Immunoglobulin light chain (LC) amyloidosis (AL) is one of the most common types of systemic amyloidosis but there is no reliable in vivo model for better understanding this disease. Here, we develop a transgenic mouse model producing a human AL LC. We show that the soluble full length LC is not toxic but a single injection of pre-formed amyloid fibrils or an unstable fragment of the LC leads to systemic amyloid deposits associated with early cardiac dysfunction. AL fibrils in mice are highly similar to that of human, arguing for a conserved mechanism of amyloid fibrils formation. Overall, this transgenic mice closely reproduces human cardiac AL amyloidosis and shows that a partial degradation of the LC is likely to initiate the formation of amyloid fibrils in vivo, which in turn leads to cardiac dysfunction. This is a valuable model for research on AL amyloidosis and preclinical evaluation of new therapies.
Introduction
Systemic light chain (AL) amyloidosis is a rare but severe acquired protein misfolding disease characterized by deposition of amyloid fibrils composed of a monoclonal immunoglobulin light chain (LC) produced in excess by a B or plasma cell clone1. The LCs involved in amyloidosis have a propensity to aggregate into the characteristic β-sheet structure of amyloid fibrils, and accumulate in the extracellular compartments of tissues, leading to organ dysfunction. Renal and cardiac manifestations are the most frequent, the latter being associated with poor outcomes2. The molecular mechanisms that lead to the aggregation of LCs have been extensively studied, primarily in vitro due to the lack of other models reproducing the early stages of the disease. The variable domain (VL) of LCs has long been suspected to be the causative part of AL amyloid fibril formation since only a few VL germline genes account for most of the cases3,4 and destabilizing mutations acquired during affinity maturation in the V domain have been shown critical for amyloidogenicity in vitro5,6. Consequently, our knowledge on LC aggregation process has been mostly obtained from isolated amyloidogenic VL since full-length LCs seem to be resistant to aggregation under physiological conditions6,7,8. Accordingly, cryo-electron microscopy (Cryo-EM) structures of ex vivo AL amyloid fibrils confirmed that the cross β-structured interactions within the core of the fibrils are primarily established by the VL domains, occasionally augmented by a few amino acids from the N-terminal part of the CL9,10,11. The remaining portion of the LC, comprising most of the constant domain (CL), seems to be disorganized outside the fibrillar structure and partially or totally cleaved12. This prompts the question of whether proteolysis in the CL is required to initiate amyloidosis formation or just a subsequent degradation of non-fibrillar parts of the LCs. The multiple cleavage pattern in sites not accessible in the native dimers suggests a fragmentation subsequent to aggregation12. However, the high amyloidogenicity of some fragmented species, as opposed to the stability of full-length LCs, suggests that proteolysis of the LCs could also be required to initiate amyloidosis formation8.
In addition to the mechanical stresses caused by the accumulation of amyloid fibrils in tissues, the soluble form of LCs may also contribute to cardiac toxicity. Patients responding to treatments that aim at reducing circulating LCs show a significant decrease in NT-proBNP concentrations, correlated with improved cardiac function, in spite of the absence of a significant decrease in the amyloid burden13. Studies conducted in vitro using cardiomyocyte and cardiac fibroblast cultures, as well as in C. elegans and Zebrafish models, support this theory14,15,16,17,18. Exposure of these models to soluble amyloidogenic LCs leads to cellular stress through internalization of the LC increased production of reactive oxygen species (ROS) and the activation of a non-canonical MAPK pathway. As a result, lysosomal dysfunction, autophagy impairment, and mitochondrial damage were observed, followed by cell death. Although these studies provide valuable insight into LC toxicity on cardiac cells, they neither reproduce the cellular complexity, the microenvironment, and the shear stress observed in human tissue, nor the deposition and accumulation of amyloid material composed of the LCs. Translating these findings to human physiology remains challenging.
Several approaches to creating rodent models of AL amyloidosis have been attempted over the years19: Firstly, by the massive injection of Bence-Jones proteins, purified fibrils from patients or the so-called “amyloidoma” composed of a crude grinded human tissue containing amyloid material20,21. Although localized amyloid material can be observed in these models, their application is limited to therapeutic studies since they poorly reproduce the human pathophysiology and organ involvement. Lately, transgenic approaches to produce amyloidogenic LCs endogenously in rodents with mostly disappointing results22,23. One of them succeeded at reproducing amyloid deposition, even though deposits were localized in the stomach of aged mice, suggesting a destabilization of LCs in the local acidic environment which promoted aggregation rather than a physiological amyloid formation in classically involved organs24. This apparent resistance to amyloidosis in mice was also observed for other types of systemic and localized amyloidosis and was attributed to better proteostasis and a rapid turnover of proteins25,26,27. One of the main limits of these models is the levels of circulating free LCs (fLC) which are much lower than those observed in patients, likely failing to reach the threshold needed to initiate fibril formation28. To overcome this limit, we have developed a unique transgenic approach that allows to achieve high levels of circulating pathological LCs into the mice29. This strategy has been successful for modeling non-amyloid monoclonal LC-related deposition diseases affecting the kidneys, including light chain deposition disease and light chain renal Fanconi syndrome30,31.