This is Part 2 of a two-part story. See also Part 1.

17 January 2008. Part 1 of this story covered the growing evidence linking diabetes with Aβ pathology and cognition. But what about the other major hallmark of Alzheimer disease, neurofibrillary tangles? Researchers led by Akihiko Takashima at the Institute for Physical and Chemical Research, Saitama, Japan, used a mouse model to ask whether diabetes may be linked to tau hyperphosphorylation, a prelude to tangle pathology. First author Emmanuel Planel and colleagues used the cancer drug streptozotocin (STZ), which destroys insulin-secreting islet cells in the pancreas, to induce experimental diabetes in 3- to 6-month-old mice. At 10-day intervals for the next 40 days, the researchers examined the mice for signs of tau pathology, using antibodies to probe for phosphorylation at multiple epitopes. The researchers found that initially, tau hyperphosphorylation increased slightly over the first 30 days, but by day 40 it had risen robustly in brain extracts. Modified sites included serine (S) 202/S205 (AT8 antibody), S199, threonine 231(AT180), S262, S356, S396/S404 (PHF-1), S400, and S422. Though total tau levels did not change, phosphorylation increases ranged from threefold in the case of S262, to 40-fold in the case of AT8, an early marker of tau dysfunction, to over a hundredfold in the case of S400 and S422, which are also phosphorylated in AD brain. However, the researchers found no signs of tau aggregation in the hippocampus. Tau protein remained localized to axons as opposed to being redistributed to neuronal soma and dendrites, as seen in AD. “Thus, insulin dysfunction appears to induce axonal and neuropil tau hyperphosphorylation, reminiscent of incipient AD, but without observable somatodendritic relocalization,” write the authors.

What explains this increased tau phosphorylation? Planel and colleagues examined the mouse brains for changes in activity of several likely kinases and phosphatases. Of GSK-3β, Cdk5, MAPK, JNK, and CamKII kinases, only the last was activated by day 10 when tau phosphorylation started creeping upward. And though inhibitory phosphorylation (S9) of GSK-3β and activating phosphorylation of JNK were apparent at day 40, the overall pattern of kinase change did not match that of tau phosphorylation. “Thus, specific kinase overactivation is probably not the primary mechanism leading to tau hyperphosphorylation at either 10 or 40 d after STZ treatment,” write the authors. Instead, they found the protein phosphatase 2A to be less active after 10 days of STZ treatment, and suggest that tau hyperphosphorylation during diabetes results from PP2A inhibition. The investigators attribute this effect mainly to STZ-induced hypothermia, since the rampant over-phosphorylation did not occur when the animals stayed in a heated chamber to keep their body temperature at 37 degrees. Interestingly, the mice maintained their normal body temperature for the first 26 days following STZ treatment, suggesting that insulin dysfunction first causes mild tau phosphorylation via a temperature-independent mechanism and then a rampant phosphorylation that does depend on temperature. It is not clear how hypothermia connects to tau phosphorylation. One hypothesis is that it may have to do with mitochondrial dysfunction, since these organelles play a major role in thermoregulation. It has been postulated that impaired insulin signaling may lead to mitochondrial impairment, which may, in turn, be linked to Aβ and tau pathology (see recent review by Rhein and Eckert). This is not the first time that thermoregulation has been linked to tau phosphorylation. Planel and colleagues previously showed that anesthesia causes tau hyperphosphorylation via hypothermic inhibition of phosphatase activity (see ARF related news story), and reversible tau phosphorylation has been linked to hibernation, which is normally accompanied by a drop in body temperature (see Arendt et al., 2003).

Whether or not hypothermia might link diabetes to dementia in humans remains to be seen. In an interview, Craft cautioned that though the model used by Planel and colleagues is good for type 1 diabetes, there are no good mouse models of type 2 diabetes (T2DM), which is more relevant to AD. “I think the peripheral component is important, as well as the brain component. In type 2 diabetes, you have high levels [of insulin] in the periphery, you have a lot of inflammation, and you have lipid abnormalities. You don’t really get those with type 1 diabetes,” she said.

The importance of peripheral and also vascular effects is one issue the field is grappling with. For example, Craft presented data at the 2007 annual conference of the Society for Neuroscience last November in San Diego, which suggests that the link between T2DM and dementia may be related to circulatory problems (see Abstract). Her group has compared dementia in people with and without T2DM. She found that patients with this form of diabetes suffer greater cognitive impairment at lower plaque burdens, and, interestingly, these patients also have more microvascular lesions and higher soluble Aβ and neurofibrillary tangles. “It appears that these patients are more vulnerable to toxic effects at lower plaque burdens,” said Craft.

Contrasting studies taking into account peripheral and cerebrovascular changes are others that claim that insulin signaling in the brain itself is compromised. Research headed by Konrad Talbot of the University of Pennsylvania in Philadelphia has shown that as people progress from non-demented to mild cognitive impairment to full-blown AD, their brain insulin receptors become progressively inactivated. That data was also presented at this year’s SfN meeting (see Abstract). The scientists used phospho-specific antibodies to assess insulin receptor activation in postmortem tissue from the University of Pennsylvania and also from the Religious Order Study at Rush University in Chicago. They looked at two different phosphorylation sites on the insulin receptor (IR): one in the catalytic site and one in the site that binds to insulin receptor substrate 1, a protein that mediates downstream signaling. The scientists found that both sites are inactivated in MCI (n = 30) and even more so in AD patients (n = 30).

If the insulin receptor shuts off in AD, what happens to downstream molecules? “This is where the story gets complicated,” said Talbot. “We thought that the whole downstream signaling system would be impaired, just as it is in diabetes, but in fact, phosphorylation is markedly up,” he said. Phosphorylated kinases include phosphatidylinositol-3-kinase, AKT, and GSK-3β at the inhibitory serine 9. “We are not the only people who have seen this. A very important paper by Rebecca Griffin working in Cora O’Neill’s lab in Ireland showed that there is increased activation downstream,” said Talbot (see Griffin et al., 2005). Talbot suggested that this may be a compensatory mechanism to correct for decreased IR activation.

Greg Cole at the University of California, Los Angeles, said he considers these findings compelling. Cole has studied if similar signaling abnormalities occur in mouse models of AD. He told ARF that his group sees exactly the same kind of changes in the triple transgenic model created by Frank LaFerla’s group at University of California, Irvine, and in the 5x transgenics created by Robert Vassar’s group at Northwestern University, Evanston, Illinois. “This suggests that, for starters, the phenomenon may be related to Aβ,” Cole said.

How might Aβ influence insulin signaling? Cole, Craft, and Talbot all said that recent work from Bill Klein’s lab at Northwestern University is germane. Klein’s group has shown that Aβ oligomers can alter the distribution of insulin receptors in the cell (see ARF related news story). Work from Dennis Selkoe’s lab at Harvard Medical School has shown that Aβ can also directly affect the insulin receptor, preventing autophosphorylation (see same ARF related news story). Cole is not convinced that the downstream signaling effects hinge on insulin, in part because there is little insulin in the brain. “Insulin-like growth factor 1 (IGF-1) may be more important, and there is a fair amount of that in the brain,” Cole said.

Cole agrees with Talbot that the observed changes in signaling molecules downstream of the insulin receptor are most likely compensatory. Cole also sees another side to compensation, one that works at the cellular level. “What you see in AD is a reduction in arbor in the face or neuronal loss,” he said. This differs from some other neurodegenerative diseases, such as Parkinson disease, where there is substantial neuron loss masked by a compensatory outgrowth. Failure to compensate may be due to a lack of trophic support, he suggested. Insulin is commonly used as a trophin in neuronal culture media, and may provide such support in vivo. Craft’s lab has shown, for example, that intranasal insulin can improve cognition in older adults (see Reger et al., 2005), which could be related to its trophic activity, suggested Cole “But it is hard to know whether this [effect of insulin] is because there is a deficiency or simply because you are supplying additional trophic support,” said Cole, adding that “the fact that you can get improvement in cognitive function is an indication that there is an opportunity to get some function out of the remaining neurons.”

Current data leave open the question of what might cause the decline in trophic support to begin with. Cole noted that proinflammatory cytokines cause resistance to trophic factors. For example, Carl Cotman’s group at University of California, Irvine, showed that interleukin 1β impairs trophic signaling in the brain (see Tong et al., 2007; Soiampornkul et al., 2008). TNFα, another such cytokine, has also been implicated in insulin resistance, and Talbot noted that it can activate IRS-1.

Could this link back to hematopoietic changes during diabetes, for example, increased inflammation? “The role of insulin in inflammation has not been well studied,” said Craft. Likewise, people are beginning to appreciate the role of insulin and insulin resistance in vascular function, she said, noting that the concept of the “neurovascular unit” may also be important. This unit refers to the way vascular function is coupled with the energy demands of the brain to ensure adequate blood flow and adequate glucose for whatever cognitive action is going on.

Craft suggested that the field needs to take this understanding beyond diabetes. “Diabetes is a neat construct that we are easily able to focus on, but we need to address the broader role of insulin and insulin resistance,” she suggested. Luchsinger expressed a similar sentiment. “I think something that has been neglected is how pre-diabetes is related to disease. Many people do not get diabetes but have cognitive problems. We often classify people as normal but they are not, because they may have insulin resistance or are obese. Investigators are now showing that the influence of diabetes and pre-diabetes is important even in teenage years,” Luchsinger said.

“People are beginning to appreciate the complexities,” agreed Craft. “If we understand how the different roles of insulin are interrelated, then we can begin to develop a more comprehensive model and then potential therapies, which is the ultimate goal,” she said.—Tom Fagan

See Part 1.


  1. Tau is a MAP and indirectly a thermometer, modified by temperature, stress, hypo- or hypermetabolism, etc. This is an interesting paper showing that modeling tauopathy for AD studies is tricky.

    View all comments by Andre Delacourte

Make a Comment

To make a comment you must login or register.


News Citations

  1. Diabetes-Insulin Roundup: Dementia Connection Grows Stronger, Part 1
  2. Anesthesia and AD: Phospho-Tau Surges in Sleeping Mice
  3. How Does Aβ Do Harm? New Clues on Insulin Signaling, Spines, Inflammation

Paper Citations

  1. . Effects of Alzheimer's amyloid-beta and tau protein on mitochondrial function -- role of glucose metabolism and insulin signalling. Arch Physiol Biochem. 2007 Jun;113(3):131-41. PubMed.
  2. . Reversible paired helical filament-like phosphorylation of tau is an adaptive process associated with neuronal plasticity in hibernating animals. J Neurosci. 2003 Aug 6;23(18):6972-81. PubMed.
  3. . Activation of Akt/PKB, increased phosphorylation of Akt substrates and loss and altered distribution of Akt and PTEN are features of Alzheimer's disease pathology. J Neurochem. 2005 Apr;93(1):105-17. PubMed.
  4. . Effects of intranasal insulin on cognition in memory-impaired older adults: modulation by APOE genotype. Neurobiol Aging. 2006 Mar;27(3):451-8. PubMed.
  5. . Interleukin-1 beta impairs brain derived neurotrophic factor-induced signal transduction. Neurobiol Aging. 2008 Sep;29(9):1380-93. PubMed.
  6. . Interleukin-1beta interferes with signal transduction induced by neurotrophin-3 in cortical neurons. Brain Res. 2008 Jan 10;1188:189-97. PubMed.

External Citations

  1. Abstract
  2. Abstract

Further Reading


  1. . Metabolic syndrome and dementia risk in a multiethnic elderly cohort. Dement Geriatr Cogn Disord. 2007;24(3):185-92. PubMed.
  2. . Coupling of mammalian target of rapamycin with phosphoinositide 3-kinase signaling pathway regulates protein phosphatase 2A- and glycogen synthase kinase-3 -dependent phosphorylation of Tau. J Biol Chem. 2008 Jan 4;283(1):100-9. PubMed.
  3. . Common pathological processes in Alzheimer disease and type 2 diabetes: a review. Brain Res Rev. 2007 Dec;56(2):384-402. PubMed.
  4. . Amyloid beta oligomers induce impairment of neuronal insulin receptors. FASEB J. 2008 Jan;22(1):246-60. PubMed.
  5. . Relation of diabetes to mild cognitive impairment. Arch Neurol. 2007 Apr;64(4):570-5. PubMed.

Primary Papers

  1. . Intake of sucrose-sweetened water induces insulin resistance and exacerbates memory deficits and amyloidosis in a transgenic mouse model of Alzheimer disease. J Biol Chem. 2007 Dec 14;282(50):36275-82. PubMed.
  2. . Insulin dysfunction induces in vivo tau hyperphosphorylation through distinct mechanisms. J Neurosci. 2007 Dec 12;27(50):13635-48. PubMed.