16 Jun 2016

In the June 1 Nature Communications, scientists led by Richard Morris (of water maze fame) at the University of Edinburgh, Scotland, report that young mutant APP transgenic mice that have not yet accumulated amyloid plaques learn a spatial navigation skill normally, but rapidly forget it. Their combined failure to consolidate the newly learned memory over the course of the following week, and to then retrieve it, is mirrored by weaker FDG glucose metabolism during this time. The finding offers a new testing paradigm that might guide efforts to detect in humans delayed forgetting at earlier stages of Alzheimer’s disease than current memory tests do, the authors suggest.

Most prior studies in APP/PS-transgenic mice focus on learning deficits, not long-term memory storage, claim the authors. How these finding relate to preclinical Alzheimer’s disease—which staging diagrams depict as having extensive amyloid accumulation years prior to cognitive deficits—remains to be seen. The authors maintain that a focus on testing delayed memory retrieval might help with diagnosis of preclinical AD by picking up deficits earlier in the time course of AD pathogenesis than current test batteries do. For example, the widely used “apple-penny-table” test uses a brief delay between learning and retrieval of only 20 minutes.

The study drew a mixed response. Researchers were surprised by the data; some agreed it could open a window to new preclinical tests, others cautioned that the findings might reflect developmental abnormalities or transgene effects rather than neurodegeneration.

Morris told Alzforum he decided to undertake this study because he was intrigued by the gradual shift in the field toward studying earlier markers of disease rather than end-stage pathology. With that in mind, first author Vassilios Beglopoulos and colleagues searched for subtle memory deficits in three- to four-month-old PDAPP mice. They trained 59 transgenic animals and 68 controls in the classic Morris water maze until the animals were able to find the hidden platform within 20 seconds. On average, the transgenic mice took about five days to master this skill, one day more than controls. Next, the authors tested the mice’s short- and long-term memories.

Ten minutes after reaching the learning criterion, the transgenic mice still performed as well as controls in a probe test. Alas, when tested again one week later, it was a different story. While the control mice remembered where the hidden platform was, the PDAPP animals had forgotten (see image at left). 

Since these transgenic mice have few plaques at this age, what might explain their memory lapse? To measure their neural activity Beglopoulos and colleagues used 2-deoxyglucose (2-DG) autoradiography, a postmortem measure of glucose uptake akin to fluorodeoxyglucose PET scans used in people. They injected radioactive 2DG at specific times and then quickly sacrificed the animals for analysis. While baseline 2DG binding across 32 brain regions was comparable between PDAPP mice and controls, the transgenic animals bound less during the delayed, seventh-day probe test. The reduction was predominantly in the midline cortical region and the hippocampus, where memory consolidation happens. The findings correlated reduced neural activity with the delayed memory deficit. Morris and colleagues believe the mice have trouble consolidating memories in the first place rather than retrieving a stored memory, because trained PDAPP mice bound less 2-DG than controls during the time up to the seventh-day probe test as well as during the probe test itself. This contradicts a recent finding of impaired memory retrieval in young APP/PS-1 transgenic mice (Roy et al., 2016). The authors report preliminary evidence that Aβ immunotherapy corrects the memory and metabolic deficits, suggesting that soluble Aβ may be to blame.

Glucose Uptake. PDAPP mice (bottom) show equally active neural metabolism at baseline (left) as controls (top), but impaired metabolism during a memory retrieval task (right). [Courtesy Richard Morris and Nature Communications.]

Could these subtle memory and metabolic effects reflect deficits in people with preclinical AD? “We are well aware of all the limitations in drawing such implications, but because the ADAS Cog, MMSE, and other memory tasks were all designed to test over a period of 30 minutes or so, I worry that we may be getting lots of false negatives,” said Morris. He believes, as do many in the field, that treating people prior to mild cognitive impairment may prove more successful and that more sensitive memory testing could help evaluate such therapies.  

Researchers were intrigued if not puzzled by the findings. Mathias Jucker, German Center for Neurodegenerative Diseases, Tuebingen, called it a fresh idea for the field. “I like that they assessed potentially relevant clinical readouts in the mice,” he wrote to Alzforum. (Jucker’s lab recently reported progress on such readouts with fluid markers; see Jun 2016 news). But he wondered if Morris’ result can be repeated in other models. 

Mony de Leon, New York University Medical Center, praised the paper. “It shows that considering the duration of delay intervals in memory paradigms is potentially of great importance,” he told Alzforum. He noted that memory studies have been pretty good at predicting who among people with mild cognitive impairment might be on the way to Alzheimer’s dementia, but that memory studies on normal aging have been less informative. “Incorporating a long delay might make the difference,” he said.

De Leon also noted a similarity between mice learning to set criteria and practice effects in people. Cognitively normal people get better on repeat testing, while people with preclinical AD tend not to show such practice effects. Richard Caselli, Mayo Clinic, Scottsdale, Arizona, agreed. He believes that erosion of the practice effect over time may be the earliest cognitive indicator of impending AD in ApoE4 carriers (see Caselli et al., 2013; Caselli et al., 2009). “The difference in the seven-day recall was reminiscent of the blunted test-retest effect we see in our longitudinal human cohort studies,” he wrote. De Leon noted that while such concepts have been around for a while, they have not been part of mainstream testing in the past and should be re-examined and exploited for preclinical trials.

De Leon said the study’s combining memory testing with glucose uptake analysis argues for considering the two together in people at risk. This is not always done in ongoing clinical studies. In recent years, amyloid PET has partly replaced FDG PET because FDG PET is less specific to AD and is sensitive to fluctuations from minor daily variation such as drinking coffee, etc.

Others were cautious in their interpretation. Takaomi Saido, RIKEN, Wako, Japan, told Alzforum that he believes the results are likely due to artifacts of overexpression of the APP transgene. Saido has developed knock-in APP models that are now being more widely used in the field (see Alzforum webinar). Eric Reiman, Banner Alzheimer’s Institute, Phoenix, wondered if the findings reflected early AD progression. More than a decade ago, Reiman and colleagues studied FDG autoradiography extensively in PDAPP mice. They found not only reduced uptake in posterior cingulate cortex and other regions compared to control animals, but that the mice had gross brain developmental abnormalities, losing up to a third of their posterior corpus callosum and hippocampus and almost all of the fornix (see Reiman et al., 2000; Gonzalez-Lima et al., 2001; Valla et al., 2008). “Finding early indicators of memory decline in transgenic animals, the biomarker measurements to which they are related, and an impact of an anti-amyloid treatment on some of these measurements is interesting and potentially important,” said Reiman.  “But based on our findings in PDAPP mice, and our findings in infants, children and young adults at genetic risk for AD, it would be helpful to know if these mouse deficits are due to a development problem or due to progressive neurodegeneration.” One way to test for the latter would be to test the animals in the delayed-learning paradigm when they are older to see if they do even worse, he said.

Morris acknowledged that he cannot rule out a contribution from structural brain changes in these mice. He considers a developmental explanation unlikely because learning, recall 10 minutes later, and baseline 2DG uptake were all normal in the three-month old mice, as was Nissl staining of the whole brain. Morris agreed that it will be important to determine whether the memory deficits detected at this young age, in PDAPP and other strains, will predict future decline. —Tom Fagan

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

Research Models Citations

1. PDAPP(line109)

News Citations

1. Blood NfL Looks Good as Progression and Outcome Marker 9 Jun 2016

Webinar Citations

1. Good-Bye Overexpression, Hello APP Knock-in. A Better Model? 11 Apr 2014

Paper Citations

1. Roy DS, Arons A, Mitchell TI, Pignatelli M, Ryan TJ, Tonegawa S. Memory retrieval by activating engram cells in mouse models of early Alzheimer's disease. Nature. 2016 Mar 24;531(7595):508-12. Epub 2016 Mar 16 PubMed.
2. Caselli RJ, Locke DE, Dueck AC, Knopman DS, Woodruff BK, Hoffman-Snyder C, Rademakers R, Fleisher AS, Reiman EM. The neuropsychology of normal aging and preclinical Alzheimer's disease. Alzheimers Dement. 2013 Mar 26; PubMed.
3. Caselli RJ, Dueck AC, Osborne D, Sabbagh MN, Connor DJ, Ahern GL, Baxter LC, Rapcsak SZ, Shi J, Woodruff BK, Locke DE, Snyder CH, Alexander GE, Rademakers R, Reiman EM. Longitudinal modeling of age-related memory decline and the APOE epsilon4 effect. N Engl J Med. 2009 Jul 16;361(3):255-63. PubMed.
4. Reiman EM, Uecker A, Gonzalez-Lima F, Minear D, Chen K, Callaway NL, Berndt JD, Games D. Tracking Alzheimer's disease in transgenic mice using fluorodeoxyglucose autoradiography. Neuroreport. 2000 Apr 7;11(5):987-91. PubMed.
5. Gonzalez-Lima F, Berndt JD, Valla JE, Games D, Reiman EM. Reduced corpus callosum, fornix and hippocampus in PDAPP transgenic mouse model of Alzheimer's disease. Neuroreport. 2001 Aug 8;12(11):2375-9. PubMed.
6. Valla J, Gonzalez-Lima F, Reiman EM. FDG autoradiography reveals developmental and pathological effects of mutant amyloid in PDAPP transgenic mice. Int J Dev Neurosci. 2008 May-Jun;26(3-4):253-8. PubMed.

Further Reading

No Available Further Reading