16 Oct 2018

The LilrB2 receptor on the surface of neurons binds Aβ and is thought to mediate some of its toxicity. Researchers led by Lin Jiang and David Eisenberg at the University of California, Los Angeles, now detail the part of Aβ involved, and the specific LilrB2 pocket it fits. As reported in the October 8 Nature Chemistry, designing a molecule that competes for the pocket blocks Aβ toxicity in neurons. "It's nice that they could figure out how the molecules fit, create a small molecule to disrupt that interaction, and decrease toxicity,” wrote David Teplow, now professor emeritus at UCLA, who was not involved in the work. “This appears to be a ‘complete story.’”

Scientists resolved the crystal structure of Aβ-binding receptor LilrB2.
They found a binding pocket and complementary region on Aβ.
A compound binds the pocket, blocking Aβ and protecting neurons.

“This work validates the structural reality of a true interaction between the LilrB2 receptor D1-D2 domain and β-amyloid,” said Carla Shatz, Stanford University, California, who originally discovered that Aβ binds the receptor (Kim et al., 2013). Shatz found LilrB2 signaling was altered in the brains of patients with Alzheimer’s disease. Genetically knocking out the mouse equivalent, PirB, saved APP transgenic mice from Aβ-induced memory problems and protected cultured cortical neurons from Aβ toxicity, giving credence to the idea that the Aβ-LilrB2 interaction could make a good therapeutic target.

To design an inhibitor, first author Qin Cao and colleagues had to determine exactly where Aβ bound on the extracellular side of the receptor. They used a 200-residue stretch of the extracellular domains D1 and D2, where Shatz reported that Aβ binds. They narrowed down binding to a six-amino-acid stretch between amino acids 16 and 21 of Aβ—KLVFFA. Oligomers, but not monomers, of this peptide bound the receptor, suggesting this region of Aβ binds in oligomeric form as well. Since oligomers of Aβ are heterogeneous and transient, Cao used a more stable tandem repeat of the KLVFFA fragment to study binding.

Crystal Structure. A pocket between extracellular domains D1 and D2 of LilrB2 holds four benzamidines (Ben 1 to 4), similar to the phenylalanine side chains of Aβ. [Courtesy of Cao et al., 2018. Nature.]

The group then grew crystals of D1D2 in the presence of the Aβ tandem. X-ray analysis of those revealed a pocket, nestled in between D1 and D2, that was the right size, shape, and electrostatic environment for Aβ. It bound two benzamidines, added for crystal optimization, which mimicked two phenylalanines lying close to one other, as they would in a tandem repeat version of KLVFFA. One at a time, the authors mutated a few amino acids lining those pockets and found that for each, less than half the Aβ bound there compared with wild-type domains. This gave them more confidence that they had hit on the right binding site. They confirmed these binding predictions by computationally modeling interactions between the LilrB2 protein and Aβ.

Cao then searched a library of 32,000 small molecules—some approved as drugs, some currently in clinical trials, and some natural compounds already tested in animals and humans—to find compounds that could inhibit the interaction between LilrB2 and Aβ and might be safe for use in people. He selected 12 based on three-dimensional structure and probability of fitting the Aβ-binding pocket of D1D2. The researchers then narrowed this down to the six that best bound LilrB2 and prevented Aβ binding in vitro.

In HEK293T cells expressing full-length LilrB2, 10μM of each inhibitor reduced Aβ binding by 50 to 70 percent. Computational software predicted that one of these inhibitors, ALI6, best bound LilrB2 and elbowed out Aβ. In cultured mouse neurons, ALI6 reduced Aβ binding by 60 percent, but no more, suggesting the peptide attaches to other receptors on the neuronal surface as well. ALI6 is also known as fluspirilene, and is an antipsychotic drug used to treat chronic schizophrenia.

In HEK293T cells and cultured mouse neurons, ALI6 decreased Aβ-induced cell death by 90 and 66 percent, respectively. It completely blocked the downstream dephosphorylation of the actin-depolymerizing factor, cofilin, which results from LilrB2-Aβ binding. Shatz had reported that cofilin dephosphorylation caused synapse loss.

Together the results imply that inhibiting the LilrB2-Aβ interaction could prevent some of the toxic effects of Aβ and stave off damage to neurons. The authors acknowledge that the high concentrations of ALI6 used in this study would be hard to achieve in vivo. Liang proposed optimizing a compound to have higher affinity for the receptor. He plans to test ALI6 in transgenic mouse models of AD to see if it can protect them from Aβ-caused toxicity or symptoms.

“In the future it will be very exciting to see if inhibiting the Aβ interaction with LilrB2 in the brain curtails synaptotoxic signaling, leading to improved synaptic plasticity, memory, and cognitive function,” wrote Iryna Benilova, University College London, to Alzforum.

“The authors are quite properly cautious about the long-term therapeutic implications since their molecule is a long way from being a drug,” wrote Gregory Petsko, Weill Cornell Medicine, New York, to Alzforum. “Perhaps the approach taken in this paper will, if not lead to a single curative or preventive agent, eventually produce one that may have value in combination with other treatments.”—Gwyneth Dickey Zakaib

Comments

- Gregory Petsko
  Harvard Medical School and Brigham&Women's Hospital
- Posted: 17 Oct 2018
The study is well-done, which is not surprising considering who did it. The authors are quite properly cautious about the long-term therapeutic implications since their molecule is a long way from being a drug; nevertheless, many projects have begun this way and ended up producing a therapeutic. If nothing else, it is a fine example of how such projects can be undertaken in their earliest stages with today’s methods.

I would quibble with the idea that docking “validates” anything about binding sites. False positives abound. The mutagenesis experiments, though not conclusive either, are more encouraging regarding the binding site. Too bad they weren’t able to soak the benzamidine out of the crystals and replace it with their compound—I assume they tried.

The serendipitous discovery of biologically relevant binding sites and bound protein conformations from adventitious binding of buffer components to proteins in the crystalline state has a long history with many fun examples, of which this is the latest. I like to say that the ontogeny of adventitious binding in macromolecular crystals recapitulates the phylogeny of native ligand binding in vivo. It has happened a number of times in my own lab (e.g.: “An anion binding site in human aldose reductase: mechanistic implications for the binding of citrate, cacodylate, and glucose 6-phosphate,” Harrison et al., 1994, and “The 1.20 A resolution crystal structure of the aminopeptidase from Aeromonas proteolytica complexed with tris: a tale of buffer inhibition,” Desmarais et al., 2002).

Although the authors quite properly present some of the reservations about the amyloid hypothesis as a basis for Alzheimer’s therapy, it seems clear that they buy into it themselves. It’s important for the field to consider more carefully the possibility that the hypothesis is right about APP misprocessing being connected to the onset and progress of the disease, while still being a poor guide to therapeutic strategy. This would be true if, for example, Aβ aggregation was only a strong driver for the familial form of the disease and more a consequence of disease in the sporadic form. It would also be true if, for example, Aβ pathogenesis was only part of a larger panoply of cell biology problems that both caused and drove the disease, so that interfering with it would be of minimal benefit. I can think of a number of other examples.

There is a curious Manichean character to many scientists, who seem to be most conformable with the idea that something must be either completely true or completely false, or that only one of a set of competing hypotheses can be true. The amyloid hypothesis can be true and still not be all that is going on, or even the most important thing that is going on in terms of finding a treatment. In that case, as in the case of many cancers and viral diseases, combination therapy needs to be considered carefully. Perhaps the approach taken in this paper will, if not lead to a single curative or preventive agent, eventually produce that in combination with other treatments that may have value.

References:

Harrison DH, Bohren KM, Ringe D, Petsko GA, Gabbay KH. An anion binding site in human aldose reductase: mechanistic implications for the binding of citrate, cacodylate, and glucose 6-phosphate. Biochemistry. 1994 Mar 1;33(8):2011-20. PubMed.

Desmarais WT, Bienvenue DL, Bzymek KP, Holz RC, Petsko GA, Ringe D. The 1.20 A resolution crystal structure of the aminopeptidase from Aeromonas proteolytica complexed with tris: a tale of buffer inhibition. Structure. 2002 Aug;10(8):1063-72. PubMed.

View all comments by Gregory Petsko

- Iryna Benilova
  University College London
- Posted: 17 Oct 2018
In this paper, researchers led by Jiang and Eisenberg identified an Aβ oligomer binding site on the LilrB2 receptor, a noncanonical major histocompatibility complex class I immune receptor implicated in amyloid-mediated synaptotoxicity (Kim et al., 2013). Using an LilrB2 ectodomain-based ELISA assay with different Aβ fragments, region AA16-21, a so-called “steric zipper” was mapped as the Aβ core that binds the receptor. Remarkably, only a tandem repeat but not a single copy of the steric zipper was recognized by the receptor, implying some conformation specificity.

Using the crystal structure of LilrB2 D1-D2 ectodomains complexed with four benzamidine molecules that mimic four phenylalanines in the “minimal oligomer” required for LilrB2 binding, Cao et al. identified two binding pockets between D1 and D2 domains of LilrB2 and validated them as a binding site for wild type Aβ AA16-21 tandem, using mutagenesis and docking simulations. An interesting mechanistic question that this study opens is if Aβ bearing mutations in the region 16-21/22, e.g. K16N, A21G (Flemish) or an array of E22 mutants (Dutch, Arctic, etc.) attenuate or exacerbate Aβ/LilrB2-mediated toxic signalling.

LilrB2 is one of the few Aβ binding partners whose role in Alzheimer’s disease remains poorly studied, in contrast to the earlier discovered receptors such as PrP(c) (Purro et al., 2018). Interestingly, emerging evidence suggests that LilrB2 receptor can robustly bind different Aβ aggregates with nanomolar affinity warranting more research (Kim et al., 2013; this paper; Benilova, De Strooper, unpublished results with Aβ oligomer preparations as per Kuperstein et al., 2010).

Cao et al. also looked out for small molecules that could inhibit the Aβ-LilrB2 interaction. Twelve were selected from a library that included approved drugs or compounds in clinical trials. To assess the inhibitors experimentally, the authors opted for the classical cell death assays such as MTT and TUNEL, which was a sensible choice for a proof-of-principle study but perhaps not as informative as a functional test in a model of LilrB2-dependent synaptic plasticity would be. In the future, it will be very exciting to see if inhibiting the Aβ interaction with LilrB2 in brain curtails synaptotoxic signalling (including the one mediated by other Aβ receptors), leading to the improved synaptic plasticity, memory, and cognitive function.

References:

Kim T, Vidal GS, Djurisic M, William CM, Birnbaum ME, Garcia KC, Hyman BT, Shatz CJ. Human LilrB2 is a β-amyloid receptor and its murine homolog PirB regulates synaptic plasticity in an Alzheimer's model. Science. 2013 Sep 20;341(6152):1399-404. PubMed.

Kuperstein I, Broersen K, Benilova I, Rozenski J, Jonckheere W, Debulpaep M, Vandersteen A, Segers-Nolten I, Van Der Werf K, Subramaniam V, Braeken D, Callewaert G, Bartic C, D'Hooge R, Martins IC, Rousseau F, Schymkowitz J, De Strooper B. Neurotoxicity of Alzheimer's disease Aβ peptides is induced by small changes in the Aβ42 to Aβ40 ratio. EMBO J. 2010 Oct 6;29(19):3408-20. PubMed.

Purro SA, Nicoll AJ, Collinge J. Prion Protein as a Toxic Acceptor of Amyloid-β Oligomers. Biol Psychiatry. 2018 Feb 15;83(4):358-368. Epub 2017 Nov 21 PubMed.

View all comments by Iryna Benilova

- Susan Catalano
  Cognition Therapeutics Inc.
- Posted: 17 Oct 2018
It is exciting to see the research community pursue anti-oligomer drug discovery and follow up on the original findings that LilrB2 is part of the oligomer receptor complex. It is a reflection of the renewed interest in the Aβ oligomer hypothesis and the willingness of labs to participate in the search for cures for this devastating disease.

Aβ oligomers are challenging to handle and quantify in a drug-screening context; different preparation and purification conditions lead to different size ranges and the resulting oligomers must be rigorously compared with AD patient-brain-derived oligomers with several methods. Acidic domains on many extracellular proteins have been shown to efficiently induce fibril formation, and care must be taken to ensure that screens do not identify small molecules that non-specifically inhibit aggregation (i.e., fibril blockers or busters), since these have not had clinical success.

It is interesting that the authors identified CDPPB, a positive allosteric modulator of mGluR5, as an anti-oligomer binding compound; this suggests there might be direct interactions between LilrB2 and cellular prion protein as constituents of an oligomer receptor complex.

View all comments by Susan Catalano

- Takaomi Saido
  RIKEN Center for Brain Science
- Posted: 25 Oct 2018
Aβ is a sticky peptide that binds to almost any membrane protein. I thus find it difficult to believe in vitro results. Besides, few data have proven the presence of Aβ oligomers and their toxicity in vivo. In my view, the published results are likely artifacts. Aβ is a garbage by-product of physiological APP processing.

View all comments by Takaomi Saido