Global loss of Neuron‐specific gene 1 causes alterations in motor coordination, increased anxiety, and diurnal hyperactivity in male mice

Abstract The Neuron‐specific gene family (NSG1‐3) consists of small endolysosomal proteins that are critical for trafficking multiple receptors and signaling molecules in neurons. NSG1 has been shown to play a critical role in AMPAR recycling from endosomes to plasma membrane during synaptic plasticity. However, to date nothing is known about whether NSG1 is required for normal behavior at an organismal level. Here we performed a battery of behavioral tests to determine whether loss of NSG1 would affect motor, cognitive, and/or affective behaviors, as well as circadian‐related activity. Consistent with unique cerebellar expression of NSG1 among family members, we found that NSG1 was obligatory for motor coordination but not for gross motor function or learning. NSG1 knockout (KO) also altered performance across other behavioral modalities including anxiety‐related and diurnal activity paradigms. Surprisingly, NSG1 KO did not cause significant impairments across all tasks within a given modality, but had specific effects within each modality. For instance, we found increases in anxiety‐related behaviors in tasks with multiple stressors (e.g., elevation and exposure), but not those with a single main stressor (e.g., exposure). Interestingly, NSG1 KO animals displayed a significant increase in locomotor activity during subjective daytime, suggesting a possible impact on diurnal activity rhythms or vigilance. Surprisingly, loss of NSG1 had no effect on hippocampal‐dependent learning despite previous studies showing deficits in CA1 long‐term potentiation. Together, these findings do not support a role of NSG1 in hippocampal‐dependent learning, but support a role in mediating proper neuronal function across amygdalar and cerebellar circuits.


| INTRODUCTION
NSG family members are small, brain-enriched proteins that arose specifically during the evolution of the vertebrate clade 1 suggesting a potential role in more advanced cognitive, emotional and/or motor behaviors. Neuron-specific gene 1 (NSG1; NEEP21) is a 21 kDa, single transmembrane-containing protein critical for trafficking multiple cargo proteins through the secretory and endolysosomal system, exclusively in neurons. NSG1-dependent cargos include L1 cell adhesion molecule (L1CAM), 2 Neurotensin receptors 1-2, 3 as well as amyloid-precursor protein (APP). 4 However, the role of NSG1 in activity-dependent AMPAR trafficking during synaptic plasticity is the most well-characterized. NSG1 forms a multi-molecular complex involving interactions between GRIP1, GLUA2, and Syntaxin 13. 5 Antisense-mediated down-regulation of NSG1 impedes recycling of GluA1 and GluA2 in hippocampal neurons following treatment with NMDA. 5 NSG1 knockdown, or expression of a dominant negative NSG1 fragment (aa129-164), reduces both miniature AMPARmediated excitatory postsynaptic currents (mEPSCs) as well as evoked currents. Further, disruption of NSG1 function causes reductions in the degree of long-term potentiation (LTP) in response to high frequency stimulation in organotypic hippocampal slices. 6 Together, these findings strongly support NSG1 is a critical regulator of neuronal function and plasticity.
While NSG1 is expressed widely throughout the brain, 7 the role of NSG1 in specific neural circuits remains unknown. NSG1 is one of three members of the Neuron-specific gene family (NSG), the others being NSG2 (P19) and NSG3 (Calcyon, Caly). 8 While less is known about NSG2, 9,10 NSG3/Caly has been well characterized, and behavioral studies may provide clues as to the brain circuits regulated by other NSG family members. In contrast to NSG1, NSG3/Calcyon is critical for clathrin-mediated endocytosis of AMPARs, where overexpression reduces AMPAR surface expression and knockout (KO) impairs long-term depression (LTD; [11][12][13][14]). Overexpression (OE) of NSG3 causes hyperactivity when exploring an open field apparatus in addition to stereotypies, rearing and sniffing behaviors. 15 NSG3 OE animals also display reduced anxiety, spending more time in the light areas of a light-dark box and open areas of the elevated plus maze. 15 Further, NSG3 OE causes reductions in response inhibition during the extinction phase of context-dependent fear conditioning, despite normal acquisition of learning on this task. 13 Furthermore, NSG3 OE also caused significant perseverative errors on a version of the Morris Water Maze task which required animals to flexibly learn a new platform location following acquisition of an initial location.
While NSG3 OE animals could do this, probe trials revealed that OE animals spent far more time searching for the initial platform location, suggesting long-term inflexible behavior. 13 Thus, behavioral analysis of NSG1 KO animals could provide significant insight into how NSG1 regulates neuronal circuits, and possibly complements that of other NSG family members. Based on its relatively overlapping expression patterns with other NSG family members, 7,16 KO

| Immunohistochemistry
Wild-type and NSG1 KO mice were anesthetized and perfused with phosphate buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were removed and immersed in 4% PFA, 20% sucrose, and 30% sucrose, each for 12-24 h. Sagittal sections were sliced on a sliding knife microtome (American Optical). Free-floating sections were permeablized and blocked simultaneously in 0.5% Triton X-100

| Anxiety-related tasks
The open-field test was used to assess exploratory behavior, anxiety, and gross locomotion according to previous methods. 17  The EZM 20 was used to assess anxiety and exploratory behavior and is considered to be more anxiety specific than the open field test 21 and may pick up on subtle motor differences compared with the EPM. 22 The EZM was performed largely as described previously. 23 The appara-

| Motor-related tasks
The rotarod task was used to assess motor performance and fatigue resistance in rodents using the Panlab Rota Rod model LE8205. The rod was initiated to rotate at a constant initial speed of 4 RPM. Once the mouse was positioned, the rod accelerated to 40 RPM in 300 s.
The time spent on the rod and the RPM reached at the time of falling was captured for each of five trials. A 30-s intertrial interval (ITI) was used throughout the experiment. The chamber was cleaned using 70% ethanol between animals to eliminate olfactory distractions.
The CatwalkXT system (Noldus) was used to analyze gait and fine motor coordination according to methods previously described. 24 Mice were placed at one end of a 1.3 m glass runway and allowed to walk across to the other end to reach a dark goal box. Footprint detection occurred as animals passed through a black tunnel illuminated from one side by a reflected green fluorescent light. Footprints were captured by a high-speed camera across an area that was manually adjusted to capture at least three full stride-lengths. Three trials were captured for each animal during one testing session. Trials were classified as being compliant by the software if the mouse crossed the recording area under 5 seconds and did not show a maximum speed variation greater than 60%. The glass plate was cleaned using 70% ethanol between mice to eliminate olfactory distractions. Compliant trials were analyzed via automatic detection but were also reviewed manually; if the mouse stopped or turned in mid-run the trials were excluded. Analysis was performed using Catwalk XT 8.1 Software. Gait and movement were analyzed for right fore (RF), left fore (LF), right hind (RH) and left hind (LH) paws on the following variables: paw print area (size of paw print area during a full stance), stride length (distance between two consecutive paw placements of the same paw), swing (time interval between two consecutive paw placements of the same paw), and swing speed (velocity of an individual paw between two consecutive placements). The percentage of step patterns categorized as cruciate, alternate or rotary was analyzed as previously described. 25

| Learning and memory
Pavlovian trace fear conditioning (TFC) was used to measure associative learning using methods adapted from previous publications. 26 and followed by a 20-s interval that terminated with a 2 second 0.7 mA footshock (unconditioned stimulus [US]). After the first tone/shock pairing there was a 90-s no stimulus interval followed by another tone/ shock pairing. The tone/shock pairings were repeated 5 times in total with a randomized interval averaging 120 s. Each session was ended 60 s after the final pairing. To eliminate olfactory distractions, the chamber was cleaned using 70% ethanol between trials.
Memory for the CS was tested absent the shock in a novel environmental context 24 h after conditioning. The novel context was a 25 x 28 x 22 cm 3 chamber with black and white checkered walls and a different metal rod floor. The chamber was cleaned using 70% ethanol between trials to eliminate olfactory distractions. The protocol used was the same as the day prior without the US present.

| Validation of NSG1 KO mice
We first validated the loss of NSG1 protein in null animals. Figure 1A displays an agarose gel in which PCR amplicons of genomic DNA were visualized from an NSG1 heterozygote (lane 2), four NSG1 wild-type (lanes 3-6), and four NSG1 KO animals (lanes 7-10). Loss of protein expression was first validated by western blot of brain extracts taken from four wild-type and four NSG1 KO animals that were probed for NSG1 ( Figure 1B Figure 1D). Wild-type animals displayed similar expression patterns for both NSG1 (green) and NSG2 (red) across multiple brain regions.
One major exception is the robust expression of NSG1 in cerebellar Purkinje neurons, with a near absence of NSG2 expression, consistent with previous findings. 7 In contrast, sections from NSG1 KO animals showed complete absence of specific staining in all regions ( Figure 1D, green lower right panel). Thus, the gene targeting strategy employed in Barford et al. 7 specifically eliminated NSG1 protein expression throughout the brain.

| NSG1 KO mice display subtle motor coordination deficits but retain gross motor function
Because of the unique expression of NSG1 in cerebellum, 7 we first tested whether NSG1 KO animals would display motor deficits using two complementary methods. First, we used the rotarod test as a measure of gross motor impairment and learning, and predicted that NSG1 KO animals would display reduced latencies to fall off the beam during acceleration. Consistent with previous reports, 28  Bonferroni post-hoc test revealed a significant reduction in crucient step pattern (Ca; p = 0.04) with a concomitant increase in the complementary crucient step pattern (Cb; p = 0.024). In addition, we found a significant interaction between genotype and print area ( Figure 2D; , and DAPI (blue) on parasagittal brain sections from a wild-type (C) and KO animal (D). Note the robust cerebellar staining of NSG1 antibody in wild-type brain section while NSG2 signal is relatively low, as well as lack of NSG1 signal in a section from an NSG1 KO animal (bottom right panel in D) loss of NSG1 causes significant gait abnormalities but that these do not impair animals on gross motor coordination or learning compared with controls.

| NSG1 KO animals display increased anxiety on elevated tasks
We next tested for altered affective behavior using multiple tasks designed to evoke anxiety or fear. We first used the open field test to assess general ambulation, anxiety, and exploratory behavior. NSG1 KO mice showed no differences compared with wild-type animals in total distance traveled, suggesting there was no gross ambulation def- We next used a complementary method to assess whether NSG1 KO truly displayed heighted anxiety. The EZM has been purported to reflect purely anxiety-related behaviors by removing choice of which arms to enter and was found to be a more accurate anxiety assessment. 22 Furthermore, animals tend to spend more time in the open areas of the EZM, reducing the likelihood of a floor effect. In this task we found no significant difference in total distance traveled (p < 0.05) or number of entries (F [1,23]

| NSG1 KO mice do not display differences in associative learning and memory
Because NSG1 has been shown to be critical for the recycling of postsynaptic AMPA receptors 5 and stable LTP, 6 we tested KO mice on an associative memory task, thought to require proper trafficking of

| Loss of NSG1 increases activity during wake cycles
As sleep-wake cycles and circadian rhythms have become increasingly associated with neurological disorders, we used homecage monitoring to determine whether loss of NSG1 causes alterations in sleep-wake activity across several circadian cycles. Homecage monitoring serves as a beneficial platform for analysis of circadian locomotor activity without introducing an aversive environment like maze-based tasks. Seventy-two hours of continuous activity monitoring were captured to analyze spontaneous activity. Two complete circadian cycles following an acclimation period were used for analysis ( Figure 6A Figure 6C; p = 0.99), indicating that loss of NSG1 did not affect activity during subjective night. Together, these data implicate NSG1 as critical for maintenance of circadian amplitude and wake activity.

| DISCUSSION
Here we demonstrate that while KO of NSG1 does not significantly impact gross motor or cognitive function, NSG1 KO affects anxiety- As NSG1 is uniquely expressed in cerebellar Purkinje neurons among NSG family members, 7 we were particularly interested in its importance for motor function. While we did not observe significant deficits on gross motor function or learning on the rotarod task, we saw significant alterations in gait and coordination. These took the form of overall decreased paw print size as well as altered step patterns, which are observed in genetic insults that cause cerebellar dysfunction. 29 31 On the other hand, stimulating the vermis induces fear responses, such as increased amplitude of the acoustic startle response. 32 Future studies using the conditioned eyeblink task across multiple NSG family member KO mice (e.g., NSG2-3), could help determine whether the cerebellum may be a locus for anxiety-related behavioral changes specifically in NSG1 KO mice.
The current model of NSG-mediated AMPAR trafficking within dendritic spines suggests a primary role of NSG3 (Calcyon) with the F I G U R E 5 NSG1 KO animals display normal associative memory. (A) Percentage of time spent freezing is illustrated across five acquisition trials. (B) Compared with wild-type mice, NSG1 mutants exhibited no difference in the overall percentage of time spent freezing during the trace period (B; p > 0.05) nor during the tone itself (C; p > 0.05) when presented in a novel context 24 h following training. Results are expressed as mean ± SEM. n = 12/genotype F I G U R E 6 NSG1 KO selectively increases diurnal activity. (A) Homecage behavioral monitoring using photocell beam breaks to measure horizontal activity across 48 h. The first 9 h were considered acclimation and excluded from subsequent analyses. There was a highly significant effect of phase (light vs. dark) where animals showed significantly more activity during dark periods (p < 0.0001). (B) NSG1 KO animals were significantly more active than wild-type animals during the dark phase (p < 0.05), while no difference was found for the light phase (p > 0.05). n = 11 controls and 12 NSG1 KOs endocytosis of surface AMPARs, and NSG1 with recycling of internalized AMPARs back to the plasma membrane during plasticity-inducing stimuli. 8  While activity data support the inverse roles of NSG1 and NSG3, data from other tasks appear to contradict this model. For instance, NSG3 OE animals displayed reduced anxiety, spending more time in the light areas of a light-dark box. 15 In contrast, NSG1 KO animals showed enhanced anxiety in similar tasks (EPM/EZM). Furthermore, the impacts of NSG1 loss compared with NSG3 OE appear to diverge with respect to hippocampal-dependent learning and the molecular mechanisms for these circuits. Animals with NSG3 OE displayed enhanced, perseverative freezing in fear conditioning tests as well as perseverative exploration in Morris water maze when platform location was changed. 13 In contrast, NSG1 is dispensable for hippocampaldependent associative learning according to our data. All of these behaviors are largely dependent on processes thought to involve alterations in post-synaptic AMPARs. 33 As mentioned, NSG1 is thought to play a primary role in AMPAR recycling following potent NMDA receptor activation or those that cause LTP. 5,6 However, recent data by Yap and colleagues found robust colocalization between NSG1 and NSG2 in early and late endosomes, but not robust localization with Rab11 + recycling endosomes. 10 Further, our recent findings place NSG2 at a relative minority of synapses, 9 which predicts that NSG1 may lack universal synaptic localization as well. Regardless, the robust colocalization of NSG1 and NSG2, 10 along with data showing that overexpression of NSG2 increased synaptic strength, 9 suggest a possible compensatory effect of NSG2 in NSG1 KO animals. Our current data argue that NSG2 alone may be sufficient to promote AMPAR exocytosis to synapses under plasticity-inducing conditions which may occur via a distributed set of vesicular compartments. 34 Thus, future studies should revisit the involvement of NSG1 in post-synaptic AMPAR regulation in hippocampal plasticity and examine whether NSG2 has a similar or disparate role in AMPAR recycling and exocytosis.
NSG family members are involved in a number of other cellular processes in addition to AMPAR trafficking during synaptic plasticity including trafficking of L1CAM, 2 Neuregulin, 35 Neurotensin receptors 1-2, 3 as well as β-APP. 4 Furthermore, while NSG1 and NSG2 appear to be expressed at similar levels to NSG3 in adults, they show their highest expression during development. 36,37 Thus, additional factors need to be incorporated into models of how they affect overall circuit function. Regardless of mechanism, future studies should examine whether NSG1 OE or NSG3 KO produce the opposite results in more complex cognitive tasks of learning, memory and flexibility to see what circuits utilize NSG1 and NSG3 in a convergent or divergent manner. In addition, we recently showed that NSG2 promotes synaptic strengthening, 9 potentially by promoting AMPAR insertion into a subset of post-synaptic specializations. Thus, future studies using NSG2 single KO and NSG1/NSG2 double mice should inform whether behavioral abnormalities are because of the interaction of both proteins on plasticity-induced changes to AMPAR surface expression at a subset, or all synapses. In addition, although sex differences have not been published for expression or function of NSG1/2, future studies should examine whether female animals show similar or unique phenotypes following KO.