Nonenzymatic acetylation of ubiquitin Lys side chains is modulated by their neighboring residues

Abstract

Nonenzymatic acetylation of Lys side chains (Lys-SCs) by various in vivo reactive molecules has been suggested to play novel regulatory roles. Ubiquitin (UB) has seven Lys residues that are utilized for synthesis of specific poly-UB chains. To understand the nature of these Lys-SC modifications, the chemical acetylation rate and pK_a and Hill coefficient of each UB-Lys-SC were measured. Mutagenesis studies combined with the determination of activation energy indicated that specific neighboring residues of the Lys-SCs have a potential catalytic activity during nonenzymatic acetylation. Based on the shared chemistry between nonenzymatic Lys acetylation and ubiquitylation, the characterized chemical properties of the UB-Lys-SCs could be a reference for deciphering both mechanisms. Our NMR approaches could be useful for studying general nonenzymatic Lys acylations of various proteins.

Keywords: 2-acetylthio acetamide; NMR; nonenzymatic acetylation; ubiquitin; ubiquitin Lys pKa.

Figure 1. Ribbon models of UB structures (PDB code, 1D3Z)

The polar and charged residues that are located within 6 Å distance to the N_ε atom of 7 Lys residues (blue) are shown. The neighboring residues of each Lys-SC, of which distance to each N_ε atom is less than 4 Å (black) and 6 Å (gray), are listed at the top, and the measured distances (Å) are shown in parenthesis.

Figure 2. Non-enzymatic acetylation of UB-Lys-SCs and assignment of the resulting Lys-SC_amide peaks using ^13C/15NLys-UB

(A) The HSQC spectra of ^15NUB before (left) and after (right) the acetylation by 2-acetylthio acetamide (2AA). The incubation of ^15NUB with 2AA produced the de novo HSQC peaks of in the UB-Lys-SC_amides at distinct region (boxed region), which are well separated from those of the BB_amides. Although the heterogeneous and incomplete acetylation of the Lys-SCs resulted in a very complicated HSQC spectrum, the HSQC peaks of the Lys-SC_amides were well resolved for analyses. (B) The HSQC spectra of the highly (left) and moderately (right) acetylated ^13C/15NLys-UB by acetic anhydride (AA) and 2AA, respectively, are shown. The original BB_amide HSQC peaks without acetylation are indicated using the prefix, “o”. The HSQC peaks of the BB_amides were split by the partial acetylation of the neighboring Lys-SCs (indicated by a prime). The HSQC spectrum of the highly acetylated ^13C/15NLys-UB at pH 6.0 clearly resolves three Lys-SC_amide peaks (K29, K33, and K33’) (left panel, inset). A possible candidate of the K29-SC_amide in the right panel is indicated by gray text. (C) The selected regions of the 3D CNH-NOESY spectrum that were recorded using the mild acetylated ^13C/15NLys-UB. The correct assignment of the K33/K33’-SC_amide peaks is supported by the NOE-connectivity shown in the selected ¹⁵N-slices of the K33/K33’-SC_amide peaks and the corresponding BB_amide peaks.

Figure 3. Determination of acetylation rate constants of the UB-Lys-SCs

(A) The time-course acetylation of the Lys-SCs was done with 15 mM 2AA at 35°C. The reaction mixture was sampled at each time point, and then the HSQC spectra were recorded. The increasing volume integrals of the Lys-SC_amide peaks with increasing reaction time were fitted to the equation of first order reaction kinetics. To determine the rate constants, the initial velocities, the slopes at time zero, were derived from the fitted parameters. The errors of rates are estimated from the SEM values of the fittings. (B) The HSQC spectra of ^15NUB were measured after incubation with 13.5 mM 2AA and 10 mM acetyl-CoA for 6 and 42 h at 35°C, respectively. To determine the rate constants mediated by acetyl-CoA, the time-course HSQC spectra were recorded. The rate constant of each Lys-SC was estimated from the slope of the time-dependent increasing peak volumes. Although the acetylation rates of the UB-Lys-SCs by acetyl-CoA were 4~20 fold lower than those by 2AA, the positions of the produced HSQC peaks were the same with those produced by 2AA.

Figure 4. The acetylation amounts of the UB-Lys-SCs are linearly dependent on the concentration of 2AA

The acetylation reactions were stopped by adding excess amount of DTT after 1-h reaction incubation. The peak volumes of the Lys-SC_amides were normalized to that of the G76-BB_amide peak (gray circle). The overall acetylation of the Lys-SCs decreased as the reaction progressed due to the decreased amounts of the native Lys-SCs. The volume integrals were adjusted by the assumption that the early reaction part followed the pseudo-first order kinetics, which increased the linear correlation between the acetylation amounts of the Lys-SCs and the concentrations of 2AA.

Figure 5. The chemical shifts (CSs) assignment of the ^13C/15NLys-UB and the pH-dependent CS changes

(A) The CSs of UB-Lys-SCs were determined using the 2D versions of the 3D HCCH-COSY and 3D CCH-TOCSY spectra. The peaks of the Lys-¹³C_ε (CE) in the 2D CCH-TOCSY spectrum are split due to the incomplete decoupling of the ¹⁵N_ε atom (inset), and thus their CSs were averaged for the pK_a analysis of the Lys-SCs. (B) The pH-dependent ¹³C-CSs (top) and the ¹H-CSs of K48 (bottom) show that ^13C/15NLys-UB maintains its native structure even at pH greater than 12. The CSs of the non-degenerated two ¹H peaks attached to ¹³C_β, ¹³C_γ, ¹³C_δ, and ¹³C_ε were averaged for the analysis (QB, QG, QD, and QE). The chemical shift values of ¹³C_δ (CD) and ¹H_ε (QE) were most sensitive to the variation of pH values, but those of ¹³C_α (CA) and ¹H_α (HA) did not change too much. The variation of the ¹³C_α- and ¹H_α-CSs is less than 1.0 and 0.1 ppm, which showed that the native structure of UB was maintained in all used pH values.

Figure 6. Determination of pK_a and n values of the UB-Lys-SCs and the deprotonated fractions of the Lys-SCs to the acetylation rate constants

(A) The pH-dependent ¹³C_δ-CSs of the Lys-SCs are shown as representative. The pK_a and n values were obtained from the global fitting of the pH-dependent ¹³C-CSs. Detailed values from the fitting are shown in Table 1. (B) The rate constant (k) does not exactly correlate with the f-NH₂, [NH₂]/([NH₂]+[NH₃⁺]), at pH 7.5, in which the lowest n of K48 greatly increases the f-NH₂. The error ranges of the f-NH₂ were estimated from the errors of the pK_a and n values (Table 1). The correction of the f-NH₂ with the R_SASA, the ratio of the SASAs of the Lys-N_ε to that of the K6-N_ε, improved this correlation, but not complete. The values of the R_SASA were calculated for the crystal structure of UB (1UBQ) using 1.4 and 2.5 Å probe radius are shown as red and blue letters, respectively. The lowest K27-acetylation seemed to result from the smallest SASA of K27-N_ε atom.

Figure 7. Effect of neighboring residues on the acetylation of each Lys-SC by 2AA molecule

(A) The neighboring residues of each Lys-SC were mutated into Ala, and then a limited acetylation reaction was done with 3 mM 2AA at 35°C overnight. The peak positions of the Lys-SC_amides show that the specific mutations only affected the peak position of the corresponding Lys-SC_amide, except for K11 (1, K6; 2, K11; 3, K29; 4, K48). The E34A mutation greatly affected the K11-SC_amide peak and resulted in overlap with the K33- and K33’-SC_amide peaks. (B) By limited acetylation with 3 mM 2AA, the acetylation amounts of each Lys-SCs were assessed for the point-mutation of each neighboring residue. The point-mutations of the neighboring residues specifically changed the acetylation of the corresponding Lys-SCs (T12A and H68A for K6; N25A for K29; R54A, Y59F, N60A, and N60D for K48), except for K11-SC (E34A). The overlapped peaks of K11-, K33-, and K33’-SC_amide resulted from the acetylation of UB^E34A were treated as a single peak group. The acetylation amount of this peak group (0.33) is ~2.3-fold higher than that of the wild-type UB (0.14), and is higher than the limit of the Y-scale (open circle). The peak of the K29-SC_amide was too weak to be analyzed properly for the N25A mutation.

Figure 8. Determination of the activation energy (E_a) for the non-enzymatic acetylation of the UB-Lys-SCs

(A) The non-enzymatic acetylation rates of the Lys-SCs were measured as varying the temperature, and then the E_a and SEM values were determined from the linear fitting using the Arrhenius equation. The K29-SC_amide peak was too weak to be integrated properly, and thus the peak intensities, instead of the peak volumes, were used for the analysis. (B) The activation energy (E_a) for the non-enzymatic acetylations of the UB-Lys-SCs were plotted against -Ln (k) at 35°C (left), f-NH₂ at pH 7.5 (middle), and SASAs (right). The -Ln (k) values likely have a linear correlation with the E_a when the values of the pre-exponential factor, A, are identical. The E_a do not apparently correlate with any of the f-NH₂ or SASAs of the UB-Lys-SCs. (C) The E_a values of the UB^H68A-Lys-SCs are plotted against those of the UB-Lys-SCs. A slightly different experimental temperature likely caused a systematic difference of the E_a between the UB^WT and UB^H68A protein, which might result from different temperature calibrations of the NMR instruments. The H68A mutation specifically increased the E_a of the K6-acetylation about 1.45 kcal/mol. (D) Models for the catalytic roles of H68 (top) and E34 (bottom) during the K6- and K11-acetylation, respectively, are illustrated. The presence of E34 increased the K11-pK_a, which decreased the K11-acetylation rate probably due to low A value. However, the proximity of the E34-carboxylate to the K11-N_ε atom (2.9 Å) could decrease the E_a by providing a catalytic activity.