Clinicopathologic Dissociation: Robust Lafora Body Accumulation in Malin KO Mice Without Observable Changes in Home-cage Behavior

Abstract

Lafora Disease (LD) is a syndrome of progressive myoclonic epilepsy and cumulative neurocognitive deterioration caused by recessively inherited genetic lesions of EPM2A (laforin) or NHLRC1 (malin). Neuropsychiatric symptomatology in LD is thought to be directly downstream of neuronal and astrocytic polyglucosan aggregates, termed Lafora bodies (LBs), which faithfully accumulate in an age-dependent manner in all mouse models of LD. In this study, we applied home-cage monitoring to examine the extent of neurobehavioral deterioration in a model of malin-deficient LD, as a means to identify robust preclinical endpoints that may guide the selection of novel genetic treatments. At 6 weeks, ~6–7 months and ~12 months of age, malin deficient mice (“KO”) and wild type (WT) littermates underwent a standardized home-cage behavioral assessment designed to non-obtrusively appraise features of rest/arousal, consumptive behaviors, risk aversion and voluntary wheel-running. At all timepoints, and over a range of metrics that we report transparently, WT and KO mice were essentially indistinguishable. In contrast, within WT mice compared across the same timepoints, we identified age-related nocturnal hypoactivity, diminished sucrose preference and reduced wheel-running. Neuropathological examinations in subsets of the same mice revealed expected age dependent LB accumulation, gliosis and microglial activation in cortical and subcortical brain regions. At 12 months of age, despite the burden of neocortical LBs, we did not identify spontaneous seizures during an electroencephalographic (EEG) survey, and KO and WT mice exhibited similar spectral EEG features. However, in an in vitro assay of neocortical function, paroxysmal bursts of network activity (UP states) in KO slices were more prolonged at 3 and 6 months of age, but similar to WT at 12 months. KO mice displayed a distinct response to pentylenetetrazole, with a greater incidence of clonic seizures and a more pronounced post-ictal suppression of movement, feeding and drinking behavior. Together, these results highlight a clinicopathologic dissociation in a mouse model of LD, where the accrual of LBs may latently modify cortical circuit function and seizure threshold without clinically meaningful changes in home-cage behavior. Our findings allude to a delay between LB accumulation and neurobehavioral decline in Lafora Disease: one that may provide a window for treatment, and whose precise duration may be difficult to ascertain within the typical lifespan of a laboratory mouse.

Graphical Abstract

Introduction

Lafora disease (LD) is a recessively inherited disorder of glycogen metabolism that gives rise to a syndrome of adolescent-onset progressive neurocognitive decline and intractable myoclonic epilepsy^1–5. The seizures of LD classically begin in children of adolescent age groups and may be of focal (occipital) or generalized onset and may coincide with a decline in school performance ^2,6. Over the next several years, as myoclonic (and other) seizures become more frequent and drug-refractory, declines in motor and cognitive functions progressively impose significant limitations to essentially all activities of daily living^7,8. LD is pathologically defined by the buildup of Lafora bodies (LBs): intracellular periodic acid-Schiff (PAS)-positive inclusions of polyglucosan⁹ that are found in neurons and astrocytes, as well as muscle, liver and skin^1,10. LD is caused by mutations in either of two genes located on chromosome 6: EPM2A (encoding laforin, a carbohydrate binding dual-specificity phosphatase¹¹) or NHLRC1 (encoding malin, an E3-ubiquitin ligase¹²). Current working models of pathophysiology hypothesize that malin and laforin function as a quality control complex within growing glycogen polymers, preventing the generation of very long branches that promote precipitation^1,2,13,14. A deeper knowledge of these mechanisms remains an active area of research, focusing on a detailed understanding of the substrates of laforin and malin’s phosphatase and E3-ligase activities, respectively. In the absence of either malin or laforin, insoluble aggregates of glycogen are hypothesized to trigger a cascade of neuroinflammation, neuronal dysfunction and hyperexcitability. Despite our advanced genetic understanding of LD, today’s standard of care remains supportive, with death commonly occurring within 10 years of initial diagnosis, typically due to status epilepticus or aspiration pneumonia^2,15.

Multiple LD mouse models have been developed through targeted constitutive deletions of either EPM2A^9,16 and NHLRC1^17–20. Laforin- and malin knockout (KO) mice generated in this fashion faithfully display age- and tissue-dependent accumulation of LBs, astrogliosis and microglial activation^9,20,21, complementing their construct validity with robust pathological validity. However, evidence of obvious face/symptomatic validity remains considerably less robust, as neither malin nor laforin KO mice display progressive and/or disabling myoclonic seizures, obvious cumulative frailty, or reproductive/survival deficits²². In one study, approximately one-year old malin KO mice displayed mild open field hypoactivity, subtle rotarod abnormalities and poorer object recognition scores¹⁷, together taken to reflect motor and memory deterioration. With video-electroencephalopgraphy (EEG), both myoclonic jerks and spike/wave discharges were observed in KO mice, but these often occurred independently/asynchronously. KO mice displayed an increased sensitivity to the chemoconvulsant pentylenetetrazole (PTZ), resulting in more frequent and rapidly occurring myoclonic and generalized seizures^23,24. However, neither study compared KO mice to littermate controls^9,17,23. A contemporaneously (but independently) generated second line of malin KO mice was found to display open field hyperactivity (interpreted as “reduced anxiety”), together with enhanced long-term potentiation (LTP) and unchanged rates of operant learning²⁰. These KO mice displayed more pronounced and frequent clonic seizures following kainic acid injections but were not surveyed for spontaneous seizures. A third line of malin KO mice¹⁹ was not tested on the open field, but was found to display elevated rates of context-dependent freezing responses²⁵, potentially in accordance with enhanced LTP rates. Through video observations alone, these mice were found to have elevated rates of spontaneous myoclonus (~1.5 jerks/min, compared to 0.5 in WT littermates). And finally, an independent fourth line of KO mice^18,22 has been extensively utilized for biochemical studies^26–28 but has never been examined for neurobehavioral or epileptic abnormalities.

A wide range of exciting targeted genetic^29,30 or biochemical^22,31 treatment strategies for LD are on the horizon. To rapidly test and improve upon these therapeutics in the preclinical domain, detailed histopathological endpoints in LD genetic models could ideally be complemented by rigorously obtained endpoints of neurobehavioral decline. To this end, we sought to identify robust evidence for progressive neurobehavioral deterioration in malin KO mice¹⁹ using home-cage monitoring (HCM), a technique that continues to gain popularity in behavioral neuroscience. As an alternative to traditional “out of cage” assays (e.g., open field/elevated plus maze tests), HCM platforms emphasize the automated and experimenter-free collection of prolonged behavioral recordings that encompass the nocturnal period³², obtained in an enclosure adopted by the mouse as its home-cage. Advances in videotracking and home-cage instrumentation now permit the simultaneous assessment of multiple streams of scalar variables (e.g., feeding and wheel-running and licking, etc), allowing investigators to appraise spontaneous behavior across a variety of time scales^33,34. Programmed provocative maneuvers/stressors, applied within the home-cage, further mitigate the observer effects associated with human exposure^35,36. HCM has been applied in this manner to uncover phenotypes in mouse models of pervasive neurodevelopment^37–40, aging⁴¹ and neurodegeneration^42–44. In this study, we conducted a systematic home-cage behavioral assessment of malin KO mice and wildtype littermates at 6 weeks, ~6 months and 1 year of age, in search of objective behavioral endpoints that may correlate with LB accumulation. At the final timepoint, we surveyed WT and KO mice for spontaneous seizures, and applied home-cage monitoring to examine the acute and subacute behavioral responses to PTZ. At each stage, we qualitatively assessed markers of microglial activation, astrogliosis and LB accumulation, and quantitatively appraised neocortical circuit dysfunction through in vitro recordings of bursting behavior⁴⁵.

Methods

Mice.

All protocols were approved by the UTSouthwestern Medical Center (UTSW) and Baylor College of Medicine (BCM) Institutional Animal Care and Use Committees and conducted in accordance with USPHS Policy on Humane Care and Use of Laboratory Animals. 7 wildtype (WT) and 9 KO¹⁹ were shipped from UTSW to BCM, from which heterozygous breeding pairs were generated (perpetuating their existing genetic background). PCR-based genotyping was performed on tail DNA at ~p16, and mice were weaned into gender-matched cages at p21. No mice were excluded.

Home-cage Monitoring:

Cohorts of age-matched WT and KO mice were transferred from the vivarium to Noldus Phenotyper home-cages (30×30×47cm) within a designated satellite study area^37,38. The “6 week” cohort were 6.08 ± 0.14 weeks of age [mean ± standard deviation]), “6 months” were 28.77 ± 1.95 weeks of age (delayed due to COVID restrictions on satellite use) and mice in the “1 year” cohort were 50.75 ± 1.3 weeks of age. Sample sizes for each genotype are provided within figures, and a total of 13 WT mice and 16 KO mice were examined serially at all three timepoints (Figure 5). 16 home-cages were employed in groups of four (“quad units”). Each cage contained (i) two lickometered water sources (0.8% sucrose-drinking water Vs drinking water), (ii) an infrared (IR)-lucent shelter, an aerial IR camera and IR bulb arrays, (iii) a beam-break device to measure entries into a food hopper, (iv) a detachable running wheel, and (v) a 2300Hz pure tone generator and an LED house light. Satellite temperature (20–26C), humidity (40–70%) and light cycle settings (ON between 0500–1700) matched vivarium conditions. White noise was played continuously, and satellite access was restricted to gowned, gloved, face-masked and capped personnel to minimize olfactory variations. Mice were distributed randomly to home-cages ensuring that one gender or one genotype was not over-represented within a single quad unit. When conducting within-cage daytime tasks described (e.g., positioning the running wheel), operators were blinded to genotype.

Data Acquisition and Modular Design of Recordings:

Live videotracking (Noldus Ethovision XT14) sampled the arena-calibrated x-y coordinates of the object centerpoint at 15Hz, providing rich location time series data, enabling heat maps and measures of sheltering, horizontal displacement and “sleep”. “Sleep” (enquoted to emphasize the noninvasive assessment) was defined as a period of continuous immobility lasting ≥40s, previously validated to provide >90% agreement with neurophysiologically-determined sleep^46–49. Lickometer-registered epochs of fluid consumption were registered as lick bouts/occurrences and durations. Similarly, both feeding entries/occurrences and feeding entry durations were tallied. A modular design^37,39 was applied (Figure 2b), beginning with a 2h-long initial habituation study (“Intro”) followed by two consecutive 23h long baseline recordings (1500–1400). Then, we conducted a five-hour long (1800–2300) evening recording termed “Light Spot and Beep”, during which a bright LED light is illuminated between 1900 and 2000 (the “light spot test”^37,38,50). An hour later, a single 60s long 2300Hz monotone was presented (“beep”^37,39). We concluded with a single daytime 2-hour long “cage swap”^38,39 provocation, manually swapping each mouse into a different home-cage previously inhabited by a conspecific of the same sex.

Pentylenetetrazole (PTZ) Injections:

A separate cohort of 12–13 month old WT and KO mice were acclimated to home-cage chambers for 24h, following which they all received intraperitoneal injections of PTZ (Sigma-Aldrich, 30mg/kg) within the home-cage at ~1200h, as previously described³⁸. We collected a 3h long “ictal” recording beginning immediately following injection, followed by a more prolonged “post-ictal” recording (from 1600 to 1100 the following day). Through the integrated visualization of video and high resolution mobility data³⁸, the first 20 minutes of the “ictal” recording was scored manually for the occurrence and latency of clinically evident seizure semiologies as classically described⁵¹, including features of phase 2 (partial clonus), phase 3 (generalized clonus with sudden loss of upright posture) and phase 4 (tonic-clonic maximal seizures, with or without hindlimb extension). Examples are provided in Supplemental Movies 1–3.

Electroencephalography:

To survey for spontaneous seizures, ~12–13 month old WT and KO mice were implanted with EMKA easyTEL S-ETA devices under sterile precautions and isoflurane anesthesia³⁹. Biopotential leads (2) were affixed epidurally in right frontal and left posterior parietal regions using dental cement, with wires tunneled to a transponder positioned in the subject’s left flank. Wireless EEG was acquired at 1000Hz sampling rate with IOX2 software (EMKA Technologies) via easyTEL receiver plates placed underneath home-cages, and EEG signals were inspected with LabChart reader using a bandpass filter (1–30Hz). Spectral analysis of unfiltered EEG (EEG ToolKit, Matlab) was conducted on randomly chosen 10-minute segments of wakefulness, collected between 1700 and 2000. Recordings were analyzed for power between 2 and 200 Hz, providing absolute power (AP) in dB (log10 (μV²/Hz)) at 1 Hz intervals. Relative power (RP) was calculated by dividing the absolute power for each frequency by the total power (TP), and then normalized with a log transformation before comparison between mice (RP=AP/TP), similar to methods described previously^52,53.

Histology.

For histological assessments, mice were rendered comatose with a single intraperitoneal injection of 0.1ml Beuthanasia-D (phenytoin 50mg/ml/pentobarbital 390mg/l), and subsequently transcardially perfused with ice cold phosphate buffered saline (PBS) followed 10% neutral buffered formalin (NBF). Mice that had been exposed to PTZ or implanted with EEG devices were not utilized for histology. Brains were extracted, post-fixed for another 24 hours in NBF and then stored in a solution of 70% ethanol. Paraffin-embedded brain tissues were sectioned and stained using standard histological technique including diastase-digested periodic acid-Schiff (PASD) staining to visualize polyglucosan bodies²⁵. Stained slides were scanned using a Hamamatsu Nanozoomer 2.0 HT digital slide scanner (40 x objective). For co-immunostaining, paraffin-embedded brain sections were deparaffinized and rehydrated by processing with xylene, decreasing concentrations of ethanol in water, and subjected to antigen retrieval using citrate buffer pH6.0 (C9999, Sigma-Aldrich). Sections were blocked with 5% normal donkey serum (in 0.1% Triton X-100, PBS) for 1 h and incubated for 48 h at 4°C with primary antibodies diluted in blocking solution, including those targeted against glycogen synthase 1 (Gys1, rabbit, 1:400, ab40810, Abcam), glial fibrillary acidic protein (GFAP, mouse, 1:500, BD556330, BD bioscience) and ionized calcium binding adaptor protein 1 (IBA1, goat, 1:350, ab5076, Abcam). Sections were then washed with PBS and incubated at room temperature for 2 h with secondary antibodies (ThermoFisher Scientific) diluted in blocking buffer: Alexa Fluor 488 donkey anti-mouse IgG (H+L) (1:500, Invitrogen A-21202), Alexa Fluor 488 donkey anti-goat IgG (H+L) (1:500, Invitrogen A-11055) and Alexa Fluor 594 donkey anti-rabbit IgG (H+L) (1:500, Invitrogen A-21207). After incubation with DAPI, sections were mounted using Aqua-Poly/Mount (Polysciences, Inc., US). Images of the different brain regions were taken on Zeiss LSM880 Airyscan confocal microscope at 40x magnification (zoom factor 0.6) with an average of 19 sections taken with a z-stack of 0.45 μm.

Electrophysiology.

Cortical slices were prepared from 3–12 month old male and female mice as previously described⁴⁵ with some modifications. Mice were deeply anesthetized with a solution of xylazine (20mg/kg) and ketamine (150mg/kg) and transcardially perfused with ice-cold dissection buffer (in mM: 87 NaCl, 3 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 7 MgCl₂, 0.5 CaCl₂, 20 d-glucose, 75 sucrose, 1.3 ascorbic acid) and decapitated. Brains were transferred into ice-cold dissection buffer aerated with 95% O₂-5% CO₂. 400 μm-thick thalamocortical slices were cut on an angled block⁵⁴ using a vibratome Leica VT 1200S, and transferred to an interface recording chamber (Harvard Instruments) and allowed to recover for 1 h in nominal “low activity” artificial cerebrospinal fluid (ACSF) at 32°C containing (in mm): 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 2 MgCl₂, 2 CaCl₂, and 25 d-glucose. Slices were then perfused with a “high activity” ACSF which contained (in mm): 126 NaCl, 5 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 1 MgCl₂, 1 CaCl₂, and 25 d-glucose. Slices remained in “high activity” ACSF for 30min and for recordings. Slices from 6- and 12-month old mice were exposed to Mg-deficient ACSF containing (in mM): 126 NaCl, 7 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, and 25 d-glucose for 30 min before and during recordings. Spontaneous extracellular multiunit recordings were performed using 0.5 MΩ tungsten microelectrodes (FHC) placed in layer 4 of primary somatosensory cortex (5 minutes per slice). Recordings were amplified 10,000-fold, sampled at 2.5 kHz, and filtered on-line between 500 Hz and 3 kHz. All measurements were analyzed off-line using custom Labview software. For visualization and analysis of neuronal activity bursts, traces were offset to zero, rectified, and low-pass filtered with a 0.2 Hz cutoff frequency. Using these processed traces, the threshold for detection was set at 15-times the root mean square noise. A burst was defined if the amplitude of the signal remained above the threshold for at least 50 ms. The end of the burst was determined when the amplitude decreased below threshold for >600 ms. Two bursts occurring within 600 ms of one another were grouped as a single event. Event amplitude was defined based on the filtered/rectified traces. For power analysis calculated for a single recording, the same offset, rectification, and low-pass filtering were performed as described for event detection. The power spectrum of this processed signal was performed over the entire 30 seconds of each trace, and then averaged over all traces. The following frequency bands were examined: low delta (0.2–1 Hz), delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz), low gamma (30–59 Hz) and high gamma (61–100 Hz). Power in each frequency band was normalized to total power (0.1–100 Hz).

Statistics:

Data were graphed and analyzed with Prism Graphpad 9, depicting mean ± standard error of the mean. For all home-cage behavioral phenotypes, we combined data from both sexes as we did not observe sex-dependent changes between WT and KO mice. Final sample sizes by sex are as follows: 6 weeks (WT:9M-8F, KO:10M-9F), 6 months (WT:10M-12F, KO:16M-12F) and 1 year (WT:8M-11F, KO:12M-15F), where M/F depict male/female sex. Lomb-scargle periodograms (Matlab) were applied to calculate the power and peaks of ultradian oscillations in activity. Behavioral endpoints were compared using two-tailed, unpaired student’s T (two groups), one-way ANOVA (three groups, analyzed post-hoc with Tukey’s multiple comparisons test) and restriction maximum likelihood mixed models when assessing for genotype x time interactions. UP-state measures (Fig. 6) were compared using the Mann-Whitney U test⁴⁵. *, **, ***, **** depict p<0.05, <0.01, <0.001 or <0.0001 respectively.

Results

To confirm the presence of age-related LB accumulation and CNS neuroinflammation, we examined WT and KO brain samples at ~6 weeks, ~6–7 months and ~1 year of age through PAS-D staining and immunohistochemical assessments of glycogen synthase 1 (GYS1, which accumulates within LBs), GFAP and IBA1 expression. LBs, GFAP and IBA1 induction were prominently seen in several brain regions, including the hippocampus, piriform cortex, striatum, cerebellum and the cochlear nucleus (Figure 1). These qualitative changes align well with our previous quantitative measurements in the same malin KO line⁵⁵.

Figure 1. Histopathological Changes in Malin KO Mice.

(a) Between 6 weeks and 1 year of age, Malin KO mice display a progressive increase in the burden of Lafora bodies (LB, which accumulate GYS1 or glycogen synthase-1) and astrogliosis (GFAP staining). (b) Microglial activation (assessed through IBA1 staining). (c) At 1 year of age, these changes are prominently observed in the hippocampal dentate gyrus (left) and the cerebellar granule cell layer (right). Top panels show PAS-D staining patterns. Scale bar: 50 μm. Images are representative of 2–6 mice per genotype, per age.

To obtain prolonged, automated and experimenter-free behavioral recordings, we employed 30×30cm instrumented home-cages^37–39 capable of measuring mouse movement and location (through videotracking), as well as drinking and feeding behavior through lickometers and feeding meters, respectively. We chose a commercially available home-cage platform that has been extensively applied previously to study within- and across-strain differences in mouse behavior^56,57, age-related changes⁴¹ and models of neuropsychiatric illness^32,40,58. Compared with these previous efforts, we adopted a configuration that includes an infrared-lucent shelter, two waterspouts (measuring sucrose water Vs water intake) and reserved the final ‘wall’ to attach/detach a running wheel (Figure 2a). To appreciate a wide range of neurobehavioral endpoints over a several day-long recording period, all mice were channeled through an identically ordered set of home-cage recordings in a modular design (Fig. 2b).

Figure 2. Behavioral Responses to a Novel Home-cage.

(a) Cartoon schematic of home-cage, featuring two lickometered water spouts (water vs 0.8% sucrose-water), a food hopper and an infrared-lucent shelter. RIGHT: representative aerial view from infrared camera with centerpoint tracking. (b) Schematic of the modular design applied for all home-cage assessments, designed to capture a mouse’s initial response to cage novelty (“Intro”), patterns of unperturbed spontaneous behavior (“baseline” recordings), and responses to various provocative maneuvers (see Figure 4). (c) 1-year-old WT and KO mice were similar in body weight (p=0.2), and featured similar patterns of cage exploration. Heatmaps (left) and trackmaps (right) are shown for a representative WT and KO mouse. TOP: Raster plot of distances traversed every minute of trial for every mouse. BOTTOM: Total measures of horizontal displacement (p=0.2), sheltering (p=0.7), water licks (p=0.9), sucrose licks (p=0.8) and feeding entries (p=0.6) were not significantly different between WT and KO mice. (d) Body weights (p=0.8), distances and sheltering curves for 6-week-old mice. (e) Body weights (p=0.9), distances and sheltering curves for 6-month-old mice. Mean ± s.e.m shown for all.

At 1 year of age, KO and WT littermate mice were of similar body weight and responded similarly during an initial 2-hour long introduction to home-cage chambers (“Intro”, Figure 2c). This module approximates a more conventional open field test, but utilizes a significantly longer recording period in an arena that is more enriched than a typical open field, with bedding, a shelter and two water sources. We observed no significant differences in kinematic measures of home-cage exploration or objective measures of spout/feeder engagement (measured through lickometers and infrared beam-breaks, respectively). Even within this initial relatively brief epoch, both genotypes displayed a clear preference for sucrose containing fluid (Figure 2c). Similar results were obtained in younger cohorts measured at 6 weeks of age (Figure 2d) and 6 months (Figure 2e).

Next, we conducted two prolonged 23-hour-long “baseline” recordings beginning at 1500h and presents results from the second baseline day. WT and KO showed comparable patterns of nocturnality, total and hourly horizontal displacement, a similar mid-active phase dip in movement and nearly identical ultradian rhythms of activity (Figure 3a). The timing and total amounts of “sleep” and “sleep bouts” were also equivalent (Figure 3b). Over the entire trial, averaged time budgets^33,37–39 allotted to sheltering, feeding and licking behavior were similar (Figure 3c), and both groups of mice demonstrated a similar micro/macrostructure of licking/feeding bouts (Figure 3d,e). Baseline behavior was also similar between WT and KO mice at 6 weeks (Figure 3f) and 6 months of age (Figure 3g). Together, these data illustrate that the constitutive loss of malin (and associated accumulations of LBs) do not significantly alter the spatiotemporal structure of spontaneous behavior within a task-less home-cage environment.

Figure 3. Baseline Recordings.

(a) On the second 23h-long baseline recording, 1-year-old WT and KO mice displayed similar patterns of hourly horizontal distances (hour x genotype, F_(22,968) = 0.7, p=0.7), total distances (p=0.3) and ultradian oscillations in activity. Raster plot depicts distances moved every minute of the day. (b) 1-year-old WT and KO mice were indistinguishable in measures of total daily “sleep” (p=0.9) and the number of “sleep” bouts per day (p=0.4), without a change in the distributions of “sleep” bout duration and timing. (c) Averaged time budgets for WT and KO mice. (d) WT and KO mice displayed similar sucrose preference (p=0.3), total water licks (p=0.8) and sucrose licks (p=0.5). The overall macrostructure of licking was preserved, with WT and KO mice displaying an identical distribution of lick durations and interlick bouts. (e) At the same age, hourly changes in feeding entries were comparable (hour x genotype, F_(22,968) = 0.7, p=0.8), as were total daily feeding durations (p=0.5). (f) Raster plots of horizontal activity in WT and KO mice at 6 weeks of age, time budgets and total daily distances (p=0.9), feeding durations (p=0.2), “sleep” (p=0.4), sucrose preference (p=0.4) and total daily licks (p=0.5). (g)) Raster plots of horizontal activity in WT and KO mice at 6 months of age, time budgets and total daily distances (p=0.2), feeding durations (p=0.5), “sleep” (p=0.3), sucrose preference (p=0.4) and total daily licks (p=0.1). Mean ± s.e.m shown for all. See Fig.1A for sample sizes. Room lights are off between 1700 and 0500.

We then applied a set of provocative maneuvers to reveal other latent phenotypes. During a light spot test^37–39,50, a bright ceiling-mounted home-cage LED was illuminated for 60 minutes in the early nocturnal period (1900). 1-year-old WT and KO mice displayed a similar initial hyperlocomotor response followed by gradual shelter engagement (Figure 4a). In comparison, KO responses to a 60-s long auditory tone (“beep”) were relatively blunted, with a weakened startle response and less vigorous shelter engagement (Figure 4b). Light spot and beep responses for 6-week- and 6-month-old cohorts are shown Figure 4c,d. On the following day, we affixed detachable running wheels within home-cages. KO mice displayed a subtle reduction in wheel-running at 6 months of age, but not at 6 weeks or 1 year (Figure 4e). Finally, during a “swap” protocol³⁹ where every mouse is repositioned within a cage previously inhabited by a sex-matched mouse, WT and KO mice at all age groups displayed similar curves of novel cage acclimation in response to this geometrically similar (yet olfactorily distinct) environment (Figure 4f).

Figure 4. Home-cage provocative maneuvers.

(a) 1-year-old WT and KO mice displayed comparable locomotor suppression and shelter engagement in response to an hour-long light spot stimulation. (b) In response to a single 60s long monotone stimulus (“beep”), KO mice displayed a blunted response, featuring lower initial peak velocities (startle) and less emphatic shelter engagement. (c) Light spot and beep responses at 6 weeks of age. (d) Light spot and beep responses at 6 months of age. (e) When presented with a running wheel, 1-year-old WT and KO mice displayed patterns of running wheel engagement (hour x genotype, F_(22,950) = 1.02, p=0.4) and overall wheel running (p=0.5). Similar results were seen in 6-week old cohorts (hour x genotype, F_(22,700) = 0.5, p=0.9, total wheel rotations, p = 0.7). 6-month-old KO mice did display fewer wheel rotations overall (hour x genotype, F_(22,960) = 1.8, p<0.05). (f) In all age groups, WT and KO mice responded comparably to a 2h-long daytime cage-swap maneuver. Mean ± s.e.m shown for all, with * depicting p<0.05. Room lights are off between 1700 and 0500.

Since our cross-sectional comparisons of WT and KO mice at various time points did not reveal any objective evidence of progressive behavioral or motor deterioration, we next sought to explore whether our home-cage platform could identify any age-dependent changes in home-cage metrics in 6-week-, 6-month- and 1-year-old WT mice. We acknowledge two main caveats to such a comparison. First, WT mice of differing ages were not studied simultaneously (we emphasized synchronous assessments KO and their WT littermates). Second, to reduce⁴⁴ the number of animals employed, 13 WT mice were studied serially across all three timepoints. Compared to 6-week-old mice, those aged 6–7 months and older displayed a significant reduction in total feeding duration (but not feeding entries, Figure 5a). We observed a mild statistically significant reduction in sucrose preference in 1-year-olds Vs 6-week-old mice. Total horizontal distances were also considerably lower at this timepoint, replicating previous results^41,59 (~587m/d, compared with ~726m/d [6 weeks], 701 m/d [6 months]), although this effect was not statistically significant (p = 0.08, Figure 5a). Estimates of total sleep time, timing and sleep structure remained largely unchanged (Figure 5b). Compared with 6-week-old mice, older cohorts displayed more sustained shelter engagement during the light spot test (Figure 5c) and a less pronounced initial startle response to auditory stimulation (Figure 5d). An age-dependent stepwise decline in overall wheel-running interest was also observed (Figure 5e).

Figure 5. Age-dependent changes in home-cage behavior.

(a) On the second baseline day 2, hourly rates of total distances moved (hour x genotype, F_(44,1229) = 2.9, p<0.0001), sheltering (hour x genotype, F_(44,1229) = 2.8, p<0.0001) and feeding duration (hour x genotype, F_(44,1229) = 1.6, p<0.01). Body weights (F_(2,56) = 62.2, p<0.001), sucrose preference (F_(2,56) = 2.9, p<0.05) and feeding durations (F_(2,56) = 10.55, p<0.001) varied by age, while total daily distances (F_(2,56) = 2.5, p=0.08), sheltering (F_(2,56) = 0.4, p=0.6) and feeding entries (F_(2,56) = 0.18, p=0.3) were not significantly altered. (b) Averaged time budgets in WT mice across age groups with sample sizes shown. Total daily “sleep” (F_(2,56) = 0.93, p=0.4) and “sleep” bouts/day (F_(2,56) = 1.12, p=0.3) remained stable with age, without a change in the distributions of “sleep” bout duration and timing. (c) Responses to light spot stimulation across 6 weeks, 6 months and 1-year cohorts. (d) Responses to beep stimulation across 6 weeks, 6 months and 1-year cohorts. (e) Hourly rates of voluntary wheel running varied by age (F_(44,1120) = 4.5, p<0.0001), as did total wheel rotations (F_(2,56)=9.1, p<0.001). *, **, ***, **** refer to post-hoc comparisons and depict p<0.05, <0.01, <0.001 or <0.0001 respectively. Room lights are off between 1700 and 0500.

To survey for spontaneously occurring epileptic seizures at 1 year of age, a total of 4 WT and 7 KO mice underwent implantation of wireless EEG electrodes³⁹, with each mouse receiving at least 96 hours of single channel EEG recording (Figure 6a). Spontaneous seizures were not observed across either genotype. EEG spectral signatures (assessed during waking periods) were similar, featuring a peak in power at ~4Hz⁶⁰ (Figure 6b). To ensure that our EEG recording apparatus could detect ictal or interictal epileptiform discharges, a subset of WT and KO mice received intraperitoneal injections of PTZ at convulsant doses (60mg/kg). As shown in Figure 6c, PTZ-induced electrographic seizure activity was similar between WT and KO mice, beginning with quasiperiodic discharges (0.1–0.2Hz), progressing ultimately to an epoch of evolving rhythmicity, followed by post-ictal amplitude suppression.

Figure 6. EEG and PTZ Responses.

(a) Representative single-channel wireless electrocorticography from 1-year old WT and KO mice. (b) EEG power spectra calculated during wakefulness. (c) Representative EEG responses to a single intraperitoneal injection of PTZ (60mg/kg), demonstrating a prolonged epoch of spike/wave discharges, followed by a discrete epoch of evolving rhythmicity. Red bars annotate epochs of absent EEG signal while the mouse receives the intraperitoneal injection. (d) Distance and sheltering responses to a single subconvulsant PTZ injection (30mg/kg, provided at approximately 12 noon) measured within home-cages, with a manual tally of convulsive events (inset) for both WT (n=14) and KO (n = 18). (e) Home-cage metrics during the post-ictal period, defined as 1600 to 1100 the following day. WT and KO distances measured during the baseline trial are shown in the background. After PTZ injections, WT and KO mice displayed a similar decline in locomotor activity (F_(18,522) = 1.18, p=0.3). Total sheltering times were higher in KO mice, but this effect did not vary with time (F_(18,522) = 0.97, p=0.5). Malin KO mice also displayed fewer feeding entries in the post-ictal period (F_(18,522)= 1.95, p<0.05). Mean ± s.e.m shown for all. * depicts p<0.05.

We then took advantage of our home-cage monitoring platform to objectively compare WT and KO behavioral responses to PTZ. Using videotracking data, we have previously conducted a detailed quantification of the remarkable immobility that is observed following a single intraperitoneal injection of subconvulsant dose PTZ (30mg/kg, administered at ~1200). Since much of this early immobility occurs outside the shelter at a time of the day when mice are largely confined to their shelters³⁸, this response quantifies the severity of the acute [ictal] encephalopathy induced by PTZ. As shown in Figure 6d, changes in horizontal activity and shelter engagement were similar between genotypes. To tally the occurrences of clonic (phase 2, 3) and tonic-clonic seizures (phase 4)⁵¹, we manually examined video recordings and high-resolution actograms for each subject during the first 20 minutes following the PTZ injection (supplemental movie 1–3). KO mice displayed a greater incidence of generalized clonus and were the only genotype to display generalized tonic-clonic seizures. To recognize genotypic differences during a more extended post-ictal period, we profiled behavioral patterns over the remainder of the day. Between 1600–1100, KO mice displayed significantly greater shelter engagement and fewer feeding entries. While both genotypes displayed impressive post-ictal hypoactivity, this was similar between WT and KO mice (Figure 6e).

Finally, in a separate group of WT and KO mice (without prior PTZ or other experiences), we asked whether LB accumulation and associated neuroinflammatory changes were associated with in vitro neurophysiological evidence of cortical circuit dysfunction. To this end, we utilized field recordings in neocortical slices to record the expression of “UP states”^45,61,62 (Figure 7a), which are epochs of synchronous depolarization driven by local recurrent excitation and inhibition within all neurons in a cortical region⁶³. While UP states can be evoked with thalamic stimulation⁶⁴, spontaneously occurring UP states are thought to underlie neocortical slow oscillations during slow wave sleep⁶⁵. Spontaneous UP states are more prolonged in mice with deletions of Fmr1 (modeling fragile-X syndrome⁴⁵), while significantly shorter and less frequent UP states are seen in a mouse model of Down syndrome⁶⁶. We found that UP state durations were significantly larger in KO mice at 3 and 6 months of age, but we observed no differences in slices from 1-year-old mice (Figure 7b). The amplitude (but not frequency) of these UP state bursts progressively increased with age⁶² (Figure 7c), and in the oldest age group, bursts displayed a similar spectral composition (Figure 7d). Together, these results identify a potential neurophysiological correlate/consequence of LB accumulation and cortical circuit dysfunction in malin KO mice that may precede frank motor impairment and seizure risk.

Figure 7: UP States in WT vs KO Somatosensory Cortex.

A: Example traces of extracellular recordings in brain slices exhibiting spontaneously occurring activity bursts (from 12-month-old mice). B: The average duration of activity bursts in MKO slices is longer at 3 and 6 months of age, but not at 1 year (Mann-Whitney test, *p< 0.05) C: UP State burst amplitudes and frequencies. D: Relative power over all activity. Mean ± s.e.m shown for all. Sample sizes at 3 months: WT (9 slices, 4 mice), KO (13 slices, 5 mice). 6 months: WT (19 slices, 6 mice), KO (29 slices, 7 mice). 1 year: WT (12 slices, 6 mice), KO (15 slices, 5 mice).

Discussion

In this study, we examined one¹⁹ out of four^17,18,20 available mouse models of malin-deficient Lafora disease to ask whether brain LB accumulation and associated astrogliosis/microglial activation are associated with any robust alterations in home-cage behavior. The home-cage approach affords several benefits that improve rigor and reproducibility of behavioral phenotyping^32,44. This includes the automated acquisition of prolonged recordings in an experimenter-free setting, providing a particularly transparent window into the rich expressions of spontaneous behavior during the murine night. To avoid the potential observer effects associated with prolonged social isolation, we applied a modular design over a 4–5-day observation period designed to gauge behavioral wellbeing across multiple dimensions, including rest/arousal systems (distances, “sleep”), rhythms of consumptive behavior, responses to potential threats (e.g., light spot, swap) and reward responsiveness (e.g., wheel-running, sucrose preference). While abundant LBs were evident at 1 year of age (Fig. 1A, 7), KO mice were largely similar to WT mice across a range of scalar home-cage endpoints. As a comparison, we did identify important age-related changes in many of the same metrics within WT mice, including with older mice displaying lower sucrose preference, feeder engagement and wheel-running drive. These findings, in conjunction with previously published home-cage phenotypic distinctions across inbred strains of mice^56,57 and disease models^{32,37–39,41–43} studied using the same apparatus, argue against platform insensitivity as an explanation for our negative findings.

In a slice preparation, we did find evidence for LB-associated neocortical dysfunction at 3 and 6 months of age, where UP state bursts were significantly prolonged without changes in burst frequency or amplitude. This may reflect a unique window of time for follow up studies designed to test the effects of LB scavenging or other therapies on neocortical dysfunction. However, differences in burst duration were not identified at 12 months of age. The lack of differences in slices from 1-year-old mice may reflect age-dependent changes in the ability of cortical neurons to survive the hypoxic stress associated with slice preparations. In vivo EEG recordings did not identify significant changes in spectral composition between WT and KO mice at this time point. Nevertheless, in response to PTZ, KO mice displayed evidence of greater convulsive and behavioral seizure severity, reproducing earlier results^23,24. While PTZ seizure induction paradigms can be quite heterogeneous across laboratories⁶⁷, these data suggest that detailed analyses of PTZ responses may serve as a potential readout of cortical function.

We posit three possible non-mutually exclusive explanations for the clinicopathological dissociation that we uncover. First, laboratory mice may display a species-specific immunity to the neurobehavioral consequences of abundant LB accumulation. Second, clinically meaningful changes in neuronal dysfunction may substantially lag behind neuropathological abnormalities at a time scale (years?) that cannot be practically explored in laboratory mice. This explanation is compatible with LD in human and canine subjects^6,68,69 (for whom serial neuropathological assessments are impossible), where epilepsy and neurocognitive decline occur after years of seemingly normal brain development. Third, there remains the possibility that LB accumulation, while impressive, is not the proximate cause of neuronal dysfunction related to the loss of malin, whose functions beyond glycogen metabolism remain unknown. We regard this as an unlikely possibility, since recent work has shown that malin’s predominant (if not exclusive) subcellular localization is at glycogen, where it is tightly scaffolded to the carbohydrate binding domain of laforin¹⁴, the deletion of which produces an identical clinical syndrome. In a number of LD mouse models (including the one studied here), preventing LB formation through downregulation of glycogen synthesis or removing LB by digesting them with a CNS-delivered amylase, prevents or corrects the neuroinflammatory, neurometabolic and brain protein glycation defects that characterize the neuropathology of LD^{22,31,55,70–74}. However, the neurobehavioral correlates of these biochemical rescue strategies have not been explored in as great detail.

We identify three main limitations to this work. (i) Our neurobehavioral survey did not include any classical measures of learning and memory. As in humans, this is a multilayered construct in mice, with an array of available tests that are designed to assay fear memory (e.g., fear conditioning), spatial memory (e.g., Morris water maze), object memory (e.g., object recognition testing) or procedural memory (e.g., illuminated radial arm maze). Our results suggest that any learning/memory phenotypes, if present, cannot be explained by (or are not associated with) concurrent deficits in sleep, motor function, or grossly assayed visual or auditory function. (ii) Our insights could have been strengthened further by utilizing mice of a different genetic background or by comparing two or more mouse models of LD, to more definitively correlate LB accumulation with home-cage behavioral deficits. (iii) Extending our recordings to even older mice (e.g., 21–27 months of age⁴¹) may have revealed additional phenotypes. (iv) And finally, as with any several-day long EEG survey in mice, the absence of spontaneous seizures does not necessarily confirm seizure freedom, as KO mice may display extremely rare seizures that require longer EEG surveillance. Spontaneous seizure occurrence in mice is frequently associated with seemingly unexplained premature mortality⁷⁵, which was not seen in our KO mice.

In conclusion, given the urgent need for LD treatments, it remains reasonable to apply mouse models to screen for therapies that impart biochemical improvements in LB burden and related neuropathological changes. Our results find little evidence for robust changes in home-cage behavior in malin-deficient mice aged to approximately half their typical laboratory lifespan. Identifying holistic neurobehavioral endpoints to practically validate a pipeline of disease-modifying LD treatments may require us to innovate strategies that go beyond the laboratory mouse⁶⁹, and justifiably invest in protocols that involve prolonged trial durations (years) that more closely mirror the temporal progression of human LD.

Supplementary Material

Movie S1

Supplementary Movie 1: Representative video capturing two events scored as “phase 2” (4 and 8s into the recording), demonstrating clonic activity affecting the forelimbs.

Movie S2

Supplementary Movie 2: Representative video capturing a single “phase 2” followed by a “phase 3” event (7s into the recording), capturing generalized clonus manifesting as a sudden loss of upright posture.

Movie S3

Supplementary Movie 3: Representative video capturing a “phase 4” event (12s into the recording) featuring a maximal seizure without hindlimb extension.