Cystatin C (cysC), a 13 kDa soluble protein, circulates in the plasma and is a major component of cerebrospinal fluid (CSF), where it carries out myriad functions. One of its main functions is to inhibit cysteine proteases such as cathepsin B (catB). Cathepsins are primarily localized in the endosomal/lysosomal system, where they degrade internalized peptides and proteins, but they also play a role in degradation of extracellular matrix in some disease states. CysC is primarily in the circulation, so it serves to strongly suppress unwanted proteolytic activity from escaped or released cathepsins. CysC, and particularly a single-point mutant L68Q, occasionally deposit as fibrillar aggregates in the cerebrovasculature, causing a disease called cerebral amyloid angiopathy. CysC has also been found associated with beta-amyloid (Abeta) plaques in Alzheimer’s patients. Several reports suggest that cysC inhibits Abeta fibrillogenesis, but rigorous experimental evidence is lacking. Whether cysC- Abeta interaction has biological consequences has not yet been definitively determined.
Human cysC has a single strongly curved 5-stranded beta-sheet and a single alpha-helix that bundles against the sheet. The active portion of the protease inhibitor is in the two loops at the end of the beta-sheet. About a third of the protein is disordered; this portion is on the opposite side of the protein from the active domain. There are 4 cysteines and 2 disulfide bonds per monomer. In CSF, the protein is believed to be monomeric, but excursions in temperature or pH can induce formation of a domain-swapped dimer that lacks protease inhibitory activity. Crystal structure studies have been complicated by the invariable appearance of this domain-swapped dimer; monomer crystal structure has been obtained only with a stabilized mutant. Conventionally, human cysC is purified from urine, but it is expensive and has a very ragged N-terminus.
Our objective in this work is to systematically characterize the interaction of cysC with Abeta. We successfully established an intein-based recombinant system to produce purified protein at good yield. The recombinant protein was characterized by standard methods, including mass spectrometry, gel electrophoresis, circular dichroism, and enzyme inhibition. By all these measures, the protein was natively folded, monomeric, and active. However, by crosslinking/gel electrophoresis and dynamic light scattering, we detected definitive evidence of oligomers. We determined that oligomers form at concentrations above ~20 µM. Furthermore, as the concentration increased, the oligomers grew from small globules to long flexible filaments that are not amyloid. We developed a technique to fractionate oligomers and monomers, and demonstrated that the oligomers are not linked by non-native disulfide bonds, and are not domain-swapped. Once formed, the oligomers are stable to dilution but unstable to SDS. Importantly, the oligomers retain native fold and protease inhibitory activity. We hypothesize that the oligomers are formed via interactions of the disordered domain of cysC.
We next examined the interaction of cysC with Abeta using several techniques, including crosslinking, fluorescence, light scattering, and electron microscopy. Our data show that both monomeric and oligomeric cysC bind to Abeta, but they differ significantly in their effect on Abeta aggregation. Specifically, monomeric cysC accelerates the rate of growth of Abeta but changes the morphology to off-pathway amorphous aggregates. In sharp contrast, oligomeric cysC completely stops Abeta aggregation. We interpret these results in terms of a generalized concept of homotypic and heterotypic interactions among amyloidogenic proteins.