An in vitro model yields “importin” new insights into chronic traumatic encephalopathy: damaged astrocytes stop “thrombospondin” to the injury

Abstract

This Editorial highlights a study by Jayakumar and colleagues (2016) in the current issue of Journal of Neurochemistry. The authors introduce an in vitro model of chronic traumatic encephalopathy (CTE) to explore the mechanistic underpinnings of CTE pathogenesis, including investigation of how traumatized astrocytes affect traumatized neurons through the release of secreted factors. The model recapitulates two key features of the human post-mortem CTE brain: neuronal tauopathy and TDP-43 proteinopathy—the respective accretion of hyperphosphorylated tau and cytoplasmic hyperphosphorylated and ubiquitinated TDP-43. Oxidative stress and casein kinase 1 episilon (CK1ε) are identified as key upstream regulators of cytoplasmic TDP-43 phosphorylation, and this phosphorylation is found to correlate with decreased importin-β protein level and a decline in synaptic integrity. RNA silencing of importin-β is sufficient to mimic both the phospho-TDP-43 accumulation and synaptic injury observed after mild in vitro trauma. Strikingly, Jayakumar et al. find that thrombospondin-1 (TSP-1), a protein secreted by traumatized astrocytes at elevated levels during the initial 5 days after damage, can attenuate CK1ε phosphorylation of TDP-43 and synaptic injury. However, TSP-1 secretion by astrocytes is lost at 10–15 days post-injury, and neurons succumb to unchecked TDP-43 pathogenesis.

Graphical abstract

Highlight

Chronic traumatic encephalopathy (CTE) is a neurological disorder initially considered to be a rare disease associated with repeated sports-related head injuries (Angoa-Perez et al. 2014). However, emerging studies now document CTE in a wider array of patients that includes military veterans and victims of domestic abuse, making it a larger public health issue than originally anticipated. The symptoms of CTE often manifest years after single or repetitive mild to moderate traumatic brain injuries and include the development of devastating cognitive and neurobehavioral deficits that in the worst cases can result in suicide (Angoa-Perez et al. 2014). Despite intense public interest, driven in large part by the post-mortem diagnosis of CTE in numerous former National Football League players, there are no tools for definitively diagnosing CTE in trauma victims prior to death. In addition, there are large gaps in our knowledge of the molecular and cellular mechanisms involved in disease progression.

CTE is distinguished from other neurodegenerative disorders by a distinct pattern of tau protein deposition observed in the post-mortem brains of patients (Angoa-Perez et al. 2014). This made tau the lead suspect in the search for an instigator of CTE pathology. However, more recently, elevated levels of transactivating DNA-binding protein-43 (TDP-43) were documented in the brains of CTE patients (McKee et al. 2010;King et al. 2010). TDP-43 is a transcription factor which, after phosphorylation, translocates into the nucleus where it regulates the expression of a wide array of genes. Dysregulation of TDP-43 localization and function leads to TDP-43 “proteinopathy,” consisting of cytoplasmic aggregation of hyperphosphorylated and ubiquitinated TDP-43 that becomes neurotoxic (Xu et al. 2010). TDP-43 proteinopathy is implicated in the pathogenesis of amyotrophic lateral sclerosis, frontotemporal lobe dementia, Alzheimer’s disease, and other neurodegenerative disorders, while its role in CTE is unclear. A better understanding of how TDP-43 proteinopathy occurs and whether it contributes to CTE pathogenesis will help forward the ultimate goal of developing therapeutic strategies for this incapacitating disease.

To elucidate cellular disease mechanisms, in vitro models that closely mimic the in vivo pathology are essential. In their article, Jayakumar and colleagues demonstrate that mild fluid percussion trauma in neuronal cell culture causes progressive phospho-tau (p-tau) and phospho-TDP-43 (p-TDP-43) accumulation that is exacerbated by multiple impacts (Jayakumar et al. 2016). Delayed loss of synaptic integrity and cell death are the result, as measured by disappearance of postsynaptic density protein 95 (PSD95), loss of the NR1 NMDA receptor subunit, and release of the cytoplasmic enzyme lactate dehydrogenase. Although this in vitro barotrauma model has been used previously (Shepard et al. 1991;Jayakumar et al. 2011), this is the first time that mild repetitive injury to cultured neurons is shown to reproduce the p-tau and p-TDP-43 hallmarks of human CTE. Furthermore, a possible role for astrocytes in the pathogenesis of CTE had not been previously explored. Expanding the utility of their in vitro model, Jayakumar et al. now injure astrocytes using the same fluid percussion device that is used to traumatize neurons. Injured neurons are then exposed to conditioned medium from damaged astrocytes—cleverly after different periods of astrocyte injury—to evaluate the effect of secreted astrocyte factors on neuronal proteinopathy and viability.

The first notable finding is that a single mild trauma is sufficient to cause accumulation of both p-tau and phosphorylated and ubiquitinated TDP-43. Thus, similar to the human disease, a single event can cause manifestations of CTE at the cellular level, while repetitive trauma increases severity. Importantly, whereas the sequence of proteinopathies is hard to untangle from post-mortem brain tissue, data from the in vitro model show that cytoplasmic p-TDP-43 accumulation precedes hyperphosphorylated tau (Fig. 1a). This is noteworthy because although p-tau is known to inhibit proteasomal degradation, the sequence of events argues against a mechanism by which hyperphosphorylated tau causes ubiquintated p-TDP-43 accumulation due to proteasomal inhibition. Lack of effect of the proteasome inhibitor MG132 on cytoplasmic p-TDP-43 accumulation further rules out this mechanism and suggests that p-TDP-43 is upstream of p-tau in the toxic chain of events leading to CTE. Having made a case against the lead suspect tau as the earliest instigator of CTE, the authors go on to establish probable cause mechanisms by which ubiquitinated p-TDP-43 accrues in the cytosol.

Fig 1.

(a) A putative time course of traumatic events in chronic traumatic encephalopathy (CTE) based on the in vitro model of Jayakumar et al. (2016). Following trauma, an initial increase in astrocyte thrombospondin-1 (TSP-1) synthesis and release protects neurons from toxic proteinopathy. As damaged astrocytes decrease TSP-1 secretion to neurons over time, hyperphosphorylated transactivating DNA-binding protein-43 (p-TDP-43) begins to accumulate in the neuronal cytoplasm (green line) and synaptic integrity (blue shading) begins to decline. Accumulation of hyperphosphorylated tau (p-tau, red line) is observed later in the sequence of traumatic events, subsequent to the start of cytoplasmic p-TDP-43 accumulation. (b) Schematic depiction of the trauma-induced TDP-43 pathogenic cascade regulated by astrocyte-secreted TSP-1. Early after trauma, astrocytes increase their synthesis and secretion of TSP-1, a guardian of synaptic integrity that antagonizes the trauma-induced upregulation of neuronal casein kinase-1epsilon (CK1ε), preventing TDP-43 phosphorylation (left panel). Over time, damaged astrocytes lose the ability to synthesize and secrete TSP-1, resulting in increased CK1ε expression and activity (middle panel). A decrease in the expression of importin-β, a carrier protein that translocates phosphorylated TDP-43 (p-TDP-43) to the nucleus, is also observed when astrocyte TSP-1 is absent. Consequently, p-TDP-43 accumulates in the cytoplasm due to elevated CK1ε and low importin β levels. Cytoplasmic p-TDP-43 buildup leads to its ubiquitination, a hallmark of CTE TDP-43 proteinopathy, and precedes a reduction in synaptic proteins PSD95 and the NR1 subunit of the NMDA receptor (NMDA-nr1, right panel) that indicates a loss of synaptic integrity. TDP-43 pathogenesis appears to start prior to and independent of p-tau accumulation, another CTE hallmark observed in this in vitro trauma model.

After phosphorylation, TDP-43 is normally translocated to the nucleus by the carrier protein importin-β (Shindo et al. 2013). However, Jayakumar et al. (2016) observe that cytoplasmic p-TDP-43 increases after trauma without a change in nuclear p-TDP-43. This finding raises the possibility of impaired p-TDP-43 nuclear import instead of (or in addition to) impaired turnover as the cause of the progressive increase in cytoplasmic p-TDP-43. An association between p-TDP-43 accumulation in the cytoplasm and decreased importin-β protein level was previously observed after ischemic insult (Shindo et al. 2013), further implicating importin-β as a possible player in the p-TDP-43 cytoplasmic rise. Consistent with the observations following ischemia, Jayakumar and colleagues demonstrate a decrease in importin-β protein in their in vitro CTE model. Importantly, importin-β knockdown by siRNA is sufficient to cause the cytoplasmic buildup of p-TDP-43 and loss of synaptic integrity in neurons even without a trauma stimulus, suggesting a causal relationship between loss of importin-β and neurotoxic p-TDP-43 accumulation (Fig. 1b). It is not yet known why importin-β levels decrease after a traumatic event or how this may affect the other cellular functions it regulates. Experimentally preventing the decrease in neuronal importin-β following trauma, either by overexpression or inhibition of proteolytic degradation, would provide additional evidence that failure of importin-β-mediated nuclear import is one of the principle pathways leading to TDP-43 proteinopathy in CTE. The findings in this work highlight the need for more research to elucidate the mechanisms that regulate the function of the aptly named importin-β.

Although identifying the immediate cause of cytoplasmic p-TDP-43 aggregation in CTE is crucial, the discovery of upstream regulators is likely to yield additional, and perhaps more accessible, opportunities for therapeutic intervention. The casein kinase 1 (CK1) family of protein kinases can be activated by oxidative stress, which is also linked to TDP-43 proteinopathy. Jayakumar and colleagues observe an increase in both CK1ε and c-Jun N-terminal kinase (JNK) 1/2 levels in neurons after traumatic injury. However, a CK1ε inhibitor or treatment with the antioxidant N-acetyl cysteine but not a JNK1/2 inhibitor reduces the buildup of p-TDP-43 and significantly rescues synaptic integrity.

Impressively, the work goes on to provide a convincing mechanism by which astrocyte-to-neuron communication may impact neuronal TDP-43 proteinopathy and ultimately neuronal integrity. This was done by exposing neurons to the conditioned medium of astrocytes separately subjected to the same trauma stimulus as the neurons. Medium from astrocytes collected 3 days post-injury (“early stage trauma”) prevented CK1ε elevation, TDP-43 phosphorylation, importin-β decline, and loss of synaptic proteins whereas medium from astrocytes collected at 15 days post-injury failed to prevent this pathological sequence of events (Fig. 1b).

Thrombospondin-1 (TSP-1) is a neuroprotective factor released by astrocytes following trauma (Tran and Neary 2006), making it a prime candidate for mediating the protection afforded by the conditioned medium. Consistent with this possibility, Jayakumar and colleagues find that TSP-1 levels are initially upregulated in astrocytes at 1–5 days post-injury, but then fall below baseline by 10 and 15 days post-injury (Fig. 1a). Elegantly, they show that TSP-1 immunodepletion abrogates the protective effect of the conditioned medium from the “early”-injured astrocytes, while astrocyte TSP-1 overexpression restores a neuroprotective effect of conditioned medium from the “15 day”-injured astrocytes, along with TSP-1 protein levels. In addition, recombinant TSP-1 mimics the neuroprotective effect of conditioned medium, further supporting the hypothesis that astrocyte-secreted TSP-1 within the conditioned medium is solely responsible for inhibiting the p-TDP-43 cascade in traumatized neurons. Although neuroprotective effects of TSP-1 have been previously documented, this paper by Jayakumar and co-workers is the first to show that TSP-1 acts to impair the cytoplasmic buildup of phosphorylated TDP-43. Intriguingly, while the attenuation of the trauma-induced CK1ε increase by TSP-1 can explain the suppression of TDP-43 phosphorylation, the decline in importin-β was also prevented by TSP-1. Thus, a further analysis of TSP-1 signaling cascades in neurons may provide clues as to the mysterious disappearance of importin-β following trauma.

Overall, one of the most unique aspects of this study is the investigation of the effects of trauma on both astrocytes and neurons, and the interaction between the two cell types. An implication of the work is that following trauma, astrocytes lose their ability to support neuronal integrity over time, and this astrocyte failure contributes to the development of TDP-43 proteinopathy in CTE. While the separate traumatizing of astrocytes and neurons in the study helped parse out the effect of astrocyte-secreted factors, additional in vitro models might be developed that take into account cell-to-cell contact and also reverse signaling from neurons to astrocytes. It was reported that purinergic signaling induces TSP-1 expression in astrocytes (Tran and Neary 2006). Therefore, as neurons begin to die and release ATP, astrocyte TSP-1 expression and secretion may be reinvigorated.

This study is also important for its translational implications. Therapeutic strategies focused on direct neuroprotection have had little success. Identifying the mechanisms underlying the defect in astrocyte synthesis and release of TSP-1 following trauma, and finding ways to recover this endogenous protective mechanism may yield new avenues for neuroprotection. Interestingly, Jayakumar et al. (2016) show that metformin, the first-line drug for treating type 2 diabetes, is able to increase astrocytic expression of TSP-1 and ameliorate the neuronal trauma-induced p-TDP-43 cascade. Thus, this paper highlights and reinforces the need for more research into not only the study of neurons in CTE, but also the integrity and function of surrounding glia.

Finally, what actually causes the loss of synaptic integrity in CTE? Although this paper strongly implicates accretion of cytoplasmic ubiquinated p-TDP-43 in the ultimate neurotoxicity following in vitro trauma, a causative role and mechanism of action remain to be established. Intriguingly, it was recently found that neuronal TDP-43 becomes localized to mitochondria in subjects with ALS and frontotemporal dementia (Wang et al. 2016). It is easy to imagine that inhibition of TDP-43 nuclear import following trauma-induced loss of importin-β leads to an increased association of TDP-43 with mitochondria in the cytoplasm. Recent evidence suggests that TDP-43 toxicity may be due at least in part to disrupting the expression and assembly of mitochondrial Complex I subunits (Wang et al. 2016), much like pathogenic α-synuclein is thought to impair Complex I in Parkinson’s disease (Di et al. 2016). As research into the toxic mechanisms of TDP-43 and tau protein aggregates accelerates, it will be interesting to see whether CTE joins the large list of neurodegenerative pathologies rooted in mitochondrial dysfunction.

Abbreviations

CK1ε

casein kinase 1 episilon

CTE

chronic traumatic encephalopathy

JNK

c-Jun N-terminal kinase

PSD95

postsynaptic density protein 95

TDP-43

transactivating DNA-binding protein-43

TSP-1

thrombospondin-1