August's ALS Research Roundup

A Microfluidic Model of AQP4 Polarization Dynamics and Fluid Transport in the Healthy and Inflamed Human Brain: The First Step Towards Glymphatics-on-a-Chip

Soden et al., 2022

As one may expect, it is extremely difficult to study a living brain. Even post-mortem brains are rare, and these still fail to show the full mechanics of the most complex living organ human have yet encountered. Various hormones, molecules and fluids move into and out of the brain, with specific mechanisms required to avoid the blood-brain barrier which separates the central and peripheral nervous systems. When considering neurodegenerative disease research the problems here are compounded, with both a small number of patients capable of donating and often unique changes to the brain and its support systems on not just a disease, but a patient level. One pathway to studying a semblance of such a living system is the use of neuromorphic chips, in which circuits are used to mimic the biological architecture of the brain. In theory these may one day be able to fully model the brain, but as of the current day it is difficult to even model individual pathways. A ‘perfect’ neuromorphic chip would be an irreplaceable tool to not only understand how the brain works, but also to model various diseases and virtually test treatments.

This study is a stepping stone towards modelling one particular part of the brain: the glymphatic system. The glymphatic system is a lymph-like system which uses glial cells (glymphatic = glia + lymphatic) to allow the flow of harmful and waste materials from the blood vessels in the brain to the cerebrospinal fluid (CSF) which surrounds the brain and spinal cord to the interstitial fluid surrounding cells, and then back out into the periphery. The transition between fluids is not completely characterised, but appears to involve the actions of astrocytes, glial cells with a supporting role, and AQP4, a water channel protein that is found at the endfeet of these astrocytes. This study sought to model the dynamics underlying AQP4 movement and fluid transport, with the end goal of creating a chip-based model of the glymphatic system. Understanding this process may play a key role in ALS research, as one of the materials cleared via the glymphatic system is the array of misfolded proteins which are believed to drive the initiation and spread of ALS and similar diseases. The astrocytes and AQP4 have also been shown to be affected in many ALS cases, depolarising AQP4 (moving it away from the endfeet where it is required to function) and preventing proper glymphatic flow.

The actual content of the study was the creation of a “gliovascular unit (GVU)-on-a-chip” consisting of a network of human astrocytes seeded around a pair of microchannels, one a bioengineered blood vessel seeded with human brain microvascular endothelial cells (HBMECs). The whole structure is contained within a 2.7cm/2cm enclosure, with numerous microchannels feeding into a 6mm fluid reservoir. Through this the fluid drainage efficiency and AQP4 polarisation was able to be monitored and manipulated, and the GVU-on-a-chip was exposed to various stressors and drivers of pathology. These included direct AQP4 inhibitors, lipopolysaccharides (LPS; bacterial toxins which induce inflammation) and oligomeric Aβ42, a pathological misfolded protein involved in Alzheimer’s disease with a similar mechanism to many ALS-driving proteins. By understanding how healthy glymphatic flow works alongside various forms of dysfunction, they hoped to better model how it functions both regularly and in disease conditions.

When implemented, the astrocytes spontaneously migrated to the bioengineered blood vessels and extended AQP4-containing endfeet, mimicking processes seen in living cortical blood vessels. The GVU-on-a-chip was able to drain fluid faster than a model that was not seeded with cells, modelling the astrocyte/AQP4-dependent glymphatic flow. ‘Unhealthy’ chips subjected to inflammatory LPS and Aβ42 oligomers had this flow slowed, to the point that the flow rate was near-identical to inflammation-affected chips with only dead cells. Given the presence of neuroinflammation in many neurodegenerative diseases including ALS, this could be a strong model for at least part of how these diseases interrupt clearance of proteins from the CSF. This was shown to be the result of AQP4 depolarisation, with inflammation resulting in dispersal of AQP4 from the endfeet where they are needed for glymphatic function. Interestingly, only chronic and not acute inflammation were able to achieve this effect, reinforcing the connection to chronic neurodegenerative disorders such as ALS.?

The glymphatic system is quite difficult to study within the brain, not only because of the difficult in accessing the brain but also because it is a widely dispersed system bordering numerous tiny blood vessels which feed the brain. As a result, there has been a great deal of conflicting research and opinions around exactly how glymphatic flow works, and just how large a role it plays. This self-contained model with living cells appears to recapitulate many features, although it doesn’t fully model the segregated compartments of a ‘true’ living system. However, the way in which the cells themselves migrated to ‘construct’ the system was intriguing, and promising for the development of such models. These kind of semi-contained biological models are the first step to creating truly representative laboratory models, which may form the core of essential disease research in the future.

TAGS: model, astrocytes, glymphatic, inflammation, non-ALS, human_cell

Deep learning methods to predict amyotrophic lateral sclerosis disease progression

(Pancotti et al., 2022)

A tremendous amount of genomic, clinical and demographic data has been gathered from ALS patients across the world, with the largest database, PRO-ACT, carrying over 11,600 patient records. However, this quantity of data is useless unless we have ways to properly utilise it. Unfortunately, the human brain isn’t equipped to work on that sort of scale, and it is nearly impossible for researchers to draw out broadly representative data. Fortunately, recent years have seen a tremendous development in machine learning and data analytical methods. One result of this is the development of deep learning models, a type of machine learning/artificial intelligence model with a huge capacity for feature detection and classification from raw data.

This study sought to investigate how deep learning models perform in predicting ALS progression using the PRO-ACT data. Three models were created: feed-forward (data only flows in one direction), convolutional (mainly used for image/object recognition and classification) and recurrent (trained to recognise patterns across time using internal memory). These were used to predict decline in the ALS functional rating score (ALSFRS) based on various forms of data collected in PRO-ACT. The performance of the deep learning model was compared to two “state-of-the-art” models developed using Random Forest Regressor (RF) and a Bayesian Additive Regression Trees (BART) methods.

Unfortunately, the deep learning models used were not definitively better than the pre-existing models which they were compared against. It had a slight advantage in one metric (root mean squared error (RMSD)), but a slight disadvantage in another (Pearson correlation coefficient (PCC)). This was the same for each of the deep learning models. The authors proposed that this may represent a possible ‘upper limit’ to the precision with which we can predict ALSFRS, although it is possible that superior methodologies will emerge to disprove this. It is worth noting that these models, while not clearly superior to prior ones, were still quite effective at predicting clinical severity and distinguishing fast progressing ALS with worse survival from slower progressors.

TAGS: predictive, machine_learning, human?

TMEM106B Acts as a Modifier of Cognitive and Motor Functions in Amyotrophic Lateral Sclerosis

(Manini et al., 2022)

One of the few consistent facts which has emerged with regards to ALS is that it is inconsistent. A massive range of different genes, molecules and systemic effects have been found to contribute to the disease in different ways, most of which are still far from fully understood. One such gene is transmembrane protein 106B (TMEM106B). TMEM106B was first associated with frontotemporal dementia (FTD), a disease which shares several features with ALS but specifically targets cognitive rather than motor function. So closely are the two linked that it has been suggested that ALS and FTD lie along a shared disease spectrum of cognitive-motor symptoms, and a significant amount (~15-20%) of patients have a shared syndrome with both. TMEM106B has been linked to another shared feature: the presence of pathological aggregates containing TDP-43, an RNA-binding protein and one of the most consistent ALS features. However, while the gene has been previously studied with regards to FTD, it has not been studied in ALS.

This study investigated the effects of two genetic variants (alleles) at a specific genetic location in TMEM106B. These were the common variant (major allele; A) and minor allele (rarer mutation; G). 865 ALS patients were assessed for these variants, as well as resultant demographic and clinical features.

Interestingly, variation in this one nucleotide resulted in significantly different manifestations of ALS. This variation indicates that this gene is likely to be disease-modifying rather than disease-causing, altering the way in which ALS symptoms manifest. Patients homozygous (both genetic strands having the same allele) for the major allele (AA) experienced higher levels of functional loss, an earlier age of onset and a higher risk of bulbar onset, which produces non-limb symptoms such as issues with speech and swallowing and typically comes with faster disease progression. This variant actually appeared protective against the deterioration of limb-associated lower motor neurons (LMN), although this could just mean that even a single G allele is damaging to LMNs. Homozygous GG patients had lower motor functional loss, but instead had higher levels of cognitive loss in terms of language, verbal fluency, executive function and memory.

Typically when a gene is associated with disease it is considered in terms of its ability to cause the disease. However, the variation in contributing genes in ALS nearly guarantees that a great many instead modify the way in which it manifests, and TMEM106B here is a perfect example of that. A single nucleotide base change shifts the rate of cognitive symptoms, upper v lower motor neuron pathology and the site of onset. While unlikely to help with finding a ‘cure’, collating such mutations may contribute greatly to our ability to categorise ALS cases and develop more targeted treatments, as well as predicting how an individual’s unique ALS course may develop.

TAGS: predictive, stratification, FTD, TMEM106B, human

Relationship between Dietary Total Antioxidant Capacity and the Prognosis of Amyotrophic Lateral Sclerosis

(Eom et al., 2022)

While it is tempting to rely on large-scale data analysis and cutting-edge drugs to address ALS, these are unlikely to solve the problem tomorrow, or even in the next year. As a result, lifestyle changes which slow the progression of ALS can be an important means of minimising the harm done to patients in the meantime. This study investigated the effects of different families of foods which contain antioxidants, substances which can prevent or delay certain types of cell damage. In terms of slowing ALS pathology these are most likely to target reactive oxygen species (ROS), toxic molecules which are overproduced in the course of ALS and which cause damage to cells and DNA.

In contrast to previous studies which investigated specific antioxidants, this one targeted the dietary total antioxidant capacity (DTAC) of different food groups. Higher DTAC has previously been associated with a lower risk of several diseases including stroke, metabolic syndrome, cardiovascular disease and several cancers. This was a retrospective study incorporating data from 301 ALS patients over 10 years, primarily assessing disease progression and survival time. ?

The largest benefit observed was in vegetables and legumes, where higher DTAC was associated with slower disease progression and longer survival in ALS patients. This mirrors earlier findings that vegetable consumption is associated with slower disease progression in several neurodegenerative diseases including Parkinson’s and multiple sclerosis as well as protecting cells from ROS damage in ALS mice. Legumes are known to be sources of vitamin E, which has previously been associated with longer survival and slower progression in ALS patients, suggesting a possible mechanism of action. Total DTAC and fruit DTAC did not have a strong enough association to reach statistical significance. This may be due to the way in which fruit juices are processed which removes many flavonoid antioxidants, as well as sugar increasing ROS generation in general.

While studies such as this seem simple in nature, lifestyle changes are a way for patients to both benefit their own clinical condition and to contribute to their own treatment. However, one of the most persistent psychological issues in ALS can be the lack of means to help their own treatment. Having some means of acting can provide benefits beyond physical health, assisting people in regaining some control over their own body.

TAGS: lifestyle, ROS, human?

Frequency of Parkinson’s Disease Genes and Role of PARK2 in Amyotrophic Lateral Sclerosis: An NGS Study

(Vacchiano et al., 2022)

While the umbrella term of ‘neurodegenerative diseases’ groups together a variety of different conditions, they are typically explored as completely separate. However, there is a not insignificant body of research into the common features between them. This ranges from similar types of functional loss to shared mechanisms of action, such as the prion-like protein pathologies in Alzheimer’s, Huntington’s and Parkinson’s diseases and ALS. There are further commonalities between the latter pair, with Parkinson’s and ALS patient sharing a prevalence of Lewy bodies (abnormal deposits of the protein α-synuclein). Furthermore, Parkinsons-like features were observed in approximately 30% of ALS patients and genetic analyses have already found some connections between the two.

This study explored the frequency of 13 key Parkinsons-associated genes in 130 ALS patients compared to 100 Alzheimer’s patients and over 1600 healthy controls. Overall, Parkinson’s-related genes were found to be altered in 26.2% of ALS, 20% of Alzheimer’s patients and 19.2% of healthy controls. However, autosomal recessive genes were significantly more involved in ALS compared to the other groups. These are gene variants which require the same variant to be inherited from both parents and so are both rarer than most mutations and often linked to genetic diseases. However, these variants were not found to modify the ALS cases’ clinical symptoms. They highlighted increased levels of a specific mutation in parkin 2, a gene which is likely the cause of an autosomal recessive variant of Parkinson’s. This is consistent with prior literature, which found reduced parkin levels in ALS animal models and patients. It was also connected to several other neurodegenerative diseases, particularly those linked to decreased levels of the protein TDP-43 such as ALS and frontotemporal dementia (FTD). Given that TDP-43 aggregates are present in nearly all ALS cases, even those without mutations in the gene which produces it, this protein is of a high level of interest in ALS. However, the direction of effect is not clear, so it could be that TDP-43 depletion results in lower levels of parkin or vice versa.

While this study did not demonstrate a direct connection between ALS and Parkinson’s, it does present a possible mechanism of interaction between the two. If loss of parkin interferes with TDP-43 in some way, then mutations which lead to Parkinson’s may create a better environment for the development of ALS through the avenue of TDP-43 pathology. There is a huge range of environmental and genetic features which appear to contribute to ALS, and while other diseases have been statistically linked before it is interesting to have a mechanistic pathway. Not only does this provide a possible target for risk factor screening in ALS, it may even be worthwhile to consider a family history of Parkinson’s as a factor to help with prediction, and thus pre-empting of ALS.

TAGS: risk_factors, mechanistic, TDP-43, Parkinsons, human

A low-power stretchable neuromorphic nerve with proprioceptive feedback

(Lee et al., 2022)

It is easy to think that if we can one day figure out a ‘cure’ for ALS, things will go back to normal for people suffering from this debilitating disease. However, even if we are able to stop the progression and further damage, by the time we can even detect it a fair degree of damage has been done. As such, it can be important to pay attention to means by which we can help with recovery, or at least rehabilitation once further damage has been prevented. Neuroprosthetics are an interesting focus within this research, connecting brain signals to devices to help restore function. However, some have taken a different approach. They worked to create a stretchable neural implant which can carry brain signals to the muscles and so facilitate voluntary movement: not a prosthetic limb, but a prosthetic nerve. Given the loss of motor neurons in ALS, this could be a remarkable means of recovering lost function, bypassing or replacing the lost neurons.

This study tested their neuromorphic nerve in the legs of mice, assessing various functions. They showed their ability to carry motor control signals to the limbs, interfacing with different muscles and inducing weak and strong muscle contractions. They were able to do so with enough precision to drive walking as well as kicking of a ball. It is also important to note that they were able to do so with gradual increases in muscle force, compared to earlier systems which had more of an ‘all or nothing’ approach which could result in abrupt, painful movements. Another advancement was the ability to mimic proprioception, the body’s ability to sense its own movement, action and location. They implemented a method mimicking the biological stretch reflex, which is able to detect leg movement and prevent overstretching by limiting muscle activation in real time. This resulted in more stable movement, as well as reducing strain on the muscle. The resistance produced by this strain sensor also reduced power consumption. This, combined with the use of organic components rather than traditional machines, resulted in two orders of magnitude less power use, as well as being able to function for a longer period of time (approximately 70 days).

The most consistent and debilitating component of ALS pathology is the loss of motor neurons, rather than degeneration of the muscles themselves. This type of technology could represent a means of bypassing these lost neurons, helping to recover lost motor function before the muscles themselves atrophy from underuse. The biological nature of these neuromorphic nerves means that they are likely to be less physically invasive, and their size and power consumption is reduced compared to earlier nerve control devices. It may be that these, or more likely further technologies derived from them, will become a common means of stabilising ALS patients and helping to maintain their control and quality of life.

TAGS: device, mouse

Shared mechanisms across the major psychiatric and neurodegenerative disease

(Wingo et al., 2022)

While disorders which affect the brain are generally considered distinct, some research has revealed some unexpected connections between them. Studies have indicated that psychiatric conditions can quadruple the risk of developing dementia or neurodegenerative diseases later in life, while around 65% of people with neurodegenerative diseases develop some sort of psychiatric symptoms.

This study assessed a range of different psychiatric disorders (major depressive disorder, bipolar disorder, schizophrenia, anxiety disorders, post-traumatic stress disorder, problematic alcohol use) and neurodegenerative diseases (Alzheimer’s disease, Lewy body dementia, Parkinson’s disease, amyotrophic lateral sclerosis, and frontotemporal dementia). They performed structural analysis of the brain, genome-wide association studies (GWAS) which identify genes associated with a particular disease/trait, and LD score regression which estimates genetic correlation between traits. Groups of causal proteins were established for psychiatric disorders, neurodegenerative diseases and both.

394 proteins were linked to psychiatric illnesses, and 31 to neurodegenerative diseases. A collection of 13 proteins were identified in both groups. Five of these had previously been linked to one or more of these disorders, typically Alzheimer’s disease. These covered a range of functions including synaptic transmission, immune function, mitochondrial function and vesicle-mediated intercellular communication through the SNARE complex. Mitochondrial dysfunction and consequently issues with synaptic transmission are frequently observed in several of the conditions assessed here, and so are of particular interest. It is also worth noting that mitochondrial dysfunction often appears early in disease processes, implicating it further in either causative or at least early disease stages. Physical interactions between these proteins were also studied, finding that there were 2.6-fold more interactions between the psychiatric protein and neurogenerative proteins than were expected by chance. The possibility of more physiological interactions between the proteins driving these disorders could explain their shared incidence, although this does depend on the interactions themselves contributing to the disorders.

While the possibility of founding a ‘unified theory of neurological disorders’ sounds tempting, it is a difficult principle to properly support. While this study had a fairly large sample size, with over 700 patients’ data used, the large range of disorders and variation within and between them likely means that some were disproportionately represented. With that being said, the genes which were identified across the range of disorders may be valuable targets for future research. Even if they are not causative in any way, they may contribute to producing a biological environment which promotes the development of one or more of these disorders. The high level of physical interactions between psychiatric- and neurodegenerative-linked protein was also of interest. It could be that a few highly reactive proteins were picked up by both, or there could be shared pathways activated across disorders. If the former is true, these types of broadly reactive, disease-associated proteins could be considered as possible contributors to, or at least risk factors in various disorders. ?

TAGS: risk_factors, human

References

Eom, J., Son, B., Kim, S.H., and Park, Y. (2022). Relationship between Dietary Total Antioxidant Capacity and the Prognosis of Amyotrophic Lateral Sclerosis. Nutrients 14, 3264.

Lee, Y., Liu, Y., Seo, D.-G., Oh, J.Y., Kim, Y., Li, J., Kang, J., Kim, J., Mun, J., Foudeh, A.M., Bao, Z., and Lee, T.-W. (2022). A low-power stretchable neuromorphic nerve with proprioceptive feedback. Nature Biomedical Engineering.

Manini, A., Ratti, A., Brusati, A., Maranzano, A., Fogh, I., Peverelli, S., Messina, S., Gentilini, D., Verde, F., Poletti, B., Morelli, C., Silani, V., and Ticozzi, N. (2022). TMEM106B Acts as a Modifier of Cognitive and Motor Functions in Amyotrophic Lateral Sclerosis. International Journal of Molecular Sciences 23.

Pancotti, C., Birolo, G., Rollo, C., Sanavia, T., Di Camillo, B., Manera, U., Chiò, A., and Fariselli, P. (2022). Deep learning methods to predict amyotrophic lateral sclerosis disease progression. Scientific Reports 12, 13738.

Soden, P.A., Henderson, A.R., and Lee, E. (2022). A Microfluidic Model of AQP4 Polarization Dynamics and Fluid Transport in the Healthy and Inflamed Human Brain: The First Step Towards Glymphatics-on-a-Chip. Adv Biol (Weinh), e2200027.

Vacchiano, V., Bartoletti-Stella, A., Rizzo, G., Avoni, P., Parchi, P., Salvi, F., Liguori, R., and Capellari, S. (2022). Frequency of Parkinson's Disease Genes and Role of PARK2 in Amyotrophic Lateral Sclerosis: An NGS Study. Genes (Basel) 13.

Wingo, T.S., Liu, Y., Gerasimov, E.S., Vattathil, S.M., Wynne, M.E., Liu, J., Lori, A., Faundez, V., Bennett, D.A., Seyfried, N.T., Levey, A.I., and Wingo, A.P. (2022). Shared mechanisms across the major psychiatric and neurodegenerative diseases. Nature Communications 13, 4314.