating. (f) The termination response to minute-long green light stimulation (green lines; Left: 9.72 μW/mm², Right: 8.03 μW/mm²) of the CDNs is potentiated as a mating progresses, similar to the response to real-world challenges like heat threats (orange lines).

Figure 2:. The CDNs integrate multimodal inputs over longer timescales as mating progresses

(a) Unlike longer challenges, the termination response to brief (500 ms and 1 sec), strong pulses of CDN stimulation does not increase as mating progresses. (b) Left: flies are given two strong 500 ms pulses of CsChrimson stimulation separated by an inter-pulse interval. Right: illustration of the probability of terminating the mating in response to two independent (top) or integrated (bottom) pulses. In the independent case, each pulse (first, red; second, blue) would terminate ~32% of matings, and so in total would terminate ~54% of matings. If information from the first pulse is integrated into the response to the second, then the overall proportion terminating the mating will be > 54% (in this example, 72% where the second pulse is augmented to 59%). (c) The probability of terminating the mating in response to two strong pulses of CDN stimulation is greater than expected if the pulses did not interact (grey line = 2p-p², estimated using the single pulse response value (p)) so long as the inter-pulse interval is sufficiently short. The ability of pulses to be integrated over longer timescales is increased at 15, relative to 10, minutes into mating. (d) Male flies become progressively more likely to stop mating in response to a 350-millisecond gust of wind. Mating pairs are not blown apart by the force of the wind; rather, the male terminates the mating with some delay after the end of the wind gust (Extended Data Figure 2 and see example in Video 4). (e) Constitutively silencing the CDNs with tetanus toxin (Tnt) prevents termination in response to wind (see example in Video 5). (f) Paired pulses of wind gusts (650 ms at 10 min, 250 ms at 15 min) are integrated over a longer timescale when delivered at 15 minutes into mating. (g) Six seconds of optogenetic grooming neuron stimulation causes termination with increasing propensity as the mating progresses. Grooming-induced termination is prevented at all time points by blocking CDN output with tetanus toxin (Tnt). (h) Two bouts of brief grooming neuron stimulation separated by 5 seconds results in a potentiation of the response to the second pulse, but only at 15 minutes into mating. A single brief pulse terminates the same fraction of matings at 10 and 15 minutes. The predicted value if the two pulses did not interact is indicated by the grey line. (i) Silencing the CDNs during a first demotivating stimulus prevents its integration with a later stimulus. Left: two bouts of grooming neuron stimulation were delivered at 15 minutes into mating, separated by 5 seconds. The CDNs were silenced during the first pulse of grooming neuron stimulation, as well as 2 seconds after the pulse offset. Right: silencing the CDNs during only the first of two optogenetic grooming pulses induces the same level of termination as if only one pulse (with no CDN inhibition) had been delivered. (j) Response of a model system to input pulses of equal strength but varying duration. The instantaneous probability of terminating a mating in response to a stimulus is: p₀τ(1 − exp(−t/τ)). p₀= perceived intensity, τ = time constant of integration, t = time since start of stimulus. Left: for pulses shorter than the integration window, the peak response is approximately the same regardless of τ, but the amount of information retained after pulse offset is greater with a greater τ. Right: for pulses longer than the integration window, information accumulates faster and peaks higher with a greater τ. Error bars represent estimated pointwise 95% coverage intervals. (k) Plotting the cumulative termination after the onset of green light and fitting the data (left) reveals a higher time constant of integration, τ, at 15 minutes into mating compared to 10 minutes (right). Error bars represent one standard error of the parameter fit estimated using the Cramér-Rao bound (see Supplementary Note 2, Extended Data Figure 5,6).

Figure 3:. CaMKII activity in the CDNs sets the timescale of integration

(a) When the CDNs are tonically stimulated with white light throughout courtship and mating, copulation duration is reduced to ~11 minutes. Out of ~500 genetic manipulations (mostly RNAi), several manipulations of CaMKII in the CDNs strongly altered copulation duration. The average duration of each genotype (consisting of at least 6 flies) is rounded down and binned to a whole number, and bins containing CaMKII manipulations are indicated with colored arrows. (b) Knocking down CaMKII in the CDNs of males increases the sensitivity of matings to heat threats early into mating. (c) Expressing constitutively active CaMKII (T287D) in the CDNs protects matings from heat threats even late into mating. (d) Decreasing CaMKII activity via RNAi increases the rate and overall fraction of flies terminating to green light at 10 minutes whereas increasing CaMKII activity with T287D expression decreases the rate and overall fraction of flies terminating to green light at 15 minutes. (e) Knocking down CaMKII more strongly potentiates the termination response to sustained green light stimulation (8.03 μW/mm²) than a short pulse of red-light stimulation, though each protocol terminates the same proportion of control flies. (f) Knocking down CaMKII in the CDNs allows paired wind gusts to be integrated across a 10 second inter-gust interval at 10 minutes into mating. Single pulse lengths were calibrated so that flies would terminate at a rate of ~30% for each genotype. (g) Cumulative fraction of flies terminating after each light pulse at 10 (top) and 15 (bottom) minutes into mating. At both 10 and 15 minutes into mating, males integrate pulses that are separated by less than 2.5 seconds, but integration across longer intervals was seen only at 15 minutes. CaMKII knockdown allows flies to integrate pulses at 10 minutes as if it were later into mating, while constitutively active CaMKII suppresses the time over which the CDNs can integrate pulses. (h) Ten 250 millisecond red light pulses with a set amount of time between each pulse were delivered to CDN>CsChrimson flies at either 10 or 15 minutes into mating.

Figure 4:. Dopamine restricts integration by facilitating CaMKII activation by calcium

(a) Dopamine motivates matings by restricting integration by the CDNs. Sustained thermogenetic stimulation of dopaminergic neurons with TrpA1 protects the mating against optogenetic stimulation of the CDNs at 15 minutes into mating. Each black stripe in the ethograms represents a single mating. (b) Thermogenetic stimulation of dopaminergic neurons drives release of dopamine onto the CDNs as measured by changes in the brightness of the genetically encoded fluorescent dopamine reporter GRAB-DA3m. Top: example abdominal ganglion (outlined with a black dashed line) at room temperature, with dim GRAB-DA3m fluorescence (left), and the same AG after increasing the temperature to stimulate local dopaminergic neurons (right). Bottom shows quantification of changes in fluorescence with temperature in flies expressing TrpA1 in the dopaminergic neurons (purple) or without TrpA1 (black). (c) Dopamine promotes the activation of CaMKII as reported by the fluorescence lifetime of the FRET sensor green-Camuiα. Left: ~10-milliwatt, 920 nm laser stimulation of the Channelrhodpsin-2 variant ChR2-XXM only slightly increases CaMKII activity in the CDNs (black); subsequent perfusion of 100 μM dopamine allows the same stimulation of ChR2-XXM to strongly increase CaMKII activity (blue). NP5270-Gal4 is used as the CDN driver for green-Camuiα and GCaMP6s imaging experiments (see Methods [Imaging experiments, Region of interest]). Each bold trace is the mean of the light traces. Middle: A representative sample trace. Right: fluorescence lifetime map of green-Camuiα signal in CDN axons from the example tracebefore and after dopamine perfusion. (d) Dopamine cannot increase CaMKII activity without ChR2-XXM stimulation. (e) Dopamine does not increase calcium influx in the CDNs as measured by GCaMP6s. Baseline fluorescence is calculated as the mean number of photons for the first 5.12 seconds of pre-perfusion recording, which was done at the same 10-milliwatt laser power (see Methods [Imaging experiments, Optogenetic stimulation while imaging/Calcium imaging]). (f) Increasing dopamine concentration increases the ability of ChR2-XXM stimulation to activate CaMKII. Each trace is a single fly. For the 0 μM concentration, saline was perfused while imaging.

Figure 5:

A motivating dopamine signal acts through CaMKII to bias behavioral choice by controlling retention of decision-relevant information.