Core Value of Amyloids

For decades, experts in brain disorders have fixated on the toxic properties of amyloids, the generic name for misfolded proteins stuck together in a particularly indestructible way. After all, amyloid plaques in the brain, formed by the protein named amyloid β, are the original defining pathological feature of Alzheimer's disease (AD). A counterintuitive new view is emerging in which some amyloids may wield a potential for doing good—and to a far greater extent than anyone suspected, according to the latest evidence from a mouse model of multiple sclerosis (MS) published 3 April in Science Translational Medicine.

In the new study, researchers saw remarkable therapeutic results using only amyloid-forming "cores" from a dozen different proteins. When injected with the smallest protein pieces that can reliably form amyloid fibers, mice disabled by an MS-like neuroinflammatory disorder could walk again (Kurnellas et al., 2013).

The paralysis returned with a vengeance when the treatments ended. That suggests that an entire category of sticky amyloid-forming cores may be active biological agents with therapeutic potential in MS and other neuroinflammatory diseases, say Lawrence Steinman, a neuroimmunologist at Stanford, and his co-authors. (A related story at AlzForum addresses the implications of the findings for AD.)

"This is an important piece of science," says Terrence Town, a neurobiologist at the Zilkha Neurogenetic Institute at the University of Southern California in Los Angeles, who was not involved in the study. "This paper is interesting because it shows what might be pathological in one disease can be beneficial and therapeutic in another." Indeed, the findings pull together several lines of research in MS, immunology, and amyloids in unexpected ways, say researchers contacted for this article. Scientists from different fields are finding that amyloid-forming proteins can assemble into many varying structures, some of which may be extremely neurotoxic and others that may be benign or even beneficial. "This paper takes that idea to the next level," Town says.

Following the trail to amyloid in MS

All proteins contain short segments that could, in theory, aggregate to form amyloid fibrils, but few proteins can indulge their dark sides, says Ulrich Hansmann, a biophysical chemist at the University of Oklahoma in Norman. In a normally folded protein, the crucial short sticky segments that can form the core (or “spine”) of amyloids are usually tucked away too deep to be tempted to aggregate with other such regions in neighboring proteins. When exposed by damage or misfolding of the protein, though, the amyloid segments may irresistibly huddle together side by side in an interlocking pleated strip called a beta sheet. The amyloids can assume various shapes and sizes from there, based on the other parts of the protein. Even with the most nefarious amyloids implicated in human diseases, scientists do not know which structural versions might be toxic nor how they are causing disease (Eisenberg and Jucker, 2012).

In the case of MS, one of the amyloid-forming proteins investigated by the Stanford group—αB-crystallin (HspB5), which belongs to the small heat shock protein family—first appeared on the radar 2 decades ago. Heat shock proteins are also called "chaperones," because they help to prevent partially folded or assembled proteins from aggregating in damaging amyloid clumps. The full-length αB-crystallin protein has advanced to early phase II clinical testing in Europe as a potential therapy for MS, based on studies that suggest it might silence autoimmune attacks against the myelin sheath. The potential therapeutic is under development by Delta Crystallon BV in Leiden, The Netherlands. "This study adds another layer of beneficial activity of the protein," says Hans van Noort, a biochemist at the drug company, of the new Stanford work. "We're working with the entire protein, but this shows the ability of bits and pieces of it to promote resolution of the problem in the brain."

Paradoxically, van Noort first identified αB-crystallin as a potential trigger of destructive inflammatory activity in MS nearly 20 years ago (van Noort et al., 1995) after taking a fresh look at tissues from a brain bank to ask what might be provoking an immune attack. Working with brain samples from individuals with MS lesions and from others without neurological disease, he and his co-authors separated the proteins from the myelin-forming cells into different test tubes. They then added white blood cells—extracted from people with or without MS—to each tube and looked for a reaction. It turned out that αB-crystallin, which was isolated only from the MS lesions, stimulated a dramatic response in both sets of white blood cells.

Until then, the scientific interest in αB-crystallin had been mostly the domain of evolutionary biologists, van Noort says. The 10-member crystallin family originally included all the proteins found in the eye lens of most vertebrates, from fish to people. They are named by molecular weight, starting with alpha for the biggest ones.

In MS, the obscure heat-shock protein appears to accumulate in myelin-forming cells before immune cells infiltrate the brain, van Noort says. “It is therefore not a secondary response to inflammation associated with an MS lesion. Instead, it triggers the formation of an MS lesion." He remains convinced that αB-crystallin is the main target of the immune attack in MS (van Noort et al., 2010).

Why would the immune system start picking a fight with this protein? The trigger might be an infection with Epstein-Barr virus (EBV), the only known link between MS and an infectious agent, van Noort speculates. It turns out that B cells spit out αB-crystallin when infected with EBV, perhaps inadvertently teaching T cells to fight both the virus and the protective protein (van Sechel et al., 1999). In the continuous immune surveillance of the brain, "hundreds of thousands of T cells are rummaging around, testing the waters," van Noort says. A problem in the brain, particularly in the myelin-forming cells (one of the few other cell types that produce αB-crystallin under stress), may lead to a buildup of the same protein that some T cells were programmed to attack in concert with EBV. Under siege, myelin-forming cells may spew out even more of the protein—attempting to protect the cellular machinery, block cell death, and recruit microglia as cellular bodyguards, but instead further inflaming T cells in a vicious cycle.

"If you want to stop the MS lesion from developing and stop T cells in the brain from causing all this trouble, you don’t need to suppress the entire immune system," van Noort says. "The only thing you need to address is the T cell reaction against this single protein." He and his colleagues are pursuing a strategy of building tolerance to the protein using small doses designed to reprogram errant T cells.

Steinman was an enthusiastic reviewer of van Noort's 1995 paper, but the Stanford scientist began studying αB-crystallin only after genomic and proteomic studies documented its abundance in MS lesions and its absence in normal brain tissue. Steinman, a co-author on some of those profiling studies, wondered about a potentially protective role for the protein. "In response to inflammatory injury, the brain doesn’t roll over and play dead," he says. "We've heard brains can't reproduce neurons or bounce back from injury. That's not true. The brain produces guardian molecules that counter [inflammatory damage]."

In a first round of experiments more than 6 years ago, Steinman and his colleagues showed that mice missing the gene for αB-crystallin developed worse experimental autoimmune encephalomyelitis (EAE) (see "Animal Arsenal") than did their normal counterparts. Injections of the protein into peripheral circulation made both types of mice better, reducing the severity of paralytic disease and attenuating inflammation in the brain (Ousman et al., 2007). Absence of αB-crystallin also aggravated disease in experimental models of stroke and brain trauma, they found, while injections of it appeared therapeutic, suggesting a more general anti-inflammatory role for the small heat shock protein.

Now the question was: How were injections having a therapeutic effect in the EAE mice? How was αB-crystallin affecting the immune system? In the Steinman lab, neuroimmunologist Michael Kurnellas and biochemist Jonathan Rothbard picked up the investigation. First, they mixed αB-crystallin with blood samples from people with MS, rheumatoid arthritis, or amyloidosis (a rare condition of protein buildup in organs), as well as from mice with EAE. The heat shock protein selectively trapped a broad, common set of inflammatory molecules and effectively disarmed them. At slightly higher temperatures, such as those found at a site of inflammation, the quantities of the trapped inflammatory molecules doubled or tripled (Rothbard et al., 2012). The data were consistent with the molecule's chaperone function.

The unexpected idea that the molecule might be functioning in an amyloid state first dawned on the members of the Steinman team when a Japanese group reported that a larger related crystalline protein in the eye lens worked as a chaperone only when it could form amyloid fibers. The amyloid-forming fragments of the full-size heat shock protein were both sufficient and essential, the other paper showed, for its ability to bind potentially harmful molecules (Tanaka et al., 2008). When the Steinman team tested αB-crystallin to learn if the same was true in their system, the answer was yes. They reported last September that a 20-amino-acid-long snippet of the protein that contains the amyloid-forming section was as potent as the full protein in reducing paralysis in EAE mice (Kurnellas et al., 2012). "When we altered one amino acid, we disrupted formation of the amyloid fibril, and there was no therapeutic activity," Kurnellas says. "We were excited that the amyloid itself could be therapeutic."

In a parallel project, another research team in the Steinman lab reported last August on the anti-inflammatory and potentially therapeutic properties in EAE mice of amyloid-forming pieces of amyloid β (Grant et al., 2012; “Split Personality”).

Amyloid versus amyloid

Even with the hints of therapeutic potential, the researchers remain concerned about possible toxic properties of the amyloid pieces. Some smaller amyloid-forming fragments known as oligomers have a newly discovered ability to form a cylinder shape that can pierce cell membranes and are thought to be among the most neurotoxic amyloid structures, Steinman says (Laganowsky et al., 2012). For the new study in Science Translational Medicine, the investigators wanted to strip down the peptides to the minimum string of amino acids needed to assemble amyloid fibrils to test in the EAE mice.

Fortunately, structural biologists have been making rapid progress in sorting out the atomic details of disease-related amyloid-forming proteins of various sizes and shapes (Eisenberg and Jucker, 2012). The Steinman team turned to the amylome, a database of all the proteins with regions that can zip together in the amyloid spine configuration. The resource contains information about experimentally solved and computationally predicted regions compiled by David Eisenberg, a structural biologist at the University of California, Los Angeles, and his colleagues (Goldschmidt et al., 2010).

The Steinman team tested 18 amyloid-forming sequences, each six amino acids long—including those from αB-crystallin, amyloid β A4, tau, major prion protein, amylin, serum amyloid P, and insulin B chain. First, Rothbard and Kurnellas checked that each peptide sequence actually assembled into amyloid structures in solution by using a dye that scatters light differently when bound by an amyloid fiber. Based on atomic structural data from others, the researchers believe the hexamer peptide strands in the test tubes likely fuse together in the side-by-side interlocking and layered beta sheet strips, six amino acids wide, that form the core amyloid spine (Sawaya et al., 2007). In these experiments, the dynamic fiber assemblies seemed to reach a steady-state length of about 20 to 25 strands for each paired beta sheet on average, with some dropping off and others joining in the zipper-Velcro-like assembly. "They don't form infinite ribbons," Rothbard says.

The team then injected each batch of different amyloid cores into the abdominal cavities of EAE mice, where the proteins moved into the bloodstream and circulated throughout the body. The animals’ symptoms improved, usually moving from complete hind limb paralysis to hind limb weakness or tail weakness. "It doesn't act immediately," Rothbard says. "It takes two to three injections. Once you stop the injections, the symptoms come back a day or two later." The researchers found no additional evidence of toxicity in the major organs of the mice.

The researchers also checked the chaperone activity of two representative hexamer peptide fibrils in a routine test with damaged insulin molecules. The small fibers prevented the insulin proteins from aggregating, but the amyloids could not undo those that had already knit themselves together in a larger misshapen amyloid structure. The results showed a correlation between the chaperone function and protection in mice by the amyloid fibers. "One amyloid can reduce another amyloid," Rothbard says.

At this stage, the researchers only have hints about how the short amyloid peptides are working. The fibrils seem to subdue the immune system by reducing levels of circulating inflammatory molecules, but multiple mechanisms may be at work, the authors and other researchers say. "They definitely have an effect on the immune system," Kurnellas says. "We're not even sure the fibrils are getting into the brain. The therapeutic effect may be an indirect result of what is being targeted in the periphery."

In further tests using blood plasma from three people with MS, the scientists observed that amyloid fibers of tau protein removed most of the same innate and adaptive immune proteins as had αB-crystallin in the experiments described in their September 2012 paper. In mice, the tau amyloid injections also reduced levels of cytokines, particularly interleukin 6, which stimulates additional pro-inflammatory mediators.

The Stanford team is now preparing a follow-up paper for publication that probes the actions and mechanisms of the small amyloid fibers in more detail, including their effect on white blood cells. For now, the long-held view of amyloids as unrelenting villains in neurodegenerative diseases makes their investigations about a potentially therapeutic role of amyloids a hard sell among their colleagues, the researchers say.

"If judges at the supreme court of chemistry were adjudicating the guilt or innocence of amyloids sitting as the accused in the dock, then they can't rule that this suspect—amyloid—is guilty," Steinman says. "There are too many extenuating circumstances that show this molecule is a good player and innocent." Those who remain undecided can count on reviewing additional evidence before passing a verdict.

Correction (3 May 2013)
The final paragraph has been modified to clarify its meaning.

Key open questions

• What are the mechanisms underlying the therapeutic effect of the amyloid fibrils in EAE mice? What cell types are involved?
• Are the injected amyloid fibrils reaching the brains and spinal cord of mice?
• What effect do naturally occurring amyloid fibrils in the brains of people with MS have on the immune system?
• Do the amyloids formed from the short hexapeptides have any toxicities?

Image credit

Thumbnail image on landing page. Courtesy of Heather McDonald.