An efficient protocol for incorporation of an unnatural amino acid in perdeuterated recombinant proteins using glucose-based media

Abstract

The in vivo incorporation of unnatural amino acids into proteins is a well-established technique requiring an orthogonal tRNA/aminoacyl-tRNA synthase pair specific for the unnatural amino acid that is incorporated at a position encoded by a TAG amber codon. Although this technology provides unique opportunities to engineer protein structures, poor protein yields are usually obtained in deuterated media, hampering its application in the protein NMR field. Here, we describe a novel protocol for incorporating unnatural amino acids into fully deuterated proteins using glucose-based media (which are relevant to the production, for example, of amino acid-specific methyl-labeled proteins used in the study of large molecular weight systems). The method consists of pre-induction of the pEVOL plasmid encoding the tRNA/aminoacyl-tRNA synthetase pair in a rich, H₂O-based medium prior to exchanging the culture into a D₂O-based medium. Our protocol results in high level of isotopic incorporation (~95%) and retains the high expression level of the target protein observed in Luria-Bertani medium.

In vivo incorporation of unnatural amino acids into proteins (Wang et al., 2001; Liu et al., 2007) has significantly expanded frontiers in protein engineering. To date, over 50 unnatural amino acids have been genetically encoded (Liu and Schultz, 2006; Xie and Schultz, 2006; Brustad et al., 2008; Brustad et al., 2008; Cellitti et al., 2008), permitting the manipulation of physicochemical, biological, and pharmacological properties of proteins with exquisite control over structure.

The advantages of creating proteins with unnatural amino acids for NMR studies have been recently discussed (Jones et al., 2010). For example, the introduction of unnatural amino acids bearing post-translational modifications (Liu and Schultz, 2006; Xie et al., 2007; Guo et al., 2008) can be particularly important for structural and dynamic investigations of signaling pathways. In addition, site-specific incorporation of NMR active (¹H, ¹³C, ¹⁵N and ¹⁹F) unnatural amino acids into silent backgrounds dramatically reduces spectral complexity, thereby facilitating the study of large proteins by in-vitro and in-vivo NMR techniques (Cellitti et al., 2008; Jones et al., 2010; Lampe et al., 2010; Li et al., 2010; Watt et al., 2011). Particularly attractive are methods that employ unnatural amino acids to design binding sites for paramagnetic metal ions (Lee and Schultz, 2008; Lee et al., 2009) or nitroxide spin-labels (Fleissner et al., 2009). Indeed, incorporation of unnatural amino acids having bio-orthogonal chemical functionality in the context of the native protein can facilitate production of site-specific, paramagnetically-labeled samples, critical for the characterization of transient, sparsely-populated states (Clore and Iwahara, 2009) and very valuable for resonance assignment (Pintacuda et al., 2004; Venditti et al., 2011) in large proteins and protein complexes.

Incorporation of unnatural amino acids is accomplished by co-expressing a bio-orthogonal tRNA/aminoacyl-tRNA synthetase pair specific for the relevant unnatural amino acid together with the protein of interest whose DNA sequence has been modified by substitution of a TAG amber codon at the site of desired incorporation (Wang et al., 2001). However, lower protein yields are generally obtained by this method owing to translation termination. Thus, to reduce truncation problems, an optimized E. coli incorporation system consisting of a plasmid that overexpresses the aminoacyl-tRNA synthetase at the time of induction has been developed (Cellitti et al., 2008), and the effects of the degree, timing, and temperature of induction have been extensively tested (Young et al., 2010). These studies demonstrated that the simultaneous overexpression of the aminoacyl-tRNA synthetase and the target protein at 30°C dramatically reduces truncation problems, enabling production of sufficient quantities of protein for NMR studies using minimal or rich media (Cellitti et al., 2008; Young et al., 2010). More recently, the expression of the protein FAS-TE incorporating the unnatural amino acid p-methoxy-phenylalanine (p-OMePhe) in deuterated M9 medium supplemented with 0.5% d-glycerol and 0.1 mM FeCl₃ has been also reported (Jones et al., 2010). However, the yield of the protein was modest (~25% of the level obtained in rich medium) and, to the best of our knowledge, no other successful incorporation of unnatural amino acids into deuterated proteins have been reported to date.

Here, we investigate the expression levels for a mutant (C575A) of the C-terminal domain (A261-C575) of the E. coli enzyme I (EIC) of the bacterial phosphotransfer system (PTS), incorporating the unnatural amino acid p-acetyl-L-phenylalanine (pAcF) at position M469 in protonated and deuterated media (Table 1). pAcF is particularly relevant to the NMR community since it reacts quantitatively with a new hydroxylamine functionalized nitroxide spin-label, known as HO-4120 (Fleissner et al., 2009), enabling production of site-directed, paramagnetically-labeled proteins even in the presence of surface exposed Cys residues which interfere with standard site-directed spin-labeling techniques. We focused our attention on glucose-based deuterated media since these are routinely employed for the production of amino acid-specific, methyl labeled proteins (Tugarinov et al., 2006) that are extremely useful in NMR studies of high molecular weight systems (Mittermaier and Kay, 2006).

Table 1.

Composition of the various culture media used in the present work.^a

Medium	Composition (per 1 L)
LB	10.0 g bacto-tryptone, 5.0 g yeast extract, 5.0 g NaCl, 200 μl 5 M NaOH
M9H2O	6.5 g Na2HPO4, 3.0 g KH2PO4, 0.5 g NaCl, 1.0 g 15NH4Cl, 2.0 g D-glucose, 120 mg MgSO4, 11 mg CaCl2, 10 ml 100X MEM vitamins mix (VWR)
M9D2O	6.5 g Na2HPO4, 3.0 g KH2PO4, 0.5 g NaCl, 1.0 g 15NH4Cl, 2.0 g D-glucose-1,2,3,4,5,6,6-d7, 120 mg MgSO4, 11 mg CaCl2, 10 ml 100X MEM vitamins mix (VWR)
M9ISO	6.5 g Na2HPO4, 3.0 g KH2PO4, 0.5 g NaCl, 1.0 g 15NH4Cl, 2.0 g D-glucose-1,2,3,4,5,6,6-d7, 120 mg MgSO4, 11 mg CaCl2, 10 ml 100X MEM vitamins mix (VWR), 1.0 g 15N-2H IsoGro (Isotec)

^aThe LB and M9^H2O media were prepared in water, while M9^D2O and M9^ISO media were prepared in 99.9% D₂O. The MEM vitamin mix used in the deuterated media was lyophilized and resuspended in 99.99% D₂O before usage.

We initially followed the optimized protocol for incorporation of pAcF into proteins using a protonated medium (Young et al., 2010). Briefly, BL21 star (DE3) E. coli cells were co-transformed with pET11a/EIC (containing the desired TAG mutant) and the pEVOL plasmid that contains the orthogonal tRNA and aminoacyl-tRNA synthetase specific for pAcF (Young et al., 2010). Transformed bacteria were then plated onto an LB-agar plate containing ampicillin (100 μg/ml) and chloramphenicol (34 μg/ml) for selection. A single-colony was used to inoculate 50 ml of Luria-Bertani (LB) medium containing the aforementioned antibiotics and grown overnight at 37°C. The following day the starter culture was diluted to A₆₀₀ ~0.05 in 50 ml of each medium listed in Table 1 and grown at 37°C to A₆₀₀ ~ 0.6. pAcF was then added to each culture to a final concentration of 2 mM. After 1 hour, the cultures were moved to 30°C and the pET and pEVOL plasmids were induced with 1 mM isopropyl- -D-1-thiogalactopyranoside (IPTG) and 0.02% L-arabinose, respectively. After 12 hours, 1 ml of each culture was harvested by centrifugation, resuspended in 300 μL of B-PER (Pierce) and the soluble fraction was checked by SDS-page.

Thick overexpression bands are clearly observed for both the full length and truncated proteins in LB medium (Fig. 1A). In contrast, only a small amount of truncated EIC with no detectable band for the full-length protein was obtained from cultures in protonated and deuterated minimal media (Figs. 1B–D). This finding cannot be ascribed to an intrinsically low expression of EIC in minimal media since ~ 15 mg of U-[²H-¹⁵N] wild type EIC was obtained from 50 ml cultures in M9^D2O (table 1). An alternative explanation may be the reduced expression of the tRNA/aminoacyl-tRNA synthetase pair which is under control of the ara promoter and could be largely inhibited at the high glucose concentration used in the tested minimal media. To test this hypothesis the starter culture was diluted to A₆₀₀ ~ 0.05 in 50 ml of LB medium and grown at 37°C to an A₆₀₀ ~ 0.4. The pEVOL plasmid was then induced with 0.02% L-arabinose for 2 hours, while still in protonated, LB medium. Bacteria were collected by gentle centrifugation (170 g for 25 min) and resuspended in 50 ml of two different recipes of deuterated media, M9^D2O and M9^ISO (Table 1), to a final A₆₀₀ of ~0.8. Subsequently, 0.5 g/l pAcF (final concentration 2 mM) and 0.02% L-arabinose were added to each culture. After 1 hour of incubation at 30°C the expression of EIC was induced with 1 mM IPTG and the two cultures were shaken at 30°C for 12 hours. In stark contrast to the near zero yield afforded by the standard protocol, a large amount of full-length protein was obtained using the latter protocol, with yields comparable to the one obtained in LB medium (Figs. 1E,F and Table 2). These data clearly indicate that pre-inducing the pEVOL plasmid in LB medium is extremely beneficial for protein expression.

Fig. 1.

Expression of EIC M469pAcF/C575A in different media. Lanes A, B, C and D correspond to expression in LB, M9^H2O, M9^D2O and M9^ISO media (see Table 1), respectively. Lanes E and F correspond to expression performed by pre-inducing the pEVOL plasmid in LB medium prior to moving the culture to the deuterated M9^D2O and M9^ISO media, respectively. The left-hand most lane displays standard molecular weight markers.

Table 2.

Final optical density (A₆₀₀) and yield of EIC M469pAcF/C575A measured for the 50 ml test cultures. LB-M9^D2O and LB-M9^ISO indicate cultures pre-induced in LB medium with L-arabinose and then moved to deuterated media (see text).

Medium	Final A600	Yield (mg)a
LB	~2.4	~9.1
M9H2O	~2.0	-
M9D2O	~1.5	-
M9ISO	~1.6	-
LB-M9D2O	~2.0	~8.4
LB-M9ISO	~2.2	~10.2

^aMeasured for purified EIC M469pAcF/C575A (residues 261–575) using an extinction coefficient ε₂₈₀ = 19.94 mM⁻¹ cm⁻¹. In all cases conversion to mg units was carried out by using the theoretical mass for the EIC domain at natural abundance (35,206 Da).

The degree of isotopic incorporation in the proposed protocol was tested by liquid chromatography/mass spectrometry on purified EIC M469pAcF/C575A. The results are summarized in Table 3, demonstrating that high levels of isotopic incorporation (~95%) are obtained in both the deuterated media tested.

Table 3.

LC-MS analysis of the natural abundance and U-[²H,¹⁵N]-labeled EIC M469pAcF/C575A samples. Data were acquired under denaturing conditions with a liquid chromatography (Zorbax 300SB-C3 column) mass spectrometry electrospray ionization time-of-flight (LC-MS ESI-TOF) instrument (HP 1100 HPLC/MSD). Protein samples (6 μM final concentration) were diluted in water and acidified with 10% acetic acid. Mass spectra were recorded in the positive ion mode (mass range 500–2000) and results were analyzed using the HP ChemStation software. In the last row (WT-M9^D2O) LC-MS data for U-[²H,¹⁵N]-labeled wild-type (WT) EIC expressed in the M9^D2O medium are reported.

Medium	Theoretical mass (Da)a	Observed mass (Da)	Theoretical number of isotopic substitutionsb	Observed number of isotopic substitutionsc	Isotopic incorporation (%)
LB	35,206	35,205	-	-	-
LB-M9D2O	37,551	36,428	2,345	2,223	94.8
LB-M9ISO	37,551	37,441	2,345	2,236	95.4
WT- M9D2O	37,547	37,530	2,368d	2,351d	99.3

^aTheoretical masses were calculated assuming 100% isotopic incorporation and complete back exchange to ¹H at labile hydrogen positions.

^bThis is the expected number of ¹⁵N and ²H atoms incorporated into EIC and is calculated by taking the difference between the theoretical mass in LB and deuterated media.

^cThis is the observed number of ¹⁵N and ²H atoms incorporated into EIC and is calculated by taking the difference between the observed mass in LB and deuterated media.

^dCalculated using the theoretical mass in LB for the natural abundance wild-type EIC (35,179 Da).

As a final quality test on the produced protein, a comparison of the ¹H-¹⁵N TROSY spectra of the wild type and mutant proteins is reported in Fig. 2. The two spectra show excellent overlap (Fig. 2A), differing only in cross-peaks assigned to residues in the vicinity of the introduced mutation (Fig. 2B), indicating that the introduction of the unnatural amino acid pAcF at position 469 does not alter the overall folding of EIC.

Fig 2.

(A) 600MHz ¹H-¹⁵N TROSY spectra of wild-type (blue) and M469pAcF/C575A (red) EIC. The NMR spectra were acquired at 37°C. Sample conditions were as follow: 400 μM EIC, 20 mM Tris, pH 7.4, 100 mM NaCl, 4 mM MgCl₂, 1 mM EDTA. Under these conditions EIC is present as a 70 kDa dimer (Patel et al., 2006). Resonance assignment for cross-peaks showing large chemical shift differences in the two EIC constructs is reported on the wild-type spectrum. (B) X-ray structure of the wild-type EIC dimer (PDB code 2HWG) (Teplyakov et al., 2006). The monomeric subunits are colored in cyan and orange. Residues whose resonances show large ¹H_N/¹⁵N chemical shift differences (>0.1 ppm) as a consequence of the M469pAcF and C575A mutations are indicated as blue spheres. M469 is shown as yellow sticks.

In conclusion, we have presented a new protocol for incorporating the unnatural amino acid pAcF into deuterated EIC using the recently developed pEVOL vector (Young et al., 2010). We have shown that pre-inducing the expression of the tRNA/aminoacyl-tRNA synthetase pair in LB medium prior to switching the culture to a deuterated, glucose-based medium results in protein yields comparable to that obtained in rich medium, with no significant effects on the isotopic incorporation level. Similar results were also obtained for an EIC construct incorporating pAcF at position A494, as well as for the full-length EI (subunit molecular weight of 64 kDa) incorporating pAcF at position M469 (Fig. 3), suggesting that our procedure is of general applicability. We therefore fully expect the overall protocol presented here to be applicable to other unnatural amino acids and protein constructs. It is worth noting, however, that the reduced expression levels obtaining with standard protocols is probably related to the high glucose content of the tested minimal media. Thus, pre-induction of the pEVOL plasmid in LB media may not be needed when using other carbon sources.

Fig 3.

Expression of full-length (residues 1-575) EI M469 pAcF/C575A (left panel) and EIC A494pAcF/C575A (right panel) in different media. Lanes A, B and C correspond to expression in LB, M9^H2O and M9^ISO media (see Table 1), respectively. Lane D corresponds to expression performed by pre-inducing the pEVOL plasmid in LB medium prior to moving the culture to the deuterated M9^ISO medium. The left-hand most lane in each panel displays standard molecular weight markers.