Phosphorylated TDP-43 near blood vessels is linked to Alzheimer’s disease and loss of blood vessel support proteins CD146 and Aquaporin-4

Accumulation and deposition of amyloid‐beta (Aβ) is a key event in Alzheimer's disease (AD) thought to underly the neuronal loss and brain atrophy seen in these patients. The accumulation can be due to an increased Aβ production, as in the case of familial AD, but also to an impeded clearance of the peptide. Such dysfunction is thought to underly sporadic forms of AD and arises when systems involved in clearing excess Aβ are disrupted or altered. The blood–brain barrier (BBB), formed by pericytes, astrocytic endfeet, and endothelial cells, is one of these clearance systems known to be affected in AD. The impairment of the system is characterized by a reduced amount of endothelial‐pericyte tight junctions, loss of pericytes [35, 36], and increased infiltration of fibrinogen [23] and albumin from the periphery [18, 37, 45]. The glymphatic system is another system that is known to play an important role in Aβ clearance. Its function is dependent on the astrocytic endfeet which forms a fluid‐filled perivascular space around the brain vessels. The perivascular fluid is transported into the parenchyma via Aquaporin‐4 channels (AQP4) expressed on the astrocytic endfeet, and a bulk flow of interstitial fluid is created. This bulk flow pushes the fluid, along with unwanted soluble toxic substances such as Aβ and islet amyloid polypeptide, toward veins where it is taken up and transported away [33]. Patients with AD show both reduced expression of AQP4 and increased perivascular space [5], pointing toward a malfunctioning glymphatic system. Impairment of both clearing systems (i.e., BBB/pericytes and the glymphatic system) contributes to an accumulation of the amyloid peptides, which start to seed each other and deposit in the vessel walls and brain parenchyma [16].

Trans‐active response DNA binding protein of 43 kDa (TDP‐43) is a ubiquitous protein essential for the development of the central nervous system and known to regulate pathways that affect cell survival, metabolism, mitochondrial function, and synaptic function dependent on the tissue [1, 29]. The majority of TDP‐43 is nuclear during physiological conditions, with a small portion being shuttled between the nucleus and the cytoplasm. During stress conditions (such as oxidative stress) and inflammatory events [11], TDP‐43 is released to the cytoplasm where it regulates the formation of stress granules, ribonucleoprotein transport granules, translation, and other processes [32]. Under such pathological conditions, TDP‐43 can be hyperphosphorylated (pTDP‐43) and ubiquitinated [9], and despite being labeled for degradation, pTDP‐43 starts to accumulate in the cytoplasm in the form of inclusions. These inclusions are foremost associated with frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS), as neuronal inclusions of the pTDP‐43 are a neuropathological hallmark of these diseases. However, in recent years it has become evident that more than 57% of all AD patients demonstrate hippocampal pTDP‐43 inclusions, and that the inclusions are associated with faster cognitive decline and a more aggressive disease course [27]. The pTDP‐43 inclusions have foremost been reported to occur in neurons, where they cause neurotoxicity and cytoskeletal dysfunction [13, 41], but the peptide can also aggregate and deposit in astrocytes [26, 40]. The consequences of the astrocytic pTDP‐43 inclusions are little known, although a few studies suggest that the neuroprotective properties of astrocytes are attenuated when pTDP‐43 accumulates and forms inclusions inside the cells [24, 25]. Interestingly, a study using electron microscopy has found small TDP‐43 inclusions associated with vessels in brain tissue from FTLD patients [26]. The authors of this article suggested that they were located inside the astrocytic endfeet. Although this finding needs to be verified, it raises the question of whether TDP‐43 inclusions affect the astrocytic endfeet function. Indeed, patients with FTLD show BBB permeability [18, 43] and inclusions of pTDP‐43 have been found in the vessel walls in the frontal cortex of ALS patients [14]. Moreover, mice overexpressing TDP‐43 show BBB permeability [43] and the loss of TDP‐43 function, which happens when TDP‐43 mislocates and aggregates, causes a reduction in blood circulation and vessel mispatterning in zebrafish [34]. Whether TDP‐43 has a direct implication on glymphatic failure is not yet determined, but a study on a TDP‐43 ALS mouse model showed that the glymphatic system is already impaired in the early stages of the disease [42]. In addition, TDP‐43 reduces the efficacy of AQP4 integration into the cell membrane [22]. The aim of the current study was to further investigate the presence of pTDP‐43 inclusions in astrocytic endfeet and determine if the accumulation of inclusions is associated with AD pathology, BBB damage, and alterations in the glymphatic system.

Our study demonstrates the presence of pTDP‐43 inclusions in astrocytic endfeet in the CA1 region of AD patients. The study also shows an association between the load of these inclusions and AD pathology and disease severity. Our findings further link the accumulation of pTDP‐43 inclusions to alterations in BBB integrity and reduction in astrocytic AQP4 expression.

The presence of pTDP‐43 inclusions in astrocytes aligns with previous research documenting similar findings in other neurodegenerative disorders like FTLD and ALS [2, 12, 14, 28]. Our study also confirmed the presence of TDP‐43 specifically within astrocytic endfeet, as suggested by a previous study investigating microvasculopathy in FTLD using immunoelectron microscopy [26]. The perivascular pTDP‐43 inclusions were more pronounced in AD patients. Whether this is a result of increased pTDP‐43 accumulation due to the AD pathology or if the inclusions themselves induce the AD pathology is yet to be determined, but a previous mouse study demonstrating an exacerbation of AD pathology in animals injected with TDP‐43 [38] is in favor to the latter hypothesis. It is further interesting that the inclusion load correlated positively with NFT Braak stages and Aβ burden. Although the specific role of perivascular TDP‐43 inclusions in AD has not been studied before, it is noteworthy that our findings support the previous epidemiological research indicating a worse prognosis in AD patients with TDP‐43 inclusions [19, 20] and patients with quadruple misfolded proteins (Aβ, tau, alpha‐synuclein, and TDP‐43) [21]. However, it should be noted that our study failed to find a correlation between perivascular pTDP‐43 and pTDP‐43 in the pyrimadal region of CA1, even though the pTDP‐43 inclusion load in this region is also increased in AD patients. Such an increase has been reported before and the pTDP‐43 inclusions were primarily found in the pyramidal neurons of the hippocampus [3, 15]. It may thus be that neuronal accumulation of pTDP‐43 is only partly associated with the astrocytic perivascular accumulation of pTDP‐43 and involves slightly different pathological events.

Our results also demonstrated an inverse relationship between perivascular pTDP‐43 and alterations in CD146 expression. This adhesion molecule is located at the intercellular junctions of endothelial cells and promotes BBB formation by mediating the interaction between endothelial cells and pericytes [4]. During neuroinflammatory conditions, CD146 is shed from the cell surface, and elevated levels of soluble CD146 in cerebrospinal fluid are associated with BBB damage [39]. The link between TDP‐43 pathology and BBB alterations has been shown before. Patients with FTLD show increased BBB permeability [18, 43] and loss of mural cells (which include BBB‐forming pericytes) has been found to correlate with TDP‐43 pathology in the cortex of AD patients [6]. Preclinical studies further show that overexpression of TDP‐43 in mice causes BBB permeability [43] and vessels in TDP‐43 zebrafish models appear distorted [34]. These findings collectively indicate an association of TDP‐43 pathology with BBB alterations. Of note, our studies did not find a correlation between IgA area fraction and pTDP‐43 inclusions, which suggests that pTDP‐43 is not associated with major BBB damage such as leakiness, but rather minor alterations in the BBB integrity. Additionally, although our IgA analysis, in line with previous studies [18, 37, 45], suggests changes in the BBB integrity in AD patients, the slight decrease in CD146 expression in AD patients did not reach significance when compared to NDC. The lack of detectable changes in the CD146 expression may be due to a lack of power and the fact that ImageJ is a rather crude analysis method but could also imply that changes in the CD146 expression are not an AD specific event. We also found a reduced presence of AQP4 around small vessels of AD patients compared to NDC along with a negative correlation with age and the severity of AD pathology both before and after correcting for age. These findings are in line with a previous study which also found a reduction in AQP4 in the frontal cortex of AD patients and negative correlations between AQP4 expression and NFT Braak stages as well as Aβ burden after correcting for age [44]. Given that APQ4 plays a crucial role in the glymphatic system, and thereby clearance of excessive extracellular Aβ and p‐tau, it is tempting to speculate that the glymphatic system is malfunctioning in AD patients and that this underlies the formation of tangles and accumulation of Aβ in the brain of these patients. The loss of AQP4 in AD may be due to several pathological events, but it is interesting that we found a negative correlation between perivascular pTDP‐43 and AQP4 area fractions after correcting for age. The association between AQP4 expression and age has been described before, but most studies have found a positive relationship [30, 44], rather than a negative, between normal aging and AQP4 expression. However, the positive correlations with age found in the former studies were with the overall AQP4 expression, not specifically the perivascular AQP4, in the analyzed brain regions of non‐demented individuals. However, one of the studies also analyzed the perivascular AQP4 and found no significant correlation with age, but, interestingly, the trend of the correlation was negative (r = −0.210, p = 0.150) [43]. In view of these previous findings and our own results, we speculate that AQP4 relocates from the astrocytic endfeet to other compartments of the cell, leaving the glymphatic system affected also in the aged brain.

It should further be noted that a correlation was not found between TDP‐43 and AQP4, which suggests that the AQP4 downregulation may be foremost related to inclusions of the TDP‐43 phosphorylated at Ser409/410. Although the relationship between AQP4 and pTDP‐43 needs to be investigated further, it is tempting to speculate that pTDP‐43 pathology may contribute to the AQP4 loss in AD patients. This idea is reinforced by a previous cell culture study showing a correlation between the depletion of nuclear TDP‐43 and the reduction of AQP4 localized on the endfeet membranes [22]. Additionally, studies on a TDP‐43 ALS mouse model show that the glymphatic system is impaired already in the early stages of the disease [42].

Lastly, the limitation of this study foremost concerns the relatively small sample sizes in different groups, which may increase the likelihood of false negatives and/or loss of potential relationships. Therefore, additional studies with larger cohorts are necessary to enhance the validity and generalizability of our findings. Another limitation concerns the limited neuropathological evaluation information. The brain tissue from the cases included in this study was collected and evaluated before 2014 and hence the TDP‐43 staging analysis has not been performed. Instead, the cases were neuropathologically evaluated using Braak stages for both tau and Aβ pathology, which at the time was practiced at the NBB. Hence, we have neither CERAD nor Thal staging values, which would be of interest to further explore the associations between perivascular pTDP‐43 pathology and other amyloid peptide aggregates. Finally, since our study is primarily observational, we encourage future experimental studies to unravel the temporal aspect of TDP‐43 accumulation, AD pathology, and alterations in the BBB and the glymphatic system.

Our study suggests that pTDP‐43 inclusions in the astrocytic endfeet can play a role in AD pathology progression by affecting brain clearance functions such as BBB integrity and the glymphatic system. Future studies investigating the direct impact of pTDP‐43 on AQP4 regulation and BBB are warranted as they might shed light on mechanisms implicated in the unfavorable prognosis of AD patients with TDP‐43 inclusions and point out possible therapeutic targets.

J.S. and M.W. contributed to the study concept and design. J.S. performed the immunostainings and analyzed the data with support from D.P. and was a major contributor in writing the manuscript. D.P. performed and analyzed the IgA staining. The NBB contributed to the brain tissue and neuropathological evaluation. All authors have read and revised the manuscript for intellectual content and agreed to the final version of the manuscript.

The study was funded by the Brain Foundation (FO2023‐0113), Crafoord Foundation (20230519), Olle Engkvists Foundation (227‐0209), Dementia Foundation (2023), Greta and Johan Kockska Foundation (2022), and the Åhlén Foundation (233002). None of the funders have been involved in the design of the study and collection, analysis, interpretation of data, and/or in the writing of the manuscript.

The authors declare that they have no competing interests.

Informed consent for using brain tissue as well as clinical data for research purposes was obtained from the patients or from their closest relatives in accordance with the International Declaration of Helsinki and the Code of Conduct for Brain Banking. The tissue collection protocols were approved by the medical ethics committee of VU Amsterdam and the Swedish Ethical Review Authority approved the study (Dnr 2021/04270). All data were analyzed anonymously.

The authors would like to thank Christina Nuñez‐Diaz for demonstrating ImageJ and discussing the analysis approach and Josefine Andreasson for pilot TDP‐43 work.

Santiago J, Pocevičiūtė D, The Netherlands Brain Bank , Wennström M. Perivascular phosphorylated TDP‐43 inclusions are associated with Alzheimer's disease pathology and loss of CD146 and Aquaporin‐4. Brain Pathology. 2025;35(2):e13304. 10.1111/bpa.13304

All data generated or analyzed during this study are included in this published article.