Amyloid-Beta42: The Good, the Bad and the Alzheimer’s
Plaque Versus Soluble Amyloid
Amyloid has dominated the disease mechanism hypothesis of Alzheimer’s since its discovery in 1984 by Glenner and Wong. Or perhaps since Alois Alzheimer observed it in the brain of his patient in 1906. Its presence in the brain is required for a diagnosis, initially by autopsy, later by amyloid PET scan and most recently, by plasma biomarkers that have been validated by their accuracy in predicting amyloid PET. Amyloid was what researchers could see, and the research followed.
We now know that amyloid plaque in the brain can be detected 15 years or more before cognitive symptoms occur. However, amyloid deposits without cognitive impairment may or may not mean that Alzheimer’s is on its way.
In the 1990s, the field discovered that cognitive decline and synaptic deficits were more correlated with soluble amyloid than with the amyloid plaque deposits, which appeared to be inert cellular debris, or the leftover evidence of a pathogenic cascade. In a landmark 1991 paper, Terry showed that amyloid plaque load correlated very poorly with cognitive decline; whereas, the density of synapses was powerfully correlated. What was causing the density of synapses to decline?
In 1991, Pike showed that amyloid beta (Aβ) peptide, starting with soluble monomers, only became toxic to hippocampal neuronal cultures when it was allowed to aggregate over two to four days. We now know that the aggregated, high molecular weight species he observed included both soluble oligomers and insoluble amyloid protofibrils and fibrils, the precursors to plaque.
In 1992, Hardy and Higgins proposed the “Amyloid Cascade Hypothesis,” which held that the aggregation and accumulation of Aβ caused Alzheimer’s disease. They specifically meant the aggregation of amyloid beta protein and formation of insoluble amyloid plaque.
In 1993, Jarrett and Lansbury showed that it was the Aβ42 peptide (42 amino acids long) that was sticky and prone to aggregate, not the more common Aβ40, due to two extra hydrophobic amino acids. Zeroing in on Aβ42 as the prime suspect, researchers still thought it was insoluble fibrils or plaques, not soluble amyloid, causing the disease.
Two seminal 1999 papers published two months apart by Lue and by McLean showed that it was soluble amyloid, not insoluble amyloid fibrils or plaque, that correlated with both synaptic deficits and cognitive decline. Lue showed that cognitively intact individuals with high plaque burden did not have corresponding high tissue levels of soluble Aβ42 and Aβ40. Importantly, high tissue levels of soluble Aβ meant reduced synaptic density — they were highly inversely correlated.
McLean showed that insoluble Aβ did not correlate with disease severity, but soluble Aβ did correlate with cognitive impairment and was three times higher in Alzheimer’s brains than in healthy controls. McLean pooled both Aβ42 and Aβ40 in these calculations but also showed that Aβ42 was crucial to the cognitive impairment. In 2000, Näslund also confirmed that brain levels of soluble Aβ42 and Aβ40 correlated with cognitive decline.
The Aβ42 Fragment
Aβ42 is cut out of the larger amyloid precursor protein by two enzymes, gamma secretase on one end and beta secretase on the other. We have long been taught that the Aβ42 fragment is toxic, but the focus on insoluble fibrils and plaques did not shift to soluble oligomeric forms of Aβ42 until around 2000. Aβ42 aggregates easily into dimers, trimers and larger oligomers. Dennis Selko showed that these oligomers were toxic to synapses, impairing long-term potentiation (LTP), the process of enhancing neuronal firing with repetition that underlies learning. Aβ42 monomers or the insoluble plaque forms did not impair LTP.
Injecting Aβ42 dimers and trimers from human Alzheimer’s brains into healthy mice impaired memory and caused a loss of neuronal dendritic spines, further validating small oligomers of Aβ42 as toxic. They injected Aβ42 concentrations of 1–5 nanomolar, 10–50 times higher than Aβ42 in healthy brains.
While the field was predominantly focused on amyloid as a causative agent in Alzheimer’s disease, a smaller camp was focused on tau, a microtubule-associated protein that becomes dysfunctional and aggregates when hyperphosphorylated, forming intracellular tangles, the other hallmark of Alzheimer’s disease. Hoau-Yan Wang was the first to show that Aβ42 induces tau hyperphosphorylation by binding and signaling through a receptor critical for learning and memory: the α7 nicotinic acetylcholine receptor.
Soluble Aβ42, mostly dimers and trimers, hijacks the α7 receptor to activate the kinases ERK and JNK1 to hyperphosphorylate tau, causing tau to lose its normal function of stabilizing microtubules, which are the transport lines inside cells. With intracellular transport impaired, cells are stressed and synapses starved, leading to neuronal degeneration and eventual cell death. While the neurons are declining in health, the dysfunctional tau aggregates into intracellular tangles that are the second hallmark of the disease.
Two Ways Soluble Aβ42 Can Form Plaque
There are two ways that plaque can form. The first is entirely extracellular, involving Aβ42 aggregation into oligomers, then protofibrils and fibrils, and finally plaque. Once a “seed” or small fibril is formed in the extracellular fluid, it rapidly captures free-floating Aβ42 monomers, causing the fibril to elongate and form plaque. Injecting pre-formed fibrils into brains of transgenic mice causes robust extracellular plaque formation, demonstrating that these fibrils act as templates to assemble plaque outside the cell.
Amyloid plaque also forms by binding and signaling through the α7 nicotinic acetylcholine receptor. Aβ42 small oligomers bind the α7 receptor with picomolar or sub-picomolar affinity, piling onto the receptor and signaling to hyperphosphorylate tau. As more and more Aβ42 piles onto the receptor, it internalizes into the cell with the receptor, depositing intracellularly in the endosomal-lysosomal system. In the acidic lysosome, the high concentration of Aβ42 rapidly assembles into fibrils, which are resistant to lysosomal degradation by proteases. They accumulate and choke the lysosome.
Meanwhile, the hyperphosphorylated tau also stops stabilizing microtubules, and the cells degenerate, leaving behind the amyloid fibril seeds, which rapidly catch free-floating monomers to grow and condense into plaque. As supportive evidence of this plaque-forming pathway, the pyramidal cells of cortex and hippocampus highly express α7 and also have the highest concentration of intracellular Aβ42. The plaques left behind also mirror the size and shape of the dead neurons. A final piece of evidence is that debris from the dead neurons, including lysosomal proteins and the α7 receptor, is consistently found inside dense cores of amyloid plaques, suggesting that the plaque was born inside cells. It then attracts more extracellular Aβ42.
How Soluble Aβ42 Concentration Increases
The toxic signaling of soluble Aβ42 and its disruption of synaptic plasticity occur at brain tissue concentrations in the nanomolar range. In healthy individuals, the brain tissue concentration is 100–200 picomolar (picomolar is one-thousand times lower than nanomolar), and it exists mostly as unaggregated monomers. But we’ll come back to that healthy tissue concentration later.
When the concentration of monomers increases enough locally in the brain tissue’s interstitial fluid (the fluid between cells), the monomers begin to alter shape, aggregating into oligomers.
Monomers can be easily cleared into cerebrospinal fluid (CSF) due to their small size, but oligomers are less easily cleared. Oligomers are not only bulkier, they also expose highly hydrophobic patches that stick to surrounding tissue, preventing clearance. As the disease takes hold, loose monomers disappear from the interstitial fluid and oligomers spike.
Aβ42 oligomers also actively damage the glymphatic or rinsing system of the brain, trapping the oligomers in the brain while the monomers drain out. They do this several ways. Aβ42 oligomers activate astrocytes, changing them from supportive cells to reactive inflammatory cells. When astrocytes transition to reactive, the aquaporin-4 water channels are no longer concentrated on their end-feet, which wrap around blood vessels to allow interstitial fluid to pass directly from the brain into the bloodstream. With fewer aquaporin-4 channels surrounding blood vessels, fluid transport into the bloodstream is impaired.
CSF is normally cleared into blood several times a day, via a complicated pressure system of exchange from the subarachnoid space into the bloodstream via arachnoid granulations that protrude through the dura, the outer membrane of the brain. It is a one-way, pressure-driven filtration system. CSF can also drain directly into meningeal lymphatic vessels embedded in the dura. Aβ42 oligomers impair the CSF-to-blood exchange by directly clogging the arachnoid granulations and by destroying meningeal lymphatic transporters.
Get Lindsay Burns, PhD’s stories in your inbox
Join Medium for free to get updates from this writer.
Just to confuse you, and despite CSF drainage issues, the concentration of soluble Aβ42 is known to be high in brain tissue but low in CSF of Alzheimer’s patients — because it has been aggregating into oligomers and sticking to tissue. Plaques, once formed, also appear to be a sticky sink for loose Aβ42. As more Aβ42 oligomers stick to brain tissue, particularly to the widespread, femtomolar-affinity binding sites of α7 nicotinic acetylcholine receptors, there is less to wash away into the CSF or bloodstream.
Toxic Effects of Soluble Aβ42
Let’s now look at what Aβ42 oligomers are doing in the brain tissue. As Aβ42 dimers and trimers increase in concentration, they cross a threshold where they induce toxic signaling through the α7 receptor, a receptor critical to cognition. Higher concentrations also impair LTP and synaptic plasticity, processes fundamental to learning and memory.
Aβ42 oligomers become active ligands at the α7 receptor, initiating pathogenic signaling to activate kinases that hyper-phosphorylate tau. This binding site is the only sub-picomolar affinity binding site of Aβ42 and so is likely the most critical to disease. As mentioned above, hyperphosphorylated tau, even before aggregating into neurofibrillary tangles, can no longer stabilizes microtubules, impairing intracellular transport. All this causes neuronal degeneration. But Aβ42 oligomers don’t stop there.
Aβ42 oligomers also interact with the cellular prion protein and crosslink it to the mGluR5 receptor, and this tripartite complex over-activates Fyn kinase, which in turn, hyper-phosphorylates the NR2B subunit of extra-synaptic NMDA receptors to cause their unregulated overactivation and excitotoxicity by a massive influx of calcium. Fyn kinase also activates Pky2 kinase, which directly hyperphosphorylates tau.
Aβ42 oligomers also degrade NMDA receptors in the synapse by binding EphB2 receptors, which tether to NMDA receptors to stabilize them. Aβ42 binding of EphB2 receptors causes them to internalize and degrade, and they bring along with them the NMDA receptors, depleting them from the synapse to undermine LTP and synaptic plasticity.
Aβ42 oligomers bind sigma-2 receptors in a complex with three other membrane components at very low affinity, 0.5 micromolar, but this interaction is reportedly stable and high volume, because it causes upregulation of sigma-2 receptors and more docking sites for Aβ42 oligomers in the synapse. These interactions purportedly impair membrane trafficking of AMPA receptors and other dendritic spine proteins, impairing synaptic plasticity.
Aβ42 oligomers also activate inflammatory receptors TLR4 (via its co-receptor CD14), TLR2, TREM2 and RAGE, and they are pathological modulators of several chemokine receptors. Aβ42 oligomers also act as damage-associated molecular patterns (DAMPs), directly binding pattern recognition receptors, innate immune receptors that detect their misfolded beta-sheet structure, on the surface of microglia and astrocytes. Aβ42’s multiple pathways to induce neuroinflammation are seemingly never ending.
In the long list of destructive processes, Aβ42 oligomers cause signaling that hyperphosphorylates tau, cause excitotoxicity at extra-synaptic NMDA receptors, deplete synaptic NMDA receptors, impair synaptic densities and synaptic plasticity by impairing trafficking of synaptic receptors and proteins, and initiate neuroinflammation via a variety of pathways. These destructive functions occur once the concentration of Aβ42 small oligomers cross a certain threshold, which is self-induced by their actions that impede clearance into CSF or blood.
What Good Is Aβ42?
By now you understand that Aβ42 is bad, very bad. It causes neurodegeneration and neuroinflammation and, above all, impairs synaptic plasticity, the key process for learning and memory.
Not so fast. Aβ42 is only toxic at higher concentrations. In fact, it is necessary at lower concentrations.
Wait, what?!
I know, I first told you that Aβ42 impairs synaptic plasticity. Now I’m telling you that it actually helps this very process that it hurts, just at lower concentrations.
In a seminal 2008 paper by Daniela Puzzo, low picomolar concentrations of Aβ42 — physiologic concentrations — enhanced LTP and improved memory. These low concentrations were delivered to normal mice by hippocampal infusion. Next, Puzzo showed in 2011 that blockade of Aβ42 by antibodies or by ablation of APP expression in normal, healthy mice impaired LTP and worsened memory. These deficits were rescued by injecting exogenous Aβ42 into the hippocampus of these mice. In 2019, Puzzo showed that these 100–200 picomolar concentrations of Aβ42 increased vesicle docking and pre-synaptic transmitter release and increased dendritic spine density. Remember, at nanomolar or higher concentrations, Aβ42 decreases dendritic spine density and impairs LTP and synaptic plasticity.
These researchers were the first to propose a hormetic response of Aβ42: low concentrations are critical for synaptic plasticity and memory, while high concentrations impair it. The “paradox” of hormesis is defined as a biologically adaptive response to a low dose of a stressor that is harmful in higher doses. Hormesis is some version of “what doesn’t kill you makes you stronger.”
Possibly predicted by Puzzo’s mouse data, clinical evidence that Aβ42 is necessary in low concentrations comes from several Phase 3 clinical trials of inhibitors of enzymes that create Aβ42 from the larger APP protein. In five separate Phase 3 trials, efforts to prevent the production of Aβ42 by inhibitors of gamma secretase or beta secretase (BACE) worsened cognition. These repeated unfortunate results likely occurred due to suppressing even the physiologic concentration of Aβ42 that is important to synaptic function and synaptic plasticity. With high concentrations of Aβ42 still toxic, a refined approach may be secretase modulators, which aim to lower the concentration into the physiologic range.
Development of BACE inhibitors accelerated after the discovery of the Icelandic mutation in APP, which prevented BACE from cleaving APP into the usual Aβ42 peptide. The lucky carriers of this mutation were protected against Alzheimer’s disease, retained cognitive function well into their 80s and 90s, and showed no signs of amyloid plaques as centarians. Aβ peptides were reduced overall in these individuals by about 40%, but most importantly, a mutant Aβ42 peptide called AβICE, appeared to be protective. With an A2T mutation (the second amino acid, alanine is replaced by threonine), AβICE didn’t aggregate and it protected the normal Aβ42 by mixing with it. The BACE inhibitors couldn’t recreate this protective mutant AβICE; they could only prevent production of all Aβ42 peptide. And that was likely their problem.
Other Physiologic Functions of Aβ42
In addition to promoting synaptic function and plasticity, there are several other physiologic roles of Aβ42. Moir and Tanzi showed that Aβ42 functions as an antimicrobial peptide, fighting infections, by trapping viruses or bacteria by rapid oligomerization. APP knockout mice have substantially higher vulnerability to lethal brain infections.
Aβ42 has been shown to control hyperexcitability and seizures. Malinow showed that normal neuronal activity drives release of Aβ42, which then acts as a negative regulator to prevent over-excitation.
Aβ42 can function as a survival factor: completely blocking Aβ42 production by secretase inhibitors causes rapid neuronal death of cells under serum deprivation, rescuable by adding picomolar concentrations of Aβ42. Aβ42 improves cell survival by binding IGF-1R, activating the PI3K/Akt survival pathway to increase neurotrophic factors such as BDNF.
Finally, during hypoxia or ischemia, hypoxia-inducible factor (HIF-1) spikes, increasing beta secretase and Aβ42 production. Blocking BACE increased the ischemic damage. With continued hypoxic events, the rescue pathway of increasing Aβ42 is constant, resulting in higher concentrations that can then aggregate and become toxic.
The Bottom Line
What causes Aβ42 concentration to increase, to cross the line? Perhaps cell stress, decreased clearance, or overproduction, all of which can interact and snowball. But it is important to remember that Aβ42 is not strictly a villain. At lower, healthy concentrations, it actually helps learning and memory.
Hormesis may actually be the wrong word to describe the effects of Aβ42, because Aβ42 might not be a stressor that is helpful in small doses, like exercise, fasting or cold plunge. It appears to be a functional peptide that unfortunately is easily prone to going awry. Just like tau, which stabilizes microtubules when phosphorylated correctly but loses that healthy function when hyperphosphorylated in Alzheimer’s disease, Aβ42 is healthy and needed in a certain low concentration but becomes toxic as it crosses a threshold, aggregates and accumulates. As a drug developer, I see Aβ42 not as a purely toxic peptide but more as a therapeutic peptide with a very narrow therapeutic window, above which is a toxic threshold. Like many drugs.
