Functional characterization of the AMPA receptor interacting proteins Shisa6 and
Shisa7
Schmitz, L.J.M.
2018
document version
Publisher's PDF, also known as Version of record
Link to publication in VU Research Portal
citation for published version (APA)
Schmitz, L. J. M. (2018). Functional characterization of the AMPA receptor interacting proteins Shisa6 and
Shisa7.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal ? Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
E-mail address:
Chapter
4
Learning & memory in Shisa6, Shisa7 and
Shisa6/7 KO mice
4
Abstract
Stimulus-induced long-term strengthening and weakening of synapses depends on expression and conductance of AMPARs and these processes lead as the cellular model for memory encoding and storage. Fear conditioning studies revealed that stimulus-dependent synaptic potentiation in the hippocampus and amygdala coincides with memory consolidation, supporting the concept that glutamatergic synaptic plasticity is crucial for learning and memory. Several AMPAR auxiliary subunits, which regulate surface expression, traffi cking and biophysical properties of the receptor, have been characterized. In addition, recent gene deletion studies indicated that both AMPAR-associated proteins Shisa6 and -7 aff ect hippocampal synaptic plasticity. Hence, these Shisa proteins might contribute to learning and memory. Here, we subjected Shisa6,
Shisa7, and Shisa6/7 double KO (dKO) mice to a series of cognitive
hippocampus-dependent tests to study spatial and contextual memory. In addition, anxiety-like and locomotor behavior was monitored to comprehensively characterize their behavioral phenotypes. We concluded that Shisa6 KO mice have impaired long-term spatial memory in the Morris water maze, whereas Shisa7 KO mice show impaired contextual fear memory, working memory, and home-cage based discrimination learning. Shisa6/7 dKO animals recapitulated all phenotypes as observed in the single gene deletions, indicating that the observed memory impairments are likely specifi c to the Shisa gene deleted and not due to compensation of the other Shisa gene. Together, these fi ndings support the hypothesis that Shisa6 and -7 aff ect experience-dependent plasticity and stress the unique contribution of each of the Shisa family members to diff erent aspects of learning and memory.
Introduc� on
Synaptic plasticity is the process of activity-dependent strengthening or weakening of synapses that modifi es effi cacy of signal transduction over time. Many forms of synaptic plasticity have been identifi ed in diff erent cell types and synapses, of which classical long-term potentiation (LTP) in combination with its counterpart long-term depression (LTD) have become the primary cellular model of memory encoding and storage48,67,254.
Recently, the causal link between synaptic plasticity and memory was demonstrated by the use of direct optic stimulation to induce potentiation and depression of synaptic strength106. Manipulating synaptic strength of auditory nuclei input to the lateral
amygdala in a fear conditioning paradigm directly modulates memory expression. More specifi cally, experience-dependent potentiation of synapses in the amygdala100,255 and
hippocampus101,256 by increased AMPA receptor (AMPAR) expression occurs during
4
studies with GluA1 knockout (KO) mice revealed the necessity of the GluA1 subunit for spatial working memory, but not long-term spatial reference memory2457-259.
Recently, a wide variety of transmembrane AMPAR-interacting proteins has been characterized as auxiliary subunits with a distinct profi le of regulation of AMPAR function152,153, including the transmembrane AMPAR regulatory proteins (TARPs)158,172,
the Cornichon homologs (CNIH-2 and CNIH3)163,167, Germ Cell-Specifi c Gene
1-Like (GSG1L)168,176, SynDIG1164, porcupine (PORCN)166 and the Shisa family of
proteins165,170,178,180. Considering the diversity of AMPAR auxiliary proteins involved in
modulation of membrane localization, surface traffi cking, receptor mobility92,158,205 and
synaptic plasticity172,260,261, it is assumed that the composition of AMPAR complexes in
terms of auxiliary protein function will aff ect memory encoding and storage by altering AMPAR-dependent synaptic plasticity.
Despite functional and behavioral characterization of mutants with deletions in AMPAR auxiliary proteins, the role of these proteins in cognition and memory remains largely elusive. A single mutation in the TARP γ-2 gene or combined deletions of other members of the TARP family causes seizures and ataxia, phenotypes that confound cognitive testing (reviewed by Jackson and Nicoll152). For TARP γ-8 KO mice no
gross behavioral changes or cognitive dysfunction have been described so far, despite the reported loss of hippocampal synaptic plasticity in these mutants172. However, a
recent study in rats revealed that blocking the interaction of TARP γ-8 and the AMPAR pharmacologically, aff ected spatial memory and working memory mildly241. In addition,
TARP γ-8 knock-in mice, in which the CaMKII phosphorylation sites are mutated,
showed reduced hippocampal LTP and reduced contextual and auditory fear memory226.
In contrast with TARP γ-8, Th e GSG1L KO rat shows enhanced AMPAR transmission, hippocampal LTP and no spatial memory defi cits in the Morris water maze168. Strikingly,
spatial object recognition memory is disturbed in these KO rats, stressing the importance of balanced AMPAR-mediated signaling for learning and memory.
Shisa6 and -7 have been characterized as AMPAR-interacting proteins with distinct brain region specifi c expression and unique eff ects on AMPAR gating and synaptic plasticity178,252,262. Expression of Shisa6 is primarily restricted to the
hippocampus and cerebellum, whereas Shisa7 expression includes the hippocampus, cortex, the striatum and the amygdala. Ex vivo, Shisa6 traps receptors at the postsynaptic site and modulates short-term plasticity by inhibition of synaptic depression upon high-frequency stimulation. As observed in GSG1L and TARP γ-8 mutant mice, Shisa6 and -7 deletion enhances and reduces hippocampal LTP respectively252,262. Hence,
4
compensatory eff ects that might mask the phenotype of single gene deletion.
When examining learning and memory behavior, several phases of memory formation can be distinguished. Starting with the encoding of information, a specifi c network of neurons that forms a memory engram is activated263. Subsequently, the
molecular stabilization of this engram over time, called consolidation, leads the transition from short-term memory (STM) to long-term memory (LTM)68. To explore during
which process Shisa proteins interact with memory formation, mice were subjected to memory tests with diff erent temporal profi les as well as diff erent training stimuli and brain regions involved. In addition, knowing that the hippocampus is involved in processing of both spatial information264 and context representation265,266 to encode
episodic memories, we aimed to separate these processes using diff erent hippocampus-dependent memory tests, including the Morris water maze and contextual fear conditioning244,267–270. General anxiety and activity tests were included in the behavioral
screening for the interpretation of cognitive phenotypes.
Here, we report that Shisa6 and -7 both infl uence hippocampus-dependent memory in a unique manner, underscoring the functional relevance of their modulatory role in AMPAR-mediated synaptic plasticity.
Methods
MiceShisa6178 and Shisa7252 KO mice and their respective WT littermate controls, bred from
heterozygous parents for each genotype, were used for behavioral testing between 10-14 weeks of age. Shisa6/7 double KO (dKO) mice were generated by breeding of Shisa6 and -7 single KO mice, and they were tested together with age-matched WT mice from the Shisa6 and -7 lines. Shisa6 and -7 KO mice were tested when 9–12 weeks old, and dKO mice when 10-14 weeks of age. All mice were housed individually for at least 1 week before the start of behavioral testing with free access to food and water in a 12-h light/dark cycle (7:00 AM lights on), and experiments were performed during the light phase (8:30 AM– 4:00 PM). Mice were tested using a test battery (Supplementary fi g. 1), starting with monitoring spontaneous activity and discrimination learning behavior in an automated home-cage design. Subsequently, mice were introduced to a test battery including the following tests in fi xed order: open fi eld, dark-light box, rotarod and grip strength measurement, T-maze and Morris water maze (MWM). Only for Shisa7 KO mice and littermates a separate batch was used for the MWM. All fear conditioning experiments were performed in distinct batches of naïve mice.
Morris water maze
4
which was colored opaque with white nontoxic dye and kept at a temperature of 24– 26°C. An escape platform (ø: 9 cm) was placed at 30 cm from the edge of the pool submerged 0.2 cm below the water surface. Visual cues were located around the pool at a distance of 1 m in a dimly lit room (20 lx). Mice were trained for 5 consecutive days, 4 trials per day, and the latency to fi nd the platform was examined using video tracking (VIEWER 2, BIOBSERVE, Bonn, Germany).
During each trial, mice were fi rst placed on the platform for 30 s, and then placed in the water at 1 of 4 start positions, randomly selected over trials, and allowed to swim for max 60 s to fi nd the platform271. Mice that could not fi nd the platform were guided
to the platform by the experimenter. All mice remained at the platform for 15 s, and this trial was immediately followed by another swimming trial. Mice were placed in their home-cage for 2 min between trials 2 and 3. On day 6, the platform was removed and a probe trial was initiated by placing the mice directly into the water opposite to the platform location. For analysis, the pool was divided in 4 quadrants and time spent in the quadrant areas was recorded as a read-out of long-term spatial memory. Only
Shisa6/7 dKO mice were subjected to an additional visual platform training session that
included 4 trials in which a black visual object (ø: 1 cm, height: 10 cm) marked the location of the platform.
Fear conditi oning
All experiments were carried out in a fear conditioning system (TSE Systems). Mice were trained and tested in a Plexiglas chamber (36x21x20 cm) with a stainless steel grid fl oor with constant illumination (100–500 lx) and background sound (white noise, 68 dB sound pressure level), which was placed in gray box to shield it from the outside. Before each training or testing session, the chamber was thoroughly cleaned with 70% ethanol. During retrieval tests freezing was defi ned as the lack of any movement besides respiration and heart beat during a 2-s period, calculated based on automatically detected laser beam breaks.
Contextual fear conditioning – Training consisted of an exploration period of 180 s, after
which a mild 2-s foot shock (0.7 mA) was delivered through the grid fl oor. Mice were returned to their home cage 30 s after the shock ended. Memory tests consisted of re-exposure (180 s) to the training context (conditioned stimulus), at 2 h after conditioning to assess short-term memory (STM) and at 24 h to assess long-term memory (LTM).
Auditory fear conditioning – In the conditioning box, a high-frequency loudspeaker
4
test was performed after 24 h by returning the mice to the conditioning box, followed by an auditory memory test in a novel context (Context B) after a 2-h time interval. Mice were placed in the novel context and after 180 s (pre-tone) the tone was played for another 180 s. Context B consisted of a similar Plexiglas cage that was covered with additional visual cues and cleaned with 1% acetic acid before testing. Th e novel context consisted of a smooth fl oor (lacking a grid) in a bright environment (380–480 lx). No background noise was presented during testing.
T-maze
Mice were introduced into the start arm (base) of a T-maze (white PVC, arms 30x10 cm, walls 35 cm high) to explore the maze. During the sample phase, mice were forced to choose the left or right goal arm by a removable wall (17 cm long) protruding into the start arm. After entering the goal arm a guillotine door at the end of the start arm was lowered. At the test phase, mice were kept at the start arm using another guillotine door. Th e test trial was initiated by removing the guillotine doors, allowing the mice to explore freely. Alternation was considered successful if the mice entered the goal arm that was not visited during the sample phase. Th e sampling and test phase were repeated 6 times in total (3 trials per day, 1-h inter-trial interval).
Open fi eld
Mice were placed in a corner of a white square open fi eld box (50x50 cm, walls 35 cm high, 200 lx). Exploration behavior was recorded for 10 min (Viewer 2, Biobserve GmbH, Bonn, Germany). Th e surface area was divided into nine equally sized squares, and the center square was used as center area. Time spent in the center area and total distance moved were measured using video tracking (Viewer 2, Biobserve, Bonn, Germany).
Dark-light box
Mice were placed into the dark compartment of a dark-light box (25x25 cm, walls 30 cm high, <10 lx) and kept for 60 s until a sliding door provided access to the light compartment (25x25 cm, 30 cm high, 625 lx). Th e mice could explore 60 s freely in any of the compartments. Time spent in the light compartment and latency to enter the light compartment were measured using video tracking (Viewer 2, Biobserve, Bonn, Germany).
Automated home-cage general acti vity & Cogniti onWall learning
4
interference for 7 days. Water was provided ad libitum, and food was available for three days during habituation and recording of spontaneous behavior. General activity of the fi rst three 12 h dark phases was compared for all genotypes as a measure of general activity.
In the subsequent four days, mice earned their daily food from the pellet dispenser by performing a discrimination and reversal learning test, described in detail in Remmelink et al. (2016)272. In brief, on day 4 in the cage a CognitionWall with three
holes (opaque Perspex, H=25 cm, W=17 cm, ø holes=3.3 cm) was placed in front of the pellet dispenser, and ad libitum food was removed. For two consecutive days, mice underwent a discrimination learning protocol, during which they learned to earn food pellets (Dustless Precision Pellets, 14 mg, Bio-Serve, Frenchtown, NJ, USA) by entering through the left hole of the CognitionWall. Entering through the middle or right hole was not rewarded. After two days, a reversal learning phase was initiated by switching the rewarded hole to the right hole, and this phase lasted for two more days. In both phases of the task, mice were given a reward on every fi fthcorrect entry (FR5).
Th e learning criterion was reached when 80% of the entries were correct, computed using a sliding window over the last 30 entries. Genotype diff erences were tested for signifi cance using the Gρ weighted log-rank test for diff erences between two or more Kaplan-Meier survival curves. During reversal learning, an entry through the left hole was quantifi ed as a perseverative error, and through the middle hole as a neutral error. Th ese data were tested using Student’s t-test, or Mann-Whitney U test, when criteria for normal distribution were not met.
Grip strength
To evaluate neuromuscular function, grip strength was measured by assessing peak force (N) mice applied when grasping a pull bar connected to a force meter (1027DSM Grip Strength Meter, Columbus instruments, Columbus, OH, USA). Grip strength was determined by taking the median of 5 repetitions of grasping the bar with only front paws or front and hind paws.
Accelerati ng rotarod
4
Results
Both Shisa6 and -7 are expressed in the hippocampus where they aff ect AMPAR-mediated plasticity. Hence, we fi rst determined whether hippocampus-dependent cognitive function was altered in Shisa KO mice. Th erefore, mice were subjected to a series of cognitive learning and memory tests that focus on spatial and contextual memories (Supplementary fi g. 1). Spatial memory was studied in a Morris water maze, a classical test for memory acquisition and long term memory based on distal cues to locate a hidden platform in a swimming pool. Contextual short-term and long-term memory was assessed in a contextual fear conditioning paradigm. Both forms of memory depend on the hippocampus and are aff ected by hippocampal lesions244,268,273,
yet, distinct neural correlates and molecular mechanisms have been identifi ed274,275.
In addition, cognitive tests with a distinct temporal profi le were performed, to further explore whether Shisa proteins contribute to initial memory encoding or the transition from short- to long-term memory. To this end, besides short- and long-term fear memory, a T-maze spontaneous alternation paradigm was performed, to include a spatial working memory test276,277. Finally, the time required for acquisition of memory
was monitored in a home-cage paradigm, the CognitionWall, to provide a continuous readout of learning behavior under semi-natural conditions.
Th e test battery memory tests (Supplementary fi g. 1) depend on locomotion and the initiative of mice to explore. Nonetheless the main readout of memory performance for these tests is not a direct measure of locomotion, but a percentage of all activity; respectively the fraction of correct over incorrect activities, percentage of alternation and fraction of time spent in each quadrant. In the fear conditioning paradigm however, general activity does directly infl uence freezing behavior that is taken as measure for memory expression. Th erefore, we assessed general activity in a home-cage design, as this allowed to track locomotion throughout all phases of the circadian rhythm. Secondly, exploratory behavior was monitored in a novel context in the open fi eld setup. A dark-light box was used to characterize both activity and anxiety-like behavior.
In contrast with Shisa7, Shisa6 is abundantly expressed in the cerebellum where it might contribute to locomotor behavior. Th erefore, the accelerating rotarod was included in the test battery, as a motor learning test to detect gross locomotor phenotypes under challenging conditions. Th is collection of behavioral tests can aid to improve insight in the general activity pattern of the Shisa KO mice, their exploratory behavior in a novel context and motor function that might infl uence memory formation and expression.
Shisa6, but not Shisa7 KO mice show defi cits in spati al memory
long-4
0 10 20 30 40 50 60 70Training 1 Training 2 Training 3 Training 4 Training 5
Es ca pe la te nc y (s ) 6 platform left right opposite Training 1 2 3 444 5 Probe day 0 10 20 30 40 50 60
Training 1 Training 2 Training 3 Training 4 Training 5
Es ca pe la te nc y (s ) 0 10 20 30 40 50 60
Training 1 Training 2 Training 3 Training 4 Training 5
Escape latency (s ) 0 10 20 30 40 50 60
Trial1 Trial2 Trial3 Trial4
Es ca pe la te nc y to v is ua l pl at fo rm (s ) a b c d e f g h 0 10 20 30 40 50
platform right left opposite
Ti m e sp en t i n qu ad ra nt (% ) WT Sh6/7 KO 0 10 20 30 40 50
platform left right opposite
Ti m e sp en t i n qu ad ra nt (% ) 0 10 20 30 40 50
platform left right opposite
Ti m e sp en t i n qu ad ra nt (%) P=0.006 P=0.023 Genotype: P<0.001 Shisa6 KO WT Shisa7 KO WT Shisa6/7 KO WT
4
term memory was assessed after a 24 h retention interval in a probe session, in which the platform was removed. During task acquisition, the escape latencies of Shisa6 KO mice were not diff erent from WT mice (P=0.420; Fig. 1a) and neither was their swimming speed (Supplementary fi g. 2a). However, the probe session revealed that Shisa6 KO mice had an impaired long-term spatial memory compared with WT controls (P=0.006; Fig. 1 b). Shisa7 KO mice, in contrast, had normal spatial memory acquisition (P=0.228; Fig. 1 c, Supplementary fi g. 2b) and long term memory (P=0.831; Fig. 1d). Finally,
Shisa6/7 dKO mice displayed more profound memory impairments compared with Shisa6 KO mice as the latencies to reach the platform increased during training
(P<0.001, Fig. 1e) and time spent in the target quadrant during the probe session was lower than WT controls (P=0.023, Fig. 1f). Th ese results suggest that in contrast with single gene deletion, the double deletion of Shisa6 and -7 induces impaired spatial memory acquisition. To elucidate whether the observed diff erences in escape latencies during training were dependent on general activity in the swimming pool, swimming velocity was measured (Supplementary fi g. 2). Th e Shisa6/7 dKO mice showed changes in velocity, with a lower velocity during training (P<0.001; Supplementary fi g. 2c). Th is suggests that changes in memory acquisition might result from changes swimming behavior. To distinguish between disturbed spatial processing and motor problems possibly contributing to the observed changes in swimming behavior, a visual platform test was performed, using a cue to signal the platform. Swimming velocity was the same for Shisa6/7 dKO and WT mice during the visual platform trials (P=0.353, Fig. 1g). Th is indicates that during training Shisa6/7 dKO mice did not reach the WT performance level due to spatial learning defi cits and not as a result of disturbed anxiety-like or swimming behavior.
Shisa7, but not Shisa6 KO mice have impaired contextual fear memory
In addition to spatial Morris water maze learning, we subjected mice to fear conditioning tests to assess contextual and cued memory (Fig. 2). In order to discriminate between short-term memory (STM) and long-term memory (LTM) separate batches of mice were tested either 2 hours (Fig. 2a-c) or 24 hours after training (Fig. 2d-f). Shisa6 KO mice did not have any fear memory defi cits (Fig. 2b,e). In contrast, Shisa7 KO mice (Fig. 2c,f)252 showed a strong reduction in freezing levels both 2 hours (P<0.001) and
24 hours (P<0.001) after training. After cued-conditioning with an auditory stimulus, we could confi rm a defi cit in contextual fear memory as Shisa7 KO showed a reduction
4
in freezing upon context re-exposure (WT: 22.8 %, KO: 10.0 %, U=39.500, P=0.035), but neither Shisa6 KO, nor Shisa7 KO237 mice showed a defi cit in auditory fear memory
(Fig. 2g–i).
Working memory defi cits for Shisa7 KO and Shisa6/7 dKO mice
Besides testing short- and long-term memory, we assessed spatial working memory in all three mutants. To this end, mice were subjected to a T-maze test, which is based on
0 10 20 30 40 50 Fr ee zi ng % 0 10 20 30 40 50 Pre-tone Tone Fr ee zi ng % WT Shisa7 KO KO Pre-tone Tone WT Sh6 WT KO Sh6KO 0 0 10 20 30 40 50 Pre-tone Tone Fr ee zi ng % WT Shisa6 KO p=0.004 0 10 20 30 40 50 Fr ee zi ng % WT CS CS 2 h US 0 10 20 30 40 Fr ee zi ng % A 24 h US CS B CS pre-tone tone B 0 10 20 30 40 50 Fr ee zi ng % P<0.001 a b c d e f CS CS 24 h US g h i WT Sh7 KO WT Sh7KO P<0.001
4
spontaneous alternation behavior (Fig. 3). In this type of maze (Fig. 3a), mice naturally tend to move towards the arm opposite to the one they have just recently explored, which allows monitoring of their working memory. Th e percentage of alternation measured during 6 consecutive trials was not diff erent between Shisa6 KO mice and WT controls (Fig. 3a). Yet, alternation was signifi cantly lower in Shisa7 KO (P=0.004, Fig. 3b) and Shisa6/7 dKO mice (P=0.035, Fig. 3c) compared with the respective WT mice. Th ese results suggest that besides short- and long-term contextual memory, Shisa7 also contributes to spatial working memory.
Shisa7 KO mice and Shisa6/7 dKO mice display impaired discriminati on learning We next subjected mice to a reward-based learning paradigm in an automated home-cage, the CognitionWall DL/RL task, which diff ers from classical memory tests in terms of duration, motivation and experimenter-induced interference. We separated discrimination learning (DL) (day 4–5) from reversal learning (RL) (day 6–7). Whereas the learning curve for Shisa6 KO mice was not diff erent from WT mice (P=0.691; Fig. 4a), Shisa7 and -6/7 (d)KO mice required more entries to reach the learning criterion (Fig. 4b,c; P=0.007, P=0.006, respectively) compared with WT controls, indicating impaired discrimination learning. Subsequently, reversal learning was tested in the following two days. Surprisingly, a trend for faster reversal learning was observed for
Shisa6 KO mice (P=0.090; Fig. 4d), suggesting that cognitive fl exibility was enhanced
in these mice. Th is was further supported by a decrease in number of neutral errors (P=0.028; Table 1) and a reduction in perseverative errors (P=0.026; Table 1) during reversal learning. Impaired initial discrimination learning in Shisa7 KO and Shisa6/7 dKO mice coincided with an increase in number of errors (Table 1). Despite this,
Figure 3. Shisa7 KO and Shisa6/7 dKO mice have working memory impairments
a) Schema� c representa� on of the T-maze setup, displaying the start of the sample and test phase of all trials. b-d) Spontaneous alterna� on in the T-maze was averaged over 6 trials, presented is the percentage correct alterna� ons for Shisa6 KO (b), Shisa7 KO (c) and Shisa6/7 dKO (d) with their respec� ve WT controls. Spontaneous alterna� on was not aff ected in Shisa6 KO mice, which implies that working memory is intact (nWT=12, nKO=12; U=50.000, P=0.219). However, Shisa7 KO mice do show a signifi cant reduc� on in correct alterna� ons (nWT=12, nKO=12; U=23.500, P=0.004), similar to Shisa6/7 dKO mice (nWT=10, nKO=10; U=22.000, P=0.035).
4
d a b e f c 0 200 400 600 800 1200 0.0 0.2 0.4 0.6 0.8 1.0Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa6 KO WT P= 0.691 Discrimination learning 0.0 0.2 0.4 0.6 0.8 1.0
Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa7 KO WT 0 400 800 1200 1600 P= 0.770 Reversal learning 0.0 0.2 0.4 0.6 0.8 1.0
Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa6 KO WT 0 400 800 1200 1600 P= 0.090 Reversal learning 0 200 400 600 800 1200 0.0 0.2 0.4 0.6 0.8 1.0
Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa7 KO WT P= 0.007 Discrimination learning 0 200 400 600 800 1200 0.0 0.2 0.4 0.6 0.8 1.0
Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa 6/7 KO WT P= 0.006 Discrimination learning 0.0 0.2 0.4 0.6 0.8 1.0
Number of entries to 80% criterion
Fraction of mice that reached criterion
Shisa6/7 KO
WT
0 400 800 1200 1600
pP= 0.428
Reversal learning
4
their discrimination performance at the end of the acquisition phase was not diff erent from WT level (Shisa7 KO, P=0.951; Shisa6/7 dKO mice, P=0.143; Table 1). Th erefore, reversal learning was examined as well, which turned out to be unaff ected by genotype for Shisa7 (P=0.770; Fig. 4e). Interestingly, also the dKO (P=0.428; Fig. 4f) required an equal number of entries to reach the reversal learning criterion (80% correct). Th us, the absence of the Shisa6 gene in Shisa6/7 dKO mice did not introduce a reversal learning phenotype. Together, the reward-based discrimination learning results were comparable to the phenotypes obtained in fear conditioning and the T-maze.
Shisa7 KO and Shisa6/7 dKO, but not Shisa6 KO mice show increased general acti vity
in a home-cage
To identify changes in general activity that may infl uence the results of cognitive tests, the activity pattern of mice was tracked in the PhenoTyper home-cage for 60 hours (Fig. 5a-c). Mice were introduced to the new home-cage 2 hours before the start of activity tracking, and the fi rst 3 hours were analyzed as measure of novelty-induced activity. No diff erences in general activity were observed for Shisa6 and -7 KO mice (Fig. 5a,b) during these fi rst
Table 1. Number of errors and performance in the Cogni� onWall DL/RL task. Presented are the number of errors mice made un� l they reached criterion (80% correct) during discrimina� on learning, and the percentage of correct entries in the last 30 trials of the discrimina� on learning phase per genotype. Both Shisa7 and -6/7 (d)KO mice made more errors before they reached the learning criterion. (P=0.032, P=0.003 respec� vely). However, there was no diff erence in performance in the last 30 trials before the start of the reversal learning phase. Subsequently, the number of errors made un� l mice reached reversal learning criterion (80% correct) were categorized as persevera� ve and neutral. A persevera� ve error is the entry into the hole that was previously rewarded with a food pellet, and a neutral error is the entry into the hole that was not rewarded before. Shisa6 KO mice displayed a signifi cant reduc� on in both persevera� ve (P=0.026) and neutral errors (P=0.028), confi rming that they have be� er reversal learning performance.
Discrimination learning Reversal learning errors performance correct before
reversal
perseverative
4
3 hours, whereas Shisa6/7 dKO mice (P=0.003; Fig. 5c) displayed increased activity. Th e total activity measured over the dark phases during the total period of 60 hours was not diff erent for Shisa6 KO mice (Fig. 5a,b). In contrast, Shisa7 KO mice showed a signifi cant increase of general activity in the fi rst and third dark phase and a trend for enhanced activity during the second dark phase (resp. P=0.021, P=0.068, P=0.032; Fig. 5b). Likewise, in addition to increased novelty-induced activity, the general activity of Shisa6/7 dKO mice was enhanced throughout all three 12-h dark phases (P=0.003, P=0.010, P=0.027; Fig. 5b). Th ese data indicate that deletion of Shisa7 leads to increased general activity that is enhanced by further deletion of Shisa6, suggesting that interpretation of activity-dependent cognitive tests such as the fear conditioning test requires caution.
Introducti on in a novel context reveals no gross changes in exploratory behavior
As locomotor behavior in cognitive tests is not merely dependent on general activity, but is also dependent on the strategy of animals to cope with handling and novelty-induced stress, we examined exploratory locomotor and anxiety-like behavior in a more classical manner, namely upon introduction in an open fi eld (Fig. 6a-f). During exploration in the open fi eld, Shisa6 KO mice spent signifi cantly less time in the center area of the open fi eld (P=0.001, Fig. 6a), indicative of increased anxiety-like behavior in these animals. In contrast, this anxiety-like phenotype was not observed for Shisa7 -and -6/7 (d)KO mice (Fig. 6b,c). No genotype diff erence was observed in the total distance moved for any of the genotypes tested (Fig. 6d-f). We further dissected the putative anxiety-like behavior of Shisa6 KO mice in a dark-light box (Fig. 6g-l). Neither Shisa6 KO, nor Shisa7 KO showed a diff erence in the latency to cross from the dark to the light compartment (Fig. 6g-i), or the total time the animals spent in the light compartment (Fig. 6j-l). In this test, only the Shisa6/7 dKO animals spent signifi cantly less time in the light compartment (P=0.002; Fig. 6i), but the latency to transfer from the dark to the light compartment was not aff ected. Taken together, these results indicate a mild increase in diff erent types of novelty-induced anxiety-like behavior in mice with a deletion of the Shisa6 gene in comparison to WT mice.
Shisa6 KO mice display defi cits in motor functi on
4
0 1000 2000 3000 4000 5000 6000 7000 0 12 24 36 48 60 D is ta nc e m ov ed ( cm ) WT Shisa6 KO 0 1000 2000 3000 4000 5000 6000 7000 0 12 24 36 48 60 D is ta nc e m ov ed ( cm ) WT Shisa7 KO a b 0 2000 4000 6000 8000 10000 12000 14000 0 12 24 36 48 60 D is ta nc e m ov ed ( cm ) WT Shisa6/7 KO c Time (h) Time (h) Time (h) P=0.021 P=0.068 P=0.032 P=0.007 P=0.010 P=0.027 P=0.0034
Figure 6. Anxiety-like behavior in the open fi eld and dark-light box. a-f) Novelty-induced anxiety-like behavior and locomotor ac� vity was measured in the open fi eld test. a-c) Shisa6 KO animals spent signifi cantly less � me in the open fi eld center area (a; nWT=22, nKO=22; F(1,42)=13.588, P=0.001) than controls, whereas Shisa7 KO mice (b; nWT=25, nKO=26; F(1,49)= 0.241, P=0.625) and Shisa6/7 dKO mice (c; nWT=10, nKO=9; F(1,17)= 2.864, P=0.109) showed no diff erence. d-f) The total distance moved was not changed for any of the genotypes (d; Shisa6 KO: nWT=22, nKO=22; t(42)=1.248, P=0.219) (e; Shisa7 KO: nWT=26, nKO=25; U=406.000, P=0.127) (f; Shisa6/7 dKO: nWT=10, nKO=9; U=53.000, P=0.459). h-l) Time spent in the light compartment and latency measured in the light compartment in a dark-light box. The � me spent in the light compartment was not diff erent for Shisa6 KO (g; nWT=23, nKO=22; t(43)=-1.079, P=0.268) or Shisa7 KO (h; nWT=17, nKO=16; t(31)=0.374, P=0.711), however it was lower in Shisa6/7 dKO mice compared with WT (i; nWT=9, nKO=8; t(15)=3.875, P <0.001). The latency to visit the light compartment of the dark-light box was not signifi cantly diff erent from respec� ve WT mice for Shisa6 (j; nWT=23, nKO=22; U=183.000, P=0.112), for Shisa7 (k; nWT=17 nKO=16; U=138.000, P=0.958) and Shisa6/7 dKO mice (l; nWT=9, nKO=8; U=47.000, P=0.321). increased ac� vity during the second dark phase (F(1,53)=3.466, P=0.068). c) Likewise, Shisa6/7 dKO mice showed an increase in spontaneous ac� vity both during the ini� al 3-h of the dark phase (nWT=8, nKO=7; F(1,13)=12.664, P=0.003), and the three 12-h dark phase cycles (F(1,13)=10.069, P=0.007; F(1,13)=9.169, P=0.010; F(1,13)=6.202, P=0.027). Note that the scale of the total distance moved for Shisa6/7 dKO (c) is diff erent from Shisa6 or -7 KO (a,b).
4
0 5 10 15 20 25 R PM re ac he d WT Sh6/7 KO genotype: P=0.003 0 0.5 1 1.5 2 G rip s tre ng th (N ) WT Sh6 KO d 0 5 10 15 20 25 R PM re ac he d WT Sh7 KO 0 0.5 1 1.5 2 G rip s tre ng th (N ) WT Sh7 KO a b e f 0 5 10 15 20 25 R PM re ac he d genotype: P<0.001 Sh6 KO WT WT Sh6/7 KO 0 0.5 1 1.5 2 2.5 G rip s tre nt gh ( N ) c4
Discussion
Here we aimed to explore learning and memory function of Shisa6, -7 and -6/7 double KO (dKO) mice. In literature both positive and negative correlations between LTP and hippocampus-dependent memory performance have been reported278. As LTP in the
Schaff er collaterals of Shisa7 KO mice is strongly reduced252, but increased in Shisa6 KO
animals249 we expected hippocampus-dependent learning processes to be disturbed in both
genotypes. Indeed, Shisa6 KO mice had impaired long-term spatial memory in the Morris water maze, although contextual fear memory, working memory and home cage-based discrimination learning were not changed. Surprisingly, in these mice a trend for increased cognitive fl exibility was observed upon reversal of the discrimination task. In contrast,
Shisa7 KO mice showed poor contextual fear memory expression, but normal
long-term auditory and spatial memory. Shisa7 deletion seemed to disturb the early process of memory acquisition, as working memory and short term memory were diminished, and discrimination learning was slowed. Th e Shisa6/7 dKO mice had memory defi cits in all paradigms tested, showing an accumulated phenotypic eff ect of the deletion of both
Shisa6 and -7. Th is suggests that the behavioral phenotypes of the single KO are specifi c to
the Shisa gene deleted and not due to compensation of the other Shisa gene.
Locomotor behavior and anxiety-like behavior interfere with the analysis of memory performance
Behavioral screening has been used extensively to assess the impact of genetic manipulations279. Here the objective was to distinguish the eff ect of Shisa deletion on
learning and memory from exploratory and emotional behavior, the latter of which may aff ect the analysis of cognition and memory. Th e combination of long-term video-tracking of spontaneous behavior with the open fi eld as classical tests of locomotor behavior revealed that although mice with a Shisa7 deletion show enhanced spontaneous activity, introduction into a novel context did not lead to changes in locomotor behavior. Hence, the increase in activity measured as reduced freezing upon fear memory recall in the Shisa7 KO mice is likely a consequence of reduced memory formation and expression, and not increased general activity during memory recall. Th is is supported by the fi nding that during retrieval of the auditory fear memory freezing behavior was normal.
As Shisa6 shows strong cerebellar expression, motor function and general activity was investigated. Swimming behavior in the Morris water maze as well as spontaneous and novelty-induced activity revealed no changes in locomotion and swimming behavior in the Shisa6 KO. Only in the accelerating rotarod, when motor learning is required, Shisa6 deletion induces disturbed motor function.
Spatial memory and contextual fear memory are infl uenced by stress and emotional arousal280–282. Stress and corticosteroids can promote switching from a spatial learning
4
in memory tests discussed in this study. Th erefore, we monitored whether novelty-induced anxiety-like behavior was changed by deletion of the Shisa genes in an open fi eld and dark-light box paradigm. Shisa6 KO mice spent less time in the center region of the open fi eld area, and the Shisa dKO mice spent less time in the light compartment of the dark-light box, together indicative of a Shisa6 dependent change in anxiety. Th e ventral hippocampus is a major player in anxiety-like behavior273,284. As we observed
that Shisa6 expression is restricted to the hippocampus and cerebellum, it is likely that ventral hippocampal dysfunction could account for the increased anxiety-like phenotype. Increased anxiety and emotional arousal during training in the Morris water maze may suppress spatial learning286, which might account for the Morris water maze
defi cits we observed for Shisa6 KO mice. In line with this, we did not observe learning defi cits for Shisa6 KO mice in the discrimination learning phase of the CognitionWall, when experimenter and novelty-induced stress is minimized. Strikingly, under these low arousal conditions, Shisa7 KO and -6/7 dKO mice still have a reduced learning rate, implying a cognitive defi cit.
Spati al and contextual memory processing in the hippocampus and corti cal regions
Th e distinct profi le of learning and memory behavior in Shisa6 vs. Shisa7 KO mice might be explained by the brain region-specifi c expression pattern of the proteins. Whereas expression of Shisa6 is primarily restricted to the hippocampus and cerebellum, expression of Shisa7 is more widespread and includes cortical regions, the striatum and the amygdala.
Th e hippocampus displays spatially localized fi ring patterns, thereby generating a cognitive map of the spatial representation of the environment288. It has become evident
that hippocampal neurons adapt their spatial fi ring behavior upon environmental changes in context, like odor and texture or tasks demands265,287. In turn, each context is
represented by a diff erent subset of neurons, a neuronal ensemble288. Classical contextual
and spatial memory tests separate these functions of the hippocampus, and previous research has revealed that at the systems level the neural substrate of these paradigms is distinct275. Burwell and colleagues have demonstrated by lesioning of peri-, ento-
and postrhinal cortical regions that spatial navigation does not require full neocortical input, and that contextual memory is more aff ected by these lesions. Although the hippocampus is the key region to integrate contextual information and is required for contextual fear conditioning270, other areas, such as the striatum289,290, prefrontal
cortex291,292 and the entorhinal cortex293, also contribute to the formation of contextual
fear memory. Lesion studies provided evidence for anterograde amnesia if hippocampal damage precedes contextual fear conditioning243,294. However, hippocampal lesions do
not always prevent learning of conditioned fear192,244, suggesting that alternative systems
can acquire contextual fear memories295. Th erefore, we suggest that although in Shisa6
4
over contextual information processing, leaving contextual memory intact. Because the contextual fear memory system is likely to be more severely impaired due to the widespread expression of Shisa7, hippocampal dysfunction might not be compensated for by cortical regions.
Lesion studies have demonstrated that spontaneous alternation as measured in the T-maze requires integrity of pathways including hippocampus, prefrontal cortex and dorsal striatum, which all express Shisa7 (for review277). Likewise, the learning process in
the CognitionWall DL/RL task probably involves both procedural and spatial learning, and presumably requires integrity of the hippocampal and striatal system. Th is might explain poor performance of Shisa7 KO mice, and the absence of a memory phenotype in these tests for Shisa6 KO mice.
Not only the diff erence in neural correlate, but also the also single trial versus multi trial experimental design could explain the contrast between fear conditioning and spatial memory tests. A lesion study in rats revealed that although hippocampal lesions reduced fear memory after a single conditioning trial, learning was normal after multiple conditioning trials295. In line with these results, we postulate that the Shisa7
KO animals, which have severe memory defi cits in a single trial fear conditioning experiment, might function normal in a multi trial learning. Hence, the multitude of training sessions in the Morris water maze might mask long-term memory defi cits in these animals. Discrimination learning in the CognitionWall DL/RL task demonstrate that Shisa7 KO mice have a slower learning curve, but eventually reach WT performance levels.
In contrast with the single KO mice, the Shisa6/7 dKO mice displayed disturbed memory acquisition in the Morris water maze. As each trial during training starts by placing mice on the platform, mice are not fully dependent on long-term spatial memory to fi nd the platform. Hence, increased escape latencies in Shisa6/7 dKO mice could be the result of impaired working memory or short-term memory, possibly in combination with long-term memory defi cits.
Behavioral phenotypes associated with deleti ons of AMPAR (auxiliary) subunits
AMPAR auxiliary subunits have been extensively studied in light of AMPAR gating and plasticity, yet little is known about their role in experience dependent plasticity in vivo. Recently, TARP γ-8 KO mice have been subjected to several behavioral tests, reporting an hyperactivity, increased antidepressant behavior in a forced swim test, but no changes in anxiety-like behavior240. Whereas until now there are no reports on memory
performance of TARP γ-8 KO mice, the eff ect of TARP γ-8 on spatial memory has been investigated pharmacologically by administration of drugs disrupting TARP γ-8 to AMPAR binding241. Th is induces in a very mild negative eff ect on Morris water maze
4
the absence of phosphorylation at these sites could account for most of the phenotypic eff ect of TARP γ-8 deletion on LTP. Moreover, it was shown that this mutant displayed a severe impairment in acquisition and expression of both contextual and auditory fear memory.
Another auxiliary subunit that has been characterized at the behavioral level is GSG1L168, an AMPAR-associated protein that enhances basal AMPAR transmission
and LTP. Th e GSG1L KO rat has normal spatial memory in the Morris water maze, although spatial object recognition is impaired168. Th is suggests that both enhanced
plasticity might induce an imbalance in network activity that results in impaired cognitive function. Such imbalance might account for long-term spatial memory defi cit observed in Shisa6 KO mice.
In contrast with the lack of data on behavioral implications of AMPAR auxiliary subunits, phenotyping of GluA1-, GluA2- and GluA3-defi cient mice has stressed the role of AMPARs in behavior and cognition. Furthermore, these studies indicated that deletion of GluA1 and GluA2 uniquely acts on memory processes with a distinct temporal profi le. Mice lacking the GluA1 subunit have severe problems with spatial working memory in a Y-maze257,258, but normal spatial long-term memory296, despite
the abolishment of classical hippocampal LTP induction in synapses of the Schaff er collaterals297. GluA3-defi cient mice display learning and memory rates similar to WT
animals in Morris water maze and Y-maze tests and normal anxiety-like behavior. Interestingly, both GluA1 and GluA3 seem to modulate social behavior like aggression and sociability298,299. On the other hand, mice lacking the GluA2 subunit display reduced
open fi eld novelty-induced exploration, disrupted motor coordination in the rotarod300
and enhanced hippocampal LTP, which are similar to our observations in Shisa6 KO mice. A more recent study using selective deletion of GluA2 in the CA1 pyramidal cells confi rmed that hippocampal LTP is increased these mice, but rotarod and open fi eld behavior is normal301, indicating that these behaviors are not controlled by hippocampal
4
Shisa6 (Chapter 2) and Shisa7 KO mice (Chapter 3).With use of the CognitionWall DL/RL task we could distinguish initial discrimination learning from reversal learning. Shisa6 KO mice showed enhanced cognitive fl exibility as displayed by a trend for signifi cance in the number of entries required to achieve the reversal learning criterion and a reduction in neutral and perseverative errors. Cognitive fl exibility in the CognitionWall DL/RL task is dependent on the orbitofrontal cortex, as confi rmed by a recent lesion study272. Shisa6 expression however,
is low in cortical regions, suggesting that brain regions other than the orbitofrontal cortex might contribute to the enhanced rate of reversal learning in Shisa6 KO mice178. Th is is
supported by the fi nding that OFC lesions increase the number of perseverative errors and not neutral errors, whereas in Shisa6 KO mice, both the number of perseverative and neutral errors were reduced. Spatial reversal learning has been associated with the hippocampus in previous studies, as both disturbed hippocampal synaptic transmission induced by SynGAP1 deletion302 and reduced synaptic depression elicited by inhibition
of AMPAR endocytosis303, aff ect reversal learning in the MWM. Hence, altered synaptic
plasticity in the hippocampus, or disturbed hippocampal-cortical circuitry activity in the Shisa6 KO might contribute to altered cognitive fl exibility. Moreover, behavioral characterization of mice with deletions of the Discs Large Homolog (DLG) family of postsynaptic scaff olding proteins revealed that DLG2 and DLG3 (SAP102) contribute to complex cognitive processes like cognitive fl exibility304. Interestingly, DLG3 has been
identifi ed as Shisa6 interactor and the DLG3 KO displays higher cognitive fl exibility, measured by rate of extinction in an inhibitory response control task.
4
Supplementary informa� on
Supplementary fi gure 1. Experimental design behavioral phenotyping. Shisa6, -7 and -6/7 (d)KO mice were fi rst introduced in a PhenoTyper home cage to monitor general spontaneous ac� vity for three days, followed by discrimina� on learning and reversal learning. Subsequently, mice con� nued with a behavioral test ba� ery including the open fi eld, dark-light box, rotarod and grip strength measurement and T-maze. Shisa6 and -6/7 (d)KO mice con� nued with the Morris water maze, whereas for Shisa7 KO a separate batch of mice was tested in the Morris water maze. To account for diff erences in habitua� on and handling these mice were handled extensively before training started. Fear condi� oning experiments were performed with separate batches of Shisa6 and -7 KO mice, which were not handled before the start of the experiment.
Test battery
Separate batches:
Contextual short term memory Contextual long term memory Auditory long term memory
Fear conditioning Open field Dark-light box Rotarod T-maze PhenoTyper General activity CognitionWall Shisa6 Shisa7 Shisa6/7 Shisa6 Shisa7 no handling >7 days handling Shisa7
Morris Water Maze
Shisa6 Shisa6/7
4
0 5 10 15 20 25Training1 Training2 Training3 Training4 Training5
Ve lo ci ty (c m /s ) WT Shisa6/7 KO 0 5 10 15 20 25
Training1 Training2 Training3 Training4 Training5
Ve lo ci ty (c m /s ) WT Shisa6 KO 0 5 10 15 20 25
Training1 Training2 Training3 Training4 Training5
Ve lo ci ty (c m /s ) WT Shisa7 KO
