There is considerable epidemiological and animal data suggesting that testosterone protects neurons. For example, the Baltimore Longitudinal Study of Aging found that lower circulating levels of the hormone are associated with increased incidence of AD (see Moffat et al., 2004), backing up an earlier cross-sectional study (see Hogervorst et al., 2001). Pike and colleagues found an inverse relationship between brain testosterone and soluble Aβ in older men with neuropathological AD (see Rosario et al., 2009). In male animal models, androgen depletion accelerates AD-like pathology, while testosterone replacement reverses it (see ARF related news story). But while the evidence for a beneficial effect of testosterone grew, so did the question of whether it worked directly or had to be converted to estrogen first.

Believing the latter to be more likely, Li and colleagues set out to prove it by crossing APP23 mice with a heterozygous aromatase knockout (Ar+/-). Earlier, she found accelerated AD pathology in female crosses (see Yue et al., 2005), but in the male crosses it was the exact opposite. “It was a complete surprise,” she told ARF.

The APP23/Ar+/- animals have no detectable aromatase activity. Joint first authors Carrie McAllister, Jiangang Long, and colleagues found that both plasma and brain testosterone levels were higher compared to APP23 controls. “Looking back, it was clear that these animals probably had elevated testosterone, because they fought a lot and could not be housed in the same cage unless from a very early age,” said Li.

Aromatase knockout also protected the animals from AD pathology. APP23 animals typically have elevated Aβ42/40 levels by nine months and easily detectable cortical plaques by 18 months. Aβ ratios in nine- and 18-month-old APP23/Ar+/- animals were the same as at three months, however, and lower than in the APP23 controls. At 18 months, the crosses contained hardly any Aβ plaques and no neuronal loss as judged by NeuN staining, whereas APP23 animals did have some neuronal loss in the CA3 field of the hippocampus. The crosses outperformed APP23 mice in the hole-board maze test of spatial memory (see He et al., 2007), which could be due to the lack of Aβ pathology. However, the crosses also outdid wild-type mice, suggesting that the testosterone boost aided cognition irrespective of the APP transgene.

How does the testosterone boost prevent amyloid pathology? “We know that androgens can enhance Aβ clearance via neprilysin,” noted Li. The neprilysin gene promoter contains an androgen response element, and sure enough, neprilysin levels and activity were both up in the crosses compared to APP23 animals. Surprisingly, BACE expression was lower. The BACE gene contains an estrogen response element that suppresses expression, making Li expect increased BACE expression sans estrogen, if anything. “We don’t know if androgens cross-react with those binding sites,” Li said, but she thinks there may be specific androgen and estrogen effects. She plans to characterize the BACE promoter to see if it has binding sites for androgen receptor complexes.

Pike cautioned that this work does not rule out a beneficial role for estrogen in the male brain. “I’m not sure you want to go after aromatase inhibitors as a therapeutic strategy. That would get you elevated levels of testosterone, but you don’t want to deplete the brain of estrogen,” he said. Li said she is collaborating with a Japanese research group to study the potential benefit of boosting brain-specific aromatase in females, which would boost estrogen and protect the brain. For males, Pike suggested that selective androgen receptor modulators (SARMs) may be the way of the future. These could boost androgen responses in specific tissues such as the brain, without evoking deleterious effects in other organs, such as the prostate. “You want to be sure androgen effects are mediated by androgen receptors. Li’s paper helps by showing, very importantly, that testosterone can have protective effects in the absence of estradiol,” said Pike.

In a similar vein, Craig Atwood, University of Wisconsin, Madison, raised the possibility that the effects might be mediated by other hormones, including luteinizing hormone (LH). His group has shown that LH promotes APP processing (see Bowen et al., 2004), and more recently, others showed reduced amyloid load in LH receptor knockout mice (see Lin et al., 2010). “I was expecting that the increase in testosterone would be accompanied by a decrease in LH, and that that would have explained Li’s results,” said Atwood. “But instead, there is actually an increase in LH in this [Ar+/-] animal. The data are getting difficult to reconcile.” Atwood noted that it is unknown what happens to LH in the APP23 background. Li said she plans to study LH, follicle stimulating hormone, and others. “We want to look at a whole hormone panel to see if the balance is the final key,” she said.

One thing Li, Pike, and Atwood agreed on was that the study takes the spotlight off estrogen. “This is great news for us guys because so many studies have focused on estradiol,” said Atwood. Pike agreed. “For years estradiol got all the glory as being the neuroprotective drug, but in this case, this is more evidence that it is testosterone.”

“People tend to forget about andropause,” said Li, “but now maybe it will get attention and attract more funding.” The NIA is currently enrolling volunteers for a study that will examine the effects of testosterone and an aromatase inhibitor in men age 65 or older.—Tom Fagan.

Reference:
McAllister C, Long J, Bowers A, Walker A, Cao P, Honda S-I, Harada J, Staufenbiel M, Shen Y, Li R. Genetic targeting of aromatase in male amyloid precursor protein transgenic mice down-regulates beta-secretase (BACE1) and prevents Alzheimer-like pathology and cognitive impairment. J. Neurosci 2010 May 26; 30:7326-7334. Abstract