Cell Damage in Light Chain Amyloidosis

Light chain (AL) amyloidosis is an incurable human disease characterized by the misfolding, aggregation, and systemic deposition of amyloid composed of immunoglobulin light chains (LC). This work describes our studies on potential mechanisms of AL cytotoxicity. We have studied the internalization of AL soluble proteins and amyloid fibrils into human AC16 cardiomyocytes by using real time live cell image analysis. Our results show how external amyloid aggregates rapidly surround the cells and act as a recruitment point for soluble protein, triggering the amyloid fibril elongation. Soluble protein and external aggregates are internalized into AC16 cells via macropinocytosis. AL amyloid fibrils are shown to be highly cytotoxic at low concentrations. Additionally, caspase assays revealed soluble protein induces apoptosis, demonstrating different cytotoxic mechanisms between soluble protein and amyloid aggregates. This study emphasizes the complex immunoglobulin light chain-cell interactions that result in fibril internalization, protein recruitment, and cytotoxicity that may occur in AL amyloidosis.

Light chain (AL) 2 amyloidosis is a protein misfolding disease characterized by extracellular deposition of immunoglobulin light chains (LC) as amyloid fibrils. LC proteins are comprised of two distinct domains: the variable (V L ) and constant (C L ) domains (also called LC full-length (FL) protein to differentiate with the V L domain). In patients with AL amyloidosis, the LC aggregate and deposit in vital organs, causing organ failure and death (1). The factors governing deposition in individual tissues are unknown. Patients with cardiac AL amyloidosis have the worst prognosis, with a median survival of less than a year (2,3).
The finding that the V L was the primary component of amyloid fibrils influenced previous biophysical studies (4,5). Recent proteomic studies have demonstrated that amyloid deposits are likely heterogeneous in nature and can be formed by FL, V L , C L , or mixtures of all types of LC fragments (6 -8). Thermodynamic studies proposed a stabilizing role for the 3C L domain in the stability and a modulating effect on fibril formation (9). Recently, our laboratory has demonstrated that the C L domain modulates the amyloid formation reaction but has no effect on the stability of the protein (10).
Within the amyloid field, it is widely accepted that oligomeric species are potentially more toxic than mature fibrils (17)(18)(19)(20). However, toxicity associated with amyloid fibrils may also be pathologically relevant. Engel et al. (21) described a mechanism in which growth of islet amyloid associated polypeptides fibrils is responsible for membrane disruption. Gharibyan et al. (22) demonstrated that lysozyme amyloid fibrils induce cell death. LC amyloid deposits are proposed to be the most common cause of amyloid cardiomyopathy (2, 23); 6 LC amyloid fibrils, but not the soluble precursor proteins, severely impair AC10 cardiomyocyte metabolism (24).
Our laboratory has compared the internalization rates of recombinant LC proteins. Levinson et al. (25) demonstrated that all proteins studied shared a common internalization pathway into lysosomal compartments.
In the present work, we have studied the mechanism of internalization into human AC16 cardiomyocytes of an amyloidogenic AL-09 protein and the non-amyloidogenic control I O18/O8 (IGKV 1-33) (hereafter called I for simplicity). Both soluble and fibrillar species and the FL and V L proteins have been compared by using real time live cell image analysis. Using endocytic inhibitors, we elucidated the mechanism of internalization of soluble LC and fibrils. Cell-mediated seeding of FL and V L was shown by incubating preformed aggregates with soluble protein in the presence of AC16 cardiomyocytes. Soluble protein and, to a larger extent, fibrillar aggregates induce cytotoxicity in cultured AC16 cells; however, the toxic effect was mediated via different mechanisms.
Our study highlights the complex aspects behind LC internalization and cytotoxicity in AL amyloidosis, underlying the importance of the amyloid fibrils in the process. Our experiments model the cellular mechanisms that may occur during the early events in AL amyloidosis.

Results
Soluble LC Internalize into Human Cardiomyocytes in a Sizedependent Manner-Using real time live cell imaging, we followed the kinetics of soluble protein internalization in live cells without external perturbation. Fig. 1A shows that the Oregon Green (OG) conjugated AL-09 V L ( OG AL-09 V L ) and I FL ( OG I FL) proteins associate with and are increasingly internalized into human AC16 cardiomyocytes over a 48-h period. As shown in Fig. 1B, all OG-labeled soluble proteins tested appeared inside cells after 4 h of incubation at 37°C and substantially increased over time. Both V L domains show a higher rate of internalization over time compared with the FL proteins as can be observed in Fig. 1B. After 12 h, OG LC soluble protein was found in the majority of the cardiomyocytes imaged (Fig.  1A). As a control, unconjugated OG was added to cell cultures and did not enter the cells. We do not see soluble protein accumulation on the plasma membrane. The internalized protein is predominantly accumulated and localized in perinuclear compartments as evidenced by confocal microscopic analysis (supplemental Fig. S1), in agreement with the co-localization studies conducted previously with HL-1 mouse cardiomyocytes (25).
To determine the importance of the C L on the internalization rates, we compared the kinetics of internalization of FL proteins with their V L counterparts. Red fluorescent protein (RFP)-AC16 cells were incubated with 1 M of either FL or V L domains. For proper comparison, the fluorescence intensity of each protein was normalized by its degree of OG labeling (see "Experimental Procedures"). Fig. 1B shows that V L domains have higher amounts of intracellular protein than FL proteins, suggesting a size-dependent correlation with the rate of inter- Because I FL has more OG binding sites than AL-09 V L , I FL appears more fluorescent than AL-09 V L . Green fluorescence signal was normalized for a correct quantification and comparison. B, quantification of soluble AL protein internalization over time. V L domains internalize faster than FL proteins. Also, both V L and FL AL-09 proteins internalize faster than the both I germlines. C, quantification of decrease in protein-associated fluorescence emission. Internalized AL soluble protein fluorescence decreases rapidly after the OG LC-rich medium is replaced with OG LC-free medium. D, representative images of AC16 cells with decreasing amounts of intracellular OG I FL protein fluorescence after 12 and 24 h of OG LC-rich medium wash. Green fluorescence intensities were normalized for each protein as a function of their degree of labeling. Samples were set up in triplicate in four independent assays (n ϭ 4) with the average values and error bars as means Ϯ S.E. *, two-tailed t test; p value Ͻ 0.05. nalization. As shown in Table 1, the amyloidogenic protein AL-09 internalizes faster than the germline I, for both V L and FL proteins (although the differences are not statistically significant between AL-09 V L and I V L ; AL-09 FL and I FL, see Table 1 for details). For a correct quantification of the protein internalized, OG LC-rich medium was replaced with proteinfree medium before each live cell imaging time point. After 24 h, the OG LC medium was replaced with medium alone. We followed the trafficking of the fluorescent protein for an extended period of time. We observed substantially decreased intracellular protein fluorescence over time (Fig. 1, C and D). We ruled out fluorescence quenching or signal degradation as a potential explanation for the reduction of fluorescence because we observed intracellular fluorescence in cells that have been incubated with OG LC for 48 h (see "Experimental Procedures").
We suggest that the internalized protein leaves the AC16 cells via an exocytosis mechanism. Extracellular fluorescence is diffused and not detected by our imaging system. Protein Internalization Is Mediated by a Macropinocytic Mechanism-To understand the mechanism of LC internalization, studies with endocytosis inhibitors were performed. Before OG LC proteins were added, RFP-AC16 cells were treated with: Dynasore (DYN), a dynamin-GTPase inhibitor that blocks clathrin-mediated endocytosis (26 -28); tetradecyl trimethyl ammonium bromide (MiTMAB), a dynamin inhibitor that specifically blocks receptor-mediated endocytosis in nonneuronal cells (29,30); genistein (GEN), a GTPase inhibitor that prevents the clathrin-independent endocytic pathway; and cytochalasin D (CYT), an inhibitor of actin polymerization required for membrane-ruffling and macropinosome formation (31).
After 28 h of cell treatment with 50 M DYN, both V L domains (AL-09 and I) decrease their uptake over 25% in comparison with the untreated controls at the same time point (Fig.  2, A and B), whereas the FL protein internalization remained practically unaltered (Fig. 2, C and D). Cell treatment with 10 M MiTMAB decreased the internalization for both V L and FL proteins. The effect reached 60% inhibition for both germline I proteins (Fig. 2, B and D), and ϳ50% inhibition for both AL-09 proteins (Fig. 2, A and C, green bars). In contrast, 50 M DYN, MiTMAB, and GEN had no significant effect on protein internalization, whereas CYT reduced protein fluorescence by 75% in all cases. E, representative images of AC16 cells with intracellular OG I FL protein after 28 h of incubation. DYN and GEN show similar green fluorescence intensity compared with no treatment. MiTMAB shows a slight reduction in fluorescence, which correlates with the decreased number of cells. CYT clearly shows a significant reduction of intracellular green signal, which indicates the inhibition of a macropinocytic mechanism. Samples were set up in triplicate in three independent assays (n ϭ 3) with the average values and error bars as means Ϯ S.E. *, two-tailed t test; p value Ͻ 0.05 with respect to the corresponding controls. #, two-tailed t test; p value Ͻ 0.05 between values at same condition.

TABLE 1 LC protein internalization rate
The samples were set up in triplicate in four independent assays (n ϭ 4). The data on the table are means Ϯ S.E. determined by two-tailed t test. The p values are Ͻ0.05 between AL-09 V L and AL-09 FL and between AL-09 V L and I FL. GEN does not show any substantial effect on the internalization of the four variants compared with the untreated cells (Fig. 2,

LC protein
Relative to the respectively cell number, cell treatment with DYN and GEN increased the uptake of FL proteins. Cells treated with 1 M CYT significantly reduced the internalization rates to values less than 25% for both V L and I FL proteins and ϳ30% for AL-09 FL protein (Fig. 2, A-D). The images in Fig. 2E show changes in cell morphology and viability caused by treatment with the inhibitors in the presence of OG I FL. The observation that DYN and MiTMABs were causing a decrease in the cell growth could be interpreted as an inhibitory effect. CYT also decreases the cell viability by 50%; the green signal reduction is not completely related to the reduced number of cells in the wells but rather to the inhibition of the internalization process. When we compared the quantified cardiomyocyte viability (percentage of red cells) (Fig. 2, A-D, red bars) with the internalized protein (Fig. 2, A-D, green bars), we were able to assess the real effect of each inhibitor on the AL internalization process. As shown in Fig. 2E with OG I FL protein, for example, the green signal inside the cells is practically nonexistent in the presence of CYT. MiTMAB treatment, to a lesser extent, also decreased the protein internalization. External Fibrillar Aggregates Interact with Cell Membranes and Recruit Soluble Protein-After examining the endocytic pathway for soluble LC protein, we sought to describe the behavior of amyloid fibrils incubated with AC16 cardiomyocytes. OG-labeled fibrils composed of I FL were added to RFP-AC16 cells, and their growth was monitored over 48 h (Fig. 3). Unlike LC soluble protein, external aggregates show a strong attraction to the plasma membrane. Fibrils rapidly surrounded the cardiomyocytes and promoted cell clustering and confined cell growth. Fig. 4 demonstrates a significant reduction in OG I FL fibril internalization by CYT, indicating that fibril uptake also occurs predominantly via macropinocytosis. As seen for soluble protein, MiTMAB treatment slightly decreased the intracellular green signal, suggesting a possible secondary pathway.
We then asked whether AL fibrils could compete with and reduce AC16 internalization of the soluble protein. Conjugation with OG does not affect the LC fibril morphology, as observed by EM (data not shown). Fig. 5 shows OG I FL protein as a representative example of the competition experiments in which we incubated each of the species (soluble protein, amyloid fibrils, or their co-incubation) with the cells. Soluble OG I FL protein internalized and localized in perinuclear compart-ments as described above (Fig. 5A). When cells were incubated with OG I FL or non-labeled fibrils, fibrils were observed associated with the cells surface. Fibril labeling highlights the internalized and perinuclear fibrils (Fig. 5, B and C). Fig. 5D demonstrates that unlabeled fibrils become fluorescent, suggesting that amyloid aggregates act as recruitment points for soluble protein, potentially allowing cell-mediated amyloid fibril elongation. Fig. 5E describes how this behavior increased over time for each condition examined. No significant differences were observed between the V L and FL proteins, suggesting that the presence of the C L domain did not significantly affect amyloid fibril elongation. However, AL-09 FL soluble protein showed a delayed seeding effect (24 h). The three other proteins showed clear seeding after 8 h of co-incubation (Fig. 5E). Confocal microscopy experiments using OG AL-09 V L co-incubated with Texas Red-labeled fibrils in AC16 cells (not co-transfected with RFP protein) corroborated the co-localization of soluble AL-09 with the fibrils (supplemental Fig. S2).
LC Fibrils Are Toxic to AC16 Human Cardiomyocytes-We next tested the toxicity of both soluble proteins and amyloid fibrils in AC16 human cardiomyocytes. For these experiments, we included two more recombinant AL patient-derived proteins: AL-T05 V L and rV6Wil proteins (hereafter called Wil for simplicity), belonging to the V1 1b (IGLV 1-51) and V6 6a (IGLV 6 -57) LC family, respectively. AL-T05 V L has the fastest fibril formation kinetics of all  proteins tested in our laboratory (32); Wil has been extensively studied by Wall et al. (33,34). Wil fibrils have been recently identified to cause metabolic dysfunction in AC10 human cardiomyocytes (24). Fig. 6 shows the effect of soluble and amyloid fibrils on RFP-AC16 human cardiomyocyte growth rates. At concentrations of 1 and 12 M, soluble protein had no effect on cell growth rates (Fig. 6, A and B). Supernatant fractions of fibril formation reactions (which may contain oligomeric species or soluble aggregates) do not have any toxic effect on cells (data not shown). In contrast, fresh I V L and Wil 1 M fibrils (monomer equivalent concentration) prevented cell growth over 80 h of analysis; however, the other variants had no apparent effect on cell growth (Fig. 6C).
We next examined the effect of fibril kinetic stability on cell growth of AC16 cardiomyocytes. Incubation of fibrils at 4°C, followed by a freeze/thawing cycle, showed more inhibition of AC16 cell growth (Fig. 6, D and E). The FL protein fibrils were generally less inhibitory than their V L counterparts. The images in the insets of Fig. 6 visually demonstrate that RFP-AC16 cell viability decreases with increasing toxicity of AL-09 V L fibrils, as evidenced by a decrease in cell number (Fig. 6, C-E, insets). The structure of freshly prepared and incubated fibrils was assessed by transmission electron microscopy (TEM) (Fig. 7). The aggregate morphology is in agreement with what has been published by our laboratory previously, where small clusters of short fibrils aggregate and interact in large conglomerates, the only difference is that we are presenting low resolution images (for representative high resolution images of these aggregates, please review Ref. 65). Distinct morphological and apparent concentration differences are clearly seen between fresh and mature fibrils, which may explain the differences in cell toxicity. Freshly prepared fibrils formed large clusters of aggregates that broke up into smaller clusters over the incubation at 4°C and the freeze/thaw cycle. We propose that fibril toxicity changes over time and depends on the size and the level of clustering of the aggregates.
Caspase 3/7 activity was measured in the presence of the soluble protein and fibril-treated AC16 cells (Fig. 8) after 80 h of cell treatment (end point of toxicity experiments; Fig. 8, A-C). Cells incubated with 1 M soluble protein did not undergo apoptosis, as compared with the untreated cells (Fig. 8A). This result correlates with the cell growth curves. Cell treatment with 12 M soluble protein did not affect cell growth (Fig. 6B); however, both AL-09 V L and FL soluble proteins showed a significant increase in caspase activity (Fig. 8B). The cells incubated with 1 M fibrils did not increase their caspase activity (Fig. 8C). The decreased caspase activity found for I and AL-09 V L domains, AL-T05, and Wil fibrils, compared with control, is directly correlated with the reduced number of cells alive after 80 h of treatment (Fig. 6). Thus, the caspase activity for the highly cytotoxic fibrils was extremely low compared with the both non-toxic FL fibrils confirming the high cytostatic effect of LC fibrils. These results suggest that soluble protein can activate apoptotic events in human cardiomyocytes in a concentration-dependent manner, especially those variants that are more amyloidogenic.
ThT Is Not Able to Detect Cytotoxic Fibrillar Species-Fibrillar Wil was the most toxic species in our studies. Therefore, we conducted a titration study to determine the minimum concentration required to observe a toxic effect in AC16 cells. In vitro fibril formation reactions were followed by monitoring the fluorescence intensity of thioflavin T (ThT) dye, which is enhanced when ThT binds to amyloid fibrils (35). The ThT fluorescence emission decreased as a function of fibril concentration (Fig. 9A). We grew RFP-AC16 cells in the presence of the dilution series of Wil fibrils (Fig. 9B). We observed a decrease in cell growth at the highest concentrations of Wil fibrils. Fig. 9C shows that ThT fluorescence signal and percent-age of cell growth intersect between 0.2 and 0.4 M, where the cells grew 50% with respect to the control and the ThT fluorescence is barely above the buffer baseline. A series of cell images incubated with 1 M of Wil fibrils showed the effect of fibrils on RFP-AC16 cells from time 0 to 64 h (supplemental Fig. S3).  From these results, we conclude that low ThT signal in fibril formation does not necessarily reflect the absence of toxic fibrillar species and that other fibril detection methods should be employed in addition to ThT fluorescence, particularly at low fibril concentrations.

Discussion
In this study we demonstrate the cellular internalization of LC soluble proteins-AL-09 V L , FL, I V L , and I FL-and their corresponding amyloid aggregates. Soluble proteins and amyloid fibrils internalize via macropinocytosis. In addition, our results uncover that amyloid fibrils are one of the cytotoxic species responsible for the loss of AC16 cell viability, and the C L domain modulates AL protein internalization in addition to fibril cytotoxic behavior. We have observed a novel behavior where external aggregates attach to the cells, confine them, and trigger a seeding effect that is significantly accelerated when compared with in vitro seeding experiments.
The presence of the C L domain delays the internalization process, indicating a size-dependent mechanism, as described previously in cardiac fibroblasts (16). The amyloidogenic protein AL-09 internalizes faster than the germline I, which could be correlated with its lower thermodynamic stability and its higher amyloidogenic propensity (13,36). Soluble proteins do not accumulate on the plasma membrane. Rather, LC proteins are rapidly internalized into the cardiomyocytes without clear evidence of undergoing a membrane binding step, unlike the findings in human renal mesangial cells (15). Our experiments using endocytic inhibitors exclude any caveolin-mediated pathway for LC internalization. We suggest that the LC proteins tested are taken up into AC16 cardiomyocytes primarily through a macropinocytic pathway. The increased FL protein internalization, found when cells were treated with DYN or GEN, suggested that these inhibitors may favor other internalization pathways. Monis et al. (16) also found that CYT inhibits the LC internalization into primary cardiac fibroblasts. Macropinocytosis involves membrane ruffling events that occur in response to actin polymerization near the plasma membrane (37). Macropinosomes fuse with the cellular membrane and are rapidly transported along the endocytic pathway, merging with lysosomal compartments (37,38), as we reported in mouse HL1 cardiomyocytes (25).
Macropinosomes mediate the cellular internalization of external amyloid aggregates, as has been found in neurodegenerative diseases (39 -41). The size of amyloid aggregates would preclude any vesicular endocytosis. The inhibitor effect of MiT-MAB suggests a phagocytic mechanism as a secondary pathway for amyloid fibrils internalization (42). Macropinocytosis has also been associated with amyloid transcellular propagation (43). Many studies have reported a cell to cell transfer of misfolded protein and aggregates, triggering the progression of the neurodegenerative disease throughout the brain (39, 40, 44 -48). Per Westermark and co-workers (49) have provided evidence that serum amyloid A or secondary amyloidosis (AA) is a transmissible disease. In our study, we described an excretion mechanism by which the internalized protein decreases over time. Protein excretion may depend on low extracellular protein concentration. Gupta and Knowlton (50) first described the release of exosomes by human cardiomyocytes. In the context of amyloid propagation, exosomes are involved in transmission of misfolded and aggregated protein, and further, they are capable of entering cells via macropinocytosis (51, 52). Apoptosis reagents were added after 80 h of treatment and were incubated for 1 h at 37°C before reading plate. Caspase activity is indicated as relative fluorescence units (RFU). Samples were set up in triplicate in two independent assays (n ϭ 2) with the average values and error bars as means Ϯ S.E. *, two-tailed t test; p value Ͻ 0.05 with respect to the corresponding media controls. SEPTEMBER 16, 2016 • VOLUME 291 • NUMBER 38

JOURNAL OF BIOLOGICAL CHEMISTRY 19819
Amyloid fibrils have been considered to play a secondary role in cell toxicity, yielding the toxic role to the soluble species (12,13,53). LC fibrillary species were not considered in the toxicity landscape of AL amyloidosis until recently (24). Here we demonstrated the high cytotoxic potential of LC amyloid fibrils when incubated with AC16 cardiomyocytes. Interestingly, we observed a different mechanism of cell toxicity followed by amyloid fibrils and soluble protein. Whereas the LC amyloid fibrils exhibit an efficient inhibition of the cell growth and division, the LC soluble proteins allow cell growth but cause cell dysfunction and apoptosis in AC16 cardiomyocytes. The toxicity potential of both AL-09 VL and FL soluble LC could be correlated with their higher amyloidogenic propensity and faster cell internalization rate compared to the germline I LC. We confirmed that the concentration of monomeric species within the fibril sample did not increase as the cell viability diminished (using the fibril sedimentation assay reported by Wetzel and co-workers (54)), excluding any reversible fibril process that will shift the equilibrium toward formation of cytotoxic soluble species (data not shown).
The cytotoxic effect of amyloid fibrils increases as they age. TEM images clearly show morphological changes between fresh and aged fibrils, which may explain the differences in cell toxicity. Fragmented amyloid fibrils possess an enhanced cytotoxic potential when compared with longer fibrils (55)(56)(57). We propose that fibril toxicity changes over time as fibrils fragment into smaller fibril clusters, which could be related to the ease at which macropinosomes engulf aggregates into cardiomyocytes. This behavior is also in agreement with the possibility that certain amyloid fibril structures may be more pathogenic than others in Alzheimer's disease (58). Fibril size, arrangement, and conformation open the question about whether or not fibril toxicity is dependent on cell internalization and whether CYT would be able to inhibit such cytotoxicity.
LC fibril toxicity could also be correlated with their strong attraction for the plasma membrane, causing a particular cell confinement. Cells surrounded by fibrils are likely to be excluded from cell-cell contact, a vital mechanism for maintaining cell viability. Cell treatment with trypsin (40,48,59), chondroitinase ABC, and heparinase II (60, 61) did not detach the external amyloid from the cell membrane (data not shown). We propose that membrane surfaces facilitate fibril attachment and act as an anchor point for cell-mediated seeding mechanism. Extracellular soluble protein would interact with amyloid aggregates on the cellular surroundings.
In vitro seeded acceleration of protein fibrillation has been reported in many proteins (62,63). We previously demonstrated that the presence of preformed aggregates in vitro accelerate the fibril formation reaction (64,65). In this study we observed soluble protein recruitment to amyloid fibrils in half the time observed in vitro, which leads us to propose that seeding is significantly accelerated in a cell culture environment. Because fibrils are highly toxic to cells, the seeding mechanism deserves attention, because it could become an exponential trigger of cell toxicity, propagating fibril elongation throughout the cellular environment.
Future studies including other amyloidogenic LC proteins both in the V L and FL forms will be necessary to confirm our observations and strengthen the proposed mechanisms of internalization and toxicity observed in this initial study. Our results suggest that AL amyloid internalization, propagation, seeding, and toxicity mechanisms, as well as the role of CL domains on LC proteins, are correlated, deserving more attention for further studies in AL amyloidosis.

Experimental Procedures
Protein Preparation-The V L sequences for I O18/O8 and AL-09 were deposited previously under GenBank TM accession numbers EF640313 and AF490909, respectively (36). There is only one C L sequence (protein accession number P01834). V L domains and FL proteins were expressed and purified as previously described (25,36). I FL sequence was mutated at position C214S (end of C L domain) to avoid the formation of nonnative disulfide bonds. Alternatively, the Cys 214 position was kept for AL-09 FL because it displayed a better protein expression and higher extraction yield without changing any other biochemical and biophysical properties. Briefly, V L domains were expressed in Escherichia coli BL21 (DE3) Gold competent cells. I O18/O8 V L was extracted from the periplasmic space by breaking the cells through one freeze-thaw cycle using PBS buffer, pH 7.4. AL-09 V L was extracted from solubilized inclusion bodies using 5 M urea and refolded by dialysis against 10 mM Tris-HCl, pH 7.4. FL proteins were expressed in E. coli Rosetta Gami competent cells. AL-09 and I FL were extracted from solubilized inclusion bodies using 5 M urea and refolded by dilution (1:10) in ice-cold refolding buffer (10 mM Tris/HCl, 1 M L-arginine, 7 mM GSH, 0.7 mM GSSG, 2.5 mM EDTA, and 1 mM PMSF protease inhibitor, pH 8.5) for 48 h at 4°C. All proteins were purified using size exclusion chromatography in 10 mM Tris buffer, pH 7.0, at 4°C (HiLoad 16/60 Superdex 75 column) on an AKTA FPLC (GE Healthcare). Eluted fractions were checked by SDS-PAGE, and their protein concentration was determined by UV absorption at 280 nm using an extinction coefficient (⑀) calculated from the amino acid sequence (14,890 and 25,940 cm Ϫ1 M Ϫ1 for I VL and FL proteins, respectively; 13,610 and 24,660 cm Ϫ1 M Ϫ1 for AL-09 VL and FL proteins, respectively). Far UV CD scan and thermal unfolding were obtained as a protein quality control. Proteins were aliquoted at concentration below 100 M, flash frozen, and stored at Ϫ80°C. Proteins were thawed at 4°C, filtered, and/or ultracentrifuged before they were used for each study.
Amyloid Fibril Formation-Because the presence of preformed aggregates may accelerate the fibril formation kinetics, protein samples were ultracentrifuged prior to the fibril formation assay following the protocol described by DiCostanzo et al. (66). Fibril formation assays were performed in triplicate using black 96-well polystyrene plates and shaken continuously at 300 rpm at 37°C in a New Brunswick Scientific Innova40 incubator shaker. Each well contained 260 l of 20 M protein, 150 mM NaCl, 10 M thioflavin T (Sigma-Aldrich), 0.02% NaN 3 in 10 mM sodium acetate, boric acid, and sodium citrate (ABC) buffer at pH 2.0 or 3.0 for V L domains and FL proteins, respectively. These fibril formation solution conditions were chosen because they are the only conditions in which all proteins (V L and FL) form fibrils with similar morphologies. Making fibrils under physiological solution conditions (pH 7.4) was not a possibility because I V L does not form fibrils under those conditions, and I FL forms a mixture of amorphous aggregates with amyloid fibrils at that pH. ThT fluorescence was used to follow the fibril formation kinetics (67,68) and was monitored daily on a plate reader (Analyst AD; Molecular Devices, Sunnyvale, CA) with an excitation wavelength of 440 nm and an emission wavelength of 480 nm, until the reaction reached the plateau (ϳ600 -800 h). A fibril formation reaction was considered completed when ThT fluorescence enhancement reached a plateau and was considered positive when ThT fluorescence readings reached at least four times the lowest ThT reading of that particular reaction (usually 200,000 counts/s). Fibrils were washed in PBS, pH 7.4, and stored at 4°C until they were used in cell culture.
Oregon Green Soluble Protein Conjugation-Oregon Green 488 (Invitrogen) conjugation reactions were conducted as reported by Levinson et al. (25). Protein samples were thawed at 4°C. Tris buffer was exchanged to PBS using a 10,000 molecular weight cutoff centrifugal filter units (Millipore, Billerica, MA). 200 -500 l of protein solution was used in each labeling reaction, at a concentration of Ͻ2 mg/ml. 100 mM NaHCO 3 , pH 8.5, was added to each protein sample to raise the pH of the reaction mixture. 1 mg of OG was solubilized in 100 l of DMSO (10 mg/ml). The volume of OG dye stock solution to be added was calculated as follows.

OG stock solution (l)
ϭ [(mg/ml protein ϫ ml protein)/Mr protein] The molecular weight (M r ) of the proteins used was as follows: 11,930 and 23,504 g/mol for both VL domain and both FL proteins, respectively. The M r of OG is 509 g/mol, the conversion factor (CF) is 100, and the molar ratio (MR) of dye to protein in the reaction mixture is 20.
Soluble proteins were labeled overnight at 4°C, protected from light. Free OG was removed from the soluble labeled protein by consecutive rounds of protein concentration and dilution with PBS, pH 7.4, through a 10,000 molecular weight cutoff centrifugal filter. Samples were then filtered using a 0.45 m filter and the protein concentration and degree of labeling were calculated by the following formula, where DF is the dilution factor (when applied), 0.12 is a correction factor to account for absorption of the dye at 280 nm. ⑀-Protein is the molar extinction coefficient of each protein (14,890 and 25,940 M Ϫ1 cm Ϫ1 for I O18/O8 VL and FL proteins, respectively; 13,610 and 24,660 M Ϫ1 cm Ϫ1 for AL-09 VL and FL proteins, respectively); ⑀-OG is the molar extinction coefficient of OG (70,000 M Ϫ1 cm Ϫ1 ).
Oregon Green Amyloid Conjugation-At the end of fibril formation reaction, fibrils were collected, pelleted, and washed three times with PBS buffer by centrifugation at 14,000 rpm, 5 min at room temperature. 200 -500 l of PBS resuspended fibrils were used in each labeling reaction, at a concentration range of 0.5-1 mg/ml. Fibrils were incubated with OG for 2 h at room temperature, protected from light. Free OG was removed from labeled fibrils by centrifugation. Supernatant was removed and quantified to determine the concentration of soluble protein left after fibril formation. Final fibril concentration was adjusted to that number. The degree of labeling was determined for each conjugation as described for the soluble proteins and used to normalize the fluorescence intensities of cellular experiments.
Cell Culture-AC16 human primary ventricular cardiomyocytes were purchased from Dr. Mercy Davidson at Columbia University. This cell line has been immortalized by fusion with SV40 transformed fibroblast cell line devoid of mitochondrial DNA (69). The cells were maintained with DMEM/F12 medium (Life Technologies Inc.) supplemented with 12.5% FBS (Mediatech, Manassas, VA) and 1% penicillin/streptomycin (Invitrogen). AC16 cells co-transfected with plasmid expressing RFP in the nucleus were also used (RFP-AC16 cells). For the nuclear cell labeling, the IncuCyte TM NucLight TM lentivirus reagent has been used (Essen Bioscience). NucLight Lentiviruses drive the expression of nuclear localization signal-tagged fluorescent proteins with an EF-1␣ promoter. The NucLight red version expresses mKate2 in the nucleus of the cells. Cell culture experiments were carried out under sterile conditions. AC16 cells are not listed in the database of commonly misidentified cell lines maintained by ICLAC. As a control of viability and differentiation, cell morphology was always checked before each experiment, and the number of cell passages after thawing was limited to 20.
Cell Internalization Experiments-Internalization experiments were carried out using the IncuCyte ZOOM (Essen Bioscience, Ann Arbor, MI) incubator. The microscope incorporated into the incubator supports two different fluorescence channels. We took advantage of the two color setup and used RFP-AC16 cells incubated with soluble proteins or fibrillar aggregates conjugated with OG dye ( OG AL-09 V L , FL, and OG I V L , FL). AC16 cardiomyocytes were plated at a concentration of 2,000 cells/well (10,000 cells/ml) in a 96-well Corning polystyrene plate and allowed to grow overnight (Ͻ20 h) into the IncuCyte ZOOM incubator (5% CO 2 at 37°C) for cell attachment. 200 l of cell culture medium was replaced with 100 l of fresh medium. 100 l of OG soluble LC or OG fibrils were added to the cells at a final concentration of 1 M. To assess the effect of amyloid fibrils on the protein internalization process, the cells were co-incubated with OG soluble LC mixed with nonlabeled fibrils, at a ratio of 1:1. Plates were scanned every 4 h. Internalization data were collected 4, 8, 12, and 24 h after addition of the protein. For this, 10 min before the desired time point, cell medium-containing OG soluble LC/fibrils were replaced with free fresh medium to remove the fraction not internalized. Red (RPF-AC16) and green ( OG soluble LC/fibrils) fluorescence channels were selected, and 5% of the red signal was removed from the green signal to avoid spectral mixing. A four-field scan pattern was used for each well, and the average data were collected as a green counts or red counts per well. Each condition was set up in triplicate. For a correct comparison between different AL proteins, green fluorescence intensities were normalized for each protein as a function of their degree of labeling determined after each conjugation as described above.
After 24 h, the cell medium-containing OG soluble LC/fibrils of three 200-l wells was replaced with free fresh medium. The cells were scanned every 4 h for a longer period of time, which allowed us to follow the decrease of intracellular soluble protein. After 48 h, the media of three different 200-l wells were replaced and also scanned every 4 h. The fluorescence intensity of the two different time sets helped us to discern between a quenching effect and secretion mechanism.
To remove the extracellular aggregates, cells were incubated with 0.01% or 0.5% trypsin-EDTA (Life Technologies Inc.) in DMEM/F12 medium for 2 min and washed with DMEM/F12 medium for deactivation of the trypsin. The cells were also incubated with chondroitinase ABC and heparinase II (Sigma-Aldrich) at 12.5, 6.25, and 3.12 milliunits/ml in DMEM/F12 medium for 1 h at 37°C. Because of our inability to detach the fibrils from the cell membrane, we were unable to quantify the amount of fibrils internalized into AC16 cells.
Protein Internalization Inhibition Assays-Prior to soluble protein internalization assay, RFP-AC16 cells were incubated for 30 min at 37°C with 50 M DYN, 10 M MiTMAB, 50 M GEN, or 1 M CYT. Thereafter, OG soluble LC/fibrils were added to the cells and followed throughout time in presence of inhibitor. The data were collected as described above. Green counts and red counts per well in the presence of inhibitor were compared with the data in the absence of inhibitors.
Cell Viability Assays-Experimental setup was followed as described for the internalization experiments, except that both proteins and fibrils were unconjugated. RFP-AC16 cells were incubated with 1 or 12 M of soluble protein, or with 1 M of amyloid fibrils. Fibrils are stored at 4°C after the fibril formation reaction is completed. Freeze/thawing cycles has been done before each experiment in all cases except the experiments using fresh fibrils. The changes in cell growth were followed by red counts per well (or percentages of red cells per well) every 4 h until cells become over confluent (Ͼ80 h).
The apoptotic index has been assessed by using the homogeneous caspase assay fluorimetric kit (Roche). We first used the apoptotic reagent CellPlayer TM kinetic caspase-3/7 (Essen Bioscience) for use on the IncuCyte ZOOM TM imaging systems, which kinetically quantify cell proliferation over time in a nonperturbing way. Unexpectedly, the apoptotic reagent binds to the external amyloid aggregates, releasing green fluorescent signal. To avoid false positives, we used the caspase assay kit. In brief, at the end of the cell viability assays (80 h), 100 l of substrate working solution was added to each well and incubated for 1 h at 37°C, 5% CO 2 . The plate was read on the Analyst AD plate reader (Molecular Devices) at an excitation wavelength of 480 nm and an emission wavelength of 520 nm, medium attenuator mode, and continuous lamp. The average data were collected as relative fluorescence units/well. Each condition was set up in triplicate, and each well was read three times.
Transmission Electron Microscopy-Amyloid fibril morphologies were confirmed by transmission electron microscopy. A 5-l fibril sample was placed on a 300-mesh copper Formvar/ carbon grid (Electron Microscopy Science, Hatfield, PA), and excess liquid was removed. The samples were negatively stained with 4% uranyl acetate, washed once with sterile H 2 O, and air-dried. Grids were analyzed on a Philips Tecnai T12 transmission electron microscope at 80 kV (FEI, Hillsboro, OR).
Confocal Microscopy-For co-localization experiments with OG soluble LC and Texas Red-labeled fibrils, the AC16 cardiomyocytes used were not co-transfected with RFP protein (no red nucleus). the Cells were previously fixed with 4% paraformaldehyde-PBS solution for 30 min at room temperature. An LSM 780 confocal microscope (Carl Zeiss Microscopy) was used to image the cells with a 40ϫ differential interference contrast lens using a water immersion objective (Zen software). Laser wavelengths of 488 and 561 nm were used. For Z stacks series, 15 slices that were 0.5 mm thick were taken. The images were captured using Zeiss LSM Image version 3.2SP2. Images were collected with ϫ4 averaging. Detector gain and amplitude offset were determined for each experiment to maximize the linear range without saturation and were kept consistent for comparable experiments. The images were prepared using ImageJ.
Determination of Monomer Concentration by Reverse Phase HPLC Assay-Quantification of monomer concentration on fibril supernatant samples was performed using a HPLC sedimentation assay as previously described (54). Briefly, a 100-l fibril sample in 10 mM ABC buffer, pH 2.0 or 3.0 (see "Amyloid Fibril Formation") was taken before each fibril toxicity experiments. Before injection into the analytical reverse phase HPLC chromatography (BioLogic DuoFlow Pathfinder 20 system), samples were ultracentrifuged at 90,000 rpm for 45 min at room temperature to remove aggregates. 70-l supernatant fractions were injected into a C8 column to determine the concentration of monomers and eluted by a linear gradient of acetonitrile in aqueous 0.05 TFA. The area under the curve of the chromatography peak at specific retention time was integrated, and the concentration was determined from a standard curve done for each AL protein.