Multiparametric Assays Capture Sex- and Environment-Dependent Modifiers of Behavioral Phenotypes in Autism Mouse Models

Background Current phenotyping approaches for murine autism models often focus on one selected behavioral feature, making the translation onto a spectrum of autistic characteristics in humans challenging. Furthermore, sex and environmental factors are rarely considered. Here, we aimed to capture the full spectrum of behavioral manifestations in 3 autism mouse models to develop a “behavioral fingerprint” that takes environmental and sex influences under consideration. Methods To this end, we employed a wide range of classical standardized behavioral tests and 2 multiparametric behavioral assays—the Live Mouse Tracker and Motion Sequencing—on male and female Shank2, Tsc1, and Purkinje cell–specific Tsc1 mutant mice raised in standard or enriched environments. Our aim was to integrate our high dimensional data into one single platform to classify differences in all experimental groups along dimensions with maximum discriminative power. Results Multiparametric behavioral assays enabled a more accurate classification of experimental groups than classical tests, and dimensionality reduction analysis demonstrated significant additional gains in classification accuracy, highlighting the presence of sex, environmental, and genotype differences in our experimental groups. Conclusions Together, our results provide a complete phenotypic description of all tested groups, suggesting that multiparametric assays can capture the entire spectrum of the heterogeneous phenotype in autism mouse models.


Supplementary figure 3. Behavioral phenotyping of autism mouse models using classical behavioral assays. A) Time spent in the inner zone of an open field test in Shank2 mice (left),
Tsc1 mice (center) and L7-Tsc1 mice (right).Data presented as mean with SEM (three-way ANOVA with Holm-Sidak's multiple comparisons test).B) Ratio of the time spent in the chamber containing a novel mouse divided by the time spent in the chamber with a novel object in the social chamber test, in Shank2 mice (left), Tsc1 mice (center) and L7-Tsc1 mice (right).Data presented as mean with SEM (three-way ANOVA with Holm-Sidak's multiple comparisons test).C) Time spent burying during the marble burying assay in Shank2 mice (left), Tsc1 mice (center) and L7-Tsc1 mice (right).Data presented as mean with SEM (three-way ANOVA with Holm-Sidak's multiple comparisons test).* p≤0.05, ** p≤0.01, *** p≤0.001
At the end of the experiments when animals were euthanized ear tissue was collected and all mice were re-genotyped using the same primers as described above.
Mice were anesthetized with 0.2 ml pentobarbital (60 mg/ml) and perfused with 0.9% NaCl followed by 4% paraformaldehyde (PFA).Brains were dissected from the skull and stored in 4% PFA at room temperature (rT) for 1.5 hours.Finally, sections were rinsed in 0.1M PB, placed on coverslips using a solution of 0.1M PB and Chromium(III) potassium sulfate, and mounted on slide glasses with Mowiol®.
Sections were imaged with a Zeiss AxioImager 2 (Carl Zeiss, Jena, Germany) at 10x.Tile scans were taken from whole brain sagittal slices (Supplementary fig.1A).Additional images were taken with the Apotome at 10x and 20x (Supplementary fig.1B).

Additional experimental information on mice used in the experiments
Different sizes of experimental groups are due to the fact that we experienced a prolonged Covid-19 related breeding stop in 2020, which affected the experiments all the way to 2021.
Unfortunately, we also experienced a mechanical failure of one of our storage drives, before it was backed up.It was not possible to add more animals to individual tests because of the standard housing and environmental enrichment weaning schedule, which resulted in the experimental time of 10 weeks per cohort.The dimensionality reduction analysis was therefore only used on animals that underwent all tests.
The choice of housing 3 mice per cage was motivated by the size of the standard cages; a larger group could lead to overcrowding and possible fighting among males.Therefore a decision was made to house 3 mice per cage for all conditions.All experimental mice were always housed in one of the two configurations -1 mutant mouse with 2 control mice, or 2 mutant mice with one control mouse.
Experimenters were blinded to the genotype of the animals during testing.Genotyping was done prior to the separation of mice into the SH and EE housing, by a lab technician not involved in this study (see acknowledgments).All animals were also re-genotyped after the conclusion of the experiments

Single-trait behavioral assays
All experiments except for the water Y-maze were performed in a behavioral box as described in (1).Single-trait behavioral assays were recorded with a fixed camera (acA 1300-600 gm, Basler AG) positioned above the arena, operated with Bonsai (2) with a frame rate of 25 frames per second (fps).Positional and locomotor analyses were done using the open-source software OptiMouse (3).Animals were habituated to the testing room for 1 hour before experiments.Experimental arenas were cleaned with 70% ethanol between testing of different animals.Data from single-trait behavioral assays was normalized to create the radial plots.

Social chamber test
A three-chamber social preference test was performed in a 63 × 42.5 × 21 cm transparent acrylic arena divided into three separate chambers, with doors between them.First, the test animal was placed in the central chamber.Next, the doors were opened and the animal could explore all chambers.Lastly, the experimental mouse was guided back into the middle chamber and two novel wire cups were placed in opposite chambers, with one containing a novel mouse.The experimental animal was left to explore for 10 minutes in each condition.Data was analyzed using Optimouse (3).Variables used in dimensionality reduction analysis were the fraction of time the animal spent in close vicinity of the novel mouse, the fraction of time the animal spent in the chamber with the social partner, the ratio between the time spent in close vicinity with the novel mouse divided by the time spent in the chamber with the novel cup, and the ratio between the time spent in the chamber with the novel mouse divided by the time spent in the chamber with the novel cup.

Elevated plus maze
The maze consisted of two open and closed arms, each with 29.5 × 8.5 cm white-opaque acrylic base, which were raised off the ground by cylindrical poles 30 cm in height.The closed arms were surrounded by black 20 cm high walls.Mice were placed in the center of a maze and left to explore for 10 minutes.Data was analyzed using Optimouse.Variables used in dimensionality reduction analysis were the time spent in the open arms, the total number of transitions and the distance moved.

Open field test
Animals were placed in a 50 x 50 x 35 cm white-opaque acrylic arena for 30 minutes.Data was analyzed using Optimouse.Variables used in dimensionality reduction analysis were the time spent in the inner zone, the number of transitions into the inner zone and the distance moved.

Marble burying test
The marble burying test was performed as described previously (4).Burying behavior during the test was quantified using JAABA (4).Variables used in dimensionality reduction analysis were the number of burying bouts, the average bout duration and the total burying duration.

Y-maze test
A water Y-maze paradigm was used to examine cognitive flexibility (5,6).A Y-shaped opaque polycarbonate apparatus (symmetrical arms: 32 cm length × 9 cm width × 20 cm height), was filled with white, non-toxic, water-based paint (Basic color, 21 white 30081, Creall).On day 0 animals underwent 3 60 s habituation trials.During the acquisition days (Days 1-2), a platform was placed at the end of one of the arms for 4 sessions of 5 consecutive 60 s trials.This location was kept the same on the test day (Day 3; 1 session of 5 trials).Mice were required to have 80% success rate on the test day to progress to the reversal day, with the location of the platform switched.During the reversal (Days 4-5), mice were exposed to 4 sessions of 5 consecutive 60 s trials followed by a 5th "forced" session where a door was placed in the initially learned arm of the maze.Mice were kept in a clean cage under the heating lamp to dry in between the sessions.
Videos were recorded with a PsEye camera (Sony) at 30 fps.Data was scored by experimenters.
Variables used in dimensionality reduction analysis were the area under the curve (AUC) of the acquisition phase, the AUC of the reversal phase, the number of correct choices, and the number of times the plateau was not reached during a session.To normalize the LMT data for each mutant group, we divided the time spent in each LMT behavior for the mutant mice with the mean time spent in each LMT behavior for WT mice with the same sex and housing conditions.This created a value of the fold change in behavior for the mutant mice relative to the WT mice.

Dimensionality reduction & machine learning classification
The measurements from the single-trait behavioral assays and the multi-parametric assays were gathered in a singular dataset.Dimensionality of the data was reduced using a two-stage PCA-LDA algorithm, as Howland and Park (2004) have shown that using Principal Component Analysis (PCA) as a pre-processing step for Linear Discriminant Analysis (LDA) when performing dimensionality reduction allows for effective discrimination while successfully decreasing the dimensionality (7).Variance captured by each individual single-trait and multi- and to identify possible global deletions, only present after crossing with a Cre-line, we used the third primer 5' AGGAGGCCTCTTCTGCTACC 3'.The PCR products gave band sizes of 190 bp (wildtype), 230 bp (mutant) and 400 bp (deletion).The L7Cre strain was genotyped using the same protocol using the four primers: (oIMR1084) forward primer 5' GCG GTC TGG CAG TAA AAA CTA TC 3', (oIMR1085) reverse primer 5' GTG AAA CAG CAT TGC TGT CAC TT 3', (oIMR7338) internal positive control forward primer 5' CTA GGC CAC AGA ATT GAA AGA TCT 3' and the (oIMR7339) internal positive control reverse primer 5' GTA GGT GGA AAT TCT AGC ATC ATC C 3'.The band sizes were 100bp (Cre transgene) and 324 bp (internal positive control).Behavioral experiments were performed on Cre positive heterozygous for Tsc1 transgenic mice of both sexes.
Data from single-trait behavioral assays was normalized to create radial plots showing an overview of the behavioral phenotype per autism mouse model.Measures used were: sociability, ratio of time spent near novel object divided by the time spent near novel object in the social chamber test; exploratory behavior, ratio of time spent in open arms divided by time spent in closed arms of the elevated plus maze; anxiety, time spent in inner zone of an open field test; hyperactivity, distance moved in an open field test; learning, area under the curve of the percentage of correct trials during the acquisition phase of the Y-maze; reversal learning, area under the curve of the percentage of correct trials during the reversal phase of the Y-maze; repetitive behavior, total time spent burying during the marble burying test.The data was normalized per test, using xnormalized = (x -xminimum) / (xmaximum -xminimum).
They were next changed to a 10% sucrose solution and left overnight at 4°C.Brains were embedded in 12% gelatin and 10% sucrose and left in a solution with 30% sucrose and 4% PFA in PBS at rT for four hours.Next, they were transferred to a 30% sucrose solution in 0.1 PB and kept at 4°C.Whole brains were sliced at 50 μm with a microtome and slices were kept at 0.1M PB.