Alternative bacterial two-component small heat shock protein systems

Two sHsps of D. radiodurans Are Active as Molecular Chaperones in Vitro.

Hsp17.7 and Hsp20.2 were expressed in E. coli and purified to homogeneity. To ensure that both tag-free proteins were properly folded, their secondary structures and stabilities against thermal unfolding were monitored by CD spectroscopy (Fig. S1B). Similar to other sHsps (33), Hsp17.7 and Hsp20.2 show CD spectra typical for β-sheet proteins and melting temperatures of 67.3 °C and 57.8 °C, respectively (Fig. S1B). The structural integrity of the proteins is not influenced at temperatures below 48 °C.

To assess the chaperone activity of Hsp20.2 and Hsp17.7, we first tested their ability to suppress protein aggregation at elevated temperatures, a key trait of sHsps (4, 8, 34). We used citrate synthase (CS, 48 kDa), firefly luciferase (61 kDa), insulin (3 kDa), and lysozyme (14 kDa) as model substrates. Hsp17.7 was able to suppress the thermal aggregation of both CS and luciferase in a concentration-dependent manner, as well as the aggregation of reduced lysozyme (Fig. 1A and Fig. S2). Additionally, the reduction-induced aggregation of the insulin B chain at 25 °C and at 43 °C was suppressed, showing similar ratios of sHsp: substrate needed for half-maximum suppression at both temperatures (Fig. 1A and Fig. S2 E and G). Thus, it seems that the chaperone activity of Hsp17.7 is not affected by temperature. Because the amount of substrate molecules to sHsps is dependent on the size of the substrate, for peptide substrates like insulin, usually substoichiometric concentrations of sHsps are sufficient (35, 36). The ratio of half-maximum aggregation suppression, however, does not correlate with the binding stoichiometry in this type of assay, because the decrease of the concentration of aggregation-prone insulin is sufficient to inhibit aggregation even if not the majority of insulin molecules is captured by the sHsp. In comparison with Hsp17.7, Hsp20.2 is the more potent chaperone because lower amounts of Hsps20.2 were sufficient to fully suppress the heat- or reduction-induced aggregation of the substrate proteins (Fig. 1A and Fig. S2).

In the case of CS, monitoring the kinetics of thermal inactivation allows conclusions on the interaction of a molecular chaperone with unfolding intermediates and on the stability of the chaperone–substrate complex (37). For Hsp17.7, we observed a deceleration of CS inactivation (Fig. 1B), indicating a transient interaction between Hsp17.7 and CS unfolding intermediates, as was described for Hsp26 and Hsp42 (38). In contrast, the thermal inactivation of CS was not affected by the presence of Hsp20.2, pointing toward a more stable complex between Hsp20.2 and its substrate.

Quaternary Structures of Hsp17.7 and Hsp20.2 Are Strikingly Different.

To determine the size of the Hsp17.7 oligomer, we used size exclusion chromatography (SEC). The resulting single peak had an apparent molecular mass of 150 kDa, which could correspond to ∼8 subunits (Fig. 2A). Next we investigated the influence of heat stress on the oligomeric size of Hsp17.7, because several sHsps dissociate at elevated temperatures (39, 40). At 43 °C, Hsp17.7 eluted with an identical apparent mass as observed at 25 °C (Fig. 2A), demonstrating that the quaternary structure of Hsp17.7 is unaffected by changes in temperature. A more detailed analysis of the quaternary structure of Hsp17.7 by analytical ultracentrifugation (aUC) sedimentation velocity experiments showed a surprisingly low S₂₀w value of 2.1 S, corresponding to a molecular mass of ∼40 kDa and a frictional coefficient (f/f0) of 1.78, independent of the protein concentration used (Fig. 2C and Fig. S3A). Thus, the aUC experiments suggest that Hsp17.7 is an elongated dimer. Further refinement of the aUC analysis using a 2D grid analysis (2-DSA) and Monte Carlo simulations revealed a single species with an S₂₀w value of 2.08 S, a frictional coefficient of 1.78, and a molecular mass of 37 kDa (Fig. S3B), confirming that Hsp17.7 is an elongated dimer in solution that does not change its size at elevated temperatures. This sets Hsp17.7 apart from most other sHsps studied to date.

In contrast, Hsp20.2 exhibited one main peak in SEC (Fig. 2B), corresponding to a large oligomeric complex of 600–800 kDa, corresponding to roughly 36 subunits and a peak shoulder, indicative of an 18mer. No temperature-dependent shift to smaller oligomeric species was observed at 43 °C. aUC experiments yielded a sedimentation coefficient of 22 S for the high molecular weight oligomer and 11 S for the smaller oligomer (Fig. 2D and Fig. S3C). Using interference optics we were able to fit the data to a frictional coefficient of 1.23 and a molecular mass of ∼700 kDa and ∼350 kDa, in accordance with a 36mer and an 18mer.

To investigate whether the assembly state of the two sHsps is similar in vivo, their separation behavior in D. radiodurans extracts was analyzed by native PAGE. Both proteins were detected at the same position as the purified proteins (Fig. S3D). For Hsp20.2, the two visible bands correlated well with the observed two main species. For Hsp17.7, a prominent dimer band was detected. When purified Hsp20.2 was added to cell lysate of D. radiodurans at physiological conditions (25 °C), an additional high molecular mass band appeared, most likely indicating the formation of substrate complexes with unfolding proteins (see below). Furthermore, immunoblot analysis suggests that the two proteins do not form mixed complexes, and aUC experiments with mixtures of the two proteins support this finding (Fig. S3E).

To elucidate the quaternary structure of the oligomers in more detail, the samples were visualized by negative stain electron microscopy (Fig. S4 A and B). We could confirm the small size of Hsp17.7, with dimensions in agreement with the postulated dimers (Fig. 3A and Fig. S4B). Hsp20.2 formed heterogeneous, mostly spherical oligomers with diameters between 9 and 18 nm (Fig. 3A and Fig. S4B). The main population had a diameter of ∼13 nm.

For Hsp17.7, crystallization conditions could be determined that resulted in a crystal diffracting to 2.4-Å resolution (Table S1). The atomic structure obtained after molecular replacement shows that Hsp17.7 crystallized as a dimer. Similar to other structures of sHsps, the N-terminal region (amino acids 1–45) and the C-terminal extension (amino acids 148–166) of Hsp17.7 are flexible and therefore lack defined electron density. The monomers adopt the characteristic α-crystallin core fold, consisting of two antiparallel β-sheets, which are packed in parallel layers (Fig. 3B and Fig. S4). In agreement with other family members (1, 16), the peripheral β6-strand in the extended loop stabilizes the dimer of Hsp17.7 by interacting with the β2-strand of the partner monomer. A backbone superposition (Fig. S4E) of sHsps from different organisms showing this typical strand-exchanging dimer interface revealed structural divergence for the extended loop. Interestingly, the backbone of the extended loop of Hsp17.7 structurally matches best with HspA from the plant pathogenic bacterium Xanthomonas sp. (XAC1151) (41, 42). However, XaHspA assembles to a 36mer in vitro. In comparison with XaHspA, MjHsp16.5, and TaHsp16.9, the extended loop of Hsp17.7 adopts the narrowest geometry (Fig. S4E), caused by the sequence gap between the β5- and β6-strands (Fig. S4C). The Hsp17.7 dimers are packed in a helical, line-up order (Fig. S4D). Interestingly, the C-terminal IXI/V motif still forms a crystal contact to the neighboring dimer, which resembles the oligomerization interface also found in the structures of MjHsp16.5 and TaHsp16.9 (4). However, the β10-strand formed by the IXI/V motif is oriented parallel to the β4-strand, which is opposed to its orientation in MjHsp16.5 and TaHsp16.9 oligomers (Fig. S4F). In this context it should also be noted that the C-terminal extension of Hsp17.7 is peculiarly short, showing a sequence gap between the α-crystallin domain and the IXI/V motif (Fig. S4C).

Substrate Complexes of Hsp20.2 and Hsp17.7.

To determine the active chaperone species and the stoichiometry of the substrate interaction, aUC experiments were performed. We exploited the tremendous spectral differences between Hsp20.2 (A₂₈₀ = 4,470 M⁻¹⋅cm⁻¹) and lysozyme (38,000 M⁻¹⋅cm⁻¹) and conducted a multisignal sedimentation velocity analysis (43). A 16-S complex was detected, which did not change when the Hsp20.2 concentration was varied (Fig. 4A). Intriguingly, the observed size of the Hsp20.2–lysozyme complex lies between the Hsp20.2 18mer and the 36mer, indicating that the 18mer is the substrate binding species. Collecting aUC data with absorbance and interference optics in parallel, we were able to spectrally resolve the complex. It showed a molecular mass of approximately 550 kDa and a stoichiometry of 3:1 Hsp20.2 to lysozyme. Thus, the complex most likely consists of 18 subunits Hsp20.2 and six molecules of bound substrate.

To analyze the interaction of Hsp17.7 with lysozyme, fluorescently labeled Hsp17.7 was incubated with lysozyme, and complex formation was analyzed by aUC. In the presence of reduced lysozyme, the sedimentation behavior of Hsp17.7 did not change compared with Hsp17.7 alone (Fig. 4B). Thus, no stable Hsp17.7–substrate complex could be detected consistent with the results obtained for CS inactivation (Fig. 1B).

The depletion of IbpA/B in E. coli leads to enhanced aggregation of proteins (7) and to a reduction of maximum cell numbers in the stationary phase. The plasmid-based overexpression of IbpA/B in the respective strain is able to complement both phenotypes (Fig. S5 C and D). When Hsp17.7 or Hsp20.2 was overexpressed no complementation of the IbpA/B phenotype was achieved (Fig. S5 C and D), indicating that both systems are not comparable to each other and that the D. radiodurans sHsps cannot even compensate partially for the E. coli Ibps.

To investigate the chaperone activity of Hsp20.2 and Hsp17.7 in vivo, we analyzed aggregated protein in stressed and unstressed D. radiodurans lysates (Fig. S6A). Interestingly, only Hsp20.2 was associated with aggregates, whereas Hsp17.7 stayed in the soluble fraction. Additionally, when D. radiodurans lysates were separated by native PAGE after 30 min at 43 °C (Fig. S6B), Hsp20.2 was found in the high molecular weight aggregates. To determine which substrates are bound by Hsp20.2, the high molecular aggregates were analyzed by mass spectrometry (see below and Table S2). This analysis further confirmed the presence of Hsp20.2 and the absence of Hsp17.7 in these complexes, and thus differences in substrate binding. When we isolated sHsp–substrate complexes by immunoprecipitation, only for Hsp20.2 substantial protein amounts were coisolated from D. radiodiurans lysates (Fig. S6C). Subsequent mass spectrometry analysis allowed us to identify ∼100 potential substrate proteins bound by Hsp20.2 (Table S2), including a number of metabolic enzymes, ribosomal proteins, and elongation factors. These aggregation-prone proteins seem to be the basic substrates of sHsps in D. radiodurans. Similar to other sHsps (38, 44), the molecular mass distribution of the substrates of Hsp20.2 is very broad, ranging from 15 to more than 100 kDa. Interestingly, besides these potential substrates, a number of S-layer proteins including the gene DR1115 (NP 294839.1), which is encoded directly upstream of hsp20.2, were detected. Usually these proteins are extracellular, but they might be recognized as substrates on their secretion pathway. Within the identified proteins, DnaK and GroEL were also detected, indicating a potential cooperation of these chaperone systems in substrate release similar to observations for E. coli (7, 10). To investigate potential synergistic effects with other molecular chaperones, we analyzed the ability of the D. radiodurans sHsps to cooperate with the E. coli Hsp70 (DnaK/DnaJ/GrpE; KJE), Hsp100 (ClpB), and Hsp60 (GroEL/GroES) chaperone systems. We determined the refolding of substrates from preformed sHsp complexes with thermally unfolded firefly luciferase (Fig. 4C and Fig. S7) and chemically denatured CS (Fig. 4D). A few percent of luciferase were reactivated spontaneously (below 2%) or in the presence of the sHsps only (below 5%). However, when Hsp20.2 was present during the inactivation reaction, 10–15% of the luciferase activity was regained in presence of GroEL or ClpB and 15–20% when KJE was added. The combination of the ClpB and the KJE systems resulted in the refolding of nearly 40% of luciferase. Similar to results for E. coli IbpB (7), the combination of all three ATP-dependent chaperone systems indicated that the GroE system is able to compensate for the role of KJE to some extent, but no additive or synergistic effects on luciferase refolding were achieved when both GroE and KJE were present. In the presence of Hsp17.7, on the other hand, only minor levels of luciferase activity were regained, independent of the combination of chaperone systems added. Hsp17.7 seems to interact only transiently, releasing its substrate spontaneously without the help of other chaperone systems. When both D. radiodurans sHsps were present in the reaction, no additive or synergistic effects were observed, indicating that the two sHsps work independently of each other. Interestingly, in experiments with chemically denatured CS, which was diluted in a solution containing KJE and Hsp17.7 or Hsp20.2, respectively, we detected a slight acceleration of the refolding kinetics in the presence of Hsp17.7, whereas the total amount of recovered activity was similar, independent of the sHsp present (Fig. 4D). Again, the two sHsps did not show any synergistic effects in this assay. The finding that Hsp20.2 and Hsp17.7 act independently of each other is in clear contrast to studies on E. coli IbpA/B, which work together synergistically in refolding with the help of other, ATP-dependent chaperone systems (24, 29).

Materials.

E.coli DnaK, DnaJ, GrpE, and pig heart CS were purified as described elsewhere (8, 50). Yeast SUMO (small ubiquitin-like modifier) protease representing the C-terminal fragment (403–621) of ubiquitin-like-protease 1 (51) was amplified by PCR from cDNA and cloned into pET28b (Invitrogen), yielding N-terminally His₆-tagged protein. The protein was purified by Ni²⁺-chelating chromatography (GE Healthcare) according to the manufacturer’s protocol. Bovine insulin and hen egg lysozyme were obtained from Sigma.

Expression and Purification of sHsps.

For purification of Hsp17.7 and Hsp20.2, N-terminal 6× His plus SUMO fusions were constructed. The proteins were expressed in E. coli BL21 and purified by Ni²⁺ chelating and SEC according to the manufacturer's protocols (GE Healthcare). Details are provided in SI Materials and Methods.

Analysis of Chaperone Activity.

Thermal aggregation of model substrates, inactivation, and refolding of CS was performed as described elsewhere (8). For refolding of chemically denatured CS by the DnaK chaperone machinery, 2 µM DnaK, 2µM DnaJ, and 0.5 µM GrpE were used in the presence of 1 µM Hsp17.7 or Hsp20.2.

Inactivation and reactivation of firefly luciferase (Photinus pyralis; Promega) was assayed as described elsewhere (8). For each measurement, native luciferase was diluted to a final concentration of 4 µg/mL (80 nM) and incubated for 10 min at 43 °C in the presence or absence of sHsps. Then different combinations of chaperones were added, and the reactions were shifted to 25 °C for 65 min for reactivation. The activity of luciferase was compared with the activity of untreated, native luciferase, which was set to 100%. Luciferase activity was measured using a thermostated Tecan GENios Microplate Reader.

Analytical SEC.

SEC was performed as described previously (35). Buffer conditions were 40 mM Hepes/KOH (pH 7.4) and 150 mM NaCl, and a flow rate of 0.5 mL/min was used. The TSK 4000 pw column (TosoHaas) was calibrated using the Sigma gel filtration molecular weight marker kit (Sigma) including carbonic anhydrase, 29 kDa; albumin, 66 kDa; alcohol dehydrogenase, 150 kDa; amylase, 200 kDa; apoferritin, 443 kDa; and thyroglobulin, 669 kDa.

aUC Sedimentation Velocity Experiments.

Analytical ultracentrifugation was carried out with a ProteomLab XL-I (Beckman) supplied with absorbance and interference optics. Sample sedimentation was monitored at 280 nm or 230 nm, continuously scanning with a radial resolution of 30 microns. Data analysis was carried out with the Ultrascan and the USLIMS platform (52). Details are provided in SI Materials and Methods.

Protein Crystallization and Structure Determination.

Crystals of sHsp17.7 were grown at 20 °C by using the sitting drop vapor diffusion method. The drops contained equal volumes of protein (12 mg/mL) and reservoir solution [0.2 M ammoniumacetate, 0.1 M bis·Tris (pH 6.5), 25% PEG 3350]. The dataset was collected on a Bruker Microstar/X8 Proteum (Bruker AXS) with a Cu rotating anode (λ = 1.54 Å) and processed with the Proteum software suite (Bruker AXS). Hsp17.7 crystallized in the trigonal space group P3₁21 with cell parameters of a,b = 51.3 Å, c = 80.5 Å. For structure determination, molecular replacement was performed in Phaser (53) using the coordinates of HspA from Xanthomonas sp. (XAC1151; PDB-ID: 3GLA) as a starting model. Details are provided in SI Materials and Methods.

Electron Microscopy and Image Processing.

For negative stain electron microscopy, 5 µL of the protein sample (0.05 mg/mL protein in 50 mM Hepes/KOH, 75 mM NaCl, pH 7.4) were adsorbed for 2 min onto carbon-coated grids that were glow discharged in air. Samples were negatively stained with uranyl acetate [1.5% (wt/vol) at pH 4.5], and electron micrographs were recorded at a calibrated magnification of 58,000 using a JEOL JEM 100CX electron microscope operated at 100 kV. Details are provided in SI Materials and Methods.