Altered social behavior and ultrasonic communication in the dystrophin-deficient mdx mouse model of Duchenne muscular dystrophy

Adult male mice of each genotype were confronted with control (WT) females for 3 min during four consecutive days, to ensure that all subjects reliably emitted USVs. Each test male was confronted with a different female in every trial. For each testing day, test males were first familiarized to the test cage (transparent Plexiglas box, 20 × 30 × 14 cm, with floor filled with fresh bedding) for 30 min, an unknown control female was introduced, and social interaction was recorded for 3 min. Female estrous stage was verified each day after behavioral testing by vaginal smear cytology. Behavioral responses were analyzed during the first session. Behavioral responses were videorecorded, and the sniffing or snout contacts elicited by the resident males were quantified manually using event-recorder keys in ANY-maze software (Stoelting, USA). The latency of first contact, mounting attempts, frequency, and duration of contacts were analyzed. Contacts were classified as orofacial (directed towards the head/neck/mouth area), orogenital (towards the anogenital area), or directed towards other body parts (flank area).

Test mice were first acclimatized to the testing room and cage for 30 min and then tested during two consecutive days. Each daily session consisted in three consecutive trials (5-min duration) with a 5-min inter-trial interval, consisting in the successive presentation of female mouse urine (trial 1), distilled water (trial 2), and mouse urine from a different female (trial 3). A 20-μl drop of urine or distilled water on a cotton swab was placed in the testing arena hanging from a telescopic clamp close to the microphone (about 15 cm above the floor). Fresh urine was previously collected from WT female donors by holding them by the scruff onto a metal grid, above a clean aluminum foil. In case handling was not sufficient to stimulate urination, the mouse ventral area was gently stroked in the anteroposterior direction [28]. The urine on the foil was immediately pipetted into Eppendorf tubes for storage at −80 °C and classified according to individual estrous stage (determined by vaginal smear cytology after urine collection). On the day of testing, frozen urine from proestrous-estrous females (matching peak levels of estradiol and progesterone) was thawed at room temperature and pipetted onto clean cotton swabs. Mice were videotracked (ANY-maze, Stoelting, USA) to analyze the distance travelled and the frequency and duration of mobile episodes. Sniffing of the cotton swab was manually quantified using event-recorder keys in ANY-maze and the percentage time sniffing compared between genotypes.

The test mice were first acclimatized to the testing room and cage for 30 min. Then, a daily session consisted in two consecutive 3-min trials. During the first trial, bedding from a females’ cage (approximately 25 ml, cage unchanged for 1 week) was presented in one of two identical recipients (4 × 4 × 1 cm) placed at opposite corners of the testing cage. In trial 2, the previously empty recipient was filled with new bedding collected from a different females’ cage. Location of olfactory stimuli used in trials 1 and 2 was balanced across animals. The distance travelled and the frequency and duration of mobile episodes were quantified. The frequency and duration (expressed as percentage time) of bedding recipients sniffing were manually scored using event-recorder keys in ANY-maze (Stoelting, USA).

Following acclimatization to the test cage as above, the test mice were confronted to an anesthetized control female (anesthesia: IP injections of 100 mg/kg ketamin and 12 mg/kg xylazin) during two 3-min successive trials performed on the same day with an inter-trial interval of 2 min. A different female was introduced in the test cage in each trial. The frequency and duration (percentage time) of contacts performed by the test mice were manually quantified using event-recorder keys in ANY-maze (Stoelting, USA).

Apparatus and paradigm have been previously described [29]. The apparatus was a rectangular three-chambered box fabricated from clear polycarbonate. Dividing walls with manual retractable doorways (width 10 cm; height 20 cm) enabled access into each chamber (20 × 40 × 22 cm). The box was cleaned with alcohol, and fresh bedding was added between trials. The middle chamber was used to confine test mice at the beginning of test trials, while each of the two side chambers contained an empty transparent polycarbonate tube (8 cm in diameter) in which social (stranger mouse) and nonsocial (object) stimuli could be confined. These confinement tubes were drilled with holes (1-cm diameter; 2-cm spaced) to allow limited contact with the confined stimulus. Control male mice used as social stimuli (strangers) were familiarized to the confinement tubes during 4 days (30 min per day) before the start of the experiment. Then, testing consisted of three trials: For trial 1 (habituation), the test mouse was placed in the middle chamber for 2 min, the doorways were then opened, and the mice could freely explore the test box for 10 min after which they were confined in the middle chamber for 30 s before proceeding with the next trial. During trial 2 (sociability), an unknown mouse (stranger 1) was introduced in one of the confinement tubes in one side chamber, whereas an object made of Lego® pieces was enclosed in the other tube in the opposite side chamber. Location of stranger 1 in the left or right side chamber was balanced across subjects. Upon door re-opening, the test mouse was allowed to explore the test box and tubes for 10 min and was then confined again in the middle chamber before the next trial. During trial 3 (preference for social novelty), the object present in trial 2 was replaced by a novel unfamiliar mouse (stranger 2), and the test mouse was then given the possibility to interact with both stranger 1 and stranger 2 for 10 min. Behavior was videotracked during all trials (ANY-maze software) to analyze general activity (distance travelled, rearing and leanings, self-grooming, periods of activity and inactivity). Social behavior was reflected by the percentage time spent and number of entries in the side chambers containing the confined stimuli, as well as by the frequency and duration (percentage time) of sniffing episodes directed against the confinement tubes.

The protocol was adapted from previously described methods [30]. Test mice of both genotypes (mdx and WT littermates) were initially changed to individual caging (20 × 30 cm cages) for 2 months without bedding change. Three distinct strains of mice were used as intruders, based on their known differences in social behavior: male C57BL/10 mice aged 3–4 months, which were of the same genetic background as the resident test mice; males of the C3H strain, known as non-aggressive mice; and males of the balb/c strain known for their high intraspecific aggressiveness [31]. Each test mouse was successively confronted to these three types of intruders within a 2-week period (5-day intervals between trials).

Behavior was videorecorded for 6 min after placing the intruder in the resident’s home cage. The following behavioral parameters were coded manually using event-recorder keys (ANY-maze software): contacts initiated by the resident, contacts initiated by the intruder, fights, pursuits, rearings, dominant behavior (putting a paw on the body of the other mouse and/or mounting), as well as periods of immobility either during or between social contacts. The specific nature of the contacts was further specified as orofacial, orogenital, and body sniffing.

The social behavior of mdx resident males differed depending on the sex and genotype of the intruder. As shown in Fig. 3a, mdx mice initiated fewer contacts with females as compared to WT mice (p < 0.005; Fig. 3a), which was not associated with changes in specific types of contacts. In male–male encounters (Fig. 3b–d), the mdx resident mice also displayed fewer contacts towards mdx male intruders (p < 0.05) but conversely initiated more contacts than control residents when confronted to WT intruders (p < 0.05) (Fig. 3b). In any cases, no fights were observed and the latency of the first contact (all <10 s) and the number of dominant acts initiated by resident test mice (Fig. 3c) were comparable regardless of the intruder’s genotype (p > 0.05). However, the increased number of contacts with WT intruders, as shown in Fig. 3b, was associated with the initiation of an increased number of pursuits directed against the WT intruders (p < 0.05). This was not observed when mdx residents were confronted with mdx intruders (p > 0.05; Fig. 3d), showing that pursuit behavior, as well as the number of contacts, varied depending on the intruder’s genotype.

Interestingly, the intruder’s behavior also varied depending on the resident’s genotype (Fig. 3e–g). The behavior of mdx intruders was not influenced by the resident’s genotype while, in contrast, the behavior of WT intruders varied when confronted with mdx residents: indeed, the WT intruders initiated more contacts (Fig. 3e), dominant acts (Fig. 3f), and pursuits (Fig. 3g) during interactions with mdx residents than with other WT residents (all parameters, p < 0.05). This suggests that mdx mice, when acting as passive (or receptive) residents, were showing more submissive responses compared to controls.

To determine whether changes in intruder’s aggressiveness influenced the variable behavior of mdx resident mice in the experiments above, a different group of male residents was submitted to a longer period of social isolation (2 months before testing) and forced encounters with intruders of three distinct genotypes were performed in the resident’s home cage (instead of a novel cage) to increase agonistic interactions. Fights were scarce with C57BL/10 and C3H intruders but more consistently observed with balb/c intruders (Fig. 4a). Although pursuits and fights were mainly initiated by residents, the intensity of aggressiveness was thus dependent on the intruders’ behavior. Still, the incidence of fights was relatively low, which allowed reliable recording of social behaviors. Self-grooming episodes were scarce and very short (<4 s; no significant genotype effect). In this experiment, the intruders’ genotype also influenced the quantity of social interactions initiated by residents, as revealed by significant intruder’s genotype × dependent variable interactions (contact latency: p < 0.05; number: p < 0.01; duration: p < 0.001). Figure 4b shows the longer duration of contacts in residents of both genotypes in presence of low-aggressiveness C3H intruders (p < 0.001).

Conversely, intruder mice displayed shorter latencies to interact with mdx as compared to WT residents (genotype: p < 0.05; genotype × intruder interaction: p = 0.3; Fig. 4c) associated with a slightly longer duration of contacts (p = 0.09; Fig. 4d). To better understand why the intruders behaved differently when confronted with mdx residents, we compared the quality of the social interactions when intruders were of the C3H and balb/c genetic backgrounds during the first 3 min of testing (Table 2). There was an increased expression of a specific set of behavioral responses, such as orofacial contacts, pursuits, dominant acts, and freezing behavior, when resident mice were confronted with C3H intruders as compared with interactions involving balb/c intruders. Moreover, the non-aggressive C3H intruders displayed more dominant acts towards mdx mice, while the more aggressive balb/c mice directed dominant acts preferentially towards WT residents (Fig. 4e). Periods of immobility (freezing), which likely reflected fear responses, were more frequent and longer during physical contacts in mdx residents compared to WT, while the quantity of rearings was conversely decreased (Table 2). Interestingly, immobility was comparable between genotypes during periods with no physical contacts, suggesting that the enhanced freezing in mdx mice was selectively triggered by social contact. Enhanced fear/defensive responses in mdx mice likely have favored intruder’s dominant behavior, as freezing was more prominent when mdx mice interacted with the C3H intruders (Fig. 4f), which also showed the largest amount of dominant behaviors towards mdx mice (Fig. 4e).

The study of ultrasonic communication in mice is a useful means to unveil phenotypic traits in mouse models of diverse speech and neurodevelopmental disorders including autism [27, 36‐38]. Mouse USVs consist in different syllable types produced in bouts that follow a song-like structure, based on call-pattern mechanisms similar to those observed in primates, thus providing a mammalian model to study vocal learning and human disorders associated with speech or communication impairments [39]. Mice normally produce complex ultrasonic vocalizations in response to various sexual and social stimuli [25, 37], which is interpreted as a communicative behavior enabling mother mice to retrieve their pups after removal from the nest during the postnatal period and modulating mice social interactions in adulthood [40, 41]. Here, we show that the production of isolation-induced USVs progressively increases from PND3 to PND9 in both mdx and WT pups, as expected [24]. However, transient alterations were observed in mdx pups at specific postnatal ages, as also demonstrated in other autism-relevant mouse models [24, 42]. First, there was an abnormal increase in call peak frequency in mdx pups at PND3. According to playback experiments [43], mother mice may show a stronger response towards a 65–45 kHz signal than to a 75–55 kHz signal, perhaps because high frequencies propagate less than low frequencies, suggesting that high peak frequencies decrease the functional value of pups “alarm” calls. The second alteration found in mdx pups was a change in the frequency modulation patterns at PND3–4 and PND6–7. Postnatal development was associated with a progressive increase in the proportion of highly frequency-modulated calls, such as sinusoidal (i.e., complex) and composite (i.e., frequency jump) calls, and such frequency modulations are believed to be critical for maternal behavior [44]. However, mdx pups emitted less “simple” downward calls at PND3 and peak calls at PND6–7, while they displayed more composite calls at PND4 compared to WT mice. This suggests that neonate mdx mice use a more complex repertoire of calls compared to WT mice. Interestingly, such changes in mice vocalization repertoire have been previously associated with changes in emotional reactivity or arousal and are consistent with the profile of communication alterations in other rodent models of autism [24, 40].

Context-specific alterations in USVs were also found in adult mdx mice and were characterized by a selective reduction in the quantity of vocalizations emitted in presence of anesthetized females and during interactions with other mdx male mice. A more modest and non-significant reduction in call rate was also observed during interaction with freely moving females. In contrast, the USVs emitted in low-noise situations in response to potent volatile sexual olfactory stimuli (proestrous-estrous female urine on a cotton swab) or to a combination of volatile and non-volatile sexual stimuli (exploration of female cage bedding) did not reveal major changes in call rate and duration, suggesting that main and accessory olfactory systems are unaltered in mdx mice [45, 46]. Also, a putative influence of motor or muscular defects is unlikely. Indeed, the intrinsic laryngeal muscles involved in vocal production by controlling the position and tension of the vocal folds are spared from the dystrophic process in the mdx mouse [47]. Although a deficient motor coordination in mdx mice [48] could alter the genesis of USVs, as suggested by studies in the Foxp2-deficient mouse [36, 49], this could not explain the selectivity of the deficits in mdx mice. Moreover, the recordings performed in presence of an anesthetized female enabled USV analysis in low background noise and ruled out possible biases due to the putative production of USVs by intruders. Our data therefore show that USVs are specifically affected in mdx mice in social contexts and more likely depended on altered social motivation.

The USVs normally emitted in the presence of females are thought to facilitate approach behavior and copulation [50], while during male–male encounters they have been associated with expression of affiliative behaviors [25]. When mdx mice were submitted to dyadic interactions with either WT females or mdx male intruders in a novel environment, the reduced number of vocalizations was associated with a reduced number of contacts initiated by mdx residents. Conversely, a slight increase in call rate was associated with an increased number of contacts initiated by mdx residents towards WT male intruders. In these situations mdx calls had shorter durations and amplitudes, two parameters remarked as prosodic elements conveying critical emotional or motivational information for social interactions [51]. Accordingly, call duration has been shown to be modulated by social motivation [25] and reduced in mouse models of autism [27]. Here, we also demonstrate that adult mdx mice used an abnormal vocal repertoire in specific social contexts, i.e., a reduced expression of peak and composite calls in presence of anesthetized females and of sinusoidal and simple calls in male–male encounters. The composition of an adult mouse repertoire normally depends on the context and previous social experience [25]. Short and composite calls are considered as “basic” calls found in many behavioral context whereas upward, frequency jump, u-shape, flat, and chevron (i.e., peak) calls are predominantly found in social situations and therefore considered as more “informative” of the behavioral, emotional, or motivational content of these specific situations. Here, the altered vocal repertoire during male–male encounters was associated with a reduction in the number of contacts, suggesting a functional link between social behavior deficits and altered USVs in mdx mice. The selectivity of the observed deficits likely depended on a change in social motivation. This might reflect an altered neural control of call production, perhaps due to the lack of Dp427 in neocortical principal neurons [52], in particular in the anterior cingulate cortex, which takes part in the volitional circuits associated with call onset [38]. Interestingly, some models of autism such as neuroligin-deficient mice have been shown to exhibit deficits in the production of complex calls during social interactions [53], while they were able to produce these calls in other contexts, which also suggests that the altered use of the vocal repertoire may reflect a deficit in behavioral responsiveness to social stimuli eliciting USVs [38].

We show for the first time the presence of context-specific disturbances in social behavior and ultrasonic communication in the dystrophin-deficient mdx mouse. Hence, mutations selectively affecting the expression of the full-length Dp427, a common genetic alteration in all DMD patients, appear to be sufficient to significantly alter social behavior and communication. This is in agreement with a case study reporting ASD in a DMD patient holding a mutation that selectively impairs Dp427 expression [15]. Mutations that affect expression of the other C-terminal brain forms of dystrophin have been associated with mental retardation. Therefore, the general observation of social behavior deficits in DMD patients that do not display mental retardation also supports a role for Dp427 in social behavior [10]. Nevertheless, patients with intellectual quotients in the normal range may exhibit deficits in executive functions [3], which might contribute to the emergence of both social behavior and communication deficits.

In mdx pups, transient alterations in USVs are readily detectable during the first postnatal week, while social behavior deficits in the adults are associated with the use of an abnormal vocal repertoire of vocalizations in specific social contexts, suggesting alterations in the executive control of social motivation, adaptive behavior and communication. Strikingly, delayed speech development, limited expressive and receptive vocabulary, reduced verbal fluency, and verbal short-term working memory have been described in DMD patients [54, 55] and associated with alterations in adaptive and social skills [10, 12]. Even when ASD is not diagnosed, DMD patients may display moderate weaknesses in social behavior and communication, including language difficulties, tendency of being withdrawn, avoidance of eye contact, difficulties in interpreting facial affect, and problems with theory of mind [56]. Social behavior relies on the functional integrity of the prefrontal cortex and cerebellum [57, 58]. Dysfunctional cortical-cerebellar circuits have been associated with autism [59], and this has also been proposed as a neural basis of the defective social and executive functions in DMD patients [2]. While impairments in executive functions have not been clearly characterized in the mdx mouse, this model displays memory deficits, altered synaptic plasticity and enhanced fearfulness [34, 60], suggesting that both cognitive and conative disturbances due to Dp427 loss could contribute to the social behavior and USV alterations.

The mechanisms responsible for emergence of autistic symptoms in DMD patients are still unclear, although one likely hypothesis is an alteration of the molecular interactions between the dystrophin complex and the trans-synaptic neurexin-neuroligin complex in central inhibitory synapses [18]. Dystrophin is selectively involved in the organization of central inhibitory postsynaptic scaffolds by contributing to the recruitment of neuroligin-2 (NLGN2) [19, 61, 62], a synaptic cell adhesion protein involved in structural remodeling of connectivity networks which has been proposed as a candidate gene for ASD [63]. Synaptic molecular alterations specifically due to Dp427 loss in mdx mice are characterized by a delocalization of NLGN2 and presynaptic vesicular GABA transporter (VGAT) and a reduction of NLGN2 colocalization with alpha-1 subunit-containing GABA_A receptors within hippocampal synaptic layers, suggesting a modification of the molecular mechanisms that normally underlie precise spatio-temporal pattern of GABAergic transmission [61]. Dystrophin loss also alters synapse ultrastructure, clustering of distinct subtypes of GABA_A receptors, central inhibitory function, and synaptic plasticity, which has been associated with disturbances in fear-related behaviors, anxiety, and cognitive functions [34, 64‐66]. Interestingly, rodent models overexpressing NLGN2 display deficits in reciprocal social interactions and stereotypies [67, 68], while in the knockout model, a reduced production of isolation-induced calls has been reported in pups [69], but no overt impairment in social behavior was found in the adult. Future pharmacological approaches aimed at compensating altered GABAergic function or restoring dystrophin function in mdx mice might be helpful to further delineate the physiopathology of autistic traits in DMD, to highlight the importance of the neuroligin and neurexin complex in ASD and to unveil new leads to alleviate behavioral symptoms in DMD patients.