Structure of an engineered intein reveals thiazoline ring and provides mechanistic insight

Abstract

We have engineered an intein which spontaneously and reversibly forms a thiazoline ring at the native N-terminal Lys-Cys splice junction. We identified conditions to stablize the thiazoline ring and provided the first crystallographic evidence, at 1.54 Å resolution, for its existence at an intein active site. The finding bolsters evidence for a tetrahedral oxythiazolidine splicing intermediate. In addition, the pivotal mutation maps to a highly conserved B-block threonine, which is now seen to play a causative role not only in ground-state destabilization of the scissile N-terminal peptide bond, but also in steering the tetrahedral intermediate toward thioester formation, giving new insight into the splicing mechanism. We demonstrated the stability of the thiazoline ring at neutral pH as well as sensitivity to hydrolytic ring opening under acidic conditions. A pH cycling strategy to control N-terminal cleavage is proposed, which may be of interest for biotechnological applications requiring a splicing activity switch, such as for protein recovery in bioprocessing.

Keywords: cleavage control; molecular switch; protein splicing; thiazoline crystal structure.

Figure 1.. Splicing pathway and first reaction step for the RecA intein.

(A) Schematic sequence for the RecA intein with N- and C-exteins. Only the residues with numbers and motif locations of interest are shown. Vertical arrows identify the N- and C-terminal splicing junctions. This construct represents a RecA mini-intein from which the endonuclease has been deleted between block B and F. (B) Overall spicing pathway: The process begins with a nucleophilic attack on the adjacent peptide bond in the precursor (P) by Cys1 of the intein (first step, shaded), which subsequently undergoes a transesterification mediated by Cys+1 of the C-extein, resulting in a branched intermediate (BI). The intein is then excised via cyclization of the terminal Asn. The thioester linking the exteins is resolved to a peptide bond by nucleophilic attack of the C-extein amine. (C) Detailed pathway for the first step of splicing (shaded). This step results in formation of a thioester (iv) from a free thiol (i), as well as a postulated aberrant pathway, which results in formation of a thiazoline ring (vi). The fate of the tetrahedral intermediate (ii, TI) is dependent on protonation of the nitrogen (iii) or oxygen (v) of this intermediate. TE, thioester; TR, thiazoline ring.

Figure 2.. Electrospray ionization (ESI) mass spectrum of RecA intein with T70A/N139A mutations.

An 18-Da mass reduction of the T70A/N139A mutant represents dehydration associated with thiazoline ring protonation. For the sake of clarity, only the region containing the 7+ charge state is shown. The most intense signal (arrow) corresponds to thiazoline containing intein (monoisotopic M_obs = 16,454.60 Da versus M_calc = 16,454.72 Da calculated from the sequence). The expected mass of the T70A/N139A mutant without the thiazoline ring was not detected (M_calc of 16,472.73 Da). Additional signals were detected for corresponding T70A forms in which Arg-to-Lys (K-R) substitutions were induced during protein expression, which resulted in a mass decrease of 28 Da. Typical salt adducts were detected for Na⁺, K⁺ and Na⁺/K⁺ (DM of +22, +38, and +60 Da, respectively. It should be noted that the high resolution afforded by these determinations allowed us to resolve the isotopic envelope of each detected species. In this way, it was possible to determine the respective charge state from the mass-over-charge ratios (m/z) spacing between contiguous isotopic signals: z = 1/D_m/z, where z is the charge state and D_m/z is the m/z spacing. Inset shows the structure of the thiazoline ring. This construct had the His tag removed by proteolytic cleavage.

Figure 3.. Crystal structure of the ΔΔI-TM intein with the thiazoline ring.

(A) Electron density map of a thiazoline ring. The ring is formed between the first Cys residue of the intein (C1) and the preceding N-extein Lys residue (K-1) in the ΔΔI-TM (T70A/N139A) construct. (B) and (C) Comparison of H-bonding in the crystal structures of spliced and thiazoline ring containing intein, respectively, with faint background structures comparing T70 with A70. (B) In the native intein the T70 backbone amide is H-bonded to the Cys1 carbonyl carbon, and the sidechain OH forms H-bonds with both the Cys1 carbonyl carbon and amide nitrogen. (C) In the thiazoline mutant the A70 sidechain H-bond to the Cys1 carbonyl oxygen is absent and, interestingly, the H-bond to the amide nitrogen is replaced by a water, which is not present in the thiazoline-less crystal structure. (D) Location of critical residues in the vicinity of the N-terminus of the intein. Asp121 corresponds to Asp422 in the full-length RecA intein (Van Roey et al. 2007; Wood et al. 1999).

Figure 4.. Characteristics of RecA mini-intein mutants investigated by ESI-MS.

Full mass range spectra on which masses are based are shown in the Supplementary Figs. S3–S9. From the top, T70 wild type with Thr at position 70 prevents thiazoline ring formation and restores N-terminal cleavage competence (Fig. S3). Wild type Asn at the ultimate 139 position restores C-terminal cleavage but does not prevent thiazoline ring formation (Fig. S4). The T70A, T70V, T70C, and K-1G mutants with N139A all spray as solely thiazoine ring-containing species (Figs. S5, S6, S8 and S9, respectively). Interestingly, the T70S spectra contain both thiazoline ring-containing precursor and N-terminal cleavage product, suggesting partial rescue of the thiazoline phenotype (Fig. S7). The His tag (part of the N-extein) was an integral component of the constructs, explaining the increased masses reported over those in Figures 2 and 6 (see Materials and Methods for sequence).

Figure 5.. Conservation of residues in the TxxH motif.

(A) Canonical splicing inteins. Thr is highly conserved in inteins that splice by the canonical mechanism, replaced occasionally by Ser or Asn. These alternate residues, like Thr, are able to form hydrogen bonds which are expected to play an important role in orientation of the tetrahedral intermediate. (B) Non-canonical splicing inteins. Thr conservation is significantly reduced in inteins that splice by alternate mechanisms. Source references (Novikova et al. 2016; Perler 2002).

Figure 6.. Stability of thiazoline ring as a function of pH and time.

(A) Thiazoline ring stability by mass spectrometry analysis. Red bars denote thiazoline ring-containing species and blue bars denote hydrolyzed species. The ring is highly stable at 4°C and at pH 7 (i) but hydrolyzes at acidic pH 2 for 15 min (ii), and at pH 2 for 3 h (iii). Calculated monoisotopic mass of thiazoline-ring containing protein: 16,454.72 Da; the calculated monoisotopic mass of hydrolyzed protein: 16,472.73 Da. (B) Kinetic profiles of ring hydrolysis at pH 7 (i), 4 (ii), and 2.5 (iii) for 7 days. (C) Fraction of thiazoline ring after 6.5 h as a function of solution pH. This construct had the His tag removed by proteolytic cleavage as in Fig. 2.

Figure 7.. Behavior of two hydrogen bond-forming mutants in the TxxH motif.

Hydrogen bond-forming mutants (T70S/N139A and T70D/N139A) with the His tag as part of the N-extein, exhibit a propensity to form thiazoline ring but retain the ability to form a thioester when thiazoline ring is hydrolyzed at 4°C. The labile thioester formation is identified by DTT-induced cleavage. Lack of cleavage by TCEP, a non-nucleophilic reducing agent, clarifies that trapping is thiazoline based and not redox based. Cleavage incompetent (C1A) and fully N-terminal cleavage competent (native T70/N139A) inteins serve as controls at 4°C. BI = branch intermediate, P = precursor, C = cleavage product.

Figure 8.. Proposed separation protocol: pH cycling to produce a desired product.

We propose that the T70A mutant is used to prevent N-terminal cleavage via the formation of a thiazoline ring. A fusion construct (i) comprising a desired protein (N-ext), a modified mini-intein (int) and a binding domain (C-ext, shown in Fig. 1 but omitted here) for later purification, is over-produced in a bioreactor at pH 7–8. The pH of the solution is then reduced to 2 to hydrolyze the thiazoline ring (ii) and then increased to 7.4 with addition of a nucleophile (e.g. DTT, hydroxylamine etc) to induce irreversible cleavage to obtain the desired product (iii). For those molecules that did not cleave but formed thiazoline ring structures (i), the solution is again dropped to pH 2 to re-enter the cycle. The product-containing solution with the free N-ext plus the int/C-ext fusion (iii) is then passed through an affinity column or membrane to remove the fusion and purify the product of interest.