Gordon Holmes Syndrome Model Mice Exhibit Alterations in Microglia, Age, and Sex-Specific Disruptions in Cognitive and Proprioceptive Function

Gordon Holmes syndrome (GHS) is a neurological disorder associated with neuroendocrine, cognitive, and motor impairments with corresponding neurodegeneration. Mutations in the E3 ubiquitin ligase RNF216 are strongly linked to GHS. Previous studies show that deletion of Rnf216 in mice led to sex-specific neuroendocrine dysfunction due to disruptions in the hypothalamic–pituitary–gonadal axis. To address RNF216 action in cognitive and motor functions, we tested Rnf216 knock-out (KO) mice in a battery of motor and learning tasks for a duration of 1 year. Although male and female KO mice did not demonstrate prominent motor phenotypes, KO females displayed abnormal limb clasping. KO mice also showed age-dependent strategy and associative learning impairments with sex-dependent alterations of microglia in the hippocampus and cortex. Additionally, KO males but not females had more negative resting membrane potentials in the CA1 hippocampus without any changes in miniature excitatory postsynaptic current (mEPSC) frequencies or amplitudes. Our findings show that constitutive deletion of Rnf216 alters microglia and neuronal excitability, which may provide insights into the etiology of sex-specific impairments in GHS.

Functionally, CHIP/STUB1, RNF216, and OTUD4 encode for classes of protein enzymes that are responsible for adding and removing an abundant post-translational tag called ubiquitin to other proteins.Ubiquitin is a 76 amino acid protein that is covalently attached to other proteins in the cell.This posttranslational modification regulates diverse biological processes that include signal transduction, protein degradation, cell cycle regulation, DNA repair, inflammation, and neural development (Hershko and Ciechanover, 1998;Mabb and Ehlers, 2010;Upadhyay et al., 2017;Mabb, 2021).This ATP-dependent reaction utilizes an enzymatic cascade consisting of a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin ligase (E3).These three types of enzymes act in concert to attach diverse configurations of ubiquitin to substrate proteins (Komander and Rape, 2012;Zheng and Shabek, 2017).Protein ubiquitination can also be reversed by deubiquitinases (DUBs), which remove a variety of ubiquitin configurations from substrates (Komander et al., 2009).CHIP/STUB1 and RNF216 are both classified as E3 enzymes, whereas OTUD4 is characterized as a DUB.In GHS, missense mutations within the catalytic regions of RNF216 and CHIP/STUB1 decrease the ability of these E3s to ubiquitinate their targets (Shi et al., 2014;Husain et al., 2017).Cumulatively, these findings strongly imply that disruptions in the ubiquitination of substrates facilitated by RNF216 and CHIP/STUB1 most likely contribute to GHS pathogenesis.
In a previous study, constitutive Rnf216 KO mice were found to have sex-specific alterations in reproductive phenotypes that were accompanied with altered microglia in the preoptic area of the hypothalamus (George et al., 2022).Given that GHS is a rare disorder with reported clinical differences in males and females, comorbidity of motor and learning impairments, and limited longitudinal tracking for pathogenesis, we monitored motor and cognitive functions in Rnf216 male and female KO mice up to 1 year of age.Here, we found that KO females displayed abnormal limb clasping without any outstanding motor phenotypes in KO males and females.When evaluating cognitive function, Rnf216 constitutive KO mice exhibited spatial and associative learning impairments that were age dependent.These alterations in learning and memory were associated with sex-specific changes in reactive microglia in the hippocampus and cortex.Although male and female KO mice had no changes in CA1 hippocampal miniature excitatory postsynaptic current (mEPSC) frequencies or amplitudes, KO males had a more negative resting membrane potential (RMP).Our findings show that deletion of Rnf216 in mice leads to alterations in microglia and neuronal excitability, which may lead to impairments in select behavioral outputs.

Materials and Methods
Animals.Mice were kept in standard housing with littermates, provided with food and water ad libitum, and maintained on a 12 h light/dark cycle.All behavioral tests were conducted in accordance with the National Institutes of Health Guidelines for the Use of Animals.Mice were treated in accordance with the Animal Welfare and Ethics Committee (AWERB), and experiments were performed under the appropriated project licenses with local and national ethical approval.All experimental mice used in this study were housed separately by sex and in groups of 4-5 with balanced WT and KO genotypes.Sample sizes for behavior and immunohistochemistry experiments were calculated using variance from previous experiments to indicate power, in which statistical analysis for significance was set at 95%.All behavioral studies and isolation of body tissue for biochemical experiments were approved by the Animal Care and Use Committee.
Generation of Rnf216 knock-out mice.Embryonic stem (ES) cell clones were generated to target exons 4-5 of the Rnf216 gene on mouse chromosome 5, which prevents the production of all isoforms (International Knockout Mouse Consortium).ES cell clones were injected into blastocytes and implanted into pseudopregnant mice.These mice were crossed to heterozygous mice for Flp recombinase to excise out the LacZ/neomycin cassette to obtain one Rnf216 allele flanked by loxP (fl) sites to generate C57BL/6N-Rnf216<tm1c(EUCOMM)Wtsi>/Tcp (Rnf216 wt/fl ; Canadian Mouse Mutant Repository at the Hospital for Sick Children).Homozygous floxed conditional male mice (Rnf216 fl/fl ) were crossed with homozygous CMV-CRE +/+ female mice (The Jackson Laboratory, JAX stock #006054) to allow the CRE to excise out exons 4 to 5, creating a dysfunctional gene.To breed out the CRE, we bred female offspring that were heterozygous for loxP and CRE (Rnf216 wt/fl ::CMV-CRE −/+ ) with WT males to generate heterozygous Rnf216 −/+ mice.Male Rnf216 −/+ mice were then bred with WT females to generate Rnf216 −/+ mice.Rnf216 −/+ male and female mice were then bred together to generate the experimental mice used for this study [Rnf216 +/+ (WT), Rnf216 −/+ (HET), and Rnf216 −/− (KO)].
Righting reflex.Mice were tested at ages postnatal day (P) 7 and P12 for developmental motor impairment.Briefly, mouse pups were separated from their mother and placed in a small paper box.One pup was removed from the box and placed in a supine position on a paper surface.Upon release of the mouse, the experimenter measured the time it took for the pup to return to a prone position (assessed by ensuring all four paws touched the ground surface).Pups were placed back in the box to rest for 1 min, and the procedure was repeated one additional time.Following the final time, the pup was immediately returned to its mother so as not to induce excessive stress.
Grip strength.Mice were tested in grip strength for muscular strength at ages P60 and P365.Briefly, we measured the peak force applied by the forelimbs of mice using a grip strength meter (Ugo Basile).Mice were acclimated to the testing room 30 min prior to performing the test.To measure forelimb strength, we gently lowered the mouse onto the grid with its two forepaws.The investigator gently pulled back on the tail of the mouse, ensuring that the torso of the mouse remained horizontal and peak force was measured.The mouse was placed back in its home cage for at least 5 min to rest.The procedure was repeated two additional times, and the apparatus was cleaned with 70% ethanol between each mouse.
Rotarod.Mice aged P65 and P370 were tested for balance, coordination, physical condition, and motor planning using the Rotarod apparatus (Ugo Basile).Tests were performed for 5 consecutive days, and mice were returned to standard housing daily after each day of testing.Briefly, mice were acclimated to the testing room for at least 30 min prior to performing the assay.Mice were placed horizontally on a rotating cylinder (rod) which was suspended 4-6 inches above the cage floor and spinning at 4 rpm.Thick bench paper was placed on the bottom of the rotarod to minimize stress if the mouse fell off.The rotarod speed was increased gradually (20 rpm/min for a maximum speed of 40 rpm), and the time (latency) it took for the mouse to fall off the apparatus was measured.Any mouse that fell off the apparatus was retrieved by the investigator and placed back in its home cage.For the first session (day 1), mice were given three trials with a 60 s rest in between each trial.
Open field.Mice aged P62 and P367 were tested in the open field activity apparatus (Med Associates), which uses an array of infrared (IR) photo beams to measure locomotion and motor activity in a square arena.The apparatus consisted of three sets of matching IR photo beams that projected across the open field along three axes: x, y, and z.The software, Activity Monitor 7, detected when and where the photo beams were disrupted by the presence of the mouse.The software also breaks the field arena into zones, or areas of interest, depending on the experiment; data analysis displayed the measurements accordingly.Briefly, mice were acclimated to the testing room for at least 30 min prior to performing the assay.Mice were placed in the novel open field chamber for a total time of 15 min.After the assay, mice were returned to standard housing.
Clasping.A separate cohort of mice was used to evaluate clasping to avoid any changes related to behavioral tasks.Mice from this cohort were tested for clasping at 3, 9, and 41 weeks of age.A portion of this cohort was used for brain harvesting and measurements of cerebellar weights.Mice were suspended in the air by their tail for two sessions, 30 s each.A score of 0 was defined as no clasping with both forelimbs and hindlimbs extended; a score of 1 was flexion of forelimbs without clasping of hind limbs; a score of 2 was assigned when forelimbs and hindlimbs were clasped.
Barnes maze.Mice aged P70 and P375 were first habituated to the Barnes maze 2 d prior to testing.On the first day of habituation, each mouse was allowed to explore the maze for ∼30 s and then gently directed to a random hole (not used during testing).The mouse stayed in the hole for 1.5 min and then was returned to their home cage.The maze was sprayed with 70% ethanol between each mouse.One day prior to testing, the mice were given a chocolate-flavored pellet treat (Bio-Serv Supreme Mini-Treats) in their home cage.Each mouse was tested only once per day.Prior to testing sessions, mice were acclimated to the room 30 min.Before each session (including the testing session), the mice were held in an opaque cylinder in the middle of the maze before recording.Each session cutoff was set to 15 min.If the mouse did not find the exit hole during the 15 min, then the investigator guided the mouse into the hole.The learning task was a 10 day duration.On days 1-5 (training), the exit hole contained a chocolate treat to associate entry into the exit box as a positive and motivating experience.Once the session was completed, the mouse was held in a holding cage until all mice in the home cage were tested.On days 6-10 (learning), the chocolate-flavored treat was given in the home cage after each mouse in the cage was done testing.During the reversal phase on days 11-16 (reversal), the exit hole was rotated 180°, and the mice were still given a chocolate-flavored treat in the home cage after testing.Mice were excluded from the study if they did not enter the correct exit hole within the 15 min time limit by day 7.The scoring and criteria for search strategy was performed as previously described (Wall et al., 2018).
Fear conditioning.Mice aged P90 and P395 were acclimated to the testing room for at least 30 min prior to performing the assay.On day 1 (conditioning), each mouse was placed in the testing chamber that contained a sound-attenuating box, and the mouse was allowed to explore for 2 min.Mice were then exposed to a 30 s tone (80 dB), followed by a 2 s scrambled foot shock (0.4 mA).Mice received two additional shock-tone pairings, 80 s between each pairing.Mice were then returned to a new, clean cage after testing.After all mice from one cage were tested, they were returned to the original home cage.On day 2 (context-dependent recall), mice were placed back in the original conditioning chamber for a test of contextual learning across a 5-6 min session.There was no shock or tone.On day 3 (cue-dependent recall), mice were placed back in the original conditioning chamber with new inserts attached to the walls of the chamber to provide a different environment.There was an auditory cue similar to day 1 without the shock in a 5-6 min session.
Western blotting.Mouse tissue was thawed on ice and 300-500 µl of RIPA buffer was added.The tissue was then homogenized using sterile pestles.The samples were then centrifuged at 14,000 rpm for 15 min at 4°C, and the supernatant was collected for protein quantification.The protein concentration of the soluble fraction was measured using the Pierce 660 nM Protein Assay Kit (Thermo Fisher).The samples underwent SDS-polyacrylamide gel electrophoresis and were transferred on a nitrocellulose membrane (Bio-Rad) for 1 h at 70 mV.The blots were incubated overnight at 4°C with blocking buffer [Intercept (TBS) blocking buffer, Li-COR].Membranes were then probed with the following primary antibodies prepared in a 1:1 ratio of TBST (1× TBS, 0.1% Tween 20) and blocking buffer solution containing a 1:1,000 dilution of 20% NaN 3 , rabbit polyclonal anti-RNF216 [LifeSpan Biosciences (amino acids 100-150), 1:1,000] and mouse anti-β-actin (GeneTex, 1:3,000), and then incubated overnight at 4°C.Membranes were washed three times in distilled water.The following secondary antibody dilutions were prepared in a 1:1 ratio of TBST and blocking buffer solution with 1:2,000 20% SDS, IRDye 680RD goat anti-mouse IgG (H + L; Li-COR, 1:20,000) and IRDye 800CW goat anti-rabbit IgG (H + L; Li-COR, 1:15,000), and were incubated for 1 h at room temperature.Membranes were washed two times in TBST and then two times with distilled water.Blots were imaged using the Odyssey CLx Imaging System (LI-COR) with a resolution of 169 µm, medium quality, and a 0 mm focus offset.Images were processed using the Gel Analysis tool in ImageJ using individual channels.Briefly, boxes were drawn around each band.Once the lanes were labeled and plotted, the area of the peaks was selected and measured.For each blot, proteins of interest were normalized to the Beta-ACTIN loading control.
Microglia staining, imaging, and analysis.Male and female WT and KO mouse brains were sectioned at 40 µm.Sections for each mouse were selected based on stereological matching using the mouse brain stereotaxic coordinate reference atlas.Samples were then blinded and processed as follows: Free-floating sections were rinsed with 3% hydrogen peroxide two times for 7 min each to remove any endogenous peroxidases and then washed six times in 1× KPBS (in mM, 1.6 NaCl, 0.4 K 2 HPO 4 , and 0.09 KH 2 PO 4 dissolved in ddH 2 O to make a 10× solution and diluted to 1×) at room temperature.Sections were then incubated overnight at room temperature in 1× KPBS containing 1.0% Triton and 1:50 K dilution of goat anti-Iba1 (Novus Biologicals).Sections were then washed 10 times with 1× KPBS before incubation in 1× KPBS containing 0.4% Triton and 1:600 dilution of donkey anti-goat biotin-SP (Jackson ImmunoResearch) for 1 h at room temperature.Sections were washed five times in 1× KPBS before being incubated in avidin-biotin-peroxidase complex (1:10, ABC Elite Kit, Vector Laboratories) for 1 h at room temperature.After rinsing sections three times in 1× KPBS and three times in 0.175 M sodium acetate buffer, Iba1 immunoreactivity was visualized with nickel sulfate enhanced 3,3 ′ -diaminobenzidine (DAB) solution (0.2 mg/ml) containing 0.08% hydrogen peroxide in 0.175 M sodium acetate buffer.Sections were incubated in the DAB solution for 15 min before rinsing three times with 0.175 M sodium acetate buffer followed by three times in 1× KPBS.Sections were then mounted onto gelatin-subbed slides, air-dried, then dehydrated in a graded series of alcohols, and cleared with xylenes.Slides were then coverslipped with Permount mounting media.Bright-field images of the dorsal hippocampus and S1 cortex were acquired with the researcher blinded to condition on a Keyence BZ-X700 (Keyence) microscope using a 20× Plan Apochromat 0.75 N.A. objective with a z-step size of 1 µm.For densitometry, background subtraction and thresholding were first performed on the images using FIJI/ ImageJ (National Institutes of Health).Cell count analysis was performed using CellProfiler 4.2.1 (Broad Institute; Carpenter et al., 2006).As the morphology of multiple microglia were analyzed from one mouse, the hierarchical bootstrapping and permutation tests were employed as they do not require the assumption of sample independence (Saravanan et al., 2020).The area of microglia soma in the S1 cortex and dorsal hippocampus was measured on maximum intensity projection images by using the freehand selection tool in FIJI ImageJ2 version 2.9.0/1.53t.Normal distribution scores for the mean were then analyzed by hierarchical bootstrapping using the ClusterBootstrap package program (Deen and de Rooij, 2020) in R Software version 4.2.3 (Free Software Foundation).For quantification of microglia branching, microglia were traced using the Simple Neurite Tracer (SNT) plugin for FIJI ImageJ2.The intersections of skeletonized microglia were then measured using the Sholl analysis plugin in FIJI ImageJ2 with the concentric circle set to 1 µm increments.The area under the curve (AUC) for each microglia was then assessed using the ClusterBootstrap package program in R. The permutation testing was set at 1,000 random shufflings, which allowed us to determine the robustness of the estimated means.Data were visualized as a probability density function implemented in MS Excel 2023 version 16.75.2(Microsoft).Representative microglia reconstructions were created from maximum intensity projection images using the SNT plugin tool fill option.Constructed images were exported and cropped using identical scales.
Electrophysiology.Male and female WT and Rnf216 KO (P28) mice were sacrificed, and hippocampal slices were prepared for electrophysiological recordings.After decapitation, the brains were dissected out quickly and immediately immersed in oxygenated (95% O 2 and 5% CO 2 ) ice-cold artificial cerebrospinal fluid (aCSF) containing the following (in mM): 126 NaCl, 2.5 KCl, 2.4 CaCl 2 , 1.2 NaH 2 PO 4 , 1.2 MgCl 2 , 11.1 glucose, 21.4 NaHCO 3 .Hippocampal transverse slices (300 µm thick) were cut by using a Leica VT1000S Vibratome (Leica Microsystems).The brain slices were transferred to a holding chamber filled with oxygenated aCSF and kept at room temperature (25°C) for 1 h before recording.For recording, a single brain slice was transferred to the recording chamber and continuously perfused with oxygenated aCSF maintained at 30°C.
All recordings were performed using an Axopatch 700B amplifier and a Digidata 1440A data acquisition system (Molecular Devices).Whole-cell voltage-clamp recordings of mEPSCs were performed on an Olympus upright microscope (BX51WIF) with a 40× water immersion objective.The holding potential of all neurons was −70 mV.Recordings were filtered at 10 kHz and digitized at 10 kHz.The electrode internal solution contained the following (in mM): 120 K-gluconate, 10 KCl, 2 MgCl 2 , 10 HEPES, 0.1 CaCl 2 , 1 EGTA, 3 MgATP, and 0.5 NaGTP, pH 7.2 (with KOH).Only recording pipettes with a resistance of 2-3 MΩ after being filled with internal solution were used.Synaptic blockers APV (25 µM), picrotoxin (100 µM), and action potential blocker tetrodotoxin (1 µM TTX) were added in aCSF during the recordings.Series resistance was monitored, and the cells were discarded when the series resistance increased >15%.The liquid junction was −8 mV detected from the intracellular solution and subtracted from the membrane potential values in the final result.The mEPSCs were analyzed using MiniAnalysis with threshold events set at 5 pA.Electrophysiology recordings and data analysis were conducted with the experimenter blinded to genotype.
Statistical analysis.Statistical analyses applied were the post hoc Student's t test, one-way ANOVA, chi-square test (for categorical limb clasping data), Pearson's correlation, and two-way and three-way ANOVA with multiple comparisons.For k = 2, data were analyzed using Student's t test.When k > 2, data were analyzed for significant main effects of genotype, age, and sex using ANOVA.When time (trials) was a factor or there were missing values, data were analyzed using a mixed model ANOVA.If no significant main effect of sex was observed, then data were collapsed to increase power.Sidak's or Tukey's tests were used for comparing group means only when a significant F value was determined.For all comparisons, significance was set at p < 0.05.Data presented in figures and tables are means (± SEM).See Extended Data Table 1 for statistical reporting table.
Code accessibility.The code for microglia normal distribution scores by hierarchical bootstrapping is provided as Extended Data and is freely available online at https://github.com/mabblab/Microglia-analysis.Code was executed on a MacBook Air running the macOS Ventura 13.4.1 with hardware specifications of 8 GB of RAM and a 1.6 GHz Dual-Core Intel Core i5 Processor.

Results
No gross motor impairments in Rnf216 KO mice Rnf216 knock-out (KO) mice have reproductive impairments with KO males having reduced testicular weights and females displaying alterations in estrus cycling (George et al., 2022).In addition to these neuroendocrine abnormalities, GHS individuals with mutations predicted to be deleterious for RNF216 also exhibit ataxia and dementia with corresponding cortical and cerebellar atrophy (Margolin et al., 2013;Alqwaifly and Bohlega, 2016;Calandra et al., 2019).However, there are no studies that have established a role for Rnf216 in motor function and learning, which are core phenotypes that are disrupted in individuals with GHS (Holmes, 1908;Margolin et al., 2013).As expected, we found a significant decrease in RNF216 in the hippocampus, cortex, and cerebellum in male and female Rnf216 KO mice (Fig. 1A).
To establish a role for Rnf216 deletion on GHS-related phenotypes, we devised a battery of behavior tests to evaluate motor and memory function in adult and middle-aged KO mice (Fig. 1B).Righting reflexes were found to be normal in P7 and P12 KO males and females suggesting that motor development was not grossly impaired in developing mice (Fig. 1C).We next measured peak force using the grip strength assay but found no significant differences (Fig. 1D,E, left), indicating that KO mice do not have deficiencies in neuromuscular strength.Based on these data, we sought to determine if KO mice had alterations in motor learning and coordination.We measured the total distance traveled and rearing movements in an open field test in adult and middle-aged mice (Fig. 1D,E, middle and right).Although there was no significant difference in the distance traveled in adult mice (Fig. 1D, middle), there was a sex-dependent reduction in the number of rearing movements in adult female mice (Fig. 1D, right).Middle-aged KO mice had an increase in the distance traveled in the center of the open field (Fig. 1E, middle), but there were no significant differences in the number of rearing movements between sex or genotype (Fig. 1E, right).Next, we measured another metric of motor function using the accelerating rotarod but found no differences in the latency to fall or maximum speed in adult and middle-aged KO mice (Fig. 1F,G).
Because cerebellar ataxia mouse models are known to display abnormal limb-clasping reflexes (Lalonde and Strazielle, 2011;Kojic et al., 2018), we measured the limb-clasping reflex in adult and middle-aged mice.Under normal conditions, adult mice splay out their forelimbs and hindlimbs away from their abdomen when held upside-down by their tail.However, limb reflex abnormalities result in the clasping of the forelimbs and hindlimbs when held upside-down.Limb-clasping was scored in WT and KO mice at different developmental ages that included weaning (3 weeks), young adult (9 weeks), and middle-aged (41 weeks).Although there was no significant difference in KO males (Fig. 2A), a small percentage (4.17%) of KO females had abnormal forelimb and hindlimb clasping that appeared to emerge as early as 9 weeks and increased at 41 weeks of age (11.54%;Fig. 2B).The onset of abnormal clasping in females could be related to reductions in cerebellar volume.However, we did not find any genotypic differences in cerebellar brain weights in 16-and 52-week-old male and female KO mice (Table 1).Taken together, these findings indicate that motor learning and coordination are intact in male and female Rnf216 KO mice up to 1 year of age.However, given the limb-clasping phenotype, a subset of female Rnf216 KO mice may have compromised proprioception.

Middle-aged Rnf216 KO mice reveal nuances in hippocampal-dependent learning and memory
Because GHS individuals with RNF216 mutations exhibit dementia, we reasoned that Rnf216 KO mice would develop impairments in spatial learning and memory.To explore the role of RNF216 in hippocampal-dependent spatial learning, we used the Barnes maze task (Eales et al., 2014).Mice were tested for 16 consecutive days to evaluate training (days 1-5), learning (days 6-10), and reversal learning (days 11-16; Fig. 3A).In both WT and KO mice, we found that the distance traveled significantly decreased during the training and learning phases (days 1-10) and during the reversal phase (days 11-16), indicating that all mice were able to learn the location of the exit hole.Evaluation of distance traveled in each phase did not reveal significant differences in sex or genotype (Fig. 3B).Next, we measured the number of errors performed by each mouse during training, learning, and reversal but found no sex-or genotype-specific differences (Fig. 3C).We also calculated the quadrant bias ratio as a measure of the frequency of visitation to the quadrant containing the exit hole during the training/learning phases (days 1-10) but also found no significant differences (Fig. 3D, left).Finally, we calculated the perseverance ratio as a reflection of the frequency of visitation to the previous location of the exit hole during the reversal phase but again found no significant differences (Fig. 3D, right).The employment of specific strategies to evaluate efficiency can also be utilized in the Barnes maze task.As described previously, we divided navigation strategies into three different categories: spatial, serial, and random (Wall et al., 2018).When separated into these different phases, we found no sex-or genotype-specific differences in strategy selection across phases (Fig. 3E-G).
Given the lack of phenotype in adult KO mice, we next measured learning phenotypes in middle-aged mice (>1 year).Although there were no genotypic differences in KO males or females in total distance traveled, quadrant bias ratio, or perseverance ratio (Fig. 4A-D), there were sex-specific differences in the number of errors selectively during reversal with a significant effect found between male and female KO mice (Fig. 4C).Upon separating the assay into time blocks of phases, sex differences were not observed in strategy usage between groups (Fig. 4E-G, Extended Data Table 1).However, there was a main effect of genotype on serial and random strategy usage (Fig. 4F,G).Rnf216 KO mice had a significant increase in using the serial strategy across training, learning, and reversal phases with a reduction in the random search strategy during the training phase.These findings show that upon aging, Rnf216 KO mice selectively adopt alternative search strategies in a hippocampal-dependent spatial task.
In addition to testing hippocampal-dependent spatial memory, we evaluated another type of learning that included associative learning through fear conditioning.In our fear conditioning task, mice were given three pairs of a tone associated with a shock (day 1).On day 2, mice were tested for context recall (same environment), and on day 3, mice were tested for cue recall (different environment + tone).Although there were no differences in time spent freezing, number of freezing episodes, or length of bout during conditioning, context recall, or cue recall in adult male and female KO mice (Fig. 5A-C), middle-aged KO mice had an overall decrease in freezing time during context recall (Fig. 5E).
We next compared Barnes maze strategy behavioral alterations between adult and middle-aged mice.In both WT and KO mice, we observed significant shifts in correlations suggesting a transition in the dependence of strategy usage as mice age (Fig. 6A,B).WT middle-aged mice overall had reduced use of the serial strategy and adopted the inefficient random search strategy, which was confirmed by a strong significant correlation between spatial and serial strategy use in adult WT (Fig. 6A, top) and significant correlations between spatial and random strategies in middle-aged WT mice     (Fig. 6B , top).Although distinct patterns between adult and middle-aged mice emerged in both genotypes, the patterns in middle-aged KO mice were highly like those of adult WT mice, where the correlation between spatial and serial strategy use was strongest (Fig. 6B, bottom).Cumulatively, these findings suggest that middle-aged Rnf216 KO mice exhibit strategies that are more tuned to adult WT mice, which could reflect a developmental stunting of strategy formation.Overall, we conclude that middle-aged KO mice have a breakdown in performance necessary to support the use of age-appropriate strategies.Moreover, middle-aged Rnf216 KO mice begin to develop subtle fear learning impairments, which is concordant with the development of dementia in individuals with GHS (Margolin et al., 2013;Alqwaifly and Bohlega, 2016).
Rnf216 KO males display lower intrinsic excitability in hippocampal CA1 neurons Due to the role of RNF216 in synaptic plasticity and receptor trafficking (Mabb et al., 2014;Wall et al., 2018;George et al., 2022) and our findings related to fear conditioning, we sought to determine if neurotransmission was altered in Rnf216 KO mice.We measured mEPSCs in CA1 neurons of acute hippocampal slices (Fig. 7A) in P28 mice.In both males and females, there were no differences in amplitude or frequency (Fig. 7B-D); however, there were sex differences in rise and decay times (Fig. 7C,D).Notably, Rnf216 KO males had a more negative RMP (Fig. 7E), suggesting a decreased probability of firing action potentials in male KO mice.There were no significant differences observed in females.Adult Rnf216 KO mice have altered microglia in the hippocampus and cortex RNF216 has a function in innate inflammatory signaling (Chuang and Ulevitch, 2004;Fearns et al., 2006;Nakhaei et al., 2009) and phenotypes related to neuroinflammation were previously observed in the preoptic area of the hypothalamus of adult Rnf216 KO male mice (George et al., 2022).We analyzed microglia density, area, soma size, and morphology in the dorsal hippocampus and S1 cortex of adult Rnf216 WT and KO mice, by labeling microglia with the microglia-specific marker Iba1.Although there were no sex or genotypic differences in Iba1 total cell area (Fig. 8B, left), we observed a significant increase in the density of microglia selectively in Rnf216 KO female mice (Fig. 8B, right).We further identified sex differences between WT male and female microglia soma size and a sex-specific reduction in male KO soma size (Fig. 8C).When evaluating microglia morphology using the Sholl-based method, sex differences in male and female microglia were observed with no genotypic differences.Overall, female WT and KO microglia at this age had an increase in the number of intersections compared with males (Fig. 8D,E).Surprisingly, there were no significant differences between WT and KO male microglia (Fig. 8D,E).When evaluating microglia in the S1 cortex (Fig. 8F), there were no differences in Iba1 area (Fig. 8G, left).Like the hippocampus, KO females had a selective increase in microglia density (Fig. 8G, right).Significant decreases in soma size were still observed in male KO mice, but there were no sex differences (Fig. 8H).Upon inspection of microglia morphology using Sholl analysis, there were significant changes between all groups (Fig. 8I,J), with male KO mice exhibiting a dramatic reduction in the number of intersections (Fig. 8I,J).Taken together, our findings indicate that global deletion of Rnf216 leads to sex-and region-specific alterations in microglia characteristics in select brain regions.These alterations in microglia profiles over time may lead to proprioceptive phenotypes and increase susceptibility to cognitive disruptions observed in middle-aged GHS model mice.

Discussion
In this study, we evaluated changes in motor and memory using Rnf216 constitutive KO mice over the time course of a year.We speculated that deletion of Rnf216 would result in motor and learning impairments that increase in severity upon age, similar to phenotypes observed in GHS patients (Margolin et al., 2013).In GHS, the onset of motor disturbances ranges in age from teens to middle adult age.Our 1-year-old mice are similar to middle-aged adults, which is why we selectively chose this time point for our behavior analysis (Dutta and Sengupta, 2016).However, to our surprise, we were unable to detect motor impairments or other measures of behavior that would be indicative of ataxia in these mice, which is a core feature of GHS (Gonzalez-Latapi et al., 2021).This lack of phenotype was also coupled with our inability to detect decreases in cerebellar mass that is a distinguishing feature in all late-stage GHS patients (Seminara et al., 2002;Margolin et al., 2013;Alqwaifly and Bohlega, 2016;Calandra et al., 2019;Chen et al., 2022).Both male and female Rnf216 KO mice developed normal motor functions as demonstrated by righting reflexes and hindlimb clasping at 3 weeks of age.Deficits in the rotarod, open field, and grip strength tasks were also not observed, except for middle-aged KO mice traveling an increased distance in the center of the open field.However, Rnf216 KO females did develop abnormal limb clasping beginning at 9 weeks of age, which affected a larger fraction of female KO mice at almost 1 year, suggesting that females may begin to develop proprioceptive deficits.
Surprisingly, adult Rnf216 KO mice did not exhibit learning deficits in the Barnes maze spatial task, but middle-aged KO mice exhibited a bias in usage of a serial strategy across all task phases.Increases in serial strategy use were also observed in a previous transgenic mouse model for the RNF216 substrate Arc (Mabb et al., 2014;Wall et al., 2018).Here, mutation of RNF216-dependent Arc ubiquitination sites led to selective reversal learning deficits in juvenile male constitutive Arc knock-in mice (ArcKR; Wall et al., 2018).However, in contrast to our findings here, ArcKR mice had an increase in error number that was also correlated with enhanced perseverance ratios and increased serial strategy use in the reversal phase only.Reasons for these discrepancies could involve the age in which we conducted this task (ArcKR mice were tested at the juvenile stage, whereas Rnf216 KO mice were tested in early and late adulthood), E3 ligase compensation for Arc ubiquitination, additional requirements of RNF216-dependent Arc ubiquitination sites, or involvement of other RNF216 substrates.When comparing strategies during aging, middle-aged WT mice exhibited an increase in random over serial strategy search.However, Rnf216 KO mice exhibited serial strategy correlation profiles that were more similar to adult WT mice, suggesting that KO mice have delays in age-appropriate use of the random search strategy.These subtle differences in strategy use may be reflective of age-related changes in the output from the hippocampal system to associative areas such as the prefrontal cortex, which is selectively active during spatial navigation strategy and responsible for goal location and route planning (Poucet et al., 2004;Negrón-Oyarzo et al., 2018).The fear conditioning task in our study had two components: context-and cue-dependent recall, both of which have overlapping and nonoverlapping microcircuits underlying each type of recall.Although there were no differences seen during cue recall, middle-aged Rnf216 KO mice displayed decreased freezing time only during context recall.The underlying circuitry of fear conditioning includes projections from ventral hippocampal CA1 to the basal amygdala that encode context-dependent recall (Kim and Cho, 2020).However, during cue recall, some studies show that an auditory cue excites PV + interneurons that indirectly disinhibit the basolateral amygdala (Wolff et al., 2014), while other studies posit that auditory stimuli evoke the pathway from the thalamus to the lateral amygdala (Bordi and LeDoux, 1994).From this, we gather that RNF216 may not directly impair whole brain regions but rather influence transmission between distinct brain regions.This is consistent with GHS clinical cases that show neuropathologies with loss of neurons and atrophy in specific brain regions and hippocampal inclusion bodies, indicative of neurodegeneration.Because there are multiple brain regions affected in GHS individuals, it is possible that the connections between them become dysfunctional.
The unique feature of this study compared with the clinical literature that identified specific RNF216 mutations is that we used a constitutive KO model.Not all GHS patients have deletions, but rather point mutations that create dysfunctional RNF216 proteins (Margolin et al., 2013), which may account for the presence of a motor phenotype.However, male and female individuals with RNF216 mutations that were predicted to be deleterious still exhibit ataxia, so using Rnf216 KO mice is deemed an appropriate disease model.As an example, individuals with an E205fsX15+ C59X mutation or P606L mutation demonstrate severe limb, gait, appendicular, and truncal cerebellar ataxia with prominent cerebellar atrophy (Mehmood et al., 2017;Calandra et al., 2019).Homozygous variants predicted to be deleterious such as c.2251C>T and c.2149C→T found within the catalytic domain of RNF216 still display progressive ataxia with cerebellar atrophy (Margolin et al., 2013) indicating regardless of the type of mutation, ataxia is still present.Along with progressive ataxia, some individuals with heterozygous mutations of G13GfxX74 and C597X+E205DfsX15 that lie outside the catalytic domain also display chorea which involves involuntary jerking movements, distinct from ataxia.The human RNF216 gene differs from mouse in that RNF216 has five predicted isoforms (TRIAD3A, TRIAD3B, TRIAD3C, TRIAD3D, and TRIAD3E) that are invariably spliced at the N terminus (Chuang and Ulevitch, 2004), while mouse Rnf216 has only three predicted isoforms (TRIAD3A, TRIAD3B, and TRIAD3C; Consortium, 2020).It is possible that although both humans and mice express TRIAD3A, TRIAD3B, and TRIAD3C, the additional human isoforms such as TRIAD3D/E (that are also differentially expressed in the brain) have a more dominant role in specific pathologies such as ataxia (Chuang and Ulevitch, 2004).Our Rnf216 KO mouse is predicted to remove all three mouse isoforms of RNF216, but it is possible that these mice still express truncated versions of different isoforms.Our inability to detect any measure of brain volume loss in this study and in previous work (George et al., 2022) may point to additional distinct differences in how RNF216 is utilized in rodents compared with humans.This could in part be resolved by studying human models that include patientderived iPS cells.Another possibility is that compound mutations in other genes in addition to RNF216 may contribute to a more severe phenotype.Due to identification of digenic recessive mutations in GHS individuals that contain OTUD4 and RNF216 (Margolin et al., 2013), other unidentified genes may be participating in this complex syndrome.Environmental factors may also drive some of the severe phenotypes in human patients, which could be further pursued in our mouse model.
Another layer of data uncovered in this study is that electrophysiological and microglia alterations in constitutive male and female Rnf216 KO mice were found to be region and sex specific.Functionally, microglia are involved in synaptic maturation through synaptic engulfment during postnatal development and are thought to be important for neuroprotection as a synaptic "stripper" (Trapp et al., 2007;Paolicelli et al., 2011;Kettenmann et al., 2013).It is known that microglia profiles differ between males and females (Schwarz et al., 2012;Bordt et al., 2020).For example, as early as P0, sex differences in microglia morphologies are observed in the CA1 hippocampus and the paraventricular nucleus of the hypothalamus (Schwarz et al., 2012).The pro-and anti-inflammatory cytokine profiles also differ between sex and region upon aging (Schwarz et al., 2012).Although males had lower RMPs and reduced microglia soma size in the hippocampus, females had no change in RMP but an increased density of microglia.The sex differences we have reported with microglia in the hippocampus are concordant to those reported in young adult mice (Schwarz et al., 2012;Bordt et al., 2020).Upon examination of microglia branching using Sholl, there were no differences in the number of intersections between WT and KO males.However, KO females had an increase in the number of intersections around the soma compared with WT females.It is possible that the select microglia alterations in male mice might lead to reduced RMP in this region with the alternative interpretation that these are unrelated observations.Microglia express a myriad of receptors that are classically involved in neurotransmitter and cytokine signaling (Kettenmann et al., 2011) and also act as active sensors in the CNS for "on" and "off" receptor-mediated signaling (Hanisch and Kettenmann, 2007).Therefore, it is possible that the hyperpolarized state observed in early male hippocampal neurons disrupt microglia receptor-mediated signaling, preventing their appropriate function throughout development.Compared with the hippocampus, adult male KOs displayed altered microglia branching (decreased intersections via Sholl analysis) and decreased soma size in the cortex.The collective alterations in microglia in these different brain regions reflect their diversity in function, namely, in the remodeling of neuronal circuits which may become more pronounced with age (Hanisch and Kettenmann, 2007).Cumulatively, these data indicate that removal of Rnf216 differentially alters microglia states in male and female KO mice.
One account for microglia differences could also be related to alterations in gonadotropin and steroid hormone release.Gonadal hormones are thought to play a role in microglia sex differences (VanRyzin et al., 2018).Although we did not measure estradiol directly in these mice, previous work did not identify ovarian failure in KO female mice (George et al., 2022).However, our previous studies also demonstrated that male KO mice selectively have elevated baseline FSH levels and reductions in inhibin B compared with WT with signs of increased neuroinflammation as evidenced by alterations in microglia area and increases in the pro-inflammatory cytokine IL-1β (George et al., 2022).This is consistent with our current work showing males, in contrast to female KO mice, display reduced microglia soma size and branching.Importantly, testosterone gives rise to sexual differentiation in the vertebrate nervous system (Morris et al., 2004), and estradiol, which is aromatized from testosterone, has effects on hippocampal morphology (Lephart, 1996).Accordingly, there may be a lack of organizational effects in KO males during the perinatal or puberty time point leading to gonadal failure and altered hormonal release.This may also explain male-specific deficits in RMP.It is important to note that the electrophysiological and microglia-related phenotypes precede any of the learning deficits that we have observed.These early changes may alter neural circuit development and changes in microglia status, which collectively may render KO mice susceptible to behavioral alterations upon aging.
Taken together, our findings suggest that RNF216 may have region-and sex-specific effects on anatomical and behavioral outputs.These alterations may also be present in male and female GHS individuals, which have yet to be evaluated in the clinical setting.Our findings of microglia alterations also suggest that a neuroinflammatory environment supersedes any display of neurodegeneration or behavioral impairments.Currently, there are limited treatment options for GHS whose pathologies worsen over time.Our study points to possible targets for therapeutic intervention specific to activated or reactive microglia.Identifying mechanisms and substrates involved in RNF216 ubiquitination would help elucidate the cause of neuroinflammation and different aspects of behaviors.Likewise, restoring behavioral impairments in Rnf216 KO with these targets would help shed light on interventions in other disorders with similar dysfunctions.
significant differences.N = 11-25 mice per genotype.Cerebellar weights were normalized to brain weights.Error bars are ± SEM.

Figure 3 .
Figure 3. Lack of spatial learning deficits in adult Rnf216 KO mice.A, Schematic of Barnes maze task to evaluate spatial and reversal learning.Learning phase consisted of training (days 1-5) followed by assessments of learning (days 6-10).During the reversal phase (days 11-16), the exit hole was rotated 180°.B, Adult (∼P70) WT and KO mice show no difference in total distance traveled on the maze before finding the exit hole.C, No difference in the number of errors during the training, learning, or reversal phase.D, Left, No differences in quadrant bias ratio.Right, no difference in perseverance ratio in KO mice.E, Top, No differences in spatial strategy during each day across phases.Bottom, No differences in spatial strategy consolidated for each phase.F, Top, No differences in serial strategy during each day across phases.Bottom, No differences in serial strategy consolidated for each phase.G, Top, No differences in random strategy during each day across phases.Bottom, No differences in random strategy consolidated for each phase.N = 20 for WT and N = 21 for KO.Error bars are represented as ±SEM on top graphs in E-G and as box and whisker plots on bottom graphs.

Figure 5 .
Figure 5. Decreased context recall in middle-aged Rnf216 KO mice.A, No differences in time spent freezing during conditioning on day 1 in adult WT and KO mice or average number of freezing episodes (bottom left) or length of bout (bottom right).B, No differences in time spent freezing during context recall on day 2 in adult WT and KO mice or average number of freezing episodes (bottom left) or length of bout (bottom right).C, No differences in time spent freezing during cue recall on day 3 in adult WT and KO mice or average number of freezing episodes (bottom left) or length of bout (bottom right).N = 10-11 mice per sex/genotype.D, No differences in time spent freezing during conditioning on day 1 in aged WT and KO mice or average number of freezing episodes (bottom left) or length of bout (bottom right).E, Middle-aged KO mice spent less time freezing during context recall on day 2. F Time(4.038,121.1)= 12.15, ****p < 0.0001; F Genotype(1,30) = 5.380, *p = 0.0274; F Time*Genotype(5,150) = 1.089, p = 0.3690.Two-way ANOVA with Sidak's multiple comparisons.There were no differences on average number of freezing episodes (bottom left) or length of bout (bottom right).F, No differences in time spent freezing during cue recall on day 3 in middle-aged WT and KO mice or average number of freezing episodes (bottom left) or length of bout (bottom right).N = 7-9 mice per genotype/sex.Top, Error bars are represented as ±SEM.Bottom, Data are represented as box and whisker plots.

Figure 6 .
Figure 6.Middle-aged Rnf216 KO mice exhibit search strategies that are more tuned to adult WT mice.A, Pearson's correlation in adult WT (top) and KO (bottom) mice using behavioral output parameter-related strategy search in the Barnes maze.B, Pearson's correlation in middle-aged WT (top) and KO (bottom) mice using behavioral output parameters related to open field, strategy search in the Barnes maze.*p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.00005.

Table 1 .
Cerebellar weights in 16 and >52 week mice