Sequence analysis
Protein sequence alignment was performed using MUSCLE 3.8 (https://www.ebi.ac.uk/Tools/msa/muscle/) and adjusted manually.
Peptides
Untagged and biotinylated synthetic Paramecium tetraurelia histone H3 peptides (Fig. 6c) were synthesized by Proteogenix (Schiltigheim, France) and used without further purification.
Expression and purification of GST-tagged CRDs
Synthetic DNA fragments (Integrated DNA Technologies) encoding Pgm(692–768) and its C712S and H701S + C712S mutant derivatives, PB(538–594), PgmL1(463–539), PgmL2(540–614), PgmL3a(471–550), PgmL4a(856–931) and PgmL5a(761–840) were PCR amplified and cloned between the EcoRI and XhoI restriction sites of plasmid pGEX6p1 (resulting plasmid sequences in File S2).
For NMR spectroscopy, BL21-Gold (DE3) E. coli cells expressing GST-Pgm(692–768) were grown at 37 °C in M9 minimal medium with 15NH4Cl (1 g/L) (Cambridge Isotope Laboratories, USA) and either D-glucose-13C6 (Cambridge Isotope Laboratories, USA) or unlabeled D-glucose (3 g/L) for expression of 13C/15N- or 15N-labeled Pgm CRD, respectively. At OD600 = 0.6 to 0.8, cells were induced with 0.2 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) for 3 h30 in the presence of 1 mM ZnSO4 and were then collected and suspended in buffer A (25 mM Tris–HCl pH 7.5, 0.15 M NaCl, 10% glycerol, 10 mM 2-Mercaptoethanol) supplemented with 0.5 mM phenylmethane sulfonyl fluoride (PMSF). Cells were lysed with a French press and the cleared supernatant was filtered through a 0.45 μm syringe filter before loading onto 2.5 ml of Glutathion Sepharose™ 4B resin (GE Healthcare) and incubation for 1 h at room temperature. The resin was washed five times with 30 mL of buffer A. The Pgm (692–768)* CRD peptide (Fig. 2e) was cleaved from the GST-tag using 100 units of PreScission protease overnight at 4 °C in 10 mL of buffer A. The supernatant was diluted 3-fold in 25 mM HEPES pH 7.6, 5% glycerol buffer (degassed solution under vacuum) and loaded onto a 1-mL HiTrap™ Q HP column (GE Healthcare). The CRD peptide was eluted through a linear gradient of 50 to 1000 mM NaCl in 25 mM HEPES pH 7.6, 5% glycerol buffer (25 column volumes). CRD-containing samples were identified by SDS-PAGE then pooled and dialyzed twice against NMR solution (5 mM (if 15N-labeled) or 0.5 mM (if unlabeled) HEPES pH 7.5, 25 mM NaCl) using a Spectra/Por3 dialysis membrane, prior to storage at − 80 °C. The whole purification procedure is summarized in Fig. S11.
For DNA or histone binding assays, BL21-Gold (DE3) E. coli cells expressing GST-CRD fusions or GST alone were grown at 37 °C in LB medium supplemented with 0.1 mM ZnSO4. Exponentially growing bacteria expressing each GST fusion protein were induced overnight at 16 °C in the presence of 0.1 mM IPTG or at 30 °C in the presence of 0.5 mM IPTG, then collected and suspended in buffer B (10 mM Tris pH 7.4; 500 mM NaCl; 100 μM ZnSO4; 1 mM DTT; 1% Triton + 1 cOmplete™ tablet EDTA-Free Roche EasyPack). Cell lysis was performed using sonication with a BioRuptor (Diagenode). Five hundred μl slurry glutathione sepharose beads 4B (GE healthcare) were pre-equilibrated 3 times in 2.5 ml of buffer B. Cleared lysates (10 mL) were loaded onto the beads and incubated overnight at 4 °C on a rotating wheel. After incubation, the beads were washed 4 times with 2.5 mL of buffer B (last wash on wheel for 10 min at 4 °C). Proteins were eluted twice with 500 μl 10 mM Tris pH 8 and 20 mM glutathione for 15 min at room temperature. Purified protein concentration was assessed with BiCinchoninic acid Assay (BCA kit ThermoFisher). The whole purification procedure is summarized in Fig. S11.
Expression and purification of MBP fusion proteins
Synthetic DNA fragments (Integrated DNA Technologies) encoding MBP-PgmL1(463–539), MBP-PgmL2(540–614), MBP-PgmL3a(471–550), MBP-PgmL4a(856–931), MBP-PgmL5a(761–840) and the wild-type and mutant versions of MBP-Pgm(692–768), were PCR-amplified and inserted between the EcoRI and PstI sites of the pMAL-c2X vector (New England Biolabs, see plasmid sequences in File S2). Expression and purification of MBP-CRD fusions in BL21-Gold (DE3) E. coli cells was performed as described above for GST-CRD fusions, except that each cleared supernatant was loaded onto a 1-ml MBP-Trap HP Prepacked Column (GE-Healthcare) for affinity purification and the MBP-CRD was eluted with 10 mM maltose. All purified preparations are shown in Fig. S11. For those preparations, in which some of the free MBP tag was released upon over-expression or during bacterial lysis, the fraction of the full-length protein was estimated from the ratio of band intensities obtained from Instant blue-stained gels or from western blots. Band intensities were quantified using the Imagelab software (BIORAD). Only the amount of full-length fusion protein was taken into account to calculate the amount of input CRD in histone-binding assays.
For the expression of MBP-PgmD401A, the GAC codon encoding Asp401 was replaced by an alanine codon (GCC) in plasmid pVL1392-MBP-PGM [61]. To produce recombinant baculoviruses, the resulting plasmid (File S2) and those constructed previously to express MBP-Pgm, MBP-PgmΔCR and MBP-PgmΔCC [34] were each transfected into High Five insect cells together with BD BaculoGold Linearized Baculovirus DNA (BD Biosciences). The purification of MBP-tagged Pgm derivatives from recombinant baculovirus-infected cells was carried out as described [34], with a final elution step in buffer A supplemented with 10 mM maltose.
NMR spectroscopy
High-quality NMR data were obtained for Pgm(692–768)*. NMR protein samples were concentrated to 0.15–0.5 mM in 90% H2O/10% D2O containing 5 mM HEPES, 25 mM NaCl, pH 6.8, or lyophilized and resuspended in 99.99% D2O. NMR experiments were performed at 293 K on 800 MHz or 950 MHz AVANCE III HD Bruker spectrometers equipped with TCI cryoprobes.
1H-15N HSQC spectra of Pgm(692–768)* were acquired at 800 MHz in 90% H2O/10% D2O with 15N carrier set to 120 ppm in order to observe the backbone amide groups. The INEPT half delay was set to 2.8 ms for observation of single-bond coupling (1JHN ≈ 90 Hz).
Optimized long-range 1H-15N HSQC spectrum was recorded in 90% H2O/10% D2O with a 21 ms delay for the selective observation of the long-range proton (Hδ2 and Hε1)-nitrogen (Nδ1 and Nε2) correlations of histidines (2JHN = 6–12 Hz) and the 15N carrier set to 200 ppm.
1H-13C HSQC spectra of Pgm(692–768)* were acquired in 99.9% D2O with 13C carrier set to 45 ppm and a half delay of 1.72 ms corresponding to a 1JHC = 145 Hz, in order to observe the aliphatic protons. 1H-13C HSQC spectra with a half delay of 1.25 ms corresponding to a 1JHC = 200 Hz were also recorded in order to observe the histidine Cδ2-Hδ2 and Cε1-Hε1 cross-peaks. In this case the 13C carrier was placed at 130 ppm with a spectral width of 40 ppm. A total of 256 data points were acquired with 4 transients per point with Echo-Antiecho quadrature in the indirect dimension.
Backbone (N, H, CO, Cα Cβ, Hα, Hβ) and side-chain (Cγ, Cδ, Cε, Hγ, Hδ, Hε) assignments were obtained using standard triple resonance assignment experiments [59] on a 15N-13C labeled protein sample: CBCA (CO) NH, HNCACB, HNCA, HN (CO) CA, HNCO and HN (CA) CO to assign backbone resonances, and H (CCO) NH, (H) C (CO) NH and (H)CCH-TOCSY to assign all the carbon and proton atoms in a given residue. Side-chain assignments were completed using a lyophilized sample redissolved into 99.99% D2O. 15N- and 13C-edited 3D NOESY experiments were acquired on samples in 90% H2O/10% D2O or 99.99% D2O with a mixing time of 150 ms to obtain nuclear Overhauser effect crosspeaks (NOEs) for structure determination. NMR data were collected and processed using TOPSPIN 3.5 software (Bruker) and analyzed using the CcpNmr software [41]. Backbone amide resonances were assigned for all non-proline amino acids except Lys740, for which no cross-peak was detected in the 1H-15N HSQC. We were able to assign 96 and 81% of the backbone and sidechain carbon atoms, respectively, 97 and 84% of the backbone and sidechain hydrogens, respectively, and 96% of the nitrogen atoms composing Pgm(692–768). Pgm(685–768)* assignments were deposited into BMRB (BMRB ID: 34527).
Hydrogen–deuterium exchange experiments were acquired to determine residues protected from solvent through hydrogen bonding as follows. A 500-μL sample of 0.5 mM protein in the protonated NMR solution was lyophilized. The sample was resuspended in the equivalent volume of 99.99% D2O and quickly transferred to an NMR tube. A first 1H-15N HSQC was recorded in 20 min just after the dissolution, and a second one 3 days after the dissolution to observe hydrogen-to-deuterium exchange of backbone amide protons.
The 15N R1 and R2 relaxation rates and {1H}-15N heteronuclear NOE were measured at 20 °C on a 950 Avance III HD spectrometer equipped with a TCI cryoprobe. The 15N R1 and R2 relaxation experiments were based on the refocused 1H-15N HSQC relaxation experiments and recorded in an interleaved pseudo-3D method with an inter-scan delay of 5 s. For the determination of R1 relaxation rate constants, 13 total datasets were collected at relaxation delay times of 10, 50, 100, 200, 300, 400, 600, 800, 1100, 1500, 2000, 2500, 3000 ms. For the determination of R2 rate constants, 13 datasets were collected at delay times of 16.96, 33.92, 50.88, 67.84, 84.8, 101.76, 118.72, 135.68, 152.64, 169.60, 220.48, 271.36, 323.24 ms. R1 and R2 spectra were recorded as 128 × 2126 complex data points. For the backbone {1H}-15N heteronuclear NOE two different spectra were recorded as 512 × 2048 complex data points in an interleaved manner with and without a 5 s proton saturation pulse. The R1 and R2 rates and heteronuclear NOE values and their associated errors were determined from the peak intensities using the CcpNmr software [41]. Relaxation parameters were analyzed with the model-free formalism of Lipari and Szabo [42, 43], using the TENSOR2 program [62] to extract internal dynamical parameters: order parameter S2 describing the amplitude of the motions; internal correlation time τε on the ps-ns timescale and Rex reflecting exchange contribution on the μs–ms timescale. The isotropic tumbling model was selected since no improvement was found with the anisotropic model.
NMR structure calculation
Inter-proton distance restraints were derived from the NOESY spectra (two-dimensional 1H NOESY and three-dimensional 15N- and 13C-NOESY). The Pgm(692–768)* structure was calculated in a semi-automated iterative manner using CYANA 2.1 [63]. Intra- and inter-residue NOEs were manually picked from the 3D NOESY experiments. The backbone dihedral angle restraints (Φ and Ψ angles) were generated using the chemical shift analysis software TALOS+ [40]. Hydrogen bond restraints were determined by hydrogen-deuterium exchange experiments and observation of NOE cross-peaks characteristic of α-helices and β-sheets. NOE peak lists, dihedral angle restraints, hydrogen bond restraints, and chemical shift assignments were used as input for CYANA 2.1. We used the standard CYANA protocol of seven iterative cycles of NOE assignment and structure calculation, followed by a final structure calculation. In each cycle, the structure calculation started from 200 randomized conformers, and the standard CYANA simulated annealing schedule was used with 10,000 torsion angle dynamics steps per conformer. The first calculations with only NOE restraints defined the general fold of the domain and revealed the two Zn2+ coordination modes. In the final refinement stage, distance restraints were added for Zn-Sγ (2.25–2.30 Å), Zn-Nδ1 (2.35–2.40 Å) and Zn-Nε2 (2.35–2.40 Å) and for the other bonds between the four height coordinating atoms, to ensure tetrahedral Zn2+ coordination geometry. Graphic representations were prepared with PyMOL [64]. The structures were deposited into the wwPDB (PDB ID 6ZOP).
In vitro DNA binding assays
Differential Radial Capillary Action of Ligand Assay (DRaCALA) is a simple and rapid filter binding assay allowing the detection of protein-ligand interactions [45]. This method includes no wash step, which represents an advantage over standard filter binding assays by avoiding sample loss and limiting complex dissociation. Briefly, a mixture of a protein and a labeled ligand is spotted directly onto a nitrocellulose membrane. When bound to the protein, the ligand remains at the center of the spot, while free unbound ligand moves by capillarity from the center to the periphery of the spot. We performed DRaCALA assays as described [19]. A 32P-labeled 80 bp double-strand substrate carrying IES 51A1835 from the surface antigen A51 gene was obtained following annealing of complementary top and bottom strand oligonucleotides (Eurofin MWG Genomics, Fig. S3). Only the top strand was labeled at its 5′ end using 32P-γ ATP and T4 polynucleotide kinase (New England Biolabs). Annealing was performed by heating to 95 °C followed by slow cooling to room temperature. The labeled dsDNA substrate (25 nM final concentration) was mixed in 25 mM HEPES pH 7.5, 0.1 mg/ml BSA, 0.5 mM DTT and 100 mM NaCl-containing buffer with an excess of each purified protein at the following final concentrations: GST (5 μM); N-terminal GST fusions for PB(538–594), Pgm(692–768), PgmL2(540–614) and PgmL4a(856–931, 1 to 3 μM), N-terminal MBP fusions for PgmL1(463–539), PgmL3a(471–550) or MBP-PgmL5a(761–840, 400 to 900 nM); full-length Pgm or its mutant derivatives (400 nM). Complexes were loaded onto a Nitrocellulose Hybond ECL membrane. Following air drying, the membrane was exposed to a Phosphorimager screen, which was scanned using a Typhoon scanner (GE Healthcare Life Sciences).
Electrophoretic Mobility Shift Assays (EMSA) using purified CRDs were performed as described [35], using the same 32P-labeled 80-bp dsDNA substrate as above for IES 51A1835 and a 32P-labeled 70-bp dsDNA fragment carrying the left end of IES 51A4404 and its flanking MAC-destined sequences (Fig. S3). After electrophoresis in a 5% acrylamide, 0.5x TBE gel, free DNA and protein-DNA complexes were visualized following exposure of the dried gel to a Phosphorimager screen (see above).
NMR analyses of the interaction of Pgm(692–768)* with DNA and histone H3
Two double-stranded DNA substrates (Eurofins Genomics, Ebersberg, Germany) (Fig. S3), were used in NMR interaction studies. They were obtained through annealing of complementary oligonucleotides, by heating to 95 °C followed by slow cooling to room temperature. Proper annealing was confirmed by the presence of imino protons at 12 to 14 ppm in 1D 1H-NMR spectra (Fig. S12). The interactions between these DNA molecules with Pgm(692–768)* were probed by 1H-1H NOESY experiments with excitation sculpting water suppression [65] with 200 ms mixing time on an AVANCE Bruker 950 MHz spectrometer, with a spectral width of 18,043 Hz with 2202 complex points in t2 and 688 t1-increments. Spectra were acquired on Pgm(692–768)* in the absence and presence of an equimolar amount of DNA.
A titration of 100 μM 15N-Pgm(692–768)* with 50 to 1000 μM of the H3(1–19) histone peptide was followed by 1H-15N-HSQC in 5 mM HEPES pH 6.8, 25 mM NaCl, 95% H2O / 5% D2O at 20 °C, acquired on an AVANCE Bruker 800 MHz spectrometer with 24 scans, 11,160 Hz and 2 k complex points, and 2513 Hz and 240 points in the 1H and 15N dimensions, respectively. A π/2 phase-shifted squared sine bell window function was applied before the FT. The dissociation constant (KD) was determined by least squares fitting of chemical shift changes between free and bound states (Δδobs) to the non-linear equation (adapted from [66]):
$$ {\Delta \delta}_{obs}={\Delta \delta}_{max}\frac{\left({K}_D+{\left[H3\right]}_0+{\left[P\right]}_0\right)-\sqrt{{\left({K}_D+{\left[H3\right]}_0+{\left[P\right]}_0\right)}^2-4\left({\left[P\right]}_0{\left[H3\right]}_0\right)}}{2{\left[P\right]}_0} $$
where Δδobs and Δδmax are the measured and maximum chemical shift perturbations for a given resonance, respectively; and [H3]0 and [P]0 are the H3(1–19) and Pgm(692–768)* concentrations, respectively.
The final titration point (100 μM Pgm(692–768)* and 1000 μM H3(1–19)) was used to observe the effect of this interaction on the peptide. 1H-1H TOCSY experiments were acquired on this sample and a 1000 μM sample of H3(1–19) in the same buffer at pH 6.8. These experiments were acquired on a Bruker 800 MHz spectrometer with 60 ms mixing time, 11,160 Hz spectral width, 96 scans, 2 k and 512 complex points in the 1H and 15N dimensions, respectively.
Paramecium cell culture, nuclei isolation and histone extraction
Autogamous cultures of Paramecium tetraurelia strain 51new [67] were grown according to [68]. Ten hours after 50% of cells became autogamous, they were harvested by centrifugation. The resulting cell pellet was flash-frozen in liquid nitrogen and stored at − 80 °C prior to nuclear extraction. The cell pellet was thawed on ice with 2 volumes of lysis buffer (0.25 M sucrose; 10 mM MgCl2; 10 mM Tris pH 6.8; 0.2% Nonidet P40; 4X CalbioChem protease inhibitor cocktail set I). Cells were lysed with a Potter-Elvehjem homogenizer. The lysate was placed in an Eppendorf tube and subsequently centrifuged at 1000 g for 2 min in a swinging centrifuge and the nuclei-containing pellet was washed 3 times in 5 volumes of wash buffer (0.25 M sucrose, 10 mM MgCl2; 10 mM Tris pH 6.8; 4X CalbioChem protease inhibitor cocktail set I). Acid extraction of histones from the nuclei was performed according to [69]. The purity of histone preparations was monitored on SDS gels (Fig. S4) and their protein concentration was measured using a BCA assay. Around 100 μg of histones were extracted from a 1-L cell culture.
Glutathione S-transferase (GST) and maltose binding protein (MBP) pull-down assays
Expression of GST-CRD and MBP-CRD fusions was carried out as described above. Four hundred μl of MBP/MBP-CRD-containing lysates or 200 μl of 1/10 diluted GST/GST-CRD-containing lysates in buffer B were loaded onto 100 μL pre-equilibrated slurry glutathione sepharose beads 4B (GE healthcare) or amylose resin (New England Biolabs), respectively, and incubated for 1 h at 4 °C. Beads were then washed 3 times in 200 μl of buffer B. Five μg of Paramecium histones were diluted in 100 μl of buffer B, added to the beads and incubated overnight on a rotating wheel at 4 °C. Because buffer B contains 500 mM NaCl, we can exclude that any observed histone/CRD interaction is indirectly mediated through DNA binding, because, as shown by NMR and DRaCALA experiments (Fig. 5), no interaction is detected between the Pgm CRD and DNA in the presence of as little as 100 mM NaCl. After incubation, the beads were washed 4 times with 200 μl of buffer B (last wash on wheel for 10 min at 4 °C). The beads (~ 50 μL) were boiled for 10 min following addition of 25 μl 4X Laemmli sample buffer (Bio-Rad) supplemented with 1.4 M β-mercaptoethanol.
Enzyme-linked immunosorbent assays (ELISA)
Each well of a Pierce streptavidin coated high capacity 96-well plate was washed 3 times with 200 μL of buffer C (25 mM Tris pH 7.4; 150 mM NaCl; 0.1% BSA; 0.05% Tween 20). C-terminally biotinylated H3 peptides (4 μM in 100 μL Wash Buffer) were applied into wells to saturate their surface and the plate was incubated at room temperature for 1 h. The plate was washed three times with 200 μL buffer C and then 100 μL of purified MBP-CRD or GST-CRD fusion proteins (10 to ~100 nM) were added to each well. The plate was incubated at 4 °C overnight, then washed as above to remove unbound proteins. One hundred μL of monoclonal anti-MBP-HRP antibody (NEB #E8038S; diluted 1/10,000 in buffer C) was added to each well and the plate was incubated for 30 min at room temperature and then washed three times. Peroxidase enzyme activity was measured with tetramethylbenzidine (1-Step™ Ultra TMB-ELISA, ThermoScientific) using a TECAN Infinite 200 pro reader at 450 nm.
Microwave plasma atomic emission spectroscopy
The relative amount of zinc complexed to GST-fused Pgm(692–768) and its mutant versions carrying the H701S, H738S or C712S mutations was determined using microwave plasma atomic emission spectroscopy. Proteins were extracted from the bacterial lysate using glutathione-sepharose beads and washed with lysis buffer as described above. The beads were then washed with 2 mM EDTA to remove excess non-complexed zinc. They were then treated with nitric acid to dissolve the glutathione-sepharose beads. The solutions were analyzed by microwave plasma atomic emission spectroscopy using an Agilent 4200 MP-AES spectrometer. Zinc concentrations were determined by comparison to a standard curve and normalized with respect to initial protein concentration.