Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Biosyst. Author manuscript; available in PMC 2010 June 1.
Published in final edited form as:
Published online 2009 April 21. doi:  10.1039/b903144h
PMCID: PMC2790073

Metal Ions Binding to RecA Inteins from Mycobacterium tuberculosis


Zinc has been found in the crystal structure of inteins and zinc ion can inhibit intein splicing both in vitro and in vivo. The interactions between metal ions and three minimized recA inteins have been studied in this work. Isothermal titration calorimetry (ITC) results show that the zinc binding affinity to three inteins is in the order of ΔI-SM > ΔΔIhh-SM ~ΔΔIhh-CM, but much weaker than to EDTA. These data explain the reversible inhibition and the presence of zinc only in the crystal structure of ΔI-SM of recA intein. A positive correlation between binding constants and inhibition efficiency was observed on the titration of different metal ions. Single-site binding mode was detected in all interactions, except ΔΔIhh-CM which has two Zn sites. Zinc binding sites on ΔΔIhh-CM were analyzed by NMR and ITC titration on inteins with chemical modifications. Results indicate that the Cys1 and His73 are the second zinc binding sites in ΔΔIhh-CM. CD study shows the metal coordinations have negligible influence on protein structure. This work suggests that the mobility restriction of key residues from metal coordination is likely the key cause of metal inhibition on intein splicing.


Self-catalytic protein splicing is a post translational process in which the intervening protein, intein, is cleaved from the precursor proteins with the concomitant ligation of flanking sequences.14 Since its discovery in 1990,5 intein has been widely applied in biotechnology and provides a novel tool in protein engineering.3, 6 Inteins are composed of 134 to 608 amino acids, and they are found in all three domains of life: Eukaryotes, Bacteria, and Archaea.7 Over 400 inteins have been identified to date;8 most inteins possess two functional domains, the homing endonuclease domain and the splicing domain. The homing endonuclease domain, which recongnizes and acts on host genes, normally inserts into the splicing domain between two sub-domains.8 These two domains are functionally independent, and evidences from minimized intein indicate that the splicing function is not prevented by the removal of homing endonuclease domain.9, 10 Thus, such minimized inteins, composed of solely splicing domain, are typically employed for studying the mechanism of protein splicing.

Zinc ions were found to interact with residues in the catalytic core in several crystal structures of intein.1113 Splicing activity studies showed that zinc reversibly inhibited protein splicing, while EDTA can relieve such inhibtion.14, 15 In addition, several other divalent metal ions, such as Ni2+ and Co2+, demonstrated weaker inhibitory activity than Zn2+on protein splicing.15 Nearly no inhibition was detected for Ca2+ and Mg2+.14, 15 These observations indicate the inhibition efficiency is highly dependent on the property of metal ions. Thus it is of great interest to explore the mechanism of metal-intein interaction and their correlation with inhibition of protein splicing.

The three intein mutants studied in this paper, ΔI-SM, ΔΔIhh-SM and ΔΔIhh-CM, have been derived from the Mycobacterium tuberculosis recA intein. The wide-type recA intein consists of 440 amino acid residues. The deletion of central endonuclease domain created a minimized intein with 168 amino acid residues (ΔI).16, 17 This minimized intein has lower splicing activity than the full length protein. A V67L mutation could enhance the activity of minimized intein and result in a splicing mutant ΔI-SM. (see Scheme 1) Additionally, the D422G mutation could promote C-terminal cleavage activity with enhanced pH dependence, which resulted in the cleavage mutant (ΔI-CM). 16, 17 It was observed in the crystal structure of ΔI-SM that residues in the linker region between two sub-domains were disordered.13 By replacing the 36 residues in the disordered region with a seven-amino-acid β-turn sequence (VRDVETG) from hedgehog proteins, a further minimized intein ΔΔIhh-SM was created. This 139-residue intein has comparable splicing activity to that of ΔI-SM.17 ΔΔIhh-CM, generated in a similar way with 139-residues, also retains the cleavage properties of ΔI-CM (see scheme 1).

Scheme 1
Protein sequences of three minimized intein used in this work,ΔI-SM, ΔΔIhh-SM and ΔΔIhh-CM. The bold letters denote the conserved residues in the active core, Cys1, His73, Asp422, His429, H439 and Asn440. The two ...

The zinc coordination in the crystal structure of ΔI-SM is shown in Fig. 1.13 Two zinc atoms are located at the interface of two protein molecules and each zinc binds to four residues from two different proteins, two His and one Glu from one protein, and one His from the other protein. Such zinc coordination induces a pseudo-dimer with 1:1 stoichiometry for the ΔI-SM mutant in the asymmetric unit. (Fig. 1). Three of the binding residues are highly conserved in intein sequences, C-terminal residue (Asn440), penultimate histidine (His439) and F-block histidine (His429). Interestingly, zinc ions were present in the crystal structure of ΔI-SM, but not in the crystal structure of ΔΔIhh-SM and ΔΔIhh-CM.

Fig. 1
Zinc coordination in the crystal structure of ΔI-SM (pdb: 2IMZ). (A) Ribbon representation of the protein. Two protein molecules are shown in different color and zinc atoms are shown in blue spheres. (B) Zn binding core in the crystal structure. ...

Isothermal titration calorimetry (ITC) is an efficient technique in the determination of thermodynamic parameters of ligand binding to biomolecules.18 The measurement detects the heat released or absorbed during the binding process, so that the enthalpy change (ΔH) and binding affinity (K) can be obtained directly, from which the free energy change (ΔG) and entropy change (ΔS) can be derived. This technique has also been used for studying the interaction of metal ions to proteins, which could quantify the stoichiometry and coordination constants of each binding step.19

Interactions of five metal ions Zn2+, Ni2+, Co2+, Mn2+, and Cd2+ to three intein mutants have been studied in this work using ITC. The binding sites were analyzed by NMR and the ITC titration on the chemically modified proteins. The coordination constants were measured, and the correlation between binding affinities and the different inhibition efficiencies of metal ions was observed. Data suggested that the metal coordination to residues in the active core is the key cause of the inhibition on protein splicing, probably due to the immobilization of key residues.


Thermodynamic Studies of Zinc Binding to Inteins

The interactions of zinc ion with three intein mutants were measured by ITC at pH 7.4 in 20 mM buffer of Tris-HCl or MOPS. The representative ITC data in Fig. 2 demonstrate the isotherms of Zn binding to three inteins. While similar profiles are seen for the interactions of ΔI-SM and ΔΔIhh-SM (Fig. 2A and 2B), ΔΔIhh-CM shows a rather different titration pattern (Fig 2C). The least-square fitting of titration data resulted in a single-site binding model for ΔI-SM and ΔΔIhh-SM, however, the titration to ΔΔIhh-CM was fit with a two-site binding model (Table 1). This result indicates that an additional zinc binding site appears in the cleavage intein mutant (CM) relative to the splicing mutant (SM). The binding stoichiometry was confirmed by the measurements from equilibrium dialysis. The ratio of zinc/intein was detected as 1:1 for ΔΔIhh-SM and 2:1 for ΔΔIhh-CM in the equilibrated proteins (ESI, Table S1).

Fig. 2
ITC data for the titration of Zn2+ into intein mutants (A) ΔI-SM, (B)ΔΔIhh-SM, (C) ΔΔIhh-CM. In each panel, the upper portions contain the baseline corrected raw data, and the lower portions indicate the concentration ...
Table 1
Binding constants of Zn2+ to three intein mutants measured by ITC titration at 25 °C.

Transition metal ions typically coordinate to buffers and form metal-buffer complexes.19 The dissociation of metal ions from buffer complexes occurs while the metal ions bind to protein. Therefore, the buffer coordination to metal ions contributes to the overall enthalpy change (ΔHITC) and entropy change (ΔSITC) measured from the ITC titrations. The weak coordination of metal to buffer (Kbuffer) has been measured by ITC titrations of metal ions into buffer. The metal binding constant to proteins is calculated as Kbind = KITC×Kbuffer, and the dissociation constant Kd = 1/Kbind.

Table 1 summarized the best-fit binding constants from Zn2+ titration to inteins. Titrations were carried out in two different buffers for avoiding possible buffer-specific effects. Results clearly show that the order of binding affinity for Zn2+ is ΔI-SM >ΔΔIhh-SM ~ΔΔIhh-CM, with the dissociation constants of 0.29 nM, 3.1 nM and 6.8 nM in Tris buffer, respectively. Comparable binding constants are observed in MOPS buffer. The binding affinity of Zn2+ to ΔI-SM is one order of magnitude higher than to ΔΔIhh-SM or ΔΔIhh-CM, which could explain the unique presence of zinc atoms in the crystal structure of ΔI-SM but not in ΔΔIhh-SM and ΔΔIhh-CM.13

The values of enthalpy change (ΔH) and entropy change (ΔS) indicate different driving forces for zinc binding among three intein mutants. (ESI Fig. S1) The zinc binding to ΔI-SM gives large negative ΔH, indicating the enthalpy drive in the reaction. Whereas the entropy change is more important in the zinc binding to ΔΔIhh-SM and ΔΔIhh-CM. Data suggest that the unstructured loop region in ΔI-SM has a negative contribution to the overall entropy change during the zinc binding, compromising the entropy increase relative to the zinc binding to ΔΔIhh-SM and ΔΔIhh-CM. This contribution is compensated by the ΔH in the reaction of ΔI-SM, resulting in the largest value of ΔG (negative) in the zinc titration.

Mapping Zinc Binding Sites in Inteins

Although Zn coordination has been revealed in the crystal structure of ΔI-SM, crystal packing forces may alter the zinc coordination mode from solution. In addition, zinc binding was detected in solution to ΔΔIhh-SM and ΔΔIhh-CM, but not in the crystal structures. The zinc binding between two different molecules in ΔI-SM causes protein dimerization, while ΔΔIhh-SM and ΔΔIhh-CM are in monomeric conformation without zinc coordination in the crystal structures. To map the zinc binding sites in recA intein mutants in solution, ITC measurements were carried out on inteins with chemical modification on histidine and cysteine residues. NMR was also performed to provide site specific binding information.

Diethyl pyrocarbonate (DEPC) was used for histidine modification and iodoacetamide was used for cysteine modification.20 Data in Table 1 shows that the histidine modification significantly decreased the zinc binding affinity to ΔI-SM (Kd from 0.20 nM to 55 nM), while the interference by the cysteine modification is much less (Kd = 3.0 nM) (see Table 1). This result indicates that histidines are critical for Zn coordination while cysteine plays a less important role. This observation is consistent with the zinc coordination mode in the crystal structure of ΔI-SM. (Fig. 1) Although cysteine is not directly involved in zinc binding in the crystal structure, Cys1 is close to the C-terminal residues (His439 and Asn440),13 so that the cysteine modification may alter the local structure for zinc binding.

Two-dimensional (2D) 1H 15N HSQC NMR spectroscopy was employed to further identify the zinc binding sites. The complete assignment of NMR spectra has been reported previously.21 While the majority of signals of intein were not affected in the presence of zinc ions, several peaks was considerably reduced by zinc binding (labeled in Fig. 3). Consistent with the coordination in the crystal structure, the zinc binding to Glu424, His429, His439 and Asn440 significantly affects the signal of Val425, T430, Asn440 amides and Asn440 sidechain. In addition to the decrease of these four signals, the His73 NεH peak was reduced considerably by the second zinc binding to ΔΔIhh-CM. (see the discussion below).

Fig. 3
Two-dimensional 1H 15N HSQC NMR spectra for determining the zinc binding sites in the intein ΔΔIhh-CM. Overlay of the spectra of zinc bound intein (blue) on the apo-intein (red). Zinc binding reduces the NMR signal from residues near the ...

Structural analysis indicates that the zinc binding to the four residues in intein (Glu424, His429, His439 and Asn440) must cause protein dimerization. Therefore, the protein size could provide an additional confirmation for the zinc binding to these four residues. Dynamic light scattering (DLS) detections were carried out to determine the protein size in solution. The measurements were performed on inteins in the absence and in the presence of 500 μM Zn2+. The data were analyzed using Dynamics V6.2 software and the apparent molecular masses were calculated based on hydrodynamic radii (Rh) detected. (ESI Table S2) Results show that all three intein mutants become dimers in the presence of zinc, suggesting the zinc coordination structure in the crystal structure of ΔI-SM remains in solution for all three intein mutants.

Other Divalent Metal Binding to Inteins

It has been reported that several divalent metal can inhibit protein splicing in addition to zinc, though with lower inhibitory efficiencies. Table 2 shows binding constants measured for four other metal ions to three mutants. (Mg2+ was not considered in this work due to its low capability of coordination to the relevant residues). Here the KITC is the relevant binding constants for the comparison because the inhibition assay has been performed in the buffer.15 The KITC of five metal ions to ΔI-SM roughly follow the order of Zn2+ [dbl greater-than sign] Cd2+ > Ni2+ > Co2+ > Mn2+, which correlates with the inhibitory efficiency15 The inhibition could result from either the metal coordination to the key residues (as observed on zinc binding), or the metal induced protein conformation change. CD spectra were measured for all reactions between five metal ions and three inteins. Results exclude any significant change of protein secondary structure upon metal binding (data not shown). Therefore, inhibition of protein splicing is likely caused by metal coordination to active site residues instead of structural changes of protein.

Table 2
Binding constants measured by ITC titration in Tris buffer at 25 °C.

Binding site analysis has shown that histidine residues are crucial for metal binding to inteins. Based on the HSAB principle (Hard Soft Acid-Base), this binding site is obviously unfavorable for hard metal ions, such as Mg2+ and Ca2+. Therefore, low inhibition effect of these metal ions can be expected. Indeed, the activity study showed that Mg2+ had very low inhibition efficiency on the recA intein (~ 4 % at 2 mM).15 Crystal structure detected zinc binding to DnaE intein at a rather soft pocket (to Cys, His and Glu). Accordingly, no significant inhibition of Mg2+ and Ca2+ was observed on DnaE intein.14 These results further confirm the correlation between binding affinity and intein inhibitions.


The effect of zinc on intein activity the has been studied in vivo and in vitro.14, 15 Results showed that both cis- and trans-splicing protein splicing of recA intein were almost completely inhibited by 2 mM ZnCl2 in vitro.15 By using the E coli containing a mutated ALS gene, the protein splicing was assayed in vivo, and the inhibition of Ssp DnaE intein by ZnCl2 was detected.14 Both studies demonstrated that the inhibitory effect by zinc ion could be reversed by the metal chelating agent EDTA.

In the crystal structure of minimized recA intein, zinc was detected in ΔI-SM. However, no zinc was added during the protein preparation and crystallization. The treatment of EDTA prior to crystallization did not result in a zinc-free crystal structure.13 In the mean time, zinc was not detected in the other two crystal structures of ΔΔIhh-CM and ΔΔIhh-SM with the same preparation and crystallization methods.

The thermodynamics measurements from ITC titration in this work indicated that the zinc binding affinity to ΔI-SM is over one order of magnitude higher than to ΔΔIhh-SM or ΔΔIhh-CM (Kd of 0.29 nM relative to 3.1 nM and 6.8 nM), which accounts for the unique presence of zinc in the crystal structure of ΔI-SM. However, this binding constant is far smaller than that of Zn-EDTA (log K ~ 16.5). These data indicate that the binding affinity of zinc to intein is reasonably high for non-metallo-proteins, but is still much weaker than to EDTA, which explains the reversible inhibition by zinc.

Based on the coordination constants measured in this work, the treatment of EDTA in the intein purification should remove Zn2+ in the protein. Indeed, we measured the zinc concentration in proteins which have been treated by EDTA followed by the dialysis. No zinc was detected by atomic absorption spectroscopy (AAS). (ESI Table S1) The origin of zinc in the crystal was discussed in the literature.13 Based on our AAS results and the binding constants comparison, it can be clarified that the presence of zinc in the crystal of ΔI-SM was from contaminations.

Zinc has been detected in several intein crystal structures. However, structural comparison has showed large variation around the zinc binding sites among inteins13, which may arise from the substantial sequence difference among the inteins studied. For the recA intein ΔI-SM, zinc coordinates to Glu424, His429, the penultimate histidine and the C-terminal residue in the crystal structure. This is different from the zinc coordination in two inteins from PI-SceI VMA and dnaE, which have the thiol coordination from C-extein C+1.11, 12 The only cysteine residue in recA intein, Cys1, does not coordinate to zinc in the crystal structure of ΔI-SM. Consistent with this coordination structure, ITC measurements on the chemically modified intein indicated that histidines are critical for Zn binding to ΔI-SM, while cysteine modification affects the zinc binding to a much less extent. (Table 1)

Although zinc was not present in the crystal structures of ΔΔIhh-SM and ΔΔIhh-CM due to the weak binding affinity, the interaction in solution indicates the zinc coordination also caused protein dimerization of these two inteins, based on the DSL measurements. This observation implies that the binding sites in ΔI-SM very likely remain in the two further minimized inteins (although an additional zinc binding site in ΔΔIhh-CM). This suggestion was confirmed by 2D 1H-15N HSQC NMR spectra, which provide detailed information of each residue in solution.

While most titration data can be fit with a single-site binding mode, two-site binding mode was obtained for fitting the data from zinc titration to ΔΔIhh-CM (Fig. 2C). We propose the additional binding site is at Cys1 thiol in ΔΔIhh-CM. In ΔΔIhh-SM and ΔI-SM, Cys1 coordination is prevented by a hydrogen bond between Cys1 thiol and carboxylate group of D422 (supported by unpublished NMR results). This hydrogen bond does not exist in ΔΔIhh-CM due to the D422G mutation, making the Cys1 thiol group more accessible for zinc coordination. To confirm this idea, we carried out ITC measurement in ΔΔIhh-CM with cysteine modification. The Cys1 is the only cysteine in recA intein. The restoration to one binding site with cysteine modification in ΔΔIhh-CM provides the direct evidence of the second zinc binding site at Cys1, whereas no such effect was observed on ΔΔIhh-SM under the same condition (only minor change of Kd from 6.4 nM to 7.5 nM, see Table 1). The His73 imidazole ring is only a few Angstrom away from Cys1 thiol group in the structure, which could form a chelation for the zinc coordination. 2D 1H 15N HSQC NMR spectra of ΔΔIhh-CM indicate that the His73 NεH signal becomes significantly weaker with zinc binding, which confirms His73 coordination to zinc (Fig. 3). Although Cys1 amide is not detectable on the HSQC map, Leu2 signal decreases ~50% due to Cys1 binding.

The Zn binding affinity of ΔΔIhh-SM is about one order of magnitude lower than that of ΔI-SM. Comparing these two mutants, the ΔΔIhh-SM was generated by replacing the 36-residue peptide in an unstructured loop region from ΔI-SM with a 7-aa peptide β-turn. Although this mutation decreases the zinc binding affinity, the loop region is rather distant from the zinc binding sites. The overall structure and the position of four coordination residues are nearly identical in two inteins, therefore, the different binding affinity is unlikely from the local structural difference for the zinc coordination. CD spectroscopy confirmed that the protein conformation changes induced by zinc binding were negligible (data not shown). This conclusion is also supported by NMR spectra, in which the zinc binding does not shift any peaks on the HSQC maps. (Fig. 3) However, TΔS change from positive in ΔΔIhh-SM/Zn binding to negative in ΔI-SM/Zn binding, while ΔH value is considerably larger for the zinc binding to ΔI-SM than to ΔΔIhh-SM. (ESI Fig. S1) These data suggest the loop region has dominant contribution in enthalpy change in the formation of dimeric conformation with zinc binding. In the mean time, the negative entropy change (ΔS) in ΔI-SM/Zn interaction may facilitate the protein crystallization. The ΔI-SM crystal was obtained slowly after zinc was recruited (most likely from glassware contamination),13 showing the important role of zinc binding in the crystallization. In contrast, ΔΔIhh-SM could be crystallized without zinc binding.13

It is generally accepted that intein splicing consists of four steps: (1) N-S or N-O acyl shift and the formation of a linear ester intermediate; (2) Transesterification and formation of a branched ester intermediate; (3) Succinimide formation and peptide cleavage by the cyclization of the asparagines residue adjacent to the C-terminal splicing junction; (4) Succinimide hydrolysis and rearrangement of the ester linking two extein segments to form a peptide bond.24 Several bond rearrangements involved in the reaction process, and mechanism shows the significance of three conserved residues at the cleavage junctions, Cys1, C-terminal asparagine and Cys+1. Even though not illustrated in this four-step mechanism, several other conserved residues play important roles in the protein splicing based on the activity studies, such as His73, Asp422 and His443 in the recA intein.13, 17 Although the detailed function of these three residues in the recA intein is not clear, NMR studies show that the protonation status of His73, Asp422 and His443 are crucial for the protein dynamic properties and the stabilization of protein structures (unpublished results).

While zinc binding to the active core residues in intein, the coordination must restrict the mobility of these residues. Among the four zinc binding residues in ΔI-SM, three of them are conserved residues in intein sequences, C-terminal N440, penultimate H439 and F-block histidine H429. Therefore, the bond rearrangements for new peptide bond formation are restrained by zinc coordination and consequently the protein splicing is inhibited. In such condition, the metal binding affinity must associate with the inhibition efficiency. ITC measurements in this work demonstrated that the observed binding constants KITC of different metal ions have reasonable correlation with the inhibition efficiency in literature.15

Protein structures are not perturbed by metal ion coordination based on CD measurements. This is consistent with the result from crystallography study on the recA intein.13 By comparison of crystal structure of zinc-bound ΔI-SM and zinc-free ΔΔIhh-SM, the overall structures of two mutants are nearly the same, even the local structures around zinc binding site are very similar. This result indicates the zinc binding does not affect the protein structures. Therefore, we can conclude that the restrain of key residues by metal coordination is likely the key cause of the inhibition of the protein splicing instead of structural changes.


This work has quantified the binding affinity and thermodynamic properties of metal ions binding to Mtu recA intein mutants. The higher affinity of zinc binding to ΔI-SM relative to the other mutants is consistent with the unique existence of zinc in crystal structure of this mutant. The binding constants indicated that the zinc binding to intein were strong but still much weaker than to EDTA, which explains the reversible inhibition. The restriction of key residues from metal coordination is probably the major cause of inhibition, based on the correlation of the inhibition efficiency and binding constants, while metal induced protein structure changes were negligible. The zinc binding sites are the same in the crystal structure of ΔI-SM as in solution for ΔI-SM, ΔΔIhh-SM and ΔΔIhh-CM. However, an additional zinc binding site was detected in ΔΔIhh-CM with coordination to Cys1 thiol and His73 imidazole.


Protein Expression and Purification

The inteins were overexpressed in E. coli host strain JM101 in Luria-Bertani (LB) medium as described previously.17 For preparing the 15N isotope labeled intein for NMR measurements, the M9 minimal medium containing 15NH4Cl as the sole nitrogen source was used. The intein expression plasmids were generous gifts from Marlene Belfort. The expressed protein is a fusion of the chitin binding domain (CBD) at the N-terminus and intein at the C-terminus. The purification of inteins were performed on affinity chromatography with chitin beads (New England Biolabs).13 After washing out impurities, inteins were released from the chitin beads by 50 mM (for CM) or 200 mM (for SM) dithiothreitol (DTT), which could induce the N-terminal cleavage between CBD and inteins. 10 mM EDTA was present in the eluents for removing metal ion contaminations. Proteins were concentrated by ultrafiltration in a stirred cell (Millipore Amicon), and DTT and EDTA were removed by buffer exchange under N2. To prevent the oxidation of cysteine residue in inteins, air must be avoided once DTT was removed from proteins. No metal ions were detected in the purified inteins by atomic absorption spectrometer (AAS) measurements. The protein purity was verified by SDS gel, which gave single bands for all inteins. NMR spectra show no impurity peaks (Fig. 3, comparing with the reference21).

Protein modification

Chemical modification of inteins was carried out as described.22 Diethylpyrocarbonate (DEPC) and iodoacetamide were used for histidine and cysteine modification, respectively. Reactions were carried out with 50 μM protein in 20 mM Mops buffer at pH 7.4 with 500 molar-equivalent of freshly prepared DEPC or iodoacetamide. The protein was treated with tris(2-carboxyethyl)phosphine (TCEP) prior to the reaction with iodoacetamide for the cysteine modification. The formation of N-carbethoxy imidazole with DEPC was monitored by UV spectrum at 25 °C by measuring the absorbance at 242 nm. The excessive DEPC and iodoacetamide were removed from protein by dialysis. Size-exclusion chromatography and gel electrophoresis were performed to verify the purity of intein after modification.

ITC measurement and data processing

ITC measurements were carried out at 25 ± 0.2 °C on an isothermal titration calorimeter (VP ITC MicroCal). All solutions were filtered through 0.22 μm filters and degassed prior to use. Samples were buffered at pH 7.4 with either 20 mM Tris or 20 mM MOPS, and the ionic strength was adjusted to 100 mM with NaCl. The metal ion concentrations of stock solutions were verified by titrations against standardized EDTA solutions. Typically, 30 or 50 times of 8 or 5 μl metal solution was injected into the protein in the sample cell during each titration. 120 s delay between injections was allowed for the equilibration. The heat generated or absorbed during interactions was measured by the instrument. The titration of metal ions to buffer solution was carried out along each experiment as a reference for baseline corrections in the data process. All experiments were repeated at least three times.

The ITC data were analyzed with the Origin 7.0 software package from MicroCal. A baseline correction was applied to each experiment by subtraction of data from titration of metal ion solution into a buffer blank correlating to the heat of dilution. A binding isotherm was fitted to the data using a nonlinear least squares method to minimize χ2 values. The detected enthalpy change (ΔHITC) and association constant (KITC) were obtained from the best fit parameters.

The binding constants of metal ions to buffer (Kbuffer) have been measured by ITC. The same experimental method and data process are used as for the titration to proteins. The titration was carried out by the injection of metal ion solution into buffer solution in the sample cell. Pure water was used as the reference correlating to the heat of dilution for the baseline corrections.

NMR Measurement

NMR spectra were collected at 25 °C on a Bruker 600 MHz spectrometer equipped with a cryogenic probe. NMR samples were prepared in phosphate buffer at pH 7.0 containing 10% D2O. 500 μL of 0.3 mM uniformly 15N-labeled inteins were used. Two-dimensional 1H 15N HSQC spectra were recorded in the absence and presence of zinc ion. Data were obtained with a spectral width of 10800 Hz in the 1H dimension and 2400 Hz in the 15N dimension and 32 scans of 2048 real time points for each of 128 t1 increments were recorded. The 15N dimension was zero-filled to 256 points. The data were processed with NMRpipe and analyzed with Sparky. Complete peak assignments have been reported previoursly.21

Dynamic Light Scattering (DLS)

DLS measurements were carried out at 25 °C on a DynaPro MSTC800 DLS instrument (Wyatt Technology Corporation, America) fitted with a 624.4-nm, 50-mW laser by measuring scattering at 90° angle. 20 μM intein samples were prepared in 20 mM Tris-HCl buffer with 200 mM NaCl at pH 7.4. All samples were filtered through a 100 nm membrane before the measurements. The hydrodynamic radii (Rh) of inteins were measured in the absence and in the presence of 500 μM Zn2+. The data were analyzed using Dynamics V6.2 software.

Equilibrium Dialysis

The stoichiometry of metal binding to inteins were determined by equilibrium dialysis. Metal ions were measured by either atomic absorption spectrometer (AAS on Aanalyst 800, Perkin Elmer, USA), or inductively coupled plasma atomic emission-mass spectroscopy (ICP-MS on PlasmaQuad 3, VG Elemental, USA). Every sample was prepared by adequate equilibrium dialysis and the reference samples were collected from the solution outside of dialysis bag.

Supplementary Material



This work was supported by the Cultivation Fund of the Key Scientific and Technical Innovation Project, Ministry of Education of China (No. 707036), NSFC (No. 20873135) and NIH (GM44844 & GM081408). Y. L. would thank Chinese Academy of Sciences for the One Hundred Talent Project. We are grateful to Dr. Marlene Belfort for the continuous support and helpful discussions.


Electronic Supplementary Information (ESI) available: [Thermodynamic parameters, results from equilibrium dialysis and DLS measurements].


1. Paulus H. Annual Review of Biochemistry. 2000;69:447–496. [PubMed]
2. Xu MQ, Evans TC. Current Opinion in Biotechnology. 2005;16:440–446. [PubMed]
3. Muralidharan V, Muir TW. Nature Methods. 2006;3:429–438. [PubMed]
4. Paulus H. Chemical Society Reviews. 1998;27:375–386.
5. Kane PM, Yamashiro CT, Wolczyk DF, Neff N, Goebl M, Stevens TH. Science. 1990;250:651–657. [PubMed]
6. Hahn ME, Muir TW. Trends Biochem Sci. 2005;30:26–34. [PubMed]
7. Gogarten JP, Senejani AG, Zhaxybayeva O, Olendzenski L, Hilario E. Annual Review of Microbiology. 2002;56:263–287. [PubMed]
8. Perler FB. Nucleic Acids Research. 2002;30:383–384. [PMC free article] [PubMed]
9. Derbyshire V, Wood DW, Wu W, Dansereau JT, Dalgaard JZ, Belfort M. Proceedings of the National Academy of Sciences of the United States of America. 1997;94:11466–11471. [PubMed]
10. Shingledecker K, Jiang SQ, Paulus H. Gene. 1998;207:187–195. [PubMed]
11. Poland BW, Xu MQ, Quiocho FA. Journal of Biological Chemistry. 2000;275:16408–16413. [PubMed]
12. Sun P, Ye S, Ferrandon S, Evans TC, Xu MQ, Rao ZH. Journal of Molecular Biology. 2005;353:1093–1105. [PubMed]
13. Van Roey P, Pereira B, Li Z, Hiraga K, Belfort M, Derbyshire V. Journal of Molecular Biology. 2007;367:162–173. [PMC free article] [PubMed]
14. Ghosh I, Sun L, Xu MQ. Journal of Biological Chemistry. 2001;276:24051–24058. [PubMed]
15. Mills KV, Paulus H. Journal of Biological Chemistry. 2001;276:10832–10838. [PubMed]
16. Wood DW, Wu W, Belfort G, Derbyshire V, Belfort M. Nature Biotechnology. 1999;17:889–892. [PubMed]
17. Hiraga K, Derbyshire V, Dansereau JT, Van Roey P, Belfort M. Journal of Molecular Biology. 2005;354:916–926. [PubMed]
18. Leavitt S, Freire E. Curr Opin Struct Biol. 2001;11:560–566. [PubMed]
19. Wilcox DE. Inorg Chim Acta. 2008;361:857–867.
20. Rahimi Y, Shrestha S, Banerjee T, Deo SK. Anal Biochem. 2007;370:60–67. [PubMed]
21. Du ZM, Liu YZ, Zheng YC, McCallum S, Dansereau J, Derbyshire V, Belfort M, Belfort G, Van Roey P, Wang CY. Biomol Nmr Assigm. 2008;2:111–113.
22. Follmer C, Carlini CR. Arch Biochem Biophys. 2005;435:15–20. [PubMed]