Consider a study in last week’s online PNAS a reminder that one lab’s trash could be another’s treasure. The “trash” in this case—a transgenic mouse generated during a search for genes regulating morphological development (Lee et al., 1992)—has become a goldmine for researchers looking to implicate a specialized protein-trafficking complex in Alzheimer disease. As it turns out, the animals succumb to a host of AD-like brain defects revealed only recently by New York and Wisconsin scientists who learned that the transgene in these mice fortuitously disrupts the endogenous gene for vacuolar protein sorting (VPS) 26. VPS26 is a key component of the retromer, a protein complex that shuttles cargo along the endosome-trans-Golgi circuit and has recently come under closer scrutiny for its suspected role in AD. In the new study, the researchers analyzed VPS26 knockout mice as well as retromer-deficient flies expressing the human wild-type genes for amyloid precursor protein (APP) and β-site APP-cleaving enzyme (BACE). The data bolster a growing body of evidence suggesting that retromer deficiency can contribute to late-onset AD pathogenesis.

For Scott Small of Columbia University, New York, who led the new research, early hints of a connection between retromer trafficking and AD emerged from reams of microarray data he and colleagues were analyzing in 2003. At that time, his team was hunting for molecules with brain expression patterns that seemed to fit the spatiotemporal profile of AD. The microarray data revealed decreased expression of VPS35 and VPS26, the retromer’s two core components, in brain areas selectively vulnerable to AD (Small et al., 2005). But “frankly, we weren’t sure if that was an interesting finding because most of the retromer studies up to that point had been done in yeast,” Small told ARF, noting that microarray experiments “can only suggest a clue or potential hit.”

Then, along came a pair of Journal of Cell Biology (JCB) papers (Seaman, 2004; Arighi et al., 2004)—“two blaring headlines that retromer is relevant to mammals,” Small said. With renewed vigor, he and colleagues confirmed the microarray expression patterns of VPS35 and VPS26 first by Western blotting and later in cell cultures using small interfering RNA to show that reducing expression of the retromer components led to increased Aβ levels (Small et al., 2005). These expression data provided the first of three confirmations Small deems necessary to validate a microarray finding.

Confirmation number two, as Small sees it, came last year when an international collaboration (Rogaeva et al., 2007) reported that genetic variants of the neuronal retromer-binding receptor SORL1 are associated with late-onset AD (see ARF related news story). And the new PNAS paper, Small said, provides the third and final confirmation establishing the significance of retromer defects in AD pathogenesis.

For these studies, first author Alim Muhammad and colleagues wanted an animal model in which they could look at how retromer deficiency affects the brain. In the 2004 JCB papers, the researchers found references to a fibroblast line derived from transgenic mice with a serendipitous VPS26 gene disruption. As luck would have it, those mice lived right down the street in Frank Costantini’s lab, also at Columbia University. Homozygous mutants lack a clear developmental phenotype and die before birth—features that would relegate many mouse lines to the morgue. But to the delight of Small’s team, Costantini had actually kept a breeding pair alive for 15 years.

In a radial-arm maze test for hippocampal-dependent memory, the Columbia team found that VPS26 heterozygous knockout (VPS26+/-) mice (which, unlike homozygotes, survive to adulthood and do not have gross developmental defects) performed worse than did wild-type littermates. The VPS26+/- mice also had faulty long-term potentiation, as revealed by electrophysiological techniques that measure synaptic integrity of hippocampal neurons. Ottavio Arancio’s group, also at Columbia University, helped with the electrophysiology experiments. Correlating the memory defects with a key molecular feature of AD—elevated levels of soluble Aβ peptide—researchers in the lab of Karen Duff at Columbia found higher brain levels of endogenous Aβ40 and Aβ42 in the VPS26+/- mice, relative to wild-type littermates.

“It was nice to show that retromer deficiency accelerated the accumulation of murine wild-type Aβ,” Small said. “But at the same time, the actual disease is characterized by accelerated accumulation of human wild-type Aβ. That's why we turned to a second model that was closer to the relevant players.”

The approach Small’s team took to generate that second model diverged from what most in the field would have expected at the time. Many colleagues suggested crossing the retromer-deficient mice to a transgenic line expressing a human gene known to increase AD risk, such as APP. The reason given was usually a practical one, Small recalled. Until recently, measuring wild-type murine Aβ was tricky. Mice do not produce much Aβ, and these low amounts could only be detected by a specialized ELISA. So analyzing the retromer deficiency in a mouse producing human Aβ seemed more reliable.

The problem, as Small saw it, was that those transgenic lines express mutant versions of the culprit genes. Yet those autosomal-dominant mutations play no role whatsoever in more than 95 percent of AD cases—those with “sporadic” late-onset AD. In these patients, Aβ levels are elevated in the presence of non-mutated APP and the pair of enzymes (BACE and γ-secretase) that cleave Aβ out of it. Furthermore, sequence differences in mouse and human forms of APP and BACE significantly affect their intracellular transport and APP processing, making it hard to establish disease relevance in experiments measuring mouse proteins. “If you’re trying to develop a model for late-onset AD (LOAD),” Small said, “you want the molecular constituents to be the ones present in LOAD.”

But putting wild-type human versions of both APP and BACE into the retromer-deficient mouse would require considerable time and effort. So Small and colleagues turned to the fruit fly, an organism much easier to manipulate genetically. These studies, using retromer-deficient flies that express human wild-type APP and human wild-type BACE, were done with help from Barry Ganetzky at the University of Wisconsin, Madison.

Compared with control siblings, retromer-deficient flies—carrying one instead of two copies of VPS35—had higher brain levels of human Aβ and increased neurodegeneration. On average, the retromer-deficient flies also died younger than did control flies: half were dead at age seven days versus 16 days for controls.

“This is an intriguing study,” wrote Thomas Willnow, Max Delbrueck Center for Molecular Medicine, Berlin, via e-mail. “It lends further proof to the emerging concept that regulation of protein trafficking in neurons is an important mechanism in neurodegenerative disease processes.” (See below for detailed comment.)

Extending the PNAS work, Small and colleagues are crossing the retromer-deficient mice with a transgenic line that expresses wild-type human APP. These and other studies should help further clarify the cell biology of the retromer pathway, which could eventually guide the development of AD therapeutics. “If you can keep BACE out of the endosome by affecting protein sorting, that might be a good way to reduce Aβ production,” Small said.—Esther Landhuis

Esther Landhuis is a science writer in Dublin, California.


  1. Overall, this is an intriguing study. It lends further proof to the emerging concept that regulation of protein trafficking in neurons is an important mechanism in neurodegenerative disease processes. In line with this notion, a number of factors involved in regulation of neuronal protein transport such as sorting receptors (SORLA/LR11), sorting adaptors (GGA), or retromer components have all been implicated in processes related to Alzheimer disease in patients and animal models.

    It is striking to note that a moderate decrease in expression of retromer components VPS26 and VPS35 in the retromer-deficient mouse model (50 percent reduction) has such profound consequences for APP processing as well as for functional integrity of the hippocampus. As such, these findings seem to indicate a central role for the retromer complex in neurodegenerative processes.

    It will be interesting to see whether deficiency in retromer expression (as described for mouse and fly models herein) also coincides with abnormal trafficking of APP, of secretases, or of presumed retromer targets SORLA and sortilin—an observation that would ultimately provide a mechanistic model for the disturbances in neuroanatomy and behavior in retromer-deficient animal models reported here.

  2. Small and colleagues (1) had previously shown that VPS26 and VPS35 are decreased in AD brains and that siRNA-mediated depletion of VPS26 and VPS35 produces an increase in Aβ levels in vitro. In this interesting paper, Muhammad and colleagues report that VPS26 haploinsufficiency increases levels of Aβ in both mouse and fly models. They also show that retromer deficiency produces memory deficits and synaptic dysfunction in VPS26+/- mice. Moreover, retromer deficiency results in neurodegeneration in a fly expressing human APP and BACE. A large number of proteins have been found to be altered in postmortem brain samples from AD subjects. However, the significance of such alterations needs to be tested in vivo, and this paper represents an excellent example. The finding that sAPPβ and Aβ levels are increased suggests that β-secretase activity is increased in VPS26+/- mice. He and colleagues (2) have previously shown that in vitro depletion of VPS26 results in accumulation of BACE in endosomes, most likely due to defective transport of BACE from endosomes to the trans-Golgi network. Our group has shown that depletion of the trafficking molecule GGA3 produces an increase in BACE and Aβ levels owing to impaired lysosomal degradation of BACE (3,4). Furthermore, we have found that GGA3 levels are decreased in AD brains and inversely correlate with BACE levels (3). Several groups have shown that BACE protein levels and activity are increased in AD brains. Many mechanisms have been proposed to explain such increases. Thus, it would be interesting to test whether retromer deficiency alters BACE protein levels or affects β-secretase activity by altering its subcellular localization in vivo.


    . Model-guided microarray implicates the retromer complex in Alzheimer's disease. Ann Neurol. 2005 Dec;58(6):909-19. PubMed.

    . GGA proteins mediate the recycling pathway of memapsin 2 (BACE). J Biol Chem. 2005 Mar 25;280(12):11696-703. PubMed.

    . Depletion of GGA3 stabilizes BACE and enhances beta-secretase activity. Neuron. 2007 Jun 7;54(5):721-37. PubMed.

    . BACE is degraded via the lysosomal pathway. J Biol Chem. 2005 Sep 16;280(37):32499-504. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. SORLA Soars—Large Study Links Gene to Late-onset AD

Paper Citations

  1. . Identification and characterization of a novel, evolutionarily conserved gene disrupted by the murine H beta 58 embryonic lethal transgene insertion. Development. 1992 May;115(1):277-88. PubMed.
  2. . Model-guided microarray implicates the retromer complex in Alzheimer's disease. Ann Neurol. 2005 Dec;58(6):909-19. PubMed.
  3. . Cargo-selective endosomal sorting for retrieval to the Golgi requires retromer. J Cell Biol. 2004 Apr;165(1):111-22. PubMed.
  4. . Role of the mammalian retromer in sorting of the cation-independent mannose 6-phosphate receptor. J Cell Biol. 2004 Apr;165(1):123-33. PubMed.
  5. . The neuronal sortilin-related receptor SORL1 is genetically associated with Alzheimer disease. Nat Genet. 2007 Feb;39(2):168-77. PubMed.

Further Reading

Primary Papers

  1. . Retromer sorting: a pathogenic pathway in late-onset Alzheimer disease. Arch Neurol. 2008 Mar;65(3):323-8. PubMed.
  2. . Retromer deficiency observed in Alzheimer's disease causes hippocampal dysfunction, neurodegeneration, and Abeta accumulation. Proc Natl Acad Sci U S A. 2008 May 20;105(20):7327-32. PubMed.