Hsp70 and Hsp40 bind to the FQNLF motif in cells

To validate in cells the observations made in vitro HEK293T cells were transfected with full length AR tagged with the FLAG epitope (FLAG-AR) and its interaction with endogenous Hsp40 and Hsp70 was assessed by proximity ligation assay (PLA). This immunofluorescence-based technique is particularly useful for detecting transient PPIs in cells where the putative complexes are of intermediate or low stability and therefore unlikely to withstand the washing steps that are required for co-immunoprecipitation experiments⁴⁵. Briefly, the method relies on the use of primary antibodies that recognize the two proteins (e.g. AR and Hsp70), and on secondary antibodies that are conjugated to complementary oligonucleotides. When the proteins are in close proximity, these oligonucleotides hybridize and are ligated to form a circular DNA template that can be amplified by DNA polymerase and detected by fluorescence probes. The PLA signals can therefore be detected by fluorescence microscopy as red puncta in the PLA experiment.

Using this approach, we observed a clear interaction between FLAG-AR and both Hsp40 and Hsp70 in the cytoplasm prior to activation but found that DHT treatment induced an almost complete loss of the interaction (Fig. 3a, c) combined with nuclear translocation (Supplementary Fig. ⁴). These results agree with the notion that activation of AR by DHT causes the dissociation of the cytosolic complexes that AR forms with molecular chaperones prior to nuclear translocation^3,8. To test whether the interaction detected by PLA occurs in cells via the region of sequence containing the FQNLF motif detected by NMR, we repeated these experiments using a version of FLAG-AR with the FQNLF motif mutated to AQNAA (FLAG-AR-mut) and a deletion mutant, where the region of sequence spanning residues 21 to 35, which includes the FQNLF as well as adjacent similar VREVI motif, was deleted (FLAG-AR-del). We observed that while FLAG-AR-mut could still interact with both Hsp40 and Hsp70 (Supplementary Fig. ⁵), the interaction of FLAG-AR-del with both molecular chaperones was substantially weaker (Fig. 3a, c). These results suggest that mutation of the three residues at the core of the FQNLF motif is not sufficient to abolish the affinity for Hsp70 or Hsp40, that instead requires the complete deletion of this motif and the ca. 10 residues immediately flanking it. This result is consistent with the 10-fold weakened affinity of the AQNAA mutant for Hsp70 observed in the NMR and FP experiments (see Supplementary Fig. ³). It should be noted that our PLA experiments were designed to detect the interaction of molecular chaperones with the NTD of AR and are therefore unlikely to report on whether they interact with the LBD. In addition, the fact that both FLAG-AR-mut and FLAG-AR-del readily translocate to the nucleus upon activation suggests that these variants retain some early AR functions (Supplementary Fig. ⁴). Collectively, these results confirm that, in cells, prior to activation by androgens, the molecular chaperones Hsp40 and Hsp70 bind to AR via a region of sequence including the motif FQNLF and the residues flanking it.

Open in a separate window

Fig. 3

Removal of the FQNLF motif and flanking residues prevents the interaction of cytosolic AR with Hsp40 and Hsp70 in cells. a The interaction between FLAG-AR as well as FLAG-AR-del with endogenous Hsp70 in the absence or presence of 1nM DHT was studied in transfected HEK293T cells by PLA (n=3). EV stands for empty vector, FLAG-AR/Hsp70 complexes appear as red dots and DAPI (blue) indicates the localization of nuclei. b Analysis by Western blotting of the levels of FLAG-AR and FLAG-AR-del in transfected HEK293T cells in the absence or presence of 1nM DHT (n=3). β-actin immunoblot is shown as a loading control. Log2 transformed values were used to assess differences using two-tailed unpaired t-tests and ***p<0.001. c Assay equivalent to that presented in a to study the interaction with the endogenous Hsp40 (n=3). The scale bars in a, c represent 10μm. Source data for the comparison of protein levels shown in b are available as Source Data file

Hsp70 and Hsp40 have both been implicated in turnover of AR. To probe the role of their interaction with the FQNLF motif in this process, we compared the levels of FLAG-AR-mut and FLAG-AR-del with those of the wild type protein prior and following to activation with DHT. We observed that mutation and, especially, deletion of the FQNLF motif caused an increase in basal AR levels in the absence of hormone (Fig. 3b, Supplementary Fig ^5B). Also, recent studies have shown that disrupting AR’s N/C interaction delays disease onset in a mouse model of SBMA⁴⁶. These studies utilized mice that expressed a mutant form of the polyQ AR where the Phe residue of the FQNLF motif was mutated to Ala. To determine whether this mutation had an effect on the interaction between the AR NTD and molecular chaperones, we carried out experiments with the equivalent mutant of NTD_1-155, named AQNLF-NTD_1-155. An analysis of the interaction by NMR showed that the AQNLF mutant interacts in vitro with molecular chaperones with an affinity that was indistinguishable from that of the WT (Supplementary Fig. ⁶). Thus, our results confirm that amelioration of the SBMA phenotype in mice expressing this mutant is likely due to a weakening of the N/C interaction rather than an altered interaction with molecular chaperones.

Hsp70 increases the solubility of the AR NTD

Because Hsp70 and Hsp40 are released from the NTD_1-155 after addition of the androgen dihydrotestosterone (DHT), the hormone-dependent phenotype of SBMA and ligand-dependent aggregation of the polyQ AR strongly suggest a protective role for these molecular chaperones in limiting AR aggregation^47–50. To investigate this possibility by solution NMR, we analyzed the oligomerization properties of the NTD_1-155 construct by measuring ¹H,¹⁵N-HSQC spectra at two different concentrations, 125 and 450μM at 278K. NMR can be used to monitor concentration-dependent oligomerization because the monomer-to-oligomer exchange process contributes to NMR linewidths, such that the R₂ ¹⁵N relaxation rates are increased⁵¹. In NTD_1-155 we first confirmed that the position of the backbone amide chemical shifts did not change with concentration. Then, we measured the R₂ ¹⁵N relaxation rates and found that some residues had higher values at 450μM, when compared to 125μM (Fig. 4a). A closer examination of these linewidth changes showed that the affected residues were clustered into regions of locally high AABUF values, suggesting that hydrophobicity could be the main driving force of oligomerization. Indeed, the most affected region was localized to the first 50 residues of the construct, with a maximum in the FQNLF motif. This observation is conceptually equivalent to that made by Wetzel and co-workers in huntingtin, in which it was shown that the early stages of aggregation are not driven by inter-molecular interactions involving the polyQ tract, but rather by hydrophobic, coiled coil-like interactions in the sequence flanking it at the N-terminus^52,53. Further, these results raise the intriguing possibility that binding of molecular chaperones to this region of AR might directly suppress its oligomerization.

Open in a separate window

Fig. 4

Hsp70 prevents aggregation of the AR NTD. a Plot of the difference in ¹⁵N R₂ relaxation rates of the NMR resonances of the NTD_1-155 construct at 125 and 450μM and of the AABUF⁴² and disorder propensity, as predicted by PONDR⁶⁹, of the NTD_1-155, indicating the presence of NTD_1-155 oligomers stabilized by hydrophobic interactions. b Decrease in the average intensity of the resonances of the NTD_1-155 construct as a function of incubation time at 298K in the absence and in the presence of 1 molar equivalent of Hsp70, indicating that this molecular chaperone abolishes the aggregation of NTD_1-155 at 125μM at 278K and in the absence of nucleotides. Data are mean±S.D. from the values obtained for all residues whose intensities are not affected by Hsp70 binding. Source data for panel b are available as Source Data file

We then carried out NMR experiments to directly study whether the interaction of Hsp70 with the NTD_1-155 construct influenced its solubility. For these experiments, we prepared samples of NTD_1-155 at 125μM and added human Hsp70 at 1 molar equivalent, using the same conditions presented in Fig. 1, and measured the corresponding ¹H,¹⁵N-HSQC spectra as a function of incubation time at 298K. In the absence of chaperone, we observed a progressive decay of signal intensity, so that after 6 days of incubation, ~25 % of the signal intensity was lost. This signal decay was likely caused by formation of aggregates whose size precluded their detection by NMR, as evident by the turbidity of the samples over time. By contrast, the samples containing Hsp70 did not show any decrease of signal intensity, suggesting that the complex remained soluble (Fig. 4b). These results suggest that molecular chaperones limit self-interactions between FQNLF motifs in different AR NTD molecules, diminishing its oligomerization and maintaining solubility.

NMR Sample preparation

The protein solutions stored at −80°C were thawed and dialyzed for 16h at 4°C against a phosphate buffer (sodium phosphate (20mM), tris(2-carboxyethyl)phosphine (TCEP, 1mM), pH 7.4). Finally, 10% D₂O and 0.015mM DSS were added to the samples.

Acquisition and analysis of NMR spectra

Two-dimensional ¹H,¹⁵N-HSQC NMR experiments were recorded at 5°C on a Bruker 600 or 800MHz spectrometer equipped with cryogenic probes. Spectra were processed with NMRPipe⁶⁶ and analyzed using CcpNMR Analysis⁶⁷. The NMR samples used for the experiments reported in Fig. 1, Supplementary Figs. ¹ and ⁵ contained 33µM ¹⁵N single-labelled NTD_1-155 protein and either 16.5µM or 33μM unlabeled Hsp70 and Hsp40 in 20mM phosphate buffer, pH 7.4, 1mM TCEP, and 10% (v/v) D₂O. The NMR samples used for the experiments shown in Supplementary Fig. ¹ were equivalent to those used for the experiments shown in Fig. 1, but contained NTD_144-450 instead of NTD_1-155.

The NMR samples used for the experiments reported in Fig. 4a contained 125 and 450µM ¹⁵N single-labelled NTD_1-155 constructs and those used for the experiments reported in Fig. 4b contained 125µM ¹⁵N single-labelled NTD_1-155 constructs and 125μM unlabeled Hsp70. The ¹⁵N R₂ rates of NTD_1-155 were measured with 10 different relaxation delays (t) and fit to the equation $I=I_o\exp \left(-R_{2}t\right)$ and only the values of R₂ determined with less than 5% error were considered sufficiently reliable for the analysis shown in Fig. 4a. The ratio of signal intensities shown in Fig. 4b corresponds to the total signal intensity of the signals that are not affected by chaperone binding (Fig. 1) measured at a given time divided by that measured after 1 day of equilibration.

Fluorescence polarization assays

For competition FP studies, 2μM Hsp70_SBD and 25nM FAM-LVEALY (Anaspec) were incubated with varying concentrations of AR protein for 30min at room temperature in assay buffer (100mM Tris, 20mM KCl, 6mM MgCl₂ pH 7.4). After incubation, fluorescence polarization was measured (excitation: 485nm emission: 535nm) using a SpectraMax M5 plate reader. In the case of the FAM-NRLLLTG tracers, 12μM Hsp70_SDB, and 25nM tracer were used and experiment was performed in a similar manner. These experiments were repeated in duplicate and repeated on three independent days.

Isothermal Titration Calorimetry

The NTD_1-155 proteins, as well as Hsp70_SBD or Hsp40, were dialyzed overnight against ITC buffer (25mM HEPES, 5mM MgCl₂, 10mM KCl pH 7.5). Concentrations were determined using a BCA Assay (Thermo Fisher Scientific), and the experiment was performed with a MicroCal ITC₂₀₀ (GE Healthcare) at 25°C. AR protein (100–200μM) in the syringe was titrated into a 10–20μM cell solution of Hsp70 (SBD) protein. Calorimetric parameters were calculated using Origin 7.0 software and fit with a one or two-site binding model. Experiments were performed in duplicate and repeated twice on two independent days.

Cell culture and transfection

Human embryonic kidney 293 T cells (HEK293T cells) were obtained from ATCC (CRL-3216) and maintained in DMEM containing 4.5gL⁻¹ glucose (Glutamax, Gibco) supplemented with 10% (v/v) charcoal stripped calf serum (Gibco) and antibiotics. Cells were cultured in a humidified atmosphere containing 5% CO₂ at 37°C. Transient transfection of HEK293T cells was performed with polyethylenimine (PEI) (Polysciences) at a ratio of 1µg DNA to 3µl PEI to express androgen receptor tagged with the FLAG epitope (FLAG-AR), the mutated version (FLAG-mut) and the version lacking the region of sequence targeted by Hsp40 and Hsp70 in the NMR experiments (FLAG-del).

PC12 cells expressing tetracycline-inducible AR112Q (gift of Diane Merry, Thomas Jefferson University) were characterized previously⁵⁷. Cells were maintained in phenol red-free DMEM (21063-029, Invitrogen) supplemented with 5% charcoal stripped fetal calf serum (Thermo Scientific Hyclone Products), 10% charcoal stripped horse serum (Invitrogen), G418 (Gibco) and Hygromycin B (Invitrogen). When indicated, cells were transfected using Lipofectamine 2000 (Life Technologies) or the Neon transfection system (Thermo Scientific) to express human CHIP and HA-His-ubiquitin (gifts from Dr. Yoichi Osawa, Univ. of Michigan). AR expression was induced with 0.5µgmL⁻¹ doxycycline (Clonetech) and activated by treatment with the synthetic androgen R1881. Lactacystin was from Sigma-Aldrich (L6785) and 17-AAG was from LC Laboratories (A-6880).

Antibodies and reagents

For Western Blot analyses the following antibodies were used: anti-β-actin-HRP (ab8224, 1:10000) (Abcam), rabbit-anti-FLAG (F7425, 1:2000), and mouse-anti-FLAG M2 (F1804, 1:2000) (Sigma), mouse-anti-HSP70/72 (ADI-SPA-810) and rabbit-anti-HSP40 (ADI-SPA-400-D) (Enzo Life Science). Additional primary antibodies were used against Akt (C6E7; Cell Signaling Technology), ERK1/2 (9102, Signaling Technology), GAPDH (NB-600-502; Novus Biologicals), HSP90 (H114, sc7947; Santa Cruz Biotechnology), CHIP (PA5-32046, Thermo Scientific, Pierce), HA (MMS-101R, Covance, Biolegend), HSP25 (ADI-SPA-801-488; Enzo Life Science), and HSP70 (ADI-SPA-812; Enzo Life Science). HRP conjugated secondary antibodies were from Bio-Rad (170-6515, 170-6516). For immunofluorescence the following primary antibodies were used: rabbit-anti-FLAG (F7425, 1:200), mouse-anti-FLAG M2 (F1804, 1:200) (Sigma), AlexaFluor conjugated secondary antibodies were from Life Technologies (A11072, 1:300). Mounting medium with DAPI (H-1200, 1:500) was from Vector Labs. For immunoprecipitation, pull down was performed with protein A–agarose beads (Santa Cruz Biotechnology) and the anti-AR antibody (PG21; Millipore).

Western blotting

Cells were washed and harvested in PBS, lysed in RIPA buffer (Technova) containing phosphatase and protease inhibitors (Roche). Lysates were centrifuged at 15000g to separate soluble and pellet fractions. Total protein was quantified using BCA assay (Pierce Biotechnology). Proteins were resolved by 7.5 or 15% SDS-PAGE, transferred to PVDF or nitrocellulose membranes, and probed with specific antibodies.

Filter retardation assay was performed by sample filtration through a 0.2 μm cellulose acetate membrane (Whatman, GE Healthcare) using a slot-blot apparatus.

For ubiquitination studies, cells were homogenized in buffer (0.025M Tris, 0.15M NaCl, 0.001M EDTA, 1% NP-40, 5% glycerol, pH 7.4) containing protease inhibitors and 5 mM N-ethylmaleimide (Sigma-Aldrich). Lysates were incubated on a rotator at 4°C for 2h with AR antibody or non-immune rabbit IgG (Santa Cruz Biotechnology). Prewashed protein A–agarose beads were added, and samples were incubated for 1 additional hour at 4°C. Protein-antibody-bead complexes were washed 6 times in buffer, and proteins were eluted by boiling in SDS-loading buffer for 5min at 100°C. Samples were then resolved on by 7.5% SDS-PAGE.

Images of uncropped and unprocessed scans of blots are provided as Source Data files.

Agarose gel electrophoresis for resolving aggregates was performed as previously described⁶⁸. Briefly 150µg total protein was loaded to 1.5% agarose gels (375mM Tris buffer, pH 8.6; 0.1% SDS) and electrophoresed at 100V until the dye front had reached the bottom of the gel. Transfer of proteins was performed using a semi-dry transfer apparatus to PVDF membranes.

Immunofluorescence

For immunofluorescence, cells were seeded on sterilized 12mm microscope coverslips, fixed 15min in 4% paraformaldehyde at room temperature and permeabilized in ice-cold methanol for 10min. Cells were then washed, blocked with PBS containing 2% bovine serum albumin for one hour at room temperature and incubated with the corresponding primary antibody diluted in blocking solution for one hour. After washing with PBS, samples were further incubated with the corresponding secondary antibodies for one hour. DAPI was used to label DNA and the coverslips were washed, air-dried and mounted on slides using Prolong Gold antifade mountant (Life technologies). 6 μm cryosections of muscle were fixed in methanol. Slides were stained, mounted, and imaged by confocal microscopy (Olympus FV 500 or Nikon A1). Nuclear signal intensity was quantified by CellProfiler.

Proximity ligation assay

The Duolink in situ Proximity ligation (PLA) kit (Olink Bioscience) was used for  the detection of protein–protein interactions according to the manufacturer’s instructions⁴⁵. The assay is based on the simultaneous binding of two antibodies to the proteins of interest and makes use of a proximity ligation technique combined with an amplification reaction to detect the antibodies whenever they are less than 30nm apart. HEK293T cells were grown and transfected as indicated on 12mm microscope coverslips (Thermo). Cells were fixed in PBS containing 4% paraformaldehyde for 15min and permeabilized with methanol for 5min. Fixed cells were blocked in PBS containing 2% BSA (Sigma) for an hour and then incubated with the indicated antibodies against Hsp70 (ADI-SPA-810, 1:100), Hsp40 (ADI-SPA-400-D, 1:100) (Enzo Life Science) and FLAG (M2, F1804 and F7425, both 1:50) (Sigma). Samples were subsequently incubated for 1h at 37°C with secondary antibody PLA with and without probes for mouse and rabbit (Sigma), and then for 30min at 37°C with a ligation solution to ligate the two PLA probes and form a circular DNA template. A rolling-circle amplification was then performed by incubating the cells for 100min at 37°C with the amplification solution. The amplicons generated are recognized by fluorescent oligonucleotides present in the amplification solution. Nuclei were stained with DAPI and the coverslips were washed, air-dried and mounted on slides using Prolong antifade mountant (Life Technologies). Cells were analyzed with a ×63 objective on a Leica SP5 confocal microscope. Images were presented as maximal projection of z-stacks by using Fiji.

AR113Q mice

AR113Q knock-in mice were generated using exon 1-specific gene targeting^50,59, backcrossed to C56BL/6 (≥10 generations), and housed in a specific pathogen-free facility on a 12-h light/dark cycle with chow and water ad libitum. For analysis, AR113Q males were used at 3 months of age and randomly assigned to treatment groups. JG-294 was dissolved in 70% propylene glycol (W294004, Sigma-Aldrich) and 30% PBS (pH 7.4, Life Technologies), heated to 95°C to dissolve compound, aliquoted and frozen. Mice received intraperitoneal injections (3mgkg⁻¹) every other day for 2 weeks. 17-DMAG (S1142, Selleckchem) was dissolved in 1% DMSO (D2650, Sigma-Aldrich), 30% polyethylene glycol (202371; Sigma-Aldrich) and 1% Tween80 (P4780; Sigma-Aldrich), and administered by i.p. injection (10mgkg⁻¹) every other day for 2 weeks. At the end of treatment, mice were euthanized and quadriceps were harvested and flash frozen in liquid nitrogen or mounted on OCT (Tissue-Plus, Fisher Health Care) for cryosectioning. For western blot, lysates were obtained by homogenizing tissue in RIPA containing phosphatase and protease inhibitors. All procedures involving mice were approved by the University of Michigan Committee on Use and Care of Animals (PRO00008133) and conducted in accordance with institutional and federal ethical guidelines for animal testing and research.

Quantitative RT-PCR

Total RNA was extracted from mouse quadriceps muscles using Trizol (Sigma-Aldrich). cDNA was synthesized from extracted RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Bio Systems). Quantitative PCR was carried out on a 7500 Real-Time PCR System (Applied Biosystems) using FastStart TaqMan Probe Master Mix (Roche) and gene specific primers for AR (universal probe library #58, Roche; 5′ primer: ccagtcccaattgtgtcaaa; 3′ primer: tccctggtactgtccaaacg) and Cpsf2 (Mm00489754_m1, Applied Biosystems). Relative AR expression was determined by the standard curve method and normalized to Cpsf2.

Statistics

Statistical significance was determined by analyzing data sets using unpaired Student’s t-test, one-way ANOVA with Tukey’s multiple comparison test or two-way ANOVA with Bonferroni or Niewman–Keuls multiple comparison test in Prism6 (GraphPad). Differences between mean values were defined as significant at p<0.05.

X.S. acknowledges funding from Obra Social “la Caixa”, AGAUR (2017 SGR 324), Marató TV3 (102030), MINECO (BIO2012-31043 and BIO2015-70092-R) and the European Research Council (CONCERT, contract number 648201). IRB Barcelona is the recipient of a Severo Ochoa Award of Excellence from MINECO (Government of Spain). A.P.L. acknowledges funding from NIH (NS055746). J.E.G. acknowledges funding from NIH (NS059690). S.R.N. was supported by a Rackham Predoctoral Fellowship and by the NIH (GM007863).