Affective communication in rodents: ultrasonic vocalizations as a tool for research on emotion and motivation
Mice and rats emit and perceive calls in the ultrasonic range, i.e., above the human hearing threshold of about 20 kHz: so-called ultrasonic vocalizations (USV). Juvenile and adult rats emit 22-kHz USV in aversive situations, such as predator exposure and fighting or during drug withdrawal, whereas 50-kHz USV occur in appetitive situations, such as rough-and-tumble play and mating or in response to drugs of abuse, e.g., amphetamine. Aversive 22-kHz USV and appetitive 50-kHz USV serve distinct communicative functions. Whereas 22-kHz USV induce freezing behavior in the receiver, 50-kHz USV lead to social approach behavior. These opposite behavioral responses are paralleled by distinct patterns of brain activation. Freezing behavior in response to 22-kHz USV is paralleled by increased neuronal activity in brain areas regulating fear and anxiety, such as the amygdala and periaqueductal gray, whereas social approach behavior elicited by 50-kHz USV is accompanied by reduced activity levels in the amygdala but enhanced activity in the nucleus accumbens, a brain area implicated in reward processing. These opposing behavioral responses, together with distinct patterns of brain activation, particularly the bidirectional tonic activation or deactivation of the amygdala elicited by 22-kHz and 50-kHz USV, respectively, concur with a wealth of behavioral and neuroimaging studies in humans involving emotionally salient stimuli, such as fearful and happy facial expressions. Affective ultrasonic communication therefore offers a translational tool for studying the neurobiology underlying socio-affective communication. This is particularly relevant for rodent models of neurodevelopmental disorders characterized by social and communication deficits, such as autism and schizophrenia.
KeywordsAnxiety Fear Autism Social behavior Preparedness
Mice and rats emit and perceive calls in the ultrasonic range, i.e., above the human hearing threshold of about 20 kHz: so-called ultrasonic vocalizations (USV). This has important implications for preclinical research, ranging from practical considerations, e.g., ambient ultrasonic noise that might affect the well-being of the animal, to new approaches for the assessment of psychological states and functions in rodents that are not readily accessible with the standard paradigms in use today, e.g., USV as a measure for communication deficits in rodent models of human neuropsychiatric disorders, such as autism.
High rates of USV occur throughout the life span in a variety of socially relevant situations. They are emitted by pups during the first days of life when removed from the nest, in juveniles engaging in social play behavior, in adult females during social investigation and in adult males when exposed to females or during aggression. Moreover, we know that various USV categories exist with distinct acoustic features. Some USV categories are present in both mice and rats, whereas others are solely emitted by one species.
Affective ultrasonic communication in rats
Aversive 22-kHz USV
Communicative function: behavioral responses
Further support for an alarm function of 22-kHz USV was obtained in playback experiments. A number of studies showed that playback of natural 22-kHz USV or 20 kHz sine wave tones activate the fight/flight/freeze system of rats. The studies carried out so far indicate that the behavioral responses elicited by such ultrasonic stimuli are strain-dependent. Specifically, in response to playback, Wistar rats (Brudzynski and Chiu 1995; Burman et al. 2007; Commissaris et al. 2000; Neophytou et al. 2000; Nobre and Brandão 2004; Sales 1991; Wöhr and Schwarting 2007) and Sprague Dawley rats (Endres et al. 2007) have been found to show a reduction in locomotor activity and a limited freezing response (behavioral inhibition), whereas Lister hooded rats display bursts of running and jumping (behavioral excitation), which is characteristic of an active defensive response (Beckett et al. 1996, 1997; Commissaris et al. 1998, 2000; Finn et al. 2004; Neophytou et al. 2000; Nicolas et al. 2007; Voits et al. 1999).
However, we should emphasize that evidence in favor of strong behavioral changes occurring specifically in response to 22-kHz USV is weak. Indeed, clear behavioral responses are only seen in studies involving loud and artificial continuous sine wave tones (Beckett et al. 1996, 1997; Commissaris et al. 1998, 2000; Finn et al. 2004; Nicolas et al. 2007; Neophytou et al. 2000; Nobre and Brandão 2004; Voits et al. 1999). Conversely, mostly weak effects have been observed in studies in which natural stimuli have been used (Brudzynski and Chiu 1995; Burman et al. 2007; Endres et al. 2007; Sales 1991; Wöhr and Schwarting 2007). As early as the first study with natural 22-kHz USV as playback stimuli, Sales (1991) reported only modest locomotor inhibition. Specifically, rats entered floor squares less often during 22-kHz USV playback (ca. 55 times) than when noise was presented (ca. 70 times). However, the level of locomotor inhibition reported in response to 22-kHz USV was similar to that seen when rats were exposed to an artificial 38 kHz stimulus. Moreover, Brudzynski and Chiu (1995), who performed a similar experiment some few years later, observed no acute effects during playback (with ca. 500 activity counts per observation before and during playback). They only reported a slight but significant, decrease in locomotor activity after playback (with ca. 350 activity counts). Similarly, three more recent studies also observed only weak behavioral effects of 22-kHz USV playback. In a study performed by Burman et al. (2007), just one out of two natural 22-kHz stimuli was effective in increasing the latency to emerge from the test box; the other was without any effect. Moreover, Endres et al. (2007) observed only a non-significant and modest increase in the time rats spent freezing when exposed to 22-kHz USV (ca. 25% versus ca. 8% during silence). As in the study by Sales (1991), however, the observed response was not specific for 22-kHz USV. Endres et al. (2007) compared the behavioral change induced by playback of 22-kHz USV with those seen in response to various other stimuli, namely 50-kHz USV, 22-kHz sine wave tones, 22-kHz USV shifted to about 45 kHz and white noise in the range from 17 to 27 kHz. Comparisons between responses to 22-kHz USV and all the other stimuli demonstrated no specific response of naive rats to 22-kHz USV playback. Only when the various stimuli with acoustic characteristics close to 22-kHz were pooled together was a moderate increase in freezing observed during and after stimulus presentation. Such an induction of freezing was not seen in the other pooled groups. Similarly, we observed only a non-significant and modest decrease in the distance travelled during playback of 22-kHz USV (ca. 10 cm/min) as compared with silence (ca. 40 cm/min; Wöhr and Schwarting 2007). Finally, some studies did not detect any behavioral response to playback of natural 22-kHz USV (Bang et al. 2008; Lindquist et al. 2004; Parsana et al. 2012a, 2012b; Sadananda et al. 2008).
Moreover, in playback studies with artificial stimuli, rats were found to show a stronger behavioral response to 7 kHz or 12 kHz sine wave tones than to 20 kHz sine wave tones (Commissaris et al. 2000), indicating that the behavioral effects were clearly not related to the communicative value of the 22-kHz USV. More likely, these effects were caused by the high sound pressure levels (SPL) used in the experiments. In some studies, artificial stimuli have been presented with more than 100 dB SPL (Commissaris et al. 2000; Voits et al. 1999), which is much higher than the usual SPL of natural 22-kHz USV, with approximately 60–80 dB SPL from a distance of 20–30 cm (Wöhr et al. 2005; Wöhr and Schwarting 2008a, 2008b). A remarkable exception, however, is a study by Nobre and Brandão (2004) in which ultrasonic sine wave tones with 75 dB SPL were used and freezing behavior was specifically seen in response to tones in the range of 20 to 25 kHz but not below or above.
As only weak (Brudzynski and Chiu 1995; Burman et al. 2007; Endres et al. 2007; Sales 1991; Wöhr and Schwarting 2007) or no behavioral responses (Bang et al. 2008; Lindquist et al. 2004; Parsana et al. 2012a, 2012b; Sadananda et al. 2008) were elicited by presenting natural 22-kHz USV as US, fear conditioning studies were performed, in which 22-kHz USV served as CS (Bang et al. 2008; Endres et al. 2007). Endres et al. (2007) addressed the questions of whether the recognition of 22-kHz USV as alarm calls can be learned and whether this learning is facilitated by a preparedness to acquire defensive behavioral patterns in response to such stimuli. They showed that rats quickly learn to associate an aversive event with 22-kHz USV, retain this information longer in their memory and are more reluctant to extinguish this memory than in the case of the association of aversive events with other types of ultrasonic stimuli, such as artificial 22-kHz sine wave tones. This indicates that “rats are predisposed to acquire adaptive defensive behavior in response to alarm calls” and that “better encoding of such learning in rats leads to a stable memory which better resists extinction” (Endres et al. 2007).
Evidence in support of a predisposition to associate 22-kHz USV with aversive events was also obtained by Bang et al. (2008) using a differential fear conditioning paradigm. Here, 22-kHz USV, 50-kHz USV and various artificial stimuli created to deconstruct the 22-kHz USV into simpler acoustic features, such as frequency changes, frequency and amplitude modulation and temporal patterning, were tested. During differential fear conditioning, one of these stimuli (CS+) always co-terminated with a foot-shock (US), whereas another (CS−) was explicitly unpaired with the US. As in the study by Endres et al. (2007), 22-kHz USV did not differ from the other ultrasonic stimuli in terms of the unconditional elicitation of freezing behavior but after pairing 22-kHz USV and foot shocks, 22-kHz USV induced freezing behavior, suggesting that freezing in response to 22-kHz USV is not innate but instead emerges as a consequence of associative learning. In contrast to the study by Endres et al. (2007), however, Bang et al. (2008) found that 22-kHz USV serving as the CS+ were no more effective than 50-kHz USV or the artificial stimuli for conditional freezing. However, in favor of a biological preparedness to associate 22-kHz USV with aversive events, an asymmetrical stimulus generalization was discovered. Specifically, when 22-kHz USV served as the CS+, less generalization of fear to the CS− was seen than when 22-kHz USV served as the CS−. Under the latter circumstance, rats failed to discriminate between CS+ and CS−. By contrast, they clearly did discriminate when 22-kHz USV served as the CS+. Thus, the amount of stimulus generalization (from CS+ to CS−) depended on which stimuli served as CS+ or CS−. Despite the differences between the findings reported by Endres et al. (2007) and Bang et al. (2008), both studies support the notion that the behavioral response to 22-kHz USV is not innate but rather emerges through associative learning, which, in turn, is facilitated by predisposition. The existence of biological preparedness to associate certain stimuli over others has previously been demonstrated, e.g., in taste aversion learning (Garcia and Koelling 1966; for review see Seligman 1970).
A different approach to testing the communicative function of 22-kHz USV was applied by Kim et al. (2010). They compared the following three groups of pair-housed rats to see whether the emission of 22-kHz USV by the sender induces freezing behavior in the receiver: (1) the sender underwent conventional fear conditioning, while the partner stayed in the home cage; (2) the sender underwent conventional fear conditioning, while the partner was exposed to an aversive stimulation, namely a series of foot shocks but not fear conditioning, i.e., no tone-shock pairings were applied; (3) the sender stayed in the home cage, while the partner was exposed to the aversive stimulation. The next day, the rat pairs were placed in a novel environment and the presentation of the tone used for fear conditioning elicited the emission of 22-kHz USV by the conditioned senders. The production of 22-kHz USV by the senders led to freezing behavior in the partners exposed to the series of foot shocks the day before but not in naive partners. Importantly, partner rats exposed to the aversive stimulation but tested with naive senders not emitting 22-kHz USV in response to the tone did not show freezing behavior, indicating that 22-kHz USV emission causes freezing behavior in experienced partner rats, rather than sensitization. This conclusion is further supported by the finding that the disruption of the primary auditory pathway by lesions of the medial geniculate nucleus of the thalamus effectively blocks the freezing response in the partner when being tested with a sender emitting 22-kHz USV. The lack of freezing behavior in the partners with medial geniculate nucleus lesions shows that the emission of 22-kHz USV by the sender and not olfactory cues or visible behavioral changes elicits the freezing response in the partner. However, probably the most important finding of the study by Kim et al. (2010) is that an aversive stimulation, i.e., a fear experience, prior to 22-kHz USV exposure, is necessary for inducing freezing behavior through 22-kHz USV. The fear experience appears to prime rats to show a freezing response to 22-kHz USV. Indeed, the onset latencies of the 22-kHz USV of the partner during fear experience correlated with their freezing responses when exposed to the 22-kHz USV emitted by the sender. This suggests auto-conditioning as the underlying mechanism. To test the auto-conditioning hypothesis, Kim et al. (2010) inactivated the medial geniculate nucleus of partner rats during their exposure to the aversive stimulation but not during testing with senders. Intriguingly, partner rats did not display freezing behavior in response to 22-kHz USV emitted by the sender, meaning that they behaved as naive rats, i.e., as if not exposed to the aversive stimulation. The auto-conditioning hypothesis was recently confirmed by Parsana et al. (2012b). Using the playback approach, they demonstrated that rats that underwent an aversive experience before displayed freezing behavior in response to 22-kHz USV, whereas no freezing response was observed in rats that did not undergo such an aversive experience. Effects of aversive experience were reported to be specific for 22-kHz USV, as experienced rats were unresponsive to 50-kHz USV. Again, 22-kHz USV production during the aversive stimulation was predictive of subsequent freezing to 22-kHz USV. Parsana et al. (2012b) therefore suggest that auto-conditioning “is sufficiently rapid, reliable, and stimulus-specific to serve an adaptive defensive function in rats”. Importantly, these last two studies by Kim et al. (2010) and Parsana et al. (2012b) probably explain the inconsistencies in the literature regarding the effects of 22-kHz USV playback on the behavioral responses of the recipients and highlight the relevance of controlling for the affective experiences of the rats in studies on the communicative functions of 22-kHz USV.
Communicative function: neuronal responses
Behavioral responses elicited by playback of natural 22-kHz USV or artificial 20-kHz sine wave tones are paralleled by the activation of brain areas regulating anxiety- and fear-related behaviors. In one of the earliest studies, Brandão et al. (2001) showed that 22-kHz sine wave tones and other fear-evoking stimulations, such as a light previously associated with foot shock application, produced an increase in the amplitude of auditory-evoked potentials in the inferior colliculus, indicating that 22-kHz USV can enhance auditory processing in aversive situations. Activation of the inferior colliculus is known to cause freezing behavior (Brandão et al. 2001).
Beckett et al. (1997) performed the first comprehensive study on brain activity patterns elicited by playback of artificial 20-kHz sine wave tones. By means of immunohistochemical assessment of the immediate early gene c-fos, a marker for neuronal activity, they demonstrated that locomotor hyperactivity in Lister hooded rats in response to artificial 20-kHz sine wave tones was associated with increased neuronal activity in the periaqueductal gray (PAG), amygdala, hypothalamus, and thalamus. Specifically, the dorsal but not ventral, part of the rostrocaudal PAG was activated. Within the amygdala, the medial, basolateral, central and lateral nuclei showed an increase in neuronal activity. Within the hypothalamus, activation was observed in the dorsomedial nucleus but not the anterior and ventromedial nuclei. Finally, the stria terminalis and the paraventricular nucleus of the thalamus were also activated. In a subsequent study that addressed the question of whether strain differences in behavioral responses to artificial 20-kHz sine wave tones were paralleled by differences in brain activity, the general pattern of neuronal activation was confirmed for Lister hooded rats (Neophytou et al. 2000). Compared with them, however, playback of 20-kHz sine wave tones led to relatively weak neuronal changes in Wistar rats. Increased activity was observed in the basolateral amygdala, stria terminalis and entorhinal cortex. Most importantly, the PAG activity pattern differed from that of Lister hooded rats and paralleled substantial behavioral differences. In Lister hooded rats, playback induced locomotor hyperactivity, which was associated with neuronal activity preferentially in the dorsal region of the rostral and caudal PAG, whereas in Wistar rats, locomotor hypoactivity and freezing responses were observed, accompanied by activation in the ventral region of the caudal PAG. The PAG presumably represents the final common pathway in the behavioral expression of aversive states (Vianna and Brandão 2003) and electrical or chemical stimulation of the dorsal part of the PAG has been shown to elicit fleeing, whereas stimulation of the ventral parts of the PAG produces freezing (Depaulis et al. 1994; Morgan et al. 1998). Accordingly, inactivation of the dorsal parts of the PAG increase fear-induced freezing, whereas inactivation of ventral parts disrupts this behavior (De Oca et al. 1998). In addition to the up-regulation of activity in brain areas implicated in the regulation of fear and anxiety, a reduction in activity has been observed in the paraventricular nucleus of the thalamus and the raphé nuclei in Wistar rats.
Using a similar experimental approach, we have examined the expression of the immediate early gene c-fos after exposure to natural 22-kHz USV (Sadananda et al. 2008). In our study, the overall immunohistochemical staining pattern was similar to that obtained in studies with artificial stimuli (Beckett et al. 1997; Neophytou et al. 2000). Specifically, an increase in neuronal activity induced by 22-kHz USV is observed in the PAG, with the highest activity levels in its rostral part (Sadananda et al. 2008). In addition to PAG, increased activation is seen in the amygdala, particularly in the lateral/basolateral part (Sadananda et al. 2008). Increased amygdala activity during playback of 22-kHz USV has also been reported in a study by Parsana et al. (2012a). By means of single unit recordings in freely behaving rats, Parsana et al. (2012a) have addressed the question of whether neurons in the amygdala are tuned to respond to 22-kHz USV as an ethologically important natural stimulus and have compared firing rates and patterns between 22-kHz USV and various controls, including 22-kHz sine wave tones. They have found that approximately 40% of lateral/basolateral amygdala neurons respond to 22-kHz USV. Two attributes of the firing patterns, namely whether playback causes an increase (+) or decrease (-) in firing rates and whether the induced change is transient (phasic) or sustained (tonic) have been differentiated, resulting in four elemental firing patterns: phasic onset (+), phasic offset (-) tonic (+), and tonic (-). Most firing responses elicited by 22-kHz USV are tonic (+) responses, meaning that 22-kHz USV leads to a sustained increase in firing. Changes in firing rates have also been observed in response to 22-kHz sine wave tones. As for 22-kHz USV, the most common change is a sustained increase in firing, a tonic (+) response. Latency analysis has shown that changes in firing occur after 10–20 ms, consistent with activation via the short subcortical pathway, as opposed to the long cortical pathway (Parsana et al. 2012a). The amygdala is a key structure in affective information processing and fear has been the function most closely associated with it (Fendt and Fanselow 1999; LeDoux 2000; Maren and Quirk 2004). The findings that 22-kHz USV lead to an increase in amygdala activation add 22-kHz USV to the group of motivationally relevant and negatively valenced stimuli capable of increasing amygdala activity, such as foot shock and restrain (Duncan et al. 1996; Kovács 1998). The finding that 22-kHz USV yields an increase in the lateral/basolateral but not in the central amygdala points to the functional importance of intra-amygdaloid circuits (Pitkänen et al. 1997). The lateral/basolateral part is generally considered as the sensory gateway into the amygdala, receiving inputs from all sensory systems, whereas the central amygdala is viewed as the output region. Convergent evidence from lesion and electrophysiological studies has been accumulated in support of the hypothesis that the lateral/basolateral amygdala is critically involved in fear conditioning, i.e., in the acquisition of fear-related CRs (Fendt and Fanselow 1999; LeDoux 2000; Maren and Quirk 2004). Typically, incoming novel stimuli elicit increased firing in the amygdala, which rapidly habituates as long as they are not accompanied by biologically significant stimuli. However, pairing an initially neutral stimulus (later becoming a CS) with a biologically significant stimulus (US) leads to changes in synaptic plasticity in the lateral/basolateral amygdala. Such plasticity changes result in modifications of the intra-amygdaloid circuits that allow the CS alone to flow through its lateral/basolateral part and to activate the central amygdala. The central amygdala then orchestrates responses in order to cope with the detected biologically significant event. In the case of threat, for instance, its output connections to the PAG induce freezing behavior. The lack of an increase in neuronal activity in the central amygdala in response to 22-kHz USV (Sadananda et al. 2008) is therefore in accordance with the observation that 22-kHz USV induce only a subtle amount of freezing (Brudzynski and Chiu 1995; Burman et al. 2007; Endres et al. 2007; Sales 1991; Wöhr and Schwarting 2007) or no freezing at all (Bang et al. 2008; Lindquist et al. 2004; Parsana et al. 2012a; Sadananda et al. 2008) in rats lacking aversive experience (Kim et al. 2010; Parsana et al. 2012b). However, the clear increase in neuronal activity in the lateral/basolateral part of the amygdala as assessed by c-fos (Sadananda et al. 2008) and single unit recordings (Parsana et al. 2012a) might reflect the initiation of the synaptic changes that underlie learning processes or provide prerequisites for it. This assumption is supported by recent findings showing that the infusion of the γ-aminobutyric acid agonist muscimol into the basolateral amygdala prior to fear conditioning impairs the acquisition of fear to 22-kHz USV (Allen et al. 2008).
Furthermore, neuronal activation is also evident in the perirhinal cortex (Sadananda et al. 2008), which is adjacent to and reciprocally connected with the amygdala (Pitkänen et al. 1997). Preliminary evidence that the perirhinal cortex is implicated in the processing of 22-kHz USV was obtained in a lesion study (Lindquist et al. 2004). Perirhinal lesions, performed prior to training, were found to severely impair delay fear conditioning to a 22-kHz USV or artificial 22-kHz USV-like stimuli such as CS, whereas such lesions were ineffective when the CS was a continuous tone of the same or a lower frequency (Kholodar-Smith et al. 2008a; Lindquist et al. 2004). Based on these findings, the discontinuous nature of 22-kHz USV, i.e., its “bout” structure, has been suggested to be at least part of the reason that normal fear conditioning to 22-kHz USV requires cortical processing, whereas cortical processing is not necessary for conditioning to continuous tones: “Cortical processing may be required to integrate these discontinuous auditory stimuli across time, in order for normal fear conditioning to occur” (Allen et al. 2007). In this context, we need to note that the perirhinal cortex is required for trace fear conditioning but not delay fear conditioning (Kholodar-Smith et al. 2008b). The difference between both paradigms is that, in the delay fear conditioning, the US is presented at the end of the CS, whereas in the trace fear conditioning, the CS is followed by a trace interval, which is terminated by the US. This indicates that the role of the perirhinal cortex in trace fear conditioning is distinct from its more perceptual functions in delay fear conditioning.
As in Parsana et al. (2012a) for the amygdala, single unit recordings were used to test whether neurons in the perirhinal cortex were also tuned to respond to 22-kHz USV (Allen et al. 2007). Firing patterns were assessed in response to 22-kHz USV and to several acoustic control stimuli, namely frequency and temporally matched discontinuous tones and continuous tones with the same or lower frequencies. A comparison of the number of neurons responding to the auditory stimuli revealed no difference between 22-kHz USV and control stimuli. Overall, approximately 40% of the neurons responded to one or more of the auditory stimulus types used, out of which 69% responded to 22-kHz USV. Most of the elicited firing patterns were phasic onset (+) responses. Indeed, discontinuous tones, among them natural 22-kHz USV, sometimes elicited temporally matched firing patterns, which consisted in a transient increase in the firing frequency that was triggered by the onset or, less often, by the offset of each of the successive tones or USV within a bout. Such temporally matched firing patterns occurred more often in the perirhinal cortex (Allen et al. 2007) than in the amygdala in which tonic (+) responses were most common (Parsana et al. 2012a). In line with the observed brain region-specific differences in elicited firing patterns, Parsana et al. (2012a) suggested that phasic (+) responses indicated the detection of an event, whereas tonic (+ or -) responses reflected its valence, i.e., aversive versus appetitive.
In a subsequent study, Furtak et al. (2007) used a classic fear conditioning paradigm in which 22-kHz USV or a continuous 22-kHz sine wave tone served as CS and examined fear-conditioning-induced changes in single unit firing elicited in the perirhinal cortex. Firing changes were observed in approximately 70% of the recorded units in response to 22-kHz USV or a continuous 22-kHz sine wave tone after the stimuli had been paired with a foot shock (US). Conditioning caused widespread changes in neuronal firing regardless of whether 22-kHz USV or a 22-kHz sine wave tone served as a cue. Remarkably, approximately 30% of units that were initially CS-unresponsive became CS-responsive after conditioning. Despite these general changes, however, two differences between single unit responses elicited by the 22-kHz USV and those elicited by the 22-kHz sine wave tone were evident. First, approximately 10% of the units recorded from the rat group, which was conditioned to 22-kHz USV, displayed a precisely timed increase in firing rate during the interval in which the US occurred during conditioning. This response pattern was unique to this group and was not seen in rats conditioned to a 22-kHz sine wave tone. Second, before conditioning, the neurons started firing to both CS after circa 55 ms. Following conditioning, however, neurons started firing in response to the 22-kHz sine wave tone as early as circa 25 ms, whereas conditioning to 22-kHz USV had no effect on the firing latency. Based on these findings, the authors suggested that firing in response to both CS was mediated by cortical rather than subcortical pathways to the perirhinal cortex before conditioning but that subcortical pathways gained the control of firing through conditioning to the 22-kHz sine wave tone but not to the 22-kHz USV (Furtak et al. 2007).
Overall, strong evidence has therefore been presented that 22-kHz USV and 22-kHz USV-like stimuli activate brain regions regulating anxiety- and fear-related behaviors, most notably the perirhinal cortex, amygdala and PAG. In line with the neuronal activation pattern, behavioral responses elicited by 20-kHz USV-like stimuli have been demonstrated to be efficiently blocked by anxiolytic compounds, including muscimol and midazolam (Beckett et al. 1996; Nicolas et al. 2007; Nobre and Brandão 2004). Neuronal response studies are further in agreement with the assumption of Endres et al. (2007) who have suggested a “neural template” that better encodes the USV of conspecifics than other auditory stimuli. This conclusion was based on their behavioral study in which rats quickly learned to associate an aversive event with 22-kHz USV, retained this information longer in their memory and were more reluctant to extinguish this memory than in the case of other types of ultrasonic stimuli. Thus, perirhinal cortex and amygdala probably at least part of this “neural template”, which is responsible for the predisposition observed when 22-kHz USV are used as CS.
Appetitive 50-kHz USV
Because 50-kHz USV occur in appetitive situations, such as rough-and-tumble play (e.g., Knutson et al. 1998; Webber et al. 2012) and mating (e.g., Sales 1972b; Thomas and Barfield 1985) or in response to drugs of abuse, e.g., amphetamine (e.g., Burgdorf et al. 2001; Thompson et al. 2006), they have been suggested to reflect a positive affective state akin to joy and happiness. This view is supported by a series of experiments conducted by Panksepp and Burgdorf (2000). Inspired by the observation that rats emit 50-kHz USV during rough-and-tumble play, they decided to mimic rough-and-tumble play in rats through a human experimenter by tickling and showed that it is possible to induce 50-kHz USV by hetero-specific play (Panksepp and Burgdorf 2000). Similar to rough-and-tumble play behavior, the rate of tickling-induced 50-kHz USV is enhanced by a short period of social isolation, indicating that high levels of social motivation are associated with high levels of 50-kHz USV (Panksepp and Burgdorf 2000, 2003). Indeed, Panksepp and Burgdorf (2003) further showed that, in particular, those rats that emit high numbers of 50-kHz USV experience the tickling procedure as appetitive, as indicated by short latencies to approach the hand of the human experimenter. In addition to the tickling itself, even the presentation of cues associated with it, such as the experimenter’s hand, are effective in eliciting 50-kHz USV (Panksepp and Burgdorf 2000, 2003). In contrast, aversive stimuli, such as cat odour or bright light, reduce 50-kHz USV emission (Panksepp and Burgdorf 2003). Panksepp (2005) therefore considers 50-kHz USV as a rat homolog or antecedent of human laughter.
Like 22-kHz USV emission, the production of 50-kHz USV is characterized by huge inter-individual differences, reflecting individual dispositions (Mällo et al. 2007; Schwarting et al. 2007). For instance, in our tickling experiments, we typically see a huge proportion of rats emitting 50-kHz USV, whereas others do not emit 50-kHz USV and some even emit 22-kHz USV (Schwarting et al. 2007). Similar findings have been reported by Mällo et al. (2007). They tickled rats repeatedly during a period of 45 days and found correlation coefficients between days mainly ranging between 0.6 and 0.9, indicating a highly stable trait. Surprisingly, however, the inter-individual differences in this trait were only weakly and often inconsistently associated with individual dispositions to show anxiety- and depression-like behavior, as assessed in standard paradigms, such as the elevated plus maze and forced swimming (Mällo et al. 2007; Schwarting et al. 2007). Nevertheless, some remarkable exceptions have been reported. For instance, Rygula et al. (2012) trained rats in an operant conditioning paradigm to press a lever in response to a tone to receive a reward, namely sucrose solution and to press another lever when a different tone was presented to avoid punishment by administration of a foot shock. After training, rats were tickled and split into two groups: (1) rats that produced high rates of 50-kHz USV during tickling and (2) rats that did not emit many tickling-induced 50-kHz USV. Then, both groups of rats were exposed to an ambiguous tone with a frequency intermediate between the two tones used during training and measurements were made regarding how often the rats pressed the food-rewarding lever versus the one that they learned to press to avoid foot shocks. Rygula et al. (2012) found that rats emitting many 50-kHz USV displayed an “optimistic” bias towards the food-rewarding lever in response to the ambiguous tone. They pressed the food-rewarding lever more often than rats that produced only a few tickling-induced 50-kHz USV.
However, little is known about biological factors underlying the observed individual disposition in 50-kHz USV emission. Out of the many factors potentially involved, we decided to study hippocampal cell proliferation, as the level of neurogenesis in the dentate gyrus of the hippocampus has been repeatedly associated with affect regulation. Thus, aversive stimuli, such as submission during intermale fighting, is known to reduce hippocampal cell proliferation (Czéh et al. 2007) and Santarelli et al. (2003) have demonstrated that hippocampal cell proliferation is necessary for the antidepressant effects of selective serotonin reuptake inhibitors. We therefore tickled rats and correlated tickling-induced 50-kHz USV with hippocampal cell proliferation. In agreement with its role in affect regulation, we found that the emission of 50-kHz USV was highly positively correlated with hippocampal cell proliferation, whereas a highly negative correlation between the emission of 22-kHz USV and hippocampal cell proliferation was evident. Remarkably, we further found that hippocampal cell proliferation was strongly elevated in rats that experienced the tickling procedure presumably as appetitive, as indicated by the high numbers of 50-kHz USV. In contrast, rats with low tickling-induced 50-kHz USV rates had levels comparable with those of non-tickled controls. This suggests that tickling induces hippocampal cell proliferation in those rats that experience tickling presumably as appetitive (Wöhr and Schwarting 2009). This finding has been replicated in a more recent study by Yamamuro et al. (2010), who have also shown that the effects are long-lasting and still detectable after a 3-week survival period.
Burgdorf et al. (2005) used a different approach to unravel underlying biological factors. They selectively bred rats emitting high or low rates of 50-kHz USV during tickling. When comparing the resulting lines, they mainly found differences in social motivation and social behavior, such as rough-and-tumble play in juveniles, with the low-rate line showing deficits (Harmon et al. 2008; Webber et al. 2012). Remarkably, such deficits were found to be associated with changes in gene expression patterns in autism candidate genes (Moskal et al. 2011).
Communicative function: behavioral responses
The emission of 50-kHz USV is a prominent part of mating behavior in rats and, hence, their functional role has been extensively studied. Behavioral observations, devocalization studies and playback experiments, all indicate that 50-kHz USV play an important role in establishing and maintaining close social contact (Geyer and Barfield 1978; McIntosh et al. 1978; Sales 1972b; Thomas and Barfield 1985; Thomas et al. 1981, 1982; White and Barfield 1987, 1989, 1990). More recent studies, however, have shown that 50-kHz USV also serve important communicative functions in the non-sexual context. Panksepp et al. (2002) have found that rats spend more time with conspecifics that emit high levels of 50-kHz USV than with others producing fewer 50-kHz USV. This again indicates that 50-kHz USV serve as social contact calls to establish and maintain contact among conspecifics. Consistent with this view, social context and social stimuli have been reported to modulate the production of 50-kHz USV. For instance, the number of 50-kHz USV emitted by rats exposed to a test environment containing the odor of conspecifics is positively correlated with the number of rats leaving their odor in this environment, suggesting that 50 kHz USV emission is driven by potential social contact (Brudzynski and Pniak 2002). However, rats were found to emit 50-kHz USV not only in anticipation of social contact but also in response to social separation, with emission being the highest immediately after separation (Schwarting et al. 2007; Wöhr et al. 2008), again in agreement with the idea that 50-kHz USV are emitted to maintain social contact.
To test experimentally whether 50-kHz USV indeed serve a pro-social communicative function as social contact calls, we conducted playback experiments and exposed juvenile and adult male rats to natural 50-kHz USV recorded from a male rat while exploring a test environment containing the odor of a familiar male conspecific (Wöhr and Schwarting 2007). Background noise, 22-kHz USV and 50-kHz sine wave tones were used as additional acoustic stimuli. The results of the playback experiment clearly showed that the exposure to 50-kHz USV elicited social approach behavior and USV emission in the recipients. Social approach behavior and USV emission specifically occurred in response to acoustic stimuli within the 50-kHz USV range, namely 50-kHz USV and 50-kHz sine wave tones. No such responses were observed when rats were exposed to background noise or 22-kHz USV. However, the finding that both 50-kHz USV and 50-kHz sine wave tones led to behavioral changes in the recipients indicated that amplitude and frequency modulation carried little or no communicative information, at least under the experimental conditions tested. In contrast to amplitude and frequency modulation, the peak frequency of 50-kHz USV appears to be highly relevant for behavioral changes to occur, as shown in a subsequent study (Wöhr and Schwarting 2012). In this study, social approach behavior was seen in response to 50-kHz USV, whereas locomotor inhibition was observed when rats were exposed to time- and amplitude-matched white noise. The lack of social approach behavior in response to the latter stimulus might have been attributable to the fact that sound energy was not confined to a critical frequency band in the ultrasonic range as in the case for 50-kHz sine wave tones known to elicit social approach behavior (Wöhr and Schwarting 2007), suggesting a categorical perception mechanism. Indeed, behavioral inhibition seen in response to playback of time- and amplitude-matched white noise was probably caused by sound energy within the critical frequency range for 22-kHz USVs.
In addition to their role in establishing and maintaining social contact, 50-kHz USV might also be involved in regulating complex social behavior. For instance, the deafening or devocalizing of rats has been shown to affect reciprocal social interaction in juveniles (Siviy and Panksepp 1987), during which high rates of 50-kHz USV occur under normal conditions (Knutson et al. 1998; Webber et al. 2012). Moreover, 50-kHz USV are reported to occur during cooperative behavior in rats, with the number of cooperative behaviors and 50 kHz USV being positively correlated (Łopuch and Popik 2011).
Communicative function: neuronal responses
Little is known about brain activity patterns induced by 50-kHz USV. Using immunohistochemical labeling of the immediate early gene c-fos, we compared brain activity patterns in rats exposed to natural 50-kHz USV, 22-kHz USV, background noise and silence (Sadananda et al. 2008). As compared with 22-kHz USV, neuronal activity levels were found to be lower in rats hearing 50-kHz USV. In particular, lower activity levels were observed in the perirhinal cortex, lateral/basolateral amygdala and the rostral part of the PAG. This pattern argues for the specificity of the USV-induced changes in activity but is not surprising, as 22-kHz USV activate such brain regions. However, findings from a recent single unit recording study by Parsana et al. (2012a) show that the observed differences in c-fos labeling are not simply attributable to increased activity levels when the rats are exposed to 22-kHz USV. Parsana et al. (2012a) have found that neurons in the lateral/basolateral amygdala not only respond to playback of 22-kHz USV but also change their firing pattern in response to 50-kHz USV. Whereas most neurons display a tonic (+) response to 22-kHz USV, a tonic (-) response is the most common firing pattern when exposed to 50-kHz USV. The opposite firing patterns elicited by the two USV types are in agreement with their distinct communicative functions and with the idea that tonic responses of neurons in the lateral/basolateral part of the amygdala reflect stimulus valence, i.e., aversive versus appetitive. In our study, reductions of activity levels were found in a number of additional brain regions, such as the lateral habenula and the dorsal raphé nuclei, as compared with silence controls (Sadananda et al. 2008). However, the observed reductions were not specific for 50-kHz USV and also occurred in response to 22-kHz USV, indicating that they reflected arousal and attention rather than stimulus valence (Abrams et al. 2004; Geisler and Trimble 2008).
In addition to decreases, exposure to the playback of 50-kHz USV led to increased c-fos labeling in the frontal cortex and the nucleus accumbens (NAcc; Sadananda et al. 2008). The neuronal activation in the frontal cortex was most pronounced in the secondary motor cortex. This activation was specifically seen in rats exposed to 50-kHz USV. As described above, 50-kHz USV but not 22-kHz USV or background noise, induce strong behavioral activation, typically directed towards the sound source, termed social approach behavior (Sadananda et al. 2008; Wöhr and Schwarting 2007, 2009, 2012). Therefore, the stimulus-specific activation seen in the secondary motor cortex is most probably associated with social approach behavior.
As for the activation seen in the frontal cortex, the increased activity of the NAcc was also specific for 50-kHz USV (Sadananda et al. 2008). Again, the activation of the NAcc through 50-kHz USV is in agreement with the social approach response elicited. Indeed, the NAcc is well known for its important role in appetitive behavior and is believed to act as an “interface between motivation and action” (Mogenson et al. 1980). NAcc activity is critically modulated by its dopaminergic input. One is therefore tempted to speculate that dopamine release in the NAcc is necessary for social approach behavior to occur when rats are exposed to 50-kHz USV. Interestingly, the local administration of the catecholaminergic agonist, amphetamine, into the NAcc has been repeatedly shown to elicit the emission of 50-kHz USV but not 22-kHz USV (Burgdorf et al. 2001; Thompson et al. 2006). Hence, the NAcc might function to close a perception-and-action loop, linking mechanisms relevant for 50-kHz USV detection and production. Such a loop appears to be particularly relevant for appetitive social and reciprocal communicatory signals. This assumption is further supported by the finding that juvenile rough-and-tumble play, which is typically paralleled by high rates of 50-kHz USV, also leads to increased c-fos labeling in the NAcc (Gordon et al. 2002).
Importantly, the NAcc is also the brain region in which opioids exert at least some of their effects on social behavior (Panksepp and Bishop 1980; Vanderschuren et al. 1995c). The opioid system is well known for its important role in regulating rough-and-tumble play (Beatty and Costello 1982; Panksepp et al. 1985; Vanderschuren et al. 1995a, 1995b). The finding that it also controls social approach behavior induced by playback of 50-kHz USV therefore needs to be highlighted (Wöhr and Schwarting 2009). In agreement with studies of opioidergic effects on rough-and-tumble play, we have shown that the opioid agonist morphine enhances social approach behavior, whereas naloxone, an opioid antagonist, inhibits social approach behavior. An important role of the opioid system in ultrasonic communication is also suggested by findings from mouse studies. Moles et al. (2004) have found that mouse pups lacking the μ-opioid receptor display strongly reduced levels of isolation-induced USVs in response to separation from their mother and littermates. When adult, male mice lacking the μ-opioid receptor do not show behavioral changes in response to the playback of USVs recorded during female-female interactions (Wöhr et al. 2011a).
Behavioral studies of aversive 22-kHz USV indicate that the likelihood of emitting 22-kHz USV is largely independent of the social context but might be potentiated by the presence of conspecifics under certain circumstances, for instance, in established colonies. They further suggest that 22-kHz USV are not innately recognized as alarm calls but that they can obtain an alarm signal value as a consequence of associative learning, such as auto-conditioning, which is facilitated by a biological preparedness to associate 22-kHz USV with aversive events. In addition, neuronal studies indicate that 22-kHz USV activate brain regions implicated in the regulation of fear and anxiety in the receivers, that at least some of these regions are required for fear conditioning to 22-kHz USV and that, among the activated structures, the amygdala and perirhinal cortex might be part of the “neural template” responsible for the biological preparedness to associate 22-kHz USV with aversive events.
Behavioral studies of appetitive 50-kHz USV indicate that they facilitate mating but also that they have a communicative value in nonsexual contexts in which they appear to serve to establish or maintain contact among conspecifics. They further suggest that such a pro-social communicative function depends on acoustic stimulus configuration, most importantly peak frequency and social acoustic memory. In addition, neuronal studies indicate that 50 kHz USV decrease activation in a large number of brain areas, including the amygdala, whereas two brain regions, namely the frontal cortex, specifically, the secondary motor cortex, and the NAcc, exhibit increased neuronal activity. Neuronal activation in the former region is probably attributable to the finding that 50 kHz USV induce pronounced behavioral activation, whereas activation of the NAcc might be related to the appetitive value of 50 kHz USV. Finally, dopamine and opioids appear to be involved in pro-social ultrasonic communication.
As pointed out by Parsana et al. (2012a), opposite behavioral responses, together with distinct patterns of brain activation, particularly the bidirectional tonic activation or deactivation of the amygdala elicited by 22-kHz and 50-kHz USV, respectively, are in agreement with a wealth of behavioral and neuroimaging studies in humans involving the use of emotionally salient stimuli, such as fearful and happy facial expressions (e.g., Morris et al. 1996; Whalen et al. 1998). Affective ultrasonic communication therefore offers a translational tool for studying the neurobiology underlying socio-affective communication. This is particularly relevant for rodent models of neurodevelopmental disorders. Indeed, socio-affective communication is severely impaired in a number of human neuropsychiatric disorders, including autism and schizophrenia. By means of human neuroimaging studies, such impairments have been repeatedly associated with aberrant amygdala responses during emotion processing, both in schizophrenia (e.g., Gur et al. 2002) and autism (e.g., Critchley et al. 2000), linking disease symptoms, brain activity patterns and genetic variants underlying deficits in socio-affective communication (e.g., Meyer-Lindenberg et al. 2009).
The authors thank Nicole Yee for the photographs shown in Fig. 4a–c.
- Beckett SRG, Aspley S, Graham M, Marsden CA (1996) Pharmacological manipulation of ultrasound induced defence behaviour in the rat. Psychopharmacology (Berl) 127:384–390Google Scholar
- Burman OHP, Ilyat A, Jones G, Mendl M (2007) Ultrasonic vocalizations as indicators of welfare for laboratory rats (Rattus norvegicus). Appl Anim Behav Sci 104:116–129Google Scholar
- Critchley HD, Daly EM, Bullmore ET, Williams SC, Van Amelsvoort T, Robertson DM, Rowe A, Phillips M, McAlonan G, Howlin P, Murphy DG (2000) The functional neuroanatomy of social behaviour: changes in cerebral blood flow when people with autistic disorder process facial expressions. Brain 123:2203–2212PubMedCrossRefGoogle Scholar
- Depaulis A, Keay KA, Bandler R (1994) Longitudinal neuronal organization of defensive reactions in the midbrain periaqueductal gray region of the rat. Exp Brain Res 90:307–318Google Scholar
- Ehninger D, Sano Y, de Vries PJ, Dies K, Franz D, Geschwind DH, Kaur M, Lee YS, Li W, Lowe JK, Nakagawa JA, Sahin M, Smith K, Whittemore V, Silva AJ (2012) Gestational immune activation and Tsc2 haploinsufficiency cooperate to disrupt fetal survival and may perturb social behavior in adult mice. Mol Psychiatry 17:62-70PubMedCrossRefGoogle Scholar
- Garcia J, Koelling R (1966) Relation of cue to consequence in avoidance learning. Psychon Sci 4:123–124Google Scholar
- Geyer LA, Barfield RJ (1978) Influence of gonadal hormones and sexual behavior on ultrasonic vocalization in rats. I. Treatment of females. J Comp Physiol Psychol 92:436–446Google Scholar
- Panksepp J, Burgdorf J (1999) Laughing rats? Playful tickling arouses high frequency ultrasonic chirping in young rodents. In: Hameroff S, Chalmers C, Kazniak A (eds) Toward a science of consciousness III. MIT Press, Camebridge, Mass., pp 231–244Google Scholar
- Sales GD (1972b) Ultrasound and mating behaviour in rodents with some observations on other behavioural situations. J Zool 168:149–164Google Scholar
- Yee N, Schwarting RKW, Fuchs E, Wöhr M (2012a) Juvenile stress potentiates aversive 22-kHz ultrasonic vocalizations and freezing during auditory fear conditioning in adult male rats. Stress 15:533–544Google Scholar
- Yee N, Schwarting RKW, Fuchs E, Wöhr M (2012b) Increased affective ultrasonic communication during fear learning in adult male rats exposed to maternal immune activation. J Psychiatr Res 46:1199–1205Google Scholar