15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

Serotonergic mediation of orienting and defensive responses in zebrafish Serotonin (5-HT) is involved in arousal and defensive responses, both of which represent modulators of attention and orienting. Orienting responses (ORs), considered as a unit of attentional processing, are elicited by novel innocuous stimuli, differ from defense reflexes, as the latter are elicited by painful and threatening stimuli. Under situations of potential or distal threat, organisms can enter into a state of hypervigilance, in which innocuous stimuli can elicit defense reflexes instead of orienting responses. When zebrafish (Danio rerio) are exposed to a conspecific alarm substance (CAS), for example, innocuous visual stimuli elicit a defense reflex (DR). We treated zebrafish with either the serotonin synthesis inhibitor para-chlorophenylalanine (pCPA) or with the serotonergic precursor 5-HTP, and exposed individuals to visual stimuli to elicit either ORs or DRs. pCPA blocked CAS-elicited OR-to-DR shift, as shown both by the shift from an approach to an escape response when the stimulus was turned “ON” and by the changes in directional focus. While pCPA did not alter ORs in animals which were not exposed to CAS, 5-HTP abolished those responses. However, 5-HTP also attenuated the CAS-elicited OR-to-DR shift, suggesting that the serotonergic tone is more important in regulating the general arousal levels in zebrafish.

Orienting responses (ORs), originally described by Ivan Pavlov in the beginning of the 20th century as the “what is that?” reflex, are elicited by novel innocuous stimuli and can be considered a the unit of attentional processing (Barry, 2009). The OR is a whole-body reflex, involving adjustments in a large range of physiological systems, including visuomotor reflexes, electrodermal activity, respiratory and cardiovascular changes, and changes in whole-brain activity (Sokolov, 1960, 1963a, 1963b). While innocuous stimuli elicit ORs, painful and threatening stimuli elicit a separate class of responses, the defense reflex (DR; Sokolov, 1960, 1963a, 1963b). In spite of that, situations of potential or distal threat can lead to hypervigilance, in which innocuous stimuli can lead to DRs. For example, in zebrafish (Danio rerio), exposure to an alarm substance – an olfactory stimulus that represents a distal threat – leads to a shift from ORs to DRs elicited by an innocuous visual stimulus (do Nascimento et al., 2020).

Arousal plays an important role in this OR-to-DR shift. The arousal level of an organism can serve as an “amplifying factor”, decreasing habituation (Sokolov, 1963a), or, at higher levels, lead to “false alarms” in a state of hypervigilance. The neural bases of this OR-to-DR shift probably involve the overlap between circuits for visual attention and defensive behavior. For example, in a preparation of eye-brain-spinal cord of lampreys (both Lampetra fluviatilis and Petromyzon marinus), recordings of the ventral roots were able to discriminate between a unilateral burst response corresponding to orientation of the head toward slowly expanding or moving stimuli, particularly within the anterior visual field, that corresponds to a fictive OR, and a unilateral or bilateral burst response triggering fictive avoidance in response to rapidly expanding looming stimuli or moving bars, corresponding to a fictive DR (Suzuki et al., 2019). These fictive responses were enhanced by pharmacological blockade of brainstem-projecting neurons in the deep layers of the optic tectum (OT)(Suzuki et al., 2019). Similarly, in goldfish (Carassius auratus), unilateral tectal ablation abolishes the OR towards the contralateral hemifield, while electrical stimulation of the OT evokes contraversive eye movements (Torres et al., 2005). Habituation of ORs appear to depend on telencephalic input in goldfish, as electrical stimulation of the dorsomedial (Dm) and dorsocentral (Dc) regions of the telencephalon – the first being homologous to the associative amygdaloid nuclei of mammals (Maximino et al., 2013) – evoke cardiac and ventilatory responses, as well as bodily movements consistent with ORs (Quick & Laming, 1988). Likewise, telencephalic ablation decreased OR habituation in goldfish (Rooney & Laming, 1988), and animals with lesions in supracommissural (Vs) and postcommissural (Vp) regions of the telencephalon showed DRs to innocuous stimuli instead of ORs (Rooney & Laming, 1986). Thus, in fish, regions of the telencephalon that are part of the aversive brain system (do Carmo Silva, Lima-Maximino, et al., 2018) project to regions of the OT that organize ORs and DRs, modulating whether the animal will emit orienting or defense responses.

Both the deeper layers of the OT that project to the brainstem and the telencephalic regions involved in ORs and DRs show extensive serotonergic projections in fish (Lillesaar et al., 2009; Yokogawa et al., 2012) and mammals (Mooney et al., 1996). In rats, electrical stimulation of the optic tectum elicits head movements at lower current intensities and running and jumping responses at higher intensities; concurrent administration of the 5-HT reuptake blocker fluoxetine or the 5-HT1A receptor antagonist WAY100,635 decreases these responses, and administration of 5-HT in the midbrain, but not in the frontal cortex, produced the same effect (Dringenberg et al., 2003) . In zebrafish larvae, a brief exposure to water flow elevates activity and sensitivity to behaviorally relevant visual cues that resemble an OR (Yokogawa et al., 2012). These effects are accompanied by increased activity in superior raphe (SR) neurons; genetic ablation of these cells abolished the increased sensitivity to visual cues (Yokogawa et al., 2012). Thus, serotonergic neurons of the SR appear to be involved in arousal-elicited amplifications in ORs in zebrafish.

We have previously shown that an innocuous visual stimulus elicits an orienting response zebrafish, characterized by approaching the stimulus and projecting the body towards it (do Nascimento et al., 2020). However, when animals were exposed to a conspecific alarm substance (CAS) – a mixture released in the water when club cells in the skin are damaged, signaling threat to conspecifics in a shoal (Maximino et al., 2019) –, this OR changes to a defense reflex, characterized by increased erratic swimming and escaping the stimulus. We have previously demonstrated a role for serotonin in other components of the alarm response: after exposure to CAS, extracellular 5-HT levels are elevated in zebrafish brains (Maximino et al., 2014) and monoamine oxidase activity is decreased (Lima‐Maximino et al., 2020). Moreover, some components of the alarm response could be blocked by treatment with para-chlorophenylalanine (pCPA), a false substrate at tryptophan hydroxylase which blocks serotonin synthesis, or metergoline, a non-selective antagonist at 5-HT receptors (Lima‐Maximino et al., 2020). The present work attempts to unravel the roles of 5-HT in this arousal-elicited OR-to-DR shift in adult zebrafish. In order to do that, animals were treated with either para-chlorophenylalanine (pCPA), a false substrate at tryptophan hydroxylase which blocks serotonin synthesis, or 5-hydroxytryptophan (5-HTP), a 5-HT precursor. It was hypothesized that pCPA-treated animals would display an attenuated effect of CAS, while 5-HTP would potentiate this effect.

Animals and Housing

48 adult wild-type zebrafish from the longfin phenotype were used in the pCPA experiments, and 40 animals were used in the 5-HTP experiments. The populations used are expected to better represent the natural populations in the wild, due to its heterogeneous genetic background (Parra et al., 2009; Speedie & Gerlai, 2008). Animals were bought from a commercial vendor (Belém/PA) and collectively maintained in 40 L tanks for at least two weeks before the onset of experiments. The animals were fed daily with fish flakes. The tanks were kept at constant temperature (28 °C), oxygenation, light cycle (14:10 LD photoperiod) and a pH of 7.0-8.0, according to standards of care for zebrafish (Lawrence, 2007). Animals were used for only one experiment to reduce interference from apparatus exposure. Potential suffering of animals was minimized by controlling for the aforementioned environmental variables. Furthermore, in the all experiments the animals used were handled, anesthetized and sacrificed according to the norms of the Brazilian Guideline for the Care and Use of Animals for Scientific and Didactical Purposes (Diretriz brasileira para o cuidado e a utilização de animais para fins científicos e didáticos - DBCA. Anexo I. Peixes mantidos em instalações de instituições de ensino ou pesquisa científica, 2017). Since, in all cases, animals were individually stressed and/or individually treated, experimental unit was a single animal, and therefore all sample sizes refer to number of animals. The experimental protocols were approved by UEPA’s IACUC under protocol 06/18.

For the exposure stage, each animal was transferred individually to a container (2 L) where, after 3 minutes of acclimatization, it was carefully exposed to 7 ml of alarm substance (CAS), extracted using a standardized protocol (do Carmo Silva, Rocha, et al., 2018). As a negative control, a group with the same amount of animals was exposed to the same volume of distilled water, according to the protocol of Lima-Maximino et al. (2020). The animals remained exposed for the entire session. All stages of the experiment were performed under constant white Gaussian noise, producing an average of 58 dB above the tank. Light levels above the tanks were measured using a handheld light meter, and ranged from 251 to 280 lumens (coefficient of variation = 3.399% between subjects).

Animals were randomly allocated to groups using a random number generator (http://www.jerrydallal.com/random/random_block_size_r.htm), with each subject randomized to a single treatment using random permuted blocks. For each experiment, animals were treated and tested in the order of allocation (i.e., randomly). In all experiments, experimenters and data analysts were blinded to drugs and treatment by using coded vials (with the same code used for randomization); blinding was removed only after data analysis. Experiments were always run between 08:00AM and 02:00 PM. After experiments, animals were sacrificed by prolonged bath in ice-cold water (< 12 ºC), followed by spinal transection (Matthews & Varga, 2012).

A blue circle (R: 0; G: 0; B: 128) was used, based on the ability of zebrafish to discriminate blue from green stimuli in a simple discrimination paradigm (Mueller & Neuhauss, 2012). Stimulus size was 72º of horizontal visual angle, calculated considering the diameter of the stimulus in the screen when it is at the from of the animal, and the animal is in the centre of the tank. The size was based on previous work which show that this size is insufficient to produce an escape response in adult zebrafish that were not exposed to CAS (do Nascimento et al., 2020).

Animals were transferred to the tank and left to acclimate for 3 min before the presentation of the stimuli. The computer screen was turned on for the entire experiment, including acclimatization time. Using an animation based on LibreOffice Impress (v. 6.0.7.3), the stimulus was presented in 1-min intervals (“trials”) interspersed with 1-min stimulus-free periods. In trials in which the stimulus was “on”, the stimulus was presented for the entire duration of the trial. 10 trials were made; total session length was 22 minutes per animal. The number of trials and intertrial durations were chosen based on previous work on arousal in goldfish (Laming & Savage, 1980)

A video camera (Sony Handycam® DCR-DVD 610) was positioned above the tanks to allow recording and offline behavioral scoring. All focal fish behaviors were therefore tracked from a top-down perspective, using both )

TheRealFishTracker and fishtracker (https://github.com/joseaccruz/fishtracker ). For each behavioral video, a 2D region (arena) was defined for tracking, comprising the inner area of the tank. A region-of-interest (ROI) of 13 x 8.5 cm

was determined as “stimulus zone”. The following variables were recorded, using TheRealFishTracker:   

Total time in the stimulus zone (s);

Significant main effects of trial type, treatment, and drug were found for the time near the stimulus (Table 1; Figure 1A). In general, animals spent almost half of the each 1-min. trial near the stimulus when it was turned on (24.07 ± 15.78 s with stimulus “ON” vs. 12.54 ± 8.65 s with stimulus “OFF” [mean ± S.D.]). Significant interactions were also found between drug and treatment, drug and trial type, treatment and trial type, and a three-way interaction between all variables (Table 1). Post-hoc Tukey’s HSD tests found that the approach response to the stimulus was abolished in the presence of CAS (d = 1.13, p < 0.001); in fact, it is apparent that the approach response is turned into an escape response in the presence of CAS, as individuals in this condition spend less time near the stimulus when it was turned “ON” than when it was turned “OFF”. pCPA abolished this effect, as individuals treat with the drug and exposed to CAS spent more time near the stimulus when it was turned “ON” than when it was turned “OFF” (d = -4.81, p < 0.001).

Significant main effects of treatment, and drug were found for the erratic swimming (Table 1; Figure 1B). In general, animals exposed to CAS showed higher erratic swimming than nonexposed individuals at both trial types (20.87º ± 6.66 for controls vs. 26.49º ± 9.51 for CASexposed animals), producing a medium-sized effect (d = - 0.75, p < 0.001). A significant interaction between drug and treatment was also found (Table 1). Vehicle-treated, CAS-exposed animals showed increased erratic swimming in relation to controls (d = -1.56, p < 0.001), while pCPAtreated animals did not show differences in relation to controls when not exposed to CAS (d = 0.12, p = 0.59), and showed lower erratic swimming in relation to vehicle-treated, CAS-exposed animals (d = 1.44, p < 0.001).

Significant effects of treatment and drug were found for swimming speed, but interaction effects were absent (Table 1, Figure 1C). Treatment with pCPA generally decreased swimming speed (d = 0.17, p = 0.01), while exposure to CAS increased it (d = -2.85, p = 0.004).

By examining the lower and upper bounds of the 95% highest posterior density interval (HPD) in Table 2, it is apparent that the HPD intervals of the circular mean of all conditions overlap, and thus we conclude that there is not enough evidence to reject the null hypothesis that the circular means do not differ and that there is no effect of treatment, drug, or trial type on the average directional angle (Figure 1D).

Significant main effects of trial type and treatment were found for directional focus (Table 1, Figure 1D). Significant treatment X drug, trial type X drug, trial type X treatment, and three-way interactions were also found (Table 1). In general, directional focus increased when the stimulus was turned on (d = -0.5, p = 0.017). When vehicle-injected animals were exposed to CAS, directional focus was negative when the stimulus was turned “ON” (d = 1.7, p = 0.002 vs. stimulus “OFF”; d = 3.09, p < 0.00 vs. CTRL). Treatment with pCPA abolished this effect.

Finally, significant associations between eye use and the independent variables were found (χ²[df = 14] = 85.033, p < 0.0001; Figure 1E). No association was found when the stimulus was turned off, but significant associations were found when the stimulus was turned on. In these cases, control-exposed individuals treated with vehicle used more the left eye, and control-exposed individuals treated with pCPA used more the right eye; CAS-exposed individuals treated with vehicle showed no eye preference, and CAS-exposed individuals treated with pCPA used more the left eye.

Main effects of trial type, treatment, and drug were found for the time spent near the stimulus zone in Experiment 2 (Table 3, Fig 1A). Interactions between stimulus type and treatment, drug and treatment, and three-way interactions were found. In general, CAS-exposed animals treated with vehicle showed decreased time in the stimulus area when the stimulus was turned “ON” (d = 0.6, p = 0.003), while control animals treated with vehicle showed more time in the stimulus area when the stimulus was turned “ON” (d = -0.93, p < 0.001). This effect was attenuated by treatment with 5-HTP (d =-0.4, p = 0.356 vs. CAS-exposed animals with stimulus “ON”); however, 5-HTP also appeared to abolish the OR in animals which were not exposed to CAS (d = 0.18, p = 0.93, CTRL animals treated with 5-HTP, stimulus “OFF” vs. “ON”).

Main effects of treatment, but not drug nor trial type, were found for erratic swimming (Table 3, Figure 2B). Drug X treatment interactions were also found. In general, CAS-exposed animals had lower erratic swimming than non-exposed animals (d = 0.34, p < 0.001), an effect that appears to be mainly driven by the higher levels of erratic swimming in non-exposed, 5-HTPtreated animals regardless of stimulus type.

Significant main effects of drug and treatment were found for swimming speed (Table 3, Figure 2C). Significant interactions were also found between drug and treatment. 5-HTP-treated animals showed consistently lower swimming speeds than all other groups, regardless of stimulus type.

Main effects of trial type and drug were found for directional focus (Table 3), and interactions between trial type and drug and trial type and treatment were also found. By examining the lower and upper bounds of the 95% highest posterior density interval (HPD) in Table 3, no overlap in the HPD intervals of the circular mean of “ON” and “OFF” trials in CTRL animals (both treated with vehicle and 5-HTP), and thus we conclude that, for animals not exposed to CAS, the average directional angles were different in these conditions (Figure 2D).

Finally, significant associations between eye use and the independent variables were found (χ²[df = 14] = 23.87, p < 0.0001). As can be inferred from the association plot in Figure 2E, no association was found when the stimulus was turned off, but significant associations were found when the stimulus was turned on. In these cases, control-exposed individuals treated with vehicle used more the left eye, and control-exposed individuals treated with 5-HTP lost this preference; CAS-exposed individuals treated with vehicle showed no eye preference, and CAS-exposed individuals treated with 5-HTP used more the right eye.

The present work tested the participation of serotonin on orienting responses of zebrafish, as well as in the shift from orienting response to defense reaction (OR-to-DR shift) elicited by a conspecific alarm substance (CAS). pCPA blocked CAS-elicited OR-to-DR shift, as shown both by the shift from an approach to an escape response when the stimulus was turned “ON” and by the changes in directional focus. pCPA had no effects on erratic swimming; importantly, erratic swimming was not significantly altered by trial type, only by CAS, suggesting again that this stimulus induces an aversive state that is unrelated to visual stimulation, while simultaneously leading to the OR-to-DR shift. 5-HTP, on the other hand, attenuated the CAS-elicited OR-to-DR shift and abolished ORs in non-exposed animals, suggesting that the serotonergic tone is more important in regulating the general arousal levels. These results are consistent with the prediction that pCPA would attenuate the OR-to-DR shift, but not the prediction that 5-HTP would potentiate it.

The OR observed in the present experiments essentially replicates previous findings from our group (do Nascimento et al., 2020). The OR was characterized by an approach response (more time spent near the stimulus when it was turned “on”) and a body oriented towards the stimulus (higher directional focus). We also replicated our previous observation that CAS elicits an OR-toDR shift in zebrafish (do Nascimento et al., 2020). This shift was characterized by avoiding the stimulus zone when the visual stimulus was turned “off” and reduced directional focus. These effects are consistent with a “defensive reaction”, with CAS leading to an attentional bias in which the animal “misinterprets” the innocuous visual stimulus as a threatening stimulus. Evidence for an aversive state elicited by CAS is also present at the increased, stimulus-independent elevation in erratic swimming.

Importantly, we found that pCPA abolished the OR-to-DR shift in CAS-exposed animals. This effect was also accompanied by the abolishing of the effect of CAS on erratic swimming, suggesting that, while the serotonergic tone is permissive to ORs elicited by visual non-threatening stimuli, it is also necessary to mount a defensive response. This is consistent with previous findings with zebrafish, in which pCPA blocked the effects of CAS on responses to distal and proximal threat (Lima‐Maximino et al., 2020). The OR-attenuating effect of pCPA was also observed in rats, in which the drug significantly reduced the responding to the first presentation of a tone, but did not alter the rate of habituation (File, 1975). On the other hand, pCPA has been shown to enhance startle responses in rats (Carlton & Advokat, 1973), suggesting that the role of serotonin in orienting and defensive responses can be opposite.

Surprisingly, pCPA affected eye use during the presentation of the stimulus. Our results replicate previous results which show preference for the left eye in control animals during the presentation of the stimulus, while CAS-exposed animals showed no preference (do Nascimento et al., 2020). Moreover, we found that non-exposed individuals treated with pCPA showed a preference for the right eye when viewing the stimulus, while non-exposed animals treated with 5

HTP lost preference. When exposed to CAS, pCPA-treated animals shifted preference to the left eye, while 5-HTP-treated animals used more the right eye. In zebrafish, novel objects were investigated mainly with right frontal field, while in a second trial left frontal viewing tended to be used instead (Miklósi et al., 1997; Miklósi & Andrew, 1999). This is unlikely to be due to asymmetry in the habenular nuclei, which has been linked to preference in eye use (Dadda et al., 2010), because habenular projections to zebrafish raphe are not asymmetrical (Amo et al., 2010). Miklósi et al. (1997) suggested that the right eye is used when it is necessary to inhibit premature response in conditions of decision-making, and the left is used when it is necessary to track familiar objects. While we were not able to ascertain whether eye use changed during trials, it is possible that serotonin helps to track novelty, which could explain the effects of pCPA and 5-HTP on this variable.

A more thorough investigation of the specific mechanisms by which phasic and tonic serotonin modulate arousal and defensive responses to non-threatening visual stimuli are still needed. Given the high number of different serotonin receptors and the wide distribution of serotonergic innervation across the brain, the specific participation of receptors and brain regions is far from known. Nonetheless, these results point to an important serotonergic regulation of an emotion-cognition interface in vertebrates, which could represent a mechanism through which individuals respond to threats. Abril-de-Abreu, R., Cruz, J., & Oliveira, R. F. (2015). Social Eavesdropping in Zebrafish: Tuning of