Attention-Like Deficit and Hyperactivity in a Drosophila Memory Mutant

Behavioral paradigms: optomotor maze

Wild-type flies respond to moving visual effects by turning in the direction of image motion (Heisenberg and Wolf, 1984) (an optomotor response, as measured by the OI; see Materials and Methods). In our optomotor maze paradigm, whereas wild-type Canton S flies display an average optomotor index of 0.75 ± 0.08, radish mutants showed diminished responsiveness to moving visuals in the maze (OI = 0.10 ± 0.12 for green/black 1 cm grating moving at 3 Hz; see Materials and Methods) (van Swinderen, 2007). Because this radish¹ phenotype is not attributable to blindness [radish mutants can respond to visual objects in other conditions (Gong et al., 1998)], we wondered whether radish mutants might be characterized by distinct behavioral idiosyncrasies that interfered with choice behavior as they completed the maze. To measure short-term processes in the maze, we filmed and quantified individual behaviors as flies progressed through the eight-level choice maze (Fig. 1A). Turn stereotypy, defined as the average number of consecutive turns a fly made in the same direction (Fig. 1B) (see Materials and Methods), was tallied for 40 individuals. Wild-type flies display significant stereotypy in the direction of optomotor flow, making ∼2.5 consecutive turns in that direction compared with ∼1.8 consecutive turns against optomotor flow (Fig. 1C). This short-term hysteresis, or behavioral persistence, most likely contributes significantly to the positive optomotor score (van Swinderen and Flores, 2007). If turn behavior were random and independent across choice levels, one would see a symmetrical distribution resembling a binomial model, with ∼1.9 consecutive turns in the same direction (Fig. 1D). Interestingly, radish mutants behaved much like the binomial model, lacking any hysteresis across choice levels, with ∼1.9 consecutive turns either with or against optomotor flow (Fig. 1E). The even distribution of radish animals across the nine endpoints of the maze further suggested a defect in visual responsiveness in radish mutants (Fig. 1F).

Figure 1. The optomotor maze paradigm. A, Maze. Arrow, Grating direction and velocity. Black trace, Filmed path of an individual fly. B, Each individual fly path is quantified for turning behavior. The number of successive turns in the same direction (+, with the moving grating; −, against the grating) is tallied per fly. Reversals of direction were also counted (rev). C, The normalized frequency of consecutive turn categories is plotted as a histogram for wild-type flies (n = 40 flies; weighted average ± SEM). The average value for either direction is indicated. D, Histogram for data created by a random model (50% turn probability at each choice level), with corresponding consecutive turn averages. E, Histogram and turn averages for radish¹ mutants (n = 40 flies). F, Average ± SEM distribution of flies among the nine collection tubes at the end of the maze (n = 8 mazes of 25–30 flies for wild type and radish¹) compared with a theoretical distribution for the random model.

Weak optomotor performance in the maze caused by the radish mutation was rescued by activating a wild-type radish transgene, hs-rsh(161), on a mutant radish background. Interestingly, rescue required prolonged (5 h) activation of the radish gene during development. The heat-shock regimen (see Materials and Methods) used previously (Folkers et al., 2006) to rescue radish¹ for odor-learning defects in the adult was not sufficient for rescuing optomotor defects (before heat shock, OI = 0.09 ± 0.15; after heat shock, OI = 0.18 ± 0.14). However, radish transgene activation during the late pupal stage rescued optomotor responsiveness to our standard 3 Hz grating (OI = 0.55 ± 0.12), but activation during the third-instar larval stage did not (OI = 0.06 ± 0.4). The requirement of radish during development for normal optomotor responses resembles similar results found for dunce mutants (which display extremely high optomotor responsiveness) in which only dunce gene induction during development (but not adulthood) rescued dunce optomotor phenotypes (van Swinderen, 2007). Furthermore, dunce defects were also found to be rescued specifically at the late pupal stage (van Swinderen et al., 2009). These results suggest that wild-type optomotor responsiveness levels are dependent on cellular events that occur during a specific stage of brain development and that dunce [a cAMP phosphodiesterase (Davis, 2005)] and radish [a gene with no clear homology (Folkers et al., 2006)] may be required during the same pupal stage for the formation of circuits affecting optomotor responsiveness.

Radish mutants are not blind and respond to visual stimuli, as has been shown in other behavioral studies (Gong et al., 1998). Phototaxis appears to be normal in radish mutants when tested in response to a light source in our maze paradigm (OI = 2.4 ± 0.15 vs OI = 2.55 ± 0.08 for wild type). To fully explore whether radish¹ responds at all to moving gratings, we exposed the mutants to different grating velocities. Wild-type flies typically display a velocity curve with decreased responsiveness at very low (<1 Hz) or very high (>10 Hz) grating velocities (Fig. 2A). We were surprised to find that radish mutants displayed a shifted velocity curve compared with wild type, with peak responsiveness at low (∼1 Hz) grating velocities but no response at 3 Hz (Fig. 2A), which is the grating velocity we used in our initial behavioral characterization (above). Although this shifted optomotor effect in radish¹ requires additional explanation, the fact that radish¹ does respond to slower moving gratings allowed us to test the mutant for distraction phenotypes.

Figure 2. radish¹ optomotor responsiveness and distractibility. A, Optomotor responsiveness (means ± SEM; n = 8 maze runs for each point) for wild type and radish¹ for different grating velocities. radish mutants do not respond to gratings moving 3 Hz or faster but do respond to more slowly moving gratings (1 and 2 Hz, p < 0.05 by t test against zero). *p < 0.05, significantly different from wild type by t test of means. B, Optomotor responsiveness (means ± SEM; n = 8 maze runs for each point) to a moving grating for wild type (at 3 Hz) and radish¹ (at 1 Hz; see above) in the presence of a visual distracter. A vertical white bar (5 × 25 cm) on a red background, positioned opposite of the direction of grating movement, was varied in intensity (percentage white; a schema is shown below the graph) to quantitatively test the distractibility of flies responding to the moving grating. *p < 0.05, significantly different from wild type by t test. C, Phototaxic responses of radish¹ and wild type in the maze to the distracter alone (the same schema shown above the graph).

Behavioral paradigms: distraction in the maze

Distraction can be measured by the loss of optomotor responsiveness to a moving grating caused by competing visual objects (van Swinderen and Flores, 2007). To test radish¹ for distractibility, we used two different paradigms that presented competing objects in different ways. In the first paradigm, we modified a previously used method in which static bars presented to the side of the maze abolish optomotor responses to the moving grating displayed under the maze (van Swinderen and Flores, 2007). To better quantify distraction phenotypes, we replaced the static (cardboard) distracters of our previous paradigm with programmable displays on LCD screens and measured responsiveness to the grating in competition with increasingly salient distracters (see Materials and Methods). We found that radish mutants (responding to a 1 Hz grating) lost optomotor responsiveness at lower levels of distracter intensity than wild type (Fig. 2B), although responsiveness to the distracter alone was the same for radish¹ and wild type (Fig. 2C).

To test the effect of distracting visuals in a different way, we presented them as competing moving objects on the CRT beneath the maze. Instead of the moving green/black grating, we displayed an intercalated field of green circles and blue squares moving slowly (∼1 Hz) in opposite directions (Fig. 3A). Wild-type fly populations presented with this conflicting scene display a “U”-shaped distribution after completing the maze, suggesting that flies are following one or the other wide-field pattern for periods of time as they complete the maze (Fig. 3B, black line). In contrast, radish mutants display an “inverted U” distribution when presented with the competing optomotor stimuli (Fig. 3B, gray line). This suggests that, unlike wild type, radish¹ individuals do not follow one or the other wide-field stimulus for enough time to bias final fly distributions in the maze. radish¹ flies are nevertheless responsive to each wide-field stimulus presented independently, as are wild type (Fig. 3C,D),

Figure 3. Optomotor responsiveness to competing objects. A, A field of blue squares (1 cm wide) moving right to left at 1 Hz is intercalated with a field of green circles (1 cm diameter) moving left to right at 1 Hz (see Materials and Methods). B, Average ± SEM distribution in the maze (n = 8 maze runs) for wild type and radish¹ in response to the visual in A. *p < 0.05, significantly different proportion of flies by t test. C, Wild-type responses to the moving squares or circles presented individually, with average ± SEM optomotor indices (n = 8 maze runs). D, radish¹ responses to the moving squares or circles presented individually, with average ± SEM optomotor indices (n = 8 maze runs).

Together, results from our visual competition paradigms as well as from filmed experiments over moving gratings suggest that the behavioral defects of radish mutants pertain to short-term choice processes (e.g., increased distraction, random alternation dynamics, and attention-like deficits). To pursue this hypothesis, we further characterized short-term behavioral processes in radish mutants using an entirely different visual paradigm, the flight arena.

Behavioral paradigms: flight arena

Suspended at a torque meter, flies spontaneously initiate turning maneuvers by modulating their wing beat amplitudes, resulting in measurable torque effects around the fly's vertical body axis (yaw torque). We used the yaw torque signal to allow the flies to control the angular position of visual patterns on the arena wall (Heisenberg and Wolf, 1984). In such a feedback setup, flies are thus able to report ongoing decision-making by their choice of flight direction.

Further supporting the notion that radish mutant flies are not visually impaired, the mutants show an optomotor response in the flight arena and are able to stabilize the rotation of the visual patterns (i.e., fly straight). In our setup, we allowed the flies to choose between flying toward a pair of upright Ts and a pair of inverted Ts. Previous research has shown that radish mutants can distinguish and learn to associate these two patterns with aversive stimuli (Gong et al., 1998), but when we analyzed the flight recordings of naive animals, we discovered that wild-type flies spend longer time periods directly in front of the patterns compared with radish mutants. The median time spent by radish mutants fixating on either pattern was significantly less than wild type: only 1.3 s compared with 3.0 s for wild-type flies (p < 0.01 by Kruskal–Wallis nonparametric test for medians) (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). We next analyzed the torque trace for any abnormalities that could explain this difference. Surprisingly, we found a strong ∼1.6 Hz peak (Fig. 4A, gray line) in the power spectrum of the radish¹ torque trace, whereas wild-type flies controlling the same visual stimuli did not show such oscillatory hyperactivity in their torque behavior (Fig. 4A, black line) (for sample torque traces, see supplemental Fig. 2, available at www.jneurosci.org as supplemental material). A different variant of radish¹, hs–rsh(161) (Folkers et al., 2006), also showed this peak (supplemental Fig. 3, available at www.jneurosci.org as supplemental material), firmly establishing the mutated radish gene, rather than the genetic background, as the cause of this oscillatory hyperactivity.

Figure 4. Tethered flight. A, Average ± SEM power spectra between 0 and 5 Hz for wild-type (black line; n = 25) and radish¹ (gray line; n = 24) torque behavior in 6 min closed-loop flights with two distinct visual objects (see Materials and Methods). B, Average power spectra between 0 and 5 Hz for wild-type (black line; n = 25) and radish¹ (gray line; n = 21) torque behavior in 6 min open-loop flights without any visual landmarks.

To study whether this “fidgety” phenotype was generated endogenously or induced by the visual patterns, we analyzed the torque traces of flies recorded in the absence of any patterns. Without feedback or visual stimuli (i.e., open loop), we found that turning behavior in radish mutants was still characterized by some hyperactivity between 1 and 2 Hz. However, in this situation, hyperactivity was not characterized by a single frequency peak, being instead more broadly distributed between 0.5 and 3 Hz (Fig. 4B, gray line). Wild-type torque behavior was again not strongly periodic without visual stimuli present (Fig. 4B, black line). These tethered flight data suggest that the presentation of competing visual stimuli induces oscillatory hyperactivity in radish mutants.

Brain recording paradigms: oscillations

We were curious as to whether the flight hyperactivity of radish¹ flies was indicative of more widespread activity effects in the brain. To determine this, we recorded LFPs using a tethered preparation that has been used previously to study brain activity in flies (Nitz et al., 2002). LFPs reflect the synchronous activity of populations of neurons, and these have been associated with visual (van Swinderen and Greenspan, 2003) as well as olfactory (Tanaka et al., 2009) stimuli in Drosophila. As in the flight arena, flies are tethered by their head and thorax but are free to move their legs and wings in this electrophysiology preparation (Fig. 5A). Two glass electrodes implanted into the brain and recording a voltage differential provide an ongoing measure of brain activity in the fly. Overall LFP activity in radish mutants was not significantly different from wild type, as evidenced by spectral analysis between 5 and 100 Hz (data not shown). However, when we explored low LFP frequency domains (0–5 Hz) in radish mutants exposed to visual stimuli (van Swinderen and Greenspan, 2003), we did notice increased LFP activity between 1 and 2 Hz in most radish flies compared with wild type (Fig. 5B), although this was not significant on average for our sample.

Figure 5. radish¹ brain recordings. A, Arena setup. Visual objects rotate around the fly clockwise with a period of 3 s. B, Average ± SEM power spectrum of wild-type and radish¹ brain activity between 0 and 5 Hz (n = 14 flies for both genotypes). The large peaks below 1 Hz (off scale) represents responses to the visual objects rotating around the fly at 0.33 Hz. C, Flies were exposed for 100 s to two identical squares before one of the squares changed to a cross. The amplitude (or power) of brain LFPs bandpass filtered at 20–30 Hz was correlated with the position of either competing object as it swept in front of the fly following a novelty transition. Data for a sample wild-type fly are shown. The black vertical line indicates when the square changed to a cross. D, Sample 20–30 and 1–2 Hz bandpass-filtered wild-type data for several rotations of the visual panorama before and after a novelty transition. Circled epochs indicate successive 20–30 Hz responses to the novel cross. E, Enlarged section of the novelty response in D. F, Sample 20–30 and 1–2 Hz bandpass-filtered radish¹ data for several rotations of the visual panorama before and after a novelty transition. G, Enlarged section of the novelty response in F. The response to the novel cross (circled) does not recur selectively, as it does in wild type.

In our brain recording preparation, our animals are not flying, and therefore they are not engaged in the behavioral alternations that produce torques in the flight paradigm. To determine whether the 1–2 Hz effect in brain activity might still be attributable to behavioral twitches, we inserted an electrode in the thorax to record movement (van Swinderen et al., 2004). Comparison of raw traces of brain and thoracic activity in radish mutants revealed that the 1–2 Hz brain oscillations were often sinusoidal, whereas the thoracic potentials were mostly bursts of spike-like activity, not necessarily synchronized to the brain events (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). The source of the 1–2 Hz oscillation remains unclear; it may be widespread across the mutant's nervous system, as evidenced by a minor peak at similar frequency range from the thorax recording (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). A recent report has shown a similar ∼2 Hz flight effect in wild-type flies lacking flight experience (Hesselberg and Lehmann, 2009). Together with our data, this suggests that radish mutants might be behaving like inexperienced flies in tethered flight.

Brain recording paradigms: novelty detection

A more general explanation for the behavioral defects described so far for radish¹ would suggest that these mutants have an “attention deficit.” To address attention-like processes in the brain of radish mutants, we recorded LFPs while presenting competing visual stimuli of variable salience. As described above, introducing competing visual objects to radish mutants in tethered flight resulted in a dramatically increased 1–2 Hz torque behavior, indicating that radish mutants are responding to the competing visuals by engaging in this strongly periodic behavior. Because the net consequence of this behavior is decreased time fixating either object (discussed above), this suggested an increased rate of perceptual alternations in radish mutants. We therefore proceeded to investigate brain responses to competing visual stimuli. Brain LFP activity in the 20–30 Hz range was shown previously to be associated with visual salience effects such as novelty, with wild-type flies showing a significant selection of visual novelty and simultaneous suppression of a competing non-novel object (van Swinderen and Greenspan, 2003; van Swinderen, 2007; van Swinderen et al., 2009). The learning and memory mutants dunce¹ and rutabaga²⁰⁸⁰ are defective in their brain responses to visual novelty, and this is associated with increased optomotor behavior in the maze (van Swinderen, 2007). We probed whether brain LFP responses were also compromised in radish mutants because radish¹ performed poorly in the optomotor maze, were more distractible, and attended more briefly to visual stimuli in tethered flight relative to wild type.

In our visual novelty paradigm, flies are exposed to two identical squares rotating 180° apart around the fly for 100 s, and then one of the squares is changed to a cross (Fig. 5C). To investigate brain responsiveness to visual novelty, we bandpass filtered the LFP signal at 20–30 Hz and correlated amplitude of that frequency band to either competing object on the rotating panorama (Fig. 5C) (for a visual explanation, see supplemental Methods 1, available at www.jneurosci.org as supplemental material). Wild-type flies display a sustained increase in 20–30 Hz activity associated with a novel visual object, often lasting three to four cycles of the rotating panorama (Fig. 5D,E). In contrast, radish mutants did not appear to display a sustained response to visual novelty in the brain-recording paradigm (Fig. 5F,G). Indeed, averaging data for multiple flies (n = 14) for the 10 s after a novelty transition suggested that radish mutants do not respond to visual novelty, for any LFP frequency domain between 10 and 40 Hz (Fig. 6A, left), whereas wild-type flies show a strong responsiveness to visual novelty (Fig. 6B, left) (van Swinderen, 2007).

Figure 6. Average responses to visual novelty in radish¹ and wild type. A, radish¹. Left, Average LFP activity for the 10 s after a novelty transition for three frequency domains (10–20, black Hz; 20–30 Hz, dark gray, 30–40 Hz, light gray) for radish mutants (n = 14 flies). Right, The same 20–30 Hz radish¹ data as summarized in the left but partitioned into successive 3 s epochs after a novelty transition (*p < 0.05, significant response) (supplemental Methods 1, available at www.jneurosci.org as supplemental material). B, Wild type. Left, Average LFP activity for the 10 s after a novelty transition was calculated for three frequency domains; n = 6 flies. The direction of panorama flow is indicated. Right, The same 20–30 Hz wild-type data as summarized in the left but partitioned into successive 3 s epochs after a novelty transition [*p < 0.05, significant response (data are from van Swinderen, 2007)]. C, Average LFP responses to each of the two visual objects presented individually to radish mutants (n = 14), for the three frequency domains indicated. Average wild-type 20–30 Hz responses to the same individual objects are shown for comparison (van Swinderen, 2007) (n = 6).

Closer observation of the LFP signal suggested differences in the temporal dynamics of 20–30 Hz responses in radish¹ compared with wild type; the mutant appears to respond more often at various times of each panorama rotation without the suppression effects characteristic of wild-type flies (Fig. 5E,G). We wondered whether radish mutants might fail to detect novelty because of such defective attention-like dynamics, as also suggested by our behavioral experiments. To address this, we partitioned our data into successive 3 s epochs (approximately one panorama rotation, e.g., the individual green wedges in Fig. 5D,E), instead of averaging the response for the entire 10 s after a novelty transition as before. When analyzed on this finer temporal scale, wild-type flies show a sustained response to novelty for three to four successive rotations (∼9–12 s) of the competing images (Fig. 6B, right; and Fig. 5D,E in a sample fly). In contrast, when we looked at radish¹ responses on this finer time resolution, we found that the mutant flies did indeed respond to novelty but that this image-locked response only lasted on average 3 s, or only one exposure of the competing objects (Fig. 6A, right). This confirms that radish mutants can see and discriminate the objects and suggests that the defect here lies not in the perception of visual stimuli but rather in the duration of perception before an internally generated switch occurs. Further confirming that the defect in this memory mutant is not at the level of simple visual processing, LFP responses to the same visual objects presented singly were not defective in radish mutants (Fig. 6C). Rather, it seems to be the introduction of salient visual competition that distracts the mutant sooner than wild type.

Brain recording paradigms: attention span

A central characteristic of attention is that it alternates among competing percepts at a characteristic tempo (Carter and Pettigrew, 2003; Brascamp et al., 2005). If the 20–30 Hz response to novelty reflects attention-like processes, then the abbreviated selective response to novelty in radish¹ may be indicative of a shortened attention span in this mutant. Furthermore, if 20–30 Hz responses to visual stimuli do indeed describe an attention-like process, then these should be continuously alternating among competing percepts even without experimentally imposed novelty. To study this possibility, and with a view to better understand radish¹ defects, we exposed flies to the same two distinct objects continuously, without novelty, while recording LFPs from their brains. We then applied some simple calculations to quantify alternation tempos in the brain LFP (supplemental Methods 2, available at www.jneurosci.org as supplemental material).

Wild-type flies and radish mutants were exposed to two objects in open loop (a square and a cross) simultaneously, as before (Fig. 5), but for extended epochs (400–800 s per fly) (Fig. 7). For each recording, 20–30 Hz power was summed for the consecutive 1.5 s periods when either object was in the frontal hemifield of the fly (Fig. 7A, black and gray bars). Thus, for every full 3 s rotation of the panorama, two 20–30 Hz values were calculated, one for either object. The log ratios of these two 20–30 Hz values are plotted for a sample wild-type fly in Figure 7B, for 60 successive cycles of the panorama rotation. The 20–30 Hz activity often appeared to alternate in power between objects: when power for one object was high, it was then low for the other object during that cycle (Fig. 7B). At times (or in the special case of salience relating to novelty, as above) such partitioning of 20–30 Hz power was biased toward one object for several successive cycles before alternating. Timings (in rotation cycles) between such alternations, termed here as AT (alteration time), were tallied (Fig. 7B, gray and black numbers) and visualized by a histogram (Fig. 7C, for the same sample fly). Noticing that wild-type AT data looked “clumped” (i.e., epochs when 20–30 Hz was biased to one or the other object for successive cycles; there are 5 such clumps in Fig. 7C) (supplemental Methods, available at www.jneurosci.org as supplemental material), we quantified this LFP bias (labeled Σ AT) in the fly brain activity by summing the duration (in total rotation cycles) of each AT clump (Fig. 7C, corresponding numbers in the right column). Combined data from eight wild-type flies revealed a frequency histogram with a characteristic shape resembling a gamma distribution (Fig. 8A).

Figure 7. Attention-like bias. A, Opposing visual stimuli (a square and a cross, 180° apart) rotate around the fly at 3 s per cycle. Each object is thus in front of the fly for 1.5 s, for which summed 20–30 Hz activity is separately calculated (during the epochs symbolized by the black and gray bars). B, Log ratio of summed 20–30 Hz activity plotted for successive cycles of image rotation in a sample wild-type fly. AT, Alteration time, or the duration (in cycles) when the ratio is biased in succession for one of the objects before alternating, indicated numerically above the graph. C, Successive AT values plotted as a time series histogram in the same sample wild-type fly. The size of five contiguous AT groupings (supplemental Methods 2, available at www.jneurosci.org as supplemental material) is tallied in a column on the right (Σ AT). D, Shuffled data from the same wild-type fly, replotted as a time series histogram with five tallied AT groups (Σ AT) shown on the right. E, The same analysis performed on data from the radish¹ mutant.

Figure 8. Distribution of attention-like bias. A, Frequency distribution of AT data for eight wild-type flies exposed to the two competing objects, as in Figure 7. Σ AT, A measure of attention-like bias, is the sum of AT values within each grouping (as indicated in Fig. 7C–E). B, Frequency distribution for eight sets of shuffled wild-type data. C, Frequency distribution for summed AT data from 14 radish flies.

To determine the significance of this distribution of wild-type LFP alternation dynamics in response to two competing visual objects, we contrasted our experimental data to shuffled data derived from the same experiments. In this way, any temporal relation between 20–30 Hz values measured for successive cycles of image presentation would be lost. Shuffling of the data indeed revealed a loss of clumping, or fewer contiguous epochs of biased 20–30 Hz activity (Fig. 7D). Tallying this bias (Σ AT) by a frequency histogram revealed a significantly different distribution compared with nonshuffled data (p < 0.05, Kolmogorov–Smirnov test for distributions), which was more characteristic of a binomial distribution (Fig. 8B).

Finally, we applied the same methodology to radish¹ and found that these mutants displayed brain LFP alternation statistics that were not significantly different from shuffled wild-type data (Figs. 7E, 8C) (p = 0.85 by the Kolmogorov–Smirnov test for distributions). This suggests that radish¹ brain activity is alternating randomly between the two visual percepts (the cross and the square), without the sustained “attention” across successive cycles characteristic of wild-type flies. Curiously, heat-shock of hs–rsh(161) failed to rescue this alternation defect in radish mutants but did result in improved novelty detection (supplemental Fig. 5, available at www.jneurosci.org as supplemental material).

MPH effects

Given that we found a deficit in attention-like processes and a well defined behavioral hyperactivity in the radish mutant flies, we wondered whether drugs commonly used in humans to treat similar symptoms in perceptual timing and hyperactivity could also rescue analogous defects in flies. In humans, amphetamines are known to affect attention in a variety of ways, depending on dosage and the specific drug used. In previous studies, we have shown that methamphetamines can increase optomotor responsiveness (van Swinderen and Flores, 2007) as well as modulate LFP responses to visual stimuli (Andretic et al., 2005). Here, we asked specifically whether a related drug, MPH (“Ritalin”), could rescue radish¹ defects in the optomotor maze and recording arena. MPH treatment, targeting the dopaminergic system (Iversen and Iversen, 2007), is typically administered to patients with attention deficit and hyperactivity disorder (ADHD). To first test the effect of MPH on wild-type flies, we fed starved animals 0.5 mg/ml of drug mixed in fly food. Testing in the optomotor maze after feeding resulted in an increased optomotor response, but this became significant only after 3 h since the initial feeding (Fig. 9A, 3 h). Such an increase in responsiveness is consistent with previous results for acute methamphetamine treatment in flies (van Swinderen and Flores, 2007) and also consistent with the time course of MPH treatment in humans (Pelham et al., 1999). A full day of chronic MPH exposure further increased the wild-type response (Fig. 9A, 24 h), although flies appeared to be slower and lethargic (data not shown). For this population exposed to a chronic treatment of MPH, only 1–2 h off the drug and onto regular food completely reset optomotor responsiveness to normal levels (Fig. 9A, Recovery), suggesting that effects on the brain were transient. We decided to focus on acute MPH treatment (2–4 h) to test for phenotypic rescue in radish mutants exposed to black/green gratings moving at 3 Hz (where radish¹ showed no responsiveness). Remarkably, MPH treatment significantly increased radish¹ optomotor performance from zero to wild-type levels (Fig. 9B, radish). When we explored the generality of this effect (increased optomotor responsiveness after drug treatment), we were surprised to find that acute MPH exposure did not alter optomotor responsiveness in a group of other learning and memory mutants (Fig. 9B), including amnesiac¹, rutabaga²⁰⁸⁰, and dunce¹.

Figure 9. MPH treatment. A, Optomotor responsiveness in wild-type flies treated with 0.5 mg/ml MPH (red histograms). Starved flies were transferred to drug-laced food and allowed to feed for 10 min, 3 h, or 24 h. Control flies were similarly transferred but to food without drug and tested 10 min later (Control1) or the next day (Control2). Flies chronically exposed to drug for 24 h were transferred back to normal food for 1–2 h and tested (Recovery). *p < 0.05, significantly different from controls by t test. n = 8 runs of 25–30 flies for each experiment. B, MPH (0.5 mg/ml) was administered acutely (2–4 h feeding on drug-laced food) to a panel of learning and memory mutants, as well as to flies with altered dopamine function [Th–Gal4 (Friggi-Grelin et al., 2003) × UAS–tnt, wherein dopaminergic neurons are silenced, or × UAS–eag^Δ932, in which dopaminergic neurons are activated]. *p < 0.05, significantly different from controls (red vs blue bars) by t test. n = 8 runs of 25–30 flies for each experiment. C, Spectral analysis (0–5 Hz) of LFP activity in the brains of radish mutants treated to acute MPH (red line). Blue line, The same flies before treatment. Data are averages of z-scored spectrograms (n = 5 flies). *p < 0.05, significantly different within 0.3 Hz band range surrounding the peak, by t test. D, The effect of MPH on 20–30 Hz responsiveness (±SEM) to visual novelty in radish¹ and wild-type flies. MPH feeding (red box) resulted in a significant novelty response in radish mutants (n = 5 flies), whereas radish mutants fed without the drug (blue box) show no response to novelty (n = 14). Wild-type responses to visual novelty were similar with and without MPH (n = 4 and 6 flies, respectively). *p < 0.05 by t test of means for either competing object. A visual explanation of how these data were calculated is in supplemental Methods 1 (available at www.jneurosci.org as supplemental material).

MPH is believed to affect dopamine signaling by targeting recycling of the neurotransmitter at the synapse (Iversen and Iversen, 2007). To test the involvement of dopamine in MPH-mediated optomotor effects, we treated two classes of flies with genetically altered dopamine function (Friggi-Grelin et al., 2003): Th–Gal4/UAS–tnt (in which dopaminergic neurons have been constitutively silenced) and Th–Gal4/UAS–eag [in which excitability of dopaminergic neurons was, in principle, increased (Broughton et al., 2004)]. Previous work has shown that either manipulation abolished optomotor responsiveness under our standard maze conditions, suggesting the requirement of a balanced dopamine environment for optomotor performance in the maze (van Swinderen and Flores, 2007). If MPH modulates dopamine levels in flies, then the presumed absence of dopamine in Th–Gal4/UAS–tnt mutants should block the effects of the drug. Indeed, the drug failed to rescue optomotor responsiveness in Th–Gal4/UAS–tnt flies (Fig. 9B). In contrast, Th–Gal4/UAS–eag flies, in which dopamine function is altered but presumably not absent, showed increased optomotor responsiveness after drug treatment, like wild type and radish¹.

We proceeded to investigate brain responses in radish mutants treated with acute MPH. A spectral analysis of LFP activity in the brain revealed that the 1–2 Hz oscillation characteristic of radish mutants was gone in animals treated with the drug compared with the same animals before treatment (Fig. 9C) (supplemental Fig. 6, available at www.jneurosci.org as supplemental material) (n = 5; p < 0.05 by paired t test comparing the average 1.4–1.7 Hz frequency domain before and after treatment). We then addressed effects on selective attention more specifically by means of our visual competition paradigm (as in Fig. 7). MPH did not significantly change the ongoing attention span defect of radish mutants, with alternations still appearing random in our sample (supplemental Fig. 5, available at www.jneurosci.org as supplemental material). However, when we introduced novelty salience into this paradigm (as in Figs. 5, 6), MPH exposure improved the 20–30 Hz response in radish mutants by producing an increase in power assigned to the novel stimulus (selection) and a decrease in power to the non-novel stimulus (suppression), resembling wild-type responsiveness to visual novelty in both amplitude and temporal dynamics (Fig. 9D) (supplemental Fig. 6, available at www.jneurosci.org as supplemental material). Acute exposure to MPH in wild-type flies did not further increase the LFP effect (Fig. 9D).