366
Bioconjugate Chem. 2004, 15, 366−372
Optimized Conjugation of a Fluorescent Label to Proteins via Intein-Mediated Activation and Ligation Robert J. Wood, David D. Pascoe, Zoe¨ K. Brown, Emma M. Medlicott, Marco Kriek, Cameron Neylon, and Peter L. Roach* School of Chemistry, University of Southampton, Highfield, Southampton, Hampshire, U.K. SO17 1BJ. Received September 19, 2003
Intein-mediated ligation provides a site-specific method for the attachment of molecular probes to proteins. The method is inherently flexible with regard to either the protein sequence or the attached probe, but practical difficulties have limited the widespread use of this valuable labeling system for the attachment of small- to medium-sized molecules. We report herein studies to improve the efficiency and practical application of these reactions, including the assembly of plasmids for the expression of target-intein fusion proteins and the analysis of their reaction with a fluorescent cysteine derivative under a range of conditions. Optimal ligation of the fluorophore to the target protein is critically dependent on the degree of oxidation of the fluorescent cysteine derivative. Efficient ligation has been achieved with freshly prepared fluorescent cysteine derivative under rigorously anaerobic conditions. Similar ligation yields have also been achieved using more practically convenient conditions including anaerobic reaction with addition of thiophenol, or aerobic reaction with the further addition of tricarboxyethylphosphine.
INTRODUCTION
Studies of a wide range of proteins and enzymes could potentially benefit from the site-specific attachment of molecular probes to the protein of interest. Recently, intein-mediated activation of the C-terminus of a protein domain has been shown to provide a general method for attaching a wide range of molecules, including proteins (1), peptides (2), fluorescent labels (3), carbohydrates (4), oligonucleotides (5), and affinity tags and metal chelators (6). Intein-mediated ligation has also been used for the assembly of selectively labeled proteins for NMR studies (7, 8) and to prepare cyclic peptides and proteins (9, 10). Intein-mediated labeling reactions require two protein domains: the N-terminal domain which contains the protein sequence to be labeled (the target domain) and a C-terminal domain containing the intein (the intein domain). These domains are usually joined in a single fusion protein containing both the intein and target domains [a target-intein fusion protein Figure 1, 1]. For those intein domains beginning with an N-terminal cysteine residue, the molecular basis of intein-mediated activation is a spontaneous reversible N f S transacylation step forming thioester (2) that occurs at this cysteine residue (Figure 1, 1 f 2). The ligation of probes to target domains has been achieved using a range of cysteine derivatives (4) under appropriate conditions (2). Although the thioester (2) is more reactive toward nucleophiles than a backbone amide, direct reaction with peptides containing an N-terminal cysteine residue was unsuccessful, a result that has been rationalized in terms of steric hindrance from the bulky intein domain (2). The use of several thiols including thiophenol [195 mM (2)], ethanethiol [0.49 M (11)], or MESA1 [30 mM (6)] has been reported to enhance intein-mediated ligation reactions, presumably by trans-thioesterification to the less hin* Author for correspondence. Telephone: +44 (0)23 80595919, fax: +44 (0)23 80596805. E-mail:
[email protected].
dered intermediate (3). Under these conditions, cysteine derivatives (4) are thought to form the kinetic product (5), then rearrange to the thermodynamically more stable amide (6) (2, 12). Another potential problem with intein-mediated ligation chemistry is the possible oxidation of both the activating thiol reagent and the modified cysteine derivative to disulfides. The use of thiol reducing agents such as DTT is not possible, as these compounds effectively cleave and derivatize the target domain. This problem has been circumvented by the use of the water soluble phosphine TCEP (3), which efficiently reduces disulfides to thiols (13). Since our initial attempts to utilize intein-mediated ligation for the fluorescent labeling of proteins gave disappointing and variable yields of labeled protein, we set out to identify the factors that are important in achieving efficient and consistent labeling. Herein we report the efficient expression and purification of targetintein fusion proteins and studies to determine the optimal conditions for ligation reactions. EXPERIMENTAL PROCEDURES
Cloning and expression vectors pBAD TOPO and pBAD/HisA and TOP10 competent cells were obtained from Invitrogen (Groningen, NL). The plasmid pTYB1 from the IMPACT-CN system was obtained from New England Biolabs (Herts., UK). DNA was isolated from cultures or agarose gels using commercially available kits following the manufacturers’ instructions. Fluorescein isothiocyanate (isomer I, 90%), Supelclean LC-18 reverse phase columns (60 mL), and HPLC solvents (Riedelde Hae¨n, chromasolv HPLC grade) were obtained from 1 Abbreviations: MESA, 2-mercaptoethanesulfonic acid; FTEC, N-[2-[[[(fluorescein-5-yl)amino]thioxomethyl]amino]ethyl]-L-cysteinamide; TCEP, tricarboxyethylphosphine; CDI, carbonyldiimidazole; TFA, trifluoroacetic acid.
10.1021/bc0341728 CCC: $27.50 © 2004 American Chemical Society Published on Web 01/30/2004
Intein-Mediated Fluorescent Labeling of Proteins
Bioconjugate Chem., Vol. 15, No. 2, 2004 367
Figure 1. The mechanism of intein-mediated protein labeling. Acyl transfer of the target-intein fusion protein (1) from the R amino group to the thiol of the reactive cysteine yields thioester (2), which is attacked by a small molecule thiol to give the less hindered thioester (3). Attack of a cysteine based labeling reagent (4) yields thioester (5), which rearranges to give the thermodynamic amide product (6).
Sigma-Aldrich (Poole, UK). NMR spectra were recorded on Brucker AC300 or AC400 spectrometers. Electrospray ionization mass spectra were recorded on a Micromass Platform 2 single quadrupole MS, HP1050, in positive mode from methanol solutions. IR spectra were recorded on a BioRad FT-IR. N-Boc-S-Trityl-N-2-aminoethyl-L-cysteinamide (8). N-Boc-S-trityl-L-cysteine [1.0 g, 2.16 mmol, 7] was dissolved in DMF (10 mL), and solid carbonyl diimidazole (420 mg, 2.6 mmol) was added. After 30 min, ethylenediamine (1.4 mL, 21.6 mmol) was added in a single portion and the reaction stirred at room temperature for a further 2 h. The DMF was then removed in vacuo and the residue dissolved in dichloromethane (50 mL). The solution was washed with water (5 × 50 mL) and saturated brine (50 mL) and dried with anhydrous magnesium sulfate. The solvent was reduced to ∼10 mL in vacuo, yielding 8 as a white crystalline solid which was collected by filtration and dried in vacuo (980 mg, 90%). Mp 76-80 °C; 1H NMR (300 MHz, CDCl3): δ 7.29 (6H, d, J ) 8.1 Hz), 7.21 (9H, m), 6.65 (1H, t, J ) 5.2 Hz), 4.89 (1H, d, J ) 7.4 Hz), 3.85 (1H, dd, J ) 12.5, 6.6 Hz), 3.25 (2H, ddd, J ) 11.4, 4.7, 1.5 Hz), 2.78 (2H, t, J ) 5.88 Hz), 2.71 (1H, m), 2.54 (1H, dd, J ) 12.5, 5.15 Hz), 1.42 (9H, s). ESMS 506.3 (MH+). N-Boc-S-Trityl-N-[2-[[[(fluorescein-5-yl)amino]thioxomethyl]amino]ethyl]-L-cysteinamide (10). Compound 8 (500 mg, 0.99 mmol) was dissolved in methanol (20 mL) and a solution of fluorescein isothiocyanate [424 mg, 1.09 mmol, 9] in methanol (30 mL) added dropwise. The reaction was stirred under nitrogen for 20 h. The solvent was removed in vacuo to yield an orange residue which was redissolved in 80% MeOH/20% water (20 mL). The solution was applied to a Supelclean LC-18 reverse phase column (60 mL) that had been preequilibrated in 80% MeOH/20% water and the compound eluted in the same solvent. The purest fractions were combined and the solvent removed in vacuo to afford the slightly unstable 10 as an orange solid (815 mg, 92% yield). Mp 184-190 °C; IR (neat) 3256, 2918, 2849, 1667, 1592, 1487, 1443, 1249, 1158, 840 cm-1; 1H NMR (400 MHz, CD3OD), δ 8.0 (1H, s), 7.58 (1H, d, J ) 8.0 Hz), 7.3-7.0 (15H, m), 6.97 (1H, d, J ) 8.5 Hz), 6.66 (2H, d, J ) 9 Hz), 6.6 (2H, d, J ) 2 Hz), 6.44 (2H, dd, J ) 8.8, 2.0 Hz), 3.88 (1H, m), 3.6-3.8 (2H, m), 3.37 (2H, t, J ) 5.15 Hz), 2.48 (2H, d, J ) 6.6 Hz), 1.31 (9H, s); 13C NMR (100 MHz, CD3OD) δ 182.8, 174.0, 173.3, 171.2, 157.0, 154.7, 146.0, 145.6, 141.5, 130.4, 129.5, 128.7, 128.6, 127.7, 127.6,
127.4, 126.4, 120.9, 114.8, 112.0, 103.3, 80.7, 67.7, 64.1, 55.0, 45.0, 39.9, 35.0, 28.4; ESMS 917.5 (MNa+), 895.5 (MH+). N-[2-[[[(Fluorescein-5-yl)amino]thioxomethyl]amino]ethyl]-L-cysteinamide (11, FTEC). Water (100 µL), triisopropylsilane (92 µL, 0.448 mmol), and TFA (4 mL) were added to 10 (100 mg, 0.112 mmol) under nitrogen. The reaction mixture was stirred at room temperature for 30 min. The solvent was removed in vacuo to afford a crude oily orange residue that was triturated with diethyl ether (5 mL). The resultant orange solid was collected by filtration and washed with a further portion of diethyl ether (10 mL) to afford 11 as an orange solid (55 mg, 89% yield). Mp 190-195 °C; IR (neat) 3263, 3061, 2933, 1665, 1585, 1535, 1175, 1115 cm-1; 1H NMR (400 MHz, CD3OD) δ 8.21 (1H, d, J ) 2.0 Hz), 7.79 (1H, dd, J ) 8.1, 2.0 Hz), 7.20 (1H, d, J ) 8.1 Hz), 6.80 (2H, d, J ) 8.1 Hz), 6.77 (2H, d, J ) 2.0 Hz), 6.62 (2H, dd, J ) 8.1, 2.0 Hz), 4.07 (1H, dd, J ) 6.6, 5.1 Hz), 3.95-3.80 (2H, m), 3.62-3.49 (2H, m), 3.11-3.03 (2H, m); 13C NMR (100 MHz, CD3OD) δ 182.4, 169.9, 167.8, 161.5, 153.8, 147.0, 141.2, 130.9, 129.7, 128.3, 125.3, 120.0, 113.3, 111.0, 102.5, 65.9, 55.2, 43.8, 39.3, 25.3; ESMS 553.2 (MH+). HPLC Analysis of FTEC. The extent to which FTEC had oxidized to its disulfide was assessed by reverse phase HPLC using a Gilson System Workcenter and a Phenomenex Prodigy 5 µm ODS-2 (150 × 4.6 mm) column. HPLC Buffer A was 0.1% TFA in water; HPLC buffer B was 0.1% TFA in acetonitrile. The following conditions were used during the separation with linear gradients between time points: flow rate 0.8 mL/min; detector wavelength λ ) 480 nm; detector wavelength 2, 280 nm; t ) 0 min, 95% A, 5% B; t ) 5.0 min, 95% A, 5% B; t ) 40.0 min, 10% A, 90% B. The following retention times were recorded: FTEC, 19.7 min; FTEC disulfide, 21.3 min. Construction of Plasmids for the Expression of Target-Intein Fusions. DNA manipulations were carried out using standard protocols (14). The desired genes were amplified by PCR from E. coli, Sulfolobus sulfotaricus genomic DNA, or from the plasmid pTYBI (New England Biolabs, Hertfordshire) as required, using Pfu Turbo polymerase and oligonucleotide primers as follows: pfbioB 5′-ccatggctcaccgcccacgc; prbioB 5′-ggatccttactcgagtaatgctgccgcgttgtaatattcg; pfint 5′-ctcgagggctgctttgccaagggt; print 5′-gaattctcagtggtggtggtggtggtgttgaagctgccacaaggcaggaacgt; pffpr 5′-ccatggctgattgggtaacaggc; prfpr 5′-ttagccctcgagccagtaatgctccgctgt;
368 Bioconjugate Chem., Vol. 15, No. 2, 2004
pflipA 5′-ccatggagaaaaaagttcaaattgc; prlipA 5′-ggatcctcactcgagactatgattcttagcattttcaatcgcc. DNA encoding the VMAI intein-chitin binding protein fusion was amplified from pTYBI using primers pfint and print which add a XhoI site (italics) and a glycine codon (bold) to the 5′ end of the amplified gene and DNA encoding a His6 tag (bold) and an EcoRI site (italics) to the 3′ end. The PCR amplification product was purified and inserted into pBAD TOPO according to the manufacturer’s instructions, and the resultant plasmids were transformed into competent TOP10 cells. Ampicillin resistant colonies were screened by PCR for plasmids containing the correct insert and plasmid DNA isolated. The XhoI-EcoRI fragment containing the modified intein gene was inserted between the unique XhoI and EcoRI sites of pBAD/HisA to yield plasmid pRJW2960/82. The bioB gene was amplified from E. coli genomic DNA using primers pfbioB and prbioB which added a 5′ NcoI site (italics), removed the native stop codon, and added a 3′ XhoI site (italics) to the wild-type DNA sequence. The PCR product was then inserted into pBAD TOPO, the resultant plasmid transformed into TOP10 cells, and plasmid DNA containing the desired insert isolated. The NcoI-XhoI fragment containing bioB was ligated to the large NcoI-XhoI fragment of pRJW2960/82 to give plasmid pRJW2960/89. Plasmids pRJW2960/88 and pRJW3104/40 were constructed in similar fashion from pRJW2960/82 by inserting the E. coli fpr and Sulfolobus sulfataricus lipA sequences amplified from genomic DNA. Primers pffpr and prfpr were used to construct pRJW2960/ 88 for expression of Fpr-intein fusion protein and primers pflipA and prlipA were used to construct pRJW3104/40 for expression of LipA-intein fusion protein. The overall design of the constructs is shown in Figure 3. Expression and Purification of Target-Intein Fusion Proteins. All purification steps were carried out at 4 °C. Cell pastes and purified proteins were stored at -80 °C. Liquid and solid growth media contained ampicillin (100 mg/L) as required. Protein purity was judged by SDS-PAGE with Coomassie staining.E. coli BL21(DE3) (15) was transformed with plasmid pRJW2960/89, ampicillin resistant transformants were isolated on 2YT agar, and single colonies were grown for strain storage (14). 2YT medium (100 mL) was inoculated from stored strains and grown overnight in a shaking incubator at 37 °C, 220 rpm. This overnight culture was used as 1% innocula into fresh medium (4 × 1250 mL) and grown until the OD600 reached 0.6. The cultures were induced by addition of arabinose (20% w/v filter sterilized solution, 10 mL/ L), and growth was continued at 27 °C for 4 h. Cells were harvested by centrifugation (10000g, 4 °C, 10 min), and the cell paste was stored at -80 °C until required. The BioB-intein fusion protein was purified from cell paste (∼25 g) resuspended in 4 mL/g Lysis Buffer (50 mM Tris/HCl, 10% w/v glycerol, 0.2 mg/mL lysozyme, 50 U/mL benzonase, pH 8.1). The suspension was stirred (30 min) and then sonicated (10 × 30 s bursts) and the lysate cleared by centrifugation (22000g, 4 °C, 30 min). The supernatant was applied to a nickel charged Chelating Sepharose FF column previously equilibrated in Buffer A (50 mM Tris/HCl, 500 mM NaCl, pH 7). The column was washed with Buffer A (200 mL), and the proteins were eluted with Buffer B (Buffer A plus 500 mM imidazole). The purest fractions were pooled, concentrated to 25 mg/mL, and applied to a Superdex 200 column (33 mm ID × 80 cm) previously equilibrated in Buffer A. The column eluate was collected in fractions
Wood et al.
(9 mL), and the purest fractions stored to yield BioBintein (150 mg from ∼25 g of cell paste). The Fpr-intein fusion protein (180 mg from ∼ 95 g cell paste) was purified by a similar method from E. coli strain 821 (16) transformed with pRJW2960/88 except that (i) all buffers were adjusted to pH 8.1; (ii) Buffer A contained imidazole (50 mM); (iii) the gel filtration chromatography used Buffer C (50 mM Tris/HCl, 500 mM NaCl, pH 8.1). The LipA-intein fusion protein (140 mg from ∼50 g of cell paste) was purified from E. coli C43 (DE3) (17) transformed with pRJW3104/40 by the method used for the Fpr-intein fusion protein. General Conditions for Cleavage and Fluorescent Labeling Reactions. Reagents were added to reactions as required to the following final concentrations: BioB-intein fusion protein (0.7 mg/mL), LipAintein fusion protein (0.64 mg/mL), Fpr-intein fusion protein (1.3 mg/mL), thiols (thiophenol, 2-aminobenzenethiol, 2-methoxythiophenol, MESA, or 4-fluorobenzenethiol, 25 mM), TCEP (20 mM), FTEC (2.3 mM), Tris/ HCl buffer (50 mM, pH 8.1). Reactions were stopped by rapid freezing to -80 °C. Protein samples (10 µL) were denatured in Gel Loading Buffer (5 µL) for 5 min at 80 °C (14) and analyzed by SDS-PAGE on 10% gels. Gels were visualized under UV illumination and after Coomassie staining using a Syngene GeneGenius imager. The images were analyzed using GeneTools software (Syngene, Cambridge, UK). Where required a sample of fluorescently labeled BioB was applied to the gel to provide a fluorescence intensity calibration. Where noted, reactions were maintained under anaerobic conditions using a glovebox (Belle Technology, Portesham, UK) maintained at 20 °C under nitrogen at
