Modulation of Protein Splicing of the Saccharomyces cerevisiae Vacuolar Membrane ATPase Intein*

Protein splicing of the Saccharomyces cerevisiae vacuolar membrane ATPase intein involves four highly coordinated reactions that result in precise cleavage and formation of peptide bonds. In this study, we investigated the roles of the last N-extein residue (−1 residue) and the intein penultimate residue in modulating splicing reactions. Most of the 20 amino acid substitutions at the −1 position had no effect on overall protein splicing but could lead to significant accumulation of thioester intermediates when splicing was blocked by mutation. A subset of −1 substitutions attenuated the initiation of protein splicing and enabled us to demonstrate in vitro splicing of a mesophilic intein containing all wild-type catalytic residues. Substitutions involving the intein penultimate residue allowed modulation of the branch resolution and C-terminal cleavage reaction. Our data suggest that the N-S acyl rearrangement, which initiates splicing, may also serve as the rate-limiting step. Through appropriate amino acid substitutions, we were able to modulate splicing reactions in vitro by change in pH or temperature or addition of thiol reagents. Both insertion and deletion were tolerated in the central region of the intein although splicing or structure of the intein may have been affected.

Protein splicing involves a precise excision of an internal segment, the intein, from a protein precursor and a concomitant ligation of the flanking regions, the exteins, resulting in the production of two proteins (1). Since the initial discovery of protein splicing, more than 40 inteins have been identified (2). Sequence analysis reveals that inteins are bounded by Cys, Ser at N termini and Asn at C termini and, with a few exceptions, have His as the penultimate residue (2). The first C-extein 1 residues are invariably Cys, Ser, or Thr (2). The chemical mechanism of protein splicing of the inteins from the thermostable DNA polymerase of Pyrococcus sp. GB-D (Psp pol-1 intein) and the 69-kDa vacuolar membrane ATPase subunit of Saccharomyces cerevisiae (Sce VMA intein) suggests that a common protein splicing pathway may have evolved in thermophilic archaea and mesophilic eukaryotes (3)(4)(5). Protein splicing of the Sce VMA intein consists of the following multistep reactions: step 1, an N-S rearrangement at the intein N terminus (Cys 1 ) to form a thioester bond between Cys 1 and the last N-extein residue (Ϫ1 residue); step 2, a trans-esterification reaction by the first C-extein residue Cys 455 to form a branched intermediate; step 3, succinimide formation at the intein C terminus (Asn 454 ) to resolve the branched intermediate; step 4, a final S-N rearrangement to form a peptide bond linkage between the ligated exteins ( Fig. 1A) (5).
The recently solved crystal structure of the Sce VMA intein demonstrates that both intein terminal residues, Cys 1 and Asn 454 , are in close proximity, forming a structure consistent with their proposed roles in the splicing pathway (6). Since the crystal structure represents the excised intein, the structure of the extein-intein precursor that involves at least one extein residue (i.e. Cys 455 ) at the splicing active site is still unknown. Although the crystal structure supports our proposed splicing pathway, many mechanistic details of the splicing reactions have not been elucidated.
Previously, we examined the Sce VMA intein in an in vitro MYT splicing system (5). The rapid splicing of the Sce VMA intein in vivo precluded the isolation of the precursor and/or intermediates (5,7,8). Our studies utilized amino acid substitutions to arrest or attenuate the splicing process (5). A single substitution, Asn 454 to Ala, was shown to block both splicing and C-terminal cleavage but not N-S acyl arrangement (5). Thiols such as dithiothreitol and cysteine were able to shift the N-S equilibrium by attacking the thioester bond and initiating N-terminal cleavage (5,9). Previous studies have shown that the Sce VMA intein penultimate residue, His 453 , although conserved in almost all inteins, was not essential for protein splicing (5,7,8). A double substitution, H453L and C455S, was shown to attenuate protein splicing in vivo and allow in vitro splicing of the purified precursor (5). Although our data on the in vitro splicing reaction support our proposed splicing pathway (5), direct examination of in vitro splicing of the Sce VMA intein containing unaltered catalytic residues has been a challenge.
The Sce VMA intein also functions as a site-specific homing endonuclease which mediates gene mobility (10 -12). The central conserved dodecapeptide motifs are directly involved in DNA recognition and cleavage (13,14). The crystal structure of the Sce VMA intein suggests that the splicing and endonuclease functions may reside in two separate domains (6). We made large in-frame deletions which removed the domain containing the dodecapeptide motifs, demonstrating the remaining splicing domain was sufficient for efficient splicing (15).
This paper extends our previous in vitro studies to focus on residues that are in close proximity to the reaction center but not directly involved in the splicing reactions, i.e. the last N-extein residue (the Ϫ1 residue) and the intein penultimate residue. This approach enabled us to modulate each of the first three splicing reactions by change in pH, temperature, and/or addition of thiol reagents and allowed direct examination of in vitro splicing of the Sce VMA intein containing wild-type catalytic residues. In addition, we investigated the effect of deletion and insertion in the Sce VMA intein on splicing reactions. Our results provide strategies for modulating protein splicing and insights into the mechanism of protein splicing of a mesophilic intein.

EXPERIMENTAL PROCEDURES
Numbering of Residues in the Intein Fusion Constructs-Amino acid numbers refer to the position in the S. cerevisiae VMA intein, essentially the same as described previously (Fig. 1B) (5).
Constructions of pMYB130 Containing Different Ϫ1 Residues-The construction of pMYB129 containing the wild-type Gly Ϫ1 residue was described previously (9). pMYB129 contains an XhoI site and a KpnI site flanking the N-terminal splice junction including the Ϫ1 position (9). These unique sites allowed convenient substitution of the Gly Ϫ1 with the remaining 19 naturally occurring amino acids through linker insertion. pMYB129 was digested with XhoI and KpnI and then ligated with the complementary oligomers 5Ј-TC GAG NNN TGC TTT GCC AAG GGT AC-3Ј and 5Ј-C CTT GGC AAA GCA NNN C-3Ј that encoded each of the 19 amino acids (NNN) at the Ϫ1 position. The resulting constructs were named pMYB130(X Ϫ1 ), e.g. pMYB130(Ala Ϫ1 ) containing Ala as the Ϫ1 residue, pMYB130(Asn Ϫ1 ) containing Asn as the Ϫ1 residue, etc. The substitutions of the Ϫ1 residues in pMYB130(X Ϫ1 ) were further confirmed by DNA sequencing (New England Biolabs). Both pMYB129 and pMYB130(X Ϫ1 ) contained the N454A mutation, which blocked splicing and C-terminal cleavage and therefore were fusion constructs for the study of the N-S acyl rearrangement and N-terminal cleavage. Unless otherwise stated, all enzymes and plasmids used were the products of New England Biolabs, Inc.
Construction of pMYK(X Ϫ1 ) and pMYK Containing His 453 Substitutions-The T4 polynucleotide kinase gene (16) was synthesized by the polymerase chain reaction using the primers 5Ј-GGT GGT ACC GGT AAA AAG ATT ATT TTG ACT ATT GGC-3Ј and 5Ј-GGT GGT CTG CAG TCA AAA ATC TCC CGA AGC GAC TTG CCA-3Ј. Polymerase chain reaction mixtures (100 l) contained Vent DNA polymerase buffer (New England Biolabs), 3 mM MgSO 4 , 300 M each of the 4 dNTPs, 10 M each primer, 1 l of T4 phage particle suspension, and 1 unit of Vent DNA polymerase. Amplification was carried out for 25 cycles using a Perkin-Elmer thermal cycler at 95°C for 1 min, 50°C for 1 min, and 72°C for 2 min. The product was digested with AgeI and PstI and ligated with pMYB (15) digested with AgeI and PstI, to yield pMYK. pMYK contained the wild-type splice junction residues including Gly as the Ϫ1 residue. To construct pMYK(X Ϫ1 ) containing different Ϫ1 substitutions, pMYB130(X Ϫ1 ) was digested with XhoI and BamHI, and the resulting XhoI-BamHI fragment was used to replace the corresponding fragment from pMYK to yield pMYK(X Ϫ1 ). pMYK(X Ϫ1 ) contained each of the 20 amino acid residues at the Ϫ1 position and the wild-type splice junction residues and was used to study the effect of the Ϫ1 substitutions on splicing.
Construction of pMYT4 and Its Mutant Derivatives-The T4 phage DNA ligase gene (17) was synthesized by the polymerase chain reaction using the primers 5Ј-GGT GGT ACC GGT ATT CTT AAA ATT CTG AAC GAA ATA GCA-3Ј and 5Ј-GGT GGT CTG CAG TCA TAG ACC AGT TAC CTC ATG AAA ATC ACC-3Ј. The reaction conditions were essentially as described above. The product was digested with AgeI and PstI and ligated with pMYB (15) that had been digested with AgeI and PstI, yielding pMYT4, which contains the wild-type splice junction residues including Gly Ϫ1 . pMYT4 was digested with BamHI and AgeI and ligated with the complementary oligomers, 5Ј-GA TCC CAG GTT GTT GTA CAG AAC GCA GGT GGC CTG A-3Ј and 5Ј-CC GGT CAG GCC ACC TGC GTT CTG TAC AAC AAC CTG G-3Ј, to create pMYT4 (H453Q/C455A), in which His 453 and Cys 455 were replaced with Gln and Ala, respectively. Similarly, the same pMYT4 digest was also used with different pairs of complementary oligomers to create pMYT4(H453Q/ N454A/C455S). The pMYT4 mutant constructs were used for the study of C-terminal cleavage.
Construction of p⌬MYB(NG) (N454A) and p⌬MYB(Bam)-The construction of p⌬MYB(NG) was described previously (15). p⌬MYB(NG) was digested with BamHI and AgeI and ligated with the complementary oligomers, 5Ј-GA TCC CAG GTT GTA GTA CAC GCT TGC GGT GGC CTG A-3Ј and 5Ј-CC GGT CAG GCC ACC GCA AGC GTG TAC TAC AAC CTG G-3Ј, to create p⌬MYB(NG) (N454A), in which Asn 454 was replaced with Ala. The gene for BamHI (18) was amplified by polymerase chain reaction following a previously described protocol (9). The forward primer 5Ј-GTT GGT GCT AGC GGT GGT AAC AAC GAA GTT GAA AAA GAA TTC ATC ACT GAT-3Ј encoded an NheI site, and the reverse primer 5Ј-GGT GGT GAC GTC GGT GGT AAC AAC TTT GTT TTC AAC TTT ATC TTT CCA TTT-3Ј encoded an AatII site. The polymerase chain reaction products were digested with NheI and AatII and ligated with the NheI-AatII digested p⌬MYB(NG) to yield p⌬MYB(Bam).
Fusion Protein Expression and Purification by Amylose Affinity Chromatography-The proteins expressed from pMYB, p⌬MYB, pMYK, pMYT4, and their mutant derivatives, are referred to as MYB, ⌬MYB, MYK, MYT4 fusion proteins, respectively. Escherichia coli strain ER2426 (15), harboring these plasmids or their mutant derivatives, was used for fusion protein expression. The expression and purification followed essentially the same procedures as described previously (5). In the case of p⌬MYB(Bam), the plasmid was used to transform E. coli strain pLD2263/m.bamH1, which contained the BamHI methylase gene (19), and the expressed fusion proteins were purified on amylose resin. The protein samples were analyzed by SDS-PAGE, followed by Coomassie Blue staining and Western blot analysis using antibodies raised against BamHI. To test the ability of the splicing product MB fusion proteins to bind chitin, an aliquot (1 ml) of amylose-purified protein was loaded onto a chitin column containing 0.5 ml of resin. The resin was washed three times with Hepes column buffer containing 30 mM Hepes, 0.5 M NaCl, pH 8.0, and the bound proteins were eluted with 2% SDS and analyzed by SDS-PAGE.
DTT-induced Cleavage of Purified Fusion Proteins-All cleavage reactions of purified fusion proteins (0.5-1.0 mg/ml) were conducted in Hepes column buffer (30 mM Hepes, 0.5 M NaCl) containing 40 mM dithiothreitol (DTT) for up to 16 h. For native fusion proteins from pMYB129 and pMYB130 (X Ϫ1 ), cleavage reactions were conducted at 4 and 16°C at pH 8.0. Subsequently, each reaction mixture (0.5 ml) was loaded onto a chitin column containing 0.3 ml of resin. After washing the resin three times with Hepes column buffer to remove the unbound proteins and the residual DTT, the bound proteins were eluted with 2% SDS. The SDS-eluted samples (40 l) were mixed with 20 l of SDS Sample Buffer (New England Biolabs) and analyzed by SDS-PAGE. The percentage of cleavage was determined by comparing the MYB precursors from the DTT-treated samples with those from samples without DTT treatment. For the MYB fusion proteins with Gly, Ala, Ile, Ser, or Gln as the Ϫ1 residue, the half-times of the cleavage reaction were determined by taking aliquots (40 l) at appropriate times during the incubation with DTT. The cleavage reaction was stopped by mixing the samples with SDS Sample Buffer. To examine DTT-induced cleavage of urea-denatured MYB fusion proteins, the proteins were dialyzed against 8 M urea at pH 5.5, 7.6, or 9.5 before incubation in 8 M urea at 23°C in the absence of DTT for up to 16 h or in the presence of DTT for 4 h. The percentage of cleavage under denaturing conditions was determined by comparing the MYB precursors from the DTT-treated samples (at pH 7.6) with those from samples without DTT treatment.
DTT-induced cleavage of fusion proteins from pMYT4 (H453Q/ C455A) was conducted at 23°C at pH 8.0. Aliquots (40 l) were removed at 0.5, 1-6, and 8 h, mixed with 20 l of SDS Sample Buffer, followed by analysis on SDS-PAGE. As a control, the proteins were incubated in the same buffer in the absence of DTT for 16 h. To examine the effect of pH on the cleavage, the purified fusion proteins in Hepes column buffer, pH 8.0, were dialyzed against the same buffer except at pH 6.0. The thiol treatment was then conducted at 4°C at pH 6.0 in the presence of 40 mM DTT or 50 mM cysteine for 8 and 16 h. Aliquots of the thioltreated samples (40 l) were analyzed by SDS-PAGE. The rest of the samples were first dialyzed against Hepes column buffer, pH 8.0, to remove DTT or cysteine and shift the pH back to 8.0, followed by an additional incubation (at pH 8.0) for 16 h. The purified proteins from pMYT4 (H453Q/N454A/C455A) were incubated at 4°C in Hepes column buffer without DTT at pH 8.0 for 16 h or with DTT at pH 6.0 and 8.0 for 16 h. The samples were analyzed on SDS-PAGE as described above.
Analytical Methods-SDS-PAGE was performed in 12% Tris glycine gels (Novex, San Diego, CA), followed by staining with Coomassie Blue. For Western blot analyses, the SDS-polyacrylamide gels were blotted onto nitrocellulose membranes and analyzed by probing with polyclonal antibodies against the Sce VMA intein (New England Biolabs) or BamHI (gifts of Jurate Bitinaite and Rebecca B. Kucera) as described previously (3). The stained gels were digitized with a Microtec Scanmaker 600 ZS, and the scanned images were analyzed to determine relative protein concentration with NIH Image 1.47 software. Variations in protein sample loading (normally Ͻ10%) were normalized prior to comparisons. Protein concentrations were estimated by the method of Bradford (20).

Fusion Constructs of the Sce VMA Intein for Modulation of
Protein Splicing-Protein splicing of the Sce VMA intein was examined by fusing the intein between two independent extein domains (Fig. 1B). E. coli maltose-binding protein (21) (MBP) was used as the N-extein to facilitate purification of splicing products that contained the MBP moiety. C-exteins of varying molecular masses were used for easy resolution of splicing products on SDS-PAGE. The MYB, MYK, or MYT4 fusion systems contained Bacillus circulans chitin-binding domain (22) (CBD or B), T4 polynucleotide kinase (K), or T4 DNA ligase (T4) as the C-extein, respectively (Table I). In the case of the wild-type Sce VMA intein with Gly Ϫ1 and Cys as the first C-extein residue (residue 455), complete in vivo splicing occurred in all three fusion systems (data not shown). The MYB fusion system, including pMYB129, pMYB130(X Ϫ1 ), p⌬MYB1(NG) (N454A), and p⌬MYB(Bam), was constructed to study the effect of the Ϫ1 substitutions on thiol-induced Nterminal cleavage and N-S acyl rearrangement and the effect of deletion and insertion mutations ( Table I). The MYK fusion system, including pMYK(X Ϫ1 ), pMYK(H453X/C455S), was constructed to examine the effect of substituting the Ϫ1 residue and the intein penultimate residue on protein splicing ( Table  I). The MYT4 fusion system, including pMYT4(H453Q/C455A) and pMYT4(H453Q/N454A/C455A), was for the study of succinimide formation coupled to C-terminal cleavage ( Table I).
Effect of the Ϫ1 Substitutions on Induction of N-terminal Cleavage-An N-S acyl rearrangement at the N terminus of the Sce VMA intein forms a thioester bond between Cys 1 and the Ϫ1 residue ( Fig. 1A) (5). Exogenous thiols, e.g. DTT and cysteine, can attack the thioester bond to induce cleavage at the intein N terminus (Fig. 1A) (5,9). This thiol-induced cleavage has been previously examined in pMYB129 (9). In this study, we determined the effect of pH and temperature on the halftime and efficiency of the cleavage reaction (Table II). The rate of the DTT-induced cleavage of the fusion proteins from pMYB129 increased 4-fold at pH 8.0 versus pH 6.0 at 4 -16°C and almost 50-fold at 23°C. At pH 8.0, the rate of cleavage increased 30-fold at 23 versus 4°C (Table II).
To investigate the effect of substitutions at the Ϫ1 position on thiol-induced N-terminal cleavage and N-S acyl rearrange-FIG. 1. A, proposed mechanism for protein splicing and thiol-induced cleavage at splice junctions of the Sce VMA intein. The Sce VMA intein (white box) is inserted between the N-extein (black box) and the C-extein (stripped box). The protein splicing pathway of the Sce VMA intein (5)  V, spliced exteins. The proposed mechanism for thiol-induced cleavage reactions is shown on the upper right. The cleavage reactions were examined in the MYB fusion system, in which the intein contained an N454A substitution (not shown) or in the MYT4 fusion system in which the intein contained a double substitution, H453Q/C455A. See text for more details. The residues at the splice junctions are shown as follows: X, the last N-extein residue (Ϫ1 residue); Cys1, the first intein residue; His453, the intein penultimate residue; Asn454, the last intein residue; Cys455, the first C-extein residue. B, schematic diagram of the Sce VMA intein fusion systems. The same shading scheme as A is used to represent the intein, Nextein and C-extein. The arrows and numbers below indicate the amino acid positions. The splice junction residues are shown essentially the same as A. The dodecapeptide motifs are indicated as shaded boxes. The features of the fusion constructs are described in Table I. ment, the Gly Ϫ1 in pMYB129 was substituted with the remaining 19 naturally occurring amino acid residues to yield pMYB130 (X Ϫ1 ). Each fusion protein was expressed in E. coli and purified on amylose resin. As seen with pMYB129 (9), the majority of the Ϫ1 substitutions resulted in the production of a predominant fusion precursor (MYB) (data not shown). A small amount of MBP was also purified suggesting that cleavage at the intein N terminus in vivo or in the course of purification occurred. As estimated from the amount of the co-purified MBP, the Thr Ϫ1 , Glu Ϫ1 , His Ϫ1 , and Arg Ϫ1 substitutions resulted in 25-75% N-terminal cleavage, whereas the Asp Ϫ1 substitution resulted in 95% N-terminal cleavage (Table III). The other cleavage product, YB, was detected in the crude cell extract confirming that N-terminal cleavage occurred in vivo or in the course of purification (data not shown). Nearly complete N-terminal cleavage of the Asp Ϫ1 and Arg Ϫ1 substitutions precluded the determination of their in vitro DTT-induced cleavage rates (Table III).
In vitro DTT-induced N-terminal cleavage was examined on both native and denatured MYB precursors. As shown in Table  III, most Ϫ1 substitutions allowed 30 -95% cleavage of native MYB with the exception of the Cys Ϫ1 , Asn Ϫ1 , and Pro Ϫ1 substitutions, which inhibited N-terminal cleavage. N-terminal cleavage of denatured MYB ranged from 30 to 92% in 9 Ϫ1 substitutions, whereas the Gly Ϫ1 , Ala Ϫ1 , Ile Ϫ1 , Ser Ϫ1 , Asn Ϫ1 , His Ϫ1 , and Pro Ϫ1 substitutions resulted in only trace amounts (Ͻ5%) of cleavage and the Cys Ϫ1 and Gln Ϫ1 substitutions in 10% cleavage (Table III).
DTT-induced N-terminal cleavage of the denatured MYB precursors suggested that thioester bond formation occurred prior to the DTT treatment. To determine if certain Ϫ1 substitutions shifted the N-S equilibrium and caused accumulation of thioester intermediates, purified fusion proteins containing different Ϫ1 substitutions were denatured in 8 M urea, followed by incubation at three different pH values (5.5, 7.4, and 9.5) in the absence or presence of DTT. Only trace amounts of the Nterminal cleavage products (M and YB) were observed for the denatured MYB precursors containing Gly Ϫ1 (Fig. 2, left panel). The presence of DTT or incubation at basic pH (pH 9.5) had no effect on N-terminal cleavage suggesting that the N-S equilibrium was disrupted after the denaturation. Similar results were obtained with Asn Ϫ1 (Fig. 2, middle panel), Ala Ϫ1 , Ile Ϫ1 , Ser Ϫ1 , His Ϫ1 , and Pro Ϫ1 substitutions (data not shown). However, significant amount of DTT-induced N-terminal cleavage

In vivo In vitro
The approximate percentage of N-terminal cleavage in vivo or in the course of purification was estimated from SDS-PAGE as the ratio of MBP versus the MYB precursor in the amylose-purified proteins.
b Cleavage reactions were in 30 mM Hepes, pH 8.0, 0.5 M NaCl, and 40 mM DTT. The percentage of in vitro N-terminal cleavage was determined by comparing the MYB precursors from the DTT-treated samples with those from the samples without the DTT treatment in scanned images of Coomassie Blue-stained SDS-polyacrylamide gels. Each percentage determination is typical of 2 or 3 determinations with a variation of Ͻ10%. ND, not determined.
c Cleavage reactions of native MYB precursors were conducted for 16 h. The numbers in parentheses are half-times (hour) of the cleavage reaction, i.e. the incubation times required for DTT to cleave 50% of the native MYB precursors. d The MYB precursors were denatured in 8 M urea in 30 mM Hepes, 0.5 M NaCl, pH 7.6, followed by incubation at 23°C for 4 h with or without 40 mM DTT. e Protein splicing of the MYK fusion proteins was determined by inspection of SDS polyacrylamide gels of the splicing products in the crude extract, flow-through, and the amylose-purified proteins (indicative of splicing in vivo or in the course of purification) or samples from incubation of purified proteins (in vitro splicing). ϩ, more than 50% of the MYK fusion spliced into MK and Y; Ϫ, 10% or less MYK fusion spliced into MK and Y; C, C-terminal cleavage of the MYK fusion to yield MY and K. products (M and YB) was observed for the denatured MYB precursors containing the Lys Ϫ1 (Fig. 2, right panel). Incubation of the precursors at pH 7.4 and 9.5 in the absence of DTT yielded the same cleavage products, M and YB (in only trace amount) (Fig. 2, right panel). The missing YB and smeared bands on the SDS-polyacrylamide gel could be due to protein aggregation or precipitation during incubation in the absence of DTT (Fig. 2). These results suggest that Lys Ϫ1 position shifted the N-S equilibrium resulting in accumulation of thioester intermediates prior to induction of cleavage by DTT. Similar results were obtained with the Val Ϫ1 , Leu Ϫ1 , Thr Ϫ1 , Glu Ϫ1 , Phe Ϫ1 , Tyr Ϫ1 , Trp Ϫ1 , and Met Ϫ1 substitutions (Table  III, SDS-PAGE data not shown).
Effect of the Ϫ1 Substitutions on Protein Splicing-Since N-S acyl rearrangement initiates protein splicing (Fig. 1A) (5), it is probable that the Ϫ1 substitutions that shift the N-S equilibrium also affect protein splicing. This was examined in the MYK fusion system using pMYK(X Ϫ1 ), in which the intein contained all wild-type catalytic residues for splicing. After expression of pMYK(X Ϫ1 ) in E. coli, the splicing products were purified on amylose resin. The predicted splicing products were identified by their apparent molecular masses on SDS-PAGE. Accumulation of the linear precursors in some Ϫ1 substitutions suggests that protein splicing was attenuated in vivo despite the fact that the intein contains all wild-type catalytic residues. To examine whether these Ϫ1 substitutions allowed splicing to continue in vitro, purified proteins from the Asn Ϫ1 , Cys Ϫ1 , Val Ϫ1 , Ile Ϫ1 , and Pro Ϫ1 substitutions were incubated in Hepes buffer for up to 40 h. In vitro splicing of the linear precursors from the Cys Ϫ1 and Asn Ϫ1 substitutions (Fig. 3A, lane 8 and 11) was also examined at different pH values. As shown in Fig. 3B, incubation of the precursors containing the Asn Ϫ1 substitution resulted in almost complete splicing at pH 6.0, yielding MK and Y, whereas significant amount of the precursors remained at pH 8.5. Similarly, incubation of the purified proteins from the Cys Ϫ1 substitution at pH 6.0 allowed the remaining linear precursors to splice, but splicing was significantly inhibited at pH 8.5 (data not shown). It is apparent that both Asn and Cys as the Ϫ1 residue allowed completion of protein splicing without side reactions. More importantly, these results suggest that overall splicing reaction of the wild-type Sce VMA intein was favored at pH 6.0 versus pH 8.5. In vitro splicing (at pH 6.0) of the purified proteins from the Ile Ϫ1 substitution, on the other hand, resulted predominantly in C-terminal cleavage, whereas the Val Ϫ1 substitution produced both splicing and C-terminal cleavage products (Table  III). No significant splicing or cleavage in vitro was observed for the Leu Ϫ1 and Pro Ϫ1 substitutions (data not shown).
Effect of Substituting the Intein Penultimate Residue on Protein Splicing-Succinimide formation by the last intein residue, Asn 454 , is coupled to peptide bond cleavage at the intein C terminus, resulting in branch resolution (Fig. 1A) (5). It is conceivable that the intein penultimate residue, His 453 , may assist the function of Asn 454 . The effects of single and double substitutions of His 453 were therefore examined in the MYK fusion system (Table IV). Analysis of the purified proteins from the single substitution, H453L, indicated partial in vivo splicing and in vivo N-terminal cleavage (Table IV). Incubation of the purified precursors at pH 8.0 resulted in no in vitro splicing; however, significant cleavage at both splice junctions occurred in the presence of DTT (Table IV). Efficient splicing was observed when the H453L substitution was combined with the C455S substitution (Fig. 4). Expression of pMYK(H453L/ C455S) resulted in the production of a major 125-kDa protein corresponding to the linear precursor MYK (Fig. 4, lane 1). A slowly migrating component corresponding to the branched intermediate (MYK*) was also observed along with splicing products, MK and Y (Fig. 4, lane 1). Incubation of the purified proteins at pH 8.5 resulted in significant in vitro splicing of the linear precursor MYK, as indicated by an increase of MK and Y (Fig. 4, lanes 7-10). The rate of splicing was much lower at pH 6.0 as only a slight increase of MK and Y was observed (Fig. 4,  lanes 1-6). The amount of branched intermediate (MYK*) remained unchanged after incubation at pH 6.0 but decreased rapidly at pH 8.5 (Fig. 4), suggesting that the branch resolution reaction was favored at pH 8.5 but inhibited at pH 6.0. Although more in vivo splicing was observed resulting in less linear precursors, the double substitutions, H453Q/C455S, produced similar results as described above (data not shown). In comparison, the double substitutions, H453A/C455S and H453F/C455S, blocked splicing in vivo and allowed partial splicing and some accumulation of the branched intermediate in vitro (Table IV, (Table IV). However, efficient cleavage at both splice junctions could occur in vitro in the presence of DTT (Table IV). This DTT-induced cleavage at both splice junctions was further examined in the MYT4 fusion system as described below.
Induction of C-terminal Cleavage in Vitro-Expression of the fusion construct pMYT4 (H453Q/C455A) resulted in the production of a single 150-kDa protein corresponding to the linear precursor MYT4 (Fig. 5A, lane 1). The MYT4 precursors were very stable in vitro as no significant cleavage at both splice junctions was observed after incubation at 4°C for 72 h or 23°C for 16 h (Fig. 5, A and B, lane 2). However, treatment of MYT4 with DTT at 23°C immediately induced cleavage at the intein N terminus to yield YT4 and M (Fig. 5A, lanes 3-5). In addition, cleavage at the intein C terminus was observed, as indicated by the appearance of T4 and Y (Fig. 5A, lanes 4 -10). The amount of YT4 declined after an initial increase (Fig. 5A,  lanes 3-10) suggesting that T4 and Y were produced from the C-terminal cleavage of YT4 and that the C-terminal cleavage occurred after the initiation of the N-terminal cleavage. Consistent with this explanation, no significant amount of MY, the product of an exclusive C-terminal cleavage, was observed (Fig. 5A).
To examine the effect of pH on C-terminal cleavage, MYT4 was first incubated with DTT at pH 6.0. As shown in Fig. 5B, the incubation resulted in efficient cleavage at the intein N terminus, yielding YT4 and M, but no significant cleavage at the C terminus (lanes 3 and 4). However, efficient C-terminal cleavage occurred after incubation of the DTT-treated samples for additional 16 h at pH 8.0 (yielding T4 and Y) (lane 5). The C-terminal cleavage appeared to occur in YT4 but not in MYT4 as indicated by a decrease of YT4 and no apparent change of a Protein splicing or cleavage in vivo or in the course of purification was determined by inspection of the splicing or cleavage products in crude extract, flow-through, and amylose-purified proteins resolved by SDS-PAGE. b In vitro splicing or cleavage was determined by comparing the splicing or cleavage products after incubation of the amylose-purified proteins at 23°C for 16 h without or with 40 mM DTT (ϩDTT).  4 and 5). Since DTT was removed by dialysis prior to the additional 16-h incubation at pH 8.0, the result suggests that the higher pH, not DTT, was required for the C-terminal cleavage. Additionally, the C-terminal cleavage was inhibited at pH 6.0 as incubation of MYT4 or YT4 at pH 6.0 resulted in no significant C-terminal cleavage (lanes 3 and 4). Similar results were obtained when free cysteine was used as a nucleophile to induce N-terminal cleavage (Fig. 5B, lanes 6 -9). As a control, the MYT4 precursors were incubated at either pH 6.0 or pH 8.0 in the absence of thiols (DTT or cysteine). No significant cleavage at either splice junction was observed (Fig. 5B,  lanes 1 and 2). To ascertain whether or not the C-terminal cleavage was coupled to succinimide formation by Asn 454 , DTTinduced cleavage was conducted with purified precursors from pMYT4 (H453Q/N454A/C455A). As expected with the N454A substitution, incubation of the purified precursors at either pH 6.0 or pH 8.0 resulted in only N-terminal cleavage (Fig. 5B,  lanes 10 -13), indicating that Asn 454 is required for C-terminal cleavage (Fig. 1A).

Effect of Deletion and Insertion on Protein Splicing-We
have demonstrated efficient splicing in p⌬MYB1(NG) in which the endonuclease domain of the intein was removed by deletion (15). To determine if the intein deletion mutant could be modified to undergo cleavage reactions similar to the full-length intein in the MYB fusion system, the N454A substitution was introduced in p⌬MYB1(NG) to yield p⌬MYB1(NG) (N454A). Expression of p⌬MYB1(NG) (N454A) resulted in the production of a predominant linear precursor (Fig. 6A, lane 1). Induction of N-terminal cleavage by DTT was conducted at different temperatures. As shown in Fig. 6A, incubation of the purified precursors with DTT at 4 and 16°C resulted in significant N-terminal cleavage, yielding M and ⌬YB(NG) (N454A) (lanes  2-7), whereas incubation at 23°C blocked DTT-induced cleavage (lanes 8 -10). As a control, the linear precursors were incubated without DTT at 4 -23°C for 16 h, resulting in no significant N-terminal cleavage (data not shown). The double substitution, H453Q/C455A, which led to the DTT-induced splice junction cleavage in MYT4 (Fig. 5), was also introduced in the p⌬MYB1(NG) construct. Incubation of the purified precursors with DTT resulted in no significant cleavage at either splice junction (data not shown), suggesting that the H453Q/ C455A substitution may disrupt the structure of the intein deletion mutant for efficient cleavage.
To determine whether we could replace the endonuclease domain of the Sce VMA intein with other protein domains without significantly affecting splicing, the gene for BamHI was inserted into the deletion site of p⌬MYB1(NG) to yield p⌬MYB1(Bam). Following the expression of the fusion construct in E. coli, the proteins were purified by amylose affinity chromatography and analyzed by SDS-PAGE and Western blot analysis. As shown in Fig. 6B, the major component was a 51-kDa protein corresponding to the ligated exteins, MB (lane 4), indicating that efficient splicing occurred in vivo. The identity of MB was verified by its ability to bind chitin (lane 5) and by Western blot analysis using antibodies raised against the maltose-binding protein (data not shown). The other splicing product, a chimeric fusion of the intein splicing domain and BamHI (⌬Y(Bam)), was detected in the crude extract as a 60-kDa protein (Fig. 6B, lane 2) and by Western blot analysis (Fig. 6B, lane 1). ⌬Y(Bam) reacted specifically with antibodies raised against BamHI (Fig. 6B, lane 1) and the Sce VMA intein (data not shown). Some minor components with high molecular masses were also detected in the amylose-purified proteins (Fig. 6B, lane 4), the 2% SDS eluates (Fig. 6B, lane 5), and by Western blot analysis (Fig. 6B, lane 1), suggesting that they were probably unspliced precursors and splicing intermediates. In conclusion, efficient splicing occurred after the endonuclease domain of the Sce VMA intein was replaced with the restriction endonuclease BamHI. Thus, insertion of a heterologous protein domain in the Sce VMA intein was possible without significantly affecting the splicing activity.

DISCUSSION
In this study, we present a comprehensive analysis of the first three reactions in the protein splicing pathway of the Sce VMA intein: N-S acyl rearrangement, trans-esterification, and succinimide formation coupled to C-terminal cleavage (Fig.  1A). By substituting the Ϫ1 residue and the intein penultimate residue, we were able to attenuate the splicing process and allow investigation of each splicing reactions in in vitro systems. Appropriate substitution of the Ϫ1 residue enabled us to examine in vitro splicing of the Sce VMA intein containing all wild-type catalytic residues. Induction of branch resolution and C-terminal cleavage in vitro was made possible by substitution of the intein penultimate residue in conjunction with substitution of the first C-extein residue. In addition, the study of the effect of insertion and deletion provides the first evidence suggesting that intein can tolerate the insertion of a foreign protein domain and that the structure of the intein for protein splicing may be destabilized by the deletion or insertion mutation. Our data yield further insights into the mechanism of protein splicing and strategies for modulating protein splicing reactions.

Roles of the Last N-extein Residue and Modulation of Protein
Splicing with Ϫ1 Substitutions-Efficient splicing occurs when inteins are transferred into heterologous proteins, suggesting that the inteins plus the first C-extein residue contain sufficient information for catalyzing splicing reactions (3,5,7,23,24). On the other hand, it is conceivable that splicing in a foreign context can be affected by the proximal extein sequences since the catalytic residues of protein splicing are located at the intein termini (4,5). It has been recently proposed that splicing of the Sce VMA intein involves interactions between the intein residues upstream of the C-terminal splice junction and the proximal extein residues upstream of the N-terminal splice junction (25). The N-S acyl rearrangement initiates protein splicing, forming a thioester bond between Cys 1 and the Ϫ1 residue (5). Glycine is the Ϫ1 residue in the 69-kDa vacuolar ATPase subunit of S. cerevisiae in which the Sce VMA intein is embedded prior to protein splicing (26). The question of how substitution of the Ϫ1 residue may affect protein splicing has not been examined by previous studies. In this study, we addressed this question by substituting the Ϫ1 residue with each of the 20 naturally occurring amino acids.
Based on their effects on N-terminal cleavage and/or the N-S acyl rearrangement, the 20 Ϫ1 substitutions in the MYB fusion system can be divided into four groups. The first group of substitutions, including the Gly Ϫ1 , Ala Ϫ1 , Ile Ϫ1 , Ser Ϫ1 , Gln Ϫ1 , and His Ϫ1 substitutions, resulted in efficient (30 -95%) DTTinduced N-terminal cleavage of the native precursors but almost no (Ͻ5%) cleavage of the urea-denatured precursors (Table III). These substitutions allowed the N-S acyl rearrangement to favor peptide bond formation prior to DTT treatment. Incubation with DTT shifted the N-S equilibrium by cleaving the thioester intermediate and inducing the formation of a thioester bond. The second group of substitutions, including the Cys Ϫ1 , Asn Ϫ1 , and Pro Ϫ1 substitutions, blocked DTT-induced N-terminal cleavage of both native and urea-denatured precursors (Table III). One possible explanation is that these substitutions blocked the formation of thioester bond in the N-S equilibrium. Alternatively, these substitutions may cause a structural change in the active site that prevents the access of DTT to the thioester bond. The third group of substitutions includes the Val Ϫ1 , Leu Ϫ1 , Thr Ϫ1 , Glu Ϫ1 , Lys Ϫ1 , Phe Ϫ1 , Tyr Ϫ1 , Trp Ϫ1 , and Met Ϫ1 substitutions (Table III). Significant amounts of the DTT-induced N-terminal cleavage occurred in both native and urea-denatured precursors. N-terminal cleavage was also observed when the precursors were denatured in urea at high pH without DTT treatment (Fig. 2) or denatured in 6 M guanidinium HCl followed by DTT treatment (data not shown). The data suggest that these substitutions caused a significant shift in the N-S equilibrium, resulting in accumulation of thioester intermediates prior to DTT treatment. These thioester intermediates could include both linear precursors and branched intermediates, as cleavage of both results in the identical products (i.e. M and YB) (Fig. 1A). The MYB fusion proteins contained the wild-type first C-extein residue Cys 455 and, therefore, were capable of forming the branched intermediate (Fig. 1A). As shown in Fig. 2, there was a component in the untreated samples migrating slower than the expected linear precursor. Whether this slow-migrating component was the branched intermediate or the result of protein aggregation requires further investigation. Assuming that urea denaturation immediately disrupted the N-S equilibrium and no further thioester or peptide bond formation occurred during denaturation, the percentage of N-terminal cleavage of the ureadenatured precursors (Table III) may represent the percentage of thioester intermediates in the native fusion precursors. Nevertheless, without detailed kinetic studies, we cannot rule out the possibility that these substitutions might cause accumulation of thioester intermediates during denaturation. The fourth group of the Ϫ1 substitutions, including the Asp Ϫ1 and Arg Ϫ1 substitutions, caused substantial N-terminal cleavage in vivo or in the course of purification (Table III). It is possible that these substitutions may disrupt the structure of the splice junctions, resulting in proteolytic cleavage or hydrolysis.
The majority of the Ϫ1 substitutions in the MYK fusion system had no significant effect on protein splicing even though many substitutions had been shown to shift the N-S equilibrium in the MYB fusion ( Fig. 3A and Table III). Nevertheless, certain Ϫ1 substitutions clearly disrupted the course of normal protein splicing, and their effects on proteins splicing paralleled their effects on the N-S equilibrium (Table III). For instance, the Pro Ϫ1 substitution, which blocked DTT-induced N-terminal cleavage in the MYB fusion, also inhibited protein splicing in the MYK fusion. The Val Ϫ1 , Leu Ϫ1 , and Ile Ϫ1 substitutions, which retarded the thiol-induced N-terminal cleavage in the MYB fusion, resulted in mostly unspliced precursors and/or C-terminal cleavage products in the MYK fusion (Table  III). The effects of the Asn Ϫ1 and Cys Ϫ1 substitutions were somewhat different. Although DTT-induced N-terminal cleavage was almost completely blocked by the Asn Ϫ1 and Cys Ϫ1 substitutions in the MYB fusion, efficient splicing in the MYK fusion occurred in vitro (Table III and Fig. 3B). A combination of the N454A mutation and the Asn Ϫ1 or Cys Ϫ1 substitution in the MYB fusion resulted in inhibition of the N-terminal cleavage, whereas restoration of the intein residue 454 to the wildtype residue Asn in the MYK fusion in conjunction with the Asn Ϫ1 or Cys Ϫ1 substitution allowed completion of protein splicing reactions. In conclusion, our results indicate that glycine at the Ϫ1 position of the Sce VMA intein is not essential for efficient protein splicing and can be substituted by other amino acid residues. A subset of amino acids at the Ϫ1 position shifts the equilibrium position of the N-S acyl rearrangement thereby affecting protein splicing. Appropriate substitutions of the Ϫ1 residue enabled us to modulate the rate of protein splicing and convert splicing into C-terminal cleavage.
Effect of Substituting the Intein Penultimate Histidine-Substitution of the penultimate histidine of the Psp pol-1 intein (His 536 ) has been shown to block both splicing and C-terminal cleavage and lead to accumulation of branched intermediate (4). Substitution of the penultimate histidine of the Sce VMA intein (His 453 ), on the other hand, produced different results. Neither splicing nor cleavage was completely inhibited by the H453L substitution (Table IV). Double substitutions, such as H453L/C455S, are required in order for efficient splicing to occur without in vivo cleavage. The observations of the accumulation of a branched intermediate and in vitro splicing of the precursors in the double substitutions, H453L/C455S (Fig. 4) and H453Q/C455S (data not shown), suggest that the His 453 substitution attenuates but does not block succinimide formation or C-terminal cleavage, whereas the C455S substitution may help to allow normal splicing to proceed without side reactions. In vitro protein splicing was less efficient when His 453 was substituted with Ala or Phe (i.e. in H453A/C455S and H453F/C455S) (Table IV). Splicing was completely blocked, and efficient DTT-induced cleavage was observed when Cys 455 was substituted with Ala to block the trans-esterification in the double substitution, H453Q/C455A (discussed below). It is apparent that substituting the intein penultimate histidine residues in conjunction with the first C-extein residue provides an alternative for modulating protein splicing and converting splicing to cleavage. These results illustrate the intricate coordination of the active site residues of the Sce VMA intein in catalyzing protein splicing and the possible role of His 453 in assisting succinimide formation and branch resolution.  (Table IV) suggesting that, albeit at an attenuated rate, the H453Q substitution could still allow Cterminal succinimide formation and cleavage. However, C-terminal cleavage in the H453Q/C455A double mutant was completely blocked resulting in isolation of linear unspliced precursors (Fig. 5). As the C455A substitution rendered the mutant incapable of undergoing trans-esterification, the results suggested that by inducing N-terminal cleavage to mimic trans-esterification, we were be able to induce C-terminal cleavage. This was indeed the case when the linear precursors carrying the H453Q/C455A double substitution were treated with DTT or cysteine to generate both N-terminal and C-terminal cleavage products (Fig. 5). Unlike thiol-induced N-terminal cleavage (9), C-terminal cleavage was independent of thiols as it could occur after the removal of thiols and was inhibited at pH 6.0 but not at pH 8.0 (Fig. 5B). In addition, no C-terminal cleavage was observed before thiol treatment (Fig. 5A, lanes 1  and 2), suggesting that C-terminal cleavage occurred after the induction of N-terminal cleavage. We speculate that thiol-induced N-terminal cleavage in the H453Q/C455A double mutant triggers a conformational change in the intein structure thereby allowing succinimide formation and C-terminal cleavage to proceed at basic pH. Whether or not this reflects the actual protein splicing of the wild-type Sce VMA intein remains unanswered. Nevertheless, the induction of cleavage at both splice junctions by exogenous cysteine simulated each step of the splicing reactions as follows: the exogenous cysteine functioned as Cys 455 to attack the thioester bond formed by the N-S acyl rearrangement; C-terminal cleavage occurred after the N-terminal cleavage; and the exogenous cysteine formed a covalent bond with the N-extein through an S-N acyl rearrangement (Fig. 1A) (5).

Induction of C-terminal Cleavage and Simulation of Protein
Further Examination of the Mechanism of Protein Splicing of the Sce VMA Intein-Studies of protein splicing of a thermophilic archeal intein, i.e. Psp pol-1 intein, have been facilitated by the fact that in vivo splicing of the wild-type intein in a foreign protein context could be attenuated by low growth temperatures (12-15°C) (3,27). Consequently, sufficient precursors could be isolated to demonstrate in vitro splicing (3). The Sce VMA intein, on the other hand, underwent rapid splicing in vivo even at low growth temperatures (5, 7). We utilized amino acid substitutions of catalytic residues of the Sce VMA intein to block or attenuate the splicing process to isolate sufficient precursors and intermediates (5). Since the mechanism of protein splicing was deduced from analyses of mutants containing altered catalytic residues (5, 7), little is known about the close coupling and intricate balancing of the wildtype splicing reactions of the Sce VMA intein. In this study, we modulated protein splicing through appropriate substitutions of the Ϫ1 residue and the intein penultimate residue. As a result, we were able to examine protein splicing of the Sce VMA intein without altering its essential catalytic residues. In particular, we focused on the effect of pH on splicing reactions. Thiol-induced cleavage at the N terminus proceeded efficiently at pH values ranging from 5.5 to 9.0, but the rate of cleavage was significantly higher at pH 8.0 than pH 6.0 (Table II) (9). Similarly, our results suggest that branch resolution (Fig. 4) and C-terminal cleavage (Fig. 5B) were inhibited at pH 6.0 but proceeded efficiently at pH 8.0 -8.5. It is possible that the nucleophilic displacements at the intein upstream and downstream splice junctions are assisted by similar groups of residues that act as acid or base catalysts. Completely different pH effects were observed when in vitro splicing of the fusion precursors from pMYK(Asn Ϫ1 ) was examined. The rate of splicing was higher at pH 6.0 than pH 8.0 (Fig. 3B). As no catalytic residues of the intein in pMYK(Asn Ϫ1 ) were altered, it is possible that a similar pH profile may also apply to protein splicing of the wild-type intein in vivo. Since it has been shown that the N-S (or N-O) rearrangement favors ester bond formation at low pH and amide bond formation at high pH (28,29), the results suggest that the rate of overall splicing is primarily determined by the N-S acyl rearrangement. Consistent with this explanation, we have shown that thiol-induced N-terminal cleavage (Table II), branch resolution (Fig. 4), and C-terminal cleavage (Fig. 5B) proceed more efficiently at high pH than low pH. Therefore, both trans-esterification and branch resolution are unlikely to be the rate-limiting step in the protein splicing pathway. In addition, we were able to modulate the rate of protein splicing by substituting the Ϫ1 residue that directly affects the N-S acyl rearrangement (Table II). It has been shown that the rate of splicing for the thermophilic Psp pol-1 intein is also favored at pH 6 and inhibited at pH 9 or above (3).
Our data are consistent with the proposal that both the thermophilic Psp pol-1 intein and the mesophilic Sce VMA intein follow the same protein splicing pathway (5).
Effect of Deletion and Insertion in the Sce VMA Intein-Previously, we deleted the central region of the Sce VMA intein and showed that the remaining intein structure (splicing domain) is sufficient to catalyze protein splicing (15). In this study, we examined how the deletion might affect the structure and catalysis of the splicing domain. Our results indicated several differences between the deletion mutant (⌬Y(NG)) and the full-length intein. First, splicing efficiency of the intein deletion mutant was affected by heterologous extein domains. For instance, when the intein deletion mutant was fused between MBP and CBD in ⌬MYB(NG), more than 80% of the fusion precursors spliced in vivo (15). However, changing the C-extein into the E. coli thioredoxin (T) in ⌬MYT (NG) resulted in only 50% of the precursors splicing in vivo (data not shown). In comparison, the full-length intein underwent complete splicing in vivo in all fusion systems that we constructed (i.e. MYB, MYT, MYK, and MYT4) (5). Second, the induction temperature for protein expression affected protein splicing of the deletion mutant but not the full-length intein. In vivo splicing of the intein deletion mutant was efficient when induction of protein expression of p⌬MYB(NG) in E. coli was conducted at 15-20°C but was completely blocked when the induction was at 30°C or above (data not shown). The full-length intein, on the other hand, catalyzed efficient splicing in vivo at induction temperatures ranging from 15 to 37°C (5,9). The sensitivity of the intein deletion mutant to induction temperature could be due to the effects of the deletion on protein folding in vivo. Consistent with this explanation, expression of p⌬MYB(NG) at induction temperatures of 30°C or above resulted in only unspliced linear precursors incapable of undergoing in vitro splicing or cleavage (data not shown). Third, the amino acid substitutions that resulted in thiol-induced splice junction cleavage of the full-length intein had different effects on the intein deletion mutant. The rate of DTT-induced cleavage of the full-length intein with the N454A substitution increased significantly upon increase of the incubation temperature (between 4 and 23°C) (Table II). In contrast, DTT-induced cleavage of the deletion mutant was efficient only at 4 -16°C and was completely inhibited at 23°C (Fig. 6A). It is possible that the deletion mutation destabilized the structure of the deletion mutant, and consequently an increase in temperature perturbed the alignment of catalytic residues. The active site structure of the full-length intein, on the other hand, was more stable, and higher temperatures simply increased the rate of trans-esterification reaction by DTT.
Based on the crystal structure, Duan et al. (6) hypothesized that the Sce VMA intein may have evolved from a composite gene that resulted from the invasion of an endonuclease open reading frame into a pre-existing gene encoding a protein splicing element. In this study, we recreated this gene invasion scenario through genetic engineering. The gene encoding the endonuclease BamHI was inserted into the Sce VMA intein replacing the native endonuclease domain. Subsequently, the chimeric fusion was allowed to splice in ⌬MYB(Bam) resulting in Ͼ80% splicing in vivo (Fig. 6B). Although no endonuclease activity was detected in the spliced ⌬Y(Bam) (data not shown), possibly due to misfolding of BamHI or the intein moiety interfering with dimerization and/or DNA binding of BamHI, the result clearly demonstrated the feasibility of an endonuclease gene invasion of a protein splicing element. Other heterologous proteins, e.g. chitin-binding domain (CBD) from B. circulans (22) and green fluorescent protein (GFP) from Aequorea victoria (30), have also been inserted into the Sce VMA intein, and between 50 and 95% splicing efficiencies were obtained. 2 In addition, both CBD and GFP in the spliced chimeric fusions appeared to fold correctly as the CBD fusion was able to bind chitin and the GFP fusion was fluorescent. 2 Efficient splicing of the Sce VMA intein with insertion of different heterologous protein domains suggested a tightly packed structure in the intein splicing domain. This is consistent with the crystal structure of the Sce VMA intein which indicates that domain I, essentially the same as the splicing domain in our deletion mutant, consists of predominantly closely packed ␤-sheets (6).
Perspectives-The work presented here has advanced our understanding of the mechanism of protein splicing. As analogous reactions have been found in the autoprocessing of some biologically important proteins, e.g. hedgehog proteins (31)(32)(33), glycosylasparaginases (34), etc., our results should enhance the understanding of the biological functions of these processes. Although the crystal structure of the excised Sce VMA intein has been solved, the active site structure at the extein-intein junctions prior to splicing has yet to be determined. It is apparent that complete characterization of the residues involved in four nucleophilic displacements of the splicing pathway requires further mutagenesis and crystallographic studies.
Protein splicing illustrates a highly specific and efficient way to cleave and religate peptide bonds. It is possible that all inteins may share a similar splicing pathway. As more inteins are being identified in various organisms (2), our study of the Sce VMA intein may reveal common strategies for modulating the rate of protein splicing, converting splicing into efficient and controllable peptide cleavage, and re-engineering inteins through deletion and insertion. The results presented in this study suggest many potential applications. The endonuclease activity of an intein is responsible for the intein-mediated gene mobility that allows site-specific insertion of an intein into an intein-less allele (12). By replacing the endonuclease domain of an intein with an endonuclease of different specificity, we can potentially redirect intein lateral transmission to a designated site. Intein insertion into a heterologous protein could be used to regulate its biological function through protein splicing. Our studies of the modulation of protein splicing and the effect of the Ϫ1 substitutions may provide valuable information regarding the choice of insertion sites and the strategy for conditional gene knock-out. Our results also suggest that inteins may tolerate insertion of heterologous protein domains. By inserting a functional protein domain (e.g. GFP, transcriptional factors, protein kinases, etc.) into an intein without diminishing its splicing activity, we may be able to monitor in situ protein splicing processes or link protein splicing to other biological events. The thiol-induced N-terminal cleavage of the Sce VMA intein has been used for protein purification in which the target protein is fused to the N terminus of the intein (9). Alternatively, we can use the C-terminal cleavage activity to purify a target protein fused to the C terminus of the intein. One advantage of the C-terminal fusion is that the N terminus of the fusion protein can be optimized for high level protein expression. With similar modifications, inteins from other organisms can also be used for protein purification or other intein-based applications. As nature reveals its unique ways to break and join peptide bonds, we will undoubtedly discover more usage of this novel class of proteins.
Note Added in Proof-The crystal structure of GyrA intein has recently been solved that showed the active site structure of the intein plus Ala as the Ϫ1 residue (35).