|Home | About | Journals | Submit | Contact Us | Français|
Nuclear Magnetic Resonance (NMR) spectroscopy has emerged as one of the principle techniques of structural biology. It is not only a powerful method for elucidating the 3D structures under near physiological conditions, but also a convenient method for studying protein-ligand interactions and protein dynamics. A major drawback of macromolecular NMR is its size limitation caused by slower tumbling rates and greater complexity of the spectra as size increases. Segmental isotopic labeling allows specific segment(s) within a protein to be selectively examined by NMR thus significantly reducing the spectral complexity for large proteins and allowing a variety of solution-based NMR strategies to be applied. Two related approaches are generally used in the segmental isotopic labeling of proteins: expressed protein ligation and protein trans-splicing. Here we describe the methodology and recent application of expressed protein ligation and protein trans-splicing for NMR structural studies of proteins and protein complexes. We also describe the protocol used in our lab for the segmental isotopic labeling of a 50 kDa protein Csk (C-terminal Src Kinase) using expressed protein ligation methods.
Nuclear Magnetic Resonance (NMR) spectroscopy has emerged as one of the principle techniques of structural biology. It is not only a powerful method for elucidating the 3D structures under near physiological conditions, but also a convenient method for studying protein-ligand interactions as well as protein dynamics. As molecular weight increases, the size limitations of NMR become apparent – a slower tumbling rate reduces resolution, and adds greater complexity of the spectra. Slower tumbling of the macromolecule means a faster transverse relaxation rate resulting in poor signal to noise (s/n) of the spectrum. Larger size implies more signals appearing in the spectrum and this complicates the assignment of individual signals. In recent years, higher magnetic fields NMR machines have been constructed to increase resolution and signal to noise. Cryogenic probes are widely used to increase the signal to noise significantly. The development of multinuclear multidimensional experiments based on sophisticated pulse schemes has greatly enhanced the utility of NMR for the study of macromolecular structure and dynamics (Mittermaier and Kay, 2006; Wuthrich, 2003).
In the sample preparation aspect, the structural investigation of proteins by NMR has heavily depended on the incorporation of stable isotopes by increased (mainly 15N, 13C) or decreased (mainly 1H) occurrence of an NMR-active isotope. The combination of 15N, 13C triple-resonance spectroscopy with deuteration, however, holds the most promise for addressing the molecular weight limitations currently imposed on structure determination by solution NMR (Mittermaier and Kay, 2006). Although using deuterium labeling and TROSY type experiments (Wider, 2005; Wuthrich, 2003) addresses the issue of line width, NMR methods have still most routinely been applied to proteins under approximately 20 kDa to avoid the issues of spectral overlap and broad line widths (Card and Gardner, 2005). The segmental labeling of proteins, in which segments of the sequence with arbitrary isotopic composition are recombined, represents a significant step to reduce the spectra complexity.
The principal applications of segmental labeling (Cowburn and Muir, 2001) may be identified as (i) assignment and structure determination of large proteins and their complexes by reducing their spectra overlap, (ii) formation of labeled proteins from readily producible fragments when the complete product is impractical, or cytotoxic, for expressions and/or labeling, (iii) segmental labeling of a segment known to be a major epitope for interaction with a ligand in a fuller sequence context, (iv) observation of domain orientation and ligand perturbation in large systems using residual dipolar couplings and/or relaxation methods, (v) introduction of non-natural or modified amino acids (such as phosphotyrosine) which cannot be included in normal expression methods, and (vi) introduction of NMR unobservable tags (such as solubility enhancement tags, or lanthanide binding tags). Although there are other methods for making segmental isotopic labeled protein or RNA samples (Kim et al., 2002; Varadan et al., 2004) the main focus in this paper is expressed protein ligation and protein trans-splicing methods.
Expressed protein ligation and protein trans-splicing have been applied to a large number of different protein engineering problems. The successful implementations in segmental isotopic labeling for NMR are still somewhat limited. This may arise from several factors. If the overall aim is the reduction of spectral complexity, several issues must be addressed. First, a project involving segmental labeling will require more time and reagents than a conventional approach. If the target protein is composed of a single domain, normally a refolding procedure is required, no matter whether trans-splicing or the expressed protein ligation method is used. If the target protein contains several independent folding domains, the ligation can be conducted at native conditions; a refolding procedure is potentially not required. Compared with protein-peptide ligation, protein-protein ligation is generally slow. Because of this, purification methods are required remove the unligated precursors (Harris, 2006; Shi and Muir, 2005; Vitali et al., 2006; Zuger and Iwai, 2005).
Expressed protein ligation is a protein engineering approach that allows recombinant and synthetic polypeptides to be chemoselectively and regioselectively joined together (Figure 1A) (Muir et al., 1998). It is a modified version of native chemical ligation (NCL) (Dawson et al., 1994) in which at least one of the building blocks are made by recombinant DNA methods. The first step of this process is the chemoselective transthioesterification of an unprotected protein C-terminal α-thioester with an N-terminal Cys of a second segment. The so-formed thioester spontaneously undergoes an S→N-acyl transfer to form a native peptide bond and the resulting peptide product is obtained. The reaction can proceed in aqueous conditions at neutral pH and internal Cys residues within both peptide segments are permitted. Although this procedure is relatively simple, segmental labeling using expressed protein ligation had not been applied until the introduction of the IMPACT (Intein Mediated Purification with an Affinity Chitin-binding Tag) system (New England Biolabs) (Chong et al., 1997; Southworth et al., 1999). The IMPACT System contains a series of expression vectors (pTXBx, pTYBx, pTWINx, etc) which allow the fusion of the cleavable intein tag to either the C-terminus (C-terminal fusion) or N-terminus (N-terminal fusion) of a target protein (Xu and Evans, 2001). It provides a practical way to generate the reactive C-terminal α-thioester and N-terminal Cys segments. The combination of IMPACT system with NCL provides a powerful method for the segmental labeling of proteins for the NMR studies (Table 1).
For segmental labeling of a protein required folding, a ligation site can be selected within the domain(Ottesen et al., 2008). For a ligation without refolding process, the ligation point is always chosen to be located within the linker region that connects the two domains (Vitali et al., 2006). In both cases, the residues flanking the ligation site are of special significance. One of these is the N-terminal Cys residue of the C-terminal segment, the other is the C-terminal residue forming the α-thioester for the N-terminal portion (Figure 1A). Since the Cys residue at the junction is generally regarded as obligatory for EPL, the primary consideration is to choose a ligation site before a Cys residue. In the absence of Cys at the desired ligation site, mutation usually is introduced at sites having similar residues such as Ala or Ser. In comparing a homologous family of similar proteins, it may be desirable to about mutating highly conserved residues.
Attention should also be paid to the residue at the C-terminus of the α-thioester segment (X position). Efficiency of the protein-intein thiolysis step varies and different inteins have their own set of preferences at this position (Chong et al., 1998). For example, if the target protein is fused with Mxe GyrA intein from pTWIN1 vector, C-terminal Asp, Pro, Gly and Ala should be avoided. If the target protein is fused with Sce VMA intein from pTYB1, pTYB3 vector, C-terminal Asp and Arg should be avoided because of the in vivo cleavage; Asn, Cys and Pro should also be avoided because of the possible block of cleavage. If the Mth RIR1 intein from pTWIN2 vector was fused to the C-terminus of the target protein, it is recommended that a Gly or Ala be present at the C-terminus of the target protein while Asp and Pro are not recommended (Xu and Evans, 2001).
As an additional consideration is that the ligation reaction rate highly relies on the molar concentration of both precursors. For the segmental labeling of protein, the ligation speed is then critical since the molar concentration of both segments cannot be very high. Although all 20 amino acids are potentially suitable for ligation, Val, Ile, and Pro represent bad choices for the α-thioester because of slow ligation rates. Rapid ligation rates are observed when the α-thioester residue X is His, Cys, or Gly (Hackeng et al., 1999).
Although chemical synthetic methods are feasible (e.g. (Romanelli et al., 2004)), the recombinant intein fusion expression method is the most widely used in the generation of C-terminal α-thioester for segmental labeling, because isotopic labeling is more easily obtained, and expression is more feasible for longer segments . Currently, three modified inteins from the IMPACT system can be utilized for the generation of segment C-terminal α-thioester: Mxe GyrA(198 aa, from pTWIN1, pTXB1,3) (Evans et al., 1998), Mth RIR1(134 aa, from pTWIN2) (Evans et al., 1999), and Sce VMA(454 aa, from pTYB1,2,3,4) (Chong and Xu, 1997). Cloning a target gene into these vectors results in the fusion of the C-terminus of the target protein to the N-terminus of an intein, which is mutated to undergo thiol-induced cleavage at its N terminus (Xu and Evans, 2001). As noted above, the C-terminal residue of the target protein is critical because the in vivo and thiol-induced cleavage at the intein can be dramatically affected by it.
Three methods generally used in the preparation of segments with N-terminal Cys are chemical synthesis, cleavage of a precursor sequence expressed with the Cys residue adjacent to a cleavage site, and intein-mediated recombinant methods (Muralidharan and Muir, 2006). Several cleavage methods can be utilized: (1) Methionyl aminopeptidase (Camarero et al., 2001; Gentle et al., 2004; Iwai and Pluckthun, 1999); (2) Factor Xa protease (Camarero et al., 2002; Romanelli et al., 2004; Xu et al., 1999) (3) TEV protease (Tolbert et al., 2005; Tolbert and Wong, 2004); (4) Cyanogen bromide (Macmillan and Arham, 2004); (5)SUMO protease (Weeks et al., 2007); (6)Leader peptidase (Hauser and Ryan, 2007); (7) EnteroKinase (Hosfield and Lu, 1999); (8) Thrombin. Recently we found that the leader sequence LVPRC (modified from the optimal thrombin cleavage sequence LVPRGS) can also be cleaved by thrombin with lower efficiency. Thus thrombin cleavage can also be used in the preparation of protein segments with C-terminal Cys(Liu et al., 2008). Each cleavage method has its own limitations. For example, nonspecific proteolysis can be observed, when there are sequences within the protein similar to the cleavage site; Cyanogen bromide cleavage requires denaturation and no other Met residues in the target protein; Endogenous Methionyl aminopeptidase is often inefficient and results in low yields of the desired material.
N-terminal fusion with an intein can also be exploited to prepare N-terminal Cys segments. The modified intein Ssp DnaB (Mathys et al., 1999), Mth RIR1 (Evans et al., 1999), Mxe GyrA (Southworth et al., 1999) have all been engineered for the generation of N-terminal Cys proteins. pTWIN1 and pTWIN2 vector from IMPACT system can be utilized for this purpose using the Ssp DnaB N-terminal fusion. A target protein sequence was fused in-frame to the C-terminus of the intein. Following expression in E. coli , the fusion protein was bound to a chitin resin and the cleavage of intein fusion is induced in a pH and temperature-dependent fashion. Note that sometimes the protein undergoes significant in vivo cleavage, causing the loss of the affinity tag (Mathys et al., 1999).
The isolated α-thioester and N-terminal Cys segments can be ligated in the presence of many additives such as denaturents, chaotropes, or detergents. The optimum pH of the ligation reaction is 7–8 and the presence of a moderate concentration of the thiol reagent such as MESNA (2-mercaptoethanesulfonic acid) is always required. The thiol-induced cleavage and ligation can also be carried out simultaneously by adding both thiol and N-terminal Cys segment (Camarero et al., 2002; Vitali et al., 2006) (Table 1). A reactive C-terminal thioester can be generated on the target protein when an appropriate thiol compound is used to induce cleavage of the intein. MESNA is an attractive reagent since it provides higher ligation efficiency, and is odorless and extremely soluble in aqueous solutions (Ayers et al., 1999; Evans et al., 1998). Johnson and Kent have compared the use of a number of thiol compounds find a highly effective and practical catalyst MPAA(4-carboxylmethyl)thiophenol) (Johnson and Kent, 2006). Ligation in the presence of MPAA showed an order of magnitude (i.e. ~10-fold) rate increase over the standard catalyst mix used for ligation. The authors also suggested that MPAA will be used as a much more effective alternative to MESNA for use in the semi-synthesis of protein by expressed protein ligation. Recently a study suggests reaction of DTT-thioester protein and Tris(tris(hydroxymethyl) aminomethane) buffer can takes place, and a Tris adduct is formed, which is fused to the C-terminus of the protein by a stable amide bond (Peroza and Freisinger, 2008). Although the reactions of other thioesters/thiol systems with Tris are not reported, caution is appropriate in the use of Tris buffer.
The ligation of proteins segments can also be accomplished by trans-splicing which relies on the high affinity and catalytic activity by the two halves of an intein to ligate the two extein sequence (Figure 1B). The N- and C-terminal pieces individually have no splicing activity but that when combined; associate non-covalently to give a functional protein splicing element. The trans-splicing reaction is initiated by binding of the intein fragments together, where expressed protein ligation reaction lacks this favorable binding. Several split inteins have been studied including native split Ssp DnaE (Evans et al., 2000; Martin et al., 2001), artificially split PI-PfuI (Otomo et al., 1999b; Yagi et al., 2004; Yamazaki et al., 1998), Mtu RecA (Lew et al., 1998; Mills and Perler, 2005; Shingledecker et al., 1998), Psp Pol-1 (Southworth et al., 1998), Ssp DnaB (Brenzel et al., 2006), Sce VMA (Mootz et al., 2003; Schwartz et al., 2007). Among these split inteins, Ssp DnaB, Sce VMA, Ssp DnaE are reported to be active in protein trans-splicing under native conditions, while others require a denaturation-refolding treatment of the mixed intein halves (Otomo et al., 1999a; Otomo et al., 1999b; Yagi et al., 2004; Yamazaki et al., 1998). The naturally occurring Ssp DnaE intein has been shown to mediate protein trans-splicing in vivo under native conditions and provide a segmental isotope method in the living cells (Zuger and Iwai, 2005). It has been recommended to include several residues flanking the target protein sequence to enhance the ligation efficiency (Xu and Evans, 2001). Table 2 summarizes the typical use of protein trans-splicing methods for the segmental labeling for NMR sample preparations.
For extremely large systems, segmental labeling involving more than two domains may be necessary. Both expressed protein ligation (Blaschke et al., 2000; Cotton et al., 1999; Ottesen et al., 2008)and protein trans-splicing (Otomo et al., 1999a; Shi and Muir, 2005) methods can be utilized in this task. Using expressed protein ligation, a protecting group on N-terminal Cys of a central segment should be introduced to avoid self-ligation and/or cyclization when the central segment(s) is/are ligated with the C-terminal segment. The protection group can be a protease cleavable peptide. After the central segment is ligated with the C-terminal segment, the protecting group is removed and the resulted protein ligated with N-terminal segment α-thioester. For the trans-splicing protocol, two separate inteins, which should be orthogonal to each other, may be used to ligate N- and C-terminal segments to both the ends of middle segment.
Csk has been shown to be important in the regulation of neural development, T-cell development and regulation and cyto-skeletal organization (Cole et al., 2003; Roskoski, 2005). Topographically, Csk is similar to the Src family kinase, having a SH3 and SH2 domain pair followed by a catalytic domain. A proline-rich sequence from the protein tyrosine phosphatases (murine PEP, (Cloutier and Veillette, 1996), human PTPN22/Lyp (Bottini et al., 2004)) can binds to the Csk-SH3 domain.Csk is a 50 kDa protein that is not good target for traditional NMR studies because of its size and self association. Segmental labeling is a promising approach to overcoming the size issue. From preliminary studies, we knew the refolding of Csk is very difficult, so we decided use expressed protein ligation do the segmental labeling. We chose pTWIN1 expression vector (New England Biolabs) for the expression of both of the segments based on these considerations with a view to obtaining segmentally labeled, full length Csk. At the same time, this approach provides SH32 and kinase domain separately and conveniently. We also need a fusion system to overcome the poor expression and folding of the kinase domain; pTWIN1 vectors are designed for protein purification or for the isolation of proteins with an N-terminal cysteine and/or a C-terminal thioester. It has a modified Ssp DnaB intein as N-terminal fusion (intein 1) and Mxe GyrA intein as C-terminal fusion (intein 2). The presence of the N- and C- terminal chitin binding domain (CBD) facilitates purification.
Csk can be divided into three domains -- SH3, SH2 and kinase domain (Figure 2). The ligation point is chosen to be located within the SH2-kinase linker region that connects the two domains, residues V172-E194 (Holtrich et al., 1991; Partanen et al., 1991). Since there is no Cys in the linker region, we have to find a similar residue that can be mutated to Cys with minimal change. Ser and Ala are good choices for mutation to Cys. A178, A179 and S186 can be considered as potential mutation sites. A178 and S186 were not used because the proceeding residues, V177 and R185 respectively, would likely result in low efficiency for the ligation (Hackeng et al., 1999). So we chose A179 to mutate to Cys and A178 to form an α-thioester. In the crystal structure of Csk, A178 and A179 are located at the N-terminal of a small helix αBC (Ogawa et al., 2002) and it was anticipated that the mutations would lead to minimal structural perturbations. In the following steps, the SH32 and kinase domain sequence were cloned into the expression vector, which allowed the generation of SH32-intein2-CBD and CBD-intein1-kinase fusion proteins, respectively.
The Csk SH32 domain gene, containing residues M1-A178 was amplified from human full-length Csk gene by polymerase chain reaction (PCR) using primers 5’-GGT GGT CAT ATG TCA GCA ATA CAG GCC (SH3_intein_fw) and 5’- GGT GGT TGC TCT TCC GCA CGC CAC TGT GCC CTC CAT (SH2_intein_rev) with Nde I and Sap I cleavage sites. Note that a stop codon should not be included in the reverse primer for the sequential expression of Mxe GyrA intein (intein2). The Sap I site will be lost after cloning so that no additional residue was introduced between SH32 and intein2. All of the PCR reactions were in 50 µl final volume with pfu polymerase (Novagen), 0.5 mM of all four dNTPs (Sigma), 0.2 µM of each oligonucleotide and 50 ng plasmid DNA. PCR was carried out in a Mastercycler (Eppendorf) machine and DNA was amplified by 30 cycles under standard conditions with annealing at 60°C and enlongation at 72°C. The PCR product was purified, digested with Nde I and Sap I, and ligated to the similarly digested pTWIN1 vector. After transformation of the plasmid into E .coli DH5a, the clone was validated by DNA sequencing using T7 universal primer (5’-TAA TAC GAC TCA CTA TAG GG). The vector was named pTWIN1-SH32. The corresponding protein product was named SH32-intein2-CBD (Figure 3).
The Csk kinase domain gene containing residues A179-L450 with an A179C mutation, with or without C-terminal His-tag, was amplified from full-length Csk gene by PCR using primers containing Sap I and BamH I sites. For the cloning of kinase without His-tag, primers: 5’-GGT GGT TGC TCT TCC AAC TGC CAG GAT GAG TTC TAC CGC (kinase_intein_fw) and 5’-GGT GGT GGA TCC TTA CAG GTG CAG CTC GTG GGT (kinase_intein_rev) were used. For the cloning of kinase with C-terminal His-tag, primers: kinase_intein_fw and 5’- GGT GGT GGA TCC TTA ATG GTG ATG GTG ATG GTG CAG GTG CAG CTC GTG GGT (kinase_intein_his_rev) were used. Note that a stop codon should be included in the reverse primer. The Sap I site is lost after cloning so that no additional residue is introduced between Ssp DnaB intein (intein1) and the kinase domain. The amplified gene was cloned into the expression vector pTWIN1 and the sequences were confirmed by DNA sequencing using Ssp DnaB forward primer (5´-ACT GGG ACT CCA TCG TTT CT) and T7 terminator reverse primer (5’- TAT GCT AGT TAT TGC TCA G). The designed fusion proteins without and with a C-terminal His-tag are named pTWIN1-kinase and pTWIN1-kinase-His, respectively (Figure 3). The corresponding protein products of pTWIN1-kinase and pTWIN1-kinase-his are named CBD-intein1-kinase and CBD-intein1-kinase-his, respectively. Both of the fusion proteins contain A179C mutations.
For the expression of SH32-intein2-CBD protein, the plasmid pTWIN1-SH32 was transformed into BL21-CondonPlus (DE3)-RIL Component Cell (Stratagene, 230245). A fresh cell harboring pTWIN-SH32 was inoculated in 20 ml of LB medium containing 100 µg/ml ampicillin and cultured at 37°C until the optical density (A600) reached 0.8. The whole culture was then transferred to 1 L of the same medium and cultured at 37°C until the optical density (A600) reached 0.6. The cells were induced with 0.5 mM IPTG at 30°C. For the isotopic labeling, M9 media with 0.1% 15NH4Cl and/or 0.2% [U-13C] glucose was used. After induction at 30°C overnight, the cell pellet was collected by centrifugation at 4000 g for 20 min at 4°C, and then resuspended in 35 ml of Buffer A (50 mM Tris-HCl, pH 7.5 200 mM NaCl) and stored at −80°C for future use. No in vivo cleavage of the SH32-intein2-CBD was found during the processing procedure above. As a practical note, in systems that are susceptible to in vivo degradation we recommend that the cell growth and protein purification steps be performed in the same day to improve yields.
Cells containing the SH32-intein2-CBD protein were passed through a French press cell twice to break the cells. The cell lysate was centrifuged at 12,000 g for 20 min at 4°C. The clarified cell extract was loaded to 6 ml of Chitin Beads (New England Biolabs S6651L) which pre-equilibrated with Buffer A. The column was washed with 50 ml of Buffer A to remove the unbound protein. The column bound with protein SH32-intein2-CBD was stored at 4°C for the ligation with kinase domain. The fusion protein was stable for at least four days at 4°C with no thiol reducing agent added. The longtime storage should be at −80 °C. For the purification of the SH32 domain, the cleavage of the intein-tag was induced by equilibrating the chitin beads with 50 mM DTT, 50 mM KPi, pH 7.2. After 24 hr of on-column cleavage at room temperature, the target protein was eluted from the column by 40 ml of Buffer B (20 mM Tris-HCl, pH 8.0) and loaded onto a Mono Q column equilibrated with Buffer B at flow rate 1 ml/min (Figure 4A). The elution was conducted by linear gradient to 60% Buffer C (20 mM Tris-HCl, pH 8.0, 1.0 M NaCl) in 60 min. The major peak near 34% of Buffer C was collected and identified as SH32.
For the expression of CBD-intein1-kinase or CBD-intein1-kinase-his, the plasmid pTWIN1-kinase or pTWIN1-kinase-his was transformed into BL21-CondonPlus (DE3)-RIL Component Cell (Stratagene, 230245). A fresh cell harboring pTWIN-SH32 was inoculated in 20 ml of LB medium containing 100 µg/ml ampicillin and cultured at 37°C until the optical density (A600) reached 0.8. The whole culture was then transferred to 1L of the same medium and cultured at 37°C until the optical density (A600) reached 0.6. The cells were induced with 0.2 mM IPTG at 15°C for 16 hr (for LB or non-deuterated M9 media) or 40 hr (when deuteration applied). After centrifugation at 4,000 g for 20 min, the cells were resuspended in 35 ml of Buffer A and stored at −80°C for future use.
Partial degradation was found for the two fusion protein after the induction. Two bands 60 kDa and 30 kDa appeared on the SDS-PAGE gel corresponding to the fusion and degradation products respectively. The calculated molecular weight for the CBD-intein1-kinase, CBD-intein1, and kinase were 60, 30, and 30 kDa respectively. Since the cell growth in D2O is much slower than the H2O media, the degradation was more severe. This makes it impossible to use chitin beads to purify the kinase protein. We used CBD-intein1-kinase-his protein in the ligation experiments. Although the degradation still remains, both non-degraded fusion-kinase and degraded kinase domain can bind the metal affinity beads.
For the purification of the C-terminal kinase protein, cells containing the CBD-intein1-kinase-his and its degradation products were passed through a French pressure cell twice to break the cells at 4°C. The lysate was centrifuged at 12,000 g for 20 min at 4°C to remove the cell debris. The clarified cell extract was loaded to 10 ml of TALON beads charged with Co2+ and pre-equilibrated with Buffer A. The column was washed with 50 ml of Buffer A to remove the unbound protein whereas the degraded and intein1 fusion kinase-his can bind to the column. The bound protein was eluted by Buffer A with 150 mM imidozole and the total elusion volume was about 10 ml. The purified protein can be stored at 4°C for several weeks. For long time storage, glycerol was added to final 25% (v/v) and stored at −80°C. In the case of kinase used in ligation, no additional purification procedures were required. If the kinase was directly used in the NMR experiments without ligation, additional purification procedures are needed. For the purification of the kinase domain, the pH of elusion protein (mixture of degraded kinase-his and fusion CBD-chitin-kinase-his protein) was adjusted to 7.0 with 1 M Tris-HCl (pH 6.5) buffer and stored at 15°C for 2 days to let the fusion protein degrade. The degraded protein mixture was passed through a chitin column equilibrated with Buffer A. CBD-intein1 bind to the column and the flow through was collected, exchanged to Buffer B and loaded onto a Mono Q column equilibrated with Buffer B at flow rate 1 ml/min. The elution was conducted by linear gradient to 50% Buffer C in 50 min. The peak near 15.5% of Buffer C was collected and identified as kinase-his with an N-terminal Cysteine (Figure 4B).
An easy way to test whether the α-thioester can formed after the MESNA thiolysis is to use a fluorescent peptide. A model peptide NH2-CGRGRGRK[fluorescein]-CONH2 was chemically synthesized by New England Peptide Inc. The observed molecular weight was 1246 Da. This peptide was used to monitor reactivity of SH32 (Ayers et al., 1999; Xu et al., 1999). The peptide was designed to be highly soluble and the fluorescein probe was incorporated to provide a convenient way to monitoring the progress of the ligation reaction by using SDS-PAGE and other methods. In preliminary ligation studies, we investigated whether this peptide could be ligated with Csk SH32. Nearly quantitative ligation of the synthetic peptide to the corresponding protein domain with C-terminal thioester was observed when a large access of peptide was added (molar ratio 1:10) as indicated by SDS-PAGE. Typically, 20 µl of chitin beads with loaded SH32-intein2-CBD was used. To these beads was added a solution of synthetic peptide (1 mg/ml) in Buffer A along with 3% MESNA. The suspension was then gently agitated at room temperature overnight, the supernatant was removed and analysis performed by SDS-PAGE. The ligation of the synthetic peptide to the recombinant protein was evidenced by the production of a highly fluorescent 21-kDa protein band on SDS-PAGE. These studies thus established that the C-terminal thioester was successfully formed during the MESNA thiolysis. After 48 h of storage at 15°C, the SH32 α-thioester failed to ligate with the synthetic peptide.
The kinase mixture (CBD-intein1-kinase-his, kinase-his) from the TALON beads was concentrated to about 1 mM using Amicon Ultra centrifugal filter (Millipore, UFC901024) at 4°C. The kinase protein mixture was flushed to the chitin column bound with SH32-intein2-CBD in the presence of Buffer A. The molar ratio of SH32 α-thioester and N-terminal Cysteine kinase was estimated to be 1:1 from the SDS-PAGE of the solution. 0.5 mM EDTA and 200 mM of MESNA were added to the system and the column was stored at 15°C for 72 hr. During this time, three reactions proceed: thiolysis of SH32-intein2-CBD to form SH32 α-thioester; degradation of the fusion CBD-intein1-kinase-his to form kinase-his; and reaction of SH32 α-thioester with kinase-his with an N-terminal Cysteine. SDS-PAGE was used to monitor the ligation process. The speed of the MESNA induced cleavage is dependent on the concentration of MESNA and the period of incubation. We increased the MESNA to 200 mM to increase the cleavage speed since the Ala at the C-terminal of the SH32 is generally not recommended in the pTWIN1 C-terminal fusion. On-column ligation has been reported to increase the ligation efficiency (Vitali et al., 2006). During 3 days ligation, about 20–25% of the SH32 and kinase protein form the ligated segmental labeled protein.
Because of the low ligation efficiency, removing the unligated protein from the ligated sample is critical. Figure 5 shows the combination of Co2+ column and Q column for the purification procedure. After 72 h ligation, the ligation mixture was eluted from the chitin column using Buffer A. The amount of protein was monitored using the Coomassie Plus-200 protein assay reagent (Pierce, 23238). Elution was continued until no more protein was detected – typically about 30 ml. The eluate was dialysised against 1.0 L Buffer A at 4°C for 4 hr, twice, to completely remove the MESNA and EDTA. The solution was then loaded to 10 ml of TALON beads charged with Co2+ and pre-equilibrated with Buffer A. The column was washed with 50 ml of Buffer A to remove the unbound protein (mainly inteins and SH32). The peptide PEP-3BP1 is used to reduce the self-association of the SH3 domain (Borchert et al., 1994; Ghose et al., 2001). The protein tyrosine phosphatase PEP proline-rich peptide (PEP-3BP1, residues 605–629, SRRTDDEIPPPLPERTPESFIVVEE) was expressed and purified as described (Ghose et al., 2001). The column was washed with another 10 ml Buffer A with 1 mM PEP-3BP1 peptide in it to remove the SH32 domain bind with the ligation product due to the dimerization of SH3 domain. The bound protein (mainly SH32-kinase-his, kinase-his) was eluted by Buffer A with 150 mM imidozole and the total elusion volume was about 10 ml. 40 ml Buffer B was added to the elusion to reduce the salt concentration and the total 50 ml elusion was loaded onto a Mono Q column equilibrated with Buffer B at flow rate 1 ml/min (Figure 4C). The elution was conducted by linear gradient to 60% Buffer C (20 mM Tris-HCl, pH 8.0, 1.0 M NaCl) in 60 min. The major peak near 34% of Buffer C was collected and identified as SH32-kinase-his ligation product.
1H-15N HSQC spectra were acquired on the segmental labeled Csk sample at 25°C, 800 MHz. Figure 6(A) is the 1H-15N HSQC of [15N-SH32]-Kinase, about 160 amide peaks can be identified from this spectrum. Figure 6(B) is 1H-15N HSQC SH32-[15ND-Kinase], about 240 amide peaks can be identified from this spectrum. All NMR experiments were performed on Bruker 800 MHz, US2 spectrometers, which were equipped with triple-resonance cryo-probes. Protein solutions were prepared in the following buffer conditions: 50 mM Tris-HCl (pH 7.5), 10 mM DTT, 1.0 mM EDTA, 0.01% (w/v) NaN3, 5% D2O, 0.1 mM DSS (4,4-dimethyl-4-silapentane-1-sulfonate). All the NMR samples were sealed in the NMR tube under a nitrogen atmosphere. Pulsed-field gradient techniques with a WATERGATE pulse sequence were used for all H2O experiments, resulting in a good suppression of the solvent signal. The 1H chemical shifts were referenced to internal DSS. In the 2D 1H-15N HSQC experiments, 512 complex points were collected in the 1H dimension and 128 complex points in the 15N dimension. The corresponding sweep widths were 14 and 36 ppm in the 1H and 15N dimensions, respectively. 15N dimension were multiplied by a cosine-bell window function and zero-filled to 512 points before Fourier transformation using Topspin 1.4. The 15N chemical shift was referenced indirectly using the 1H/15N frequency ratios of the zero point 0.101329118 (15N) (Live et al., 1984; Wishart et al., 1995).
We are grateful to Dr. Michael Goger, Dr. Kaushik Dutta, and Dr. Shibani Bhattacharya for their assistance in the NMR instrumentation and Professor Tom Muir for advice and discussion. Supported by NIH grants EB001991 and GM47021.