https://orcid.org/0000-0002-9889-334X
Figures
Abstract
Autism is a common neurodevelopmental disorder that despite its complex etiology, is marked by deficits in prediction that manifest in a variety of domains including social interactions, communication, and movement. The tendency of individuals with autism to focus on predictable schedules and interests that contain patterns and rules highlights the likely involvement of the cerebellum in this disorder. One candidate-autism gene is contactin-associated protein 2 (CNTNAP2), and variants in this gene are associated with sensory deficits and anatomical differences. It is unknown, however, whether this gene directly impacts the brain’s ability to make and evaluate predictions about future events. The current study was designed to answer this question by training a genetic knockout rat on a rapid speech sound discrimination task. Rats with Cntnap2 knockout (KO) and their littermate wildtype controls (WT) were trained on a validated rapid speech sound discrimination task that contained unpredictable and predictable targets. We found that although both genotype groups learned the task in both unpredictable and predictable conditions, the KO rats responded more often to distractors during training as well as to the target sound during the predictable testing conditions compared to the WT group. There were only minor effects of sex on performance and only in the unpredictable condition. The current results provide preliminary evidence that removal of this candidate-autism gene may interfere with the learning of unpredictable scenarios and enhance reliance on predictability. Future research is needed to probe the neural anatomy and function that drives this effect.
Citation: Centanni TM, Gunderson LPK, Parra M (2025) Use of a predictor cue during a speech sound discrimination task in a Cntnap2 knockout rat model of autism. PLoS One 20(8): e0315883. https://doi.org/10.1371/journal.pone.0315883
Editor: Stephen D. Ginsberg, Nathan S Kline Institute, UNITED STATES OF AMERICA
Received: December 30, 2024; Accepted: July 25, 2025; Published: August 18, 2025
Copyright: © 2025 Centanni et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All data files are available on OSF.
Funding: NICHD 1R15HD103479-01A1 to TMC The funders of this study had no role in the study design, data collection and analysis, decision to publish, or preparation of this manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Autism is associated with many behavioral symptoms in domains including social interactions, communication, and movement, that are likely due to deficits in prediction. Given the range of prevalent deficits, individuals with autism tend to struggle more often in academics, are less likely to achieve independence, and have poorer long-term life outcomes than neurotypical individuals [1]. More specifically, individuals with autism often exhibit deficits in tasks that require creating and evaluating upcoming events. For example, children with autism often fail to catch a bouncing ball, likely due to an inability to predict its trajectory [2]. The social deficits in autism, including theory of mind, are also possibly due to an inability to predict or anticipate how another person might be thinking or feeling [3]. This prediction deficit is thought to underlie the tendency of individuals with autism to prefer predictable schedules and focus on interests that contain patterns and rules [4–7]. Deficits with prediction or creating expectations about upcoming events in several domains are likely linked with abnormalities in cerebellar circuits [8,9]. Given that the cerebellum is critically involved in evaluating anticipated movements and assisting the cortex in processing of speech, it is not surprising that the cerebellum is also one of the most consistently reported regions of anatomical and functional differences in autism [8]. The goal of the current study was to evaluate the effect of an autism gene known to impact cerebellar development on the detection of predictable vs. unpredictable rapid auditory stimuli in a rat model.
With respect to language, abnormalities in the cerebellum likely impact the development of speech sound knowledge early in life, with effects that cascade throughout the lifespan. In young adults, processing predictable sentences, for example, increases activation in the right cerebellum, which connects to the left cerebral language areas [10]. If cerebellar dysfunction in autism leads to an array of prediction deficits, understanding the direct link between the brain and behavior is important for developing earlier diagnostic tools, perhaps even prior to the onset of theory of mind in childhood. One method for probing whether prediction deficits are ubiquitous across modality and present early in life is through the study of autism-candidate genes.
Variants in the autism-candidate gene contactin-associated protein 2 (CNTNAP2) are associated with sensory deficits and anatomical differences [11]. The protein coded for by this gene, Caspr2, plays a role in axonal growth for mature myelinated neurons and, when deficient, leads to abnormal neuronal migration [12]. This gene is highly expressed in Purkinje cells, the main neuron type in the cerebellum, and variants in this gene are associated with reduced cerebellar grey matter in humans [13]. Mice with Cntnap2 knockout exhibited significant differences not only in Purkinje cell morphology, but also increased spontaneous firing, increased excitability, and earlier onset latencies in response to somatosensory stimuli [14]. Abnormalities in development of auditory perception and reactivity in Cntnap2 knockout rats [15] align with symptoms reported in humans with autism and suggests this gene should be one of interest in sensory processing studies.
Most of the prior research characterizing the Cntnap2 knockout rodent (including mice and rats) has investigated social behaviors, activity levels, and motor coordination (for review, see [16]). While cognitive flexibility has been evaluated in other genetic models, and atypical behaviors were reported in Cntnap2 knockout mice [16], this has not yet been fully evaluated in the rat model, which is capable of much more complex operant behaviors. During a pre-pulse inhibition (PPI) task, in which a cue indicates an upcoming startling event, there were mixed results on behaviors such that Cntnap2 knockout mice showed an impaired startle response [17] while the effect in rats varied based on genotype and age [15]. Adult rats with homozygous Cntnap2 knockout exhibited increasing startle response during a PPI paradigm [15], suggesting an impaired ability to anticipate upcoming events. It is unknown whether removal of this gene would also interfere with prediction during an operant task not containing aversive stimuli.
Given the inability to conduct causal genetic link studies in humans, animal model work is necessary to answer this question. The use of human speech sounds in rat models of human communication impairments is well-validated and has translational value. Since rats do not process human speech sounds as ‘language’, they are considered artificial and semantically meaningless stimuli. This provides a benefit over conspecific stimuli, which has a social component and is therefore not ideal for a study of autism given that Cntnap2 knockout rodents exhibit abnormal social behavior and vocalizations [18,19]. However, despite the lack of semantic association in the rat, the use of human speech sounds is considered to have translational value for our understanding of speech sound processing in humans. Rats are capable of learning to discriminate human speech sounds based on the initial consonant [20,21], or the middle vowel [22]. They can accomplish this task in the presence of background noise [23], with acoustic degradation [24], and at various speeds [25], while performing at levels that mimic human performance curves. In addition to their comparable behavior, using human speech sounds provides value in allowing us to probe the complex spatio-temporal patterns evoked by speech in primary auditory cortex, which is difficult to do in humans.
Comparative research using rodents (work described above), cats [26], and non-human primates [27] has agreed that at the basic auditory level, non-human animal research is informative to our understanding of primary sensory level speech sound perception in humans in a way that measures auditory perceptual deficits without the confound of the social deficit observed in humans. Studies of auditory cortex encoding of speech sounds in rat models of autism have highlighted potential mechanisms for speech perception differences in humans, including altered patterns of cortical firing [28,29]. Genetic rat models of other neurodevelopmental disorders, such as dyslexia, have led to the discovery of new neural deficits in a subgroup of children [30–32]. Although rats do not process speech sounds as language, they are a useful model for understanding the impact of genetic manipulation on operant behavior and neural encoding of speech sounds as auditory cues. Thus, the current study utilized a Cntnap2 knockout rat model to evaluate the impact of this gene on a rapid speech sound discrimination task that included unpredictable as well as predictable target presentations. We hypothesized that the knockout rat would exhibit deficits in the unpredictable condition and an enhanced performance when a reliable predictor was present.
Methods
Animals
Subjects were 17 Long-Evans rats developed, bred, and supplied by the Medical College of Wisconsin, generated with support from the Simons Foundation Autism Research Initiative. Rats were generated by the injection of CRISPR/cas9 targeting rat Cntnap2 into the embryos of a knockout/wild type cross. The rats in the current study came from 5 litters. Following data collection, fluorescent genotyping was completed by the Medical College of Wisconsin to verify each rat’s genotype. Of our 17 animals, 9 were Crispr/Cas9 Cntnap2 homozygous knockouts (N = 2 female) and the remaining 8 were wildtype littermate controls (N = 3 female). During behavioral training, rats were housed on a reverse 24-hour light:dark cycle (lights on, 7:00 pm- 7:00 am) with food restriction to ensure motivation during behavioral training but were not allowed to fall below 85% of their pre-deprivation body weight. The health and behavior of the rats was monitored twice a day to ensure their well-being across the study and minimize any potential suffering, but any animals experiencing distress that could not otherwise be remedied were euthanized with CO2 inhalation by thoroughly trained animal care staff or a senior graduate student. This was only the case for two rats originally included in the study which experienced seizures, a known risk associated with the Cntnap2 knockout [33,34], and were euthanized the day the seizing was observed and confirmed by animal care staff. No animals died before meeting endpoint criteria and no other animals met endpoint criteria prior to the completion of the study. Rats began training at three months of age and completed testing no later than twelve months of age. All animal protocols were approved by the Texas Christian University Institutional Animal Care and Use Committee.
Behavioral paradigm
The first portion of the task was identical to our prior work [25,35,36]. In brief, rats were trained to discriminate a target speech-sound (/dad/) from a randomized string of speech-sounds (/bad/,/gad/,/sad/, and/tad/) presented at varying speeds (2, 4, 5, 6.7, 10 and 20 syllables per second/sps; Fig 1A). The initial set of sounds were approximately 500 ms in length and the subsequent speeds were obtained by compressing the stimuli to 50%, 40%, 30%, 20%, and 10% of the original length [25,35,36]. Training was conducted over two 30-min sessions per day (5 days/week) in an operant chamber placed into a double-walled, soundproofed booth (Vulintus Inc., Lafayette, CO). The chamber contained a pellet dispenser and an infrared-monitored nose poke on opposing walls (Fig 1B). Speech sounds were recorded by a female native-English speaker and shifted up 1 octave into the rat hearing range. Each sound was calibrated with respect to its length so that the most intense 10% of the stimulus length was heard at 60 dB SPL. Since rats could move around the operant chamber freely, calibration was conducted at three locations throughout the chamber and averaged. Sounds were presented through a Tucker-Davis speaker mounted to the back side of the chamber and routed through a TDT RP2 (Tucker Davis Technologies, Alachua, FL).
A. The rat initiates a trial by entering the nosepoke. Removal of the rat’s nose from the nosepoke is considered a response. Responding to distractor sounds was considered a false alarm. Failing to respond to the target was considered a miss. Top row shows an example trial from the unpredictable condition. Middle row shows an example catch trial where no target was present. The bottom row shows how the random sequence example in the top row would be altered in the predictable condition, where sounds are still presented randomly, but the sound/bad/ is presented immediately before a target sound. B. The operant booth consisted of an infrared nosepoke and pellet feeder, located on opposite walls of the booth. A speaker was mounted to the back wall of the booth. The entire booth was located inside a double-walled, sound shielded booth.
The first stage of training used autoshaping to teach the rat to place their nose in an infrared-monitored nose poke to associate the sound of the target (/dad/) with the receipt of a 45 mg sugar pellet (sucrose) reward. After receiving at least 100 pellets in a single session, they were then trained to hold their nose in the nose poke during varying silent delays until the target sound/dad/ was played, at which time they should remove their nose to receive the sugar pellet reward. Progression through training stages introduced rats to individual and combinations of distractor sounds, which the rat learned to ignore while keeping their nose in the nosepoke. Throughout training and testing, if the rat removed their nose from the nosepoke in response to the target sound, a 45 mg sugar pellet was dispensed. If the rat falsely responded to a distractor sound, the booth lights were extinguished and the program paused for 6 seconds.
Movement through these stages was dictated by previously demonstrated criteria using the d’ value of each training session as an indicator of discrimination ability (Table 1). The d’ value is defined as the Z-score of the distractor response rate subtracted from the z-score of the target response rate. After 10 nonconsecutive sessions with a d’ ≥ 1.50, a series of eight training stages was used to introduce each of the four distractors and increasing levels of randomization. After completion of stage 11, the rats were trained to discriminate the target sound from the distractors at progressively increasing speeds up to 20 sps. As reported in our prior work, training on this set of stages typically takes approximately 4 months [25,35,36]. Next, rats were placed into a new training stage, where the distractor sound/bad/ was converted to a predictor sound such that whenever/bad/ was played, the target sound always followed [37]. After achieving two nonconsecutive sessions with a d’ ≥ 1.50, rats completed a final testing stage over 10 sessions, during which the predictor was played in 40% of initiated trials, with the target appearing at random without the predictor in the remaining 60% of trials. Response rates and reaction times to each stimulus were recorded as dependent variables for analysis.
Statistical plan
All analyses were performed using custom code in Matlab (version R2021b). Response rate was defined as the total number of responses to each sound at each presentation rate with respect to the total number of presentations. Reaction time was defined as the mean time to respond to a stimulus in milliseconds (ms). Repeated measures analyses of variance (ANOVA) were utilized to assess the effect of the predictor (present vs. absent), presented sound, and presentation rate on response rates and reaction times. Any significant main effects and interactions were explored with post-hoc t-tests using the Bonferroni adjustment to correct for multiple comparisons as needed. T-tests used to compare response rates across groups were one-tailed to test our a priori hypotheses that the knockout rat would exhibit deficits in the unpredictable condition and an enhanced performance when a reliable predictor was present.
Results
Knockout rats were affected by randomized distractors during training
Both groups of animals progressed through training at the same pace such that the number of days to hit criterion in each training stage was equivalent across genotype groups (one-tailed unpaired t-tests; ps > 0.10). When randomization was first introduced in training (stage 11; [25,36]), the knockout rats responded significantly more to the distractors (13.66 ± 1.41%) compared to the wildtype controls (9.78 ± 1.57%; one-tailed t-test, t (15) = 1.97, p = 0.03), but responded just as frequently to the target sound (KO: 64.89 ± 5.16 vs. WT: 61.38 ± 3.53%; one-tailed t-test, t (15) = 0.58, p = 0.28; Fig 2). As training progressed, knockout rats continued to respond significantly more often to the distractors at the slowest speed compared to the wildtype controls (ps < 0.03). To ensure this was not due to an anxiety behavior or a motor deficit, we also evaluated reaction times during training to all stimuli. Across genotypes, there were no significant differences between reaction times to the distractors (two-tailed unpaired t-test, t (15) = 0.33, p = 0.74) or to the target sound (two-tailed unpaired t-tests, t (15) = 1.00, p = 0.33). Within the wildtype control group, rats generally responded more quickly when false alarming to distractors vs. responding to the targets (two-tailed paired t-test; t (7) = 2.80, p = 0.026). Within the knockout group, there was a trend in the same direction (two-tailed paired t-test; t (8) = 2.21, p = 0.058). Two-tailed t-tests were used for reaction time analyses as we did not have an a priori hypothesis about processing speed.
In stages 11-16 of training (after rats learned the mechanism of response and reliably responded to the target sound), rats were gradually exposed to more complex unpredictable sequences and faster presentation rates [36]. During these stages of training, knockout rats responded more frequently to the distractors (increased false alarms) compared to the wildtype controls. Error bars represent standard error of the mean. * p < 0.05, ** p < 0.01.
Knockout rats can efficiently utilize a predictor cue
As expected based on our prior work [25,36,37], wildtype controls learned the unpredictable rapid speech sound discrimination task and learned to utilize the predictor cue to anticipate the target sound/dad/ (Fig 3; black lines). The KO rats similarly learned both versions of the task, exhibiting comparably high hit rates in both predictable and unpredictable conditions during final testing (Fig 3; blue lines). There were significant main effects of genotype (F (1, 1008) = 24.5, p < 0.0001), presentation rate (F (5, 1008) = 33.77, p < 0.0001), predictability (F (1, 1008) = 48.44, p < 0.0001), and sound (F (4, 1008) = 232.14, p < 0.0001). Within each genotype group, we evaluated the effect of the predictor on hit rates and false alarm rates. In the WT group, hit rates to the target sound were higher when the predictor was present at 2 sps (two-tailed, paired t-tests; t (7) = 3.6, p = 0.009) but hit rates were lower when the predictor was present at 6.7 sps (t (7) = 2.51, p = 0.04). This reduced hit rate in the context of the predictor could be due to the learned relationship between predictor and target leading to increased anticipatory responses to the target [37]. False alarm rates in the WT group were, however, higher in the unpredictable condition at the four slower speeds compared to in the presence of the predictor (ps < 0.038). In the KO group, there were no differences in hit rates in the predictable vs. unpredictable conditions (ps > 0.17). There were differences in false alarm rates in the KO group such that rats exhibited higher false alarm rates at every speed in the unpredictable condition (ps < 0.05).
Hit rates to the target/dad/ were high in both unpredictable (A) and predictable (B) contexts. C. During unpredictable trials, summed false alarm rates to the three distractors were significantly below hit rates until 20 sps (one-tailed paired t-tests; WT ps < 0.039, KO ps < 0.01), as has been shown in our prior work [36]. D. During predictable trials, KO rats responded more to the target than the distractors at all speeds (ps < 0.011) and WT rats exhibited this pattern at 4 of the 6 speeds (all except 6.7 and 20 sps; ps < 0.02). Error bars represent standard error of the mean.
Interestingly, early anticipatory response rates to the predictor sound/bad/ were significantly higher in the KO group at the two slowest and the two fastest speeds (one-tailed t-tests, ps < 0.04; Fig 4). This increased anticipatory response, however, did not impede accuracy. The KO animals responded significantly more frequently to the target sound/dad/ than the WT rats at 4 sps (KO: 65.70 ± 4.59 vs. WT: 50.91 ± 6.43%; one-tailed t-test, t (15) = 2.03, p = 0.03), 6.7 sps (KO: 41.62 ± 7.73 vs. WT: 15.81 ± 4.13%; t (15) = 3.01, p = 0.004), and 20 sps (KO: 18.20 ± 4.20 vs. WT: 9.30 ± 2.37; t (15) = 1.89, p = 0.039). This pattern of results suggest that the KO group was more efficient at utilizing the predictor cue to minimize false alarms and more efficiently respond to the target sound.
During the testing stages, there were no differences in response rates to the target sound/dad/ across genotypes in the unpredictable (A.) or predictable (B.) conditions. While there was no effect of predictability on false alarm rates in the WT rats, the KO rats exhibited higher false alarm rates during unpredictable trials (C.) compared to predictable trials (D.) During predictable trials, the response to the predictor sound/bad/ was higher in the KO group compared to the WT group at four of the six speeds (E.). Error bars represent standard error of the mean.
With respect to reaction times, in the unpredictable condition, WT rats responded significantly faster to the target/dad/ compared to the KO rats at three speeds: 5 sps (unpaired, two-tailed t-test, t (15) = 2.13, p = 0.05), 6.67 sps (t (15) = 2.34, p = 0.03), and 20 sps (t (15) = 2.32, p = 0.03; Fig 5). The same general pattern was observed during the prediction condition. During these trials, WT rats were significantly faster at responding to the target/dad/ at 6.7 sps (t (15) = 2.88, p = 0.01) and 20 sps (t (15) = 3.99, p = 0.001).
A. During the unpredictable trials, WT rats responded faster to the target sound than KO rats at three speeds (* p < 0.05). B. During predictable trials, a similar pattern was observed, where WT rats responded faster than KO rats at two of the six speeds (*p < 0.05). Error bars represent standard error of the mean.
To ensure that our pattern of results was not due to hyperactivity in the KO group, we quantified the number of trials initiated by each group. WT rats initiated 134.31 ± 33.5 trials during the final testing stage compared to 152.5 ± 14.9 trials by KO rats (unpaired, two-tailed t-test; t (15) = 1.13, p = 0.28).
Preliminary evidence for sex differences
Given the small numbers of female rats in the sample (N = 3 WT, N = 2 KO), we conducted an exploratory analysis on possible sex differences on response rates. First, we evaluated response rates in the unpredictable condition. In the WT group, female rats responded significantly more to the target/dad/ at 10 sps (29.68 ± 6.62%) compared to males (11.15 ± 3.74%; unpaired, two-tailed t-test, t (6) = 3.11, p = 0.02). In the KO group, female rats responded significantly more to the target/dad/ at 20 sps (27.88 ± 4.08%) compared to males (9.12 ± 3.80%; unpaired, two-tailed t-test, t (7) = 2.67, p = 0.03). In the predictable condition, there were no sex differences in response rates to the target/dad/ in the WT group (ps > 0.22) or in the KO group (ps > 0.16). Similarly, there were no sex differences in response rates to the predictor/bad/ in the WT group (ps > 0.19) or in the KO group (ps > 0.58). The increased response rates at faster speeds by female rats in both groups compared to their male counterparts could indicate increased temporal processing by females, an effect which is exacerbated in the Cntnap2 knockout. However, it is important to reiterate that due to the small sample of female rats, these analyses should be considered exploratory, and a replication is needed to verify these patterns.
Discussion
The goal of the current study was to evaluate the effect of Cntnap2 knockout on the ability of rats to learn and utilize a predictive cue during a rapid auditory discrimination task. Although both knockout and wildtype rats were able to learn the task and perform with comparable accuracy during unpredictable trials, the knockout rats exhibited significantly higher false alarm rates during training and higher hit rates to the target during testing when the predictor cue was available compared to their wildtype counterparts. The current results provide preliminary evidence that removal of this candidate-autism gene creates an animal that is more likely to respond to a distractor in unpredictable conditions and performs more accurately in predictable conditions. Future research is needed to probe the neural anatomy and function that drives this effect and consider translation of these findings into humans.
Cntnap2 KO rats utilize a predictor cue more effectively
In the current study, we observed evidence of improved use of a predictor cue in rats with Cntnap2 KO compared to WT controls. Our KO rats were more accurate during predictable trials at 4, 6.7, and 20 sps speeds compared to their WT counterparts, with the largest effect at 6.7 sps. While Cntnap2 KO rats exhibited more false alarms than WT rats during initial training on random sequences, this effect disappeared by the testing phases, when there was no main effect of predictability. One possible explanation for the lack of a deficit during unpredictable sequences in our study is that the prediction deficits associated with abnormal Cntnap2 expression are not universal and may be restricted to certain types of behaviors. For example, prior studies in rodent models have reported social deficits following Cntnap2 knockout [38] and social deficits are associated with this gene in humans [39]. Since we did not use conspecific vocalizations and human speech is not meaningful to rats, we cannot speculate about whether this gene impacts speech sound perception in humans.
It is also interesting to note that despite their increased accuracy during predictable trials, the KO rats in our study exhibited increased reaction times at faster speeds compared to the WT group. It was possible that the prior learning of/bad/ as a distractor could have interacted with its later meaning as a predictor. However, since rats were never rewarded for responding to/bad/ and the consequences for responding to/bad/ were the same in either context, it is unlikely that blocking occurred here. However, our task paradigm could potentially be used to study blocking in an operant conditioning context, with slight changes to the reinforcement schedule. Thus, the pattern of increased reaction times could indicate decreased processing speed capabilities in this model, requiring extra time for the rats to listen to each sound and decide whether to respond. Given evidence of slower processing speed in autism [40], this possible link to Cntnap2 should be considered in future studies.
Our result also contradicts prior findings that rats with homozygous Cntnap2 knockout exhibited increasing startle response during a pre-pulse (PPI) paradigm with age [15], suggesting an inability to anticipate upcoming events even when a predictor cue is available. This apparent contradiction may be explained by the fact that PPI tasks are often considered classical conditioning tasks, which may utilize different neural circuitry than is used for operant behaviors, as in the current task. Future research is needed to test this hypothesis about unique neural circuits based on task design and the impact of Cntnap2 knockout on these different circuits. In addition, the startling stimulus in a PPI task is often aversive (hence the startle response) whereas the target stimulus in our task was associated with a positive and appetitive reinforcer. Thus, a consideration when interpreting our findings is the influence of this gene on specific neural circuitry involved in generating and evaluating expectations for aversive versus appetitive outcomes.
Neural mechanisms for prediction
The cerebellum plays a fundamental role not only in evaluating predictions but also in adjusting to errors. For example, during walking, the cerebellum receives a copy of the movement plan and then utilizes somatosensory feedback to evaluate the execution of the plan. If an error occurs, the cerebellum registers the error and sends the signal to the cortex for a correction to be issued [41]. The removal of Cntnap2 from the rat model in our study may have selectively interfered with part of this function. During the unpredictable condition of our task, there is no known plan of action or method available to anticipate the timing of the target sound. Therefore, in this condition the auditory cortex may drive the behavior without the need for cerebellar involvement, and prior work has demonstrated that auditory cortex activation is sufficient to predict behavior in the rat during isolated and randomized speech sound discrimination tasks [25,42]. During the predictable condition however, the cerebellum may be using the predictor as the anticipated event, comparing each sound against the predictor. It is unclear, however, how the removal of Cntnap2 would lead to this enhanced ability to utilize a predictive cue.
The effect of CNTNAP2 variation on autism may be best understood at the circuit level, as supported also by its association with epilepsy and seizure activity in both humans [43,44] and rodent models [33,45]. Anatomical [46] and functional [47] connectivity are both abnormal in humans with CNTNAP2 risk variants. Mice with Cntnap2 risk alleles also exhibit abnormal neural network activity [33]. Interestingly, although CNTNAP2 is expressed in Purkinje neurons in the cerebellum, this gene has higher expression levels in frontal and temporal lobes in humans [48]. Increased expression of CNTNAP2 in both cerebellum and frontal lobe could mediate the attentional control circuit between these two brain regions and influence performance on rapid perception tasks like ours. However, this cortical bias in gene expression is not observed in the rodent brain, suggesting that the impact of CNTNAP2 variation in humans may be meaningfully different than in a rodent model. If the circuit connecting cortical language and cognitive regions to the cerebellum is functioning at increased excitability [34], it may better explain not only the enhanced reliance on predictive cues compared to the rodent model, but also the inability of humans with autism to focus attention on rapid stimuli, generate expectations in cortex, and pass those to cerebellum for monitoring.
Limitations
There are three main limitations in the current study. First, we did not collect data about brain function or anatomy. Given the known associations between Cntnap2 and cerebellar abnormalities in rodent models [14,17] as well as in humans [13,46], future work is needed to determine whether the behavioral results we present here are attributed to these published differences or another circuit. Second, we were not adequately powered for statistical analyses on possible sex differences. Though we designed the study to support sex difference analyses, two animals experienced seizures during the study period, which is a known risk associated with Cntnap2 knockout [33,45], and another failed to learn the task. Given that autism presents differently in males compared to females, future study of the biological mechanisms of autism should include well-powered comparisons of sex. Finally, human speech sounds are not ecologically relevant to rats. These stimuli have been utilized in many prior preclinical studies of human developmental disorders, including dyslexia [31,35], autism [28,29], fragile X [49], Rett syndrome [50,51], and others [21,23,24,52]. Although these prior studies have demonstrated the relevance and utility of using human speech sounds in rodent model work, we cannot say with certainty that knockout of Cntnap2 creates the same speech sound perception impairments observed in humans with autism. Thus, these results should be considered with this caveat in mind.
Conclusion
The present study utilized a rat model to determine the impact of Cntnap2 knockout on the ability to discriminate speech sounds in a predictable vs unpredictable context. We saw that Cntnap2 knockout rats did respond to distractor stimuli early in training with fully randomized presentation of the sounds but outperformed their wildtype counterparts on test trials where the target sound was preceded by a predictive cue. These findings are in line with prior work showing impaired auditory discrimination to randomly presented stimuli and, perhaps more importantly, they also present novel evidence that rats with a knockout of this autism-associated gene can utilize a predictive cue more effectively than wild-type rats, Future work is needed to probe the specific neurological mechanisms that underlie these behavioral differences and translate these observations into the human population.
Acknowledgments
The authors would like to thank Mya Conley, Alyssa Gutierrez, Sarah Grace White, and Kate Moorman-Wolfe for their assistance with behavioral training and handling of research animals.
References
- 1. Chamak B, Bonniau B. Trajectories, Long-Term Outcomes and Family Experiences of 76 Adults with Autism Spectrum Disorder. J Autism Dev Disord. 2016;46(3):1084–95. pmid:26572660
- 2. Park WJ, Schauder KB, Kwon O-S, Bennetto L, Tadin D. Atypical visual motion prediction abilities in autism spectrum disorder. Clin Psychol Sci. 2021;9(5):944–60. pmid:34721951
- 3. Baron-Cohen S. Theory of mind and autism: A review. Int Rev Res Ment Retardat. 2000;23:169–84.
- 4. Cannon J, O’Brien AM, Bungert L, Sinha P. Prediction in Autism Spectrum Disorder: A Systematic Review of Empirical Evidence. Autism Res. 2021;14(4):604–30. pmid:33570249
- 5. Gomot M, Wicker B. A challenging, unpredictable world for people with autism spectrum disorder. Int J Psychophysiol. 2012;83(2):240–7. pmid:21968196
- 6. Hellendoorn A, Wijnroks L, Leseman PPM. Unraveling the nature of autism: finding order amid change. Front Psychol. 2015;6:359. pmid:25870581
- 7. Sinha P, Kjelgaard MM, Gandhi TK, Tsourides K, Cardinaux AL, Pantazis D, et al. Autism as a disorder of prediction. Proc Natl Acad Sci U S A. 2014;111(42):15220–5. pmid:25288765
- 8. D’Mello AM, Stoodley CJ. Cerebro-cerebellar circuits in autism spectrum disorder. Front Neurosci. 2015;9:408. pmid:26594140
- 9. Stoodley CJ, Tsai PT. Adaptive Prediction for Social Contexts: The Cerebellar Contribution to Typical and Atypical Social Behaviors. Annu Rev Neurosci. 2021;44:475–93. pmid:34236892
- 10. D’Mello AM, Turkeltaub PE, Stoodley CJ. Cerebellar tDCS Modulates Neural Circuits during Semantic Prediction: A Combined tDCS-fMRI Study. J Neurosci. 2017;37(6):1604–13. pmid:28069925
- 11. Alarcón M, Cantor RM, Liu J, Gilliam TC, Geschwind DH, Autism Genetic Research Exchange Consortium. Evidence for a language quantitative trait locus on chromosome 7q in multiplex autism families. Am J Hum Genet. 2002;70(1):60–71. pmid:11741194
- 12. Canali G, Garcia M, Hivert B, Pinatel D, Goullancourt A, Oguievetskaia K, et al. Genetic variants in autism-related CNTNAP2 impair axonal growth of cortical neurons. Hum Mol Genet. 2018;27(11):1941–54. pmid:29788201
- 13. Tan GCY, Doke TF, Ashburner J, Wood NW, Frackowiak RSJ. Normal variation in fronto-occipital circuitry and cerebellar structure with an autism-associated polymorphism of CNTNAP2. Neuroimage. 2010;53(3):1030–42. pmid:20176116
- 14. Fernández M, Sánchez-León CA, Llorente J, Sierra-Arregui T, Knafo S, Márquez-Ruiz J, et al. Altered Cerebellar Response to Somatosensory Stimuli in the Cntnap2 Mouse Model of Autism. eNeuro. 2021;8(5):1–9. pmid:34593517
- 15. Scott KE, Schormans AL, Pacoli KY, De Oliveira C, Allman BL, Schmid S. Altered Auditory Processing, Filtering, and Reactivity in the Cntnap2 Knock-Out Rat Model for Neurodevelopmental Disorders. J Neurosci. 2018;38(40):8588–604. pmid:30126973
- 16. Möhrle D, Fernández M, Peñagarikano O, Frick A, Allman B, Schmid S. What we can learn from a genetic rodent model about autism. Neurosci Biobehav Rev. 2020;109:29–53. pmid:31887338
- 17. Brunner D, Kabitzke P, He D, Cox K, Thiede L, Hanania T, et al. Comprehensive Analysis of the 16p11.2 Deletion and Null Cntnap2 Mouse Models of Autism Spectrum Disorder. PLoS One. 2015;10(8):e0134572. pmid:26273832
- 18. Möhrle D, Yuen M, Zheng A, Haddad FL, Allman BL, Schmid S. Characterizing maternal isolation-induced ultrasonic vocalizations in a gene-environment interaction rat model for autism. Genes Brain Behav. 2023;22(3):e12841. pmid:36751016
- 19. Scott KE, Kazazian K, Mann RS, Möhrle D, Schormans AL, Schmid S, et al. Loss of Cntnap2 in the Rat Causes Autism-Related Alterations in Social Interactions, Stereotypic Behavior, and Sensory Processing. Autism Res. 2020;13(10):1698–717. pmid:32918359
- 20. Engineer CT, Perez CA, Chen YH, Carraway RS, Reed AC, Shetake JA, et al. Cortical activity patterns predict speech discrimination ability. Nat Neurosci. 2008;11(5):603–8. pmid:18425123
- 21. Porter BA, Rosenthal TR, Ranasinghe KG, Kilgard MP. Discrimination of brief speech sounds is impaired in rats with auditory cortex lesions. Behav Brain Res. 2011;219(1):68–74. pmid:21167211
- 22. Perez CA, Engineer CT, Jakkamsetti V, Carraway RS, Perry MS, Kilgard MP. Different Timescales for the Neural Coding of Consonant and Vowel Sounds. Cereb Cort. 2012;23(3):670–83.
- 23. Shetake JA, Wolf JT, Cheung RJ, Engineer CT, Ram SK, Kilgard MP. Cortical activity patterns predict robust speech discrimination ability in noise. Europ J Neurosci. 2011.
- 24. Ranasinghe KG, Vrana WA, Matney CJ, Kilgard MP. Neural mechanisms supporting robust discrimination of spectrally and temporally degraded speech. J Assoc Res Otolaryngol. 2012;1–16.
- 25. Centanni TM, Engineer CT, Kilgard MP. Cortical speech-evoked response patterns in multiple auditory fields are correlated with behavioral discrimination ability. J Neurophysiol. 2013;110(1):177–89. pmid:23596332
- 26. Young ED. Neural representation of spectral and temporal information in speech. Philos Trans R Soc Lond B Biol Sci. 2008;363(1493):923–45. pmid:17827107
- 27. Rauschecker JP, Scott SK. Maps and streams in the auditory cortex: nonhuman primates illuminate human speech processing. Nat Neurosci. 2009;12(6):718–24. pmid:19471271
- 28. Engineer CT, Centanni TM, Im KW, Kilgard MP. Speech sound discrimination training improves auditory cortex responses in a rat model of autism. Front Syst Neurosci. 2014;8:137. pmid:25140133
- 29. Engineer CT, Centanni TM, Im KW, Borland MS, Moreno NA, Carraway RS, et al. Degraded auditory processing in a rat model of autism limits the speech representation in non-primary auditory cortex. Dev Neurobiol. 2014;74(10):972–86. pmid:24639033
- 30. Centanni TM, Booker AB, Sloan AM, Chen F, Maher BJ, Carraway RS, et al. Knockdown of the dyslexia-associated gene Kiaa0319 impairs temporal responses to speech stimuli in rat primary auditory cortex. Cereb Cortex. 2014;24(7):1753–66. pmid:23395846
- 31. Centanni TM, Chen F, Booker AM, Engineer CT, Sloan AM, Rennaker RL, et al. Speech sound processing deficits and training-induced neural plasticity in rats with dyslexia gene knockdown. PLoS One. 2014;9(5):e98439. pmid:24871331
- 32. Centanni TM, Pantazis D, Truong DT, Gruen JR, Gabrieli JDE, Hogan TP. Increased variability of stimulus-driven cortical responses is associated with genetic variability in children with and without dyslexia. Dev Cogn Neurosci. 2018;34:7–17. pmid:29894888
- 33. Peñagarikano O, Abrahams BS, Herman EI, Winden KD, Gdalyahu A, Dong H, et al. Absence of CNTNAP2 leads to epilepsy, neuronal migration abnormalities, and core autism-related deficits. Cell. 2011;147(1):235–46. pmid:21962519
- 34. Scott R, Sánchez-Aguilera A, van Elst K, Lim L, Dehorter N, Bae SE, et al. Loss of Cntnap2 Causes Axonal Excitability Deficits, Developmental Delay in Cortical Myelination, and Abnormal Stereotyped Motor Behavior. Cereb Cortex. 2019;29(2):586–97. pmid:29300891
- 35. Centanni TM, Booker AB, Chen F, Sloan AM, Carraway RS, Rennaker RL, et al. Knockdown of Dyslexia-Gene Dcdc2 Interferes with Speech Sound Discrimination in Continuous Streams. J Neurosci. 2016;36(17):4895–906. pmid:27122044
- 36. Centanni TM, Sloan AM, Reed AC, Engineer CT, Rennaker RL 2nd, Kilgard MP. Detection and identification of speech sounds using cortical activity patterns. Neuroscience. 2014;258:292–306. pmid:24286757
- 37. Gunderson LPK, Brice K, Parra M, Engelhart AS, Centanni TM. A novel paradigm for measuring prediction abilities in a rat model using a speech-sound discrimination task. Behav Brain Res. 2024;472:115143. pmid:38986956
- 38. Scott KE, Kazazian K, Mann RS, Möhrle D, Schormans AL, Schmid S, et al. Loss of Cntnap2 in the Rat Causes Autism-Related Alterations in Social Interactions, Stereotypic Behavior, and Sensory Processing. Autism Res. 2020;13(10):1698–717. pmid:32918359
- 39. Chien Y-L, Chen Y-C, Gau SS-F. Altered cingulate structures and the associations with social awareness deficits and CNTNAP2 gene in autism spectrum disorder. Neuroimage Clin. 2021;31:102729. pmid:34271514
- 40. Zapparrata NM, Brooks PJ, Ober TM. Slower Processing Speed in Autism Spectrum Disorder: A Meta-analytic Investigation of Time-Based Tasks. J Autism Dev Disord. 2023;53(12):4618–40. pmid:36112302
- 41. Kitazawa S. Optimization of goal-directed movements in the cerebellum: a random walk hypothesis. Neurosci Res. 2002;43(4):289–94. pmid:12135772
- 42. Engineer CT, Perez CA, Carraway RS, Chang KQ, Roland JL, Sloan AM, et al. Similarity of cortical activity patterns predicts generalization behavior. PLoS One. 2013;8(10):e78607. pmid:24147140
- 43. Friedman J, Vrijenhoek T, Markx S. CNTNAP2 gene dosage variation is associated with schizophrenia and epilepsy. Molecular. 2007;13(3).
- 44. Rodenas-Cuadrado P, Ho J, Vernes SC. Shining a light on CNTNAP2: complex functions to complex disorders. Eur J Hum Genet. 2014;22(2):171–8. pmid:23714751
- 45. Thomas AM, Schwartz MD, Saxe MD, Kilduff TS. Cntnap2 Knockout Rats and Mice Exhibit Epileptiform Activity and Abnormal Sleep-Wake Physiology. Sleep. 2017;40(1):10.1093/sleep/zsw026. pmid:28364455
- 46. Dennis EL, Jahanshad N, Rudie JD, Brown JA, Johnson K, McMahon KL, et al. Altered structural brain connectivity in healthy carriers of the autism risk gene, CNTNAP2. Brain Connect. 2011;1(6):447–59. pmid:22500773
- 47. Zeeland AS-V. Altered functional connectivity in frontal lobe circuits is associated with variation in the autism risk gene CNTNAP2. Sci Translat Med. 2010;2(56):56ra80.
- 48. Abrahams BS, Tentler D, Perederiy JV, Oldham MC, Coppola G, Geschwind DH. Genome-wide analyses of human perisylvian cerebral cortical patterning. Proc Natl Acad Sci U S A. 2007;104(45):17849–54. pmid:17978184
- 49. Engineer CT, Centanni TM, Im KW, Rahebi KC, Buell EP, Kilgard MP. Degraded speech sound processing in a rat model of fragile X syndrome. Brain Res. 2014;1564:72–84. pmid:24713347
- 50. Adcock KS, Chandler C, Buell EP, Solorzano BR, Loerwald KW, Borland MS, et al. Vagus nerve stimulation paired with tones restores auditory processing in a rat model of Rett syndrome. Brain Stimul. 2020;13(6):1494–503. pmid:32800964
- 51. Engineer CT, Rahebi KC, Borland MS, Buell EP, Centanni TM, Fink MK, et al. Degraded neural and behavioral processing of speech sounds in a rat model of Rett syndrome. Neurobiol Dis. 2015;83:26–34. pmid:26321676
- 52. Ranasinghe KG, Carraway RS, Borland MS, Moreno NA, Hanacik EA, Miller RS, et al. Speech discrimination after early exposure to pulsed-noise or speech. Hear Res. 2012.