Two new animal models may help researchers clarify the link between diabetes and Alzheimer disease (AD). Researchers in Japan have crossed an AD mouse (APP23) with two types of diabetic cousins, the leptin-deficient obese mouse (ob/ob) and the spontaneously diabetic NSY mouse. As senior author Ryuichi Morishita of Osaka University and colleagues report in this week’s PNAS online, offspring from both crosses have poorer learning and memory than either of their parental strains, but without producing any more amyloid-β (Aβ) than do APP23 mice. The findings suggest that diabetic mice are more sensitive to Aβ pathology, which jibes with recent data from human patients. Interestingly, the diabetic profile is exaggerated in both crosses as well, indicating that AD pathology, in turn, exacerbates diabetes. “The ob/ob mouse is a fairly good model of type 2, insulin resistance-related diabetes. The fact that they see additive effects is a really interesting phenomenon,” suggested Suzanne Craft, University of Washington, Seattle.

Scientists have known for some time that type 2 diabetes increases the risk for Alzheimer disease, but exactly how the two conditions are linked is still unclear. “This work provides a great opportunity to identify the mechanisms by which insulin resistance accelerates dementia symptoms and, similarly, to explain how APP metabolism exacerbates the diabetic phenotype,” according to one industry commentator who requested anonymity to avoid lengthy internal review (see full comment below).

First author Shuko Takeda and colleagues report that the cognition of APP23-ob/ob mice declines rapidly. At eight weeks old, both APP23 and ob/ob parental strains perform just as well as wild-type animals in the Morris water maze, cutting the time to find the hidden platform from about 75 seconds to 25 seconds over nine training sessions. But the APP23-ob/ob mice perform poorly already. Remarkably, at the ninth training session, they take as long to find the platform as they did on the first attempt. By 12 weeks of age, they perform no better, and even have difficulty finding the platform when it is in plain view above the murky water. The authors do not think poor vision explains why these animals have trouble finding raised platforms because, after three training sessions, they gradually do improve and are able to find the platform as quickly as control wild-type animals do on the first attempt.

One might think that the poor cognitive skills of the APP23-ob/ob mice could be due to greater production of Aβ in the brain as a quirk of the diabetic comorbidity. But while the brain weighed less in 12-month-old animals than in their age-matched APP23 parents, neither strain had developed amyloid plaques in the hippocampus by that age. And though faint plaques appeared in the entorhinal cortex of both strains, total Aβ levels were no different between the two. Where Aβ may enter the picture, though, is through the blood vessels. More Aβ40 deposited in vessels of six-month-old APP23-ob/ob mice compared to APP23 controls. Whether this cerebral amyloid angiopathy (CAA) is related to the eight- and 12-week cognitive deficits is unclear. Interestingly, recent work from Suzanne Craft and colleagues suggests that diabetes sensitizes humans to Aβ pathology. That’s because dementia patients with diabetes seem to have lower Aβ burden than dementia patients free of diabetes (see ARF related news story on Sonnen et al., 2009). While Craft did not look at CAA pathology in those cases, she did find that dementia patients with diabetes had a greater number of microinfarcts, indicative of blood vessel disease. “The role of CAA in dementia is very interesting, and we don’t have a good handle on that yet in humans. This study points to some very interesting possibilities that could be investigated,” she told ARF. (For a recent review on the link between vascular risk factors, diabetes, and cognition, see Knopman and Roberts, 2010.)

One potential nexus is RAGE, or receptor for advanced glycation end products. It both mediates diabetic pathology and binds Aβ (see ARF related news story). Takeda and colleagues found that, compared to the age-matched parental strain, RAGE is elevated in the blood vessels of three-month-old APP23-ob/ob mice. Interleukin 6 (IL-6) and tumor necrosis factor α were elevated in microvessels of the brain, as well, and this is in keeping with RAGE activation setting off an inflammatory response. Astrogliosis was also prominent in these young APP23-ob/ob mice. Finally, that Aβ binds RAGE might explain why these animals have worse diabetic pathology, including higher glucose intolerance and reduced insulin sensitivity, than ob/ob parents. However, Craft also pointed out that Aβ can bind the insulin receptor. “High levels of Aβ could interfere with insulin receptor binding processes and might contribute to insulin resistance,” she said.

The ob/ob mouse is a well-accepted model of type 2 diabetes. Still, the animals differ from humans with type 2 diabetes in having a leptin deficiency and a very early emergence of pathology. This raises the possibility that leptin loss, rather than diabetes per se, may directly contribute to the phenotype, note Takeda and colleagues. However, the researchers found similar exacerbations of pathology when they crossed APP23 mice with NSY (Nagoya-Shibata-Yasuda) animals. These were developed by selective breeding of animals that show spontaneous glucose intolerance, and their diabetic features emerge with age. “Though this strain is not as widely used, it does seem to capture age-related insulin resistance, or type 2 diabetes,” suggested Craft.

Takeda and colleagues report that both glucose tolerance and insulin sensitivity are reduced in 12-week-old APP23-NSY crosses compared to NSY parents. Interestingly, while different lines of offspring had different Aβ40 levels, those levels correlated with the extent of glucose intolerance, suggesting a direct link between Aβ and metabolic dysfunction. In addition, feeding the animals a high-fat diet inflamed brain microvasculature in APP23-NSY mice (but not NSY animals), as judged by IL-6 immunoreactivity, and a deterioration in performance in the Morris water maze that was not apparent in the parental strain. These aggravated readouts emerged without any increase in total brain Aβ.

Results from both sets of crosses indicate an intimate and complex relationship between diabetes and dementia, with diabetes perhaps leading to increased CAA and dementia aggravating metabolic hallmarks of diabetes. The emergence of cognitive deficits in APP23-NSY mice on a high-fat diet may be particularly germane to human conditions. “We’ve been thinking that there has got to be a gene-environment interaction that is critical in humans, and the one that seems to make the most intrinsic sense is some diet-related factor,” suggested Craft. This work also fits with an emerging consensus from biomarker research that Aβ pathology represents a sign of impending AD, and that diabetes and/or vascular disease reflect factors that limit one’s resistance to overt expression of dementia.—Tom Fagan


  1. [Editor's note: The Alzforum occasionally allows industry scientists to post comments without attribution to avoid lengthy internal review requirements.]

    Takeda et al. have made a very important step forward in capturing the interaction between type 2 diabetes and Alzheimer disease in a transgenic animal model. This work provides a great opportunity to identify the mechanisms by which insulin resistance accelerates dementia symptoms and similarly to explain how APP metabolism exacerbates the diabetic phenotype.

    Building on evidence from Suzanne Craft, Siegfried Hoyer, Greg Cole, Suzanne de la Monte, and others, Dr. Morishita and colleagues have tested a specific hypothesis that peripheral insulin resistance causes a rapid deterioration in cognitive function in mice that overexpress APP. Eight-week-old double-transgenic mice (APP+ - ob/ob) show a profound deficit in the Morris water maze that is not observed in either single transgenic line. The inability of these mice to learn the water maze cannot be explained by diabetes-induced visual impairment. Intriguingly, this deficit is apparent prior to plaque deposition, and levels of soluble and insoluble Aβ were found to be indistinguishable from the APP23 parental line.

    Importantly, the authors support this finding in a second mouse model of diabetes (NSY mice). The effect is much less pronounced in this model, perhaps because the NSY mice have impaired peripheral insulin secretion, while ob/ob mice show severe insulin insensitivity due to a deficiency in leptin signaling in the hypothalamus.

    Nevertheless, the authors clearly demonstrate that the two diseases are interacting. For instance, Aβ shows an accelerated deposition along the cerebral vasculature, and RAGE, which has been implicated in both diabetes and AD pathology, is upregulated by three months of age. Similarly, APP overexpression induced an increase in circulating glucose and a profound insulin insensitivity compared to the ob/ob parental line. A key piece to this puzzle may prove to be the further suppression of brain insulin in double-transgenic mice (Figure 4).

    The authors identify several other hallmarks of Alzheimer disease not typically seen in APP transgenic mice. These include reduced cholinergic innervation of the hippocampus, decreased brain weight, and astrogliosis. It will be important to see whether MAPT/tau is abnormally phosphorylated and if there is overt cell loss in older animals. It will also be interesting to learn how these animals respond to treatment with glitazones, DPPIV antagonists or γ-secretase inhibitors.

    View all comments by Name Withheld
  2. The elegant paper by Takeda et al. demonstrates that crossing APP-overexpressing mice with leptin-deficient mice, a model of a metabolic syndrome including hyperglycemia, exacerbates the cognitive decline observed in APP mice (APP23). The effect was observed without changes in brain Aβ burden, but cerebrovascular amyloid deposition and associated inflammation were enhanced in the APP-ob/ob mice. Considering that cerebrovascular dysfunction is present both with defective leptin signaling (Didion et al., 2005) and in APP mice (Iadecola et al., 1999), the data suggest that the added alterations in the blood supply to the brain may play a role in the cognitive worsening of the APP-ob/ob crosses.

    This conclusion is supported by experiments in APP mice crossed with NOX2-null mice in which rescuing the cerebrovascular dysfunction ameliorated cognitive performance without altering the amyloid load (Park et al., 2008). Therefore, modulation of the brain blood supply can influence, positively or negatively, the cognitive outcome in these models independently of the amyloid load. The authors were careful to consider the possibility that alterations in insulin signaling could also play a role. Furthermore, other metabolic effects of leptin deficiency (hyperlipidemia, hypotension, etc.; see Kennedy et al., 2010) could also have influenced the outcome of these studies. Nevertheless, the findings of Takeda et al. broaden our understanding of the effect of cardiovascular risk factors on amyloid pathology, and raise the possibility that therapeutic interventions targeted to cerebral blood vessels may be beneficial in Alzheimer disease.


    . Impaired endothelium-dependent responses and enhanced influence of Rho-kinase in cerebral arterioles in type II diabetes. Stroke. 2005 Feb;36(2):342-7. PubMed.

    . SOD1 rescues cerebral endothelial dysfunction in mice overexpressing amyloid precursor protein. Nat Neurosci. 1999 Feb;2(2):157-61. PubMed.

    . Nox2-derived radicals contribute to neurovascular and behavioral dysfunction in mice overexpressing the amyloid precursor protein. Proc Natl Acad Sci U S A. 2008 Jan 29;105(4):1347-52. PubMed.

    . Mouse models of the metabolic syndrome. Dis Model Mech. 2010 Mar-Apr;3(3-4):156-66. PubMed.

    View all comments by Costantino Iadecola

Make a Comment

To make a comment you must login or register.


News Citations

  1. Special Issue Explores Link Between Metabolic Disease and Dementia
  2. Aβ Oligomers and Synaptic Dysfunction—Blame It on RAGE?

Paper Citations

  1. . Different patterns of cerebral injury in dementia with or without diabetes. Arch Neurol. 2009 Mar;66(3):315-22. PubMed.
  2. . Vascular risk factors: imaging and neuropathologic correlates. J Alzheimers Dis. 2010;20(3):699-709. PubMed.

Further Reading

Primary Papers

  1. . Diabetes-accelerated memory dysfunction via cerebrovascular inflammation and Abeta deposition in an Alzheimer mouse model with diabetes. Proc Natl Acad Sci U S A. 2010 Apr 13;107(15):7036-41. PubMed.