Traceless protein splicing utilizing evolved split inteins Steve W. Lockless1 and Tom W. Muir2 Laboratory of Synthetic Protein Chemistry, The Rockefeller University, New York, NY 10065 Edited by James A. Wells, University of California, San Francisco, CA, and approved May 7, 2009 (received for review March 18, 2009) Split inteins are parasitic genetic elements frequently found inserted into reading frames of essential proteins. Their association and excision restore host protein function through a protein self-splicing reaction. They have gained an increasingly important role in the chemical modification of proteins to create cyclical, segmentally labeled, and fluorescently tagged proteins. Ideally, inteins would seamlessly splice polypeptides together with no remnant sequences and at high efficiency. Here, we describe experiments that identify the branched intermediate, a transient step in the overall splicing reaction, as a key determinant of the splicing efficiency at different splice-site junctions. To alter intein specificity, we developed a cell- based selection scheme to evolve split inteins that splice with high efficiency at different splice junctions and at higher temperatures. Mutations within these evolved inteins occur at sites distant from the active site. We present a hypothesis that a network of conserved coevolving amino acids in inteins mediates these long-range effects. BIOCHEMISTRY directed evolution ͉ protein ligation ͉ Crk-II ͉ SCA ͉ coevolution rotein function is modulated by a variety of posttranslational Pmodifications, such as phosphorylation, ubiquitylation, and methylation (1). One of the most dramatic posttranslational mod- ifications is protein splicing, an autocatalytic process in which an intervening polypeptide sequence, termed an intein, is excised from a precursor protein with concomitant splicing of the franking sequences, known as exteins. Protein splicing has proven to be a highly versatile process for methods development and forms the Fig. 1. The mechanism of protein trans-splicing. (A) Schematic showing the hub of several protein engineering technologies (2). Inteins have basic steps in the intein-mediated protein trans-splicing reaction. IntN and IntC been successfully used in protein semisynthesis to create posttrans- refer to the N- and C-terminal parts of the split intein, N- and C-extein refer to the lationally modified proteins (2, 3), in the cyclization of peptides to N- and C-terminal parts of the spliced protein, and “CFN” refers to the amino-acid create small molecule toxins (4), in plant biotechnology to recon- sequence in the C-extein at the IntC boundary. (B) Schematic showing the bond struct proteins in vivo (5), in the segmental labeling of proteins for rearrangements catalyzed by the intein during protein splicing. NMR studies (6), and in protein semisynthesis in living cells (7). Inteins come in 2 flavors—cis splicing inteins are single polypep- the Crk-II proto-oncogenic adaptor protein from peptide frag- tides that are embedded in a host protein, whereas trans-splicing ments without the need to introduce mutations into the final inteins (herein called split inteins) are separate polypeptides that protein. In addition, this evolution scheme was used to generate mediate protein splicing after the intein pieces and their protein mutant split inteins that support efficient splicing at 37 °C and can cargo associate (8, 9) (Fig. 1A). Despite the many applications of be used in mammalian cell applications. Intriguingly, the location of split inteins in chemical biology and protein chemistry, they are the mutated amino acids in each case is distant from the active site plagued with various idiosyncratic parameters that limit their more and likely does not interact directly with extein amino acids. We general use. Most importantly, the 2 parts of the naturally split present a hypothesis for how these distant sites might affect the inteins associate and typically splice at a C-terminal junction function of amino acids in the active site. containing the canonical ‘‘CFN’’ tripeptide sequence (10, 11), Results which are the first 3 amino acids of the C-extein sequence (Fig. 1A). This tripeptide remains in the product after splicing, meaning that The Rate-Limiting Step of Splicing Using DnaE Split Inteins. We first it will most often be a mutant protein. It is currently unclear why set out to understand why protein trans-splicing is so inefficient this “CFN” sequence is required for efficient protein trans-splicing. In contrast, work from several laboratories, including our own, Author contributions: S.W.L. and T.W.M. designed research; S.W.L. performed research; indicates that the N-terminal splice junction is much more tolerant S.W.L. contributed new reagents/analytic tools; S.W.L. and T.W.M. analyzed data; and to noncanonical sequences (6, 11–13). S.W.L. and T.W.M. wrote the paper. In this study, we begin by identifying a key catalytic step in the The authors declare no conflict of interest. splicing reaction that is sensitive to the identity of the amino acids This article is a PNAS Direct Submission. at the C-terminal splice junction. To alter this rate-limiting step, we 1Present address: Department of Biology, Texas A&M University, 3141 Interdisciplinary Life describe a general cell-based selection scheme that allows the Sciences Building, College Station, TX 77845. directed evolution of mutant split inteins that can splice at nonca- 2To whom correspondence should be addressed. E-mail: [email protected]. nonical C-terminal splice junctions. We demonstrate that these This article contains supporting information online at www.pnas.org/cgi/content/full/ mutant inteins can be used for traceless protein splicing by creating 0902964106/DCSupplemental. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0902964106 PNAS Early Edition ͉ 1of6 Downloaded by guest on September 24, 2021 Table 1. Description of all constructs used in this study Protein Molecular weight Description Ssp intein 14.0 kDa (N-term), 3.9 kDa (C-term) N- and C-terminal inteins from Ssp DnaE Npu* intein 11.9 kDa (N-term), 3.9 kDa (C-term) N-terminal intein from Npu DnaE with C-terminal intein from Ssp DnaE mNpu* intein 11.9 kDa (N-term), 3.9 kDa (C-term) Npu* intein evolved with “SGV” (E15D, L25I, and L92M mutations in NpuN; D23Y mutation in SspC) mNpu37* intein 11.9 kDa (N-term), 3.9 kDa (C-term) Npu* intein evolved at 37°C (L25S and P21RC mutations) N H6-Ub-Ssp 24.4 kDa His6–tagged ubiquitin fused to the N-terminal intein from Ssp DnaE N H6-Ub-Npu 22.3 kDa His6–tagged ubiquitin fused to the N-terminal intein from Npu DnaE SspC-“CFN”-HA 5.2 kDa C-terminal intein from Ssp DnaE with “CFN” sequence preceding the HA tag SspC-“SGV”-HA 5.2 kDa C-terminal intein from Ssp DnaE with “SGV” sequence preceding the HA tag CrkII 38.3 kDa Full-length protein with domain structure of SH2-“SGV”-SH3-SH3 Flag-SH2-IntN-Ub 35.4 kDa Flag-tagged CrkII SH2 domain followed by the N-terminal intein domain and ubiquitin MBP-IntC-“SGV”-SH3-SH3-H6 68.2 kDa His-tagged CrkII SH3 domains preceded by MBP and the C-terminal intein domain when noncanonical sequences are placed at the C-terminal splice in the Ssp splicing reaction, the Npu* reaction led to the appearance junction. Split inteins catalyze a remarkable series of chemical of both the spliced product and the 16-kDa band, which was rearrangements that require the intein to be properly assembled confirmed as the branched intermediate by mass spectrometry (Fig. and folded (9) (Fig. 1B). The first step in splicing involves an N-S S2). We determined the rate constants for the branched interme- acyl shift in which the N-extein polypeptide is transferred to the side diate formation as well as its conversion to the spliced product for chain of the first residue of the intein. This is then followed by a both Ssp and Npu* by quantifying the bands on the gel and fitting trans-(thio)esterification reaction in which this acyl unit is trans- their values to the differential equations describing a sequential ferred to the first residue of the C-extein (which is either serine, kinetic reaction model (Fig. 2C). The rate of branched intermediate threonine, or cysteine) to form a branched intermediate. In the formation was similar between Ssp (0.32 Ϯ 0.02 hϪ1) and Npu* penultimate step of the process, this branched intermediate is cleaved from the intein by a transamidation reaction involving the C-terminal asparagine residue of the intein. This then sets up the A H Ub IntN final step of the process involving an S-N acyl transfer to create a 6 IntN + IntC -”SGV”- HA (22.3 kD) (11.9 kD) H6 Ub -”SGV”- HA + H Ub normal peptide bond between the 2 exteins. We monitored the 6 (11.8 kD) C overall protein splicing reaction using an in vitro protein trans- Int -”SGV”- HA (15.7 kD) (5.2 kD) splicing assay with purified proteins. Constructs were generated in which model N- and C-exteins were fused to the fragments of the B Ssp (anti-HA) Npu* prototypical DnaE split intein from Synechocytis sp. PCC6803 15kD BI (herein abbreviated to Ssp). In initial studies, we reacted an N 10kD SP N-terminal construct (H6-Ub-Ssp ) with a C-terminal construct 0:00 0:05 0:15 0:45 1:30 2:30 3:30 0:00 0:05 0:15 0:45 1:30 2:30 3:30 (hr:min) (SspC-“CFN”-HA) containing the canonical “CFN” sequence at the C-terminal splice junction (see Table 1 for a description of all C 800 800 constructs used in this study). As expected, splicing was found to be 600 600 highly efficient as read out by Western blotting against the HA tag 400 400 Density [supporting information (SI) Fig.