A potential Alzheimer’s treatment aims to whisk the amyloid precursor protein (APP) out of harm’s way to prevent it from generating Aβ. The small molecule fortifies the retromer, a large protein conglomerate that ferries APP out of endosomes and away from the clutches of BACE1, the enzyme that initiates formation of Aβ. Scientists led by Dagmar Ringe, Brandeis University, Waltham, Massachusetts; Gregory Petsko, now at Weill Cornell Medical College, New York; and Scott Small, Columbia University, New York, report April 20 in Nature Chemical Biology that their new compound pre-empted production of Aβ and other toxic fragments of APP in cultured neurons, and instead nudged APP toward processing by α-secretase.

“This paper represents the combination of expert structural biology and AD cell biology to produce simultaneous advances in both arenas,” James Lah of Emory University School of Medicine, Atlanta, wrote to Alzforum in an email (see full comment below). “Their results demonstrate rather convincingly that the identified compound is capable of stabilizing the retromer complex.”

On its way from the cell membrane to other cell compartments for processing, APP stops over in the endosome. From there, the retromer shunts APP to the trans-Golgi network or back to the cell surface. However, if the retromer falters, APP lingers in endosomes, where it is at the mercy of BACE1. Small’s group previously reported that dysfunction of the retromer leads to elevated Aβ levels in cultured neurons (see Bhalla et al., 2012). Other researchers have found that variants of retromer-related genes elevate the risk for AD (see Vardarajan et al., 2012). Conversely, previous studies suggested that boosting the amount of retromer in the cell enhances its ability to traffic proteins (see MacLeod et al., 2013). 

Based on that idea, first author Vincent Mecozzi and colleagues reasoned that if they could stabilize  retromer, they might protect it from degradation, increase its presence in the cell, and help move more APP out of the endosome. The researchers turned to structure-based drug design to find a small molecule that could help hold  retromer together. A virtual model of the complex helped identify a handful of pockets where a small molecule could bind and reinforce the connection between subunits. Then, the scientists virtually screened these sites against a library of compounds to see which would fit best and to pick some for testing in an actual chemical screen. They ended up testing 24 of these chemicals with  retromer in vitro. One, R55, stabilized the complex, raising its melting temperature by 10 degrees Celsius. 

To see what R55 did in cells, the researchers tested it in cultured hippocampal neurons from wild-type and from J20 transgenic mice. The compound increased the amount of retromer complex and reduced levels of Aβ40 and Aβ42. It also lowered the APP fragments that result from BACE1 cleavage and raised those generated by α-secretase processing (see image below). Immunohistochemistry assays confirmed that R55 reduced APP levels in the endosome.


R55 nestles between retromer subunits (turquoise/orange) and cements their connection, lowering β- and raising α-secretase cleavage of APP. [Image courtesy of Mecozzi et al., Nature Chemical Biology.]

“We see this as the first step in a drug-discovery program,” said Small. The researchers are now conducting preclinical research on R55 and its analogs, to see if they are stable in serum, can cross the blood-brain barrier, and whether they are safe in animal models. “Everything looks promising,” said Small, but he added that results are preliminary. He, Petsko, and Ringe are planning to develop this further. 

In an email to Alzforum, Samuel Gandy, Mount Sinai Alzheimer's Disease Research Center, New York, praised the work and supported clinical development. However, he doubted that a single compound would provide a silver bullet for Alzheimer’s. “That simple approach made sense before we had the computational tools to begin to address the true complexity of diseases,” he wrote in an email to Alzforum (see full comment below). He and other scientists interviewed for this article also agreed that interfering with such a ubiquitous complex as the retromer is likely to have side effects.

Small conceded the point, noting that retromer is found in every cell in the body. However, he said that preliminary evidence suggests that just a slight boost in retromer stability with R55 or its analogs is not toxic to animal models. He added that a small dose of a retromer-stabilizing compound could be added to a more target-specific therapy, such as a BACE inhibitor, to produce an additive effect, while at the same time minimizing toxicity of either compound. 

Why not stick with the BACE inhibitors currently under development (see Dec 2013 news story)? Small points out that even if a therapy successfully shifted APP processing, defects in endosomal function would persist. Also, previous studies suggest that retromer dysfunction may underlie a number of disorders, including Parkinson’s disease (see Jul 2011 news story). That suggests a compound like R55 could help treat a number of diseases. He is currently collaborating with Asa Abeliovich, also at Columbia University, to test the compounds in a mouse model of Parkinson’s. 

“That a chemical chaperone can restore functional retromer activity is a remarkable proof of concept,” Matthew Farrer, University of British Columbia, Vancouver, Canada, wrote to Alzforum in an email (see full comment below). “Further research and development might have broad clinical applications.”—Gwyneth Dickey Zakaib


  1. Scott Small, Dagmar Ringe, Greg Petsko, and their colleagues are to be congratulated on a beautiful piece of science. This paper represents the combination of expert structural biology and AD cell biology to produce simultaneous advances in both arenas. Their results demonstrate rather convincingly that the identified compound (R55) is capable of stabilizing the retromer complex through binding at a Vps35-Vps29 interaction site. The effects of R55 on APP processing also support the notion that enhanced retromer function will tend to shunt APP through a non-amyloidogenic pathway, reduce APP-BACE interaction, and limit Aβ production.

    It is worth noting that a small-molecule strategy to stabilize the retromer complex and enhance its function should affect all retromer-mediated traffic. A retromer-based therapeutic strategy for AD will require further refinement to establish target specificity. Nevertheless, this paper provides a nice proof-of-principle that indirect control of APP processing and Aβ production through manipulation of intracellular traffic is a viable approach.

  2. My reaction is that the new pro-retromer drug is brilliant and essential, and at the very least, we have a terrific new tool for cell biology at the bench. In contemplating its use in treating or preventing AD, there are some caveats that tend to keep me up at night.

    The first is that I have become convinced that the conventional one target-one drug approach is willfully naive.  That simple approach made sense before we had the computational tools to begin to address the true complexity of diseases.  Genetic predispositions have been present in humans since conception, and many equilibria will shift as a result. Instituting a one target-one drug intervention in adulthood in hopes of a clinical benefit may well turn out to be a quixotic endeavor. 

    The second caveat is readily addressable in whole animal and human trials.  One worries that interfering with such a ubiquitous event as cargo retrieval is likely to have pleiotropic side effects.  I am especially worried about the vps10 family since there are other pathogenic molecules therein (e.g., sortilin). That concern notwithstanding, I completely support moving ahead with clinical development.

    Looking at this from the perspective of a diabetologist (which I was in a previous life), since retromer-based dysfunction of the insulin receptor may underlie SORCS1-linked type 2 diabetes (T2D), I would love to see a retromer-stabilizing drug tested in Sorcs1 knockout mice and in people with SORCS1 mutations who have T2D.

  3. This study raises some intriguing possibilities. Loss of retromer expression has long been noted in AD, with consequent effects on APP processing and Aβ production. However, retromer recycling appears as a nexus for parkinsonism and dementia. Deficits in neuron transmission and early endosome protein sorting and trafficking are highlighted by the recent discoveries of mutations in DNAJC13 and VPS35, most likely mediated by dominant-negative loss of functional WASH complex. That a chemical chaperone can restore functional retromer activity is a remarkable proof of concept. Further R&D might have broad clinical applications in PD-MCI and dementia. These are exciting times.

  4. The paper by Mecozzi et al. elegantly demonstrates how a small drug, R55, is able to stabilize the retromer complex, enabling enhanced neuronal transport of APP (and sorLA) out of early endosomal compartments thereby escaping amyloidogenic processing. These novel data strongly support previous studies from several groups, and adds yet additional evidence that retromer plays a pivotal role in the neuronal trafficking of APP together with sorLA.

    The idea to boost retromer activity by selectively stabilizing the interaction between individual components is intriguing, and offers a very beautiful way to rapidly control the level of retromer and the fate of its cargo.

    One major challenge to developing this interesting compound into a pharmacologically useful drug will probably be to get R55 into the brain and show that it can exert its function in vivo. So far the report from Scott Small and colleagues only describes how R55 cross-links the subunits VPS29 and VPS35 in neuronal cultures, but they have not yet tested their potential drug in animal models.

    It will therefore be highly interesting to see in future studies whether this compound also is able to reduce amyloid production in the brain of mice. Moreover, as retromer is responsible for the retrograde sorting of many cargo receptors besides sorLA (directly) and APP (indirectly), the risk of harmful side effects from using retromer activity as a molecular target against Alzheimer’s disease is high. As the effect described in the present paper is mediated by the triad of retromer, sorLA, and APP, it is also of interest to determine next what other retromer/sorLA-dependent ligand sorting will be affected by R55. 

    These concerns still need to be addressed before we will find out whether R55 is offering the way for a new pharmacological compound or not. Nevertheless, so far the paper holds great promises, and presents a very elegant approach and innovation towards selectively avoiding amyloid production.

  5. This is a very carefully done study that combines elegant structural biology, drug discovery, cell biology, and neuroscience. Clearly, endosomal APP is an attractive target and the authors of the study have identified a pharmacological chaperone that stabilizes retromer complex and thus enhances the sorting of APP away from endosomes. This has clear implications for therapy. Of course, one needs to study how other cargoes that depend on retromer are affected and whether there is a way to confer specificity towards APP sorting. But this is the first demonstration that a small molecule could regulate retromer-dependent sorting of APP from the endosomes, and its a very elegant one.

    View all comments by Lawrence Rajendran

Make a Comment

To make a comment you must login or register.


Research Models Citations

  1. J20 (PDGF-APPSw,Ind)

News Citations

  1. Cloistered Retreat Takes the Pulse of BACE Research
  2. Sorting Out Parkinson’s: Exome Sequencing Points to Recycling Defect

Paper Citations

  1. . The location and trafficking routes of the neuronal retromer and its role in amyloid precursor protein transport. Neurobiol Dis. 2012 Jul;47(1):126-34. PubMed.
  2. . Identification of Alzheimer disease-associated variants in genes that regulate retromer function. Neurobiol Aging. 2012 Sep;33(9):2231.e15-2231.e30. PubMed.
  3. . RAB7L1 interacts with LRRK2 to modify intraneuronal protein sorting and Parkinson's disease risk. Neuron. 2013 Feb 6;77(3):425-39. PubMed.

Further Reading

Primary Papers

  1. . Pharmacological chaperones stabilize retromer to limit APP processing. Nat Chem Biol. 2014 Jun;10(6):443-9. Epub 2014 Apr 20 PubMed.