The Specialized Hsp70 (HscA) Interdomain Linker Binds to Its Nucleotide-Binding Domain and Stimulates ATP Hydrolysis in Both cis and trans Configurations

Abstract

Proteins from the isc operon of Escherichia coli constitute the machinery used to synthesize iron–sulfur (Fe–S) clusters for delivery to recipient apoproteins. Efficient and rapid [2Fe-2S] cluster transfer from the holo-scaffold protein IscU depends on ATP hydrolysis in the nucleotide-binding domain (NBD) of HscA, a specialized Hsp70-type molecular chaperone with low intrinsic ATPase activity (0.02 min⁻¹ at 25 °C, henceforth reported in units of min^–1). HscB, an Hsp40-type cochaperone, binds to HscA and stimulates ATP hydrolysis to promote cluster transfer, yet while the interactions between HscA and HscB have been investigated, the role of HscA’s interdomain linker in modulating ATPase activity has not been explored. To address this issue, we created three variants of the 40 kDa NBD of HscA: NBD alone (HscA₃₈₆), NBD with a partial linker (HscA₃₈₉), and NBD with the full linker (HscA₃₉₅). We found that the rate of ATP hydrolysis of HscA₃₉₅ (0.45 min^–1) is nearly 15-fold higher than that of HscA₃₈₆ (0.035 min^–1), although their apparent affinities for ATP are equivalent. HscA₃₉₅, which contains the full covalently linked linker peptide, exhibited intrinsic tryptophan fluorescence emission and basal thermostability that were higher than those of HscA₃₈₆. Furthermore, HscA₃₉₅ displayed narrower ¹H^N line widths in its two-dimensional ¹H–¹⁵N TROSY-HSQC spectrum in comparison to HscA₃₈₆, indicating that the peptide in the cis configuration binds to and stabilizes the structure of the NBD. The addition to HscA₃₈₆ of a synthetic peptide with a sequence identical to that of the interdomain linker (L³⁸⁷LLDVIPLS³⁹⁵) stimulated its ATPase activity and induced widespread NMR chemical shift perturbations indicative of a binding interaction in the trans configuration.

Iron–sulfur (Fe–S) clusters comprise an ancient class of enzymatic cofactors whose biosynthesis occurs in a tightly regulated and intricately complex manner, despite their chemical simplicity (e.g., [2Fe-2S] or [4Fe-4S]) and spontaneous in vitro syntheses.^1,2 Proteins that utilize Fe–S clusters (Fe–S proteins) occur throughout nature and are commonly found in essential metabolic (e.g., mitochondrial complex III and aconitase) and biosynthetic (e.g., DNA helicases and polymerases) pathways.³ Because Fe–S proteins are synthesized in their apo form, they require post-translational insertion of clusters to yield mature, holoenzymes. However, because of the toxicity of iron ions and sulfide and the instability of free Fe–S clusters, cells have evolved dedicated proteinaceous machinery to safeguard cluster synthesis and delivery to recipient apo-Fe–S proteins.⁴ This machinery minimizes the level of cytotoxic reactive oxygen species (ROS) that can inflict irreversible macromolecular and cellular damage.⁵

Prokaryotes contain an iron–sulfur cluster (ISC) assembly system that leads to the maturation of nearly all Fe–S proteins under basal cellular conditions.⁴ In eukaryotes, the homologous mitochondrion-based ISC assembly system includes more than 30 proteins that function together to cause the maturation of both mitochondrial Fe–S proteins and, with assistance from the cytosolic iron–sulfur cluster assembly (CIA) system, nonmitochondrial (i.e., cytosolic, nuclear, and endoplasmic reticulum-based) Fe–S proteins.⁶ Because of the prevalence and biological significance of many Fe–S proteins, defects in Fe–S cluster biosynthesis and delivery manifest as a myriad of human diseases, including Friedreich’s ataxia, the most common recessive ataxia.^7,8 Moreover, recent evidence suggests that dysfunctional Fe–S cluster biogenesis could play a potential role in the onset and progression of Parkinson’s disease.⁹ To understand the biological basis of Fe–S cluster-related human diseases, elucidation of the molecular mechanisms that underlie Fe–S cluster assembly and transfer will prove to be critical. Many research groups over the past 15 years have utilized the ISC assembly proteins of Escherichia coli as a model system for investigating the structural and biophysical details of Fe–S cluster biogenesis.¹⁰ Because of the homology between E. coli and eukaryotic ISC assembly proteins, results from studies on bacterial ISC proteins often can be translated to the human system.

In E. coli, the protein components necessary to synthesize and transfer Fe–S clusters are encoded by the isc operon.¹¹ Among these proteins is HscA, a specialized Hsp70-type chaperone (66 kDa) that catalyzes Fe–S cluster transfer from the scaffold protein IscU to an acceptor apo-Fe–S-protein in an ATP-dependent reaction.¹² Similar to DnaK, the general Hsp70 chaperone in E. coli, HscA contains a 40 kDa nucleotide-binding domain (NBD) that binds to and hydrolyzes ATP,^13,14 but its 25 kDa substrate-binding domain (SBD) uniquely recognizes a central LPPVK motif in IscU as the substrate.¹⁵⁻¹⁸ Furthermore, numerous kinetic differences exist between the ATPase cycles of HscA and DnaK (Table S1 of the Supporting Information, with references therein): compared to DnaK, HscA displays a 10³-fold weaker binding affinity for ATP, an augmented rate of ATP association, and a >10³-fold higher rate of ADP dissociation.¹⁹ Uniquely, and in contrast to DnaK and most Hsp70s, HscA does not require a nucleotide-exchange factor (NEF) to reset its ATPase cycle.^20,21

Similar to that of other Hsp70s, nucleotide binding in the NBD of HscA allosterically regulates substrate (IscU) binding affinity in its SBD, with ATP-HscA (K_d^IscU = 37 μM) exhibiting an IscU binding affinity weaker than that of ADP-HscA (K_d^IscU = 1.6 μM).²² HscB, an Hsp40-type cochaperone, binds to and delivers [2Fe-2S]-IscU to the SBD of HscA (Figure 1B); together, IscU and HscB synergistically enhance HscA’s ATPase activity nearly 1000-fold,^20,22 much like the synergistic activation observed in DnaK by DnaJ and substrate peptides. We have previously shown that IscU exists in two slowly interconverting,²³⁻²⁵ metamorphic conformations, one more structured (S-state) and one more dynamically disordered (D-state), and its binding partners preferentially interact with one state over the other: HscA binds preferentially to the D-state, while HscB binds preferentially to the S-state (Figure 1B).^26,27 Hydrolysis of bound ATP to ADP alters the conformation of HscA to the form that binds the D-state of IscU, promoting concomitant release of the [2Fe-2S] cluster from holo-IscU. Inhibition of ATP hydrolysis in HscA (via an ATPase deficient mutant, HscA T212V) abolishes [2Fe-2S] cluster transfer,¹² and inactivation of the hscA gene leads to diminished cellular growth rates and activities of Fe–S proteins, including succinate dehydrogenase²⁸ and aconitase.²⁹

Figure 1.

(A) Degree of sequence identity between DnaK and HscA mapped onto the structure of ADP-bound DnaK (PDB entry 2kho): red for identically conserved residues, yellow for highly similar residues, green for similar residues, and black for no sequence similarity. Roman numerals indicate the subdomains of DnaK’s NBD; note the lack of sequence similarity in subdomain IIB where HscA’s lone tryptophan is located. The highly conserved interdomain linker sequence (HscA numbering) is highlighted according to the colors noted above. Linker inset abbreviations: Ec, E. coli; Hs, Homo sapiens; Sc, Saccharomyces cerevisae. (B) Architecture of HscA and the design of variants used in this study. The sequence of the interdomain linker (truncated or otherwise) is listed next to each construct. (C) SDS–PAGE analysis of purified HscA variants. Lane 1 contains molecular weight (MW) markers with their associated MW designated at the left. Lanes 2—5 contain the respective HscA variant listed above the gel.

HscA is thus vital to the maturation of Fe–S proteins, yet the structural basis of its unique nucleotide binding and dissociation features has remained unresolved, for no atomic-level information about HscA’s NBD has been reported. Moreover, whereas studies of DnaK (E. coli) and Hsc70 (Bos taurus cDNA that was expressed in and purified from E. coli) have demonstrated that the linker modulates interdomain communication and serves as a switch that activates ATP hydrolysis,³⁰⁻³³ the question of whether this holds true for HscA, which relies on ATP hydrolysis to transfer [2Fe-2S] clusters from holo-IscU to recipient apo-Fe–S proteins, remains.¹² Thus, we sought to investigate the specific role of HscA’s interdomain linker in an effort to probe the mechanism by which HscA catalyzes ATP hydrolysis to promote Fe–S cluster transfer.

Here, we report detailed information about the isolated NBD of HscA and how the presence of a shortened linker or full-length linker peptide affects its functional properties. We also explore the effect of adding a synthetic peptide with the sequence of the full-length linker on the properties of the isolated NDB.

Materials and Methods

Design of NBD Variants

No high-resolution structural information exists for full-length HscA or its isolated NBD; however, numerous structures of homologous Hsp70s have been determined in various substrate- and nucleotide-bound states.³⁴⁻³⁸ Therefore, to create variants of the HscA NBD containing a partial or full interdomain linker, we utilized structures of ADP- and ATP-bound DnaK (PDB entries 2kho and 4jne, respectively) and a bioinformatics approach to identify conserved linker residues among various Hsp70s.^35,36 Eight of the nine residues in HscA’s interdomain linker sequence (L³⁸⁷LLDVIPLS³⁹⁵) are conserved with DnaK (E. coli), GRP75 (Homo sapiens), HSC70 (H. sapiens), and Hsp70 (Saccharomyces cerevisiae) (Figure 1A). Moreover, because actin and DnaK NBD share only ∼10% sequence homology yet both structures display similar folds characteristic of the actin/hexokinase superfamily,³⁹ we hypothesized that prior structural and functional studies of DnaK would be relevant to HscA, which shares 42.4% sequence homology with DnaK (Figure 1A).

Within the wild-type (WT) HscA-encoding pTrc plasmid,¹⁴ stop codons were introduced at L³⁸⁷, D³⁹⁰, and L³⁹⁶ to create HscA₃₈₆ (isolated NBD), HscA₃₈₉ (NBD with a truncated linker; L³⁸⁷LL³⁸⁹ motif), and HscA₃₉₅ (NBD with a full linker; L³⁸⁷LLDVIPLS³⁹⁵), respectively (Figure 1B).

Protein Production

The QuikChange II site-directed mutagenesis kit (Agilent) was used to create the HscA NBD constructs mentioned above by introducing appropriate stop codons at HscA residues 387, 390, and 396 in the WT pTrc(HscA) vector,¹⁴ and these mutations were verified with DNA sequencing. Rosetta Blue DE3 competent cells were transformed with the mutated pTrc(HscA) expression plasmids and selectively grown on medium plates containing 100 μg/mL ampicillin. For the production of unlabeled protein, cell cultures were grown in 1 L of LB medium until the optical density reached ∼1.0, upon which IPTG was added to a final concentration of 0.4 mM. Protein expression continued for 4 h, after which cells were harvested and frozen at −80 °C until they were used. For the production of [U-¹⁵N]HscA₃₈₆(T212V) and -HscA₃₈₆, cells were grown in 1 L of M9 minimal medium with 100 μg/mL ampicillin and 1 g/L ¹⁵NH₄Cl instead of ¹⁴NH₄Cl. Protein expression occurred as described above. Unlabeled IscU and HscB samples were prepared as described previously.²⁴

Protein Purification

To purify HscA NBD samples, anion-exchange chromatography was performed using a DEAE Bio-Gel column equilibrated with 50 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, and 1.0 mM DTT (TED) buffer. Protein elution occurred over a 0 to 0.25 M NaCl gradient, and fractions containing HscA NBDs of interest were identified via SDS–PAGE. HscA-containing fractions were concentrated and subsequently subjected to size-exclusion chromatography on a HiLoad 16/60 Superdex 75 column (GE Healthcare). Protein was eluted in TED buffer supplemented with 150 mM NaCl. The concentration of HscA NBDs was determined by recording the absorbance at 280 nm in a 1 cm path-length cuvette with a theoretical extinction coefficient of 17420 M^–1 cm^–1. Full-length HscA samples were quantified and purified as described previously, except that the reversed-phase size-exclusion chromatography step was eliminated.¹⁴ The purified samples of HscA variants were analyzed for purity via SDS–PAGE (Figure 1C) and mass spectrometry (data not shown).

Synthetic Peptide

A peptide corresponding to HscA residues L³⁸⁷LLDVIPLS³⁹⁵ was synthesized at the University of Wisconsin—Madison Biotechnology Center. It was partially dissolved in TED buffer [without DMSO (see below)] for the acquisition of NMR data at an upper limit concentration of 3 mM or HMKD buffer with 10% DMSO (v/v) and added to final concentrations of 5, 10, 50, 100, 150, 200, and 500 μM ATPase assays to ensure solubility. Above this linker peptide concentration, more DMSO was required to maintain the solubility; however, the resultant increase in the level of DMSO inhibited the ATPase activity of HscA₃₈₆, and we were unable to pursue soluble peptide concentrations above 500 μM.

ATPase Assays

To quantify the rate of ATP hydrolysis for each HscA variant, steady-state assays were conducted at 25 °C in HMKD buffer [50 mM HEPES (pH 7.3), 150 mM KCl, 10 mM MgCl₂, and 1 mM DTT] containing 1 μM HscA and 1 mM ATP as previously reported.¹⁹ A standard curve of known P_i concentrations versus the respective absorbance at 360 nm was generated to determine experimental ATPase activities of HscA variants. ATPase reactions were conducted with three or more replicates, and ATPase rates were calculated by continuously monitoring the amount of phosphate released using the enzyme-coupled EnzChek Phosphate Assay Kit (Molecular Probes, Life Technologies). To determine the Michaelis–Menten constants (K_m) of HscA₃₈₆ and HscA₃₉₅, we added ATP to final concentrations of 2, 4, 10, 40, 100, and 1000 μM and plotted the resultant ATPase activities versus ATP concentration. Reactions were conducted with three replicates, and data were plotted as the mean ATPase activity for each ATP concentration. Using nonlinear regression analyses, the resultant curves were fit to the Michaelis–Menten equation to yield K_m and V_max values.

NMR Spectroscopy

Solution NMR solvents contained either HMKD or TED buffer (with 10 mM MgCl₂ and 150 mM NaCl), 7% D₂O, 0.02% sodium azide, and nucleotide or peptide, as required. Multidimensional NMR data were processed with NMRPipe⁶⁵ and visualized with Sparky.⁶⁶ For the acquisition of nucleotide binding data, [U-¹⁵N]HscA₃₈₆(T212V) was added to a final concentration of 0.3 mM. ATP or ADP was added in excess (10 mM), and spectra were acquired at 25 °C with an 800 MHz Agilent Unity-Inova spectrometer. For the acquisition of linker binding data, [U-¹⁵N]HscA₃₈₆ was added to a concentration of 0.2 mM, and the synthetic linker peptide was added to a final concentration of 3 mM in TED buffer; spectra were acquired with a 900 MHz Agilent Unity-Inova spectrometer equipped with a cryogenic probe.

Intrinsic Tryptophan Fluorescence

HscA NBD samples containing 1 μM protein in HMKD buffer were excited at 295 nm, and fluorescence emission was monitored from 305 to 400 nm. As HscA contains only one tryptophan residue in its NBD (W291), the relative fluorescence emission could be directly compared between HscA₃₈₆ and HscA₃₉₅. Background fluorescence was corrected for by subtracting fluorescence from an equivalent volume of solvent alone, and fluorescence data for each sample were collected at least three times to ensure their validity. Data were acquired with a QM-1 fluorometer (Photon Technology International) with 2 nm slits, and the results were visualized with FELIX.

Differential Scanning Fluorimetry

Individual wells of a 96-well plate were filled in triplicate with 20 μL of a solution containing 2 μM protein in 5 mM HEPES buffer (pH 7.3), 10 mM MgCl₂, 10 mM KCl, and 10 μM SYPRO orange dye. A ViiA 7 real-time polymerase chain reaction (PCR) instrument was employed with Protein Thermal Shift software (Life Technologies) to conduct thermal denaturation assays over a range of 15–99 °C. SYPRO dye was excited at 492 nm, and fluorescence emission was detected at 610 nm. The melting temperature (T_m) for each sample was determined by quantifying the temperature at which the first derivative of fluorescence emission reached a minimal value (Protein Thermal Shift software). However, for the sake of visual clarity, plots were made of the fluorescence emission versus temperature.

Results

Characterization of HscA Nucleotide-Binding Domain Variants

To ensure that HscA’s isolated NBD had folded properly and was able to bind to both ADP and ATP, we prepared [U-¹⁵N]HscA₃₈₆(T212V), which included a previously identified mutation (T212V) that renders HscA unable to cleave ATP but still responsive to ATP-induced conformational changes observed in WT HscA,¹² and collected 2D ¹H–¹⁵N TROSY-HSQC spectra in the absence of nucleotide (Figure 2A), with excess ADP (Figure 2B), and with excess ATP (Figure 2C). Vast dispersion of ¹H chemical shifts (Figure 2A) indicated that the NBD had folded properly, and differences in the positions of cross-peaks in spectra of ADP-bound (Figure 2B) and ATP-bound [U-¹⁵N]HscA₃₈₆(T212V) (Figure 2C) suggested that the NBD undergoes a nucleotide-induced structural rearrangement between two (or more) states, as reported in prior solution-state NMR spectroscopic^30,40−42 (but not X-ray crystallographic)^37,38 studies of DnaK. Furthermore, because the cross-peaks in the 2D ¹H–¹⁵N TROSY-HSQC spectrum of [U-¹⁵N]apo-HscA₃₈₆(T212V) in the absence of added nucleotide did not correspond to the positions of cross-peaks in spectra of the protein following the addition or ADP or ATP (Figure 2B,C), we concluded that our purified NBD likely exists in the apo state.

Figure 2.

NMR spectroscopic investigation of nucleotide binding in HscA₃₈₆ and HscA₃₈₆(T212V). (A) 2D ¹H–¹⁵N TROSY-HSQC spectrum of 0.3 mM [U-¹⁵N]apo-HscA₃₈₆ in HMKD buffer (red). (B) 2D ¹H–¹⁵N TROSY-HSQC spectrum of 0.3 mM [U-¹⁵N]HscA₃₈₆ in HMKD buffer with 10 mM ADP (green) overlaid upon the spectrum from panel A. (C) 2D ¹H–¹⁵N TROSY-HSQC spectrum of 0.3 mM [U-¹⁵N]HscA₃₈₆(T212V) in HMKD buffer with 10 mM ATP (blue) overlaid upon the spectrum from panel A.

As found with other Hsp70s,^38,43 binding of nucleotide to the NBD of HscA was dependent on the presence of Mg²⁺: the 2D ¹H–¹⁵N TROSY-HSQC spectra of apo and ATP-bound [U-¹⁵N]HscA₃₈₆(T212V) were nearly superimposable when they were acquired in TED buffer that lacked Mg²⁺ (Figure S1 of the Supporting Information). Thus, all subsequent structural and functional assays were conducted in HKMD buffer, which contained 10 mM MgCl₂.

HscA’s Interdomain Linker Autoactivates ATP Hydrolysis but Does Not Confer Sensitivity to HscB-Mediated Stimulation of ATPase Activity

Silberg et al.¹³ had previously analyzed the ATPase activity of HscA’s isolated NBD, but because the variant used in the study was truncated at residue 382, the potential ATPase-activating role of the interdomain linker was left unexplored. We characterized the ATPase activity of each HscA NBD variant (Figure 3A) by using a steady-state spectrophotometric assay that couples the hydrolysis of ATP to the enzymatic conversion of 2-amino-6-mercapto-7-methylpurine riboside (MESG) to ribose 1-phosphate and 2-amino-6-mercapto-7-methylpurine via purine nucleoside phosphorylase (PNP). By adding 1 mM ATP to a solution containing 1 μM HscA and following the change in absorbance at 360 nm, we could directly quantify the amount of ATP hydrolyzed per minute per mole of ATPase (ATPase activity, reported in units of min^–1) over time. As shown in panels A and B of Figure 3, the steady-state ATPase activity of HscA₃₉₅ (V_max³⁹⁵ = 0.443 min^–1) was 13-fold higher than that of HscA₃₈₆ (V_max³⁸⁶ = 0.0352 min^–1) and 17-fold higher than that of WT HscA (V_max^WT = 0.0254 min^–1), whereas HscA₃₈₉ (V_max³⁸⁹ = 0.230 min^–1), which contains only the highly conserved L³⁸⁷LL³⁸⁹ motif, exhibited 7- and 9-fold increases in its respective ATPase activity. Moreover, HscA₃₉₅ autoactivated ATP hydrolysis to rates that exceeded those of WT HscA stimulated by a 50-fold molar excess of HscB (V_max^WT+HscB = 0.288 min^–1) or IscU (V_max^WT+IscU = 0.257 min^–1) (Figure 3B), HscA’s cochaperone or substrate, respectively. As expected, the addition of 50 μM IscU had no effect on the ATPase activities of any of the NBD variants (Figure 3B), because they all lack the SBD that interacts with IscU. Collectively, these results are consistent with previous investigations of DnaK^30,31 and Hsc70³² and illustrate the conserved role of the Hsp70 interdomain linker in autoactivating ATP hydrolysis, even in this specialized Hsp70-type chaperone dedicated to Fe–S cluster transfer.

Figure 3.

(A) Steady-state kinetics of ATP hydrolysis in HscA NBD variants measured spectrophotometrically. Production of inorganic phosphate (P_i) via ATP hydrolysis was directly proportional to the absorbance at 360 nm. Each sample contained 1 μM protein in HKMD buffer at 25 °C. The interdomain linker (HscA₃₉₅, green diamonds) enhanced HscA’s basal ATPase activity 13-fold in comparison to that of the isolated NBD (HscA₃₈₆, blue circles). The highly conserved L³⁸⁷LL³⁸⁹ motif in the linker (HscA₃₈₉, red squares) was sufficient to autoactivate ATP hydrolysis to rates that were 7-fold faster than those of HscA₃₈₆. Mean ATPase activities are plotted here (n = 3) vs time. (B) ATPase activities of HscA₃₈₆, HscA₃₈₉, HscA₃₉₅, and WT HscA alone (blue) and in the presence of a 50-fold molar excess of HscB (red) or under IscU-stimulated conditions (green). The ATPase activities of all three NDB variants were not affected by added HscB or IscU. By contrast, the ATPase activity of WT HscA was enhanced in the presence of HscB or IscU. The presence of the SBD in WT HscA serves to decrease the activity of the NDB under all conditions studied. The Y-axis units are reported as moles of ATP hydrolyzed per mole of HscA per minute, or turnover number. Error bars are the standard deviation (n = 3).

Reports on the matter of ATPase stimulation of isolated Hsp70 NBDs in the presence of their Hsp40 cochaperones have differed.^31,32 An NMR investigation by Greene et al.⁴⁴ confirmed that DnaK’s isolated NBD was able to bind to DnaJ’s isolated J-domain in solution, but the interaction failed to stimulate its ATPase activity.³¹ On the other hand, the ATPase activity of Hsc70 NBDs was found to be stimulated by its J-protein (auxilin) if and only if the interdomain linker was present.³² Here, we investigated the effects imparted by HscB (the Hsp40 cochaperone) on the ATPase activities of WT HscA and the three HscA NBD variants. In agreement with previously published data,^45,46 the ATPase activity of 1 μM WT HscA increased by a factor of 8 in the presence of 50 μM HscB (Figure 3B). However, similar 50:1 HscB:HscA₃₈₆, HscB:HscA₃₈₉, or HscB:HscA₃₉₅ molar ratios did not elicit any stimulation of ATPase activity (Figure 3B).

HscA’s Interdomain Linker Binds to Its NBD in the cis Configuration, Altering the Fluorescence Emission and ¹H NMR Line Width of W291 and Enhancing the Inherent Thermostability of the NBD

Given the clear role of the linker in autoactivating ATP hydrolysis, we collected 2D ¹H–¹⁵N TROSY-HSQC spectra of 0.3 mM apo-[U-¹⁵N]HscA₃₈₆ and 0.3 mM apo-[U-¹⁵N]HscA₃₉₅ in HMKD buffer to investigate potential linker–NBD interactions in the apo state. In the spectrum of apo-[U-¹⁵N]HscA₃₉₅, additional cross-peaks arose from linker residues, as expected; however, approximately 20 cross-peaks either experienced severe line broadening and were rendered undetectable or were significantly displaced from their respective positions in the corresponding spectrum of apo-[U-¹⁵N]HscA₃₈₆ (Figure 4A). This finding suggests that the linker interacts with a distinct subset of NBD residues. Moreover, linker–NBD interactions in HscA₃₉₅ significantly altered the distribution of ¹H line widths at half-height (LWHH) calculated from ∼200 well-resolved, nonoverlapping peaks in the 2D ¹H–¹⁵N TROSY-HSQC spectra. The mean ¹H LWHH values were 27.2 ± 4.5 Hz for apo-[U-¹⁵N]HscA₃₈₆ and 21.6 ± 3.3 Hz for apo-[U-¹⁵N]HscA₃₉₅ (Figure 4B).

Figure 4.

NMR spectra of apo-[U-¹⁵N]HscA₃₈₆ and apo-[U-¹⁵N]HscA₃₉₅ indicate a binding interaction in the linker’s cis configuration. (A) Overlay of 2D ¹H–¹⁵N TROSY-HSQC spectra of 0.3 mM apo-[U-¹⁵N]HscA₃₈₆ (red) and 0.3 mM apo-[U-¹⁵N]HscA₃₉₅ (blue) acquired at 900 MHz in HMKD buffer. (B) Distribution of ¹H line widths at half peak height (LWHH) measured from 200 well-resolved peaks in the spectra presented in panel A. ¹H LWHH values from apo-[U-¹⁵N]HscA₃₈₆ are colored red, and ¹H LWHH values from apo-[U-¹⁵N]HscA₃₉₅ are outlined in blue. The mean ¹H LWHH values for apo-[U-¹⁵N]HscA₃₈₆ and apo-[U-¹⁵N]HscA₃₉₅ are 27.2 ± 4.5 and 21.6 ± 3.3 Hz, respectively, with errors representing one standard deviation. (C) Comparison of the ¹H LWHH of the cross-peak arising from the ε-¹H^N on the side chain of W291. Spectra in panel A were used for this analysis. Peak contours have been colored on a red to blue spectrum to indicate intensity.

To further investigate the nucleotide-independent linker–NBD interactions, we utilized two fluorescence assays: an intrinsic tryptophan fluorescence assay that has been used to characterize structural rearrangements in DnaK^47,48 and a differential scanning fluorimetry (DSF) thermal denaturation/fluorescence assay that reports on intrinsic protein stability.⁴⁹ We probed the tryptophan fluorescence emission profiles of HscA₃₈₆ and HscA₃₉₅ to examine the local chemical environment of W291, the lone tryptophan residue in HscA’s NBD (Figure 5A). HscA₃₉₅ displayed a red-shifted, enhanced fluorescence profile in comparison to that of HscA₃₈₆ (Figure 5A), suggesting that the apo-NBD–linker interaction causes W291 to become more exposed to solvent, in concordance with our NMR data presented above. In agreement with these fluorescence data, W291’s ε-¹H^N 1H LWHH values in apo-HscA₃₉₅ and apo-HscA₃₈₆ were 15.8 and 20.2 Hz (Figure 4C), respectively, which suggested that W291 becomes more dynamic or exposed to solvent in apo-HscA₃₉₅. In HscA₃₉₅, the normalized peak intensity of the W291 side chain ¹H^N was modestly increased (∼10%) over that of HscA₃₈₆ (data not shown), indicating enhanced motions on the nanosecond time scale.

Figure 5.

Direct comparison of HscA₃₈₆ and HscA₃₉₅ by intrinsic tryptophan fluorescence, DSF, and enzyme kinetics. (A) Intrinsic tryptophan fluorescence traces of HscA₃₈₆ and HscA₃₉₅. HscA’s lone tryptophan residue (W291) is located on a β-sheet near the bottom of domain IIB, which is known to rotate by ≥14° in DnaK but does not share much sequence homology with HscA in comparison to other NBD subdomains. Note the enhanced emission and red-shifted λ_max in HscA₃₉₅. (B) Differential scanning fluorimetry (DSF) thermal melting curves for HscA₃₈₆ and HscA₃₉₅. The melting temperature (T_m) of HscA₃₈₆ was 49.25 °C, which was significantly lower than that of HscA₃₉₅ (54.35 °C), suggesting that the linker confers thermostability by binding to the NBD. The melting temperatures (n = 3 for HscA₃₈₆, and n = 3 for HscA₃₉₅) of each replicate were identical. (C and D) ATPase activity plotted as a function of ATP concentration for (C) HscA₃₈₆ (◇) and (D) HscA₃₉₅ (●). Note that the values of K_m for HscA₃₈₆ (21.4 ± 4.0 μM) and HscA₃₉₅ (22.8 ± 3.2 μM) are very similar to the K_d^ATP (26 ± 7 μM) and K_m (12.6 μM) of WT HscA. Thus, the linker has no effect on apparent ATP binding affinity, but rather on catalytic activity (V_max³⁸⁶ = 0.039 ± 0.001 min^–1; V_max³⁹⁵ = 0.446 ± 0.016 min^–1). Results are shown with error bars for the standard deviation (n = 3).

Next, HscA₃₈₆ and HscA₃₉₅ were subjected to heat denaturation in the presence of a dye (SYPRO orange) that fluoresces (excitation at 492 nm and emission at 610 nm) upon binding to hydrophobic patches of unfolded proteins. Thus, the fluorescence signal intensity is directly proportional to the amount of unfolded protein in a given sample, and protein melting temperatures (T_m) are extracted from the lowest point in the first-derivative plot of fluorescence intensity.⁴⁹ Protein aggregation at higher temperatures often results in a decay of the fluorescence signal.⁴⁹ As seen in Figure 5B, HscA₃₉₅ melted at 54.35 °C whereas HscA₃₈₆ melted at 49.25 °C, indicating that the interactions between the NBD and interdomain linker confer enhanced basal thermostability to the NBD. These results are consistent with our Trp fluorescence data (Figure 5A), which suggested that the linker binds to the NBD, as well as with results from a thermal denaturation study via circular dichroism in which the T_m of DnaK₃₉₂ (DnaK NBD and the LLL of the linker) was 3.5 °C higher than that of DnaK₃₈₈ (the isolated DnaK NBD).³⁰

The Presence or Absence of the Interdomain Linker Does Not Affect the Apparent ATP Binding Affinity of the NBD

To examine whether linker–NBD interactions enhance the NBD’s binding affinity for ATP, we conducted ATPase assays as a function of ATP concentration while holding the HscA concentration constant (Figure 5C,D). From comparison of the Michaelis–Menten (K_m) constants determined for HscA₃₈₆ (K_m³⁸⁶ = 21.4 ± 4.0 μM) and HscA₃₉₅ (K_m³⁹⁵ = 22.8 ± 3.2 μM), it was readily apparent that the interdomain linker has no effect on the apparent ATP binding affinity. Moreover, both K_m³⁸⁶ and K_m³⁹⁵ are similar to binding affinity constants previously determined for WT HscA: K_d^ATP (26 ± 7 μM) and K_m (12.6 μM).¹⁹ This further indicates that the interdomain linker does not play a role in ATP binding but rather increases the catalytic rate of ATP hydrolysis (V_max³⁸⁶ = 0.039 ± 0.001 min^–1; V_max³⁹⁵ = 0.446 ± 0.016 min^–1).

The Synthetic HscA Linker Peptide (LLLDVIPLS) Binds to the Isolated NBD and Stimulates ATP Hydrolysis in the trans Configuration

2D ¹H–¹⁵N TROSY-HSQC spectra of 0.2 mM [U-¹⁵N]HscA₃₈₆ were collected in the absence and presence of a 15-fold molar excess of the linker peptide, which was partially dissolved in TED buffer to avoid complications with excess DMSO (see Materials and Methods and Figure S2 of the Supporting Information). The presence of the peptide led to widespread chemical shift perturbations, including distinct broadening of some signals (Figure 6A, bottom right inset), indicative of protein–ligand interactions (Figure 6A). A concentration dependence was observed in the fold stimulation of the ATPase activity of 1 μM HscA₃₈₆ upon addition of increasing amounts of linker peptide [5–500 μM (Figure S2 of the Supporting Information)] dissolved in HMKD buffer with 10% (v/v) DMSO. Maximal ATPase activity reached nearly 4-fold over basal levels (Figure 6B), after which the peptide required higher DMSO concentrations to fully dissolve. The enhanced concentration of DMSO in turn disrupted the basal ATPase activity of apo-HscA₃₈₆ (data not shown), and therefore, we were unable to pursue higher linker peptide concentrations to quantitatively determine the K_m value.

Figure 6.

(A) 2D ¹H–¹⁵N TROSY-HSQC spectra of [U-¹⁵N]HscA₃₈₆ at a final concentration of 0.2 mM in the absence (blue) and presence (orange) of a 15-fold molar excess of interdomain linker peptide (LLLDVIPLS) partially dissolved in TED buffer. Spectra were acquired at 900 MHz. (B) Steady-state ATPase assays conducted under conditions identical to those described in the legend of Figure 3 (except for the addition of linker peptide). Addition of 0.5 mM peptide increased the ATPase activity of the isolated NBD by a factor of nearly 4. See Figure S2 of the Supporting Information for a linker peptide concentration titration.

HscA₃₉₅ hydrolyzed ATP at a rate that was 3-fold faster than that of HscA₃₈₆ in the presence of 500 μM linker peptide (Figure 2B). This incomplete stimulation of ATPase activity likely results from relatively weaker binding interactions between the NBD and linker peptide in the trans configuration (HscA₃₈₆ and peptide) as compared to the cis (HscA₃₉₅) configuration. Indeed, a comparison of the 2D ¹H–¹⁵N TROSY-HSQC spectra of apo-[U-¹⁵N]HscA₃₉₅ in TED buffer and [U-¹⁵N]HscA₃₈₆ in the presence of 3 mM linker peptide, which corresponds to a 15-fold molar excess, indicates that unique cross-peaks appear in the spectrum of linker-bound HscA₃₈₆ (Figure S3 of the Supporting Information). Nevertheless, identical cis and trans-linker-induced chemical shift perturbations and a decrease in the ¹H LWHH of ε-¹H^N of W291 (20.4 Hz in apo-HscA₃₈₆ vs 16.7 Hz in linker peptide-bound HscA₃₈₆ under identical buffer conditions) are apparent (Figure S3 of the Supporting Information), suggesting that the observed binding between the NBD and synthetic linker peptide does not emerge from nonspecific interactions.

Discussion

Chaperone Action in the Crowded in Vivo Environment

Hsp70s function, among other roles, to ensure that nascent polypeptides fold properly into their correct conformations, that endocytosed clathrin-coated vesicles are unassembled, and that organelle-targeted proteins arrive at the proper translocase machinery.⁵¹⁻⁵³ Misfolded and unfolded proteins are recognized by the cochaperone Hsp40 and escorted to Hsp70’s SBD, which binds to solvent-exposed hydrophobic residues and allows abnormally folded proteins to refold into their native conformations.⁵² Owing to such diverse and indispensable roles, Hsp70 is among the most highly conserved proteins, and many eukaryotic and prokaryotic genomes encode multiple Hsp70s: for example, humans have 13 Hsp70s dispersed among the cellular compartments for various roles.⁵⁴

High-Resolution Structures of DnaK Bound to ADP and ATP Provide Insight into Hsp70 Function

High-resolution structures of full-length DnaK, the canonical Hsp70 chaperone of E. coli, bound to ADP³⁶ and ATP³⁵ (Figure S4 of the Supporting Information) have provided mechanistic insight into DnaK’s allosteric interdomain communication. Numerous studies have established that substrate binding stimulates ATPase activity in DnaK’s NBD and thereby creates a mechanism that efficiently utilizes energy derived from ATP hydrolysis to conduct polypeptide refolding or other proteostatic functions.⁶⁰ Crystal structures of isolated DnaK NBDs in various nucleotide states have failed to capture the delicate intersubdomain allostery that modulates substrate binding affinity,^37,38 whereas solution-state NMR spectroscopic studies have instead demonstrated that ATP or ADP binding significantly alters NBD subdomain orientation and flexibility.^30,41−43 In the NMR structure of ADP-DnaK, the NBD and SBD are joined by a flexible linker and appear to be relatively independent of one another; they each retain structures similar to those of their isolated domains.³⁶ Mutations in the linker have been shown to disrupt chaperone action via compromised interdomain communication and cochaperone recognition.^{31,32,55−57} DnaK’s interdomain linker is therefore critical to interdomain allostery and serves as a connection between ATP binding in the NBD and substrate binding affinity in the SBD.^{32,33,35,39,58,59}

Role of the Interdomain Linker in HscA Compared to That in DnaK and Hsc70

The results reported here allow us to compare the role of the interdomain linker in HscA, an Hsp70 specialized for Fe–S cluster biosynthesis in E. coli, with prior results from other Hsp70s (i.e., E. coli DnaK and bovine Hsc70 purified from E. coli). Evaluation of the steady-state ATPase activities of the HscA NBD variants shows that the linker autoactivates ATP hydrolysis in the NBD (Figure 3A), with basal rates that exceed the rate of HscB- or IscU-mediated stimulation of WT HscA (Figure 3B). These findings are consistent with previously published studies of DnaK and Hsc70³⁰⁻³² that illustrated the conserved role of the interdomain linker in autostimulating ATPase activity. Additionally, the apparent ATP binding affinities of HscA₃₉₅ and HscA₃₈₆ were identical within experimental error (Figure 5C,D), indicating that the linker is primarily involved in catalyzing the cleavage of ATP. Moreover, the average ¹H LWHH in apo-HscA₃₉₅ (21.6 ± 3.3 Hz) was significantly more narrow than that in apo-HscA₃₈₆ (27.2 ± 4.5 Hz) (Figure 4B); as line broadening can result from, among other contributions, chemical exchange that is intermediate on the NMR time scale, it is possible that apo-HscA₃₈₆ adopts a variety of conformations in solution, whereas linker–NBD interactions in apo-HscA₃₉₅ select for one conformer or a subset of conformers. Elegant NMR studies of the NBD of DnaK have demonstrated that this ATPase domain transmits its allosteric signals via conformational selection, with apo- and ADP-DnaK NBD adopting a variety of conformations in solution. Spin relaxation experiments in combination with resonance assignments of HscA will be required to investigate this possibility in more detail.

In HscA₃₉₅, the interdomain linker enhances fluorescence emission from W291 (Figure 5A) and increases basal thermostability in comparison to that of HscA₃₈₆ (Figure 5B), suggesting that the linker interacts with the NBD in the absence of a nucleotide. Furthermore, the ¹H line width of the side chain ¹H^N in W291 has become narrower in HscA₃₉₅, which indicates enhanced mobility. This finding is not entirely inconsistent with prior data from investigations of DnaK, wherein Gierasch and co-workers reported that the truncated linker (DnaK₃₉₂) was highly flexible and dynamic in the apo state;³⁰ however, small perturbations to peak intensities and backbone chemical shifts were observed in nucleotide-free DnaK₃₉₂.^30,33 Moreover, linker–NBD interactions in a two-domain DnaK construct, while primarily nucleotide-dependent, were similarly found to occur in the apoprotein.⁵⁹ However, fluorescence results with HscA presented herein cannot be compared directly to similar measurements on DnaK, because the tryptophan of DnaK (W102) studied by fluorescence is 43 Å from the residue corresponding to W291 (M296) in ADP-DnaK (Figure S5 of the Supporting Information). M296 resides on a β-strand near the bottom of subdomain IIB that is known to rotate by 14° during DnaK’s ATPase cycle,⁶¹ perhaps indicating that this region in HscA is sensitive to linker–NBD interactions.

Our ATPase assays illustrate that the SBD of HscA represses basal ATPase activity by restricting the linker’s access to the NBD. This internal inhibition likely functions to couple ATP hydrolysis in the NBD to the binding of HscB and [2Fe-2S]-IscU, interactions that presumably remodel HscA’s tertiary structure to allow the linker to dock onto the NBD and activate ATP hydrolysis. Such a mechanism would ensure that HscA efficiently utilizes energy derived from ATP hydrolysis to specifically transfer [2Fe-2S] clusters. Silberg et al.¹⁹ previously characterized ATP binding in HscA through an absorbance assay, demonstrating that ATP binding occurs in two steps: a fast ATP binding event and a slow ATP-induced conformational change. Because ATP binding is also known to destabilize HscA–peptide complexes,^13,16 the binding of ATP and consequent activation of ATP hydrolysis exquisitely regulate the formation and dissociation of [2Fe-2S]-IscU–HscB–HscA ternary complexes and resultant [2Fe-2S] cluster transfer.

Through extensive structural and biochemical studies, DnaK’s linker was determined to play a key role in interdomain allostery.^33,55,59,62 Across all Hsp70s, the interdomain linker displays nearly ubiquitous conservation of its internal hydrophobic sequence (LLLDV-PL) with some variation of residue identity at the N- and C-termini. In the NMR structure of ADP- and peptide-bound DnaK, the interdomain linker (V³⁸⁹LLLDVTPLS³⁹⁸, DnaK numbering) was found to be highly dynamic and was not observed in the NMR spectra,³⁶ yet in the ATP-bound crystal structure, the linker had docked onto the NBD and adopted a β-sheet conformation.³⁵ Prior NMR analyses of DnaK₃₉₂ had identified docking of the truncated linker (V³⁸⁹LLL³⁹²),^30,33 which was sensitive to nucleotide binding and consistent with other structural analyses of the full-length protein.³⁹ Moreover, this indicated that the presence of the SBD restricted the ability of the linker to dock onto the NBD (i.e., the NBD and SBD were essentially isolated domains in ADP-DnaK, but coupled in ATP-DnaK) and activate ATP hydrolysis. We observed a similar SBD-mediated repression of ATP hydrolysis in our investigation of HscA NBD variants (Figure 3B). We speculate that stimulation of HscA’s ATPase activity by HscB and IscU results from enhanced linker–NBD interactions that are not present or occur with diminished probabilities in the unstimulated full-length protein.

By examining the ATPase activity of HscA NBD variants in the presence of excess HscB, we determined that neither the isolated NBD (HscA₃₈₆) nor the NBD with a full interdomain liker (HscA₃₉₅) was receptive to ATPase stimulation via HscB (Figure 3B). A recent NMR study from our lab demonstrated that the interaction between HscB and ATP-bound HscA NBD lacked vital contacts with HscB’s J-domain.⁶⁵ Full-length HscA was found to be sensitive to HscB-mediated activation of ATPase activity through ATP-dependent interactions with the J-domain of HscB.⁶⁵ From these results, it was unclear whether the linker plays a role in HscB-mediated activation of ATP hydrolysis. Here, we have shown that HscA’s linker alone is not sufficient; thus, the presence of the SBD is required to elicit a functional activation from HscB. However, the mechanism by which the SBD facilitates activation of ATPase stimulation by HscB remains unclear: does the SBD in ATP-HscA optimally position the NBD and linker for interaction with HscB, or does the SBD itself contact HscB and facilitate subsequent activation of ATP hydrolysis? Elucidation of the HscB-binding interface on ATP-HscA will help answer this question.

Our results are consistent with a prior investigation of DnaK–DnaJ interactions, which similarly showed that DnaK’s SBD is required for ATPase stimulation by DnaJ.⁵⁶ However, Jiang et al.²³ demonstrated that the ATPase activity of an Hsc70 NBD variant with a truncated linker could be stimulated by its J-protein, auxilin. Similar results were not seen with the Hsc70 isolated NBD, suggesting that linker–auxilin interactions were primarily responsible for ATPase activation by auxilin.³² Differences between these studies may arise from mechanistic distinctions between auxilin and DnaJ or HscB, or from functional disparities between Hsc70 and DnaK or HscA. Indeed, the NBD and SBD in Hsc70 are known to contact one another in the apo and ADP states,^32,66 whereas DnaK’s NBD and SBD remain relatively isolated and free from contact under these conditions.^30,36 Determining how the two domains of HscA interact under various nucleotide conditions will shed light on this apparent mechanistic distinction among Hsp70s.

Using solution-state NMR spectroscopy, we found that a peptide corresponding to HscA’s interdomain linker (L³⁸⁷LLVIDPLS³⁹⁵) interacts with HscA₃₈₆ in the absence of nucleotide (Figure 6A), and our ATPase assays demonstrated that this synthetic linker peptide stimulates the ATPase activity of the isolated NBD nearly 4-fold over basal levels (Figure 6B). Narrowing of the W291 ε-¹H^N line width in HscA₃₈₆ in the presence of the linker peptide (16.7 Hz; 20.4 Hz in apo-HscA₃₈₆ under identical buffer conditions) further confirms the specificity and WT-like binding interactions (Figure S3 of the Supporting Information) in the trans configuration.

As compared to the 2D ¹H–¹⁵N TROSY-HSQC spectrum of [U-¹⁵N]HscA₃₉₅, both identical and unique chemical shift perturbations appear in the 2D ¹H–¹⁵N TROSY-HSQC spectrum of [U-¹⁵N]HscA₃₈₆ in the presence of the linker peptide (Figure S3 of the Supporting Information). On one hand, this indicates that the linker peptide, while interacting with the NBD in a nativelike manner, could also impart non-native NBD conformations. On the other hand, these unique trans linker configuration signals could also arise from nativelike interactions that are unsaturated by the weakly binding linker peptide. Without complete backbone resonance assignments, however, interpretation of these spectra remains challenging, as no residue-specific information (other than the side chain of W291) can be obtained regarding a comparison of the linker-binding site in cis and trans configurations.

Nonetheless, this is the first indication that an exogenously added Hsp70 linker peptide can bind to its isolated NBD and stimulate ATP hydrolysis in the trans configuration (Figure 6B), presenting the unique ability to isotopically label individual residues within the linker for NMR or IR spectroscopic structural and functional studies. This system could prove useful in dissecting the structural and dynamical roles of specific interdomain linker residues upon binding to ATP-bound NBD versus apo-NBD. Prior NBD–linker investigations relied on selective labeling of individual amino acid types (e.g., ¹⁵N-labeled leucine residues in ref (30)) or distinct chemical moieties (e.g., protonated, ¹³C-labeled methyl groups in an otherwise deuterated background in ref (59)). Here, we offer the ability to further probe the roles of interdomain linker residues via solid-state peptide synthesis with any combination of isotopic labels.

Glossary

Abbreviations

2D

two-dimensional

AU

arbitrary units

DnaK

stress-inducible heat shock protein 70 kDa (Hsp70) in E. coli

DSF

differential scanning fluorimetry

DTT

dithiothreitol

HKMD

50 mM HEPES (pH 7.3), 150 mM KCl, 10 mM MgCl₂, and 1 mM DTT

HscA₃₈₆

isolated nucleotide-binding domain (NBD) of HscA

HscA₃₈₉

NBD of HscA containing a short (three-amino acid) linker peptide

HscA₃₉₅

NBD of HscA containing the full-length (nine-amino acid) linker peptide

Hsc70

constitutively expressed heat shock cognate 70 kDa (B. taurus) purified from E. coli

HSQC

heteronuclear single-quantum coherence

NBD

nucleotide-binding domain

PDB

Protein Data Bank

P_i

inorganic phosphate

ROS

reactive oxygen species

SBD

substrate-binding domain

SDS–PAGE

sodium dodecyl sulfate–polyacrylamide gel electrophoresis

TED

50 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, and 1 mM DTT

TROSY

transverse relaxation-optimized spectroscopy

[U-]

uniformly labeled.

Supporting Information Available

Structures of ATP- and ADP-bound DnaK, a comparison of the kinetic parameters in the ATPase cycles of DnaK and HscA, NMR spectra that highlight the essential role of Mg²⁺ in facilitating binding of ATP to HscA’s NBD, NMR spectra that compare the effects of linker binding in cis and trans configurations, and HscA NBD ATPase assays as a function of added linker peptide concentration. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Present Address

^∥ T.R.A.: Physical and Theoretical Chemistry Laboratory, Department of Chemistry, University of Oxford, Oxford, U.K.

Author Present Address

^⊥ J.H.K.: Max Planck Institute for Biophysical Chemistry, Göttingen, Germany.

This work was supported by National Institutes of Health (NIH) Grant U01 GM94622 and made use of the National Magnetic Resonance Facility at Madison, which is funded by NIH Grant P41GM66326 (National Institute of General Medical Sciences). Equipment was purchased with funds from the University of Wisconsin—Madison, the NIH (P41RR02301, P41GM66326, S10RR02781, S10RR08438, S10RR023438, S10RR025062, and S10RR029220), the National Science Foundation (DMB-8415048, OIA-9977486, and BIR-9214394), and the U.S. Department of Agriculture.

The authors declare no competing financial interest.

Funding Statement

National Institutes of Health, United States