As with industrious employees, it is hard to spur greater productivity in an enzyme by simply cracking the whip. However, clearing distractions might do the trick. Taking this strategy into an Alzheimer disease mouse model, scientists have boosted the activity of an Aβ-degrading enzyme by reducing levels of an endogenous inhibitor. The enzyme, cathepsin B (CatB), caught the attention of Li Gan at the Gladstone Institute of Neurological Disease in San Francisco several years ago when it appeared to prevent buildup of amyloid plaques in the brains of AD mice overexpressing mutant human amyloid precursor protein (APP). In the October 23 Neuron, Gan and colleagues now report that APP mice lacking cystatin C (CysC)—an inhibitor of cysteine proteases including CatB—have lower soluble Aβ levels and reduced Aβ-associated deficits in cognition, behavior, and synaptic plasticity compared to APP/CysC+/+ mice. By crossing the animals onto a CatB-null background, they show that these benefits depend on CatB. In light of prior human polymorphism data and recent mouse data highlighting cystatin C’s ability to interfere with amyloid aggregation, the new findings suggest a more complex character for cystatin C and make it challenging to say whether it plays a “good” or “bad” role in AD pathogenesis.

In the same previous study that found greater amyloid deposition in CatB-deficient APP mice (see ARF related news story), Gan’s group showed that CatB trims Aβ1-42 at its C-terminus, leaving behind smaller, less toxic Aβ peptides. The scientists therefore reasoned that they could enhance CatB activity by removing an endogenous inhibitor, CysC. Indeed, when they examined hippocampal lysates from CysC+/+, +/- and -/- mice (Huh et el., 1999), first author Binggui Sun and colleagues found gene dose-dependent increases in CatB activity in the CysC-deficient animals. Next, they crossed the CysC knockouts with hAPP-J20 transgenic mice, a human mutant APP overexpressing line with increased Aβ42 production. At two to four months of age, prior to Aβ deposition, CysC-deficient APP mice had lower levels of total soluble Aβ and Aβ1-42. “That gave us confidence that we were really looking at Aβ metabolism without the effects of Aβ stabilization,” Gan told ARF. Both processes influence plaque formation, she noted, making it hard to distinguish these effects in older mice where amyloid deposition is well underway.

Consistent with their findings in young, predeposition animals, the researchers found reduced Aβ deposits in immunostained hippocampi from APP/CysC-/- mice, compared to APP/CysC+/+ animals, at five to eight and eight to 10 months of age. However, in experiments using thioflavin S, which predominantly labels Aβ fibrils and spares diffuse Aβ forms, they saw Aβ reduction in CysC-null APP mice at five to eight months but not at eight to 10 months of age. This suggested to the authors that CysC may affect soluble and fibrillar Aβ forms differently.

The Aβ trends found support in several lines of functional evidence. The researchers showed that CysC ablation in the APP mice prevented premature mortality and increased expression of calbindin (a neuronal marker whose decreased levels correlate with cognitive decline in APP-J20 mice). Furthermore, loss of CysC reduced Aβ-associated cognitive defects (revealed by Morris water maze tests) and restored long-term synaptic plasticity in hippocampal circuits involving the dentate gyrus.

To show that the benefits of CysC reduction are in fact mediated by bolstering CatB activity, the researchers crossed the APP mice (expressing +/+, +/- or -/- CysC) onto a CatB-null background. As predicted, the effects on calbindin immunostaining and hippocampal synaptic function in the CysC-deficient APP mice disappeared when CatB was out of the picture. In line with these observations, CysC deficiency had no effect on total soluble Aβ or Aβ1-42 levels in the hippocampus of APP mice lacking CatB. However, on the CatB-null background, hippocampal plaque load in four- to six-month-old APP/CysC-/- mice was about 50 percent higher than in APP/CysC+/+ animals. At first glance, this trend would seem to back recent work indicating a seemingly beneficial role for cystatin C in AD mice—as an agent that interferes with amyloid aggregation (see ARF related news story). But Gan thinks the new data are “completely consistent with the argument that cystatin C fulfills at least two functions”—inhibiting CatB and blocking aggregation by binding to Aβ. “Once you’ve taken CatB off the table, cystatin C’s other effects (i.e., its influence on Aβ aggregation) will be more prominent,” she told ARF.

More work is needed to tease out which mechanism plays a bigger role in the AD brain. Meanwhile, data on human polymorphisms thus far seem to point to a protective role for cystatin C in AD. As suggested by several studies (Benussi et al., 2003; Paraoan et al., 2004), a cystatin C variant (A25T) that is associated with higher AD risk appears to reduce production and secretion of the protein. Furthermore, in a study of 761 older men published last month in Neurology (Sundelöf et al., 2008), a 0.1-mumol/L drop in serum cystatin C between ages 70 and 77 years correlated with a 29 percent higher risk of incident AD. “In humans, the evidence is that reducing cystatin C levels puts you at risk for AD,” said Paul Mathews of the Nathan Kline Institute at New York University School of Medicine. “That would be more in line with the idea that small increases in cystatin C levels could be protective.”

Mathews noted in an ARF interview that differences in the AD mouse models used for the various cystatin C studies may be important. “If cathepsin B's activity is to chew back at the C-terminus of Aβ, particularly Aβ42, then a mouse with increased Aβ42 production (the APP-J20 mouse) will show the greatest effects when cathepsin B activity is increased,” he said. “I think that this may be a somewhat more artificial situation than that in the CNS of the APP23 and Tg2576 mice, where Aβ40:Aβ42 ratios are more like in the human.” Mathews was a coauthor on a recent study (Mi et al., 2007) led by Efrat Levy, also at Nathan Kline, which showed lower Aβ deposition in Tg2576 mice that overexpress cystatin C. Published at the same time, an independent study by Mathias Jucker’s group at the Hertie Institute for Clinical Brain Research in Tuebingen, Germany, found similar Aβ-reducing effects of cystatin C in APP23 mice (Kaeser et al., 2007).

Jucker found more inconsistencies with the apparent benefits of CysC reduction in the current study. He told ARF that his group detected no change in levels of mouse Aβ40 and Aβ42 in CysC-/- mice relative to CysC+/+ animals. Furthermore, Gan said her team found a mere 20 percent reduction (not statistically significant) in mice that overexpress human cystatin C. She believes these puzzling results could stem from the high physiological concentrations of CysC relative to CatB. The CatB-dependent effects would likely be masked in CysC-overexpressing mice because endogenous levels of the protein are already in relative excess, she said.

To see whether the current observations hold in an independent mouse model, Gan and colleagues are crossing the CysC-deficient mice with transgenics expressing wild-type APP. In the meantime, they are working to nail down where in the cell the CysC-CatB interactions take place. “It’s very surprising to us that when you take out cystatin C, a major cysteine protease inhibitor, you see no toxic effects,” Gan said. “Maybe the (CysC-CatB) interaction only happens in a compartmentalized way.” She suspects the endosome of being that place, since Aβ colocalizes with cystatin C there, and its low pH is optimal for CatB activity.

On the translational front, Gan’s group is planning a high-throughput screen for small-molecule compounds that inhibit CysC-CatB interactions—either by directly blocking the association or inducing a conformational change in one protein that affects its binding to the other. “That could be a pharmacological way to unleash the CatB-dependent Aβ degradation,” she said.—Esther Landhuis


  1. Cystatin C has numerous and diverse roles via mechanisms that are either dependent or independent of cathepsin inhibition. In vitro experiments have indicated that it can inhibit the cysteine proteases cathepsins B, H, L, and S. Cystatin C is expressed by all tissues studied and secreted into all body fluids. Because it is present in high concentration in the CNS, it has been suggested that it has an important role as a cysteine protease inhibitor. However, mechanisms that are independent of cathepsin inhibition such as neuroproliferation were previously demonstrated. Furthermore, we have demonstrated that cystatin C binds to amyloid-β and inhibits amyloid-β fibril formation and oligomerization. This was shown in vitro and in mouse models overexpressing human cystatin C. Cystatin C overexpression in these mice does not affect cathepsin B expression and activity. Moreover, the polymorphism in the cystatin C gene that was liked to increased risk of late-onset Alzheimer disease involves a moderate reduction in cystatin C secretion. Alternatively, the complete absence of cystatin C in knockout mice results in increased activity of cathepsin B and enables the identification of the possible role of this enzyme in amyloid-β degradation. The significance of such a role for cathepsin B in the human brain, where high levels of cystatin C are present, remains to be demonstrated.

  2. It seems that there are two mechanisms by which CysC may regulate cerebral β-amyloidosis. One of them we and Efrat Levy’s group described previously in Nature Genetics (Mi et al., 2007; Kaeser et al., 2007), where CysC interacts directly with Aβ fibril formation and deposition. The other is described in the present study, where cystatin C regulates the proteolysis of Aβ via CatB.

    The CST3 25Thr allele has been associated with an increased risk of AD. Mechanistically, it was suggested that the 25Thr variant disturbs intracellular cystatin C processing, resulting in impaired CysC secretion and reduced levels of extracellular cystatin C. This is consistent with the lower levels of CysC observed in the plasma of 25Thr allele carriers. These observations are rather in line with the findings in Nature Genetics and are somewhat in contrast with those of Li Gan et al., since the authors do not provide data showing that such an intracellular elevation of CysC would increase Aβ levels. In fact, one would have to challenge whether this retained cystatin C would still be functional in inhibiting proteases like CatB, because the overall processing seems to be impaired. However, this does not mean the findings of Li Gan are not relevant; it just may suggest that one mechanism dominates over the other in AD brains.

    Nevertheless, there are some intriguing questions to solve before the above conclusion can be drawn. First, we have, together with Paul Mathews and Anders Grubb, looked at endogenous mouse Aβ levels in CysC-/- mice and did not find any change in the level of mouse Aβ compared to wild-type CysC+/+ mice. Thus, after the work of Gan et al., one would have to conclude that the CatB effect is specific to human Aβ, which is very interesting but also unexpected. Hence, it would be nice to see what happens if mouse CysC is overexpressed. Second, we reported in our Nature Genetics work, in the supplementary information, that APP x CysC-/- mice show a tendency toward decreased Aβ plaque accumulation, while CAA is up severalfold in these mice. Unfortunately, it appears Gan and colleagues only looked at the plaques and thus may have missed an increase in CAA and maybe total Aβ. Third, it is not clear to us why the Aβ-immunoreactive plaque load but not the thioflavin S-positive one is decreased in APP-J20 CysC-/- mice in Figure 2, while the plaque load is increased by 50 percent on a CatB-null background in Figure 6 (although this increase was not significant). Fourth, Li Gan and colleagues thus far report results only in one mouse model (APP-J20), and therefore the results may be specific to this mouse model, whereas the direct interference of CysC with Aβ amyloidosis on the other hand was reproducible in several different double transgenic mouse models. Thus, it would be nice to see the interesting results of Li Gan replicated in another mouse model.


    . Cystatin C inhibits amyloid-beta deposition in Alzheimer's disease mouse models. Nat Genet. 2007 Dec;39(12):1440-2. PubMed.

    . Cystatin C modulates cerebral beta-amyloidosis. Nat Genet. 2007 Dec;39(12):1437-9. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Role Reversal—AD Mouse Desperately Seeks CatB
  2. Genetic Risk Explained: Cystatin C Staves Off Plaque Formation in Mice

Paper Citations

  1. . Decreased metastatic spread in mice homozygous for a null allele of the cystatin C protease inhibitor gene. Mol Pathol. 1999 Dec;52(6):332-40. PubMed.
  2. . Alzheimer disease-associated cystatin C variant undergoes impaired secretion. Neurobiol Dis. 2003 Jun;13(1):15-21. PubMed.
  3. . Unexpected intracellular localization of the AMD-associated cystatin C variant. Traffic. 2004 Nov;5(11):884-95. PubMed.
  4. . Serum cystatin C and the risk of Alzheimer disease in elderly men. Neurology. 2008 Sep 30;71(14):1072-9. PubMed.
  5. . Cystatin C inhibits amyloid-beta deposition in Alzheimer's disease mouse models. Nat Genet. 2007 Dec;39(12):1440-2. PubMed.
  6. . Cystatin C modulates cerebral beta-amyloidosis. Nat Genet. 2007 Dec;39(12):1437-9. PubMed.

Other Citations

  1. hAPP-J20

External Citations

  1. cystatin C

Further Reading


  1. . Serum cystatin C and the risk of Alzheimer disease in elderly men. Neurology. 2008 Sep 30;71(14):1072-9. PubMed.

Primary Papers

  1. . Cystatin C-cathepsin B axis regulates amyloid beta levels and associated neuronal deficits in an animal model of Alzheimer's disease. Neuron. 2008 Oct 23;60(2):247-57. PubMed.