Quantitative analysis of the interplay between hsc70 and its co-chaperone HspBP1

Cell culture and heat shock

HeLa cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with antibiotics and 8% Fetal Bovine Serum (FBS). Cells were maintained at 37 °C; they were grown to 70% confluency on poly-lysine-coated cover slips and analyzed between passages 3 and 12. For heat shock, cells were exposed to 45.5 °C for 1 h; subsequent recovery was at 37 °C for different periods of time.

Immunocytochemistry

HeLa cells were washed with PBS, fixed for 10 min with 10% formaldehyde in PBS, permeabilized for 5 min with cold acetone at −20 °C and blocked for 1 h with PBS/ 2 mg/ml BSA/1 mM NaN₃ (blocking solution). All subsequent steps were carried out in blocking solution at room temperature. Samples were incubated for 1 h with primary antibodies against hsc70 (diluted 1:1,000; Stressgen SPA-815), HspBP1 (diluted 1:200; Santa Cruz sc-34252) or Crm1 (diluted 1:200, sc-5595). Following three washes, fluorescently-labeled secondary antibodies were added for 1 h. Samples were washed and nuclei stained for 2 min with 1 µg/ml 4,6′-diamidino-2-phenylindole (DAPI). Cover slips were mounted and sealed.

Microscopy and quantitative image analysis

Images were acquired in the multi-track mode with a Zeiss LSM510 confocal microscope. Appropriate filter settings were selected to minimize cross-talk between the channels.

Image quantification was performed with MetaXpress software, as described earlier (Kodiha, Brown & Stochaj, 2008, Fig. S1). In brief, the DAPI signal defined the nuclear and nucleolar compartments. Crm1, another nuclear marker, validated the use of DAPI to demarcate nuclei. The first step of image analysis was background correction (Kodiha, Brown & Stochaj, 2008). To this end, average pixel intensities from a region devoid of cells were subtracted to generate a Background Correction image. After background correction, pixel intensities of the proteins of interest were measured in the nucleus and cytoplasm. All steps were automated. At least 35 cells were quantified for each of the different conditions in every experiment. All images were visually inspected to verify the correctness of the compartment demarcation. For single-cell analysis, a minimum of 115 cells was quantified for each of the five time points. For each experiment, data were normalized to non-stressed controls.

Colocalization was measured with MetaXpress software as follows: regions of interest (ROI) that were identical for hsc70 and HspBP1 channels were selected for the nucleus or cytoplasm. For each cell analyzed, the colocalization was quantified both in the nucleus and cytoplasm (Fig. S2). The images were thresholded, and the overlap between hsc70 and HspBP1 signals was evaluated for the same pixel using the Colocalization Module. The pixel size was 0.14 µm². The same threshold value for each channel was used for all cells and maintained across the different conditions. At least 10 cells were analyzed per condition. The results show percent overlap of the hsc70 with HspBP1 area (Hsc70/HspBP1), or percent overlap of the HspBP1 with hsc70 area (HspBP1/hsc70). Different steps of the method and representative output data are shown in Fig. S2.

Western blotting

Methods for the preparation of whole cell extracts and Western blotting have been published (Mahboubi, Barisé & Stochaj, 2015). Primary antibodies were used at the following dilutions: hsc70 (1:10,000), HspBP1 (1:400), and actin (1:100,000; Chemicon; EMD Millipore, Hayward, CA, USA).

Statistics

Data were obtained for at least three independent experiments for microscopy and Western blot analyses. Significant differences (p < 0.05) were identified with One-way ANOVA and Bonferroni post-hoc analysis. Results are shown as means + SEM. Linear regression analysis was performed in Microsoft Excel; the equations and R² values are included in the scatter plots. The Pearson correlation coefficient was computed using the standard formula r = cov (hsc70, HspBP1)/[σ_Hsc70 × σ_HspBP1].

Rationale for the quantitative analyses of the co-chaperone HspBP1 and its binding partner hsc70

The objective of this study was to define the reciprocal relationship between HspBP1 and hsc70 and gain quantitative spatial information on their possible interplay under normal and stress conditions. Our work focused on hsc70, rather than other HspBP1 interactors, for several reasons. The BioGrid database (Chatr-Aryamontri et al., 2015) lists hsc70 (also referred to as HspA8) as the most frequently identified binding partner for HspBP1. Hsc70 is essential for mammalian cell survival (Daugaard, Rohde & Jäättelä, 2007 and references therein). Moreover, primate hsc70, but not inducible hsp70, supports growth of the budding yeast Saccharomyces cerevisiae (Tutar, Song & Masison, 2006), emphasizing the important and conserved role of hsc70. The chaperone is constitutively synthesized; it is thus available to interact with HspBP1 in the absence and presence of stress. In the human brain, aging and neurodegeneration, conditions associated with chronic stress, reduce the expression of HspBP1 and hsc70 genes, while hsp70 gene expression is upregulated (Brehme et al., 2014). This may indicate that HspBP1 and hsc70 act in a coordinated fashion, at least during chronic stress. Based on their importance for basic biological processes (see ‘Introduction’) and human disease, hsc70 and HspBP1 as well as their interplay are of particular interest to human physiology.

Our earlier work provided initial studies of the nuclear and nucleolar association of hsc70 (Banski et al., 2010; Kodiha et al., 2005), whereas little is known about the subcellular distribution of HspBP1. As previously, the current studies use HeLa cells as an established model system to examine heat shock proteins and their subcellular location (Banski et al., 2010; Kodiha et al., 2005). Figure 1A shows the domain organization of HspBP1 and hsc70 and the regions that participate in the interaction between both proteins.

Assessment of the subcellular distribution of hsc70 and its co-chaperone HspBP1 during different growth conditions

To define the impact of a changing environment on the chaperone/co-chaperone pair hsc70/HspBP1, we evaluated endogenous proteins by quantitative immunofluorescence. This method was selected to avoid the complications associated with cell fractionation (Kodiha et al., 2014 and references therein). The localization of hsc70 and HspBP1 was monitored in non-stressed cells, during heat shock and at different time points of post-stress recovery (1, 2 and 3 h). Hsc70 and HspBP1 localized to the nucleus and cytoplasm in unstressed HeLa cells (Fig. 1B), and the exposure to heat shock induced a robust accumulation of hsc70 in the nucleus. During the recovery period, hsc70 began to concentrate in nucleoli (Fig. 1B and Banski et al., 2010), which were identified as “dark holes” in the DAPI image (Fig. S1; Kodiha, Banski & Stochaj, 2011). Under the same stress conditions, there was little effect on the overall distribution of HspBP1 (Fig. 1B). Notably, HspBP1 remained largely excluded from nucleoli. In summary, stress markedly impinged on the steady-state distribution of hsc70, but only minor changes were observed for HspBP1.

Quantification of the compartment-specific changes in hsc70 and HspBP1 localization

The proper assessment of protein redistribution can be complicated by changes in protein abundance, such as de novo synthesis or degradation. We therefore compared the cellular concentrations of hsc70 and HspBP1 in control, heat-shocked and recovering cells. To this end, Western blotting was performed on whole-cell extracts (Fig. 2), and protein levels were quantified for the time points relevant to our microscopic studies. As compared to control cells, hsc70 and HspBP1 levels did not change significantly throughout the stress and recovery period. This is consistent with results we reported earlier for hsc70 (Banski et al., 2010). These properties distinguish hsc70 and HspBP1 from other components of the chaperone network, because many chaperone and co-chaperone concentrations increase upon stress (Hageman et al., 2007 and references therein).

To determine whether hsc70 and HspBP1 share additional properties, their subcellular location was measured with quantitative microscopy methods that we established previously (Kodiha, Brown & Stochaj, 2008; Kodiha, Mahboubi & Stochaj, 2014). As shown in Fig. 1C, the distribution of hsc70 changed in response to stress, as cells displayed a time-dependent increase in nuclear fluorescence. At the same time, cytoplasmic signals were reduced. (It should be noted that the figure shows pixel intensities/area. Therefore, the gain in nuclear fluorescence is not identical to the loss of cytoplasmic signals). Changes in hsc70 distribution were especially pronounced during the recovery period, leading to a significant increase of the nucleus/cytoplasm ratio (Fig. 1C). Interestingly, this global analysis uncovered only minor effects for HspBP1, because none of the conditions caused a significant relocation of the co-chaperone.

Taken together, our new data demonstrate significant differences in the subcellular distribution of hsc70 and HspBP1 upon stress. Under these conditions, there was no strong link between the steady-state distribution of hsc70 and its co-chaperone HspBP1. This could suggest that during heat shock and stress recovery hsc70 and HspBP1 contribute –at least in part- to different functions. Our hypothesis is supported by the growing list of HspBP1 non-chaperone binding partners that participate in cytoplasmic and nuclear functions (Chatr-Aryamontri et al., 2015).

Single-cell analyses for HspBP1 and hsc70

Results presented in Fig. 1 demonstrate significant global changes for the subcellular hsc70 distribution, with only small changes for HspBP1. However, the experiments above do not provide information on whether these changes affect all or only a subpopulation of cells. To address this point, we went beyond the analyses in Fig. 1, and data were examined at the single-cell level.

Hsc70 and HspBP1 were evaluated in at least 115 cells for each time point. Pixel intensities were quantified in the nucleus and cytoplasm for hsc70 and HspBP1 in individual cells, and results were normalized to non-stress controls. In Fig. 3, data were plotted in ascending order for each individual nucleus and cytoplasm. They show that during hyperthermia and stress recovery the relative abundance of HspBP1 (purple) and hsc70 (green) shifted for the whole cell population. This applied to the nucleus, where a global increase in hsc70 abundance was detected. Interestingly, we uncovered that HspBP1 levels in the cytoplasm rise, while they diminish for hsc70.

To identify a possible link between HspBP1 and hsc70 abundance in individual cells, we performed regression analyses. Figure 4 plots single-cell fluorescence intensities for HspBP1 as a function of hsc70 signals. Plots were generated for (i) the ratio nucleus/cytoplasm, (ii) pixel intensities in the nucleus, and (iii) pixel intensities in the cytoplasm. Figure 4 suggests a linear relationship between the hsc70 and HspBP1 levels, both in the nucleus and cytoplasm under control and heat shock conditions. Interestingly, the slope of the regression line changed upon heat shock and during recovery, and the distribution of single-cell data were much more scattered in recovering cells (Fig. 4, see graphs for 1, 2, 3 h recovery). A likely interpretation of our work is that heat shock and stress recovery alter the interplay between this chaperone/co-chaperone pair.

To obtain a better understanding of this relationship, we computed the Pearson’s correlation coefficient (r) for single-cell data sets (Figs. 5A and 5B). Heat shock substantially increased the correlation coefficient in the nucleus. During recovery, there was a strong correlation between hsc70 and HspBP1 abundance in the cytoplasm at 1 and 3 h (r > 0.8). Interestingly, we observed the lowest r value in the nucleus at 1 h recovery.

Taken together, quantitative single-cell analyses showed that the stress-induced hsc70 relocation occurs in the whole cell population. Moreover, we uncovered small shifts for HspBP1 abundance, especially in the cytoplasm during stress recovery. On the single-cell level, the relationship between hsc70 and HspBP1 can be described by a linear function for control and heat shock conditions.

Overall, our results support the idea that the hsc70/HspBP1 interplay is (i) stress dependent, (ii) different in the cytoplasm and nucleus and (iii) affects the entire cell population.

Nucleolar association of hsc70 and HspBP1 in control and stressed cells

Our earlier work (Banski et al., 2010; Kodiha et al., 2005) and Fig. 1 showed that hsc70 concentrated in nucleoli in a stress-dependent fashion. It was thus important to determine whether HspBP1 locates to nucleoli as well. This idea was tested with methods we developed previously (Kodiha, Banski & Stochaj, 2011). To this end, we applied protocols to automatically demarcate nucleoli (Fig. S1) and quantify signals within the nucleolar compartment. The measurements of nucleolar signals revealed hsc70 nucleolar accumulation during the recovery period (Figs. 1 and 6 and Banski et al., 2010). Unlike its binding partner hsc70, HspBP1 did not concentrate markedly in nucleoli at any of the examined time points (Figs. 1 and 6).

For the experiments presented here, the most pronounced hsc70 nucleolar accumulation occurred at 1 h recovery. Interestingly, this coincided with the lowest Pearson’s correlation coefficient (r = 0.11) for nuclear hsc70/HspBP1 data sets, as calculated for single-cell data (Fig. 5). The non-homogeneous hsc70 distribution in nuclei, due to nucleolar accumulation, may contribute to the low r value.

These new findings for HspBP1 are significant for the understanding of its co-chaperone function. Several studies support the idea that hsp70 family members play a role in the nucleolus, where they likely maintain and restore nucleolar organization and integrity during stress (Banski, Kodiha & Stochaj, 2010; Banski, Kodiha & Stochaj, 2011; Kodiha et al., 2005; Kodiha & Stochaj, 2013; Pelham, 1984). Some co-chaperones, such as hsp40 (Hattori et al., 1993, reviewed in Banski, Kodiha & Stochaj, 2011; Kodiha, Frohlich & Stochaj, 2012) are also present in the nucleoli of stressed cells and could thus modulate chaperone functions (Banski, Kodiha & Stochaj, 2011). The data presented here indicate a different role for HspBP1, as the co-chaperone will likely have little impact on nucleolar-specific hsc70 activities during heat shock and recovery.

Colocalization of hsc70 and HspBP1 in control and stressed cells

To further examine the interplay between hsc70 and HspBP1, we quantified their colocalization at the single-cell level and under different conditions. This goes beyond the measurements in Figs. 1, 3 and 4, as we focused on the area of overlap in nuclear and cytoplasmic compartments. To this end, we determined the colocalization for individual pixels, with a pixel size of 0.14 µm².

Two different aspects of the chaperone/co-chaperone colocalization were assessed. First, the colocalization of hsc70 with HspBP1 was quantified (Fig. 7A; Hsc70/HspBP1, green bars); second, the colocalization of HspBP1 with hsc70 was measured (Figs. 7A and 7B; HspBP1/Hsc70; purple bars and pie charts).

The most significant changes for hsc70/HspBP1 occurred in the cytoplasm during stress recovery. The colocalization increased from approximately 60% to 80%; this means that during recovery ∼80% of the area occupied by hsc70 also contained HspBP1. One possible interpretation is the retention of hsc70 in the cytoplasm through interaction with HspBP1, whereas hsc70 which does not co-localize with HspBP1 moves to the nucleus.

For HspBP1, the changes in colocalization were particularly profound in the nucleus. In unstressed cells, 38% of nuclear HspBP1 colocalized with hsc70 (Fig. 7). Nuclear colocalization increased significantly during and after heat shock, rising to >65%. By contrast, the colocalization of HspBP1 with hsc70 in the cytoplasm was high in control and heat-stressed cells, but diminished during recovery (Fig. 7; from 80% to ∼50%).

In summary, our data indicate that the interplay between HspBP1 and hsc70 is dictated by the availability of each of the two proteins in the nuclear and cytoplasmic pools. Stressed cells can quickly accomplish this by increasing the local protein concentration (for instance, cytoplasmic HspBP1, Fig. 3) or by promoting nucleocytoplasmic relocation (hsc70, Fig. 1). Thus, the regulation of chaperone/co-chaperone function not only depends on their absolute cellular concentrations, but also on their subcellular distribution and colocalization within the compartment.