Heat shock-induced chaperoning by Hsp70 is enabled in-cell

Recent work has shown that weak protein-protein interactions are susceptible to the cellular milieu. One case in point is the binding of heat shock proteins (Hsps) to substrate proteins in cells under stress. Upregulation of the Hsp70 chaperone machinery at elevated temperature was discovered in the 1960s, and more recent studies have shown that ATPase activity in one Hsp70 domain is essential for control of substrate binding by the other Hsp70 domain. Although there are several denaturant-based assays of Hsp70 activity, reports of ATP-dependent binding of Hsp70 to a globular protein substrate under heat shock are scarce. Here we show that binding of heat-inducible Hsp70 to phosphoglycerate kinase (PGK) is remarkably different in vitro compared to in-cell. We use fluorescent-labeled mHsp70 and ePGK, and begin by showing that mHsp70 passes the standard β-galactosidase assay, and that it does not self-aggregate until 50°C in presence of ATP. Yet during denaturant refolding or during in vitro heat shock, mHsp70 shows only ATP-independent non-specific sticking to ePGK, as evidenced by nearly identical results with an ATPase activity-deficient K71M mutant of Hsp70 as a control. Addition of Hsp40 (co-factor) or Ficoll (crowder) does not reduce non-specific sticking, but cell lysate does. Therefore, Hsp70 does not act as an ATP-dependent chaperone on its substrate PGK in vitro. In contrast, we observe only specific ATP-dependent binding of mHsp70 to ePGK in mammalian cells, when compared to the inactive Hsp70 K71M mutant. We hypothesize that enhanced in-cell activity is not due to an unknown co-factor, but simply to a favorable shift in binding equilibrium caused by the combination of crowding and osmolyte/macromolecular interactions present in the cell. One candidate mechanism for such a favorable shift in binding equilibrium is the proven ability of Hsp70 to bind near-native states of substrate proteins in vitro. We show evidence for early onset of binding in-cell. Our results suggest that Hsp70 binds PGK preemptively, prior to its full unfolding transition, thus stabilizing it against further unfolding. We propose a “preemptive holdase” mechanism for Hsp70-substrate binding. Given our result for PGK, more proteins than one might think based on in vitro assays may be chaperoned by Hsp70 in vivo. The cellular environment thus plays an important role in maintaining proper Hsp70 function.


mHsp70 and mHsp70K71M ATPase activity
The absorbance spectrum of PercevalHR shifts according to the nucleotide present ( Fig  B, panel A). With 2 mM ATP the absorbance peak at 500 nm increased while with 2 mM ADP the absorbance peak at 420 nm increased. The ATPase activity was measured by monitoring the emission of PercevalHR at 520 for 500 nm excitation (Fig B, panel B). As the ATP is hydrolyzed the absorbance at 500 nm decreases and hence the emission at 520 nm for the 500 nm excitation also decreases. The curves were fit using 1 st order reaction kinetics and yielded an ATP hydrolysis rate of 6.31 X 10 -1 sec -1 for mHsp70 and 3.69 X 10 -2 sec -1 mHsp70K71M.

In vitro circular dichroism and fluorometer melts
Prior to all measurements the glycerol from the frozen stocks were removed by spin filtration buffer exchange. Tryptophan fluorescence measurements and in vitro FRET binding experiments were conducted on an FP8300 spectrofluorometer equipped with Peltier temperature control S-2 (JASCO). For in vitro PGK characterization, tryptophan was excited at 295 nm, and emission spectra were collected from 290 to 450 nm. Samples were measured in 300 μL cuvettes at 5 μM concentrations, unless otherwise noted. Circular dichroism (CD) was measured using a J-715 spectropolarimeter with Peltier temperature control (JASCO). Unless otherwise noted, all spectra were recorded from 250 to 200 nm at a scan rate of 50 nm/min at 1 nm resolution and averaging five accumulations. Thermal melts were done using a 2 mm path length cuvette. Unless otherwise noted, protein concentration in circular dichroism experiments was 5 μM. Protein unfolding was monitored by measuring the change in mean residue ellipticity at 222 nm.
In vitro FRET-PGK1 unfolding with denaturant guanidinium hydrochloride (GuHCl) was measured by conducting isothermal titrations at 20 °C of 5 μM FRET-PGK1 with GuHCl between 0-0.8 M. mEGFP was excited at 485 nm and emission spectra were collected from 480 to 700 nm in 300 µL cuvettes.

Measurement of ATPase assay with PercevalHR
Prior to all measurements the glycerol from the frozen stocks were removed by spin filtration buffer exchange. ATPase assays were conducted with the ATP sensor protein PercevalHR 1 . The plasmid for bacterial expression, pRsetB-PercevalHR (Addgene plasmid # 49081), was obtained from Addgene and was a gift from Gary Yellen. The protein was expressed as described in the main text and dialyzed in 5 mM MOPS, 300 mM NaCl, pH 7.3. Absorbance measurements were made in K1 buffer with 2 mM total nucleotide (2 mM ATP, 2 mM ADP or 1 mM each of ATP and ADP) and 5 µM PercevalHR. ATPase assays were conducted also in K1 buffer with 5 µM mHsp70 or mHsp70K71M, 2.5 µM Hsp40, 10 µM PercevalHR and 1 mM ATP. The PercevalHR emission intensity was monitored at 520 nm for 500 nm excitation for 2 hours at 15 second intervals at 37 °C.

Data analysis
Data analysis was performed with MATLAB (Mathworks). Midpoint denaturant concentration ( " ) was calculated using a two-state sigmoidal fit to the experimental isothermal titration data. Native and denatured state baselines were assumed to be linear with respect to the perturbing variable (concentration) so that the signal can be estimated by: where is either the native state (N) or the denatured state (D), is the slope and is the intercept. " is the corresponding midpoint of the curve with the concentration . The total signal ( ) can then be estimated as a sigmoidal function with respect to concentration, : 2) ( ) = 2 ( ) 2 ( ) + 4 ( ) 4 ( ) , where, 2 and 4 are the populations of the N and D states, respectively.
The kinetic data for mHsp70 ATPase activity was fit using the equation: where, is the reaction rate. The mHsp70 fits yielded a value of -0.6623 for which was fixed at this value for fitting the kinetic data for mHsp70K71M. The data for mHsp70K71M was then fit to the equation:  In ATP PercevalHR shows an absorbance peak at 500 nm. As the ATP is replaced by ADP the peak at 500 nm decreases giving rise to a new absorbance peak at 420 nm. (B) ATPase activity of mHsp70 (blue) and mHsp70K71M (red) monitored by PercevalHR emission intensity at 520 nm for 500 nm excitation. The rates were fit using a 1 st order reaction rate kinetics and the rate, k, = 6.31 X 10 -1 sec -1 for mHsp70 and 3.69 X 10 -2 sec -1 mHsp70K71M.
S-6    temperature for (A) mCherry binding to ePGKs(0-3). A small change in FRET efficiency, 1%, was observed due to non-specific interaction between mCherry and unfolded ePGKs (B) mHsp70 binding to mEGFP. Inset shows the enlarged y-axis for mHsp70/mHsp70K71M non-specific interaction with mEGFP. The change in FRET efficiency for non-specific interaction of either unfolded mHsp70 or unfolded mHsp70K71M with mEGFP is <0.5%.     Fig 7B, and 4x smaller than the change seen for mHsp70K71 with ePGK in Fig 7A. S-14