Pathological mitochondria in neurons and perivascular astrocytic endfeet of idiopathic normal pressure hydrocephalus patients

The study was done according to the following approvals: Regional Committee for Medical and Health Research Ethics (REK) of Health Region South-East, Norway: (i) Brain Tissue Research Biobank (Approval no. REK 2010/1030 and 2012/1157). (ii) Brain biopsy from iNPH patients (Approval no. REK 2009/2060). (iii) Storage of brain tissue from REF patients, i.e. patients undergoing neurosurgery for various reasons (epilepsy, tumor or cerebral aneurysms) wherein removal of brain tissue is necessary as part of treatment (Approval no. REK 2011/2306). Oslo University Hospital (Approvals no. 10/6806 and 2011/19311). The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Consecutive patients were included after oral and written informed consent.

The study design was prospective and observational: Randomization of patients was not relevant; neither was a priori sample size calculation needed. The person analyzing the electron microscopy specimens (MMHO) was blinded to patient data and diagnosis.

The sampling of brain tissue was done in conjunction with placement of an ICP sensor for overnight ICP monitoring. The diagnostic procedure was done in the operation room. In local anesthesia, a skin incision was done frontally on the right side, a small burr hole about 1 cm in diameter was made, the dura opened, and the brain surface exposed. A disposable Nashold Biopsy Needle (Integra Radionics, Burlington, MA, USA) was introduced immediately below the cortical surface and a biopsy (0.9 mm × 10 mm) aspirated through the needle. Then, a solid ICP sensor (Codman MicroSensor™, Johnson & Johnson, Raynham, MA, USA) was tunneled subcutaneously, zeroed against the atmospheric pressure and introduced 1–2 cm into the frontal cortex parenchyma via the small opening that was established by the biopsy. The overnight monitoring and subsequent analysis of ICP scores was undertaken as described in detail before [2, 3].

The detailed technical procedures for the handling, processing and analysis of the frontal cortex biopsies have been reported [17, 18]. In brief, the dissected brain tissue was fixed in the operating room to avoid time delay by immersion in 0.1 M phosphate buffer containing 4% paraformaldehyde and 0.25% glutaraldehyde and cut into small specimens (typically 0.5 mm × 0.5 mm × 1 mm). The electron microscopy specimens were subjected to freeze substitution and infiltration in Lowicryl HM20 resin (Polysciences Inc., Warrington, PA, USA, Cat 15924). A Reichert ultramicrotome (Reichert Techn., Wien, Austria) was used for cutting sections of 80 nm thicknesses, which were mounted on nickel grids and further processed for counterstaining with 1% uranyl acetate and 0.3% lead citrate for ultrastructure analysis.

The electron micrographs were recorded with a FEI Tecnai™ 12 transmission electron microscope (FEI Company, Hillsboro, OR, USA) at different magnifications of 43,000×, 26,500×, 20,500×, 11,500×, 9900×, 8200×, 6000×, and 4200×. Electron micrographs were saved as 8-bit, 2048 × 2048 pixels tagged image format images (TIFFS). Images were subsequently analyzed using a customized version of a MATLAB toolbox [19] designed for evaluation of mitochondrial area (Additional file 1: Fig. S1). Images were segmented manually by drawing polygonal selections over identifiable cellular compartments and structures. Neuronal soma, and pre- and post-synaptic terminals, and astrocytic endfoot processes were identified and outlined.

The number of mitochondria per patient; area of mitochondria per neuronal soma in each patient, and mitochondrial area fraction (area covered by mitochondria divided by total cytoplasmic area).

We classified all mitochondria systematically into three categories based on the following qualitative features:

A cartoon illustrating the differentiation of normal, pathological and clustered mitochondria is shown in Additional file: Fig. S2. 1

The number, area and fraction of autophagic vacuoles were examined in a similar fashion as the mitochondrial analysis (Additional file: Fig. S1). The length of postsynaptic densities of asymmetric synapses and length of mitochondria-endoplasmic reticulum contact sites were measured interactively in the MATLAB program (Additional file: Fig. S1). 1 1

We performed light microscopic immunohistochemistry and semi-quantitative assessment of glial fibrillary acidic protein (GFAP), aquaporin-4 (AQP-4), dystrophin 71 (Dp71), and Cluster of Differentiation 68 (CD68), as previously described [17, 21].

Statistical analyses were performed using the SPSS software version 25 (IBM Corporation, Armonk, NY, USA). For each individual, the data were pooled and presented as mean ± standard deviation. Differences between categorical data were determined using Pearson Chi-square test, and differences between continuous data were determined using independent samples t-tests, Mann–Whitney U-test, or multivariate analysis to take into consideration age differences between groups. Correlation between variables was determined using Pearson correlation coefficient. Statistical significance was accepted at the 0.05 level (two-tailed).

The data presented in this work is available upon request.

The patient material consists of 9 reference (REF) individuals and 30 iNPH patients. Demographic information is presented in Additional file: Table S1. The patient cohorts were different regarding age, co-morbidity (presence of diabetes), disease duration and severity of iNPH symptoms. 1

Biopsies from the 30 iNPH patients were from grey matter of frontal cortex, while REF biospies were from the grey matter of temporal cortex in four epilepsy cases, frontal cortex in three patients (2 aneurysm, 1 tumor), parieto-occipital cortex in one epilepsy case, and the occipital cortex in another epilepsy patient.

The distance between endoplasmic reticulum (ER) and mitochondria was used as a measure of mitochondria-endoplasmic reticulum contact sites, and this distance was significantly reduced in iNPH patients, as illustrated in Fig. 3. The lengths of ER examined in these cohorts were comparable (0.67 ± 0.30 nm² and 0.79 ± 0.43 nm² in REF and iNPH patients, non-significant difference).

With regard to autophagy, we examined in individuals of REF and iNPH cohorts on average 12.7 ± 6.5 and 12.1 ± 5.8 soma, which included mean cytoplasmic areas of 16,982,978 ± 5,335,206 nm² and 21,797,964 ± 7,383,110 nm², respectively. Autophagy vacuoles (Fig. 4a, b) were seen in 2/9 REF individuals and in 16/28 iNPH cases (P = 0.068, Chi square test). The two REF individuals presenting autophagy vacuoles had long-lasting epilepsy (7 and 30 years durations), and both reported some cognitive impairment. Compared with REF subjects, the area of autophagy vacuoles per soma was 5–6-fold higher in iNPH (Fig. 4c), and the area fraction of autophagic vacuoles per soma was about fivefold higher in iNPH (Fig. 4d). Differences between groups were, however, not significant when taking into account the age differences between groups. Further examples of autophagic vacuoles are given in Additional file 1: Fig. S6.

The post-synaptic density differed between iNPH and REF individuals (Fig. 5). Hence, the post-synaptic density length was on average 0.26 ± 0.06 µm in iNPH as compared to 0.61 ± 0.09 µm in REF subjects (P < 0.001; Fig. 5c).

In biopsies from individuals of the REF and iNPH cohorts, we examined mean 18.0 ± 8.2 and 20.3 ± 10.4 astrocytic endfoot processes, respectively (not significantly different). The average area of endfoot processes were 2,040,461 ± 1,009,156 nm² and 1,587,893 ± 641,170 nm², respectively.

Figure 6 shows normal mitochondria in astrocytic endfeet of REF subjects (Fig. 6a) and normal and pathological mitochondria in astrocytic endfeet of iNPH patients (Fig. 6b). The number of normal mitochondria in astrocytic endfoot processes was significantly reduced in iNPH (Table 3). Increased mass of pathological mitochondria in astrocytic endfoot processes of iNPH individuals was revealed by significantly increased number and area of pathological mitochondria (Table 3).

With increasing astrogliosis, measured as high GFAP immunohistochemical labelling on light microscopic examinations, there was significantly reduced number of normal mitochondria (Fig. 7a) and significantly increased number of pathological mitochondria (Fig. 7b).

By using immunofluorescence, we also examined how the number of normal and pathological mitochondria related to the expression of other key molecules of the perivascular endfeet (Table 4). For all individuals of the iNPH and REF cohorts, the number of normal mitochondria was significantly and positively correlated with the expression of perivascular AQP4 and Dp71, while expression of AQP4 and Dp71 in neuropil was not correlated with number of normal mitochondria in the endfoot processes. On the other hand, the number of pathological mitochondria was significantly and negatively correlated with expression of AQP4 perivascularly and in neuropil, and significantly and negatively correlated with Dp71 perivascular (Table 4). Reduced perivascular expression of AQP4 was also significantly correlated with increased area of autophagic vacuoles in neuronal soma (Additional file 1: Fig. S7). It should be noted that the AQP4 expression referred to in Table 4 and Additional file 1: Fig. S7 refers to arbitrary units (AU); reduced arbitrary units (AU) is indicative of increased AQP4 expression because higher AQP4 causes less transmission of light and thereby reduction of arbitrary units. With regard to expression of CD68, a marker of inflammation, higher number of normal mitochondria seemed to be associated with lower expression of CD68 (P = 0.056; Table 4).

With increasing area (Fig. 8a) and fraction (Fig. 8b) of pathological mitochondria in endfoot processes, the mean intracranial pressure wave amplitude (MWA_ICP) was increased, while this was not seen for mean intracranial pressure (ICP) (data not shown).

Among the 30 iNPH patients, 14 (48%) had arterial hypertension and 10 (35%) diabetes mellitus. Except for a higher number of pathological mitochondria in the astrocytic endfoot processes of iNPH patients with diabetes (Additional file: Fig. S8), there were no significant differences between iNPH patients with or without comorbidity (arterial hypertension or diabetes) for any of the other reported mitochondria attributes. 1

iNPH is a subtype of dementia, characterized by abnormal CSF circulation homeostasis, and clinical improvement by shunt surgery, at least to some degree and for a certain period [3]. Of the iNPH patients in the present study, 27/30 individuals received a shunt. Twenty-five of these were clinical responders, qualifying for definite iNPH, according to the Japanese guidelines [22]. Therefore, we consider the present observations as representative for iNPH patients in general.

iNPH shares many similarities with Alzheimer’s disease. For instance, brain biopsies from iNPH patients show extracellular deposition of amyloid-β and cellular accumulation of neuro-fibrillary tangles of HPτ [5, 23], as well as other neuropathological similarities between iNPH and Alzheimer’s disease [4]. Previous MRI studies of extra-vascular movement of a CSF tracer showed delayed clearance of the CSF tracer from CSF spaces and from glymphatic pathways of iNPH patients [14–16]. While delayed extravascular clearance of a CSF tracer might be relevant for clearance of soluble amyloid-β, the link to cellular clearance of HPτ seems less obvious.

Mitochondria are dynamic organelles that are key producers of adenosine triphosphate, and are instrumental for maintaining cellular metabolic homeostasis, cell survival and cell death. The mitochondrial machinery consumes oxygen to generate sufficient energy for the maintenance of normal cellular processes. To protect mitochondria from damage, cells and mitochondria have developed a wide-range of mitochondrial quality control mechanisms that clear damaged mitochondrial cargo, enabling the mitochondria to repair damage and restore normal function [24].

In this study, we systematically categorized and analyzed normal, pathological, and clustered mitochondria in soma and nerve terminals of REF and iNPH patients. In neuronal soma of iNPH patients, the fraction of normal mitochondria was reduced while the numbers of pathological and clustered mitochondria were increased. This finding may be indicative of defective cellular clearance and impaired handling of damaged mitochondria in iNPH. Normally, dysfunctional and damaged mitochondria are repaired in the soma, or they become cleared by mitophagy, a process of imperative role for mitochondrial health [25]. Damaged mitochondria can also be degraded by astrocytes through translocation from neurons [26]. This process is required to maintain a functional mitochondrial biomass in cells. We also found an increased mass of clustered mitochondria in the soma of iNPH subjects, which as well may be indicative of defective mitophagy. To our knowledge, systematic electron microscopy analyses of mitochondrial health in different neuronal compartments and astrocytic endfoot processes has previously not been done in iNPH patients.

The present observations of increased occurrence of pathological and clustered mitochondria in neuronal soma further link iNPH with other neurodegenerative diseases and with Alzheimer’s dementia. In neurodegenerative diseases such as Huntington’s disease [27, 28], Alzheimer’s disease [29] and amyotrophic lateral sclerosis [30], breakdown of mitochondrial dynamics has previously been shown to affect mitochondrial biogenesis that in turn may lead to insufficient axonal mitochondrial transport. Moreover, impaired removal of damaged and dysfunctional mitochondria has been shown to facilitate pathological features of Alzheimer’s disease [31].

In iNPH individuals, we found reduced ratios of mitochondria in presynaptic terminal versus soma as well as reduced ratios of mitochondria in postsynaptic terminals versus soma. Moreover, the number of pathological mitochondria was increased in both pre- and postsynaptic terminals. These observations may be indicative of impaired mitochondrial trafficking from the nerve terminal to soma and of disturbed synaptic activity in iNPH patients. In order to maintain compartmentalized energy demands, mitochondria follow anterograde and retrograde transport, which control the mitochondrial biogenesis in axon terminals and cell bodies [32, 33]. Since the energy demand is higher in the pre- and post-synaptic terminals of neurons, sufficient number of mitochondria in the terminals is required to maintain synaptic function [34, 35]. Normally, damaged and dysfunctional mitochondria in the distal regions of the neuron are transported retrograde to the soma where they are repaired through fission and fusion mechanisms [36, 37]. The mitochondrial movement is highly associated to this mitochondrial fusion-fission machinery [38].

Breakdown of the normal mitochondrial trafficking may lead to neurodegeneration. For example, in a model of Huntington’s disease induced pluripotent stem cells (iPSCs) derived neurons, impaired mitochondrial trafficking was associated with neuronal abnormalities [39]. Abnormal mitochondrial trafficking has also been reported in other neurodegenerative diseases, such as Parkinson’s disease [40] and Alzheimer’s disease [38], which accelerates synaptic energy deprivation and oxidative stress [41]. Mitochondrial motility is severely hampered in Alzheimer’s disease due to amyloid-β and HPτ pathology. The amyloid-β pathology reduced the number of functional mitochondria in presynaptic terminal in Alzheimer’s disease [42]. The brains of Alzheimer victims show abnormal accumulation of defective mitochondria in the nerve terminal and soma [43]. Importantly, the deviation from appropriate removal of damaged mitochondria from neurons and astrocytes contribute to failure of synaptic and neuronal functions.

Another indication of defective cellular clearance of metabolites in iNPH was the evidence of shortened distance between mitochondria and endoplasmic reticulum, resulting in increased mitochondria-endoplasmic reticulum contact sites in iNPH, as compared with REF individuals. Comparable to the present findings, a previous study reported shortened distance between mitochondria and endoplasmic reticulum and increased number of mitochondria-endoplasmic reticulum contact sites per cell in iNPH patients who presented with amyloid plaques and cellular neuro-fibrillary tangles of hyperphosphorylated tau (HPτ) [44]. In addition, the authors reported that number of mitochondria-endoplasmic reticulum contact sites positively correlated with increasing age [44]. The presently reported differences in distance between mitochondria and endoplasmic reticulum between REF and iNPH individuals were significant after considering age differences.

The characteristics of mitochondria-endoplasmic reticulum contact sites play a role in Ca²⁺ signaling, mitochondrial dynamics, bioenergetics and turnover [45, 46]. In Alzheimer’s disease, mitochondria-endoplasmic reticulum contact sites were associated with amyloid-β and HPτ accumulation [47]. As previously suggested [44], the elevation of mitochondria-endoplasmic reticulum contact sites may alter Ca²⁺ homeostasis [48]. Tentatively, this mechanism might deteriorate mitochondrial function and trafficking in iNPH.

As compared with another neurodegenerative disease, amyotrophic lateral sclerosis, endoplasmic reticulum pathology was associated with defective autophagy and mitochondrial abnormalities [49].

Autophagic vacuoles were more abundant in iNPH than REF individuals. An increased amount of non-fused autophagic vacuoles may be related to defective cellular clearance of debris [50], and inefficient autophagic machinery to clean up the damaged mitochondria in the neuronal soma [31] may be one mechanism behind the elevated levels of pathological mitochondria in iNPH.

Furthermore, it may be speculated that a dumping of autophagic vacuoles in iNPH may contribute to accumulation of amyloid-β and HPτ protein in iNPH [23], as has previously been shown in Alzheimer’s disease [5, 43]. The extensive accumulation of immature autophagic vacuoles in neurites is evident in Alzheimer’s disease [51]. Since damaged mitochondrial accumulation is an early event in Alzheimer’s disease and therefore elevated levels of autophagic degradation of defective mitochondria may halt the progress of Alzheimer’s disease pathogenesis.

The average length of the postsynaptic densities was significantly reduced in iNPH, as compared with REF subjects. The postsynaptic density length provides a measure of the strength of the synaptic activity. It has been shown that oligomeric amyloid-β in close proximity to the postsynaptic area may shrink the postsynaptic density length, reduce synaptic plasticity and increase synaptic loss [52]. The presence of amyloid-β in iNPH patients and reduced postsynaptic density length may therefore be one contributing factor to the cognitive decline and synaptic loss in iNPH. In comparison, synaptic loss is considered to be an early event of Alzheimer’s disease [53].

Mitochondrial trafficking and distribution is connected to synaptic activity [54]. In the present iNPH cohort, we found that the number of pathological mitochondria was significantly increased in both pre- and post-synaptic terminals, and the area of normal mitochondria per postsynaptic terminal was significantly reduced. The increased mass of pathological mitochondria in the synaptic terminals may be related to insufficient energy supply, which may cause shrinkage of synaptic length that in turn leads to poor synaptic transmission and synaptic plasticity. In comparison, synaptic mitochondrial pathology is one early pathological feature of Alzheimer’s disease, which contribute to degeneration of synaptic functions [55].

The present results showed altered mitochondria phenotype in astrocytic endfoot processes of iNPH individuals. The number of normal mitochondria was reduced and the number and area of pathological mitochondria increased in the endfoot processes of iNPH patients. While this is a morphological study, the data may be indicative of altered integrity of the neuro-vascular unit and blood–brain barrier function in iNPH.

Failure of the neurovascular unit may alter blood flow and result in a low-grade ischemia, which has been demonstrated previously in iNPH [56]. Astrocytes seems critical for bridging the connection between neurons and blood vessels, playing key roles in brain metabolism [57]. The astrocytic end-feet wrap up the microvasculature contributing to maintenance of blood–brain barrier function [58]. In astrocytes, the positioning of mitochondria seems to influence Ca²⁺ signaling [59]. Notably, mitochondria play a crucial role in glutamate-glutamine shuttle in astrocytes [60]. However, the spatial function of mitochondria is yet to be investigated in astrocytic processes such as in perivascular astrocytic endfoot.

Proper mitochondrial function is a key factor for astrocytic energy metabolism, and mitochondrial dysfunction may lead to reactive cortical astrogliosis [61]. A recent animal study reported that mitochondrial dysfunction halts generation of new astrocytes and increases nerve cell death [61]. Transmission electron microscopic observation showed that degeneration of astrocytic processes is associated with increased presence of swollen and abnormally shaped mitochondria seen in ischemic neuronal injury during 5 to 48 h post-ischemia [62]. Notably, the electron microscopic 3D reconstruction revealed the presence of bundle of mitochondria in the endfoot processes close to perivascular endfoot membrane [12].

In this study, we also found an association between increase of pathological mitochondria in endfoot processes and loss of perivascular AQP4 in iNPH. Previously, we reported loss of perivascular AQP4 in iNPH [17, 63]. The present observations of reduced number of normal and increased number of pathological mitochondria in astrocytic endfoot processes may be indicative of defective energy metabolism in endfoot processes of iNPH, which in turn may be associated with astrocytic malfunctions and impaired water and solute exchange. In comparison, previous ultrastructural observations in chronic compressive spinal cord injury showed swollen and disrupted mitochondria in astrocytes of spinal cord [64]. Moreover, observations in animals suggest that aging aggravates mitochondrial alterations in the astrocytic endfeet [65].

The data presented in this work is available upon request.

The data presented in this work is available upon request.