Does protein aggregation drive postmitotic tissue degeneration?

Abstract

Pharmacological evidence, from clinical trials where patients with systemic amyloid diseases are treated with disease-modifying therapies, supports the notion that protein aggregation drives tissue degeneration in these disorders. The protein aggregate structures driving tissue pathology and the commonalities in etiology between these diseases and Alzheimer’s disease are under investigation.

INTRODUCTION

A collection of diseases called amyloidoses, which includes Alzheimer’s disease (AD), exhibit pathology associated with cross–β sheet protein aggregates such as amyloid fibrils that are deposited in various tissues (Fig. 1) (1, 2). Individual proteins that have the propensity to form amyloid fibrils, such as amyloid β (Aβ), can adopt distinct amyloid structures that contribute to the diversity of amyloid disease phenotypes (2, 3). In addition, modifier genes and environmental factors are likely to alter the age of onset, the severity of disease, and the tissues affected in these amyloidoses (4, 5). The process of protein misfolding and aggregation leads to the formation of a spectrum of aggregate structures (annular oligomers and protofilaments) in addition to amyloid fibrils that may include distinct aggregate conformers as yet uncharacterized (Fig. 1) (1, 3, 6). Structure-proteotoxicity relationships are lacking, so it remains unclear which misfolded and aggregated protein structures drive degenerative phenotypes in these diseases. The amyloid hypothesis posits that the process of protein aggregation that yields multiple aggregate structures causes the degenerative manifestations of the amyloidoses. A version of the amyloid hypothesis is also used to explain AD pathogenesis (4, 5).

Fig. 1. Protein aggregation and mechanisms of toxicity.

Intrinsically disordered and folded proteins undergo a conformational change that renders them aggregation competent (top). Aggregation-prone proteins form a spectrum of aggregates including amyloid fibrils (middle) that are postulated to contribute to tissue degeneration. These aggregates may be in circulation, deposited in the extracellular spaces and plasma membranes of postmitotic tissues and may enter cells through endocytic pathways. The non-native protein structures responsible for pathogenic processes leading to tissue degeneration remain unclear, but a variety of mechanisms are likely to be involved (bottom). These mechanisms include deficits in protein trafficking and transport, mitochondrial dysfunction, and activation of inflammation. The amyloid hypothesis of AD posits that the process of Aβ aggregation in the brain leads to neurodegeneration and dementia. Clinical testing of disease-modifying therapies in patients with TTR amyloidosis, light chain amyloidosis, and serum amyloid A amyloidosis provides an opportunity to investigate the amyloid hypothesis in these three amyloidoses. Clinical data suggest that protein aggregation drives tissue degeneration in these three systemic amyloid diseases, but the data from AD clinical trials remain inconclusive regarding support for the amyloid hypothesis of AD.

There are three principal systemic amyloid diseases: transthyretin (TTR) amyloidosis characterized by polyneuropathy or cardiomyopathy phenotypes, light chain amyloidosis resulting in multiple organ dysfunction, and serum amyloid A amyloidosis typified by early renal dysfunction followed by progression to multiple organ failure (Table 1). For each of these three maladies, there exist disease-modifying therapies that slow or halt aggregation of newly synthesized proteins in symptomatic patients by blocking the start of the amyloidogenic cascade, resulting in slowing of disease progression (7–21). Clinical testing of disease-modifying therapies in patients with these three disorders allows investigation of the amyloid hypothesis and the elucidation of commonalities between the systemic amyloid diseases and AD.

Table 1.

The amyloidoses.

Amyloid disease	Pathology	Symptoms
Alzheimer’s disease	Extracellular Aβ aggregates (plaques) and intracellular tau aggregates (tangles)	Cognitive dysfunction and dementia
Familial amyloid polyneuropathy	Extracellular mutant Transthyretin (TTR) aggregates causing axonal loss	Progressive sensory, motor, and autonomic axonal neuropathy
Familial amyloid cardiomyopathy	Extracellular wild-type and mutant TTR aggregates in the heart	Orthostatic hypotension, increased heart rate, and heart failure
Wild-type TTR amyloid cardiomyopathy	Extracellular wild-type TTR aggregates in the heart	Increased heart rate and heart failure
Familial leptomeningeal amyloidosis	Mutant TTR aggregates in brain meninges and parenchyma and arteries	Dementia, transient focal neurological episodes, and subarachnoid hemorrhage
Light chain amyloidosis	Extracellular light chain aggregates in multiple organs	Fatigue, shortness of breath, and heart failure
Amyloid A amyloidosis	Extracellular amyloid A aggregates in the kidney and other organs	Edema, reduced urination, and shortness of breath

PHARMACOLOGICAL EVIDENCE THAT PROTEIN AGGREGATES DRIVE DEGENERATION

Clinical studies in symptomatic patients with TTR amyloidosis demonstrate that halting TTR aggregation early during disease is an effective therapeutic strategy that is consistent with the amyloid hypothesis (7–9,15). Two different small molecules, tafamidis and diflunisal, bind to TTR and slow its aggregation in the blood (7, 15). Two double-blind, placebo-controlled clinical trials and two open-label clinical studies of these drugs (7–9, 15) have reported slowing of polyneuropathy in ~70% of patients with TTR amyloidosis exhibiting a principal polyneuropathy phenotype (TTR familial amyloid polyneuropathy; Table 1) (14). According to a placebo-controlled clinical trial and an open-label study, tafamidis also slowed progression of TTR amyloidosis displaying a primary cardiac pathology (TTR amyloid cardiomyopathy; Table 1) due to aggregation of mutant or wild-type TTR (10, 12). In the familial TTR amyloidoses, both wild-type and mutant TTR form aggregates, whereas in wild-type TTR amyloid cardiomyopathy, only wild-type TTR forms aggregates (Table 1). Tafamidis and diflunisal block the amyloidogenic cascade by slowing the first and rate-limiting step of the TTR aggregation process, that is, dissociation of the native nonamyloidogenic TTR tetramers. In those patients with TTR amyloidosis with a primary cardiomyopathy or a principal polyneuropathy, rate-limiting dissociation of TTR tetramers followed by a conformational change in the monomers and misassembly leads to TTR aggregation.

Two oligonucleotide drugs that lower TTR mRNA, patisiran and inotersen, slow disease progression in patients with TTR familial amyloid polyneuropathy according to two double-blind, placebo-controlled clinical trials (11, 13). These two drugs likely reduce the TTR concentration to be below the so-called critical concentration for aggregation and so stop initiation of the amyloidogenic cascade (11, 13). The critical concentration is defined as the minimum concentration of the monomeric amyloidogenic protein that is required to support aggregation. The mRNA-lowering agents act through substrate deprivation, in contrast to tafamidis and diflunisal, which act through pharmacological stabilization of the nonamyloidogenic tetramers (11, 13). Both patisiran and inotersen substantially lower plasma TTR concentrations at clinical doses, with an ~80% median decrease from a baseline plasma TTR concentration of ~20 mg/dl (11).

Genetic evidence supporting the amyloid hypothesis as the cause of TTR amyloidosis is compelling. More than 125 missense mutations in one TTR allele are linked to TTR familial amyloid cardiomyopathy or familial amyloid polyneuropathy (Table 1). The mutant allele leads to destabilized TTR heterotetramers that dissociate rapidly, enabling more efficient aggregation of the misfolded subunits. Patients who are compound heterozygotes carrying the common familial amyloid polyneuropathy–associated V30M TTR mutation on one allele and the T119M TTR mutation (acting as an interallelic trans-suppressor) on the other allele (22) are protected against disease or show a mild polyneuropathy due to kinetic stabilization of the TTR heterotetramers (22). A 68,602-person prospective study indicated that harboring the T119M mutation on one allele and wild-type TTR on the other allele lowered the risk of cerebrovascular disease while extending life span (23).

Clinical studies on light chain amyloidosis also support aggregation as the driver of organ degeneration (19–21, 24). Elimination of the clonal plasma cells secreting aggregation-prone light chains through chemotherapy slows the progression of organ deterioration in about half of symptomatic patients. High-dose melphalan (a DNA alkylating agent) and autologous stem cell transplantation led to durable remission and improved overall survival in these patients. However, only a minority of patients are eligible to receive this aggressive treatment due to advanced organ dysfunction by the time a diagnosis is made. Proteasome inhibitor–based treatment regimens, specifically bortezomib, are the backbone of initial therapy for patients who are not able to undergo stem cell transplantation. Bortezomib and similar treatments lower the light chain concentration below the critical concentration required for light chain aggregation, supporting the notion that light chain aggregation drives tissue degeneration. Although light chain amyloidosis generally results from somatic mutations, germline mutations in the gene encoding the antibody constant domain (the region of an antibody that is identical in all antibodies of the same isotype) may cause an inherited form of light chain amyloidosis. The notion that destabilizing mutations in the light chain constant domain cause aggregation and organ dysfunction provides additional genetic evidence in support of the amyloid hypothesis (25).

Systemic autoinflammatory diseases increase hepatic expression of the serum amyloid A protein, leading to its aberrant endoproteolysis and aggregation resulting in tissue degeneration. This amyloid A amyloidosis is initially associated with renal dysfunction, which progresses to multiple organ failure if left untreated. Secretion of the serum amyloid A protein by the liver, the precursor to amyloid A, is driven by the proinflammatory cytokines interleukin-6 (IL-6), IL-1, and tumor necrosis factor. Pharmacological evidence demonstrates that disease progression can be slowed or even stopped by ameliorating inflammation, which lowers the concentration of serum amyloid A protein below its critical concentration; this, in turn, stops serum amyloid A aggregation (16–18).

THE RELATIONSHIP BETWEEN AGGREGATE STRUCTURE AND PROTEOTOXICITY

An aggregate structure-proteotoxicity relationship has not been established for any human amyloid diseases to date. That is, there is no agreed-upon mechanism for how the process of aggregation compromises postmitotic tissues, nor have the aggregates responsible been identified. There are many data-based hypotheses for how the process of protein aggregation compromises postmitotic tissues that cannot easily regenerate (Fig. 1). The advantage of using TTR amyloidosis and light chain amyloidosis to establish aggregate structure-proteotoxicity relationships is that there are sufficient patients being treated to carry out a statistically significant clinical study. Moreover, the putative pathogenic aggregate structures are likely to be found in blood or tissues available for biopsy and so could be sampled more easily (6). We need to quantify the various aggregate structures that substantially decrease in the plasma of patients with light chain amyloidosis or TTR amyloidosis being treated with disease-modifying therapies. This quantification needs to be performed on a time scale during which tissue degeneration ceases to progress owing to pharmacological inhibition of protein aggregation. Correlating the disappearance of certain aggregate structures with a clinical response would rely on the assumption that correction of the proteinopathy would be sufficient to achieve a clinical response. Such a correlation becomes more challenging to establish if both the clearance of certain aggregate structures and correcting a process such as neuroinflammation are required for disease amelioration in a subset of patients.

Preliminary data indicate that nonamyloid oligomers circulate in patients with TTR familial amyloid polyneuropathy (6) and are reduced by treatment with kinetic stabilizer drugs such as tafamidis or by liver transplantation (6). The non-native monomers or misfolded aggregate structures that are reduced on the time scale of the clinical response could be the potential drivers of pathology (6). There is a lack of technology that can accurately quantify misfolded monomers and aggregate structures (6), although progress has been made using peptide probes (6) and enzyme-linked immunosorbent assays to detect non-native forms of TTR. Methods for quantifying annular oligomers and structurally distinct protofilaments in patient plasma, cerebrospinal fluid (CSF), and tissue need to be developed. Whereas it seems unlikely that TTR and light chain cross–β sheet amyloid fibrils are cleared on the time scale of drug-induced clinical responses, amyloid fibril clearance has not been rigorously quantified in clinical trials or published studies. Given that amyloid load can now be quantified by positron emission tomography (PET) imaging, studies can be designed to address whether a decrease in amyloid load correlates with clinical responses in patients with systemic amyloidoses (26).

EARLY PHARMACOLOGICAL TREATMENT IS KEY

Clinical trials of disease-modifying therapies for TTR amyloidosis showed that the earlier the treatment was initiated in the disease course, the better was the clinical response of patients (7, 8, 12). For example, the percentage of patients with TTR familial amyloid polyneuropathy carrying the V30M TTR mutation who responded to tafamidis decreased from 68 to 46% if tafamidis treatment was delayed by 18 months (because the patients were in the placebo arm of the trial initially) in otherwise demographically similar groups (7, 8). Among the patients with cardiomyopathy with two wild-type TTR alleles or who were heterozygous, tafamidis treatment resulted in a reduced risk of death for 64.4% of those with New York Heart Association (NYHA) class I heart failure (n = 37), compared to 39.6% of patients with NYHA class II heart failure (n = 263) and 16.3% of patients with NYHA class III heart failure (12).

New early diagnostic assays are needed to discern whether TTR amyloidosis disease progression can be fully halted in most patients if disease-modifying anti-aggregation treatments are started early enough. Tafamidis, patisiran, or inotersen treatment initiated “early” in the course of TTR familial amyloid polyneuropathy benefited about two-thirds of patients (14, 27). However, sensory and autonomic neuropathy often precedes treatment initiation, suggesting that treatment needs to start even earlier. Currently, TTR mutation carriers are considered asymptomatic for polyneuropathy although intraepidermal, sweat gland, and pilomotor nerve fiber densities are all reduced in this group relative to age-matched healthy controls (28). Moreover, asymptomatic TTR mutation carriers harboring a variety of mutations had twice as many nerve lesions relative to controls and half as many nerve lesions as patients with a tissue biopsy positive for amyloid (29).

We need to understand what differentiates patients with TTR familial amyloid polyneuropathy who respond to tafamidis versus those who do not respond. Because of the different primary end points used in placebo-controlled clinical trials testing tafamidis versus patisiran and inotersen, it is difficult to compare the effectiveness of these agents. It is crucial to identify patient characteristics that can predict clinical responses to each new therapy. Toward this end, the effectiveness of tafamidis has been scrutinized in a 210-person retrospective study (in patients not enrolled in the registration trial) over an observation period of 18 to 66 months. Clinical experts classified the patients as responders [34% with almost complete arrest of disease progression and a neuropathy impairment score (NIS) change of ≤0 over 54 months], partial responders (36% with partial to complete arrest in some disease symptoms and an NIS median increase per year of 1.8 points), and nonresponders (30% with a median NIS progression rate of 5.9) (14). This approach should enable a comparison of all three drugs using this published response-to-therapy predictive equation that will need prospective validation and further refinement (14).

Treating patients with TTR amyloid cardiomyopathy earlier should be possible because these patients often have bilateral carpal tunnel syndrome release surgery or lumbar spinal stenosis surgery several years before their cardiomyopathy is diagnosed. A recent study showed that 10% of patients undergoing carpal tunnel syndrome release surgery exhibited TTR aggregates in tenosynovial tissue that could be stained with Congo red (30). Of these, two patients exhibited positive scans with technetium pyrophosphate scintigraphy imaging (31). Such scans could potentially identify patients with cardiomyopathy for early treatment with tafamidis, if TTR involvement is suspected (30). Those patients with TTR aggregates in the tenosynovial tissue, in the lumbar spinal canal, or in other soft tissues, but showing no signs of cardiomyopathy, should be monitored carefully (31).

Although amyloid fibril deposition in the extracellular space between cardiomyocytes in the heart contributes to cardiac dysfunction especially late in disease, clinical trial data from patients with familial or wild-type TTR amyloid cardiomyopathy suggest that this is not the main driver of mortality. Tafamidis (20 or 80 mg once daily) slows progression to heart failure in these patients by all clinical trial metrics published (12), especially in less severely affected patients. This is despite the fact that there is no evidence for substantial amyloid clearance from the heart (based on heart wall thickness) on the time scale of the clinical response (30 months).

IS THERE PHARMACOLOGICAL SUPPORT FOR THE AMYLOID HYPOTHESIS OF AD?

Quantifying each aggregate structure, including amyloid fibrils, and correlating the disappearance of distinct aggregates in response to drugs have the potential to move the AD field forward by establishing the relationship between aggregate structure and proteotoxicity. This strategy avoids the need to guess which aggregate structures are driving the degenerative phenotypes in AD. A number of companies have carried out expensive AD clinical trials based on the assumption that amyloid fibrils are the main driver of disease pathology and that clearing amyloid fibrils using antibodies would modify disease progression. Until very recently, these trials had all failed, despite demonstrating a reduction in amyloid fibrils in the brain by PET imaging in some cases.

Biogen/Eisai initially declared that their two phase 3 clinical trials (EMERGE and ENGAGE) with the Aβ aggregate-targeting monoclonal antibody, aducanumab, had been stopped because of futility. Aducanumab was derived from a de-identified blood lymphocyte library collected from healthy elderly individuals with no signs of cognitive impairment. Aducanumab binds to Aβ amyloid plaques within postmortem brain tissue sections from patients with AD; it binds to Aβ40 and Aβ42 aggregates in vitro but not to the amyloid precursor protein or monomeric forms of Aβ. The company later startled the scientific and medical communities by announcing that analysis of a larger, more complete dataset showed that aducanumab reduced clinical decline in patients with early AD according to prespecified primary and secondary end points. The primary end point was the Clinical Dementia Rating Sum of Boxes Score, which, in the EMERGE trial, decreased in a dose-dependent manner by 23% in the highest dose treatment arm relative to the placebo arm (P = 0.01). Statistical significance was achieved in most of the secondary end points as well.

The similar size phase 3 ENGAGE clinical trial did not exhibit statistical significance in the same primary or secondary end points, nor did it exhibit the expected dose dependence that the EMERGE study showed. The Peripheral and Central Nervous System Drugs Advisory Committee of the U.S. Food and Drug Administration ruled that the EMERGE and ENGAGE trials did not provide sufficient evidence of aducanumab efficacy for treating AD. Without clinical evidence to support it, the Aβ amyloid hypothesis of AD has been left in limbo.

ARE WE TREATING AD TOO LATE?

Enrolling patients with early-stage AD in clinical trials, when a pure Aβ aggregation–associated proteinopathy could exist, may increase response rates. Many in the AD scientific community have suggested that anti–Aβ aggregation strategies have failed largely because we are treating the proteinopathy component of AD at a stage of the disease when other factors such as tau aggregation and neuroinflammation are playing a more important role in driving disease pathogenesis. The hypothesis currently being tested clinically is that ameliorating the proteinopathy (aggregation) component of AD at an early stage in patients with a rare early-onset familial form of AD could be effective in slowing or stopping disease progression (32). The genetic evidence supporting the hypothesis that AD is largely driven by Aβ aggregation during early disease is compelling (4, 5). Failed AD drug candidates with persuasive mechanisms of action merit reevaluation once patients with early-stage familial AD can be identified and recruited to clinical trials, which is already beginning to happen (32).

ARE ALL AMYLOID DISEASES SYSTEMIC?

Some have argued that AD is a localized brain amyloidosis and so is different from the systemic amyloid diseases. Whereas it is clear from the literature that the TTR amyloidoses have peripheral pathological manifestations, whether this is the case in AD remains controversial and needs to be scrutinized with more rigor. There are reports of peripheral Aβ deposition in AD, but questions remain regarding whether the peripheral pathology precedes the central nervous system (CNS) pathology or arises in parallel. Another critical question is whether detection of peripheral pathology could lead to treatment initiation much earlier, resulting in amelioration of AD pathology in the CNS (33). Several groups have demonstrated that some patients with AD have subclinical heart disease, including diastolic and myocardial dysfunction, apparently caused by Aβ aggregate deposition in the cardiomyocytes and interstitium (34, 35). Aβ aggregation was also observed in the large intestine in patients with AD by one group but not another, which may explain the higher incidence of serious upper and lower gastrointestinal events in patients with AD versus healthy age-matched controls in a large retrospective clinical study (36). We need to determine in patients with AD the extent to which Aβ deposition is observed in the periphery, the timing of aggregate deposition relative to CNS pathology, and the extent to which systemic pathology influences quality of life. Whether technetium pyrophosphate scintigraphy imaging can detect Aβ deposition in the hearts of patients with AD also merits investigation (31).

EMERGENCE OF CNS TTR AMYLOIDOSIS AFTER PERIPHERAL TREATMENT

More than 2000 patients with TTR familial amyloid polyneuropathy underwent liver transplantation before tafamidis became available. In this surgical form of TTR gene replacement, transplantation of healthy donor livers carrying two wild-type TTR alleles led to a decrease in secretion of destabilized mutant TTR heterotetramers into the blood of the transplanted patients (37). In liver transplant recipients, the blood concentration of the tetramer-destabilizing mutant TTR subunits was reduced by 95%, slowing the progression of peripheral and autonomic neuropathy. However, mutant TTR continued to be secreted from the choroid plexus into the CSF after liver transplantation. Many patients, who were effectively treated by liver transplantation for peripheral TTR amyloidosis for over a decade, began to exhibit TTR aggregation in the CNS as revealed by PET imaging (26). CNS TTR aggregates resulted in dementia, transient focal neurological episodes, and cerebral vascular bleeding (23, 26, 38–40). Mutant TTR aggregation continued in the CNS in transplanted patients, most likely due to biosynthesis and secretion of TTR heterotetramers by the choroid plexus. TTR aggregates compromised the meninges and arteries as well as the brain parenchyma as disease progressed (26, 38). Moreover, magnetic resonance imaging supported the hypothesis that the microbleeds associated with meningovascular amyloidosis caused the focal neurological episodes observed in patients with the V30M TTR mutation a decade or more after liver transplantation (41). TTR amyloidosis in the CNS presents much later than TTR amyloidosis in the periphery in these patients, most likely because the TTR concentration in CSF is an order of magnitude lower than that in the blood, thereby lowering aggregation efficiency.

The CNS pathology due to aggregation of TTR in heterozygotes (26, 38) is similar to that in Aβ aggregate–mediated cerebral amyloid angiopathy. It has been argued that >60% of patients with sporadic AD exhibit cerebral amyloid angiopathy (42). In this disease, fragments of the amyloid precursor protein are secreted from the brain into the CSF, where they aggregate and destroy the vasculature and the brain parenchyma, leading to cerebrovascular bleeding and dementia. The extent to which cerebrovascular bleeding contributes to dementia in classical AD deserves greater scrutiny.

Historically, patients with untreated TTR amyloid polyneuropathy or cardiomyopathy (Table 1) probably died before their CNS pathology became overt. Some patients expressing rare TTR mutations develop a familial leptomeningeal amyloidosis as their initial phenotype. The mutant TTR of these patients is too unstable to be secreted by the liver but is secreted by the choroid plexus into the CSF, where it aggregates. That liver transplantation does not appear to protect against CNS pathology supports the hypothesis that TTR secreted into the CSF by the choroid plexus penetrates the arteries from the outside in (that is, from the adventitia of the artery muscle wall to the media), while largely leaving unaffected the endothelium in direct contact with the blood, at least in early disease (26, 38). It is too early to tell whether TTR kinetic stabilizers such as tafamidis that cross the blood-brain barrier would be present in sufficient amounts to protect against TTR amyloidosis in the CNS.

CONCLUDING REMARKS

Pharmacological evidence from clinical trials in patients with systemic amyloid diseases supports the hypothesis that the process of TTR, light chain, and serum amyloid A aggregation causes degeneration of postmitotic tissues in these disorders. The disease-modifying drugs tested in these trials inhibit the process of aggregation by the newly synthesized aggregation-prone proteins, thus supporting the amyloid hypothesis for these diseases. Drugs that work in a similar fashion by blocking Aβ aggregation in AD have failed to support the amyloid hypothesis of AD so far. To date, there is little evidence that amyloid fibrils, versus other aggregate structures, are the main drivers of the degenerative phenotypes in AD or in the systemic amyloidoses.