|Home | About | Journals | Submit | Contact Us | Français|
We report the first detailed investigation of the kinetics of protein splicing by the Methanococcus jannaschii KlbA (Mja KlbA) intein. This intein has an N-terminal Ala in place of the nucleophilic Cys or Ser residue that normally initiates splicing but nevertheless splices efficiently in vivo [Southworth, M. W., Benner, J., and Perler, F. B. (2000) EMBO J.19, 5019–5026]. To date, the spontaneous nature of the cis splicing reaction has hindered its examination in vitro. For this reason, we constructed an Mja KlbA intein–mini-extein precursor using intein-mediated protein ligation and engineered a disulfide redox switch that permits initiation of the splicing reaction by the addition of a reducing agent such as dithiothreitol (DTT). A fluorescent tag at the C-terminus of the C-extein permits monitoring of the progress of the reaction. Kinetic analysis of the splicing reaction of the wild-type precursor (with no substitutions in known nucleophiles or assisting groups) at various DTT concentrations shows that formation of the branched intermediate from the precursor is reversible (forward rate constant of 1.5 × 10–3 s–1 and reverse rate constant of 1.7 × 10–5 s–1 at 42 °C), whereas the productive decay of this intermediate to form the ligated exteins is faster and occurs with a rate constant of 2.2 × 10–3 s–1. This finding conflicts with reports about standard inteins, for which Asn cyclization has been assigned as the rate-determining step of the splicing reaction. Despite being the slowest step of the reaction, branched intermediate formation in the Mja KlbA intein is efficient in comparison with those of other intein systems. Interestingly, it also appears that this intermediate is protected against thiolysis by DTT, in contrast to other inteins. Evidence is presented in support of a tight coupling between the N-terminal and C-terminal cleavage steps, despite the fact that the C-terminal single-cleavage reaction occurs in variant Mja KlbA inteins in the absence of N-terminal cleavage. We posit that the splicing events in the Mja KlbA system are tightly coordinated by a network of intra- and interdomain noncovalent interactions, rendering its function particularly sensitive to minor disruptions in the intein or extein environments.
Inteins are intervening sequences that are post-translationally excised from precursor proteins with simultaneous splicing of flanking regions, termed the exteins, to form mature proteins.(1) Standard protein splicing is believed to occur via the mechanism summarized in Scheme 1.2−4 All standard inteins utilize a Cys, Thr, or Ser residue at position 1 to perform an acyl rearrangement and form a (thio)ester linkage at the N-terminal splice junction in the first step of the reaction (Scheme 1, step 1).(5) Splicing is blocked upon nonconservative substitution of this residue.5,6 Therefore, it has long been believed that noncanonical inteins, such as Methanococcus jannaschii KlbA (Mja KlbA) intein, which harbors an Ala at position 1 (Ala1), cannot undergo splicing. Nevertheless, it has been shown that the Mja KlbA intein splices efficiently in vivo and does so by an alternative splicing mechanism (Scheme 2).(7) In this mechanism, a nucleophilic attack by the Cys located at the N-terminus of the C-extein (Cys+1) on the peptide bond at the N-terminal splice junction occurs as the first step of the splicing reaction (Scheme 2, step 1).(7) This step results in the formation of a branched intermediate with two N-termini, one of the N-extein and another of the intein. This situation is fundamentally different from what is observed in the standard intein pathway, in which the C-extein nucleophile attacks a previously formed linear (thio)ester intermediate resulting in the formation of the branched intermediate (Scheme 1, step 2).(8) In both pathways, the branched intermediate is resolved during a transamidation step performed by the C-terminal intein residue, Asn, which results in the release of the intein (Scheme 1, step 3; Scheme 2, step 2). A spontaneous S–N or O–N acyl shift, which results in the formation of a peptide bond between the N- and C-exteins, completes the reaction (Scheme 1, step 4; Scheme 2, step 3). More recently, another class of atypical inteins was identified, which splices by a third mechanism involving two branched intermediates.(9) Intein splicing mechanisms are now divided into three classes. Class 1 inteins follow the standard splicing pathway. Class 2 inteins follow the Mja KlbA splicing pathway. Class 3 inteins follow the two-branch intermediate splicing pathway.
Side reactions off the main splicing pathway have been detected in all classes of inteins, often as a consequence of improper coordination between various steps of the splicing mechanism resulting from substitution of catalytically important amino acid residues. The side products arise from cleavage at either or both splice junctions without concomitant ligation (Schemes 1 and 2, steps i–iii). Previous studies have shown that substitution of essential catalytic residues at one splice junction usually inhibits splicing and isolates the cleavage side reaction at the other junction. It has also been shown that nucleophiles such as DTT, hydroxylamine, and sodium 2-mercaptoethanesulfonate (MESNA) can intercept the (thio)ester intermediates (linear and branched), resulting in the formation of N-terminal cleavage products (Schemes 1 and 2, step ii).
In this study, we sought to define the kinetic details of the nonstandard Mja KlbA intein splicing reaction to improve our understanding of its mechanism. To this end, we designed an intein–mini-extein precursor with a redox switch and synthesized it by intein-mediated protein ligation [IPL, also known as expressed protein ligation (EPL)]. This system allowed us for the first time to dissect the kinetics of cleavage and cis splicing of this novel intein from an active precursor with no substitutions in known nucleophiles or assisting groups. The effects of pH variation and of substitution of catalytic and assisting groups on the kinetics of the overall splicing reaction and individual steps were further examined with this system. On the basis of our results, we suggest an explanation for how a direct attack on a peptide bond at the N-terminal splice site by Cys+1 is facilitated in the Mja KlbA intein when it has been blocked in all standard inteins studied to date.
Bacto yeast extract and Bacto tryptone were purchased from Becton, Dickinson, and Company (Sparks, MD). Dextrose, magnesium chloride, sodium hydroxide, tris(hydroxymethyl)aminomethane (Tris), sodium chloride, and urea were purchased from J. T. Baker Chemical Co. (Phillipsburg, NJ). Isopropyl β-d-thiogalactopyranoside (IPTG), MESNA, chloramphenicol, N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (HEPES), and Sephadex-G50 were purchased from Sigma (St. Louis, MO). Ampicillin was purchased from America Pharmaceutical Partners, Inc. (Schaumburg, IL). Sodium dodecyl sulfate (SDS) was purchased from Bio-Rad (Hercules, CA). Oligonucleotide primers, reagents for the polymerase chain reaction (PCR), restriction enzymes, 10× CP buffer, pTWIN1 and pMAL-c2X vectors, and competent Escherichia coli cells (NEB TURBO and T7 Express) were obtained from New England Biolabs (NEB). The pRIL vector was purchased from Stratagene (La Jolla, CA).
Peptides used in this study are summarized in Table Table11 and comprise native C-extein sequences. P1NC, P2AC, P3QC, P4NS, and P5NA were synthesized at NEB. P1NCs, P2ACs, and P6NT were purchased from AnaSpec (San Jose, CA). Peptides P1NC and P1NCs, and peptides P2AC and P2ACs, respectively, are identical in the nature of their functional residues but differ in their lengths (Table (Table1).1). The pure forms of the peptides were characterized as the desired products by electrospray ionization time-of-flight mass spectrometry (ESI-TOF) using a 6210 ESI-TOF mass spectrometer with a capillary electrospray ionization source (Agilent Technologies, Santa Clara, CA).
The previously described Mja KlbA intein precursor, which encodes the native intein (168 residues) flanked by seven native N-extein residues (MNTGHDG) and six native C-extein residues (CSGTLH),(10) was used as the initial PCR template. PCR was conducted using a sense primer (5′-ATAAACTGCCAGGAATTGGGGATCGG-3′), which anneals to a sequence upstream of the NdeI restriction site, and an antisense primer (5′-TGTTACGCTCTTCTGCAAGCAGCAAAACCTTCGTTTTTTCC-3′), which truncates the intein after residue 166. The antisense primer also introduces a substitution at codon Val166 [GTC (V) to GCT (A), triplet complement in bold] and a SapI restriction site (underlined). PCR resulted in a 650 bp fragment, which was then digested with NdeI and SapI, and ligated into the pTWIN1 vector (NEB) digested with the same enzymes. To introduce a maltose binding protein (MBP) tag at the N-terminus of the precursor preceding the Mja KlbA seven-amino acid native N-extein, we first cut the resultant plasmid with NdeI to yield a linear vector. Next, a HindIII site was introduced adjacent to the NdeI site by insertion of a double-stranded oligonucleotide cassette (5′-TATGCTAGGCAAGCTTGTA-3′ and 5′-TATACAAGCTTGCCTAGCA-3′). Finally, the resultant plasmid was digested with NdeI-HindIII and ligated to the 1201 bp NdeI-HindIII restriction fragment of pMAL-c2X, carrying the malE gene encoding MBP, to yield the desired expression vector, pTWIN1-MI166. MI166 corresponds to the MBP-tagged (N-terminus) Mja KlbA intein with seven native N-extein residues and with intein truncated at residue A166.
Single-site substitutions were introduced into pTWIN1-MI166 using the Phusion Site-Directed Mutagenesis Kit (NEB). The primers used are listed in Table Table2.2. PCRs were conducted according to the manufacturer’s instructions. After amplification, the parental DNA strand was digested with DpnI restriction enzyme, and the digested mixture was used to transform NEB Turbo competent E. coli cells.
The sequences of the coding regions of all plasmid constructs prepared in this study were verified to ensure the absence of unwanted mutations. Sequencing was performed at the NEB DNA Sequencing Facility.
E. coli T7 Express cells were transformed with pTWIN1-MI166 and pRIL plasmids. pRIL encodes cognate tRNAs for rare codons AGG/AGA, AUA, and CUA and confers resistance to chloramphenicol (Stratagene). Transformed cells were grown with vigorous aeration at 37 °C in LB medium containing 10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, 1 g/L dextrose, 1 g/L MgCl2 (pH 7.2), 0.25 g/L ampicillin, and 0.025 g/L chloramphenicol. Details of the growth protocol have been described previously.(10)
Ligation of MI166 into the NdeI and SapI sites of pTWIN1 results in the fusion of the Mycobacterium xenopi GyrA (Mxe GyrA) intein to the C-terminus of the expressed protein. The chitin-binding domain present on the Mxe GyrA intein allows the affinity purification of the expressed protein on a chitin resin. MI166 and variant fusion proteins were purified as instructed in the IMPACT-TWIN manual (NEB). Buffer A [20 mM Tris-HCl (pH 7.5) and 500 mM NaCl] was used during sonication of harvested cells, and buffer B [20 mM Tris-HCl (pH 7.5), 500 mM NaCl, and 50 mM MESNA] was used as the elution buffer. Fractions containing the protein of interest, as judged by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), were pooled and concentrated to ~10 mg/mL using an ultrafiltration spin column with a YM30 membrane (Vivascience). The concentration of the protein was determined with the Bio-Rad Bradford protein assay kit standardized with BSA. The MI166 protein was flash-frozen in liquid N2 and stored at −80 °C until it was used. SDS–PAGE revealed the protein to be >98% pure. A typical yield was ~10 mg/g of cell paste.
A typical IPL reaction mixture contained, in a final volume of 160 μL, 100 mM Tris-HCl (pH 8.5), 10 mM MESNA, 50–80 μM MI166 protein, and 0.5 mM peptide P. All of the steps were performed at 4 °C unless stated otherwise. The reaction sample was incubated for ~15 h for ligation to occur before it was applied to a Sephadex-G50 spin column for removal of MESNA. The column fractions were analyzed on a Novex 10–20% Tris-glycine polyacrylamide gel (Invitrogen, Carlsbad, CA). The concentrations of protein fractions containing the IPL product, MI166-P (also known as the precursor), were determined using the Bradford method. These fractions were flash-frozen in liquid N2 and stored at −80 °C until they were used.
A typical protein splicing assay mixture contained, in a final volume of 540 μL, 20 mM Tris-HCl (pH 7.5), 500 mM NaCl, 0.1 μg/μL precursor, and 1, 5, or 25 mM DTT. Protein splicing was initiated by the addition of DTT after incubation of the other components of the assay mixture at 42 °C for 5 min. Aliquots (20 μL) of the assay mixture were removed at designated times, added to blue loading buffer [6% (w/v) SDS] containing 8 M buffered ureaa [1 M Tris-HCl (pH 8.0)], and placed on dry ice until all time point samples were collected. Some of the samples were then incubated in a boiling water bath for 5 min prior to being loaded on a Novex 10–20% Tris-glycine polyacrylamide gel for analysis, whereas others were loaded directly on the gel, as illustrated in Results. Coomassie Brilliant Blue R250 staining was used to visualize protein bands.
For pH effect studies, 20 mM sodium phosphate (Na2HPO4/NaH2PO4) buffer (pH 6.0), 20 mM HEPES (pH 6.5 or 8.0), or 20 mM Tris-HCl (pH 8.5 or 9.0) was substituted for 20 mM Tris-HCl (pH 7.5) in the assay described above.
Fluorescent band signals were detected using a CCD camera (ChemiDoc XRS, Bio-Rad) and quantified using Bio-Rad’s Quantity One. Kinetic traces were plotted as [Ii] versus time (seconds), where [Ii] = Ii/(I1 + I2 + I3 + ... In) and I1 = I°1 – Ib, I2 = I°2 – Ib, I3 = I°3 – Ib, etc. (I° is the measured fluorescence intensity of the species at time t, Ib is the background intensity, and Ii is I1, I2, I3, ..., or In).
The ratio of the sum of fluorescent intensities at time t [∑(I1 + I2 + I3 + ... In)t] to the sum of fluorescent intensities at time zero [∑(I1 + I2 + I3 + ... In)0] was determined to be ~1, which ensures no loss of fluorescence intensity over the time course of the experiment.
Kinetic simulations were performed using KinTekSim (KinTek Corp., Austin, TX) according to the simulation models provided in the corresponding figures.
For N-terminal sequencing, protein bands from 10–20% Tris-glycine polyacrylamide gels were transferred to a PVDF membrane (Problott, Applied Biosystems Inc.) according to the procedure of Matsudaira,(11) with modifications described previously.(12) The membrane was stained with Coomassie Blue R250, and the desired bands were excised from the surrounding membrane and subjected to sequential degradation on a Procise 494 protein/peptide sequencer (Applied Biosystems Inc.).(12)
The in vitro characterization of protein splicing of any intein (cis splicing or trans splicing) requires the assembly of an unspliced and active form of the intein known as the precursor and a mechanism for initiating splicing at the desired time. For example, the cis splicing Pyrococcus GB-D DNA polymerase (GB-D pol) intein can be purified in a precursor form when inserted into the MBP-paramyosin ΔSal gene and its splicing initiated by an increase in temperature or pH.(8) Other examples include naturally and artificially split inteins, which are isolated as two or more separate precursor fragments and splice only upon assembly of these fragments. Attempts to identify conditions that would slow the splicing of the wild-type Mja KlbA intein in E. coli and allow for the isolation of an intact precursor were not successful. As an alternative approach, we resorted to IPL to construct a functional precursor, which would maintain the splicing efficiency exhibited by the wild-type Mja KlbA intein in vivo. The crucial design feature is a built-in redox switch consisting of a disulfide linkage between the existant Cys +1 and an engineered neighboring Cys. The approach was inspired by previous studies employing thiol–disulfide redox switches to control other protein activities.13−16
The IPL system consists of two protein fragments, the recombinantly expressed N-terminal portion of the precursor with a C-terminal thioester-activated bond and a synthetic peptide fragment bearing a Cys as the first residue. Each synthetic peptide fragment encodes the last residues of the Mja KlbA intein, when required, followed by native Mja KlbA C-extein sequences. A number of Mja KlbA site-directed variants were prepared to determine the best site for peptide ligation (Supporting Information). Two major concerns were the positioning of the engineered Cys in the proximity of Cys+1 and the identity of the C-terminal amino acid of the expressed fragment. Previous studies showed that a C-terminal Val, Thr, and Ile often diminish ligation efficiency, whereas His, Cys, and Gly give the most efficient ligation with an N-terminal Cys peptide.(17) Assessments of ligation and splicing efficiencies showed that the best position for truncation of the Mja KlbA intein is Val166, which is substituted with Ala to increase the efficiency of ligation. The peptide has Ser167 as the first residue, which is substituted with Cys to permit ligation of the peptide to the expressed protein fragment and formation of a deactivating disulfide with Cys+1. As a result, the IPL system used in our study consists of a recombinant fragment, MI166, and the chemically synthesized peptide, P. MI166 comprises MBP,b seven native N-extein residues (MNTGHDG), and the first 166 intein residues with Ala166 as the last residue. P contains the two most C-terminal intein residues (S167C and N168), 10 or 30 native C-extein residues (depending on the P used), and a fluorescein tag (FAM) (Figure (Figure1).1). The V166A substitution and the S167C substitution were tested independently in a standard precursor and shown not to inhibit splicing in vivo (Tables S1 and S2 of the Supporting Information).
MI166 expressed in E. coli, as detailed in Materials and Methods, was readily purified with a chitin affinity column using MESNA as the pTWIN vector intein–thioester cleavage reagent. The MI166-MESNA analogue (Figure (Figure1)1) was incubated with an 8–10-fold molar excess of P1NCs for ~15 h at 4 °C. Under these conditions, a strongly fluorescent major new band at an apparent molecular mass of 65.8 kDa was detected, confirming the formation of the ligated product (the splicing precursor) (Figure (Figure2). The2). The ligation reaction does not go to completion: not all of MI166 (64.3 kDa) is converted to MI166-P1NCs, as seen in Figure Figure2.2. However, neither the splicing products [ligated exteins (EN-ECc) (46.4 kDa) and free intein (I) (19.4 kDa)] nor the cleavage side products [MBP-N-extein (ENc) (44.9 kDa) and intein-C-extein (I-EC) (20.9 kDa)] form at detectable levels, indicating that the conditions under which the ligation reaction is performed are optimal. Upon removal of excess MESNA by chromatography on G50 resin, the desired disulfide of MI166-P1NCs apparently forms (Figure (Figure1),1), as the precursor is then stable until the initiating reducing agent, DTT, is added, after which the splicing reaction occurs (Figure (Figure2).2). Formation of the splicing products, EN-EC and I, is confirmed with a denaturing gel (Figure (Figure2,2, lane 7) and ESI-TOF MS (Figure S4 of the Supporting Information).
Initiation of the MI166-P1NC splicing reaction with 1 mM DTT at 42 °C results in the appearance of four new species over the course of 3 h (Figure (Figure3A):3A): EN-EC (47.9 kDa), EN (44.9 kDa), I-EC (23.1 kDa), and I (19.4 kDa). Only two of these species are fluorescent, EN-EC and I-EC (Figure (Figure3B).3B). The assignment of these species is done on the basis of: (i) their estimated molecular masses determined in comparison to a broad range molecular mass marker, (ii) the products that are expected to form during the reaction of MI166-P1NC, and (iii) the fluorescent properties of the species (only those species with intact EC should fluoresce). Kinetic traces for this reaction exhibit an exponential decay for MI166-P1NC [Figure [Figure3C3C (●)], an exponential rise for EN-EC [Figure [Figure3C3C (■)], and transient behavior for I-EC [Figure [Figure3C3C (▲)].
Mixing of MI166-P2AC, a variant precursor disabled in the C-terminal cleavage reaction by the substitution of N168 with A, with 1 mM DTT at 42 °C results in the formation of two new species (Figure (Figure4A,B).4A,B). These species are assigned as EN (44.9 kDa) and I-EC (23.1 kDa), products of the N-terminal cleavage side reaction. Species I-EC appears to be stable [Figure [Figure4C4C (■)].
We were puzzled by the transient nature of I-EC in the reaction of MI166-P1NC (Figure (Figure3C).3C). A possible interpretation is that the decay of I-EC is a result of Asn cyclization, which would lead to the release of EC and formation of free I. The detection of stable I-EC in the reaction of MI166-P2AC strengthens this interpretation (Figure (Figure4C).4C). However, our attempts to simulate the kinetic traces obtained in the reaction of MI166-P1NC with such a model were unsuccessful. It was necessary to invoke the accumulation of an intermediate, which decays via two constituent pathways to form EN-EC and I in one pathway and I-EC and EN in the other (Figure (Figure6B;6B; discussed in the next section), to reproduce the experimental data. We suggest that this intermediate is the branched thioester intermediate, as discussed below.
Studies of the in vivo splicing reaction of the Mja KlbA N168A intein variant suggest that the branched intermediate does accumulate. In an attempt to capture the branched intermediate in vitro, we omitted the heat quench step from the analysis of the MI166-P2AC reaction, anticipating that heating might result in the hydrolysis of the trapped branched thioester intermediate.(18) Interestingly, a new band at an apparent molecular mass comparable to that of the precursor accumulates, whereas species I-EC is hardly detected under the new quenching conditions (Figure (Figure5).5). Sequential Edman degradation of the new species detected the presence of two residues in each cycle: one residue corresponds to the N-terminus of MBP, and the other is equivalent to that of the intein. This result, in addition to the fact that this species is fluorescent, leads us to conclude that the species formed when heat quenching is omitted is the branched intermediate. On the other hand, sequential Edman degradation of the protein in the band assigned as I-EC revealed that the first 10 residues do indeed correspond to the sequence of the intein consistent with this assignment.
In the MI166-P1NC reaction, the species observed on a denaturing gel are equivalent, regardless of the quenching procedure (data not shown). The branched intermediate is not observed, by contrast to the MI166-P2AC reaction. However, kinetic simulations of the MI166-P1NC time course, as discussed in detail in the next section, point to the accumulation of the branched intermediate during the reaction. Therefore, we conclude that the branched intermediate in the MI166-P1NC reaction decomposes upon quenching, hydrolyzing to form I-EC and EN.
Simulation of the kinetics of the MI166-P1NC reaction at varying DTT concentrations (1, 5, and 25 mM) was used to uncover the mechanistic details of the reaction (Figure (Figure6A).6A). The mechanism for which simulated traces best agreed with the experimental data is summarized in Figure Figure6B.6B. The precursor is a heterogeneous mixture of two forms: unreactive, (MI166-P1NC)U, and reactive, (MI166-P1NC)R. Formation of the branched intermediate (Branched) from the reactive precursor is reversible with a forward rate constant (k1) of 1.5 × 10–3 s–1 and a reverse rate constant (k–1) of 1.7 × 10–5 s–1. It decays via two pathways. The first is productive and results in the formation of the spliced products, EN-EC and I. This pathway is relatively fast, with an apparent first-order rate constant (k2) of 2.2 × 10–3 s–1; the second pathway is the N-terminal cleavage side reaction. It has a pseudo first-order rate constant (k3) of (2.0 × 10–5 mM–1 s–1) × [DTT] and is slower than the productive pathway at the DTT concentration used in our experiments. The N-terminal cleavage product, I-EC, further undergoes Asn cyclization, yielding EC and I with a k4 of 1.7 × 10–5 s–1. The simulated trace for the transient species is the sum of the concentrations of Branched and IEC (Figure (Figure6C,6C, triangles, solid line), consistent with our deduction that Branched hydrolyzes upon quenching. The transient behavior is attributable to the accumulation of Branched (Figure (Figure6C, dashed6C, dashed line) rather than IEC (Figure (Figure6C,6C, dotted line).
Kinetic simulations of the MI166-P2AC reaction, performed under the same conditions as for the reaction of MI166-P1NC, reveal interesting differences between the two mechanisms (Figure (Figure7).7). First, precursor structural isomers, (MI166-P2AC)I and (MI166-P2AC)II, are both reactive, with (MI166-P2AC)I decaying to form (MI166-P2AC)II with a k0 of 1.7 × 10–5 s–1 (Figure (Figure7B);7B); second, Branched decays reversibly into another intermediate species with a reverse rate constant (k–2 = 3.3 × 10–2 s–1) that is considerably greater than the forward rate constant [k2 = (6.7 × 10–4 mM–1 s–1) × [DTT]] at the DTT concentrations interrogated (Figure (Figure7B).7B). The nature of this species is unclear, but it is proposed to be an intermediate formed between MI166-P2AC Branched and DTT, a species predicted not to form in the reaction of MI166-P1NC. We designate this intermediate (Branched·DTT) (Figure S1 of the Supporting Information). (Branched·DTT) undergoes N-terminal cleavage at a k3 of 1.7 × 10–4 s–1, which is 10-fold greater than the rate constant for N-terminal cleavage in the MI166-P1NC reaction at 1 mM DTT. As expected for this variant, no C-terminal cleavage products are detected.
Interestingly, MI166-P1NC and MI166-P2AC Branched are both less reactive to thiolysis by DTT, as evident from the extracted rate constants (Figures (Figures6B6B and and7B),7B), than has been previously observed for standard inteins such as the Sce VMA intein,(19)Ssp DnaE intein,(20) and Pab PolII intein.21,22 This explains the productive formation of spliced product in the former and the stability of Branched in the latter, despite the presence of relatively high concentrations of DTT, which have been shown to intercept the (thio)ester intermediates in the splicing pathways of other inteins.19−21,23
The lengths of peptides P1NC and P2AC were reduced from 32 amino acids to 12 amino acids (P1NCs and P2ACs) (Table (Table1)1) to study the effect of peptide length on the assigned rate constants of the various steps of the MI166-P1NC and MI166-P2AC reactions. The results obtained indicate that precursors with a longer EC exhibit higher Keq values in the equilibrium step between the reactive precursor and Branched (Figure S2 of the Supporting Information). It was not necessary to vary other rate constants to accommodate the data for the reactions of the precursors formed with either P1NC or P2AC.
Replacement of the N168 residue with Q prevents C-terminal cleavage in vivo. In vitro, reaction of MI166-P3QC in the presence of 1 mM DTT at 42 °C results in the formation of three new species (Figure (Figure8).8). The first species starts with the first 10 amino acids of MBP, as shown by N-terminal sequencing, and is not fluorescent. It further runs at the same position as EN, observed in the reactions of MI166-P1NC and MI166-P2AC, and is therefore identified as EN (44.9 kDa), the product of the N-terminal cleavage side reaction. Each of the two other species has the N-terminus of the intein and the C-terminus of EC (as designated by their fluorescence properties), which indicates that both are I-EC products (I-EC I and I-EC II). It is unclear why these species exhibit distinct mobilities on a denaturing gel.
Kinetic simulations reveal that the MI166-P3QC precursor is heterogeneous in nature, with two different conformations that are stable and undergo their own decay pathways (Figure (Figure9).9). In pathway I, (MI166-P3QC)I decays reversibly to form a branched intermediate (Branched I), which is then subject to rapid N-terminal cleavage by DTT with a k2 of (1.0 × 10–2 mM–1 s–1) × [DTT] to form I-EC I. In pathway II, a branched intermediate (Branched II) is formed reversibly from (MI166-P3QC)II, with a Keq that is slightly larger than that observed for the formation of Branched I. Branched II is then subject to thiolysis to form (Branched II·DTT) reversibly [k2 = (1.3 × 10–3 mM–1 s–1) × [DTT], and k–2 = 1.2 × 10–3 s–1]. In the last step, (Branched II·DTT) undergoes N-terminal cleavage to form I-EC II with a k3 of 1.7 × 10–4 s–1. According to the simulations, I-EC I and I-EC II form in parallel. This deduction offers an explanation for the detection of two different forms of I-EC on a denaturing gel.
Intriguingly, the intermediates in pathway I are analogous of those in the MI166-P1NC reaction, whereas the intermediates in pathway II are analogous of those in the MI166-P2AC reaction. However, the kinetic parameters indicate that Branched I and Branched II are more reactive toward DTT in comparison with the cognate intermediates in the MI166-P1NC and MI166-P2AC reactions. This increased reactivity with DTT can explain why more I-EC is formed in the MI166-P3QC reaction than in the MI166-P2AC reaction at matching DTT concentrations.
Variants with substitutions of residues implicated in the splicing reaction of the Mja KlbA intein have been examined in this study. The Cys+1 variants, MI166-P4NS, MI166-P5NA, and MI166-P6NT, which are expected to inhibit N-terminal cleavage and result in C-terminal cleavage products,(7) showed no evidence of the formation of C-terminal cleavage products in vitro (Figure S3 of the Supporting Information). Only ~4% of the precursor decays after 20 h at 42 °C to yield what might be the products of the slow intrinsic hydrolysis reaction of the scissile peptide (the bands’ identities were confirmed by N-terminal sequencing as EN and I-EC). Other variants examined include MI166/T93A-P1NCs, MI166/Y156A-P1NCs, and MI166/D147A(E)-P1NCs. In vitro, the reaction of MI166/T93A-P1NCs in the presence of 1 mM DTT at 42 °C showed the formation of the splicing products but at a very diminished rate. N-terminal cleavage products were also detected in the reaction (Figure (Figure10A,B).10A,B). This is consistent with the role of block B Thr93, which is a member of the catalytic triad (Thr93, His96, and Cys+1) thought to facilitate reactions at the N-terminal splice junction.(10) It is also consistent with in vivo reports in which replacement of Thr93 with Ala is shown to slow the splicing reaction of Mja KlbA intein without disrupting it.(7) Kinetic simulation of this reaction at 42 °C shows that only 33% of the precursor is reactive. The rate constants for both N- and C-terminal catalytic cleavage events are much lower than those observed for MI166-P1NCs. Rate constants of the side reactions remain the same (Figure (Figure11A).11A). Another variant tested was MI166/Y156A-P1NCs. Residue Tyr156 of block F is highly conserved among inteins as either Phe or Tyr. The NMR structure of the Mja KlbA intein shows that Tyr156 is centrally located in the active site.(10) Mutation studies suggest that the role performed by Tyr156 is structural and not catalytic.(10) In vivo, replacement of Tyr156 with Phe does not affect the splicing efficiency of the intein, while replacement with Ala results in equal amounts of precursor, N-terminal single-cleavage products, and free intein.(10) Interestingly, no C-terminal cleavage is observed in the reaction of MI166/Y156A-P1NCs in the presence of 1 mM DTT at 42 °C in vitro. The only products detected are those of N-terminal cleavage by DTT (Figure (Figure10C,D).10C,D). Simulation provides evidence that this substitution renders the protein mostly inactive (only 10% of the Y156A precursor is reactive) and results in the slow production of I-EC (Figure (Figure11B).11B). Last but not least, substitution of block F Asp147 with Glu or Ala [MI166/D147E(A)-P1NCs] was examined in vitro in the presence of 1 mM DTT at 42 °C. Asp147 is believed to play a role in Branched formation as well as Asn cyclization/C-terminal cleavage.(10) In vivo, replacement of Asp147 with Glu results in predominant formation of the C-terminal cleavage product, while its replacement with Ala inhibits both N- and C-terminal cleavage reactions.(10) In this study, both of these substitutions inhibit N- and C-terminal cleavage events in the reaction of MI166/D147E(A)-P1NCs (data not shown).
Despite the fact that evidence of C-terminal single-cleavage reactions, in the absence of N-terminal cleavage, has been observed for mutant Mja KlbA inteins in vivo, results from our in vitro studies of T93, D147, and C+1 variants support tight coupling between the C-terminal and N-terminal cleavage steps in the Mja KlbA intein–mini-extein precursor.
As an additional probe of the splicing mechanisms, the pH was varied in the reaction of MI166-P1NCs in the presence of 1 mM DTT at 42 °C. The effects on the rate constants of the various steps of the MI166-P1NCs reaction are summarized in Table Table3. The3. The results indicate that physiological pH (7.0–7.5) gives the most efficient formation of the spliced product (maximal values for Keq = k1/k–1 and k2 are obtained in this range). A bell-shaped effect of pH on k1 is also observed. This profile most likely reflects general acid–base catalysis during the generation of the branched intermediate. High pH disfavors Branched in its equilibrium with the precursor. This observation can be explained by the greater stability of amides relative to (thio)esters at elevated pH.24,25k2 decreases at pH >7.5, implying a decreased efficiency of Asn cyclization, possibly because of inhibition of the protonation of the amide N of the scissile bond during succinimide formation or deprotonation or misalignment of a yet to be identified residue that assists Asn cyclization. Similar observations for the decreased efficiency of splicing at high pH have been reported in various other inteins, such as the Pyrococcus DNA polymerase intein,(8)Mycobacterium tuberculosis RecA intein (Mtu RecA),(14) DnaE split intein from Synechocystis sp. PCC6803 (Ssp DnaE),(20) and Pyrococcus abyssi PolII intein (Pab PolII).(21) Conversely, the increase in k3 at pH >7.5 is explained by an increase in the nucleophilicity of DTT upon thiol deprotonation. This has also been observed for the Ssp DnaE intein.(20) pH variation appears to have no effect on k4.
A fundamental unanswered question in the noncanonical splicing mechanism of the Mja KlbA intein has been how Ala1 inteins overcome the barrier to direct attack of a distal Cys residue on the scissile amide bond, when the equivalent attack apparently cannot occur in standard inteins with a C1A substitution. In an effort to address this question, we previously determined the structure of the Mja KlbA intein by NMR spectroscopy.10,26 The structure suggests that a widening in the active site of the Mja KlbA intein as compared to standard inteins, such as Thermococcus kodakaraensis Pol-2 intein,(27)Saccharomyces cerevisiae VMA intein (Sce VMA),(28) and Mxe GyrA intein,(29) could allow the C+1 nucleophile to access the −1 carbonyl group. Steric hindrance blocks such attack at the N-terminal bond in standard inteins. In this study, we have performed a kinetic dissection of the cis splicing mechanism of the Mja KlbA intein, utilizing an intein–mini-extein system with a disulfide redox switch, to shed additional light on the various steps of this reaction. Intriguingly, our results reveal that the rate constants for formation of the branched intermediate and its resolution to spliced products in the MEI166-P1NC reaction are almost indistinguishable (Figure (Figure66 and Table Table4).4). This observation contrasts with what has previously been observed for standard inteins (cis and trans splicing), in which C-terminal cleavage (Asn or Gln cyclization) has been identified as the slowest step of the splicing reaction (a summary of rate constants from various intein systems is presented in Table Table4).4). In Pab PolII intein, C-terminal cleavage is 10-fold slower than N-terminal cleavage.21,22 In the Ssp DnaE intein, it is ~7-fold slower.(20) A recent study by the Muir group directly monitoring the formation of the succinimide product from a semisynthetic branched intermediate presents strong evidence that favors intein-succinimide formation being the slowest step in the Mxe GyrA intein splicing mechanism.(30) Another study by the same group reports rate constants for branched intermediate formation being significantly greater than the rate constants of spliced product formation in the trans splicing reactions of Ssp DnaE and the chimeric split intein, composed of the N-terminal Nostoc punctiforme DnaE intein and C-terminal Ssp intein (Npu*), both inteins with model N- and C-exteins and a C+1S substitution.(31)
It is plausible that the diminished rate at which the branched intermediate forms in the context of the splicing reaction of the Mja KlbA intein might be attributed to the greater entropy of activation relative to the entropy of attack by the neighboring Cys in the standard inteins, which adds to the enthalpic cost for conversion of an amide to a (thio)ester. Despite this discrepancy between the Mja KlbA intein and the standard inteins, the value of the rate constant for formation of the branched intermediate in the Mja KlbA intein is comparable to the rate constant for formation of the linear (thio)ester intermediate in the Mxe GyrA intein and is 10-fold greater than the values of the rate constants for formation of the branched intermediates in the Ssp DnaE and Npu* inteins (Table (Table4).4). This situation reflects a surprising efficiency of the Mja KlbA intein in directing the attack by the distant C+1 nucleophile on the scissile peptide bond, a step standard inteins are incapable of performing in the absence of the N-terminal nucleophile. It is conceivable that such efficiency is the result of a highly ordered conformation for the active form of the Mja KlbA precursor, in which the spatial orientation of C+1 relative to the scissile amide bond mimics that of C1 in standard inteins. Another possibility would be a strained geometry of the active site, which would lead to ground-state destabilization of the precursor and perhaps compensate for the energetically costly rearrangement of an amide bond to a thioester.
Several observations made in this study support the idea that the Mja KlbA intein has evolved to tightly control its active site for effective splicing via a complex network of hydrogen bonds involving not only catalytic but also second- and third-sphere residues. Results showing that Keq describing the formation of the branched intermediate is diminished with a shorter EC imply that splicing is driven forward not only by intradomain protein contacts (residues within the intein active site) but also by interdomain protein contacts (residues in the N- and C-terminal extein domains and the intein domain), which probably result in transition-state stabilization for branched intermediate formation. Furthermore, larger extein domains are probably better at excluding water or thiol molecules from the active site of the intein, which would explain why more off-pathway products are usually observed with heterologous exteins. More insight into the role of interdomain contacts in the overall splicing reaction is gained by the in vitro evidence presented for tight coupling between the N- and C-termini. Substitution of T93, which is involved in N-terminal activation, with A substantially decreases the rates of both branched formation and Asn cyclization. Substitution of C+1 or D147 (shown by an NMR structural model to be hydrogen-bonded via its side chain to Sγ of C+1(10)) with any residue abrogates both N- and C-terminal cleavage events. In light of these observations, and in consideration of the absence of an absolute requirement for N-terminal cleavage to precede C-terminal cleavage in vivo, we suggest the presence of a regulated cross communication between the two termini in the form of a network of shared bonds. The inconsistency of the in vitro and in vivo results could be explained by slight differences in the folding pattern of the intein, which would again indicate sensitivity in the described network to minor disruptions in bonding or folding patterns. Other intein systems, such as the Ssp DnaE intein, the Mtu RecA intein, and the Sce VMA intein, exhibit tight coordination between the N- and C-terminal cleavage events that is explained in terms of a conformational change during N-terminal cleavage, which optimally positions residues for C-terminal cleavage.20,32,33 Furthermore, it has been observed that residues in the N- and C-terminal motifs, flanking the catalytic center, impose a dramatic effect on the splicing reaction at both N- and C-terminal junctions in these same inteins.28,31 A similar explanation could be extended to the Mja KlbA intein, where local conformational changes could be triggered during or after branched intermediate formation to assist in positioning of essential residues for C-terminal cleavage.
The sensitivity of the Mja KlbA intein environment to changes in its catalytic and assisting residues is underscored by (i) the existence of structural isomers of the MI166-P1NC, MI166-P2AC, and MI166-P3QC precursors and (ii) their varying reactivities, as predicted by kinetic simulations. This sensitivity is clearly emphasized in the MI166-P3QC reaction, in which the folding properties of the precursor result in two parallel splicing trajectories. Moreover, a disparity in the reactivity of the branched intermediate formed in theses reactions toward DTT most likely reflects subtle conformational differences in these species. MI166-P3QC Branched I and II are both susceptible to thiolysis, whereas the MI166-P1NC and MI166-P2AC Branched species are much less so. It is quite interesting, though, that Branched I (thought to mimic MI166-P1NC Branched) undergoes faster thiolysis than Branched II (thought to mimic MI166-P2AC Branched). This greater susceptibility would indicate a more open active site for the former species and perhaps explain the stability of Branched in MI166-P2AC. Furthermore, it is intriguing that DTT-induced N-terminal cleavage leads solely to the formation of the N-terminal cleavage side reaction products in various standard inteins such as the Sce VMA intein,(19)Ssp DnaE intein,(20) and Pab PolII intein,21,22 while in the case of the Mja KlbA intein, the presence of even 50 mM DTT still leads mostly to the formation of the splicing products, ligated exteins and intein. The values of the N-terminal cleavage rate constants for the various inteins are comparable when compared at the same temperature in the presence of 50 mM DTT (Table (Table4).4). However, the rate for formation of the succinimide-intein product from the branched intermediate is what dictates the fate of the reaction products in the presence of DTT in either case. In the studied standard inteins, Asn/Gln cyclization is the slowest step of the reaction and is ~10-fold slower than N-terminal cleavage (Table (Table4).4). As a result, the reaction is directed toward formation of N-terminal cleavage products in the presence of high DTT concentrations. On the other hand, in the Mja KlbA intein, Asn cyclization occurs at a rate comparable to that of branched formation and that of N-terminal cleavage in the presence of 50 mM DTT. This results in the bifurcation of the reaction in two pathways and allows the formation of the splicing products as well as N-terminal cleavage products.
In summary, we have constructed an intein–mini-extein system, which closely mimics the native Mja KlbA precursor, with seven native N-extein residues and 30 native C-extein residues, and a triggerable redox switch. Our system has permitted a detailed kinetic dissection of the splicing mechanism of this intein with the assignment of rate constants for every step involved in the splicing reaction, a task unprecedented not only for noncanonical inteins such as the Mja KlbA intein but also for standard inteins. Importantly, our kinetic data provide a number of “firsts” that distinguish the Mja KlbA noncanonical intein from all studied standard inteins. (1) The rate constant for formation of the branched intermediate is comparable if not slightly less than the rate constant of Asn cyclization. (2) Despite the slow nature of formation of a branched intermediate in the splicing reaction of the Mja KlbA intein, it still occurs efficiently at a rate equivalent to that of linear (thio)ester intermediate formation and ~10-fold faster than branched intermediate formation in standard inteins. (3) The branched intermediate is not susceptible to thiolysis by DTT, as are the linear (thio)ester and branched intermediates in standard inteins, and this stability is ensured by comparable rates of formation of branched and succinimide-intein adducts, which guarantees that the reaction proceeds toward the productive formation of the splicing products. Therefore, although the exact molecular mechanism that allows the Mja KlbA intein to overcome the barrier present in standard inteins that prevents attack on an amide bond at the intein N-terminus is yet to be determined, the kinetics of the reaction described herein demonstrate that splicing in this Ala1 intein is not limited by kinetic parameters.
We thank Dr. Catherine L. Madinger for performing ESI-TOF mass spectrometric analysis on the purified peptides. We thank Shelley Cushing for conducting N-terminal sequencing. We also thank Dr. Christopher J. Noren for his scientific input. We thank Dr. Don Comb for support and encouragement.
This work was supported by New England Biolabs.
Determination of the site for peptide ligation in the Mja KlbA intein–mini-extein system (text and Tables S1 and S2) and proposed structures of Branched·DTT, kinetic simulations of the MI166-P1NCs and MI166-P2ACs reactions, in vitro products for the MI166-P6NT reaction, and mass spectrometric analysis of the products of the splicing reaction of MI166-P1NCs (Figures S1–S4, respectively). This material is available free of charge via the Internet at http://pubs.acs.org.
aNo splicing activity was detected for any of the assembled precursors in the presence of 6 M buffered urea.
bThe addition of an MBP tag at the N-terminus distinguishes the N-terminal cleavage product and ligated exteins on an SDS–PAGE gel.
cEN corresponds to the seven native N-extein residues with an N-terminal MBP tag.