Article Cite This: Biochemistry XXXX, XXX, XXX−XXX
pubs.acs.org/biochemistry
Amino Acid Selective 13C Labeling and 13C Scrambling Profile Analysis of Protein α and Side-Chain Carbons in Escherichia coli Utilized for Protein Nuclear Magnetic Resonance Toshihiko Sugiki,† Kyoko Furuita,† Toshimichi Fujiwara,† and Chojiro Kojima*,†,‡ †
Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan Graduate School of Engineering Science, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama 240-8501, Japan
Downloaded via UNIV OF NEW ENGLAND on June 20, 2018 at 20:45:41 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information *
ABSTRACT: Amino acid selective isotope labeling is an important nuclear magnetic resonance technique, especially for larger proteins, providing strong bases for the unambiguous resonance assignments and information concerning the structure, dynamics, and intermolecular interactions. Amino acid selective 15N labeling suffers from isotope dilution caused by metabolic interconversion of the amino acids, resulting in isotope scrambling within the target protein. Carbonyl 13C atoms experience less isotope scrambling than the main-chain 15N atoms do. However, little is known about the side-chain 13C atoms. Here, the 13C scrambling profiles of the Cα and side-chain carbons were investigated for 15N scrambling-prone amino acids, such as Leu, Ile, Tyr, Phe, Thr, Val, and Ala. The level of isotope scrambling was substantially lower in 13Cα and 13C side-chain labeling than in 15N labeling. We utilized this reduced scrambling-prone character of 13C as a simple and efficient method for amino acid selective 13C labeling using an Escherichia coli cold-shock expression system and high-cell density fermentation. Using this method, the 13C labeling efficiency was >80% for Leu and Ile, ∼60% for Tyr and Phe, ∼50% for Thr, ∼40% for Val, and 30−40% for Ala. 1H−15N heteronuclear single-quantum coherence signals of the 15N scrambling-prone amino acid were also easily filtered using 15N-{13Cα} spin−echo difference experiments. Our method could be applied to the assignment of the 55 kDa protein.
A
mino acid selective isotope (2H, 13C, 15N) labeling is an indispensable technique in modern protein nuclear magnetic resonance (NMR) studies and can assist with signal assignments and the elucidation of intermolecular interactions, dynamics, and structures.1 This technique is critical for the examination of larger proteins, membrane proteins, and intrinsically disordered proteins.1 With increasing numbers and line widths of NMR signals, signal degeneracy is greater and leads to signal overlap, and thus, unambiguous resonance assignments become more problematic. In many cases, these problems can be overcome by utilizing amino acid selective isotope labeling. For amino acid selective 15N labeling, the recombinant protein is overexpressed using Escherichia coli with M9 minimal medium containing the desired 15N-labeled amino acid and the other 19 unlabeled amino acids. For amino acid selective unlabeling, M9 minimal medium containing the desired unlabeled amino acid and 15NH4Cl is used. Consequently, NMR signals from the desired amino acid of the target protein are selectively observed or disappear if appropriate NMR techniques are used.2,3 However, the progress of metabolic reactions in host cells results in isotope dilution of the isotopeenriched amino acids and isotope scrambling to undesired residues. Isotope dilution and scrambling represent major obstacles in amino acid selective 15N labeling. © XXXX American Chemical Society
It is well-known that there are 15N scrambling-prone amino acids (e.g., Leu, Ile, Tyr, Phe, Thr, Val, Ala, Gln, Glu, Asn, and Asp) and amino acids less prone to scrambling (e.g., Lys, Arg, Cys, Met, Ser, Gly, and Trp), although the 15N scrambling between Ser and Gly/Cys is not negligible.3−5 Many 15 -scrambling-prone amino acids, such as Leu, Ile, Tyr, Phe, Thr, and Val, play a key role in protein structure stabilization by forming a hydrophobic core.6,7 Therefore, the development of selective labeling methods for 15N scrambling-prone amino acids is critical for unambiguous resonance assignments and structural studies. For amino acid selective 13C labeling, carbonyl 13C atoms experience less isotope scrambling than α-amino 15N atoms do.8,9 Amino acid selective 13C labeling of polypeptide backbone carbonyl groups is sufficiently accomplished even if 15 N scrambling is severe.8,9 These results suggest the possibility that amino acid selective 13C labeling can be more useful than 15 N labeling, as demonstrated for some 13C-labeled methyl and aromatic groups.10−12 Using “13C/2H-labeled glucose and pyruvate” as 13C sources, comprehensive information about Received: February 14, 2018 Revised: June 2, 2018
A
DOI: 10.1021/acs.biochem.8b00182 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
and the supernatant comprising culture medium was discarded. In an effort to arrest background expression and/or translation of E. coli endogenous genes and to activate the CspA promoter, cell pellets were resuspended on ice with 500 mL of sufficiently prechilled fresh M9 minimal medium containing 14NH4Cl and [U-1H,12C6]glucose as sources of nitrogen and carbon, respectively, and the desired isotope-enriched amino acids (100, 100, 100, 50, 100, 100, and 400 mg/L for Leu, Ile, Tyr, Phe, Thr, Val, and Ala, respectively) as previously reported15−17 with minor modifications. The cell suspension was further incubated on ice for approximately 30 min, and the temperature setting of the jar fermenter was changed from 37 to 15 °C during that interval. Then, the sufficiently chilled cell suspension was transferred into a prechilled jar by decantation, and fermentation was continued at 15 °C for 1 h. At this point, isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to the cell suspension to a final concentration of 0.5 mM, and protein expression was induced by fermentation overnight (approximately 16−20 h). A uniformly 13C- and 15N-labeled sample, which was used to record reference NMR spectra, was also overexpressed by the same procedure substituting 14 NH 4 Cl and [U- 1 H, 12 C 6 ]glucose with 15 NH 4 Cl and [U-1H,13C6]glucose in M9 minimal medium, respectively. Following the completion of fermentation, cells were collected by centrifugation at 3000g for 15 min at 4 °C. Harvested cells were resuspended with lysis buffer [20 mM Tris-HCl and 300 mM NaCl (pH 7.5)] and then disrupted by sonication. The cell debris was eliminated by ultracentrifugation at 35000g for 30 min at 4 °C, and the supernatant was collected. Recombinant GB1 protein was purified from the supernatant by performing affinity and size-exclusion chromatography using 1 mL of Ni-NTA resin and a Superdex 75 (16/600) column, respectively. For the preparation of NMR samples, the concentration of the GB1 protein was adjusted to 1 mM by dissolving it with a solvent containing 20 mM potassium phosphate (pH 6.8), 150 mM NaCl, and 7% D2O. For the expression of uniformly 2H-, 15N-, and 13C-labeled GF14c (1−235), the pGEX 6P-3 expression vector was transformed into E. coli Rosetta (DE3). The protein was expressed in M9 medium, where 15NH4Cl and [13C]glucose were included as the sole nitrogen and carbon sources, respectively, and salts and glucose were dissolved in D2O. For the expression of amino acid selectively 13C- and 15N-labeled GF14c (1−235), the pCold-GST expression vector was transformed into E. coli Rosetta (DE3),18 and the protein was expressed in a manner similar to that used for GB1 as described above. Following precultivation of a single colony of the E. coli transformants in 100 mL of fresh LB medium at 37 °C with overnight shaking (12−16 h), the whole cells were collected by centrifugation at 3500g for 15 min at room temperature and resuspended in 900 mL of fresh LB medium in a 2 L volume of an Erlenmeyer flask, and cultivation with shaking was continued at 37 °C. Approximately 5 h later, the whole cells were collected again by centrifugation at 3500g for 15 min at room temperature and resuspended in 1.5 L of fresh M9 minimal medium in a 3 L Erlenmeyer flask; the medium contained 14NH4Cl and [U-1H,12C6]glucose as the sources of nitrogen and carbon, respectively, and a 19-amino acid mixture that does not contain the desired amino acid. At this time, the OD600 value of the cell suspension of M9 medium would be approximately 0.6−0.8. Then, 37 °C cultivation was continued for a few hours. When the OD600 values of the cell suspension reached approximately 0.8−1.0, its flask and cell suspension
the metabolism of amino acid side chains has been provided.13,14 However, when “13C/15N-labeled amino acids” are used as 13C sources, a comprehensive investigation of the 13 C scrambling profile of the side chains has yet to be determined. Here, 13C and 15N scrambling profiles are comprehensively investigated for seven 15N scrambling-prone amino acids (Leu, Ile, Tyr, Phe, Thr, Val, and Ala), and labeling efficiencies are evaluated. Seven less 15N scrambling-prone amino acids (Lys, Arg, Cys, Met, Ser, Gly, and Trp),3−5 four more scramblingprone amino acids (Gln, Glu, Asn, and Asp), a very expensive amino acid (His), and an amino acid without an amide proton (Pro) are not treated. To develop a highly efficient and costeffective method of amino acid selective isotope labeling, E. coli cultivation procedures were developed for the 15N scramblingprone amino acids Leu, Ile, Tyr, Phe, Thr, Val, and Ala. Two distinct techniques, comprising the use of a cold-shock expression system and high-cell density fermentation (HCDF), are employed without using any specific E. coli strains, such as transaminase-deficient or auxotrophic mutant strains. Furthermore, utilizing the reduced scrambling-prone character of 13C, the 1H−15N resonances of 15N scramblingprone amino acids are selectively observed by a 13Cα spin− echo difference heteronuclear single-quantum coherence (HSQC) pulse sequence. This approach was tested on a 55 kDa protein.
■
MATERIALS AND METHODS All chemicals were purchased from Wako (Osaka, Japan) and Nacalai Tesque (Kyoto, Japan) except for isotope-enriched amino acids, which were purchased from Cambridge Isotope Laboratories (Tewksbury, MA). Sample Preparation. The pCold I expression plasmid encoding His6-tagged GB1 (T2Q mutant, whose amino acid sequence is MNHKVHHHHHHIEGRHMQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE) was used. Its amino acid composition is as follows: six Ala, one Arg, four Asn, five Asp, zero Cys, two Gln, six Glu, five Gly, eight His, two Ile, three Leu, seven Lys, two Met, two Phe, zero Pro, zero Ser, ten Thr, one Trp, three Tyr, and five Val residues. It was transformed into the prototrophic E. coli BL21 (DE3) strain (Merck, Darmstadt, Germany). Induction of recombinant GB1 protein expression, precisely at the phase of incorporation of an isotope-labeled amino acid into the target protein, was performed in a high-cell density cultivation manner at 15 °C for 16 h as described below. Following precultivation of a single colony of E. coli transformants in 10 mL of fresh LB medium at 37 °C with overnight shaking (12−16 h), the whole cell suspension was placed in 100 mL of fresh LB medium in a 300 mL baffled Erlenmeyer flask, and cultivation with shaking was continued at 37 °C. Approximately 4 h later, 50 mL (half the volume of the 100 mL cultivation) of the cell suspension was placed in 500 mL of fresh LB medium in a jar, and cultivation was continued at 37 °C with a BioFlo jar fermenter (Eppendorf, Hamburg, Germany) using an average agitation speed of 450 rpm and constant air feeding until the OD600 value reached 2− 3. With continuous monitoring of the dissolved O 2 concentration, the agitation speed was automatically controlled between 450 and 600 rpm to regulate the dissolved O2 level in the medium to >30%. The maximum OD600 value is dependent on the heterologous gene of interest. Cells were then collected by centrifugation at 3000g for 15 min at room temperature, B
DOI: 10.1021/acs.biochem.8b00182 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
The NMR measurements of GF14c for the backbone assignments were performed using a Bruker AVANCE III HD 600 MHz NMR instrument equipped with a QCI-P CryoProbe. For the backbone sequential assignment, non-uniformly sampled (NUS)-HNCA-TROSY, NUS-HN(CO)CA-TROSY, NUS-HN(CA)CB-TROSY, NUS-HN(COCA)CB-TROSY, HNCO-TROSY, and NUS-HN(CA)CO-TROSY spectra of uniformly 2H-, 13C-, and 15N-labeled GF14c were measured. All the spectra were measured with 512 complex points in the proton dimension. For NUS-HNCA-TROSY and NUSHN(CO)CA-TROSY experiments, 1275 complex points were selected from a matrix of 40 × 100 points for the 15N and 13C dimensions. For the NUS-HN(CA)CB-TROSY experiment, 314 complex points were selected from a matrix of 32 × 48 points for the 15N and 13C dimensions. For the NUS-HN(COCA)CB-TROSY and NUS-HN(CA)COTROSY experiments, 384 complex points were selected from a matrix of 32 × 35 points for the 15N and 13C dimensions. The sampling schedules for the NUS spectra were generated by the Poisson Gap Sampling Method20 using Schedule Generator version 3.0 (http://gwagner.med.harvard.edu/ intranet/hmsIST/gensched_new.html). For the HNCOTROSY experiment, 32 × 48 complex points for the 15N and 13 C dimensions were measured. The spectra acquired with an NUS method were reconstructed with hmsIST21 and processed with NMRpipe.19 The spectra acquired with a uniform sampling method were processed with NMRpipe.19 The 15N-{13Cα} spin−echo difference 1H−15N TROSY spectra were recorded using the pulse scheme shown in Figure S1. The reference and attenuated spectra were measured separately. Carrier positions for 1H, 15N, and 13C′ are the water frequency (4.7 ppm), 118 ppm, and 176 ppm, respectively. The carrier positions for 13Cα are set according to the target amino acid type (58 ppm for Arg, Ile, and Leu and 53 ppm for Ala). Sixty-four and 2048 complex points were acquired with sweep widths of 32 and 16 ppm in the t1(15N) and t2(1H) dimensions, respectively. A total of 512, 640, and 720 scans were acquired for Leu and Ile, Ala, and Arg selectively labeled proteins, respectively. All spectra were processed with TopSpin version 3.5. To investigate the magnetic field dependence of the peak intensities of GF14c, 1H−15N TROSY spectra of uniformly 2 H-, 13C-, and 15N-labeled GF14c were measured with a Bruker AVANCE III HD 600 MHz NMR instrument equipped with a QCI-P CryoProbe and a Bruker AVANCE III 950 MHz NMR instrument equipped with a TCI CryoProbe. The spectra were processed with NMRpipe.19
were promptly chilled in an ice/water mixture and incubated for 30−60 min. During this chilling incubation period, the isotope-enriched desired amino acid was added to the culture medium using the same manner described above for the GB1 protein expression procedure. After the cooling incubation had reached completion, IPTG was added to the cell suspension at a final concentration of 0.5 mM, and GF14c protein expression was induced by shaking (110 rpm of rotational shaking speed) overnight (approximately 16−20 h). Following the completion of the cultivation, whole cells were collected by centrifugation at 3500g for 15 min at 4 °C. The GF14c protein was purified as follows. Cells expressing GF14c were lysed by sonication in buffer A [50 mM Tris-HCl, 300 mM KCl, 0.1 mM EDTA, and 1 mM DTT (pH 8.0)], and the lysate was centrifuged at 35000 rpm for 30 min at 4 °C. The supernatant was loaded onto a glutathione Sepharose 4B resin (GE Healthcare), and GF14c was eluted with buffer B [50 mM Tris-HCl, 300 mM KCl, 0.1 mM EDTA, 1 mM DTT, and 50 mM glutathione (pH 7.5)]. Then, the GST tag was removed using HRV3C protease. Then, the sample was applied to a Superdex 75 (26/60) gel filtration column (GE healthcare) and eluted with buffer C [20 mM KPi, 30 mM KCl, and 1 mM DTT (pH 7.4)]. Finally, the remaining HRV3C protease was removed using glutathione Sepharose 4B resin (GE Healthcare). For NMR measurements, uniformly 2H-, 15 N-, and 13C-labeled GF14c was concentrated to 0.5 mM, and amino acid selectively 13C- and 15N-labeled GF14c was concentrated to 0.6−0.8 mM. NMR Experiments. All NMR spectra were recorded on a Bruker AVANCE III HD 600 MHz instrument equipped with room-temperature TXI and cryogenic QCI probes at 25 °C. Resonances of 1H, 13C, and 15N nuclei of the GB1 protein were assigned by measuring conventional two- or three-dimensional NMR spectra: comprising 1H−15N HSQC, 1H−13C CTHSQC, 1H−1H TOCSY, 1H−1H NOESY, HNCACB, HN(CO)CACB, H(CCCO)NH, (H)CC(CO)NH, HC(C)HCOSY, HC(C)H-TOCSY, 13C-edited NOESY-HSQC, and 15 N-edited NOESY-HSQC using uniform 13C- and 15N-labeled GB1. The 1H−13C CT-HSQC spectra were collected at 1024 data points in the direct observation dimension and 256 increments with a 26.6 ms constant-time period (T = 1/1Jcc) during the 13C refocusing delay. All NMR spectra were processed using NMRPipe19 or TopSpin version 3.2 (Bruker). Signal intensities of the two-dimensional (2D) 1H−15N HSQC and 2D 1H−13C CT-HSQC spectra were estimated using Sparky (Goddard and Kneller, SPARKY 3, University of California, San Francisco). The 15N-{13Cα} spin−echo difference 1H−15N HSQC spectra were measured by modifying the pulse sequence 15N{13Cγ} spin−echo difference 1H−15N HSQC spectra (Bruker pulse program hsqcetfpf3gpjcsi) as follows. The G3-shaped pulse for 13Cγ selective excitation and the 13Cα selectively exciting shaped pulse of the pulse program hsqcetfpf3gpjcsi were changed so that their center frequency comprised a 13Cα selective G3-shaped pulse and a 13Cγ selectively exciting shaped pulse. The carrier frequency of the shaped pulse of 13Cγ was set to 127 ppm, and that of 13Cα was adjusted to the individual amino acid selective isotope-labeled samples because the chemical shift ranges of 13Cα differ depending on the target amino acids. The spin−echo delay δ was changed from 1/ (43JNCγ) to 1/(41JNCα) by defining the mean 1JNCα value as −11 Hz.
■
RESULTS
Amino Acid Selective 13C and 15N Labeling. Amino acid selective 15N labeling using E. coli expression systems is very useful in protein NMR analysis, although two serious concerns remain. One involves isotope dilution with isotope scrambling for many amino acids, and the other is the cost for limited amino acids such as Trp and His. For amino acid selective 13C labeling, the labeling efficiency is not known very well, and the cost is high. To improve the labeling efficiency and lower costs, a pCold expression system and HCDF technique were employed, respectively. The former allowed a prolonged induction period of ∼16−20 h for amino acid selective isotope labeling, while the latter improved the yield of the overexpressed heterologous proteins by approximately 4− C
DOI: 10.1021/acs.biochem.8b00182 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 1. 1H−13C CT-HSQC spectra of (A) Leu, (B) Ile, (C) Tyr, (D) Phe, (E) Thr, (F) Val, and (G) Ala selectively 13C- and 15N-labeled GB1 protein. Red and blue colors indicate the positive and negative signal signs, respectively. (B) The asterisk indicates the 1H−13C correlation signal of the methyl group of the methionine residue. (C and D) The insets show the 1H−13C HSQC spectra of the aromatic region. D
DOI: 10.1021/acs.biochem.8b00182 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 2. continued
E
DOI: 10.1021/acs.biochem.8b00182 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 2. 13C labeling efficiency of (A) Leu, (B) Ile, (C) Tyr, (D) Phe, (E) Thr, (F) Val, and (G) Ala selectively 13C- and 15N-labeled GB1 protein shown by each amino acid type. The 13C labeling efficiency is estimated by the signal intensity ratio of selective amino acid 13C- and 15N-labeled samples to uniform 13C- and 15N-labeled samples. Asterisks indicate the signals not observed in the uniformly 13C- and 15N-labeled samples. The error bars are calculated from the root-mean-square deviation of the signal-to-noise ratio for each peak.
(CT)-HSQC spectra of 13C- and 15N-labeled proteins, following a published procedure,9 and used for the scrambling analysis. GB1 protein was used as a model protein because of its small size and the highly quantitative nature of the 1H−13C CT-HSQC spectrum,23 although the 13C labeling ratio can be overestimated by ≤12% [(1/cos{2πJ[13Cα(i − 1) − 13Cα(i)] × 0.0266}) × (1/cos{2πJ[13Cα(i) − 13Cα(i + 1)] × 0.0266}) ≤ 1.12] due to the inter-residual 13Cα−13Cα J coupling (
