Cellular Stomachaches and the Commonalities of Neurodegenerative Proteinopathies

Cellular Stomachaches and the Commonalities of Neurodegenerative Proteinopathies

Huntington's, Alzheimer's and Parkinson's disease all involve proteins that accumulate or behave abnormally, alongside cellular dysfunction and eventual cell death. But as I worked through lectures on these diseases, what interested me was the different levels of certainty in the explanations. With Huntington's, we can begin with a defined genetic change and follow an increasingly concrete causal chain. With Alzheimer's, the evidence points strongly toward amyloid beta as an upstream contributor, while much of what connects it to dementia remains complicated. And with Parkinson's, even the question of whether patients share the same initiating process becomes difficult. [1–5]

The shared category is neurodegenerative proteinopathies, diseases in which abnormal protein behavior appears to be central to the pathology. This gives us a useful way to compare them, although the category alone tells us little about why a particular protein becomes troublesome in a particular cell. A protein might have an altered sequence, be produced in excess, take an abnormal shape or accumulate because the cell cannot clear it adequately. These possibilities place the beginning of the problem in different parts of cellular life. [1,3,5]

Susan Lindquist's introductory lecture makes the physical problem easier to picture. Proteins operate in a crowded, moving environment, where a newly made chain has to acquire a useful shape while surrounded by other molecules. Cells employ chaperone proteins to help with that process and have systems that remove proteins when things go wrong. Protein synthesis can seem almost miraculous given those conditions, but its success depends on a process of continuous assistance and maintenance. The genetic sequence only supplies part of the explanation, and a broader cellular context supplies the conditions in which the protein has to be assembled in and function within. ^[7]

So the question to explore is how much of that beginning we actually understand, and what kind of experiment would let us understand more. We can identify a protein in a clump, observe a damaged neuron nearby and still have questions about which event came first. Those questions become especially important when the proposed treatment removes something that healthy cells also use. [1,2,5]

Huntington's is the most concrete example of a shared genetic starting point in this comparison. The disease begins with an expanded stretch of CAG repeats in the HTT gene, which lengthens the glutamine sequence in huntingtin, a protein that healthy cells also make. The repeat can then expand further within individual cells over a person's lifetime. As I covered in A Consensus on Precision Therapy, the emerging account connects this somatic expansion to increased production of HTT1a, a short RNA that produces an aggregation-prone exon-1 huntingtin protein. This is a useful distinction because HTT1a is the instruction for making the short protein, while full-length mutant huntingtin comes from a longer RNA. Gillian Bates develops this account directly in her Royal Society lecture, including experiments that distinguish the effects of lowering different huntingtin products. [2,8]

But even here, where the genetic cause is well defined, we still have to establish the relative contributions of the short protein, full-length mutant huntingtin and the cellular pathways that lead to cell death. We also still need to explain why certain cells, particularly the medium spiny neurons of the striatum, are so vulnerable. Knowing where the causal chain begins gives treatment a clearer direction, but the failed and incomplete interventions discussed in that earlier article show how much precision the intervening biology demands. We may know what genetic change to blame while remaining uncertain about which protein form to lower, how much normal function to preserve and the time frame when intervention would be most effective. [2,8]

Descriptions of Alzheimer's pathogenesis generally focus on both amyloid beta and tau, with the evidence placing them at different stages of a process that unfolds across several kinds of cells. The strongest case for amyloid as an upstream contributor comes from combining genetic findings with experiments that test its effects on the subsequent pathology. [3,4]

First some terminology. Amyloid describes a general form of protein assembly, while amyloid beta, usually abbreviated Aβ, names a specific peptide involved in Alzheimer's. Aβ is cut from a larger protein called amyloid precursor protein, or APP. So while all these neurodegenerative diseases may have some amyloid-forming protein implicated, it is important to understand that huntingtin, alpha-synuclein and Aβ have different identities and ordinary biological contexts. In Alzheimer's, Aβ accumulates in plaques, while abnormal tau forms tangles inside neurons. [1–4]

Diving into the mechanics of the disease, in "Alzheimer's and Other Neurodegenerative Diseases," John Hardy begins his account with evidence from genetics, including rare families in which duplication of APP causes disease. A duplication supplies an extra copy of the gene, giving the person three copies where there would ordinarily be two, which explains the roughly 50% increase in gene dosage he discusses. That additional dosage can drive excess production, linking a specific genetic change to an increased supply of the protein. ^[3]

The significance is that even the ordinary protein can become troublesome when its amount increases. Other inherited forms involve mutations in APP or in the presenilin components of the machinery that cuts it. Bart De Strooper explains how altered processing can release longer Aβ peptides that are more prone to aggregate. The relevant change can therefore concern the not just the quantity produced, but the properties of the peptide or the mixture of peptide lengths, all without requiring a general failure of protein synthesis. [3,4]

But most Alzheimer's requires an explanation beyond those rare inherited mutations. Hardy's discussion of late-onset disease turns toward the balance between production and removal, and toward the cells that respond to accumulating material. If a protein can accumulate because too much is made, impaired handling or removal can also push that balance in an unfavorable direction. Genetic associations involving APOE, TREM2 and other genes then bring lipid handling and microglia, the brain's resident immune cells, into that account. Those associations identify relevant cell biology while leaving the work to establish exactly how it contributes in each setting to future experimentation. [3,4]

The connection to lipid biology maybe particularly relevant, as Hardy makes it more central by considering where Aβ is produced. APP sits in a membrane, and the processing that releases Aβ includes cutting within that membrane, while Aβ itself has an affinity for lipids. Hardy therefore proposes that deposition may begin in a membrane, disturb it and bring in APOE and microglia as part of the response to the disturbance. On this account, membrane stress helps connect amyloid production with the lipid and immune biology implicated by genetic studies. ^[3]

Amyloid also accumulates along brain blood vessels, making its location relevant to treatment as well. Hardy discusses ARIA, amyloid-related imaging abnormalities, among the complications of antibody treatment, placing the tissue response to amyloid removal alongside the benefit of reducing its burden. The vascular complications and Hardy's membrane hypothesis both draw attention to the physical setting in which amyloid accumulates and is cleared. However, this remains largely speculative, as the initial membrane involved in his proposal remains unidentified, and any disease progression model must also account for Aβ that is secreted from cells. ^[3]

The cellular response becomes more complicated still once we ask how amyloid leads to neuronal damage. Microglia can help contain or clear problematic material, and their response can change with disease stage and local conditions. Astrocytes, another kind of glial cell, also respond around plaques and interact with microglia and neurons. De Strooper uses spatial measurements of gene activity to examine these neighborhoods, showing how the response involves different cell populations and changes over time. We therefore need to understand what those cells are doing together, including when a response helps and when it contributes to injury. [3,4]

Tau remains part of that problem. Hardy describes experiments combining amyloid pathology with tau pathology in mice, where the combination greatly increased tangles without a corresponding increase in amyloid. That result supports an upstream role for amyloid in those models. De Strooper describes human neurons transplanted into mouse brains that develop tau pathology and then identifies signs of a regulated cell-death process when exposed to an amyloid-bearing environment. Together, these experiments narrow the possible explanations, while leaving the precise relationships among amyloid, glial responses, tau and cell death open to further testing. [3,4]

Hardy's own earlier model placed amyloid accumulation before tau pathology and neuronal death. He had hoped that genetics would then reveal the intervening steps as clearly as biochemists had mapped glycolysis, the sequence of reactions cells use to break down glucose. But Alzheimer's unfolds over decades, with different cells responding and changing along the way. The mechanism that helps initiate amyloid accumulation may therefore differ from the mechanisms that sustain later decline. Once tau pathology or a damaging cellular response has developed, removing an earlier stimulus might leave part of that later process active. ^[3]

The anti-amyloid trials give us a way to test at least part of the account. In the lecanemab and donanemab results Hardy discusses, reducing amyloid was accompanied by slower average clinical decline than in the comparison groups. That is evidence that changing amyloid can change the disease's course in the patients and treatment settings studied. The improvement is partial, however, and slowing decline leaves a considerable distance to stopping degeneration or restoring functions that have already been lost. ^[3]

Treatment timing therefore matters, although even early intervention raises difficult questions. We need to know how early intervention must begin, how much reduction is sufficient and which downstream processes continue despite it. And while a long trial can show treated and untreated populations separating over time, amyloid reduction alone cannot tell us in advance how large that separation will become, nor its ultimate therapeutic accomplishments. Hardy also acknowledges uncertainty about how long an amyloid-positive person may remain without symptoms, which makes prevention a different evidential problem from treating an already symptomatic population. ^[3]

For common late-onset Alzheimer's, this leaves a less uniform initiating explanation than Huntington's. The genetic and intervention evidence gives us substantial reason to place amyloid upstream, while the progression from amyloid accumulation to dementia depends on cellular responses whose contributions remain incompletely understood. [3,4]

Parkinson's makes the distinction between a disease's clinical appearance and its underlying biology even more difficult. Loss of dopamine-producing neurons helps explain its movement symptoms, chorea, but the Parkinson's Foundation briefing by Duke researcher Dr. Sanders also emphasizes a huge breadth of sleep, gastrointestinal, cognitive and other non-motor problems. We are looking at a disease that can involve several systems, even when the diagnosis may typically occur due to the attention on movement disorders. ^[5]

The protein at the center of much of the current pathology research is alpha-synuclein, which healthy neurons use in functions associated with neurotransmitter release and the movement of small membrane-bound packages called vesicles. In the disease pathway, alpha-synuclein can misfold and assemble into larger structures, including the material associated with Lewy bodies, which these protein clumps found in cells of some patients. Smaller assemblies and fibrils may have different effects, and Sanders emphasizes continuing uncertainty about which forms are most harmful. Even the role of a large Lewy body itself can be questioned. Is the deposit itself damaging the neuron, or has the cell gathered harmful material into it to reduce that material's interactions with the rest of the cell? Sanders describes this as an unresolved part of the disease mechanism. ^[5]

There is causal evidence that alpha-synuclein can drive disease from genetic studies though. Mutations and extra copies of SNCA, the gene encoding alpha-synuclein, can cause inherited Parkinson's. Hardy's dosage argument therefore applies here as well, meaning that increasing the amount of otherwise normal alpha-synuclein can be enough to cause disease. And Lindquist describes experiments in which increasing alpha-synuclein in yeast cells causes increasingly severe dysfunction. When her team made yeast cells produce more alpha-synuclein, vesicles accumulated with the protein and stopped moving normally through the cell. Follow-up experiments found related transport defects carried over into other neuronal models. [1,3,5]

Virginia Lee supplies a more direct experiment connecting alpha-synuclein aggregation with later injury. In the work she describes, introducing preformed alpha-synuclein fibrils into otherwise ordinary mice initiated accumulation of abnormal alpha-synuclein, followed over time by loss of dopamine-producing neurons and movement deficits. The introduced material acted as a seed, encouraging the animal's own protein to adopt an abnormal form. So the result goes beyond finding protein deposits in a diseased brain, because her experimental intervention preceded a developing sequence of pathology and functional impairment. ^[9]

These experiments establish a route by which existing fibril seeds can propagate pathology, but leaves the formation of the first seed in a person as an open causal question. Lee also distinguishes forms, or strains, of abnormal alpha-synuclein that may differ in their effects. Material derived from human Lewy-body disease brains produced a different pattern of inclusions from laboratory-made fibrils in the models she describes. That makes the identity and behavior of the protein assembly part of what an experimental model of the disease pathology would need to consider. ^[9]

Further, the selected brain samples in Lee's work included Alzheimer's cases with Lewy-body pathology, showing that the protein categories can overlap within a patient. That overlap complicates a simple mapping of one diagnosis to one protein, while still leaving us to establish what each pathology contributes. Her experiments support mechanisms of seeding and propagation, and the distinction between those mechanisms and disease initiation helps explain how a field can know a great deal while still debating its root causes. ^[9]

To add to the complexity, a mutation in the alpha-synuclein gene and abnormal behavior of the alpha-synuclein protein are different observations. A person can have a typical gene sequence and still accumulate problematic forms of the protein. That was initially unintuitive to me, because it is tempting to imagine that correct genetic instructions guarantee a correctly functioning product. The protein's concentration, interactions, location and removal processes all matter however, even after those correct genetic instructions have been read. [1,5]

We also have evidence that the clinical condition can arise without the typical alpha-synuclein findings. In the Sander's discussion of LRRK2-associated Parkinson's, she describes patients who lack the expected synuclein aggregation or test results, as part of the open questions about other forms of Parkinson's expressions. She does caveat that this is only a subset of cases, and that a negative test by itself cannot guarantee that every relevant form of the protein is absent. Still, it complicates any claim that all patients must pass through the same alpha-synuclein-driven sequence. Conversely, synuclein pathology can precede recognizable clinical symptoms, leaving another gap between a molecular finding and the condition a clinician diagnoses. ^[5]

Sanders's discussion of mitochondria expands the possible starting points of the pathology. Alongside energy production, mitochondria participate in processing proteins and metabolites, regulating oxidation and coordinating signals that affect cell survival. When they become dysfunctional, a neuron can face both an energy shortfall and increased oxidative stress, which can damage cellular components. Damaged mitochondria themselves have to be removed through mitophagy, the selective clearance of mitochondria, so failure of maintenance can compound the original problem. ^[5]

Mitochondrial abnormalities appear in both monogenic Parkinson's, associated with a disease-causing change in a single gene, and idiopathic Parkinson's, where no specific cause has been identified. This gives mitochondrial function a role in the causal investigation beyond being a consequence of visible protein deposits. Sanders leaves open whether mitochondrial dysfunction initiates other disturbances or follows them, and whether that ordering differs between patients. ^[5]

Inflammation can then extend the disturbance across cells. Sanders describes microglia responding to alpha-synuclein aggregates and injured neurons, with sustained activation capable of adding further stress to the tissue. In this sequence, inflammation begins as a response to damage and becomes part of what perpetuates it. The resulting feedback helps explain why identifying the initial disturbance versus identifying the processes that sustain disease can lead to different therapeutic targets. ^[5]

Parkinson's therefore offers the least uniform causal story in this comparison. Some inherited forms have identifiable genetic causes, and experimental work connects particular disturbances with particular cellular effects. The unresolved problem is how those findings explain the broader population, including which process starts the pathogenic expression in each case and which processes become necessary to sustain it. That is a more demanding question than whether alpha-synuclein can cause harm. [1,5]

Across these diseases, I suspect that more of the explanation lies in the cellular environment than a simple chain from gene to protein aggregate allows us to see. By environment, I mean the conditions inside and around a cell, including its energy supply, membranes, clearance capacity and interactions with neighboring cells. It seems feasible that cellular metabolic dysfunction can itself help initiate protein pathology. A cell's ability to obtain energy, process materials and renew its components may deteriorate until proteins that it previously managed become a source of building, accumulating damage.

Cellular metabolism includes the building and breakdown of cellular material, and the digestive aspect of this maintenance is especially relevant. Lysosomes are compartments in which enzymes break down material delivered for disposal and recycling. Autophagy delivers portions of the cell's own contents to this system, including damaged proteins and organelles. Proteostasis, the maintenance of a functioning population of proteins, depends on this clearance alongside protein production, folding and repair. Mitochondrial function and protein maintenance therefore meet within the same continuing work of keeping a cell alive. [5,7,10]

Steve Finkbeiner's work on the dynamics of proteostasis provides a concrete example of a defect in cellular handling preceding the accumulation of an otherwise normal protein. In his discussion of frontotemporal dementia, he describes reduced progranulin, a protein encoded by GRN, and the accompanying impairment of lysosomes. In the neuronal experiments he presents, reducing progranulin slowed the turnover of TDP-43, a protein normally involved in processing RNA. TDP-43 persisted longer, accumulated outside its usual nuclear location and eventually formed deposits. In other words, the evidence suggests that TDP-43 accumulated because the cell’s clearance machinery was impaired, not that the protein itself had become a problem.^[10]

This gives the metabolic thesis a specific causal route to investigate. A defect in cellular processing can increase the time a protein remains in the cell and the burden that the cell must manage. Huntington's supplies a different entry point, with altered huntingtin placing an increased burden on that machinery. In both cases, the outcome depends partly on the relationship between the material a cell has to handle and its capacity to handle it, even if the initial disturbance occurs at a different point in the patient's life. [2,8,10]

Perhaps a cell has a bad day, part of an assembly line starts throwing off its work, and the disturbance becomes persistent. A temporary shortfall in energy or clearance might leave damaged components behind, which could further compromise the systems needed to remove them. This might lead to a cascading cellular metabolic dysfunction. The proposed cascade begins within the cell, but its consequences can extend to neighboring cells through inflammatory signals or, in some proteinopathies, the transfer of material that seeds further protein aggregation. [5,9,10]

Finkbeiner's experiments with progranulin-deficient cells illustrate why those neighbors matter. Neurons and microglia cultured separately showed much smaller changes in certain immune signals than cultures in which the cells were brought together. The interaction was particularly pronounced when the neurons came from patients with the progranulin-related disorder, including when the accompanying microglia came from control samples. He also reports greater neuronal death in these mixed cultures. The combination exposed a disturbance that studying either cell population alone would have understated. ^[10]

Long-lived neurons face particular demands in maintaining this balance. Most mature neurons do not divide, so they lack the opportunity to distribute accumulated material between daughter cells. Their extended processes also place distant regions of the cell far from much of the machinery that handles degradation. Finkbeiner describes differences in stress responses between neurons and astrocytes, and even among neurons from different regions. These observations could help explain selective cellular vulnerability, the fact that some kinds of cells deteriorate while others exposed to related stresses survive. [5,10]

A cell's prior experience can also change what it is able to withstand. In Lindquist's heat-conditioning experiments, cells exposed first to a milder stress were better able to survive a later, more severe exposure than cells given the severe exposure directly. The preparatory stress induced proteins that help other proteins retain or recover their function. Successful adaptation therefore belongs in the explanation alongside failure, particularly if we want to know why one cell recovers from a disturbance while another begins to deteriorate. ^[7]

Unfortunately, most of our measurement capability at the cellular level is like understand an economy from aggregate growth. We may see that wages are rising, unemployment is low and productivity is improving, yet a large part of the population may be under increasing strain. People who have financial assets that are appreciating could be doing well while others work harder to cover ordinary costs. And if the pressure eventually produces mass protest or systemic burnout, the aggregate indicators can leave us surprised because they failed to describe the lives in which the trouble was accumulating.

Similiarly in cell biology, a favorable average can coexist with a deteriorating subgroup, and the subgroup's importance may depend on how it connects to the rest of the system. A gene-expression signal averaged across a culture can conceal which cells generated it, while even a cell-type average can leave differences among individual cells unresolved. The experimental question is what distinguishes a cell approaching failure from a comparable cell that continues functioning, including the history that brought each of them to that state.

Longitudinal single-cell imaging gives that question a practical method. Finkbeiner describes robotic microscopes that repeatedly return to the same identified cells, with some experiments continuing for six months. The system records changes in individual cells and relates earlier measurements to later outcomes, including survival. A cell that is stable at one observation and deteriorating at the next can therefore be studied through its own sequence of changes, even when neighboring cells progress at different rates. ^[10]

Their Huntington's experiments show why this changes the interpretation of pathology. Finkbeiner's team introduced mutant huntingtin fragments into neurons and followed the formation of visible inclusion bodies. After accounting for other measured factors, cells that formed inclusions tended to survive longer, while higher levels of diffusely distributed mutant huntingtin predicted greater risk of death. The findings support the possibility that an inclusion can temporarily sequester harmful protein forms as part of a coping response. They also direct attention toward the protein outside the conspicuous deposit and toward what happens to the same cell afterward. ^[10]

A similar distinction matters when measuring clearance. An increase in autophagosomes, the compartments that carry material toward lysosomal degradation, could reflect increased delivery of cargo. It could also reflect a blockage farther along the process, leaving those compartments uncleared. The quantity present at a particular moment can look similar under either condition, while the rate of successful processing differs. That rate is the relevant question when we ask whether cellular digestion is keeping up. ^[10]

These longitudinal methods can also help clarify how different kinds of single-cell evidence fit together. For instance, in the Huntington's culture study by Charlene Smith-Geater and colleagues, single-cell sequencing traced abnormal signals of continued cell division to a persistent neural stem cell population within a mixed neuronal culture. Having established which cells generate a signal, longitudinal imaging can ask how the state of an identified cell changes, while molecular measurements help explain the machinery underlying those changes. [6,10]

But the available technology still shapes how much of a cell we can describe. A fluorescent reporter makes a selected process visible, and Finkbeiner discusses the difficulty of observing several reporters together when their signals overlap. He also reports that increasing a tagged stress-response protein changed cell survival, demonstrating how a measurement can alter the behavior it was intended to reveal. His efforts to combine imaging, molecular measurements and analyses of cell shape expand that view, while leaving a gap between a set of measurable pathways and a full account of the living cell. ^[10]

That gap is where I would place more emphasis. We can use pathway measurements to explain how a cell processes its workload while also following its function, recovery and relationships with surrounding cells. The measurements would need to connect energy availability and material turnover to sustained neuronal function, so that we characterize a cell by what it can continue doing under demand. Earlier changes that predict later death identify candidates for intervention, and controlled perturbations can test whether changing them alters the outcome. The strongest case for the metabolic thesis would be to observe a decline in cellular processing before protein pathology develops, then show that restoring the relevant function changes the subsequent course.