Thermodynamic Characterization of the Binding Interaction between the Histone Demethylase LSD1/KDM1 and CoREST

Abstract

Flavin-dependent histone demethylases catalyze the posttranslational oxidative demethylation of mono- and dimethylated lysine residues, producing formaldehyde and hydrogen peroxide in addition to the corresponding demethylated protein. In vivo, the histone demethylase LSD1 (KDM1; BCH110) is a component of the multiprotein complex that includes histone deacetylases (HDACs 1/2) and the scaffolding protein CoREST. Although little is known concerning the affinities of or the structural basis for the interaction between CoREST and HDACs, the structure of CoREST^286–482 bound to an alpha helical coiled-coil tower domain within LSD1 has been recently reported. Given the significance of CoREST in directing demethylation to specific nucleosomal substrates, insight into the molecular basis of the interaction between CoREST and LSD1 may suggest a new means to inhibit LSD1 activity by misdirecting the enzyme away from nucleosomal substrates. Towards this end, isothermal titration calorimetry studies (ITC) were conducted to determine the affinity and thermodynamic parameters characterizing the binding interaction between LSD1 and CoREST^286–482. The proteins tightly interact in a 1:1 stoichiometry with a dissociation constant (K_d) of 15.9 ± 2.07 nM, and their binding interaction is characterized by a favorable enthalpic contribution near room temperature with a smaller entropic penalty at pH 7.4. Additionally, one proton is transferred from the buffer to the heterodimeric complex at pH 7.4. From the temperature dependence of the enthalpy change of interaction, a constant pressure heat capacity change (ΔC_p) of the interaction was determined to be −0.80 ± 0.01 kcal/mol·K. Notably, structure-driven truncation of CoREST revealed that the central binding determinant lies within residues 293–380, also known as the CoREST ‘linker’ region, which is a central isolated helix that interacts with the LSD1 coiled-coil tower domain to create a triple helical bundle. Thermodynamic parameters obtained from the binding between LSD1 and the linker region^293–380 of CoREST are similar to those obtained from the interaction between LSD1 and CoREST^286–482. These results provide a framework for understanding the molecular basis of protein-protein interactions that govern nucleosomal demethylation.

As part of epigenetic regulation of gene expression, the structure of chromatin is influenced by association with non-histone proteins or multi-protein complexes, the activity of multiprotein chromatin remodeling complexes, and protein complexes responsible for the posttranslational modification of histones. Protein-protein interactions govern many events in chromatin enzymology, especially those surrounding the site-specific posttranslational modification of histones at the level of individual nucleosomes. In the nucleosome, the basic unit of chromatin, ~147 base pairs (bp) of DNA wind about a histone octamer that is composed of two copies each of the histone proteins H2A, H2B, H3, and H4 (1, 2). The highly conserved sequence of the N-terminal segment of each histone protein extends out of the nucleosome and provides sites for various posttranslational modifications, including acetylation, phosphorylation, ubiquitination, and methylation.

Posttranslational methylation of histones is a modification that has been intimately linked with both transcriptional activation and repression events; the context of these seemingly opposing activities depends on the type of histone being modified, the specific residue being modified, and on the degree of mono-, di- or trimethyl modification (3, 4, 5). Whereas the action of S-adenosylmethionine-dependent histone methyl transferases produces methylated histones, removal of methyl marks is performed by two classes of oxidative demethylation enzymes: flavin-dependent histone demethylases such as LSD1 and LSD2 (also known as KDM1/BCH110 and KDM2/AOF1, respectively); and Jumonji domain C (JmjC) family histone demethylases that utilize Fe (II) and α-ketoglutarate as cofactors.

Human LSD1 and LSD2 belong to the flavin adenine dinucleotide (FAD)-dependent family of amine oxidases, members of which include monoamine oxidase A, monoamine oxidase B, and polyamine oxidase (3, 6). LSD1 selectively catalyzes the demethylation of mono- and dimethylated lysine 4 of histone H3 (H3K4) via the oxidative process outlined in Scheme 1. The mono- or dimethylated amine of lysine is oxidized with assistance from the oxidized FAD cofactor to form an imine intermediate that hydrolyzes to produce formaldehyde and the demethylated side chain on histone H3. In a final step, the flavin cofactor is reoxidized by molecular oxygen, forming hydrogen peroxide (7). Through this mechanism LSD1 can demethylate mono- and dimethylated lysine but lacks the ability to demethylate trimethylated lysine.

Scheme 1.

General mechanism of lysine demethylation by LSD1

As a result of this catalytic activity, LSD1 disrupts methylated histone modification and leads to repressed transcription (4), resulting in severe pathological consequences inducing cancer progression. LSD1 is consistently found in various transcriptional corepressor complexes that include CoREST, CtBP (C-terminal Binding Protein), and HDACs 1/2 (Histone Deacetylase 1 and 2) (4, 5, 6), indicating that LSD1 activity can be modulated by its interacting proteins. LSD1 has also been shown to interact with p53, a tumor suppressor gene, and repress p53-mediated transcription (8). Consequently, the upregulation of LSD1 has been related to an early response to carcinogen exposure in human mammalian epithelial cells (9), a high risk of prostate cancer (10), and breast carcinogenesis (11).

LSD1 alone is sufficient to demethylate H3K4 in peptides or bulk histones, but its activity toward nucleosomal substrates is regulated by its interaction with CoREST (4, 12). CoREST is a 66 kDa protein that functions as a corepressor to the silencer REST (RE1 silencing transcription factor/ neural restrictive silencing factor) (12, 13). It consists of an ELM2 (Egl-27 and MTA1 homology 2) domain, two SANT (Swi3/Ada2/NCoR/transcription factor IIIB) domains, and the linker region between the two SANT domains (Figure 1). Recent studies have shown that the C-terminal fragment of CoREST (residues 286–482), including the linker region and SANT2 domain, is essential for LSD1-catalyzed demethylation of H3K4 within nucleosomes and that the linker region (residues 293–380) of CoREST between the two SANT domains is sufficient to bind LSD1, potentially providing stability to the helical conformation of the LSD1 tower domain (residues 418–522) (4, 6, 12). Consistently, deletional mapping studies (5) have shown that the LSD1 tower domain^418–522 contains the binding site for CoREST.

Figure 1.

Crystal structures of (a) LSD1 in complex with CoREST^286–482, (b) the functional region of CoREST (CoREST^286–482) including the linker region^293–380 and the SANT2 domain^381–450, (c) the linker region^293–380 of CoREST, and (d) their domain structures. The three dimensional structural data file was obtained from Protein Data Bank (PDB entry code 2IW5, (4)). The catalytic domain of LSD1, amine oxidase domain, is shown in green, the SWIRM domain of LSD1 is in red, the tower region of LSD1 is in blue, FAD indicated by a black arrow is in yellow, the SANT2 domain^381–450 of CoREST is in purple, and the linker region^293–380 of CoREST is in orange. All structural figures were generated using PyMoL.

While many studies have demonstrated the physical interaction between LSD1 and CoREST^286–482, little is known about the biophysical nature of the interaction and the molecular details of the interaction sites between LSD1 and CoREST. Here, on the basis of the importance of the functional region of CoREST (CoREST^286–482) in binding to LSD1 as previously studied, we describe the thermodynamic characterization of the binding interaction between LSD1 and CoREST^286–482 using isothermal titration calorimetry to provide a fundamental understanding of their interaction.

MATERIALS AND METHODS

Reagents and Materials

Media, antibiotics, and all other buffer reagents were purchased from Sigma, New England Biolabs, MP Biomedicals, and BD Biosciences. Chemically competent BL21 (DE3) Star Escherichia coli cells were purchased from Invitrogen. Chromatographic protein purifications were carried out on an AKTA FPLC (GE Healthcare), and the protein concentration was determined by UV absorbance spectroscopy. Direct binding studies were performed using isothermal titration calorimetry on a Microcal VP-ITC instrument.

Expression and Purification of LSD1

The gene encoding a truncated form of LSD1 (residues 151–852) (14) was used for expression and purification as previously described (15, 16) except that the gene was cloned into the pET15b vector instead of the pET 151-D/TOPO vector for expression. The resulting plasmid was transformed into chemically competent BL21 (DE3) Star E. coli cells, which were grown on a LB agar plate supplemented with ampicillin overnight at 37 °C. Streaks of the colonies on the plate were used to grow 6 L of cells in TB media with shaking (200 rpm) at 23 °C. When the cell density reached an OD₆₀₀ of 0.6, 0.5 mM IPTG was added to the flasks to induce LSD1 expression. The cells were allowed to grow overnight and collected by centrifugation at 4225 × g. The cell pellets were lysed with an Emulsiflex C-5 cell cracker in buffer containing 50 mM sodium phosphate, 300 mM NaCl, 5 % glycerol, and 0.4 mM PMSF (pH 7.8). LSD1 was then purified by nickel-affinity chromatography, HiPrep 26/60 Sephacryl S200 gel filtration chromatography (GE Life Sciences), and anion-exchange chromatography (Q-Sepharose Fast Flow, GE Life Sciences). The final concentration of LSD1 was determined by absorption spectroscopy at 458 nm (17) and the protein was stored at −20 °C in 80 % glycerol.

Expression and Purification of CoREST^286–482

A vector encoding a truncated CoREST gene described by Yang et al. (4) was constructed and used for expression and purification with the modifications that the gene was cloned into the pET28b vector and a gel-filtration chromatography step was added for improved purification. The vector was transformed into chemically competent BL21 (DE3) Star E. coli cells, which were used to grow for purification. Streaks of the colonies on a LB agar containing kanamycin were used to grow 6 L of bacteria. The cells were grown as described above for LSD1 purification. CoREST^286–482 was purified by nickel-affinity chromatography with a linear gradient from 50 to 500 mM imidazole in 50 mM sodium phosphate and 300 mM NaCl (pH 7.4), followed by gel filtration chromatography (HiPrep 26/60 Sephacryl S100, GE Life Sciences) with buffer containing 50 mM sodium phosphate and 100 mM NaCl (pH 7.4), and cation-exchange chromatography (CM-Sepharose Fast Flow, Sigma) with a linear gradient from 100 to 800 mM NaCl in 50 mM sodium phosphate (pH 7.4). The concentration of CoREST^286–482 was measured spectrophotometrically using the extinction coefficient of 16,950 cm⁻¹M⁻¹ at 280 nm.

Expression and Purification of the linker region^293–380 and SANT2 domain^381–450 of CoREST

The coding sequence of the linker region^293–380 of CoREST was extracted and amplified by using PCR with a forward primer (5’GCGCATATGGTCAAAAAAGAAAAACATAGCACACAAGCTAAA-3’) and a reverse primer (5’-GCGCTCGAGTTAATTACATTTCTGAATGACCTCTGG AGG-3’) under the following conditions: an initial denaturation step for 5 minutes at 95 °C, 30 cycles of denaturation for 1 minute at 95 °C, annealing for 1 minute at a gradient of 54–65 °C, elongation for 100 seconds at 70 °C, and followed by a final elongation step for 10 minutes at 70 °C. The primers were designed to contain NdeI and XhoI restriction sites at the N- and C-termini respectively to allow for facile ligation into the pET28b vector. The vector was transformed into chemically competent BL21 (DE3) Star E. coli cells, and the cells were grown overnight as described above for LSD1 purification. The cell pellets were lysed with an Emulsiflex C-5 cell cracker in buffer containing 50 mM sodium phosphate, 300 mM NaCl, 5 % glycerol, and 0.4 mM PMSF (pH 7.8). The linker was then purified via nickel-affinity chromatography with a linear gradient from 50 to 500 mM imidazole in 50 mM sodium phosphate and 300 mM NaCl (pH 7.8) and followed by cation-exchange chromatography (SP-Sepharose Fast Flow, Sigma) with a linear gradient from 100 to 800 mM NaCl in 50 mM sodium phosphate (pH 7.8). The protein concentration was measured spectrophotometrically at 280 nm with its extinction coefficient of 1,490 cm⁻¹M⁻¹.

The gene encoding the SANT2 domain^381–450 of CoREST was extracted, amplified by PCR with a forward primer (5’- GCGCATATGGCACGTTGGACTACA GAAGAGCAGCTT -3’) and a reverse primer (5’- GCTCTCGAGTTAACTGGGCCC ATTGGTCTCTTCTTTACC-3’), and ligated into the pET 28b vector. The vector was transformed into chemically competent BL21 (DE3) Star E. coli cells, and cells were grown overnight as described above for LSD1 purification. The SANT2 domain was purified via nickel-affinity chromatography with a linear gradient from 50 to 500 mM imidazole in 50 mM MES and 300 mM NaCl (pH 6.0) and cation-exchange chromatography (SP-Sepharose Fast Flow, Sigma) with a gradient from 100 to 800 mM NaCl in 50 mM MES (pH 6.0). The protein concentration was measured spectrophotometrically at 280 nm using its extinction coefficient of 13,980 cm⁻¹M⁻¹.

Enzymatic Assay

Steady-state kinetic assays to determine kinetic parameters for LSD1 activity on the dimethylated H3K4 21-mer peptide substrate with and without CoREST^286–482 were performed by employing a fluorescence assay as previously described (15, 16). All assays were performed at 25 °C, and the product was monitored by a fluorescence plate reader (Molecular Devices SpectraMax Germini EM) at 560 nm excitation and 590 nm emission wavelength. The experiment was performed in duplicate.

Isothermal Titration Calorimetry (ITC)

ITC experiments were performed using a MicroCal VP-ITC microcalorimeter (MicroCal, Northampton, MA) with a cell volume of 1.4346 ml. Protein samples were dialyzed against the buffer containing 50 mM sodium phosphate and 1 mM DTT (pH 7.4) and vacuum-degassed for at least 30 minutes before loading into the calorimeter. Approximately 1.5 ml of the degassed LSD1 and buffer were placed in the sample cell and reference cell respectively, and CoREST^286–482 was loaded into the syringe injector. In most ITC experiments, 3 µl aliquots of CoREST^286–482 at 30 µM concentration were titrated sequentially against LSD1 at 3 µM concentration in the sample cell at 25 °C. Each injection lasted for 6 s, and there was a delay of 300 s between injections. During the titration the stirring speed was 310 rpm and total 40 to 60 injections of CoREST^286–482 were titrated into LSD1. A one site binding model (Origin 5.0 software, MicroCal software, Inc) was used to fit the data. The titrations were performed under different experimental conditions such as various buffers, temperature range of 15 °C to 35 °C, different pH, and changes of buffer additives. For titrations of the linker region^293–380 or SANT2 domain^381–450 of CoREST against LSD1, 3 µl aliquots of the linker region^293–380 or the SANT2 domain^381–450 at 30 µM concentrations were titrated sequentially into 1.5 ml of LSD1 at 3 µM concentration at 25 °C under the same experimental conditions described above for the titration of CoREST^286–482 against LSD1.

Surface Plasmon Resonance (SPR) Measurements

All SPR measurements were made using a BIAcore 3000 instrument, and data analyses were performed using the BIAevaluation 4.1 software (BIAcore). The CoREST^286–482 was immobilized (~2200 RU) on a BIAcore CM5 (research grade) chip using standard amine coupling chemistry with reagents obtained from BIAcore. In a parallel flow cell, LSD1 was immobilized (~2500 RU). During the screening experiments with 50 mM sodium phosphate and 1 mM DTT (pH 7.4) as a running buffer, LSD1 was injected over the immobilized CoREST^286–482 for two minutes at a flow rate of 30 µl/minute to monitor the binding interaction. The surfaces were regenerated by injecting 10 µl glycine (pH 2.0) at 50 µl/min flow rate. The response from the LSD1 surface was used to subtract out the background (non-specific) signal. The K_d value of the LSD1-CoREST^286–482 interaction was estimated by injecting LSD1 in buffer containing 50 mM sodium phosphate and 1 mM DTT (pH 7.4) for 3 minutes at various concentrations ranging from 6.25 nM to 62.5 nM. A global fitting of binding curves at different concentrations with the Langmuir binding model was used to measure the association (k_on), dissociation rates (k_off), and the apparent affinity constant (K_a).

Calculation of the Solvent Accessible Surface Area (ASA)

Solvent accessible surface area (ASA) of the LSD1-CoREST^286–482 complex was estimated by three different programs, GetArea (18), PIBASE (19), and PISA (20) with the radius of the probe set to 1.4 Å. The structural data file of LSD1 in complex with CoREST^286–482 (4) was obtained from the Protein Data Bank (PDB entry code 2IW5) and used to calculate the parameters.

RESULTS

Expression and purification of LSD1 and CoREST^286–486

A recombinant protein variant of human LSD1 with a deletion of the first 150 amino acids at the N-terminus was expressed and purified as previously described (14, 15, 16). The purified LSD1 was visualized on a 15 % SDS-PAGE gel by Coomassie Blue staining (Figure 2-(a)). A His₆-tagged CoREST^286–482, including the linker region and SANT2 domain, was expressed in several E. coli strains under different growth conditions in order to determine the optimal yield. BL21 (DE3) Star E. coli cells exhibited the best expression of CoREST^286–482. Although a higher temperature (37 °C) led to increased expression, the cells were grown at lower temperature (23 °C) to yield soluble protein. The purified CoREST^286–482 was run on a 15 % SDS-PAGE gel (Figure 2-(b)). In order to define the crucial role of the linker region^293–380 of CoREST in binding to LSD1, the linker region^293–380 and the SANT2 domain^381–450 of CoREST were individually expressed and purified from BL21 (DE3) Star E. coli. They were run on 4–20 % gradient SDS-PAGE gels (Figure 2-(c) for the linker region^293–380 and 2-(d) for the SANT 2 domain^381–450 of CoREST).

Figure 2.

15 % SDS-PAGE gels of (a) the purified LSD1 and (b) the purified CoREST^286–482. 4–20 % gradient SDS-PAGE gels of (c) the linker region^293–380 and (d) the SANT2 domain^381–450 of CoREST. Kaleidoscope prestained standards were used (BioRad).

Enzymatic Assays

The demethylation activity of LSD1 against several methylated peptides of varying lengths has been previously reported (17, 21), and showed that the first 21 amino acids of H3 including dimethylated K4 was sufficient for the detectable LSD1 activity. As such, we evaluated kinetic parameters for LSD1 activity on peptide consisting of the first 21 N-terminal amino acids of H3 with dimethylated K4 by using the GraFit 6.0 software (Erithacus Software, West Sussex, UK); kinetic parameters and representative data are shown in Table 1 and Figure 3, respectively. Derived kinetic values are in reasonable agreement with previously reported values (17, 21). The incubation of LSD1 with CoREST^286–482 for 2 hours at 4 °C increased the initial velocity of the catalytic activity of LSD1 roughly 1.6-fold, while decreasing catalytic efficiency (k_cat/K_M). Thus, CoREST^286–482 has little impact on the catalytic efficiency of LSD1 towards peptide substrates. However, it is possible that the catalytic efficiency may be increased when nucleosomal substrates are used because CoREST^286–482 is known to stimulate the catalytic activity of LSD1 toward nucleosomal substrates.

Table 1.

Kinetic Parameters of the Catalytic Activity of LSD1 with the First 21 Amino Acids of H3 with Dimethylated K4 ^a

	Initial Velocity (µM/s)	KM (µM)	kcat (s−1)
LSD1 only	0.028 ± 0.002	2.300 ± 0.305	0.057 ± 0.003
LSD1-CoREST286–482	0.046 ± 0.001	8.833 ± 1.559	0.092 ± 0.001

^a50 mM Tris buffer at pH 7.85 and 25 °C, 19.9 mM peptide substrate used.

Figure 3.

Initial velocity curves of the catalytic activity of LSD1 using the first 21 amino acid residues of H3 with dimethylated K4 as a substrate. The assay was performed in 50 mM Tris buffer, pH 7.85 at 25 °C, and the final concentration of both proteins was 0.5 µM. Data were fitted to the Michaelis-Menten. The black curve represents the activity of LSD1, whereas the green curve represents the activity of LSD1 incubated with CoREST^286–482.

Calorimetric Titration of CoREST^286–482 against LSD1

Previous studies have demonstrated an interaction between CoREST^286–482 and LSD1 (4, 5), but little biophysical information about the interaction exists. Initial experiments to study this binding interaction were performed using a MicroCal VP-ITC microcalorimeter at 25 °C. Calorimetric data were analyzed using the Origin 5.0 software. In order to correct data for dilution, average heats observed in the last 10~15 injections were subtracted from binding data. The data was then fit to a one site binding model to give stoichiometry (n), association constant (K_a), and change in binding enthalpy (ΔH). Free energy of association (ΔG) and change in entropy (TΔS) were calculated from the known thermodynamic relationships.

ITC experiments between LSD1 and CoREST^286–482 yielded a thermogram shown in Figure 4-(a), displaying raw power output versus time of CoREST^286–482 injection into the cell containing LSD1 (top panel) and the corresponding binding isotherm where the enthalpy per mole of CoREST^286–482 as a function of LSD1 is plotted (bottom panel). Each injection of CoREST^286–482 gave rise to exothermic heats of binding, and each peak became smaller as the binding sites on LSD1 were saturated with CoREST^286–482. Thermodynamic parameters for the interaction are summarized in Table 2. The dissociation constant (K_d) for LSD1-CoREST^286–482 interaction was determined to be 15.9 ± 2.07 nM, indicating a tight binding between LSD1 and CoREST^286–482. The thermodynamic parameters of binding of LSD1 to CoREST^286–482 showed an overall favorable enthalpic (−21.3 ± 0.19 kcal/mol) and unfavorable entropic (−10.7 ± 1.40 kcal/mol) contributions near room temperature. The stoichiometry is close to 1, suggesting a simple 1:1 complex.

Figure 4.

(a) A representative calorimetric titration of 3 µM of LSD1 with 30 µM of CoREST^286–482 in 50 mM sodium phosphate and 1 mM DTT, pH 7.4. Thermogram (top panel) represents the heat released after each injection of CoREST^286–482 into LSD1 solution. Binding isotherm (bottom panel) shows the integrated peak area plotted as a function of molar ratio (CoREST^286–482/LSD1). The following parameters were observed: n = 1.14 ± 0.00, ΔH = −21.3 ± 0.19 kcal/mol, K_a = 6.29e7 ± 8.18e6 M⁻¹. The red line indicates the best fit of the ITC data to a one site binding model. (b) Surface plasmon resonance measurement curves obtained during and after injection of LSD1 on chip surfaces with immobilized CoREST^286–482. Sensograms show the association of LSD1 to and dissociation from CoREST^286–482.

Table 2.

Thermodynamic Parameters for Binding of LSD1 to CoREST^286–482 / the Linker Region^293–380 of CoREST ^b

	Kd (nM)	ΔH (kcal mol−1)	TΔS (kcal mol−1)	ΔG (kcal mol−1)	Stoichiometry
LSD1-CoREST286–482	15.9 ± 2.07	−21.3 ± 0.19	−10.7 ± 1.40	−10.6 ± 1.38	1.14 ± 0.00
LSD1-CoREST293–380	7.78 ± 3.50	−13.5 ± 0.37	−2.42 ± 1.09	−11.1 ± 5.00	0.96 ± 0.01

^b50 mM sodium phosphate and 1 mM DTT, pH 7.4

Surface Plasmon Resonance (SPR) Measurements

The tight binding interaction between LSD1 and CoREST^286–482 was confirmed by surface plasmon resonance measurements. As shown in Figure 4-(b), the increase in response units (RU) over time represents the amount of LSD1 bound to CoREST^286–482 that is proportional to the association rate constant (k_on, 5.63×10⁴ ± 1.93×10³ M⁻¹s⁻¹) of the binding interaction. After association, buffer was injected to dissociate the bound LSD1, which results in the decrease in RU over time and yields a dissociation rate constant (k_off, 7.78×10⁻⁵ ± 1.27×10⁻⁵ s⁻¹). Some noises observed in the dissociation phase after 300 seconds, which might affect uncertainty in the determination of the dissociation rate constant (k_off), were removed. These rate constant values suggest a dissociation constant (K_d) of 1.4 ± 6.61 nM, slightly lower than the dissociation constant (15.9 ± 2.07 nM) measured using isothermal titration calorimetry. The small but significant differences between binding constants obtained from two techniques reflect a small degree of enhanced binding during SPR measurement. This difference in binding constants may also be attributed to a slow off rate due to a columning phenomenon frequently observed in SPR experiments (22). Nonetheless, these techniques independently confirm that the interaction between LSD1 and CoREST ^286–482 is tight, specific, and proceeds with a 1:1 stoichiometry.

Calorimetric Titrations of CoREST^286–482 against LSD1 in Different Buffers

The observed calorimetric enthalpy may not solely arise from the binding interaction, since several other events contribute to the heats of binding (23, 24, 25). One such event is proton transfer between a protein-protein complex and buffer due to a shift in protein pK_a on complex formation and buffer ionization. Such proton transfer can be parsed by the following expression:

$$\mathrm{\Delta }{\mathrm{H}}_{\text{cal}}=\mathrm{\Delta }{\mathrm{H}}_{\text{intrinsic}}+\mathrm{N}\mathrm{\Delta }{\mathrm{H}}_{\text{ion}}$$

where ΔH_cal is the sum of calorimetric enthalpy, ΔH_intrinsic is the enthalpy of binding in absence of protonation effects, N is the number of protons transferred during binding, and ΔH_ion is the enthalpy of buffer ionization. A negative slope of the plot of ΔH_cal versus ΔH_ion indicates a net release of protons from protein to the buffer, while a positive slope indicates a net uptake of protons to protein from the buffer (23). In order to assess the contribution of heat of ionization upon binding, titrations were performed in five buffers with different heats of ionization (sodium phosphate with ΔH_ion = 0.9 kcal/mol, PIPES with ΔH_ion = 2.7 kcal/mol, HEPES with ΔH_ion = 5.7 kcal/mol, ACES with ΔH_ion = 7.5 kcal/mol, and Tris with ΔH_ion=11.4 kcal/mol) (26). Table 3 shows thermodynamic parameters for the binding between LSD1 and CoREST^286–482 in various buffers at pH 7.4. A plot of ΔH_cal versus ΔH_ion (Figure 5-(a)) yielded a straight line with a positive slope, indicating that 0.83 ± 0.09 protons are absorbed by the LSD1-CoREST^286–482 complex from the buffer at pH 7.4. ΔH_intrinsic (i.e. when ΔH_ion=0) was determined to be −19.8 ± 0.58 kcal/mol, suggesting that the binding interaction is intrinsically exothermic and enthalpy-driven at room temperature.

Table 3.

Thermodynamic Parameters for Binding of LSD1 to CoREST^286–482 in Various Buffers at 25 °C ^c

Buffer	Kd (nM)	ΔHcal (kcal mol−1)	TΔS (kcal mol−1)	ΔG (kcal mol−1)	Stoichiometry
Phosphate	17.3 ± 3.98	−19.5 ± 0.26	−8.93 ± 2.06	−10.6 ± 2.44	1.11 ± 0.01
PIPES	5.84 ± 1.63	−17.4 ± 0.32	−6.19 ± 1.75	−11.2 ± 3.17	0.91 ± 0.01
HEPES	11.7 ± 4.20	−15.2 ± 0.38	−4.33 ± 1.57	−10.9 ± 3.93	1.01 ± 0.01
ACES	21.2 ± 7.82	−12.6 ± 0.41	−2.48 ± 0.93	−10.1 ± 3.77	0.93 ± 0.02
TRIS	8.54 ± 3.50	−10.9 ± 0.28	−0.12 ± 0.05	−10.8 ± 4.45	0.92 ± 0.01

^c50 mM sodium phosphate (PIPES, HEPES, ACES, TRIS) and 1 mM DTT, pH 7.4

Figure 5.

(a) Plot of the calorimetric enthalpy (ΔH_cal) obtained from binding of LSD1 to CoREST^286–482 as a function of the buffer ionization enthalpy (ΔH_ion) at pH 7.4. The slope of the plot indicates the number of protons transferred between the complex and the buffer upon binding, and the y-intercept indicates the intrinsic ΔH without a protonation effect. At pH 7.4, 0.83 ± 0.09 protons are absorbed to the complex from the buffer. (b) Temperature dependence of the binding enthalpy change for the interaction between LSD1 and CoREST^286–482. The slope of the plot yields the binding heat capacity change (ΔC_p), which is equal to −0.80 ± 0.01 kcal/mol·K.

Calorimetric Titrations of CoREST^286–482 against LSD1 at Different Temperatures

In addition to the contribution of a proton transfer to the binding interaction, another important contribution to binding enthalpy comes from solvent reorganization upon binding (23). The thermodynamic parameter most closely associated with solvent reorganization is the change in constant pressure heat capacity (ΔC_p). The change in heat capacity (ΔC_p) was obtained by measuring ΔH over the temperature range 15 °C to 35 °C in 50 mM sodium phosphate buffer with 1 mM DTT (pH 7.4) (Table 4). At low temperature (10 °C), the signals generated upon binding were too small to fit to a binding model. Titrations at a high temperature (55 °C) yielded an irregular binding curve, suggesting protein denaturation.

Table 4.

Thermodynamic Parameters for Binding of LSD1 to CoREST^286–482 at Different Temperatures ^d

T (°C)	Kd (nM)	ΔH (kcal mol−1)	TΔS (kcal mol−1)	ΔG (kcal mol−1)	Stoichiometry
15	21.7 ± 10.9	−4.24 ± 0.17	5.86 ± 2.95	−10.1 ± 5.07	0.99 ± 0.02
25	3.18 ± 2.54	−12.1 ± 0.39	−0.49 ± 0.39	−11.6 ± 9.27	1.01 ± 0.02
30	3.70 ± 1.74	−16.1± 0.27	−4.42 ± 2.08	−11.7 ± 5.50	1.03 ± 0.01
35	8.03 ± 4.10	−20.2 ± 0.57	−8.79 ± 4.49	−11.4 ± 5.82	0.97 ± 0.02

^d50 mM sodium phosphate and 1 mM DTT, pH 7.4

A plot of ΔH versus temperature yielded a straight line with a negative slope (ΔC_p), −0.80 ± 0.01 kcal/mol·K (Figure 5-(b)). The major contributions to ΔC_p originate from hydrophobic, conformational, and vibrational effects (27), but hydrophobic interactions are dominant contributors (28, 29). The crystal structure of the LSD1-CoREST^286–482 complex shows that the binding interface between LSD1 and CoREST^286–482 is mostly populated by nonpolar residues that are presumably involved in hydrophobic interactions during complex formation (Figure 6-(b)). Also, the effect of NaCl on the change in heat capacity (ΔC_p) was studied over the same temperature range, using varying concentrations of NaCl (10 mM NaCl and 100 mM NaCl) in 50 mM sodium phosphate buffer with 1 mM DTT (pH 7.4). As the concentration of NaCl in buffer increased, larger binding heats (ΔH) were observed, but ΔC_p values did not vary significantly. The ΔC_p values with the addition of 10 mM NaCl and 100 mM NaCl were determined to be −1.04 ± 0.04 kcal/mol·K and −0.84 ± 0.1 kcal/mol·K, respectively.

Figure 6.

(a) Surface representation of the binding interface between the LSD1 tower region^418–522 and the linker region^293–380 of CoREST. The blue and orange surfaces represent the LSD1 tower domain^418–522 and the linker region^293–380 of CoREST, respectively. Based on information from the three dimensional structure of the complex, pale blue and orange surfaces indicate the surface of each protein involved in key binding interactions. (b) Nonpolar residues at the binding interface. The LSD1 tower domain^418–522 and the linker region^293–380 of CoREST are represented in a ribbon diagram, and nonpolar residues involved in the binding interface are represented by a surface diagram. (c) Stereo view of packing in the trimeric coiled coil interaction between the LSD1 tower region^418–522 (blue) and the linker region^293–380 of CoREST (orange). The side chains at a and d position of a heptad repeat model are represented as sticks. (d) Stereo top view of the binding interface. All residues involved in the binding interaction are represented by lines.

Calculation of the Solvent Accessible Surface Area

The solvent accessible surface area (ASA) has been used to predict structural information of proteins (29, 30). The ASA of the binding interface calculated by three different programs was predicted to be about 2500 Å, and the ASA buried at the interface was determined as approximately 5498.0 Å, with a polar/nonpolar ratio, ASA_{polar buried}/ASA_{nonpolar buried} = 1695.2 Å/3802.8 Å = 0.446. The amount of buried nonpolar surface area estimated was 70 % of the interfacial surface, which was presumably the main contributor to the observed negative ΔC_p.

Calorimetric Titrations of CoREST^286–482 against LSD1 as a Function of pH

The effect of pH on the thermodynamic parameters of the binding interaction was studied over the pH range 6.0 to 9.0 in a buffer system of constant ionic strength. At pH 6.0 and 7.0 the buffer consisted of 50 mM sodium phosphate, 50 mM glycine, and 1 mM DTT, whereas 40 mM sodium phosphate, 30 mM glycine, and 1 mM DTT were used for experiments at pH 8.0 and 9.0. The pH range was selected based on previous reports of the stability and activity of LSD1 (16, 17). Table 5 demonstrates that while changing pH produces no significant change in affinity, there is a trend towards a more favorable enthalpic contribution and a larger entropic penalty as pH increases. These changes in enthalpic and entropic contributions to free energy as the pH increases can be rationalized by changes in the protonation state of residues that may cause slight perturbations from the optimal binding interaction between LSD1 and CoREST^286–482. At pH 9.0, a binding isotherm was not observed. High pH presumably alters the structure of LSD1 or CoREST^286–482 or both, changing their binding interaction. Previous work has shown that the catalytic activity of LSD1 is diminished above pH 9.5 (16), in accord with our observations at pH 9.0.

Table 5.

Thermodynamic Parameters for Binding of LSD1 to CoREST^286–482 at Different pH (25 °C) ^e

pH	Kd (nM)	ΔH (kcal mol−1)	TΔS (kcal mol−1)	ΔG (kcal mol−1)	Stoichiometry
6.0	4.18 ± 0.96	−8.56 ± 0.01	2.87 ± 0.66	−11.4 ± 2.62	0.90 ± 0.004
7.0	6.75 ± 1.49	−12.9 ± 0.15	−1.54 ± 0.34	−11.4 ± 2.51	1.00 ± 0.01
8.0	7.66 ± 2.91	−15.6 ± 0.32	−4.48 ± 1.71	−11.1 ± 4.22	1.00 ± 0.01
9.0	nb	nb	nb	nb	nb

^e50 mM sodium phosphate, 50 mM glycine, and 1 mM DTT, pH 6.0, 50 mM sodium phosphate, 50 mM glycine, and 1 mM DTT, pH 7.0, 40 mM sodium phosphate, 30 mM glycine, and 1 mM DTT, pH 8.0, 40 mM sodium phosphate, 30 mM glycine, and 1 mM DTT, pH 9.0. nb = no binding curve observed

Effect of Disulfide Bond Formation on the Binding Interaction

A cysteine from the LSD1 tower domain^418–522 (residue 491) and another from CoREST^286–482 (residue 379) are present at the binding interface: these residues can presumably participate in disulfide bonds. Concomitant protein oxidation leading to the mixed disulfide bonds between LSD1 and CoREST^286–482 would confound interpretation of thermodynamic data. In order to study the effect of disulfide bond formation upon binding, titrations were performed at 25 °C (pH 7.4) with buffer containing only 50 mM sodium phosphate without DTT. The thermodynamic parameters measured under these conditions were in reasonable agreement with those obtained with the buffer containing 1mM DTT. On the basis of these data we concluded that derived thermodynamic parameters do not include contributions from disulfide bond formation.

Calorimetric Titration of the CoREST Linker Region^293–380 and SANT2 domain^381–450 against LSD1

Titrations of the linker region^293–380 of CoREST against LSD1 were performed at 25 °C (pH 7.4) in order to define the role of the linker region^293–380 of CoREST in complex formation and to study thermodynamic parameters for the binding interaction with LSD1. Table 2 shows that although diminished enthalpic and entropic change were observed in the binding interaction between LSD1 and the linker region^293–380 of CoREST compared to the those for binding between LSD1 and CoREST^286–482, there is no significant difference in affinity, with K_d of 7.78 ± 3.50 nM. Interestingly, the titration of the SANT2 domain^381–450 of CoREST against LSD1 showed no binding (Figure 1S) under the same experimental condition, indicating no or only very weak binding. This result indicates that the residues crucial for the interaction with LSD1 lie in the linker region^293–380 of CoREST, and the lack of the binding interaction observed between LSD1 and the SANT2 domain^381–450 supports the importance of the linker region^293–380 of CoREST for LSD1 binding.

DISCUSSION

In this study, thermodynamic parameters for the binding interaction between LSD1 and CoREST^286–482 were determined calorimetrically. ITC experiments demonstrate that the binding between LSD1 and CoREST^286–482 is a tight interaction with a dissociation constant (K_d) in the nanomolar range, and that important residues involved in the binding interaction are localized within the linker region^293–380 of CoREST. The high binding affinity is attributed to various non-covalent interactions between LSD1 and CoREST^286–482 upon binding as shown in Figure 6. In addition to the helical interactions between the linker region^293–380 of CoREST and the LSD1 tower domain^418–522, the N-terminal region of CoREST^286–482 wraps around the bottom of the anti-parallel helices of the LSD1 tower domain^418–522, contributing to the overall strength of the binding.

Most thermodynamic data of the binding between LSD1 and CoREST^286–482 showed the enthalpies of interaction greater than the free energies of association and the unfavorable entropies of association, which is commonly observed in biomolecular interactions. In the binding interaction between LSD1 and CoREST^286–482, the favorable change in enthalpy reflects a net increase in the number or strength of bonds formed between two proteins. Based on the known crystal structure of the LSD1-CoREST^286–482 complex (4, 5), several candidates for the formation of hydrogen bonds and salt bridges at the interface are found; interactions between residues of F315, D320, D339, K353, N356, K360, R371 of the linker region^293–380 of CoREST and Q419, K421, K424, Q438, D495, E505, E512 of the LSD1 tower domain^418–522 may contribute directly to the enthalpic gain. The entropy of association can be affected by contributions associated with proteins and solvent (31). The unfavorable entropies of binding observed most likely can be ascribed to the contributions from conformational changes upon binding. Previous studies showed a linear correlation of the change in conformational entropy with the change in total entropy of binding (31).

Conformational changes occurring in the binding interaction between LSD1 and CoREST^286–482 can affect the environment of amino acid residues and thus their pK_a values (23). This results in the transfer of protons between the buffer and ionizable groups of amino acid residues. As this process contributes to the sum of calorimetric enthalpy change, measured changes must be corrected for the proton transfer effects. In order to perform this correction, the formation of LSD1-CoREST^286–482 complex was analyzed in buffers with different ionization enthalpies. The positive slope (0.83 ± 0.09) of the plot of ΔH_cal versus ΔH_ion demonstrates that the LSD1-CoREST^286–482 complex takes up approximately one proton from the buffer system at pH 7.4. Presumably this protonation event involves either lysine or arginine in the LSD1-CoREST^286–482 complex near neutral pH.

It is widely known that protein- protein interaction is driven by hydrophobic interactions to a large extent, which results in a negative change in heat capacity (28, 29). The change in heat capacity of the binding interaction between LSD1 and CoREST^286–482 obtained from the slope of the linear temperature dependency of the change in enthalpy was negative as predicted based on the structural features of the complex (Figure 6-(b)). In addition, the change in accessible surface area (ΔASA) allows us to calculate the empirical heat capacity change by the following equation (32):

$$\mathrm{\Delta }{\mathrm{C}}_{\mathrm{p}}=0.36\text{cal}/\text{mol}·\mathrm{K}·{\AA }^2(\mathrm{\Delta }\mathrm{A}\mathrm{S}{\mathrm{A}}_{\text{nonpolar}})-0.25\text{cal}/\text{mol}·\mathrm{K}·{\AA }^2(\mathrm{\Delta }\mathrm{A}\mathrm{S}{\mathrm{A}}_{\text{polar}})$$

Where ΔASA_nonpolar is the change in nonpolar accessible surface area, ΔASA_polar is the change in polar accessible surface area, and 0.36 and −0.25 are empirical constants for nonpolar and polar surface area, respectively. The calculation of ΔASA_nonpolar and ΔASA_polar was performed using GetArea (18). The values of the changes in polar, nonpolar, and total accessible surface area are summarized in Table 1S. The value of the change in heat capacity predicted by the equation above is −1.16 kcal/mol·K, which is somewhat larger than the experimentally determined value of −0.80 kcal/mol·K, but which is in at least qualitatively agreement. This result allows us to conclude that the nonpolar surface area contributes more to the buried surface, in agreement with the negative change in heat capacity.

Our thermodynamic parameters for the interaction between LSD1 and CoREST^286–482 can be rationalized in terms of the structure of the binding interface. The linker region^293–380 of CoREST is clearly important for binding to LSD1, with thermodynamic parameters in reasonable accordance with those for the binding between LSD1 and CoREST^286–482. In contrast, no binding isotherm was observed during titration between LSD1 and the SANT domain^381–450. The LSD1 tower domain^418–522 was previously reported to be sufficient for interaction with CoREST (5). These data led us to conclude that the binding interface occurs between LSD1 tower domain^418–522 and the linker region^293–380 of CoREST, and for this reason, only the linker region^293–380 of CoREST and the LSD1 tower domain^418–522 is considered for the detailed interaction study below.

As shown in Figure 6-(c) and (d), the binding interface consists of three α-helices: one helix from the linker region^293–380 of CoREST (orange-colored) and two anti-parellel helices from the LSD1 tower domain^418–522 (blue-colored). This type of interaction is typically classified as a coiled-coil interaction, and is found in various proteins including intermediate filaments, cell surface receptors, molecular motors, and transcription factors (33). Like the interaction between the LSD1 tower domain^418–522 and the linker region^293–380 of CoREST, trimeric coiled-coil interactions have been reported in various studies of the oligomerization domains of hemagglutinin membrane glycoprotein (34), C-type mannose-binding protein (35), mechanisms of Laminin chain assembly (36), the crystal structure of GCN4-pI_QI (37), and the crystal structure of the de novo designed V_aL_d (38).

Coiled-coil interactions are usually characterized by a heptad repeat sequence (abcdefg)_n (39–42), with hydrophobic residues at a and d position whose interaction is well-known as a main driving force for the stability of helical conformations, and polar or charged residues at e and g position, providing further stability of the coiled-coil interaction through ionic interactions. However, unlike this classic periodicity, all the hydrophobic residues of three α-helices in the LSD1-CoREST^286–482 complex are not assigned at a and d position. The presence of polar residues such as asparagine, glutamine, glutamic acid, and lysine is observed at those positions, suggesting their role in the formation of hydrogen bonds and ionic interactions. This structural feature presumably compensates for the destabilizing effect of desolvation (43, 44).

For a more detailed analysis at a molecular level, the binding interface was dissected into three portions. The first consists of two α-helices of an N-terminus of the linker region^293–380 of CoREST, an N-terminus of the LSD1 tower domain^418–522, and a tail of a C-terminus of the LSD1 tower domain^418–522 (Figure 7-(a)). Most of the interactions in the first region of the interface originate from hydrophobic contacts, with two hydrogen bonds formed between D339 of CoREST and K424 of LSD1, and D320 of CoREST and K421 of LSD1. The second region of the interface consists of three α-helices, which comprise residues 433–450 and 496–513 of LSD1 and 345–365 of CoREST. Several hydrogen bonds and ionic interactions are observed at this interface, as well as hydrophobic packing (Figure 7-(b)). Interestingly, three residues – K353, N356, and K360 – in the linker region^293–380 of CoREST, which occupy periodically almost equivalent positions, are involved in both hydrogen bonds and salt bridge formation with both α-helices of the LSD1 tower domain^418–522, contributing to the tight binding interaction. The third region of the interface, defined by residues 451–495 of LSD1 and 366–380 of CoREST, includes anti-parallel α-helices of the LSD1 tower domain^418–522 with a loop and a single turn of α-helix of the linker region^293–380 of CoREST with R371 at the end of the helix (Figure 7-(c)). R371 of CoREST interacts with D495 on one helix of LSD1 at a distance of about 2.6 Å. Based on this structural study, the important binding interactions seem to be localized to the second region of the interface where all three helices are held tightly by several non-covalent interactions. However, the first and third regions of the interface also play an important role in the overall binding interaction by providing hydrophobic interactions and at least some hydrogen and ionic bonds. Future mutagenesis studies may illuminate the means by which individual residues affect the binding interaction.

Figure 7.

Detailed molecular contacts between the LSD1 tower region^418–522 (blue ribbon) and the linker region^293–380 of CoREST (orange ribbon). Residues are labeled in black for LSD1 and in red for CoREST. Hydrogen bonds are represented by black dashed lines. (a) Interface part I (LSD1 residues 418–432, 514–522; CoREST residues 308–344) (b) Interface part II (LSD1 residues 433–450, 496–513; CoREST residues 345–365) (c) Interface part III (LSD1 residues 451–495; CoREST residues 366–380) (d) Stereo view of the interface part II. Residues involved in hydrogen bonds in the interface part II are shown.

Sequence alignment analysis using the DALI database (45) revealed that many proteins overlap with either two α-helical structures of LSD1^418–522 or one helix of the linker region^293–380 of CoREST, suggesting that the helical structure is common to other proteins. Although their sequences do not match with high scores, helical conformations can have structurally similar characteristics. Notably, the trimeric helical features formed by the LSD1 tower domain^418–522 and the linker region^293–380 of CoREST were not scored as hits by DALI analysis, indicating their unique structure. This analysis suggests that although the helical structure of either the LSD1 tower domain^418–522 or the linker region^293–380 of CoREST can be used as a structural motif for existing proteins, the trimeric helical structure is less well precedented. Thus, small molecules targeting the interaction between LSD1 and CoREST may exhibit high selectivity.

The results of the thermodynamic study of binding between LSD1 and CoREST^286–482 led us to question the role of full length CoREST and the effects of other corepressor proteins such as HDACs 1/2 associated with the LSD1/CoREST complex on binding. Previous studies by Yu, Lei, Shi, and Mattevi (4, 5, 6, 21) demonstrated the importance of the C-terminal region^286–482 of CoREST in mediating the demethylase activity of LSD1. In deletion mapping studies, the linker region^293–380 of CoREST was determined as a putative binding site for LSD1, and the SANT2 domain^381–450 apparently stimulates the demethylation acitivity of LSD1 by bringing LSD1 to its nucleosomal substrates. Another study by Shiekhattar and coworkers examined the effect of full length CoREST in regulating the activity of LSD1 (46). Addition of full length CoREST increased the demethylase activity of LSD1, and the deletional mapping analysis showed the necessity of both the SANT1 and SANT2 domains for the nucleosomal demethylation, although the stimulatory activity of SANT1 was weak. The SANT1 domain may act as a bridging sequence between LSD1 and its substrates because of the structural similarity with the SANT2 domain. Thus, the SANT1 and SANT2 domains may independently facilitate LSD1-mediated demethylation. Also, the activity of HDAC1 increased by addition of full length CoREST owing to the role of an ELM2 domain of CoREST that is known to mediate the deacetylase activity of HDACs 1/2 (13, 47).

Based on these previous studies, full length CoREST seems to have similar activity to the truncated CoREST^286–482. These results let us presume that the ELM2 domain and SANT1 domain of CoREST may be located away from the stalk formed between the LSD1 tower domain^418–522 and the linker region^293–380 of CoREST, and that consequently the binding affinity between LSD1 and full length CoREST should not be significantly different from the binding affinity obtained between LSD1 and the truncated CoREST^286–482. However, the exact role or function of either domain on the binding affinity can not be rationalized until a crystal structure of LSD1 and full length CoREST is determined. Whether the presence of HDACs 1/2 affects the binding affinity between LSD1 and the full length of CoREST may depend on the location of the ELM2 domain in the complex because of its physical association with the ELM2 domain (13). Assuming that the N-terminal region of CoREST, including the ELM2 domain and SANT1 domain, is not associated with the binding region of the LSD1-CoREST complex, the presence of HDACs 1/2 may not affect the binding affinity significantly as well. However, it is also possible that HDACs 1/2 is sufficiently close to the binding region of LSD1 and full length CoREST due to its spatial occupancy upon binding to the ELM2 domain, to result in weakening the binding affinity.

In recent studies (48, 49), LSD2 has been identified as a flavin-dependent histone demethylase, with enzymatic activity and substrate specificity profile similar to those of LSD1. Unlike LSD1, however, LSD2 does not contain the tower domain^418–522, which is essential for binding to CoREST, but has a CW-type zinc-finger domain at its N-terminus between residues 130–200. Accordingly, LSD2 is not able to interact with CoREST, distinguishing it from LSD1. The role of the zinc-finger domain in LSD2 remains unclear, but we speculate with some degree of confidence that this sequence may facilitate direct binding of LSD2 to nucleosomal DNA.

In summary, our thermodynamic study has verified the tight binding interaction between CoREST, particularly the linker region^293–380 of CoREST, and LSD1, which is both enthalpically and energetically favorable. Future mutational studies will provide valuable insights into key residues involved in the tight binding interaction, supporting our structural analysis. Furthermore, based on the binding study between LSD1 and CoREST, a study of the binding interaction between HDACs 1/2 and the LSD1/CoREST complex should provide a good framework to downregulate epigenetic transcriptional derepression mechanisms associated with cancer progression.

ABBREVIATIONS

LSD1

lysine-specific demethylase 1

CoREST

corepressor to the silencer REST (RE1 silencing transcription factor/ neural restrictive silencing factor)

FPLC

fast protein liquid chromatography

KDM1

newly accepted nomenclature for the LSD1 histone demethylase

ITC

isothermal titration calorimetry

ΔH

enthalpy change

ΔS

entropy change

ΔC_p

constant pressure heat capacity change

K_d

the equilibrium dissociation constant

n

stoichiometry

K_a

the equilibrium association constant