The Ability of DNAJB6b to Suppress Amyloid Formation Depends on the Chaperone Aggregation State

For many chaperones, a propensity to self-assemble correlates with function. The highly efficient amyloid suppressing chaperone DNAJB6b has been reported to oligomerize. A key question is whether the DNAJB6b self-assemblies or their subunits are active units in the suppression of amyloid formation. Here, we address this question using a nonmodified chaperone. We use the well-established aggregation kinetics of the amyloid β 42 peptide (Aβ42) as a readout of the amyloid suppression efficiency. The experimental setup relies on the slow dissociation of DNAJB6b assemblies upon dilution. We find that the dissociation of the chaperone assemblies correlates with its ability to suppress fibril formation. Thus, the data show that the subunits of DNAJB6b assemblies rather than the large oligomers are the active forms in amyloid suppression. Our results provide insights into how DNAJB6b operates as a chaperone and illustrate the importance of established assembly equilibria and dissociation rates for the design of kinetic experiments.

The data shown in Figure 2 were fiAed using the Amylofit soCware 1 and a model including primary nucleaEon, secondary nucleaEon, and elongaEon.The coupled rate constants for primary nucleaEon and elongaEon (knk+), and for secondary nucleaEon and elongaEon (k2k+), were varied parameters, and the reacEon orders for both primary and secondary nucleaEon were fixed at 2. The fits are displayed together with the kineEc data in Figure 1S.It was noted that some parameters can be globally fiAed and sEll give a good representaEon of the data, but the current experiment was not designed to evaluate the inhibitory mechanism of JB6b, which has been explored before 2,3 .

Crosslinking of 40 nM JB6b
To monitor the dissociaEon kineEcs of JB6b oligomers upon diluEon with a complementary technique to MDS, chemical crosslinking was uElized.The experimental setup is illustrated in Figure S2, panel A. 12.8 μM JB6b was diluted to 40 nM in a volume of 100 ml.FracEons of 5 ml were removed from the container as a funcEon of Eme since diluEon, and crosslinked using either 40 μM (Figure S2, panel B) or 400 μM (Figure S2, panel C) BS3 linker (bis(sulfosuccinimidyl)suberate).The BS3 linker was prepared by dissolving in H2O to 40 mM just prior to freezing in 5 μl aliquots in 5 ml tubes (Eppendorf, protein low binding).Each tube was thawed less than a minute before the addiEon of 5 ml protein soluEon.The reacEon was stopped aCer 10 minutes by adding 40 mM ethanolamine from a 1 M stock, pH 8.0.The samples were concentrated by precipitaEon by the addiEon of 10 % (v/v) tri-chloro aceEc acid (TCA), vortexing, incubaEon on ice for a minimum of 30 min, centrifugaEon at 10 000 g at 4 °C for 20 min and pellet resuspension in 3 μl of 3 M Tris at pH 8.0. 3 μl of 4X SDS loading buffer was added to each sample and SDS-PAGE was run at 140 V for 110 min using 10-20% polyacrylamide Tris/Tricine gels (Novex).
One should note that oligomers can be partly crosslinked.Under denaturing condiEons such as in SDS-PAGE, the assemblies will dissociate into the smallest crosslinked species, which may result in bands corresponding to smaller sizes in an SDS-PAGE than the complexes they were part of under the naEve soluEon condiEon.ParEal crosslinking is a probable explanaEon for the results shown in Figure S2 panel B (40 μM crosslinker).Here, mostly monomers and low order oligomers are visible at the shortest Emes aCer diluEon (for instance in lane 2, 2 min aCer diluEon).When using 10 Emes higher concentraEon of crosslinker, Figure S2 panel C, much less monomers and low order oligomers are seen shortly aCer diluEon, but instead a large part of the protein is retained high up in the lane, not penetraEng the gel more than a few millimeters.Hence, in the case of 400 μM crosslinker, the large JB6b oligomers are crosslinked.
To quanEfy the results of the gel in Figure S2, panel B, the soCware ImageJ was used to get grey value profiles of each lane.An example of how this analysis was carried out for lane 2 is shown in Figure S2, where the analyzed part is the drawn rectangle in panel B, and the generated profile is shown in panel D. The integral of each protein band was calculated to a baseline between two chosen points on the profile.Due to the many experimental steps that leads to the protein loaded in each lane, not the same amount of protein was able to be loaded for each sample.Hence, to compare the amount of crosslinked oligomeric states between lanes, the sum of the non-monomeric integrals was divided by the sum of all integrals in that lane.The raEo is ploAed versus the Eme since diluEon in panel E, in black circles.A fit to an exponenEal decay is shown as dashed line, where the apparent dissociaEon Eme constant is k != 0.030 h -1 .The control sample without any added crosslinker is ploAed at Eme zero as a blue square and is not part of the fiAed curve.The dissociaEon Eme constant is in the same size range as obtained using MDS in Ref. 4 ( " = 0.039 h -1 ).It should be noted that the ⟨RH⟩ from MDS not necessarily relates perfectly to the amount of crosslinked oligomeric species, since the crosslinking only reports on a part of the self-associated protein.In other words, since we do not know if some aggregaEon states are more or less prone to be crosslinked, the decay rate does not need to follow the ⟨RH⟩ decay precisely, but should at least be on the same order of magnitude, which they are.Furthermore, in both measurement techniques there is a rather large measurement error.
One might ask whether a crosslinked product could be the result of two parEcles coming in proximity by Brownian moEon and get crosslinked at this moment, providing an overesEmaEon of the fracEon of oligomeric species.This hypothesis was here discarded using a monomeric control protein, with similar number of reacEve sites for the crosslinker.In this case, a protein consisEng of two copies of human S100G, joined via a single glycine residue, was expressed as a single protein chain, and used as a control.The amino acid sequence for diS100G was: MKSPEELKRIFEKYAAKEGDPDQLSKDELKLLIQAEFPSLLKGMSTLDDLFQELDKDGDGEVSFEEFQVLVK KISQGRSPEELKRIFEKYAAKEGDPDQLSKDELKLLIQAEFPSLLKGMSTLDDLFQELDKDGDGEVSFEEFQV LVKKISQ.