GroEL mediated conformational changes of DM-MBP

intermedi-0.15 M GuHCl and 20 mM KCl 0.13 M GuHCl and 20 mM KCl 0.13 M GuHCl and 200 mM KCl

A M [ms]

Figure 7.13: Kinetic of the spontaneous refolding of DM-MBP. The FRET efficiency curves of Figure 7.12 were fitted using a Gaussian function. The mean FRET efficiency of the fit and standard error are plotted versus the time after mixing for DM-MBP refolded in 0.15 M GuHCl and 20 mM KCl (black dots), in 0.13 M GuHCl and 20 mM KCl (red dot) and 0.13 M GuHCl and 200 mM KCl (cyan dots). Fitting the mean FRET efficiency using an exponential curve results in refolding times of 681± 609 ms, 509±244 ms and 428±323 ms, respectively.

ate state is independent of the salt concentration, then refolding from this point onwards to the final conformation has to be salt dependent.

is again very broad and corresponds to multiple subconformations. Thus, we conclude that GroEL unfolds a fraction of DM-MBP to a larger degree than GuHCl is able to do and the other fraction changes the conformation resulting in a slightly wider distance between the two labeling positions. These results are in line with the previously published data for the used DM-MBP mutant [Sharma et al., 2008].

Table 7.2: Distances calculated using PDA for DM-MBP with GroEL. DM-MBP (50 pM) was measured in the presence of 0.13 M GuHCl and 3 µM GroEL. The photons of a burst were divided into bins of 1 ms, summed up in a FRET histogram, and fitted using a double-Gaussian model. With photon distribution analysis (PDA), the distance d between donor and acceptor, the corresponding width σ and the relative weights of the different populations were calculated.

Subpopulation Subpopulation Goodness of the fit d [˚A] σ [˚A] % d [˚A] σ [˚A] % χ²

+ GroEL 50 7 35.5 84 15 64.5 3.6

Next, the temporal progress of refolding in the presence of GroEL was investigated by using microfluidics. DM-MBP was again denatured in 3 M GuHCl, diluted to a final GuHCl concentration of 1.8 M in a buffer containing 20 mM KCl and 0.001 % Tween20 and, then, connected to the sample inlet of the microfluidic device. The buffer, connected to the buffer inlets, contained 20 mM KCl, 0.001 % Tween20 and 3.25 µM GroEL, which resulted after mixing in a final GroEL concentration of 3 µM. The GuHCl concentration was again 0.13 M after mixing.

The FRET efficiency of the DM-MBP was determined 33 ms after mixing as ~0.6 (Figure 7.15, black line). This is a similar value as the one detected in measurements without GroEL (Figure 7.12B, black line). Therefore, the GroEL is not bound, which is in line with the timescale of GroEL binding measured with a halftime of ~100 ms [Sharma et al., 2008].

For comparison the time for mixing is estimated as the following: The channel in the mixing point is 40µm wide. The solutions are combined 1 to 12. Therefore, the width of the middel stream is approximately 3 µm. Due to the fact that the buffer is mixed from both sides with the sample stream, the GroEL and ATP only have to diffuse half of the way, i.e. 1.5 µm.

The diffusing coefficient for ATP can be approximated as 1·10^{−5 cm}_s² [Wang et al., 2012] and for GroEL as 560·10^{−9 cm}_s² [Lakowicz, 2006]. WithτD = _4D^ω2r, the time the molecule diffuses in the sample stream can be approximated as 0.5 ms for ATP and 10 ms for GroEL. Thus, mixing of GroEL is much faster than binding found in literature and our first measurement timepoint.

After 50 ms (Figure 7.15, red line), an additional population with a low FRET efficiency value appears. Two populations are detectable, one at lower FRET values, which is induced by binding of GroEL, and the second population at the same FRET efficiency as detected in the measurements in the absence of GroEL (Figure 7.12B, red line).

33 ms

50 ms

155 ms

232 ms

343 ms

512 ms

530 ms

718 ms

792 ms

1.07 s

+ GroEL (equilibrium) unfolded

Figure 7.15: Timescale of the refolding kinetic of DM-MBP mediated by GroEL. DM-MBP was unfolded in 3 M GuHCl. The GuHCl concentration was diluted to 1.8 M and then mixed in the microfluidic device 1 to 12 with a buffer containing 20 mM KCl, 0.001 % Tween20 and 3.25 µM GroEL (after mixing 3 µM GroEL) to a final concentration of 0.13 M GuHCl. The concentration of DM-MBP as to be 15 pM after mixing.

To further investigate the binding of GroEL, the measurements after 33 ms, 50 ms, 155 ms and 232 ms were analyzed with fluorescence correlation spectroscopy (FCS). In Figure 7.16 the four autocorrelation functions are depicted. Due to the fact that the measurements were done with two different flow rates, only the measurement after 33 ms and 155 ms or 50 ms and 232 ms can be easily compared. 33 ms and 50 ms after mixing the DM-MBP diffuses faster than 155 ms and 232 ms after mixing, respectively, indicating a binding of GroEL faster than 155 ms. For a detailed analysis, the FCS curves were fitted with the correlation function with flow (Equation (3.29)). The diffusing coefficients were calculated as 147.5 ^µm_s² for 33 ms, 108,4 ^µm_s² for 50 ms, 76.5 ^µm_s² for 155 ms and 82.6 ^µm_s² for 232 ms. Thus, the diffusing coefficient changes around 50 ms after mixing, which is a hint for binding of GroEL.

10^-4 10^-3 10^-2 Time Lag τ [s]

0 0.2 0.4 0.6 0.8

G(τ)

33 ms 50 ms 155 ms 232 ms -10

0 10

Weighted residuals

10^-5

Figure 7.16: FCS curves of DM-MBP mixed with GroEL. DM-MBP was unfolded in 3 M GuHCl. The GuHCl concentration was diluted to 1.8 M and then mixed in the microfluidic device 1 to 12 with a buffer containing 20 mM KCl, 0.001 % Tween20 and 3.25µM GroEL (after mixing 3µM GroEL) to a final concentration of 0.13 M GuHCl. The concentration of DM-MBP was chosen as 15 pM after mixing. FCS autocorrelation was performed 33 ms, 50 ms, 155 ms and 232 ms after mixing. The diffusing coefficients were calculated as 147.5 ^µm_s² for 33 ms, 108,4 ^µm_s² for 50 ms, 76.5 ^µm_s² for 155 ms and 82.6 ^µm_s² for 232 ms.

After binding the FRET histogram changes. The amplitude of the low FRET population is increased up to ~300 ms and the FRET efficiency value decreases over time. The high FRET population stays at its FRET efficiency value. From 343 ms onwards, the FRET efficiency is stable over time and does not change anymore. To analyze the changes in a more precise way, two Gaussian functions were fitted to the data. The means of the two FRET efficiencies are plotted versus the time after mixing (Figure 7.17). The mean of the high FRET population stays constant over time. The mean of the low FRET population can be fitted using an exponential function and results in a reaction time of 168±59 ms.

low FRET high FRET

[ms]

Figure 7.17: Kinetic of the GroEL mediated refolding of DM-MBP. The FRET efficiency curves of Figure 7.15 were fitted using two Gaussian functions. The two mean FRET efficiencies of the fit and the corresponding standard error are plotted versus the time after mixing for DM-MBP refolded in 0.15 M GuHCl and 20 mM KCl. The low FRET population was fitted using an exponential curve resulting in kinetic times of 168 ±59 ms.

Compared to the equilibrium measurement of DM-MBP with GroEL, the FRET populations are broader and molecules can be detected between the two main peaks. This could be

because of interchanging molecules between the two FRET populations or more likely due to a the low number of photons. The applied flow decreases the time the molecule spends in the focus and, thus, the number of photons is decreased.

In summary, the binding of GroEL to DM-MBP happens on a timescale of approximately 50 ms, which is larger than the diffusing time we calculated of GroEL. After binding, first the low FRET population is filled, which corresponds to unfolded proteins (Figure 7.15). Till approximately 300 ms, the low FRET peak shifts to slightly lower FRET values with a kinetic time of 168 ±59 ms. From 343 ms onwards, the conformation stays stable over time.