2-3 Results - 首都大学東京博士（理学）学位論文（課程博士）

【2-3-1 Preparation and characterization of each RRM domain】

I constructed E. coli expression systems for RRM1 (108–191), RRM2 (188–284), RRM3 (309–412) and RRM4 (397–501), and purified the proteins. I measured ¹H–¹⁵N HSQC spectra for the RRM1–RRM4 proteins, and provided well dispersed signals of RRM1, RRM2 and RRM4. Purification of the RRM3 protein was problematic. The RRM3 protein eluted in the void volume during gel-filtration, indicating the presence of protein aggregates. The buffer using measurement include high level salt (400 mM KCl) because RRM3 in low level salt (100 mM KCl) buffer is aggregated. Therefore, we could not measure RNA binding analysis. The results of SDS-PAGE in the purification are shown below (Fig. 6-13).

Nrd1 RRM1 (108-191)

Fig. 6 The 17.5% SDS-PAGE gel of Nrd1 RRM1 at Superdex 75 (26/60) gel filtration column after cleaving GST tag by HRV3C.

After concentration by Amicon-Ultra 10,000 M.W. cutoff (Millipore), the fusion protein was cleaved with HRV3C overnight at 4ºC. The cleaved sample was purified by chromatography using Superdex 75 (26/60) (GE Healthcare) with gel filtration buffer containing 30 mM HEPES (pH 7.5), 100 mM KCl, 5 mM DTT, 0.1 mM EDTA.

Nrd1 RRM2 (188-284)

Fig. 7 The 17.5% SDS-PAGE gel of Nrd1 RRM2 at Superdex 75 (26/60) gel filtration column after cleaving GST tag by HRV3C.

Nrd1 RRM3 (309-412)

Fig. 8 The 17.5% SDS-PAGE gel of Nrd1 RRM3 at Superdex 75 (26/60) gel filtration column after cleaving GST tag by HRV3C.

Nrd1 RRM4 (397-501)

Fig. 9 The 17.5% SDS-PAGE gel of Nrd1 RRM4 at Superdex 75 (26/60) gel filtration column after cleaving GST tag by HRV3C.

Nrd1 RRM1-2 (108-284)

Fig. 10 The 15% SDS-PAGE gel of Nrd1 RRM1-2 at Superdex 75 (16/60) gel filtration column.

The sample of Nrd1 RRM1-2 using SAXS was purified by chromatography using Superdex 75 (26/60) (GE Healthcare) with gel filtration buffer containing 30 mM HEPES (pH 7.5), 100 mM KCl, 1 mM TCEP, 0.1 mM EDTA, 5% glycerol.

Nrd1 RRM3-4 (309-501)

Fig. 11 The 15% SDS-PAGE gel of Nrd1 RRM3-4 at Superdex 75 (16/60) gel filtration column.

The sample of Nrd1 RRM3-4 using SAXS was purified by chromatography using Superdex 75 (26/60) (GE Healthcare) with gel filtration buffer containing 30 mM HEPES (pH 7.5), 100 mM KCl, 1 mM TCEP, 0.1 mM EDTA, 5% glycerol.

Nrd1 RRM2-3-4 (188-501)

Fig. 12 The 15% SDS-PAGE gel of Nrd1 RRM2-3-4 at Superdex 75 (16/60) gel filtration column.

The sample of Nrd1 RRM2-3-4 using SAXS was purified by chromatography using Superdex 75 (26/60) (GE Healthcare) with gel filtration buffer containing 30 mM HEPES (pH 7.5), 100 mM KCl, 1 mM TCEP, 0.1 mM EDTA, 5% glycerol.

Nrd1 RRM1-2-3-4 (108-501)

Fig. 13 The 12.5% SDS-PAGE gel of Nrd1 RRM1-2-3-4 at Superdex 200 (16/60) gel filtration column.

【2-3-2 RNA binding analysis】

I examined the RNA binding character and RNA binding site of RRM1, RRM2 and RRM4 by chemical shift perturbation (CSP) experiments. Since the interactions were weak, an excess amount of chemically synthesized non-labeled RNA (5-mer; UUCUU) was added to the ¹⁵N-labeled proteins (the molar ratio of the each protein to RNA was 1:4) in the CSP experiments. As a result, changes in the chemical shift of a number of peaks of RRM2 were observed, while signals of RRM1 and RRM4 were not significantly perturbed (Fig. 14). This suggests that RRM2 may play an important role in the interaction between Nrd1 and target mRNA, although the possibility that RRM3 possesses RNA binding activity cannot yet be excluded. The RRM2 domain was then subjected to structural analysis and mapping of the RNA binding site (Fig. 15). The RNA binding site of RRM2 distributed to four -sheets.

Fig. 14 The ¹H-¹⁵N HSQC spectra of RRM1, RRM2, and RRM3-4.

1H-¹⁵N HSQC spectra of the ¹⁵N-labeled RRM1, RRM2, RRM4 and RRM3-4 domains shown on the left, middle and right, respectively. The spectra of RNA-free and 4-fold excess non-labeled UUCUU are shown in black and red, respectively.

Fig. 15 RNA binding site of Nrd1 RRM2 domain.

(Top) Weighted average of ¹H and ¹⁵N chemical shift perturbations of residues in RRM2 induced by Cdc4 mRNA (UUCUU) binding. The dashed line indicates the threshold for mapping.

(Middle, Bottom) Significantly perturbed residues are mapped on the ribbon drawing (Middle) and on the molecular surface model (Bottom) of the RRM2 structure. Residues significantly perturbed are highlighted in red and labeled. C-terminal 7 residues are omitted for clarity.

【2-3-3 Interaction among RRMs】

By comparison of NMR spectra of RRM1 and RRM1-2 tandem region indicated that RRM1 and RRM2 interact with each other (Fig. 16). Similarly, it was indicated that RRM3 interacts with RRM4 (Fig. 16). I also tried to prepare RRM2-3-4 and RRM1-2-3-4, and succeeded in purification of them (Fig. 12, 13). NMR spectra of RRM2-3-4 and RRM1-2-3-4 implied that RRM1-2 has no interaction with RRM3-4 (Fig. 16). I tried to prepare RRM2-3 and RRM1-2-3, but they caused aggregation. This result suggest that RRM3 is unstable when RRM4 does not exist. RRM4 may contributes to stability of RRM3. Similarly, RRM1 alone was found to be not stable well. These data suggested that structure of Nrd1 can be divided into two parts. one is RRM1-2 and the other is RRM3-4.

Fig. 16 The ¹H-¹⁵N HSQC spectra of RRMs.

(Top) Overlay views of RRM1-RRM2 (black), RRM1 (red) and RRM2 (green).

(Middle) Overlay views of RRM3-RRM4 (black), RRM3 (magenta) and RRM4 (blue).

(Bottom) Overlay views of RRM1-2-3-4 (black), RRM1-2 (yellow) and RRM3-4 (blue).

【2-3-4 Structure determination of Nrd1 RRM2】

Purified RRM2 domain exhibited well dispersed signals with relatively narrow line shape in the ¹H–¹⁵N HSQC spectrum as shown (Fig. 14), and the elution volume in the gel-filtration corresponded to a molecular weight of the monomer. These results indicated

that RRM2 exists as a monomer in solution. Almost all ¹H, ¹³C, and ¹⁵N NMR signals were assigned using standard multi-dimensional NMR techniques.

The number of distance restraints derived from NOE was sufficient (more than two thousand), and the ensemble of 20 structures were in excellent agreement with a large body of experimental data (Fig. 17, Table 1). The r.m.s. deviations of the backbone and all heavy atoms of RRM2 are 0.18 and 0.53 Å, respectively, excluding disordered regions.

In particular, the conformation of the N-terminal (residues 188–203) and C-terminal (275–284) regions of RRM2 are not converged. RRM2 comprises an anti-parallel -sheet and four -helices. The -sheet is composed of four -strands, 1 (residues 207–211), 2 (233–238), 3 (242–248) and 4 (272–274). The short a-helix, 1 (196–200), is located distal to the core region of the protein while the other three a-helices, 2 (219–229), 3 (251–263) and 4 (266–269), pack against the -sheet (Fig. 17). Although the first short helix a1 is shown close to the core region of the RRM2 domain in the ribbon model, this part is not converged in the ensemble and is highly mobile in the solution state as indicated

by the ¹H–¹⁵N heteronuclear NOE experiment (Fig. 17).

Fig. 17 Solution structure of Nrd1 RRM2 domain.

(Left) Backbone superposition of the final 20 simulated annealing structures of RRM2.

(Right) Ribbon drawing of a representative structure of the RRM2 domain.

Table 1 NMR Structure Determination Statistics for Nrd1 RRM2

SA SAwater refined

Total number of distance constraints 2087

intra residue 464

short range (|i-j| = 1) 540

middle range (|i-j| = 2,3,4) 434

long range (|i-j| > 4) 605

hydrogen bond constraints 44

Dihedral constraints

φ, ψ, χ1 55, 59, 22

R.m.s. deviations from experimental constraintsc

Distance (Å) 0.024 ±3×10^-4 0.0318±9×10^-4

Angle (°) 0.344±0.01 0.834±0.05

R.m.s. deviations from idealized covalent geometry

Bonds (Å) 0.00135±1×10^-5 0.00440±1×10^-4

Angles (°) 0.284±0.001 0.530±0.02

Impropers (°) 0.166±0.002 1.33±0.06

PROCHECK Ramachandran plotd

Residues in most favoured regions (%) 92.0 93.1 Residues in additional allowed regions (%) 5.7 4.0 Residues in generously allowed regions (%) 2.2 2.9

Residues in disallowed regions (%) 0.0 0.0

Average atomic r.m.s. deviations from the average structured

Back bone (Å) 0.08 0.48

All heavy (Å) 0.18 0.53

aThese statistics comprise the lowest energy ensemble of the 20 structures obtained from 100 starting structures. Structure calculations were performed using CNS version 1.2.

bThe numbers of intermolecular NOE is shown in parentheses.

cNone of these structures exhibited distance violations > 0.5 Å, dihedral angle violations > 5°

dEvaluated for the RRM2 domain residues 204-274

【2-3-5 Structure determination of Nrd1RRM1-2】

2-3-5-1 Structure of Nrd1 RRM1-2 derived NOEs

The 2D ¹H–¹⁵N HSQC spectrum for the ¹³C- and ¹⁵N-labeled RRM1–RRM2 of Nrd1 is shown in Fig. 18, and 77% of the ¹HN and ¹⁵N resonances of backbone amide groups was assigned (Kobayashi et al., in press). For non-labile CHn moieties, 70% and 78% of ¹H and ¹³C resonances were assigned, respectively. The conformational exchange of the linker region possibly reduced the ¹HN signals in the vicinity. Judging from the backbone dihedral angles obtained using the TALOS+ program (Shen et al., 2009) and NOEs obtained from NOESY experiments, we identified six -helices and nine -strands.

According to the topology of the secondary structure, RRM1 presumably adopted a typical RRM fold. Notably, the ¹HN, ¹⁵N resonances of backbone amides of the RRM2 region of RRM1–RRM2 were significantly different from those of RRM2, although the secondary structural elements were well conserved (Kobayashi et al., 2013). The methyl resonances in 17/28 isopropyl groups of Leu and Val residues and 22/129 -methylene groups were assigned in a stereospecific manner. NOE derived structure of Nrd1 RRM1-2 is shown in Fig. 19.

Fig. 18 ¹H-¹⁵N HSQC spectrum of RRM1-2 with assignments.

Spectrum of ¹³C/¹⁵N uniformly labeled Nrd1 RRM1-2 in 93 % H2O/7 % D2O in K-phosphate buffer (pH 6.9), 100 mM KCl, 1 mM TCEP and 0.1 mM EDTA at 303 K. The spectrum was recorded at 600 MHz at ¹H frequency. Backbone assignments are annotated using one-letter amino acid codes and the sequence number.

Fig. 19 NOE derived structure of Nrd1 RRM1-2.

RRM1 fitted, but RRM2 is not converged. This caused by lacking of enough NOE.

2-3-5-2 RDC

RDC provides orientation information of the magnetic dipole-dipole interaction vector in a common reference frame. The measurement of RDC requires an anisotropic orientation via direct or indirect magnetic field orientation of the protein in solution. (Chen, et al., 2012). ¹H-¹⁵N residual dipolar couplings were measured using a DSSE experiment (Cordier, et al., 1999). Alignment media consist of 12 mg/ml Pf1 phage. I successfully obtained RDC values to analyze the orientation of RRM1-2 (Fig. 20)

Fig. 20 DSSE spectra of Nrd1 RRM1-2.

Aligned sample was made by using 12 mg/ml Pf1 phage.

2-3-5-3 PCS

I tried to obtain long-range distance information of Nrd1 RRM1-2. I used the chelater DOTA-M8, developed by professor grezesiek group to obtain PCS data (Häussinger, e al., 2009). This chelater has 8 methylene groups in addition to DOTA backbone. Thus rigid and bulky. So single point attachment causes enough PCS. Informative PCS data were successfully obtained monitoring methyl groups (Fig. 21).

Fig. 21 Pseudo contact shifts obtained by using DOTA-M8.

PCS observed in tandem RRM1-2 (monitoring methyl group). The spectra of RRM1-2-DOTA-M8-Lu (diamagnetic) in black and Dy (paramagnetic) in cyan, respectively.

【2-3-6 SAXS experiments】

A series of SAXS data and pseudo atomic model derived from SAXS data by GASBOR (Svergun, et al., 2001) indicated that RRM1-2 and RRM3-4 form compact structure and RRM1-2-3-4 adopts elongated shape consistent with NMR data (Fig. 22, 23). NMR relaxation parameters also consistent that RRM1-2 adopts well packed structure (Fig. 23, 24). The pseudo-atomic model, RRM3 part in RRM1-2-3-4 is exposed to solvent, which is suitable for the RNA binding.

Fig. 22 SAXS experimental data of Nrd1.

Log Plots, Guinier Plots, Kratky plots and pseudo atomic models of Nrd1 RRM1-2 (Top), RRM3-4 (Middle) and RRM1-2-3-4 (Bottom) derived from SAXS data.

Fig. 23 Pseudo atomic model of Nrd1 derived from SAXS.

Pseudo atomic models of Nrd1 RRM1-2 (Top), RRM3-4 (Middle) and RRM1-2-3-4 (Bottom) derived from SAXS data. The models were generated by GASBOR based on the SAXS data.

Fig. 24 T1/T2 of ¹⁵N Nrd1 RRM1-2.

The rotational correlation time (τc) of Nrd1 RRM1-2 was estimated from ¹⁵N relaxation to be 12.5 ns; this value is consistent with a 20 kDa protein.

【2-3-7 Reconstruction of Nrd1 using sortase】

Nrd1 RRM1-2 and RRM3-4 were ligated using soratse A. The yield was approximately 30% (Fig. 25).

Fig. 25 SDS-PAGE of sortase-mediated protein ligation of Nrd1.

Protein ligation was carried out at room temperature in 50 mM Tris-HCl (pH 8.0), 150 mM KCl, 10 mM CaCl2, 1 mM DTT, 10% glycerol dialyzed with same buffer.

【2-3-8 Phosphorylation of Nrd1】

To investigate the influence on the Nrd1 structure by the phosphorylation, I prepared RRM1-2 mutants (T126E and T126D). It is changed for negative charge of a glutamic acid or aspartic acid instead of phosphrylation. Comparing the spectrum of RRM1-2 T126E with RRM1-2 wild type or RRM1 wild type, it is different for chemical shifts (Fig.

26). Interestingly, the spectra of phospho-mimic mutant were close to only that of RRM2 wild type (Fig. 26). These experiments implied that the phosphorylation destroys the fold of RRM1. I also measured the spectrum of only RRM1 region mutated T126E. As a result,

NMR spectrum did not exhibit dispersed signals (Fig. 26). Nrd1 mutant mimicked phosphorylation found that interaction between RRM1 and RRM2 are lost in the cause of unfolding of RRM1 induced by phosphorylation of T126 residue.

Fig. 26 The ¹H-¹⁵N HSQC spectra of phospho-mimic mutants.

(Top) Overlay views of Nrd1 RRM1-2 T126E (black) and RRM1-2 WT (masenta), RRM1 WT (red).

(Middle ; Left) Overlay view of Nrd1 RRM1-2 T126E (black) and RRM2 (green).

(Middle ; Right) Overlay view of Nrd1 RRM1-2 T126D (black) and RRM2 (green).

(Bottom) The spectrum of Nrd1 RRM1 T126E.

Fig. 27 The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2 and unlabeled Pek1DD/pPmk1.

(A) The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2

(B) The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2 and unlabeled Pek1DD/pPmk1 on the morrow of addition to Pek1DD/pPmk1

(D) The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2 and unlabeled Pek1DD/pPmk1 at 2 h after addition to Pek1DD/pPmk1

(E) The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2 and unlabeled Pek1DD/pPmk1 at 3 h after addition to Pek1DD/pPmk1

(F) The ¹H-¹⁵N HSQC spectra of ¹⁵N Nrd1 RRM1-2 and unlabeled Pek1DD/pPmk1 at 4 h after addition to Pek1DD/pPmk1

ドキュメント内首都大学東京博士（理学）学位論文（課程博士） (ページ 31-60)