Animals

Mcoln1 knock-out mice were maintained and genotyped as previously described [25]. The Mcoln1 ^+/− breeders for this study were obtained by backcrossing onto a C57Bl6N background for more than 10 generations. Mcoln1 ^+/+ littermates were used as controls. Experiments were performed according to the institutional and US National Institute of Health guidelines and approved by the Massachusetts General Hospital Institutional Animal Care and Use Committee.

Open field testing

Open field testing was performed on naive male mice at one (n = 13, wild type; n = 9, Mcoln1 ^−/− ) and two (n = 17, wild type; n = 18, Mcoln1 ^−/− ) months of age under regular light conditions. Each mouse was placed in the center of a 27 × 27 cm² Plexiglas arena, and the horizontal and vertical activity were recorded by the Activity Monitor program (Med Associates). Data were analyzed during the first 10 mins in the arena. Zone analysis was performed to measure movements/time spent in the central (8 × 8 cm²) versus peripheral (residual) zone of the arena. Data were analyzed by two-way ANOVA (genotype × age) followed by Bonferroni post-test.

Stereological analysis and immunohistochemistry

To obtain brain tissue for histological examination two (n = 6 per genotype), three (n = 3, WT; n = 4, Mcoln1 ^−/− ) and seven month-old Mcoln1 ^−/− and control mice (n = 4 per genotype) were transcardially perfused under isoflurane anesthesia with ice-cold phosphate buffered saline (PBS) followed by 4% paraformaldehyde in PBS. Brains were postfixed in 4% paraformaldehyde in PBS for 24 hours, washed with PBS, cryoprotected in 30% sucrose in PBS overnight, frozen in isopentane and stored at −80°C. Brains were bisected along the midline and one hemisphere was examined histologically. 40 μm coronal sections were cut using a Microm freezing microtome and collected into 96 well plates containing TBSAF (TBS, 30% ethylene glycol, 15% sucrose, 0.05% sodium azide). These sections were stored at 4°C prior to any staining procedures. For Nissl staining a one-in-six series of sections were mounted onto chrome gelatine coated slides, incubated in 0.05% cresyl violet stain and 0.05% acetic acid at 60°C for 30 mins before being dehydrated in ascending concentration of industrial methylated spirits (IMS) followed by xylene. The sections were then coverslipped using DPX mountant.

Quantitative histological measurements

Nissl stained sections were used to obtain cortical thickness measurements using StereoInvestigator software. All analyses were carried out using ×2.5 objective on a Zeiss, Axioskop2 MOT microscope (Carl Zeiss Ltd, Welwyn Garden City, UK). Briefly, within each region three consecutive sections were chosen and 10 perpendicular lines were drawn from the pial surface to the white matter.

Unbiased Cavalieri estimates of regional volume were obtained using StereoInvestigator software. Briefly, for each of the regions a sampling grid size of 150 μm was superimposed on every one-in-twelfth Nissl stained section (cortex and hippocampus) or every one-in-six Nissl stained section (thalamus and striatum) and the number of points that fell within this region was recorded to provide an unbiased Cavalieri estimate of regional volume (μm³) for each brain. All measurements were carried out at ×2.5 objective on the Olympus BX50 microscope (Olympus Microscopes, South-on-Sea, UK).

Unbiased optical fractionator estimates for the number of Nissl stained neurons within the VPL-VPM, laminae IV and V of the S1BF, DLG and red nucleus were obtained using the StereoInvestigator program. A grid size of 175 μm × 175 μm was used for the thalamus, 225 μm × 225 μm was used for laminae IV and V of the S1BF, and 125 μm × 125 μm was used for the DLG and red nucleus. A counting frame of 68 cm × 38 cm was used for all regions and all measurements were carried out at ×100 objective on a Zeiss, Axioskop2 MOT microscope (Carl Zeiss Ltd, Welwyn Garden City, UK). The mean coefficient of error (CE) for optical fractionator estimates of neuron number were all between 0.05 and 0.1. All analyses were performed blind to genotype, which was only revealed once these analyses were complete. At this point the mean value for each parameter was calculated for both Mcoln1- deficient and +/+mice. Differences between genotypes were compared statistically using Student's t-tests with a P-value of ≤0.05 considered as statistically significant.

Immunostaining

Immunostaining was performed as described [27]. The primary polyclonal rat anti-CD68 (Serotec, 1:2000) and polyclonal rabbit anti-GFAP (DAKO, 1:4000) antibodies were used. For myelin FluoroMyelin™ Green fluorescent stain (Molecular Probes) was used according to the manufacturer protocol.

Thresholding image analysis

Three consecutive GFAP- and CD68- immunostained sections were chosen for each of the regions being observed and 30 non-overlapping images were captured at 40× objective by a live video camera (JVC, 3CCD, KY-F55B) mounted on Zeiss Axioplan microscope (West Germany). The lamp intensity, video camera setup, and calibration were kept constant when capturing all the images. These images were then analysed on ImageProPlus software with an appropriate threshold selected to distinguish the foreground immunostaining above the background for each of the age groups being observed. The data obtained from the thresholding analysis was plotted graphically as a mean percentage area of immunoreactivity per image.

In vivo imaging of Ca2+ concentrations in cortical neurons

To deliver the genetically encoded calcium indicator, craniotomies were performed on Mcoln1 ^−/− (n = 5) and control mice (n = 3) of 2 months and 1 week of age. A 5-mm diameter skull flap centered over the right primary somatosensory cortex was removed and 2 μl of AAV2/8 viral vector (5.6 × 10¹² viral genomes/ml) encoding YC3.60 under hybrid cytomegalovirus (CMV) immediate-early enhancer/chicken β-actin promoter/exon1/intron [28] were injected into the open brain close to the center of the craniotomy at a depth of about 0.5 mm and at a speed of 0.2 μl/min. The brain surface (still covered by intact dura matter) was kept moist with Ringer's solution at all times. Following the virus injection, the craniotomy was sealed with a glass coverslip and cemented with dental acrylic. Appropriate anesthetic and analgesic regimes were followed before, during and after the surgery. 3 weeks after the surgery, when the mice reached 3 months of age, they were anesthetized (isoflurane in balanced oxygen: 4% for 5 min induction, then maintained at 1.2%), head fixed and imaged on a multiphoton microscope. Our imaging setup was described earlier [29]. The excitation laser was tuned to 860 nm and the output power before the objective was set to 30 mW. Emitted light was collected in three channels: 460 – 500 nm (cyan channel (C), CFP fluorescence), 530 – 560 nm (yellow channel (Y), YFP fluorescence) and 575 – 630 nm (red channel (R), autofluorescence from storage material). With these settings, autofluorescent storage material yielded equal signals in cyan, yellow and red channels. Z-stacks were acquired with the resolution of 0.5 μm/voxel in X-Y dimension and the Z-step of 3 μm for imaging of neuronal cell bodies in the cortical layer II/III (512 × 512 × 270 μm stack) and with the resolution of 0.25 × 0.25 × 2 μm/voxel for imaging of dendrites in the cortical layer I (512 × 512 × 60 μm stack). Imaging settings were kept constant across mice.

Image processing was performed using the Fiji package of NIH ImageJ software (fiji.sc; rsbweb.nih.gov/ij) and MATLAB (MathWorks). Ratio of YFP to CFP signals representative of intracellular calcium concentration was calculated after subtracting the signal from background-subtracted red channel:$R=\frac{\mathrm{Y}-\mathrm{R}}{\mathrm{C}-\mathrm{R}}$. Regions of interest (ROIs) outlining cell bodies and dendrites were selected on the raw images and applied to the ratio images and to the red channel to estimate the amount of autofluorescent storage. The values for individual cells and dendrites were calculated as mean ratio in ROIs. All statistical tests were performed on R values. To convert R values to [Ca²⁺] for data presentation and interpretation, the following formula was used:$\left[C{a}^{2+}\right]=K_d^{\prime }{\left(\frac{R-R_{\mathit{min}}}{R_{\mathit{max}}-R}\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n$}\right.}$. Values for$K_d^{-}$(277 nM) and the Hill coefficient n (1.1) were described previously [28]. R _min (0.67) and R _max (2.55) were measured experimentally. To create pseudocolored images, ratio images coded with "Rainbow RGB" lookup table were converted to the RGB space weighted by intensity.

The normality of datasets of YC3.60 ratio values was tested using Kolmogorov-Smirnov method. Since the distributions were found to be non-normal, they were compared between genotypes using the Wilcoxon rank-sum test with correction for clustering of values within individual mice [30]. The correlation of YC3.60 ratio values with the amount of cytosolic storage material was performed using a mixed effects model, with random slope effects for mouse, and using mouse-specific standardized transformations of YC3.60 ratio and autofluorescence values.

Electrophysiology

Transverse hippocampal slices were prepared from 7 month-old Mcoln1 ^−/− and control male mice (n = 6 per genotype) and field recordings were performed as described [31]. Data were normalized to the baseline response and are presented as group means ± SE. One-way ANOVA and Student's t-test were used to determine statistically significant differences. For all experiments the experimenter was blind to genotype.

Electron microscopy

Seven month-old Mcoln1 ^−/− and control mice (n = 4 per genotype) were anesthetized with isoflurane and transcardially perfused with PBS followed by 2% paraformaldehyde/2% glutaraldehyde in PBS, brains were removed and 50 μm coronal serial vibratome sections were processed for EM as previously described [32]. 35 micrographs per sample from stratum radiatum per mouse were obtained on a JEOL JEM-1011 transmission electron microscope at final magnification of ×25,000. Only asymmetric macular (non-perforated) synapses that were fully present on the micrograph were included in the analysis. Digital image analysis was performed using ImageJ (NIH). The post-synaptic density length was measured. Analysis of mitochondrion volume was performed as described [33]. For myelin sheath thickness measurements two perpendicular lines were drawn through the center of each myelinated axon and four measurements of the thickness were taken at each intersection with the myelin sheath with the mean value of four measurements calculated. The normality of all data sets was tested using the Kolmogorov-Smirnov method. Non-parametric data are presented as median values with interquartile ranges. An extended Wilcoxon rank-sum test that accommodates clustered data was used for comparisons between control and Mcoln1 ^−/− groups for non-normally distributed data. Statistical significance threshold was set at p < 0.05.

Early motor and cognitive deficits in Mcoln1−/−mice

To determine if Mcoln1 ^−/− mice display early locomotor and behavioral deficits, spontaneous activity of one- and two-month-old naïve Mcoln1 ^−/− mice and wild-type littermates was tested in the open field arena. We found a significant decline in both jumping and vertical activity and an increase in the resting time in Mcoln1 ^−/− mice at two months of age (Figure 1A-E), which indicates the early onset of motor deficits. Analysis of activity within the central zone revealed that the number of central zone entries, percentage of central track length and percentage of time spent in the center were significantly lower in Mcoln1 ^−/− mice compared to wild-type controls at 2 months of age (Figure 1F-H) indicating an early decline in the exploratory activity associated with loss of Mcoln1.

Absence of brain atrophy in Mcoln1−/−brain

Neuropathology in models of lysosomal storage disorders is typically characterized by atrophy and neuronal loss within specific brain regions. To assess the progression of anatomical changes in the Mcoln1 ^−/− brain, we performed unbiased stereological measurements of the volume of four brain regions (cortex, hippocampus, thalamus and striatum) in two, three and seven month-old mice. We observed no atrophy of any of these brain regions at any of these three time points (Additional file 1: Figure S1A). Next we assessed whether more subtle effects were evident within the cortex of these mice by obtaining cortical thickness measurements from the somatosensory barrelfield cortex (S1BF), primary motor cortex (M1), lateral entorhinal cortex (LEnt) and primary visual cortex (V1) at two, three and seven months. No significant changes in cortical thickness were detected between Mcoln1 ^−/− and control mice at any age (Additional file 1: Figure S1B), suggesting that none of the examined regions of cortex undergoes overt neurodegeneration during the course of this disease.

Profound early activation of microglia and astrocytes in Mcoln1−/−mice

Since pronounced activation of microglia and astrocytes has been reported in the single available autopsy report from an MLIV patient [9], and in end-stage Mcoln1 ^−/− mice [26], we evaluated the onset and progression of these phenotypes in brain tissue from two, three and seven month-old Mcoln1 ^−/− mice. Immunostaining for the astrocyte marker GFAP revealed marked reactive astrocytosis in Mcoln1 ^−/− brains as early as two months of age. At this time we observed an intense GFAP immunoreactivity in the somatosensory barrelfield cortex, primary motor cortex and hypothalamus, with the most prominent staining intensity in the VPL/VPM region of thalamus (Figure 2A, B). Astrocytosis in Mcoln1 ^−/− mice progressed with age, and by the age of seven months intensely stained GFAP-positive cells were present throughout the entire area of thalamus and hypothalamus. These GFAP-positive astrocytes within the Mcoln1 ^−/− brain appeared hypertrophied with enlarged soma and thickened processes. However, astrocytosis appeared to progress at different rates between brain regions. For example, staining for GFAP was more pronounced in VPL/VPM at earlier stages of disease, and only subsequently became more evident in S1BF, the target cortical region for this somatosensory relay nucleus of the thalamus (Figure 2C). Thresholding image analysis in the VPL/VPM area of thalamus revealed a significant effect of age and genotype on GFAP staining intensity (F _{genotype × age} = 5.67; p < 0.0107) with increased staining in Mcoln1 ^−/− mice at two and three months of age (post-hoc t-test with Bonferroni correction: p _{2 months} < 0.05; p _{3 months} < 0.001; p _{7 months} > 0.05). GFAP staining intensity in S1BF cortex was also significantly affected by genotype and age (F _{genotype × age} =5.54; p = 0.0117), however post hoc test revealed significant increase in GFAP in Mcoln1 ^−/− mice only at the age of seven months (p < 0.01).

To examine the microglial response, brain tissue from two, three and seven month-old Mcoln1 ^−/− mice and wild-type littermates was immunostained for the microglia/macrophage marker CD68. Compared to the wild-type controls, the Mcoln1 ^−/− mice displayed pronounced activation of microglia early in disease progression, which was evident at two months of age, especially within the VPL/VPM (Figure 2D), throughout the cortex (Figure 2E), hippocampus, DLG and red nucleus (not shown). However, microglial response became less pronounced and widespread in Mcoln1 ^−/− mice at seven months of age. Microglia in control mice were faintly stained and exhibited a ramified morphology with a small cell body, whereas in the Mcoln1 ^−/− mice they were much more intensely stained and displayed more rounded and enlarged cell bodies, with short thickened processes characteristic of activated microglia or brain macrophages (Figure 2D, E). Thresholding image analysis in the VPL/VPM region of thalamus followed by two-way ANOVA test showed a significant effect of genotype (F _genotype = 11.54; p = 0.0027), and no effect of age (F _age = 0.098; p = 0.91) on CD68 staining intensity (Figure 2F). Analysis of CD68 staining in the S1BF region of the cortex revealed significant effect of genotype and age (F _{genotype × age} = 4.1; p = 0.03), with a significant increase in CD68 staining intensity at two (p < 0.001) and three (p < 0.001), but not at seven months (Figure 2E, F), when activated microglia appeared to be more localized to lamina IV. Similar changes in microglia morphology and increased staining intensity were observed with the microglial marker Iba1 (Additional file 2: Figure S2).

Ultrastructural analysis of the brain tissue (CA1 stratum radiatum of hippocampus) at seven months revealed accumulation of electron-dense storage bodies in Mcoln1 ^−/− astrocytes similar to those described previously in the cortex [25] (Additional file 3: Figure S3A). We also observed the formation of large aggregates of electron-dense organelles (or clumps). Most often they were engulfed by the plasma membrane which lacks synaptic contacts and, based on cell morphology [34], most likely belongs to microglia/macrophages (Additional file 3: Figure S3B).

Normal resting calcium concentrations in the somatosensory cortex of Mcoln1−/−mice

To assess if loss of TRPML1 and the early activation of glia resulted in sub-lethal neurotoxicity and impaired calcium homeostasis, we performed in vivo measurements of intracellular calcium concentrations in cortical neurons of Mcoln1 ^−/− and wild-type mice expressing the genetically-encoded ratiometric calcium sensor YC3.61 [28],[35]. An adeno-associated viral vector (AAV 2/1) was used to deliver YC3.61 to layer II/III neurons of the somatosensory cortex, a region where pronounced astrocytosis and microglial activation is present (Figure 2). Resting [Ca²⁺] was determined in the individual cell bodies and dendrites by in vivo multiphoton imaging of the injected area through implanted cranial window [28] in anesthetized mice. Analysis of resting calcium concentration in three month-old mice revealed no differences in either neuronal somata in layers II/III (Figure 3A, C) (3 WT mice, 888 cells; 5 KO mice 1875 cells analyzed) or neuropil in layer I (Figure 3B, D) (3 WT mice, 312 neurites; 5 KO mice 536 neurites analyzed). Resting [Ca²⁺] levels in the dendrites of both KO mice and control littermates were in good concordance with the previously reported values for wild-type mice [28]. Additionally, we observed large autofluorescent storage aggregates in the cytosol of Mcoln1 ^−/− neurons (Figure 3E) which, however, had very little effect on the levels of intraneuronal resting [Ca²⁺] (R = –0.074, p = 0.04), and even the cells with the highest amount of storage material had physiologically normal [Ca²⁺] levels (Figure 3E, F). These data show that the loss of TRPML1 does not result in neuroexcitotoxicity and provides evidence of physiological health of Mcoln1 ^−/− neurons at three months of age despite profound activation of microglia and astrocytes in the somatosensory cortex at this age.

Activation of microglia and astrocytes is not accompanied by neuronal loss in Mcoln1−/−mice

Activation of microglia has been linked to neuron loss in many neurodegenerative diseases [36]. To assess if chronic activation of microglia was associated with neuronal loss in Mcoln1 ^−/− mice later in disease progression, unbiased optical fractionator neuron counts were obtained in the brain areas with the most prominent microgliosis, i.e. the VPL/VPM region of thalamus, laminae IV and V of the somatosensory barrelfield cortex, the CA1 and CA3 subfields of the hippocampus, the DLG and red nucleus, in seven month-old mice (Additional file 4: Figure S4). We observed no significant differences in neuron number between wild type and Mcoln1 ^−/− mice in any of the examined regions, suggesting that activation of microglia was not caused by or resulted in neuronal death in the Mcoln1 ^−/− brain.

Abnormal synaptic plasticity in Mcoln1−/−mice

Since our data show no overt neuronal loss in the Mcoln1 ^−/− brain, we next examined the consequence of loss of TRPML1 on neuronal function, which might explain the severe motor and cognitive decline in MLIV. To assess synaptic function, we performed field recordings in the Schaeffer collateral pathway of acute hippocampal slices from seven month-old control and Mcoln1 ^−/− mice (n = 6). First, we examined basal synaptic transmission by measuring the synaptic input–output relationship, plotting the slope of the field excitatory postsynaptic potential (fEPSP) as a function of the presynaptic fiber volley amplitude at varying stimulus intensities. Our results revealed a non-significant trend toward decreased synaptic input–output coupling in the Mcoln1 ^−/− relative to control mice (p = 0.08) (Figure 4A). Second, we analyzed paired-pulse facilitation (PPF), a short-term form of presynaptic plasticity in which paired stimulation with a short inter-stimulus interval elicits a transient enhancement in neurotransmitter release. We observed significantly elevated PPF in the Mcoln1 ^−/− hippocampus at the shortest inter-stimulus intervals (ISI) (Figure 4B), implying that TRPML1 deficiency may affect neurotransmitter release. We also examined the impact of TRPML1 loss on synaptic plasticity by analyzing long-term potentiation (LTP), a long-lasting enhancement of synaptic strength induced by patterned high-frequency stimulation, which is thought to model synaptic modifications underlying learning and memory. Experiments in which LTP was induced by multiple trains of tetanic stimulation revealed that loss of TRPML1 led to significantly enhanced LTP (Figure 4C).

Alterations in synapse morphology in Mcoln1−/−mice

To determine if there were morphological changes associated with altered synaptic plasticity in Mcoln1 ^−/− mice, we performed electron microscopic analysis of synaptic terminals in the CA1 stratum radiatum of 7 month-old mice. We observed elongated post-synaptic densities (PSD) in excitatory (asymmetric) synapses (p = 0.024) in Mcoln1 ^−/− mice compared to control littermates (Figure 4D, E). Quantification of the asymmetric synapse density in CA1 stratum radiatum showed a trend towards a decrease in Mcoln1 ^−/− mice, but this did not reach statistical significance due to high variability (Figure 4F). Since axonal pathology is often accompanied by mitochondrial stress, we measured volume of mitochondria in neuropil of CA1 stratum radiatum in Mcoln1 ^−/− and control mice, but found no difference (Figure 4G).

Loss of Mcoln1 results in the early defects of myelination

The most remarkable ultrastructural feature observed by electron microscopy in the stratum radiatum of Mcoln1 ^−/− mice was reduced myelination (Figure 5A). Morphometric analysis revealed significant thinning of the myelin sheaths as shown by significantly lower g-ratio (axon diameter/fiber diameter, where fiber diameter is a sum of axon diameter and myelin sheath thickness) in Mcoln1 ^−/− mice (Figure 5B). No significant difference was observed in the diameter of myelinated axons between the two genotypes (Figure 5C). Impaired CNS myelination and a dysgenic corpus callosum are hallmarks of neuropathology in MLIV patients, and a decreased Luxol fast blue staining has been previously reported in the corpus callosum, deep layers of neocortex and cerebellar white matter tracts of the end-stage Mcoln1 ^−/− mice [26]. To follow the time-course of myelination deficits, we set out to measure myelination in two and seven month-old Mcoln1 ^−/− mice using FluoroMyelin Green stain. Analysis of coronal sections (Figure 5D) revealed profound thinning of the corpus callosum and reduced myelination of internal capsule in Mcoln1 ^−/− mice at both time points. Strikingly, we observed no anterior commissure (AC) in the Mcoln1 ^−/− mice at two months. However, a dysmorphic and smaller AC appeared in the sections from seven month-old mice (Figure 5D). Analysis of the FlyoroMyelin Green staining in the sagittal sections revealed malformation of the corpus callosum (CC) in the Mcoln1 ^−/− mice. Specifically, similar to MLIV patients, Mcoln1 ^−/− mice displayed a generally hypomorphic corpus callosum, with dysgenic rostrum and dysgenic or absent splenium compared to WT littermate controls (Figure 5E).