Sequence-specific ¹H, ¹³C and ¹⁵N assignments of the phosphoesterase (PE) domain of Pseudomonas aeruginosa DNA ligase D (LigD)

Abstract

DNA ligase D (LigD), consisting of polymerase, ligase and phosphoesterase domains, is the essential catalyst of the bacterial non-homologous end-joining pathway of DNA double-strand break repair. The phosphoesterase (PE) module performs manganese-dependent 3’-phosphomonoesterase and 3’-ribonucleoside resection reactions that heal broken ends in preparation for sealing. LigD PE exemplifies a structurally and mechanistically unique class of DNA end-processing enzymes. Here, we present the resonance assignments of the PE domain of Pseudomonas aeruginosa LigD comprising the N-terminal 177 residues.

Biological context

Bacteria repair double-stranded DNA breaks (DSBs) by either of two distinct pathways – RecA-dependent homologous recombination (HR) or Ku-dependent nonhomologous end joining (NHEJ) (Shuman and Glickman, 2007). Bacterial NHEJ is catalyzed by a multifunctional DNA ligase D (LigD). Pseudomonas aeruginosa LigD consists of an N-terminal phosphoesterase (PE) domain (residues 1-187), a central ATPdependent ligase (LIG) domain (residues 188-527) and a C-terminal polymerase (POL) domain (residues 533-840). POL catalyzes the incorporation of deoxyribonucleoside or ribonucleoside monophosphates at the 3’-ends of DSBs prior to their sealing by LIG. Atomic structures have been determined for the POL and LIG domains of bacterial LigD proteins bound to substrates at various functional states along their respective reaction pathways (Shuman and Glickman, 2007). On the other hand, only recently has the PE domain been characterized structurally (Nair et al., 2010).

The PE component of LigD is a 3’ phosphodiesterase that removes the short 3’-ribonucleotide tracts generated by POL at 5’-overhang DSBs, yielding a 3’-monoribonucleotide end that is the preferred substrate for sealing by NHEJ ligases (Zhu et al., 2005). PE also has a 3’ phosphomonoesterase activity that converts DNAs with non-ligatable 3’-PO₄ ends to 3’-OH ends that can be extended by POL or sealed by LIG. The 3’ phosphodiesterase and 3’ phosphomonoesterase reactions of LigD PE are characteristically dependent on Mn⁺² (Mg²⁺ is ineffective in promoting catalysis).

We are applying the complementary methods of X-ray crystallography and solution NMR spectroscopy to obtain atomic structures of LigD PE and thereby illuminate its catalytic mechanism, substrate specificity, and conformational dynamics. We recently reported the 1.9 Å crystal structure of the active phosphodiesterase core of Pseudomonas aeruginosa PE (PaePE), spanning residues 30-187 (Nair et al., 2010). PaePE has a novel fold in which an 8 stranded β-barrel with a hydrophobic interior supports a crescent-shaped hydrophilic active site on its outer surface. Six essential side chains (His42, His48, Asp50, Arg52, His84, Tyr88) were found to coordinate a manganese ion and a sulfate mimetic of the scissile phosphate. The crystal structure accounted for the manganese requirement (insofar as two of the metal ligands were histidines) and showed that PE active site is unlike that of any known DNA/RNA endhealing enzymes or any known phosphotransferases (Nair et al., 2010).

A shortcoming of the X-ray structure was that it omitted the protease-sensitive N-terminal peptide of PaePE that includes several residues important for the 3’ phosphomonoesterase activity (Zhu et al., 2005). Our aim in the present study was to obtain an NMR structure of an active version of PaePE that includes the N-terminal peptide and the core PE fold. After testing incremental truncations from the C-terminus of the PE module, we found that PaePE-(1-177) retained 3’ phosphodiesterase and monoesterase activities comparable to that of PaePE-(1-187). We exploited PaePE-(1-177) for solution NMR studies, after finding that this construct provided spectra of higher quality than PaePE-(1-187), especially in the experiments required for the assignment of side chain ¹H resonances, a crucial step toward obtaining its solution structure. Here we report the main chain and side chain ¹⁵N, ¹H and ¹³C resonance assignments for PaePE-(1-177).

Methods and experiments

Production and purification of PaePE

A pET28b-based plasmid encoding the N-terminal 177 residues of P. aeruginosa LigD fused to an N-terminal His₁₀-SUMO tag, was transformed into E. coli BL21 (DE3)-RIPL cells. Cells from individual colonies were inoculated into 20 mL of M9 medium prepared in 100% H₂O. After overnight growth, cells were subsequently transferred stepwise into 20 mL M9 media containing successively increasing proportions of D₂O (25%, 50%, 75% and finally 100%). Cells grown overnight in M9-(100% D₂O) were inoculated into 1 L of M9-(100% D₂O) containing ¹⁵N-labeled ammonium chloride and ¹³C,²H-labeled glucose as the only nitrogen and carbon sources, respectively. The cultures were incubated at 37° C with shaking at 250 rpm until the A₆₀₀ reached 0.6, then 1 mM IPTG was added to induce PaePE production. The induced culture was incubated at 16° C for 16 h. The cells were then harvested by centrifugation and resuspended in 40 mL of lysis buffer that contained 50 mM Tris-HCl (pH 7.5), 0.5 mM NaCl, 15 mM Imidazole, 10% sucrose, 0.1% Triton X-100, and one crushed protease inhibitor tablet (Roche). All subsequent procedures were performed at 4° C. The cell suspension was lysed by sonication, followed by removal of the cell debris by centrifugation at 14500 rpm for 30 min. The soluble lysate was then passed through a TALON affinity column (Clontech) pre-equilibrated with lysis buffer. The protein was eluted with 40 mL of 50 mM Tris-HCl (pH 7.5), 0.5 mM NaCl, 500 mM Imidazole, 10% sucrose. SUMO-protease Ulp1 (20 units; a kind gift from Dr. Christopher Lima, SKI) was added to the imidazole eluate and the mixture was then dialyzed overnight in 1 L of a buffer containing 50 mM Tris-HCl (pH 8.0), 400 mM NaCl, 5 mM β-mercaptoethanol. The Ulp1-treated dialysate was applied to a second column of TALON beads, and the tag-free PaePE-(1-177) protein was recovered in the flow-through fraction. The flow-through fractions were pooled, concentrated and purified further by gel-filtration through a Superdex 200 column (GE Life Sciences) equilibrated with 50 mM Bis-tris (pH 6.5), 200 mM NaCl, 5 mM DTT. The peak fraction containing pure PaePE-(1-177) uniformly-labeled with ¹⁵N,¹³C,²H was concentrated to 300 μM protein for use in NMR experiments. A similar procedure was followed to obtain ¹⁵N,¹³C-labeled PaePE-(1-177) except that the pre-adaptation D₂O steps were eliminated and the M9 containing ¹³C,¹H-glucose (as the only carbon source) was prepared in 100% H₂O.

We also confirmed the absence of any bivalent metals ions that could have been acquired through the expression and purification process by noting that the ¹⁵N,¹H TROSY spectra acquired in the absence and presence of a 10-fold excess of EDTA in the buffer were identical. Thus all further experiments were carried out in the absence of chelating agents.

NMR spectroscopy

NMR experiments were performed at 25° C using either Bruker Avance (600 MHz or 800 MHz) or Varian Inova (600 MHz) spectrometers equipped with cryogenic probes capable of applying pulsed-field gradients along the z-axis. TROSY-based HNCACB, HN(CO)CACB experiments (600 MHz; 1024, 32 and 64 complex points and sweep-widths of 14 ppm, 32 ppm and 65 ppm in the ¹H, ¹⁵N and ¹³C dimensions respectively), HNCO and HN(CA)CO (600 MHz; 512, 32 and 32 complex points and sweep-widths of 15 ppm, 36.2 ppm and 14 ppm in the ¹H, ¹⁵N and ¹³C dimensions respectively) experiments were used for backbone-directed sequential assignments utilizing uniformly ¹³C,¹⁵N,²H-labeled protein. A pre-scan delay of 1.5 s was used for all experiments. The assignment procedure was facilitated by residue type identification using HCCH-TOCSY (800 MHz; 512, 35 and 35 complex points and sweep-widths of 14 ppm, 64 ppm and 64 ppm in the ¹H, and the two ¹³C dimensions respectively; mixing time 10.8 ms) and (H)C(CO)NH-TOCSY (600 MHz; 512, 32 and 64 complex points and sweep-widths of 15 ppm, 39.5 ppm and 76 ppm in the ¹H, ¹⁵N and ¹³C dimensions respectively; mixing time = 12 ms) experiments on uniformly ¹³C,¹⁵N-labeled PaePE-(1-177). The data were processed with NMRPipe (Delaglio et al., 1995) and analyzed using NMRViewJ (Johnson, 2004). Main-chain and side-chain ¹H assignments were obtained using a H(CCO)NH experiment (600 MHz; 512, 30 and 43 complex points and sweepwidths of 13 ppm, 32 ppm and 7.5 ppm in the direct ¹H, ¹⁵N and indirect ¹H dimensions respectively; mixing time = 18 ms). Additional ¹H assignments were obtained using a ¹⁵N-edited NOESY-HSQC experiment in H₂O (800 MHz; 512, 40 and 128 complex points and sweep-widths of 14 ppm, 32 ppm and 12 ppm in the direct ¹H, ¹⁵N and indirect ¹H dimensions respectively; mixing time = 150 ms) and a ¹³C-edited NOESY-HSQC experiment in D₂O (800 MHz; 512, 38 and 99 complex points and sweep-widths of 14 ppm, 64 ppm and 10.5 ppm in the ¹H, ¹³C and ¹H dimensions respectively; mixing time = 150 ms).

Resonance assignments and data deposition

146 of 164 (excluding the Met residue at the extreme N-terminus) main-chain amide ¹⁵N, ¹H^N (89 %), 153 of 165 C_α (93 %), 139 of 150 C_β (93 %) and 146 of 165 C’ (89 %) resonances could be unambiguously assigned for the non-proline residues. For the 12 prolines, 11 of the 12 C^α, C^β and C’ resonances could be assigned. Further, 60 of 100 (60 %) C^γ, 36 of 66 (55 %) C^δ and 5 of 14 C^ε (36 %) were assigned (including proline but excluding aromatic side-chains). For ¹H resonances, 89 % of H_α, 86 % of H_β, 73 % of H_γ, 76 % of H_δ and 7% of H_ε were assigned (including prolines but excluding aromatic sidechains). The side-chain N_δ, H_δ of the only Asn, 1 of 6 Gln N_ε, H_ε and 2 of 5 Trp side-chain N_ε, H_ε pairs were assigned. 34 % of aromatic protons were assigned. The completeness of resonance assignments was highly satisfactory given the quality of spectra that could be obtained even for this optimal construct (Fig. 1).

Fig. 1.

Assigned ¹⁵N, ¹H TROSY spectrum of uniformly ¹³C,¹⁵N,²H-labeled PaePE-(1-177) in 50 mM Bis-Tris propane (pH 6.5), 200 mM NaCl, 5 mM DTT. Data were acquired at a ¹H frequency of 600 MHz at 25° C using sweep-widths of 15 ppm (512 complex points) in ¹H dimension and 36.2 ppm (128 complex points) in ¹⁵N dimension with a recycle delay of 1.5 s. The crowded central region indicated by the dashed rectangle, is shown in the inset. Assigned backbone ¹⁵N,¹H resonances have been labeled. Sidechain NH groups have not been labeled.

Of the 17 residues for which ¹⁵N, ¹H^N chemical shifts could not be assigned, 10 (Asp162, Gly163, Glu164, Ala165, Ser167, Leu168, Asp169, Asp172, Val173 and Lys175) belonged to the C-terminal region that is dispensable for the enzymatic activity (Nair and Shuman, unpublished results). Incomplete back-exchange of amide protons was unlikely to be the cause of the missing assignments, because identical ¹⁵N-¹H HSQC spectra were recorded for samples prepared in H₂O or in D₂O minimal media. However, it is possible that some of these missing resonances were present at the crowded region of the spectra (see Fig. 1) and could not be sufficiently resolved to allow unambiguous assignment. Of the residues important for catalytic activity (Zhu et al., 2005), assignments for main-chain ¹⁵N, ¹H^N resonances for His42, His48, and Tyr88 could not be obtained. It is likely that these residues are in conformational exchange in the absence of the catalytic Mn⁺² ion that is essential (Zhu and Shuman, 2005) for both the phosphomonoesterase and phosphodiesterase activities. Mn⁺² was not utilized in the present studies because of its paramagnetic nature and resultant line-broadening effects.

Secondary structural elements were identified using the TALOS+ (Shen et al., 2009) software utilizing ¹³C^α, ¹³C^β, ¹³C’ and ¹⁵N chemical shifts after correction of ¹³C^α and ¹³C^β chemical shifts for ²H-isotope effects (Venters et al., 1996). Thus, eight β strands (β₁:34-41, β₂:49-54, β₃:58-63, β₄:76-80, β₅:106-115, β₆:127-133, β₇:141-145 and β₈:154-159) were identified in addition to single α–helix (120-124). These regions of secondary structure were consistent with those seen in the crystal structure (Nair et al., 2010) as shown in Fig. 2.

Fig. 2.

Secondary structure prediction using using TALOS+ (Shen et al., 2009). The confidence values for regions predicted to be β-strand (black) or α-helix (blue) are plotted against residue numbers. Also shown are the predicted order parameters (S²) based on chemical shifts. Residues that show peak doubling (see main text) are indicated by green arrows. Secondary structure elements seen in the crystal structure of PaePE-(30-187) (PDB: 3N9B) are shown (β-strands as black rectangles and α-helices as blue rectangles) and labeled.

We noted another interesting feature upon careful analysis of the ¹⁵N, ¹H^N TROSY spectra - two sets of peaks (a major and a minor peak) corresponding to residues Leu34, Leu55, Gly57, Thr58, Leu59, Trp62, Ile106, Val107 and Trp108 were seen in a freshly prepared sample. Over time, the minor peaks disappeared and only the major peak could be detected, indicating a conformational transition that was very slow on the chemical shift timescale. These residues were located near the N-termini of β₁ and β₅, and the region between β₂ and β₃ (Fig. 2). Inspection of Fig. 2 reveals the confidence in strand prediction by TALOS+ to be quite low at the N-termini of β₁ and β₃. This is consistent with conformational dynamics on the slow to very slow timescale. Additionally, the same analysis seems to suggest a break in β₅ around residue 109 (Fig. 2). This indicates conformational heterogeneity around the edges of one face (β₅-β₁-β₂-β₃) of the β-barrel (Nair et al., 2010).

¹³C chemical shifts are key indicators of the conformation of the peptide backbone about a Xaa-Pro bond. An exhaustive statistical analysis of the differences between ¹³C^β and ¹³C^γ chemical shifts (Δ_βγ) for trans and cis conformations of the Xaa-Pro peptide bond were 4.5 ± 1.2 ppm and 9.6 ± 1.3 ppm, respectively (Schubert et al., 2002), suggesting a significantly larger difference for a cis conformation. For Pro65 and Pro72, the Δ_βγ values were both 8.4 ppm after correcting for ²H isotope shifts (Venters et al., 1996). Using the program Promega (Shen and Bax, 2009), we found that the absolute probabilities of the Val64-Pro65 and Asp71-Pro72 bonds being in the cis conformation were 91% and 86%, respectively. Notably, both these bonds are in the trans conformation in the crystal structure (Nair et al., 2010). All other prolines for which ¹³C^β and ¹³C^γ assignments were available (this excludes Pro68) were all predicted to be in the trans conformation.

The N-terminal 32-aa peptide of PaePE, which is important for the phosphomonoesterase activity, was disordered on the ps-ns timescale with average ¹⁵N-{¹H} NOE values of 0.03 ± 0.28. However, an analysis of the secondary structure propensities using SSP scores (Marsh et al., 2006) revealed significant helical tendencies (>28 %) for residues 6-11 (data not shown). S² values for this region (0.45 ± 0.02) predicted from the Random Coil Index (RCI) (Fig. 2) indicated some degree of order, though a higher degree of order (0.58 ± 0.07) was predicted for residues 14-21, a segment that includes three side chains (Arg14, Asp15 and Glu21) shown to be important for the 3’-phosphomonoesterase activity but not for the 3’-phosphodiesterase activity (Zhu et al., 2005). The ¹⁵N-{¹H} NOE values for this segment were also marginally higher (0.16 ± 0.04) than the rest of the disordered N-terminus.

A detailed analysis of the structural and dynamic features of PaePE-(1-177) in solution and its interactions with substrate DNA will be discussed at length elsewhere (Natarajan et. al., in preparation).

The main-chain ¹H, ¹³C^α, ¹³C’ and ¹⁵N and side-chain ¹³C and ¹H chemical shifts of PaePE-(1-177) have been deposited in the BioMagResBank (http://www.bmrb.wisc.edu) under accession number 17283. Corrections to account for the TROSY and ²H-isotope shifts were not applied to the deposited shifts.