Abstract
Individuals with mutations in a single copy of the SHANK3 gene present with social interaction deficits. Although social behavior in mice depends on olfaction, mice with mutations in a single copy of the Shank3 gene do not have olfactory deficits in simple odor identification tasks (Drapeau et al., 2018). Here, we tested olfaction in mice with mutations in a single copy of the Shank3 gene (Peça et al., 2011) using a complex odor task and imaging in awake mice. Average glomerular responses in the olfactory bulb of Shank3B+/− were correlated with WT mice. However, there was increased trial-to-trial variability in the odor responses for Shank3B+/− mice. Simulations demonstrated that this increased variability could affect odor detection in novel environments. To test whether performance was affected by the increased variability, we tested target odor recognition in the presence of novel background odors using a recently developed task (Li et al., 2023). Head-fixed mice were trained to detect target odors in the presence of known background odors. Performance was tested using catch trials where the known background odors were replaced by novel background odors. We compared the performance of eight Shank3B+/− mice (five males, three females) on this task with six WT mice (three males, three females). Performance for known background odors and learning rates were similar between Shank3B+/− and WT mice. However, when tested with novel background odors, the performance of Shank3B+/− mice dropped to almost chance levels. Thus, haploinsufficiency of the Shank3 gene causes a specific deficit in odor detection in novel environments. Our results are discussed in the context of other Shank3 mouse models and have implications for understanding olfactory function in neurodevelopmental disorders.
SIGNIFICANCE STATEMENT People and mice with mutations in a single copy in the synaptic gene Shank3 show features seen in autism spectrum disorders, including social interaction deficits. Although mice social behavior uses olfaction, mice with mutations in a single copy of Shank3 have so far not shown olfactory deficits when tested using simple tasks. Here, we used a recently developed task to show that these mice could identify odors in the presence of known background odors as well as wild-type mice. However, their performance fell below that of wild-type mice when challenged with novel background odors. This deficit was also previously reported in the Cntnap2 mouse model of autism, suggesting that odor detection in novel backgrounds is a general deficit across mouse models of autism.
Introduction
There is substantial interest in the expression and function of the SHANK3 gene because of its role in the Autism Spectrum Disorders (ASD) phenotype observed in Phelan–McDermid syndrome (PMS) also known as 22q13.3 deletion syndrome (Betancur and Buxbaum, 2013; Soorya et al., 2013; Holder and Quach, 2016). SHANK3 codes for a scaffold protein in postsynaptic excitatory synapses (Monteiro and Feng, 2017) and mutations in a single copy of SHANK3 gene can produce PMS (Golden et al., 2018). Haploinsufficiency of the SHANK3 gene accounts for 1 of 50 cases of ASD, and it is the most common genetic cause of ASD (Leblond et al., 2014).
Shank3 is a complex gene that includes six protein interaction domains and transcriptional regulation results in 10 different isoforms (Wang et al., 2014). Multiple mouse models have been created that alter the expression of different subsets of Shank3 isoforms (Drapeau et al., 2014; Bozdagi et al., 2010; Peça et al., 2011; Yang et al., 2012; Duffney et al., 2015; Kouser et al., 2013) and these partial isoform mice present deficits in social behaviors, which have been compared with the autism-like phenotypes seen in PMS. Other Shank3 mouse models exhibit altered expression of all the Shank3 isoforms (Wang et al., 2016; Drapeau et al., 2018) resulting in a more severe autism phenotype.
Olfaction plays an important role in social behavior in mice and could be affecting social behavior in mouse models of PMS. However, only mice with deletion of all Shank3 isoforms (Wang et al., 2016; Drapeau et al., 2018) exhibit reduced olfactory function in simple olfactory tasks when tested as homozygous but not as heterozygous mice (Drapeau et al., 2018). Partial isoform Shank3 mouse lines (Bozdagi et al., 2010; Yang et al., 2012) also do not have reduced olfactory function in simple olfactory tests as heterozygous or homozygous.
The Shank3B mouse (Peça et al., 2011) is a widely used partial isoform Shank3 mouse line used to study the role of Shank3 in autistic behavioral phenotypes. Although most of the behavioral characterization has been done in homozygous versions of these mice, Shank3B+/− mice have slower learning in visual recognition using a touchscreen (Copping et al., 2017), as well as tactile hypersensitivity, anxiety-like behavior, and lack of interest on conspecifics (Orefice et al., 2019). Surprisingly, homozygous Shank3B−/− mice do not have olfactory deficits in a simple olfactory test (Dhamne et al., 2017).
Novel odor detection and discrimination is an understudied area with relevance to social behavior as unfamiliar stimuli/individuals can cause significant stress to individuals with ASD, who otherwise have a persistence for sameness (American Psychiatric Association, 2022). We recently developed a behavioral task in mice that allows for quantitative analysis of target odor recognition within novel odor backgrounds (Li et al., 2023). We demonstrated that this is a sensitive assay that could detect olfactory deficits in the Cntnap2−/− mouse model of autism (Peñagarikano et al., 2011), despite similar performance in wild-type (WT) mice for target odor recognition in familiar backgrounds (Li et al., 2023). These results suggest that deficits in target odor recognition within novel odor backgrounds may be a characteristic feature of mouse models of ASD, independent of other features like hyperactivity or anxiety-like phenotypes. Therefore, we tested our novel task in the Shank3B haploinsufficient mouse. Shank3B+/− mice showed odor recognition deficits in novel backgrounds that are similar to the Cntnap2−/− mice. Moreover, in vivo intrinsic imaging in awake Shank3B+/− mice indicated increased variability in olfactory receptor neuron odor responses, likely contributing to the observed behavioral phenotype. Our results have important implications on sensory deficits in neurodevelopmental disorders including the role of olfactory novelty in ASD.
Materials and Methods
Animals
All procedures were approved by the Institutional Animal Care and Use Committee of the New York Institute of Technology, College of Osteopathic Medicine (protocol 2017-GOA-0142). We used heterozygous Shank3B mice breeders (Peça et al., 2011) obtained from The Jackson Laboratory (stock #017688). Heterozygous mice of both sexes (Otazu et al., 2021) were used for all the experiments. Mice between 3 and 9 months of age were used. We used age-matched C57BL6J mice as controls (WT).
Task description
The detailed task description, including training, has been previously published (Li et al., 2023). Briefly, head-fixed, water-deprived mice were trained to perform a go/no-go task and were rewarded with water (Fig. 1A). Mice were trained with a training set of odor mixtures of targets and backgrounds and were tested with a test set of odor mixtures that included the same targets combined with novel background odors. Mice performed ∼250 trials per day. On each trial, mice were presented with 50% probability with either a go mixture or a no-go mixture. If a mouse licked the water delivery tube for the go mixture, the mouse was rewarded with a small amount (2–4 µl) of water. If a mouse licked the water delivery tube for the no-go mixture, the mouse was given a time-out.
Task used for testing odor recognition in novel odor environments. A, Water-deprived head-fixed mice were trained to lick for odor mixtures that included a target Go odor and to refrain from licking for mixtures that included a target No-Go odor. B, Mixtures used for training set trials consisted of three odors, a known background odor (1 of 4 possible odors), (S)-(-)Limonene, and the target odor. The target odor could be one of four target odors (2 possible go odors and 2 possible no-go odors). The test set was identical, but (s)-(-)limonene was replaced with one of the 11 novel background odors. C, Animals were presented with the test set trial as catch trials as they performed trials with the training set.
The training set consisted of 8 different mixtures and the test set consisted of 88 different mixtures. A training-set mixture consisted of combinations of three different odors—a target odor and two background odors (Fig. 1B). The target odor was presented at a low concentration of 0.025% of saturated vapor, and the background odors were presented at a higher concentration of 0.1% of saturated vapor. The target odor determined whether the mixture was a go mixture or a no-go mixture. For the go mixtures, the target odor could be either isopropyl butyrate or propyl butyrate, and for the no-go mixtures the target odor was either isobutyl propionate or ethyl propionate. One of the background odors was a contextual background odor, selected from a group of four different odors. Contextual background odors were the same for both the training set mixtures and the test set mixtures. The second background odor was always (s)-(−)-limonene for the training set mixtures. The test set mixtures included the same target and contextual background odors. However, (s)-(−)-limonene was replaced by one of 11 possible novel background odors. Although the contextual background odors and the target odors were the same for the training set and the test set, the combinations of targets and contextual background odors were different between the training set and the test set (Li et al., 2023, Reduced Training Set).
Naive water-deprived mice were trained for 10–12 d before reaching optimal performance on the training set. Once a performance threshold of >80% was achieved for at least 50 trials, animals were tested with test set trials with novel background odors. The test set trials were presented randomly interleaved with training set trials, representing only 27.5% of the all the trials in a session. Novel background odors were presented over two days, with 88 total trials per animal.
Olfactory stimulus
The detailed description of the odor delivery system has been previously published (Li et al., 2023). Odors were delivered asynchronously. Odor stimulus started with the delivery of the contextual background odor, followed after 750 ms by (s)-(−)-limonene. After 750 ms, (s)-(−)-limonene was followed by the target odor. For the test set mixtures, (s)-(−)-limonene was replaced by one of the 11 novel background odors.
Imaging
Intrinsic optical imaging of the olfactory bulb (Meister and Bonhoeffer, 2001) was done as previously described (Li et al., 2023) using a tandem single-lens reflex system. Briefly, awake naive animals were presented with the 20 odors at the concentrations used during the behavior, four target odors at 0.025% of vapor pressure and 16 background odors at 0.1% of vapor pressure [four contextual background odors, 11 novel background odors, and (s)-(−)-limonene)]. The olfactory bulb was illuminated using a 780 nm light source. Activated glomeruli appear as reductions in the amount of reflected light. Z-scores were calculated using as baseline the 5 s preceding the odor onset. Odors were presented for 9 s. and the response was quantified as the average z score in the window between 2 and 9 s following odor onset.
Sniff recording
Sniff responses were recorded using an airflow sensor (1000Sccm AWM300V, Honeywell) connected opposite to the animals snout as previously described (Li et al., 2023). Briefly, we determined the change in sniff rate produced by novel background odors by counting the number of inhalations in a 1 s window between 300 and 1300 ms after the onset of the novel odor or the (s)-(−)-limonene odor. To quantify the sniff response to the novel background odors, we compared the presentation of mixtures with novel background odors to the preceding trial with known background odors to correct for potential nonstationarity of the sniff rate.
Surgery
Detailed surgical details have been previously described (Li et al., 2023). Briefly, animals that were used for behavioral experiments were anesthetized and implanted with titanium head bars. Animals that were used for imaging were also implanted with a 3 mm glass coverslip over the olfactory bulb. Animals recovered for at least 1 week before water deprivation or imaging occurred.
Image analysis
Image analysis was performed as previously described (Li et al., 2023). Briefly, the images obtained from the same odor were combined by averaging them over multiple repetitions (>16). To assess the response of the average image, a normalized signal df/f0 was computed. The baseline f0 was determined as the average response during the 5 s period preceding the onset of the odor. To eliminate the broad hemodynamic signal caused by odor response in intrinsic images, we used a spatial Gaussian filter with a radius of σ = 40 µm. The resulting spatially low-pass-filtered signal was then subtracted from the original signal. Additionally, to eliminate spatial noise with high spatial frequencies, the resulting images were further convolved with a Gaussian filter with a radius of σ = 12 µm. To normalize the signal, a mean z score was calculated for each pixel using the df/f0 values from the 5 s before odor onset to compute the mean and SD. Average z score odor responses were the average of the z score calculated over a period between 2 and 9 s following odor onset. Regions of interest (ROIs) corresponding to activated glomeruli were manually delineated in ImageJ (Schindelin et al., 2015) software using the average z score odor responses. To identify activated glomeruli, the minimal projection across all odors was used. Glomerular activation was quantified by calculating the mean value of the z scores across all selected pixels within an ROI.
To calculate the z score for individual trial responses, images from single trials were spatially filtered as described above. The average across all selected pixels within each ROI was calculated, and the z score was calculated using the 5 s before odor onset as baseline.
Estimation of the coefficient of variation
Estimation of the coefficient of variation (CV) was performed as previously described (Li et al., 2023). We computed the mean (μ) and total variance (σ2) of the glomerular response using the z score for individual trial responses. The total variance increased as the mean value of the evoked response increased, consistent with findings from a previous study (Mathis et al., 2016). There was also a baseline level of variability for the responses of ROI-odor pairs, which was independent of the average activation and even present in cases where the average responses were zero or positive. This baseline variability can be attributed to imaging-related noise.
The CV relates the average odor response to the observed standard deviation as follows:
Estimation of the uncorrelated coefficient of variation
The estimation of the uncorrelated coefficient of variation (CVuncorr) was conducted following the procedure previously described (Li et al., 2023). Part of the trial-to-trial variability for a given odor was shared among different glomeruli. This means that when one glomerulus produced a larger-than-average odor response in a trial, other glomeruli also tended to show larger-than-average responses in the same trial (Mathis et al., 2016).
To determine the coefficient of variation of the uncorrelated variability, we used a method previously developed (Mathis et al., 2016). In this approach, for each odor presentation, we plotted the average response of a specific ROI-odor pair across multiple trials against the response for that particular odor presentation. For each odor presentation, a line was fitted to all the simultaneously recorded ROI responses. The population correlated fluctuations were observed to follow this line because they were proportional to the average response of each glomerulus. Deviations from the fitted line represented the contribution of uncorrelated noise for that ROI in that odor presentation.
To quantify the variance of the uncorrelated noise across trials for each ROI-odor pair, we calculated the variance
However, even in intrinsic imaging, there still existed uncorrelated variance in the absence of an average odor response, which was attributed to imaging noise and denoted as
Least absolute shrinkage and selection operator deconvolution
Least absolute shrinkage and selection operator (Lasso) deconvolution, as previously described (Li et al., 2023), was conducted in the following manner. Initially, we constructed extensive dictionaries of glomerular activation patterns based on our glomerular imaging. These dictionaries encompassed the glomerular activation patterns associated with the nine individual odors that were part of the training set, which consisted of four target odors, four contextual background odors, and (s)-(−)-limonene. However, the dictionaries did not include the 11 novel background odors as the animals had not been exposed to them during training. In addition to the training set odors, the dictionaries contained glomerular patterns for other odors. To incorporate novel elements into the dictionaries while maintaining the average activation per glomerulus and interglomerular correlation, we calculated the mean and covariance matrix using the responses to all 20 odors used in the experiments. Subsequently, we generated the additional dictionary elements using a Gaussian process, using the calculated mean and covariance matrix. Despite imaging the odor responses at the concentrations used during the behavioral experiments, each dictionary element was normalized to have unit variance. We used dictionary sizes ranging from 100 to 1000 elements. The Lasso algorithm was implemented using the Lasso function from MATLAB 2017B (MathWorks), with a regularization constant (λ) set to 0.0001. For each dictionary size, we simulated 30 different dictionaries and assessed the performance of each dictionary using 13,200 test mixtures, which were composed of 88 reduced test set mixtures combined with the virtual odor mixture.
Experimental design and statistical analysis
For the imaging experiments, we used three Shank3B+/− mice (two males and one female) and five WT (female mice). For the behavioral experiments we used eight Shank3B+/− mice and six WT mice. The behavioral data from three female WT mice and imaging data from WT mice have been previously used (Li et al., 2023). Comparisons of the glomerular response strength were performed using a binary test over the 20 odors imaged. We also used linear correlation analysis to compare the correlation between glomerular responses between the two genotypes. Behavioral comparisons were done using Fisher's exact test in MATLAB 2017.
Results
Shank3B+/− mice glomerular responses had higher trial-to-trial variability compared with WT mice
The olfactory bulb receives inputs from axons of receptor neurons of the olfactory epithelium. Receptor neurons that express the same olfactory receptor gene converge into individual glomeruli in the olfactory bulb. We wondered whether the odor responses of the olfactory bulb input of Shank3B+/− mice were different from the responses of WT mice and whether these differences could lead to olfactory deficits. The Shank3b isoform is preferentially expressed in the glomerular layer of the olfactory bulb (Fig. 2A for RNA; Wan et al., 2021 for antibody staining), so reduction in the Shank3B expression might affect glomerular responses. Although for Shank3B homozygous mice, olfactory discrimination using simple odors is not different from WT mice (Dhamne et al., 2017); odor recognition in the presence of background odors is a more difficult task that might reveal a deficit/phenotype. Peripheral odor responses and their variability could affect target odor recognition in the presence of background odors. Differences in peripheral sensory activity have been shown to underlie some of the somatosensory phenotypes in Shank3B+/− mice (Orefice et al., 2019).
Glomerular responses measured using intrinsic imaging were more variable in Shank3B+/− mice than in WT mice. A, Expression of the Shank3b (ankyrin domain) in the glomerular layer of the olfactory bulb from in situ hybridization from the Allen Brain Atlas. B, Maps of average glomerular activation from hemibulbs from a Shank3B+/− and a WT mouse quantified using z score. C, D, Glomerular responses could be observed on individual trials. E, Average glomerular responses versus trial-by-trial variability for Shank3B+/− and WT mice. Purple line indicates mean fitted trial-to-trial variability and dotted lines are the 95% confidence intervals. For both genotypes, the variability increased with the average response.
We used intrinsic optical imaging (Meister and Bonhoeffer, 2001) using the setup described in (Li et al., 2023) to measure odor-evoked responses in the dorsal olfactory bulb to a panel of 20 odors, presented at the concentrations that would be used for behavioral testing. Intrinsic imaging was performed in five naive WT mice and three Shank3B+/− mice that were awake but passively exposed to 9 s odor pulse. Odors were repeated at least 20 times.
Responsive glomeruli produced reductions in the reflectance of 780 nm light and appeared as dark spots even in single odor presentations in WT mice and in the Shank3B+/− mice (Fig. 2B). Odor-evoked responses were quantified using z scores with the 5 s before odor presentation as baseline. Odor responses were quantified as the average of the z score during the odor period, a window of 2–9 s following odor onset. We recorded from 155 ± 38.1 ROIs (mean ± SD) per WT mouse (775 glomeruli total) and from 121.7 ± 36.3 ROIs (mean ± SD) per Shank3B+/− mouse (365 glomeruli total).
Glomerular responses were different in different presentations of the same odor (Fig. 2C,D). Part of this variability results from imaging noise, and the other part of the variability corresponds to intrinsic sensory variability. The effect of the imaging noise on the estimation of the average glomerular response to an odor can be mitigated by increasing the number of averaged trials. The other part of the variability corresponds to the intrinsic sensory variability, which does affect the mouse olfactory performance as animals make decisions based on individual presentations and cannot average responses over multiple odor presentations.
To separate the imaging noise from the intrinsic sensory variability, we plotted the average glomerular response versus the SD of the trial-by-trial response (Fig. 2E). The variability σnoise relates to the imaging noise and appears in the absence of an average odor-evoked response. This variability (σnoise = 1.59) was similar between the WT and the Shank3B+/− mice. Given that we used at least n = 16 repeats for each odor, the threshold for the z score glomerular responses that we could directly detect was
The SD of glomerular responses over trials increased with the average glomerular response, reflecting the intrinsic variability of the sensory response as previously described (Mathis et al., 2016). The constant of proportionality between the average glomerular response and the SD of glomerular responses is given by the CV. We estimated CV from the glomerular imaging data as previously described (Li et al., 2023). Shank3B+/− mice glomerular responses had a higher CV (0.96 with 95% CI = 0.89, 1.06) compared with WT mice (0.34 with 95% CI = 0.31, 0.36), meaning that their glomerular responses were more variable than those of WT mice. This increased trial-by-trial variability could affect odor recognition in Shank3B+/− mice.
Some of the trial-by-trial variability was shared among glomeruli, meaning that trials that produced large activity for one glomerulus produced large activity in other glomeruli (Fig. 3A,B). The other part of the variability was uncorrelated among glomeruli, meaning that correlations were not shared among glomeruli. The shared variability could be compensated by normalizing the response of a given glomerulus on a given trial using the population response of all the glomeruli on that trial (Mathis et al., 2016). In contrast, the uncorrelated variability could not be compensated. We extracted the uncorrelated variability using a previously developed method (Mathis et al., 2016). Briefly, for each odor trial and for all the recorded glomeruli, we plotted the average response calculated for all the trials against the response on that particular odor trial. If the variations on an odor trial were correlated among all the glomeruli, and if the variation of each glomerulus was proportional to its average response, all the responses would fall in a straight line with the correlated variability causing the line to have different slopes on different trials (Fig. 3C). Deviations from the line are caused by the uncorrelated variability. For each glomerulus and each odor, we calculated the variance (σ2uncorr) of these deviations for all the trials. We plotted the average z score (μ) against this variance. The σ2uncorr increased for larger average glomerular responses (μ), with the constant given by coefficient of variation (
Trial-by-trial variability of glomerular responses were partially correlated between glomeruli. A, Example map of average glomerular activation of a hemibulb in a Shank3B+/− mouse, quantified using z score, in response to ethyl butyrate at a concentration of 0.1% of saturated vapor. B, Example of the trial-by-trial responses for individual trials for ROI 13 versus ROI 23. There was a weak linear correlation (R = 0.4, p = 0.036, n = 27 repeats) indicating that trials that produced strong activation in one glomerulus would tend to produce strong activation in the other glomerulus. C, Example of the average response against the single-trial response for two different trials in response to ethyl butyrate. The average z score responses of 80 simultaneously recorded glomeruli were plotted against the response of the same glomeruli in individual trials (trial 1 and trial 25). The single-trial response could be fitted by a line. The slope of the line was different on different trials, which was caused by the correlated variability across the population. Deviations from the line corresponded to the uncorrelated variability. D, Average glomerular responses versus the SD of the trial-to-trial uncorrelated variability. There was a nonzero variability at zero average glomerular responses that corresponded to the uncorrelated variability produced by the imaging noise.
We wondered whether the increased variability in the odor-evoked responses was also present in the spontaneous activity. We calculated the coefficient of variation using the baseline signal on the 5 s preceding odor onset. For each ROI, we calculated the average coefficient of variation across all 5 s spontaneous periods (∼400 periods per ROI). The coefficient of variation for the Shank3B+/− mice was 0.29 ± 0.07 (mean ± SEM, n = 365 ROIs), and it was not significantly different (p = 0.60, double-tailed t test) from the coefficient of variation of WT mice (0.33 ± 0.03, mean ± SEM, n = 543 ROIs).
The increased uncorrelated variability of the odor-evoked glomerular responses in the Shank3B+/− mice could potentially affect odor recognition in novel environments.
WT and Shank3B+/− mice average glomerular representations were correlated
We wondered whether the average glomerular responses in Shank3B+/− mice were different from those of WT mice. We compared the odor responses for the odor at the concentrations used for testing (four target odors at 0.025% of vapor pressure and 16 background odors at a stronger concentration of 0.1% of vapor pressure). The average glomerular activation per odor (Fig. 4A) was very similar between WT (5 animals, 775 glomeruli) and Shank3B+/− mice (3 animals, 365 glomeruli) for all odors tested (p > 0.1, double-tailed t test). Odors that produced strong average glomerular activation patterns in the Shank3B+/− mice also produced strong average glomerular activation patterns in the WT mice. There was a significant linear correlation (r = 0.64, p = 0.0026, Pearson linear coefficient) between the average glomerular response for a given odor between Shank3B+/− mice and WT mice (Fig. 4B).
Average odor responses were correlated between WT and Shank3B+/− mice. A, Average glomerular responses for WT mice and Shank3B+/− mice for a panel of 20 odors used during the behavior, with each odor presented at the concentration used during the testing phase (target odors, 0.025% of vapor pressure; background odors, 0.1% of vapor pressure). Symbols correspond to average responses from individual animals. Error bars are s.e.m. B, Average evoked responses were correlated between WT mice and Shank3B+/− mice. C, Fraction of glomeruli whose activity exceeded the threshold per odor. D, The fraction activated per odor was correlated across genotypes.
The fraction of glomeruli that responded to a given odor and exceeded the detection threshold (z_score ≤ 0.42) was also similar between WT and Shank3B+/− mice (Fig. 4C). Of the 20 odors, only hexanal, an odor used as novel background odor, activated a significantly larger (p = 0.03, double-tailed t test) fraction of glomeruli in WT mice (0.53 with 95% CI = 0.50, 0.56) compared with Shank3B+/− mice (0.35 with 95% CI = 0.31, 0.39) with the rest of odors having no differences in the fraction of glomeruli activated (p > 0.07, double-tailed t test). Odors that activated a large fraction of glomeruli in the Shank3B+/− mice also activated a large fraction of glomeruli in the WT mice. There was a significant linear correlation (r = 0.70, p = 6e-4, Pearson linear coefficient) between the fraction of glomeruli activated for a given odor between Shank3B+/− mice and WT mice (Fig. 4D). Although comparing the fraction of glomeruli that responded to individual odors between WT and Shank3B+/− mice only reached significance for hexanal, there was a general trend in that odors activated a larger fraction of glomeruli for WT compared with Shank3B+/− mice (p = 0.0019, double-tailed t test, n = 20 odors).
The smaller fraction of glomeruli activated by odors in Shank3B+/− could potentially affect performances for odor recognition in novel environments.
Shank3B+/− mice glomerular variability affected performance of a deconvolution algorithm
Average glomerular responses were correlated between WT and Shank3B+/− mice, but odors activated a smaller fraction of glomeruli in Shank3B+/− mice. Shank3B+/− mice glomerular responses also had higher trial-to-trial variability compared with WT mice, which might negatively affect odor identification in novel environments, a computationally complex task. Simple algorithms (Mathis et al., 2016; Rokni et al., 2014; Dasgupta et al., 2017) that have been proposed as being implemented by the olfactory system using feedforward connections (Haberly and Price, 1977) are not able to solve our novelty task. In contrast, Lasso (Tibshirani, 1996), a deconvolution method, could perform well above chance for odor recognition in the presence of novel background odors using WT and Cntnap2-/− glomerular data (Li et al., 2023). The Lasso deconvolution determines the combination of odors from a large dictionary that could account for observed patterns of glomerular activation produced by a mixture, which is a method that has been proposed as being implemented by the nervous system (Koulakov and Rinberg, 2011; Grabska-Barwińska et al., 2017; Li and Hertz, 2000; Otazu and Leibold, 2011). The large dictionary reflects all the odors that are known to the animal.
To test whether the Lasso deconvolution performance was affected by the increased trial-to-trial variability and the smaller fraction of activated glomeruli in the Shank3B+/− mice, we simulated the Lasso using the glomerular imaging from individual odors from Shank3B+/− mice (Fig. 5). Lasso deconvolution needs a dictionary that includes the glomerular representation of all the odors that are known to the animal. We do not have access to such a dictionary, so we have simulated the response of the Lasso using dictionaries with different numbers of elements. As the dictionary represents the odors that are known to the animal, all the dictionaries tested included the responses for the target odor (four odors), the known background odors (four contextual background odors, and (s)-(−)-limonene) because the animals were exposed to them during the training phase (Fig. 1). These dictionaries did not include the 11 novel background odors because the animals were exposed only during testing and not during training. As the dictionary should reflect all the odors that are known to the animal, we simulated additional dictionary elements using a Gaussian process with the mean and the covariance from the actual measured odors. To create the mixture that the animal would be required to discriminate, we added the glomerular representation of the target and the background odors and used an experimentally measured saturating nonlinearity as previously described (Li et al., 2023).
Higher trial-to-trial variability in Shank3B+/− mice affected performance of Lasso deconvolution. A, Example of average glomerular responses to propyl butyrate (a go target odor presented at a concentration of 0.025%), ethyl valerate (a contextual background odor presented at a concentration of 0.1%), and acetal (a novel background odor presented at a concentration of 0.1%). The combinations of these three odors would create a mixture of the test set. The odor mixture was simulated by adding the glomerular representations of the individual odors and passing the result of the additions through an experimentally measured saturating nonlinearity. The variability of a single-odor presentation was simulated by adding uncorrelated Gaussian noise to each glomerulus with an amplitude proportional to the glomerular response, with the proportionality constant being given by the uncorrelated coefficient of variation
We simulated the performance of 10 dictionary sizes between 100 and 1000 elements. Performance of the Lasso using Shank3B+/− mice average glomerular odor responses with Shank3B+/− mice high variability (
The increased trial-to-trial variability of the Shank3B+/− mice caused a degradation in the performance of the Lasso, a deconvolution algorithm that can robustly identify target odors in novel environments.
Shank3B+/− mice could learn to identify target odors in the presence of background odors
We trained eight Shank3B+/− mice (five males and three females) and six WT mice (three males and three females) for target odor recognition in the presence of background odors. Shank3B+/− mice have slower learning in operant conditioning on a visual discrimination task using a touchscreen (Copping et al., 2017). We wondered whether Shank3B+/− mice could even learn to identify odors in known background odors and whether their olfactory learning rates would be slower than those of WT mice. The purpose of our training was to have mice detect weak target odors in the presence of strong background odors. This was achieved first by training animals to discriminate the pure target odors at high concentrations without any background odors. Over ∼5 d, the concentration was reduced until reaching the final target concentration of 0.025%. After that, background odors were introduced at a low concentration. Over 5 d, the background concentration was increased until reaching the final background concentration of 0.1%.
On the first day of training, both WT and Shank3B+/− mice could discriminate between isobutyl propionate (0.34%, no-go stimulus) and isopropyl butyrate (0.34%, go stimulus). Learning of this first odor pair depended not only on olfactory discrimination per se but also on the reaction of the animal to head restraint and on learning to lick to the water tube. To separate olfactory discrimination learning from these other sources of behavioral variation, we analyzed the learning curve for a second odor pair (ethyl propionate at 0.34%, no-go stimulus; propyl butyrate, 0.34%, go stimulus) on day 3 after animals have become habituated to the head restraint and have already learned to lick in response to odors (Fig. 6A). The performance of the Shank3B+/− mice reached a plateau after 50 trials. The performance on trials 50–100 was 67.3% (0.95 CI = 59.3, 74.5%; 350 trials, eight Shank3B+/− mice). This performance was significantly lower than the performance reached by WT mice (82.4%; 0.95 CI = 73.6, 89.2%; 672 trials, 6 WT mice, p = 0.0091, Fisher's exact test). The slower learning of the Shank3B+/− mice is consistent with previous reports of learning deficits in the Shank3B-/− mice (Dhamne et al., 2017).
Shank3B+/− mice and WT mice learned to recognize a target odor in background odors. Solid lines represent the mean performance calculated using a 60-trial sliding window, and dotted lines represent the mean ± SEM calculated for eight Shank3B+/− and six WT mice. A, Learning curves for the second pair of go target and no-go target odors without backgrounds at a high concentration (0.1%) after having already learned to discriminate the first pair of target odors. B, Learning curves for the first exposure of the standard background odors at 0.025% (20% of the final concentration) with the targets at their final concentration (0.025%). C, Learning curves for the first exposure of the standard background odors at the final concentration of the background odors (0.1%) with the targets at a concentration of 0.025%.
However, as the training progressed and animals were exposed to the background odors for the first time, the learning deficits of the Shank3B+/− mice ameliorated (Fig. 6B). When target odors were presented for the first time together with low concentration of background odors (0.025% of vapor saturation), Shank3B+/− mice outperformed WT mice in discriminating between go target odors (isopropyl butyrate and propyl butyrate, 0.025% of saturated vapor) from no-go target odors (isobutyl propionate and ethyl propionate, 0.025% of saturated vapor). The performance on trials 50–100 was 84.0% for Shank3B+/− mice (0.95 CI = 78.0, 89.0%; 188 trials, 8 Shank3B+/− mice) which was significantly higher than the performance of WT mice (65.7%; 0.95 CI = 59.0, 72.1%; 216 trials, 6 WT mice, p = 3.5e-5, Fisher's exact test).
When animals were exposed to the background odor at the final concentration (0.1% vapor pressure) for the first time, the learning of the WT and Shank3B+/− mice became very similar, with mice performing large number of trials at performances above criterion (Fig. 6C). The performance on trials 100–150 was 83.6% for Shank3B+/− mice (0.95 CI = 78.1, 88.2%; 225 trials, 8 Shank3B+/− mice), which was not significantly different from WT mice (83.1%; 0.95 CI = 78.0, 87.5%; 255 trials, 6 WT mice, p = 1, Fisher's exact test). Although Shank3B+/− mice had slower initial learning for pure odors than WT mice, they outperformed WT mice in learning to identify odors in weak background odors. Shank3B+/− mice could be trained to accurately discriminate weak targets in the presence of strong background odors, reaching a similar performance as WT mice, permitting us to test their performance with novel background odors.
Shank3B+/− mice had deficits in odor recognition in the presence of novel background odors
The performance of Shank3B+/− mice on the standard background odors, just before the presentation of the novel odors started, was 90.4% (0.95 CI = 88.1, 92.4%; 791 trials, 8 Shank3B+/− mice), which was not significantly different (p = 0.13, Fisher's exact test) from the performance of the WT mice (87.8%; 0.95 CI = 85.1, 90.2%; 672 trials, 6 WT mice; Fig. 7A). There was no significant difference in the performance between males and females for WT mice (WT males, 90.3%; 0.95 CI = 86.4, 93.4%; 300 trials; WT females, 85.5%; 0.95 CI = 81.8, 89.1%; 372 trials, p = 0.0761, Fisher's exact test) nor for Shank3B+/− mice (Shank3B+/− males, 89.6%; 0.95 CI = 86.6, 92.1%; 500 trials; Shank3B+/− females, 91.8%; 0.95 CI = 88.0, 94.6%; 291 trials, p = 0.3814, Fisher's exact test).
Shank3B+/− mice behavioral discrimination of targets was selectively affected in novel backgrounds. A, Performance and 95% confidence interval for eight Shank3B+/− mice (n = 791 trials) and six WT mice (n = 672 trials) on known backgrounds for the reduced training set. Circles indicate performances of individual female mice, with plus signs indicating individual male performances. Significance was calculated using a two-tailed Fisher's exact test. B, Performance in novel background for the reduced test set for Shank3B+/− mice (n = 519 trials) and WT mice (n = 432 trials); p values were calculated with a two-tailed Fisher's exact test. C, Fraction of total errors for known background odors. Bar indicates mean ± SEM for the error fraction per animal (6 WT mice and 8 Shank3B+/− mice). The most common type of error for both phenotypes were false alarms caused by erroneously licking for the no-go mixture. D, Fraction of total errors for novel background odors. The most common type of error for both phenotypes were misses, caused by lack of licking for the go mixture. E, Fraction of errors for go stimuli for known (blue bars) and novel (red bars) background odors. The fraction of errors for go stimuli increased for both phenotypes for novel odors compared with known background odors. F, Fraction of errors for the no-go stimuli. WT mice had reduced fraction of errors for the no-go stimuli, whereas the Shank3B+/− mice had an increased fraction of errors for the no-go stimuli with novel background odors.
However, when tested with novel background odors (Fig. 7B), Shank3B+/−mice performance dropped to 62.4% (0.95 CI = 58.1, 66.6%; 519 trials, 8 Shank3B+/− mice), which was significantly lower (p = 3.46e-8, Fisher's exact test) than the performance of WT mice (72.0%; 0.95 CI = 67.5, 76.2%; 432 trials, 6 WT mice). There was no significant difference in the performance between males and females in novel background odors for WT mice (WT males, 69.6%; 0.95 CI = 62.8, 75.8%; 204 trials; WT females, 74.1%; 0.95 CI = 67.9, 79.7%; 228 trials, p = 0.3185, Fisher's exact test) nor for Shank3B+/− mice (Shank3B+/− males, 60.3%; 0.95 CI = 54.2, 66.2%; 272 trials; Shank3B+/− females, 64.8%; 0.95 CI = 58.5, 70.7%; 228 trials, p = 0.3185, Fisher's exact test). Thus, Shank3B+/−mice had a selective deficit for target recognition in the presence of novel background odors.
Shank3B+/− mice reduced performance was not caused by hypoactivity
Homozygous Shank3B-/− mice are hypoactive (Dhamne et al., 2017). We wondered whether hypoactivity might cause the errors of the heterozygous mice Shank3B+/−. Hypoactivity might cause a lack of licking, resulting in a larger fraction of total errors being misses (not licking for the go stimuli) compared with false alarms (licking for the go stimulus) or early licks. We quantified the fraction of total errors that were misses (fmiss). The most common type of error made by Shank3B+/− and WT mice in known background odors were not misses but false alarms (Fig. 7C), similar to other go/no-go tasks (Rokni et al., 2014; Kuchibhotla et al., 2019).
For the known background odors, the WT mice fmiss was 12.5% (0.95 CI = 6.2, 21.8%; 80 error trials), and it was not significantly different from fmiss of Shank3B+/− mice (7.9%; 0.95 CI = 3.0, 16.4%; 76 error trials, p = 0.43, Fisher's exact test). In contrast, the most common type of error in the presence for novel background odors for the Shank3B+/− mice were misses, consistent with hypoactivity playing a role (Fig. 7D). However, the most common type of error for WT mice were also misses. Shank3B+/− mice had significantly lower fractions of misses compared with WT mice. Shank3B+/− mice fmiss was 60.5% (0.95 CI = 53.3, 67.4%; 195 error trials), and it was significantly lower than the fmiss of WT mice (81.0%; 0.95 CI = 72.9, 87.6%; 121 trials, p = 1.72e-4, Fisher's exact test). Thus, hypoactivity does not seem to play a role in the lower performance of Shank3B+/− mice in odor recognition in novel background odors.
Shank3B+/− mice performance with novel background odors was reduced for both go-stimuli and no-go stimuli
The most common type of error in the presence of novel background odors for the Shank3B+/− and WT mice were misses, whereas false alarms were the most common type of error for known background odors. Indeed, there was a significantly increased fraction of go stimuli trials without lick responses when novel background odors were present (Fig. 7E) compared with stimuli with known background odors for both WT (known, 3.5%; 0.95 CI = 1.7, 6.4%; 285 go trials; novel, 49.0%; 0.95 CI = 41.9, 56.2%; 200 go trials, p = 3.49e-34, double-tailed Fisher's exact test) and Shank3B+/− mice(known, 1.5%; 0.95 CI = 0.6, 3.3%; 390 go trials; novel, 42.4%; 0.95 CI = 35.3, 49.8%; 191 go trials, p = 5.90e-38, double-tailed Fisher's exact test). We wondered whether the low performance of the Shank3B+/− mice was because of their reluctance to lick when exposed to a novel background odor, potentially caused by their higher anxiety levels (Dhamne et al., 2017). If the Shank3B+/− mice indeed had a general reluctance to lick when exposed to novel background odors, they should also have reduced licking responses to no-go stimuli. However, the Shank3B+/− mice actually showed a significantly increased proportion of incorrect licking responses (false alarms) to the no-go stimuli with novel background odors (Fig. 7F; known, 9.3%; 0.95 CI = 6.5, 12.8%; 365 no-go trials; novel, 22.5%; 0.95 CI = 16.9, 28.9%; 200 no-go trials, p = 3.7887e-05, double-tailed Fisher's exact test). In contrast, WT mice had a decreased proportion of incorrect licking responses to the no-go stimuli with novel background odors (known, 13.9%; 0.95 CI = 10.2, 18.2%; 310 no-go trials; novel, 7.9%; 0.95 CI = 4.8, 12.24; 227 no-go trials, p = 0.039, double-tailed Fisher's exact test). Thus, the novel background odors reduced Shank3B+/− mice performance for both the go stimuli and no-go stimuli.
Shank3B+/− and WT mice increased their sniff rate for the first presentation of a novel background odor
We wondered whether the low performance of Shank3B+/− mice in the presence of novel background odors might be caused by the reduced exploration of novel background odors. Shank3B-/− mice do not explore novel environments compared with WT mice (Dhamne et al., 2017; Yang et al., 2012). Mice increase their sniff rates for novel odors (Verhagen et al., 2007). In our head-fixed behavior, WT mice explored the first presentation of a novel background odor by increasing their sniff rate compared with their sniff rate in response to mixtures with known background odors (Fig. 8A, example trace). However, Shank3B+/− mice also explored the first presentation of a novel background odor by increasing their sniff rate (Fig. 8B,C). We quantified the sniff response to the novel background odor by counting the number of inhalations in a 1 s period following the onset of the novel background odors. We also quantified the sniff responses to known background odors using a similar 1 s window following the onset of (s)-(−)-limonene. On the first presentation of the day for a novel background odor, both WT (4.9 ± 0.19 sniffs per second, n = 110 trials) and Shank3B+/− mice (5.7 ± 0.2 sniffs per second, n = 130 trials) had higher sniff rates compared with the sniff rates for known background odors (WT, 3.6 ± 0.04 sniffs per second, n = 1340 trials; Shank3B+/−, 3.8 ± 0.05 sniffs per second, n = 1164 trials). The differences between sniff rates for novel and known background odors were significant (WT, p = 1.67e-13; Shank3B+/−, p = 5.67e-28; double-tailed t test). In fact, the sniff rate for novel background odors was significantly higher for the Shank3B+/− mice compared with the WT mice (p = 0.0047; double-tailed t test). Shank3B+/− mice explored the novel background odors as much as the WT mice.
Shank3B+/− and WT mice explored first presentations of novel background odors by increasing their sniff rate. A, Example sniffing signal from a WT mouse. The presence of a novel background odor in a test set mixture induced an increased sniff rate compared with the sniff rate for a training set mixture that included only known background odors. B, Similar example for a sniffing response from a Shank3B+/− mouse. C, Mean ± SEM of the sniff responses for eight Shank3B+/− and sux WT mice for standard background odors and for the first presentation of each day for novel background odors. D, Mean ± SEM of the increase in sniff rate for WT mice for novel background odors compared with preceding trial with (s)-(−)-limonene as a function of the novel background odor presentation number (8 presentations total over 2 d, n = 66 trials per presentation); p values were calculated using the Bonferroni-corrected two-tailed t test. Increase in sniff rate adapted after a second presentation of the novel background odors. E, Shank3B+/− mice had a similar adaptation of their sniff rate (n = 88 trials per presentation).
Shank3B+/− and WT mice sniff responses similarly adapted for subsequent presentations of a novel background odor
We wondered whether odor memory was affected in Shank3B+/− mice. Memory deficits have been reported in Shank3 mice (Drapeau et al., 2018; Wang et al., 2011). Mice elevate their sniffing responses for novel odors, and this elevation was mitigated after repeated exposure (Wesson et al., 2008). Although the WT and Shank3B+/− mice both elevated their sniffing on the first presentation of a novel background odor with respect to the known background odor, their sniffing rates returned to a sniffing rate that was not significantly different from the sniff rate in response to the known background odors after a single exposure (Fig. 8D,E). The elevated sniff rate for novel background odors was not present on the second day, suggesting that both WT and Shank3B+/− mice retained a memory of the novel background odor for at least 1 d. Thus, exploratory behavior of the novel background odor was similar between WT and Shank3B+/− mice.
Discussion
In the present report, we characterize the glomerular responses and the behavioral performance of odor recognition in novel environments in Shank3B+/− mice. Shank3B+/−mice had average glomerular responses similar to those of WT mice but showed increased trial-to-trial variability. The increased trial-to-trial variability of the Shank3B+/−mice reduced the performance of the Lasso deconvolution algorithm used to detect targets in novel backgrounds. Shank3B+/−mice had a selective deficit for target odor recognition in the presence of novel backgrounds. In contrast, target odor recognition in known backgrounds was not different between WT and Shank3B+/− mice. The lower performance of the Shank3B+/− mice was not caused by hypoactivity nor by lack of sniff exploration of the novel background odors.
The presence of novel background odors increased the overall error rate in both Shank3B+/− and WT mice. However, WT mice exhibited increased error rates only for the go stimuli when exposed to novel background odors. In a prior study on WT mice (Lebovich et al., 2021), it was found that during challenging trials with a higher number of known background odors, increased error rates were observed for both go stimuli and no-go stimuli. This suggests that the impact on WT mice differs depending on whether the background odors are novel or known. Using a drift diffusion formalism, Lebovich et al. (2021) demonstrated that the specific increase in error rates for go stimuli in WT mice was most consistent with inhibition of target-associated signals. In contrast, the increase in error rates for both go stimuli and no-go stimuli was most consistent with an elevated level of noise within the decision process. Unlike WT mice, Shank3B+/− mice displayed increased error rates for both go stimuli and no-go stimuli in the presence of novel background odors, which aligns more closely with the hypothesis that novel background odors heighten the noise level within the decision process for Shank3B+/− mice.
Olfactory function in partial isoform Shank3 mouse lines has been previously evaluated using two paradigms without any reported differences compared with WT mice. One paradigm is a simple discrimination between urine and saline (Wang et al., 2011) that uses the inherent tendency of mice to explore urine. The lack of deficit in this simple odor discrimination task is consistent with our results as we found that average glomerular odor representations were similar between WT mice and Shank3B+/− mice. In addition, we found no deficit in discrimination for known mixtures in Shank3B+/− mice. The second previously tested paradigm is the habituation/dishabituation paradigm using social and nonsocial odors (Dhamne et al., 2017; Yang et al., 2012). In this paradigm, mice are presented with odors repeatedly. The first time that an odor is presented, a mouse approaches the odor to explore it. The time exploring an odor is reduced with subsequent presentations as the mouse habituates to the odor. Once a novel odor is introduced, the exploration time increases again. This paradigm not only tests odor discrimination but it also tests odor memory maintenance. We also did not find deficits in odor memory maintenance as evaluated by the reduced sniffing response on the second presentation of a novel background odor for the WT and Shank3B+/− mice. Thus, Shank3B+/− mice remembered their first encounter with a given novel background odor. Mouse lines with complete disruption of all the Shank3 isoforms (Wang et al., 2016; Drapeau et al., 2018) have more broad olfactory deficits, which can be shown in the habituation/dishabituation paradigm. The broader olfactory deficits of mouse lines with complete disruption of all the Shank3 isoforms would preclude these models from our task, not allowing a direct comparison with WT mice performance on odor discrimination in the presence of novel background odors.
Although partial isoform Shank3 mouse lines do not have reported deficits in odor recognition, we found a clear deficit for odor recognition for novel environments in the partial isoform Shank3B mice when tested in the heterozygous genotype. Novel background odors increase the computational difficulty of target recognition, making this task a more sensitive assay to detect odor deficits in mouse models of autism without overt olfactory deficits. Target detection in known background odors can be solved using simple linear classifiers (Mathis et al., 2016), whereas target detection in novel backgrounds could not be solved by linear classifiers (Li et al., 2023), requiring a more complex computation, Lasso deconvolution, which makes this task more likely to be affected by subtle circuit disfunctions produced by Shank3 haploinsufficiency.
Learning deficits have been reported for homozygous Shank3B mice on the Morris water maze (Dhamne et al., 2017), so it was possible that olfactory learning deficits affect the training of the Shank3B+/− mice. Our training paradigm gradually increased the difficulty of the task. The training started with pure targets without backgrounds odors. The target odor concentration was gradually reduced over days followed by introduction of background odors at low concentrations. Background concentrations were increased gradually (Li et al., 2023). Olfactory learning deficits in Shank3B+/− mice were apparent in the initial stage when mice were abruptly exposed to a new pair of target odors without backgrounds. Nevertheless, our gradual training paradigm resulted in both Shank3B+/− and WT mice reaching similar performance for known background odors, despite the initial delay in target odor acquisition of Shank3B+/− mice. Training paradigms with the sudden introduction of new odors could reveal odor learning deficits in mouse models of autism, as shown for the fragile X mouse model (Kuruppath et al., 2023).
Using intrinsic imaging, we found that average glomerular responses of Shank3B+/− mice were correlated with WT mice responses, but they had increased trial-to-trial variability. Intrinsic imaging reflects input from the olfactory receptors into the glomeruli (Vincis et al., 2015; Gurdenet al., 2006). Increased trial-to-trial variability in odor-evoked olfactory bulb neurons postsynaptic to the olfactory receptors (mitral and tufted cells) have been previously described (Geramita et al., 2020) in a different Shank3 homozygous knock-out line (Bozdagi et al., 2010). The increased neural variability in the olfactory bulb caused by deficits in Shank3 might originate in the olfactory receptor axon themselves. Intrinsic glomerular signals produced by odors are modulated by presynaptic GABAB and dopamine (Gurden et al., 2006; Aroniadou-Anderjaska et al., 2000). Differences in the regulation of glomerular layer interneurons could cause the larger variability of intrinsic glomerular signals. The increased variability of the glomerular responses seen in Shank3B+/− mice (
Our simulations suggest that the increased variability in glomerular odor responses in Shank3B+/− mice contributed to the lower performance for odor recognition in novel environments by affecting the performance of odor deconvolution. Go mixtures and no-go mixtures with known background odors contain different target odors, and performance was high and very similar between WT (87.8%) and Shank3B+/− mice (90.4%). It is possible that the higher variability might affect the performance of Shank3B+/− mice for more difficult types of odor classifications where animals need to detect differences in the ratios of odors present and where performance is lower (Uchida and Mainen, 2003).
Shank3 is expressed in many brain areas (Peça et al., 2011) in addition to the glomeruli, and its absence might affect odor recognition via multiple synaptic circuits. Odor recognition in novel environments is a complex computation, and it is possible that other circuit deficits that have been reported in Shank3B mice might also be contributing to their olfactory deficits. Our task cannot be solved using the information from individual glomeruli, and integration over multiple glomeruli is necessary to reach above-chance performance (Li et al., 2023). The integration of information from selected glomeruli requires synaptic plasticity, which is affected in Shank3B mice (Peça et al., 2011; Wang et al., 2011). Inhibitory circuits are also affected in Shank3B mice, resulting in sensory hyper-reactivity (Chen et al., 2020) that might contribute to the effective masking of targets by novel background odors.
In conclusion, we show that Shank3B+/− mice are impaired at target odor detection in the presence of novel background odors. A plausible mechanism for these deficits include increased odor-evoked glomerular variability as simulations with Shank3B+/− levels of variability reduced the performance of the Lasso deconvolution algorithm. Our results have implications for the study of olfaction and mouse models of PMD and ASD. First, our results in Shank3+/− mice are similar to those in Cntnap2−/− mice, suggesting that target odor detection in novel backgrounds is a general deficit in mouse models of autism. An open question relates to whether this phenotype is also present for other sensory modalities. Second, our results indicate that simple discrimination tasks are likely not a sensitive behavioral measure for identifying sensory deficits in mouse models. Finally, our task could serve as an assay for testing interventions to reverse the sensory deficits in mouse models of PMD and ASD, which might then translate into new therapies for individuals with PMD and ASD.
Footnotes
The authors declare no competing financial interests.
- Correspondence should be addressed to Gonzalo H. Otazu at ghotazu{at}gmail.com
