Modulation of intein activity by its neighboring extein substrates

Results

A System of FRET Loss to Monitor Intein Cleavage.

Protein splicing is a multistep process dependent on the sequential intein-catalyzed rearrangement of bonds flanking the intein. For identification of possible extein effects on intein activity, we constructed a system in which FRET-active CFP and YFP flanked native extein residues upstream and downstream of the intein. We term this construct C-I-Y (Fig. 1A). To enable study of the relationship between FRET loss and intein-mediated N-terminal cleavage, we created several modified intein variants: Ala substitution of the active site Asn-159 residue (N159A), called C-Ia-Y, retains N-terminal cleavage but abolishes splicing (13); and double Ala substitution C1A/N159A, called C-Iaa-Y, abolishes both cleavage and splicing (12) (Fig. 1B). The splicing competent, C-I-Y, and the inactive C-Iaa-Y constructs exhibit similar FRET values in vivo, presumably because of extein proximity in the precursor; however, the C-Ia-Y construct exhibits FRET loss (Fig. 2A). We also constructed 2 control variants resembling the N-terminal cleavage products: intein-YFP (I-Y) and CFP (C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

FRET-based N-extein cleavage assay. (A) FRET reporter construct. Splicing and N-extein cleavage pathways for a FRET-active intein precursor, CFP-Intein-YFP (C-I-Y), and relative FRET associated with the respective reaction products are shown. (B) Intein–extein structure relationship. Ssp PCC6803 DnaE intein (red) with its native N-2, N-1 (blue), C+2, and C+3 (yellow) extein residues, and catalytic residues Cys-1 and Asn-159, and Cys+1 residues (green) (modified from PDB:1ZD7).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

FRET assay reports N-extein cleavage in vitro. (A) Progress of N-extein cleavage. Recorded FRET values over 1 h with C-Ia-Y extracts and controls (C-Iaa-Y; C+IY, mixture of C and I-Y, purified separately) in the presence and absence of DTT (20 mM). (B) Mechanism of DTT-induced N-extein cleavage. (C) Gel analysis of N-extein cleavage. Unheated extracts of C-Ia-Y with (+) or without (−) DTT (20 mM) separated by 12% SDS/PAGE followed by fluorescence detection of YFP and CFP (excitation 457 nm, emission 526 nm; wavelengths overlap for CFP and YFP). Lane 1: BG, blank MC1061 lysate. Both bands in the doublet labeled “C” are CFP (lane 7), which migrates spuriously in native form on an SDS gel.

FRET values measured in vitro for the inactive C-Iaa-Y and the N-cleavage-competent C-Ia-Y were 10-fold higher than a C+I-Y (Fig. 2A). Addition of the nucleophilic 1,4-dithio-dl-threitol (DTT) as an initiator of N-terminal cleavage (Fig. 2B) resulted in a time-dependent loss of FRET with the active C-Ia-Y, but not the inactive C-Iaa-Y (Fig. 2A). Selective FRET loss is consistent with specific DTT cleavage at the intein-generated thioester bond that links C to I-Y (Fig. 2B) (14). Product analysis using SDS/PAGE corroborated the link between FRET loss and intein/extein cleavage (Fig. 2C and Fig. S2 A and B).

Similarly, in bacterial cells, the FRET signal was >3-fold lower for the cleavage-competent C-Ia-Y construct than the refractory C-Iaa-Y construct after subtraction of background cell autofluorescence (Fig. 3A, Inset). Together, these results validated the FRET approach for identifying extein residues that modify N-terminal intein cleavage in vitro and in vivo.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

FRET assay for in vivo screening and in vitro characterization of cleavage mutants. (A) Representative FRET screening of an extein mutant library. Dual emission plot of data collected from a live-cell expression library of the C-Ia-Y variants with randomly mutated N-2 and N-1 extein residues. (Inset) FRET measured in vivo for C-Iaa-Y, C-Ia-Y, and n-ND 6 h postinduction, relative to E. coli MC1061 (BG). (B) Extein effects on DTT-induced N-extein cleavage. Progress of N-extein cleavage of parental C-Ia-Y along with representative extein mutants that show suppressed (red) and accelerated (green) reactivity toward DTT in vitro. Mutants of intermediate phentoype are shown in black. Identified variants were further characterized. (C) Rates of DTT-induced N-extein cleavage for 30 variants with k_obs values that differ by >2-fold from the parental C-Ia-Y, representing 4–5% of total variants screened. Residue combination of library n refers to randomized N-2 N-1 positions, library c to C+2 C+3, and library nc to N-2 N-1 and C+2 C+3. Bold letters designate variants that were further characterized.

Extein Residues That Influence Intein Activity.

We engaged in high-throughput FRET-based screening of extein libraries after simultaneous mutation of all combinations of the N-2 and N-1 residues (n library, 400 combinations), the C+2 and C+3 residues (c library, 400 combinations), or both residue pairs (nc library, 160,000 combinations). After expression in Escherichia coli, extein libraries were screened in vivo for cleavage activity by excitation of liquid cultures at 400 ± 15 nm and monitoring of fluorescence emission at 540 ± 12.5 nm (YFP) and 485 ± 10 nm (CFP) in a 96-well format (Fig. 3A). The longer wavelength emission arises from the acceptor YFP and is due in part to FRET, whereas the shorter wavelength emission arises from the donor CFP. The ratio of these 2 emissions was used for FRET normalization across E. coli cultures. Variants having a higher FRET signal than C-Ia-Y within the cell are associated with reduced cleavage rate and vice versa for lower FRET. Extein mutant n-ND (E-2N, Y-1D), which we later demonstrated to be an accelerator of N-terminal cleavage, confirmed a lower in vivo FRET value (Fig. 3A, including Inset).

From the 3 libraries, a total of 74 variants, having either reproducibly lower or higher FRET values than that of native C-Ia-Y in vivo, were sequenced and their FRET-loss kinetics determined in vitro (Fig. 3B, Fig. S3A, and Table S1). N-extein cleavage assays were conducted by using the soluble fraction of lysed cells after addition of DTT to 20 mM. Rate constants (k_obs) were determined by fitting the change in FRET ratio to a single exponential based on 36 measurements collected over 1 h (Fig. 3B). For the majority of variants, FRET-loss curves were fit with a high correlation coefficient (r² = 0.99). However, the fit was compromised with accelerators for which extensive cleavage had occurred intracellularly (Fig. 3B, n-ND and c-SC; Fig. S3A, arrow).

DTT-induced N-terminal cleavage rates for the variants showed a wide distribution, with 30 of the 74 extein variants having k_obs >2-fold higher or lower than C-Ia-Y, with native exteins (Fig. 3C and Table S1). The k_obs value for DTT-induced cleavage with C-Ia-Y was 1.4 × 10⁻³ ± 0.03 s⁻¹, in agreement with previous reports of 1 × 10⁻³ s⁻¹ (13). The minimal k_obs for cleavage attenuation was 10⁻³-fold depressed at 1.6 × 10⁻⁶ s⁻¹, whereas the k_obs for maximal enhancement was elevated >4-fold at ≥5.7 × 10⁻³ s⁻¹.

In efforts to relate genotype to phenotype, we examined the 30 extein variants that had cleavage rates differing by >2-fold from the parental C-Ia-Y construct (Fig. 3C). Among the accelerating mutations from the n library, Asp and Arg were the most common at N-1; Asp had the most pronounced effect on cleavage rate (Fig. S3B). Mutations at N-2 had a less consistent effect, as reflected by a broader scatter of activities for each mutation. From the c library, Cys substitution at C+2 or at C+3 enhanced N-terminal cleavage, as did Arg at C+2, similar to this residue's effects at the N-extein. Attenuating mutants from the n library had the highest in vivo FRET signals with Pro, Val, or Glu substituted for N-1 Tyr, and they were relatively resistant to cleavage by DTT in vitro (Fig. 3 B and C). Although other variants in the library showed effects on intein activity (Fig. 3 B and C, Table S1, and Fig. S3B), we focused on 1 accelerator in each extein: N-1 Asp and C+3 Cys (variants n-ND and c-SC, respectively), and the strongest attenuator, N-1 Pro (variant n-RP) (Figs. 3B and 4A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Characterization of accelerators and attenuators in vitro. (A) Progress of DTT-induced N-extein cleavage. Soluble protein extracts of parental construct (C-Ia-Y), accelerator mutants n-ND and c-SC, and attenuator n-RP were incubated at pH 8 with 20 mM DTT for the times indicated. Reaction mixtures were separated by12% SDS/PAGE followed by fluorescence detection of YFP and CFP (excitation 457 nm, emission 526 nm for both CFP and YFP). (B) Single extein residue effects on N-terminal cleavage. Individual residues from 3 mutant extein pairs that produced accelerated (n-ND, c-SC; lanes 2 and 8) or reduced (n-RP; lane 5) rates of N-extein cleavage compared with native extein parental C-Ia-Y control (P, lane 1) were returned to their native congener residue (underlined). The spurious migration of CFP (C) is as in Fig. 2C. The percentage of precursor was determined by dividing the fluorescent intensity of the C-I-Y band by that of the I-Y+C-I-Y bands (excitation 488 nm, emission 580 nm). (C) Attenuated and induced N-extein cleavage with N-1 Pro. Spontaneous and DTT-induced N-extein cleavage for parental extein control (C-Ia-Y) and N-1 Pro mutant, n-RP, was determined after 18 h incubation. Protein extracts were incubated at pH 8 without (−) or with (+) 20 mM DTT (lanes 1–4) or without (−) and with (+) 0.1 M hydroxylamine (lanes 5–8), and the products separated by 12% SDS/PAGE followed by fluorescent gel imaging as above.

Acceleration by N-1 Asp and C+3 Cys, and Attenuation by N-1 Pro.

Two independent members of the n library and 2 members of the nc library with Asp residues at N-1 showed low in vivo FRET values (Fig. 3C; Table S1). Kinetic and SDS/PAGE analyses of N-1 Asp-containing variants showed that N-terminal cleavage proceeded nearly to completion in vivo (98%), in contrast to the protein with native exteins, where N-terminal cleavage proceeded only to ≈30–40% completion (Fig. 4A, compare C-Ia-Y and n-ND, 0 min). To confirm that Asp was the residue responsible for enhancing cleavage of the n-ND double mutant, we reverted each individual residue of the variant pair back to its native congener. The accelerated precursor cleavage persisted with the Glu-Asp pair, but not Asn-Tyr pair, indicating that Asp is the residue that facilitates N-terminal cleavage (Fig. 4B, lanes 2–4).

Two cysteine-containing variants at position +3 were identified in the c library, and a third was selected from the nc library (Fig. 3C; Table S1). All of these variants accelerated N-terminal cleavage to an extent similar to the N-1 Asp mutants (Fig. 4A, c-SC). A representative mutant, c-SC, was subjected to a mutagenesis protocol identical to that above, by reverting to the native C+2 and C+3 residues Phe and Asn. As apparent from Fig. 4B (lanes 8–10), the residue primarily responsible for accelerating cleavage is Cys at position C+3.

In contrast to N-1 Asp, variants substituted with Pro or Val at N-1 had N-terminal cleavage suppressed. A high, in vivo FRET signal led to our selection of a single variant with Pro at the N-1 position and 2 independent N-1 Val variants (Fig. 3 B and C; Table S1). Accordingly, these derivatives were purified almost exclusively as precursor, and N-terminal cleavage by added DTT was slowed significantly compared with the native extein control (Fig. 4A). Because the N-2 N-1 Arg-Pro pair gave the lowest k_obs, with >1,000-fold attenuation, we elected to isolate the Pro in a single mutant. Once again, we showed that this residue, whether preceded by Arg or the native Glu, strongly inhibits N-terminal cleavage (Fig. 4B, lanes 5–7).

With a view to exploiting precursor buildup of an attenuator variant for practical applications (see Discussion), we sought to restore activity to the n-RP mutant. After exploring some different conditions, we found that using hydroxylamine (0.1 M) as the nucleophile, rather than DTT, allowed complete cleavage of the N-1 Pro precursor at 25 °C (Fig. 4C, compare lanes 4 and 8).

Divergent Pathways for N-Terminal Cleavage.

In comparing pathways for accelerated N-terminal cleavage by N-extein and C-extein mutants, the steps by which an intein supports N-terminal cleavage should be considered (Fig. 5A) (15). In the first step, the intein's nucleophilic N-terminal Cys residue attacks the amide linkage preceding it, resulting in an N→S acyl shift (Fig. 5A, step 1). Once generated, the acylated N-extein is relayed to the nucleophlic side chain of the C+1 Cys residue, yielding a branched intermediate (step 2). Evidence for a branched intermediate is provided by the appearance of a slowly migrating band during SDS/PAGE of a heat-denatured intein/extein precursor (Fig. S2B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Divergent paths for N-extein cleavage. (A) Cleavage pathway in N159A (C-Ia-Y) mutant. Steps 1 and 2 are described in Divergent Pathways for N-terminal Cleavage. (B) The extents of in vivo N-extein cleavage. The n-ND and c-SC derivatives and their Ala point mutants at the catalytic intein Cys (C1) or extein Cys (C + 1) residues were analyzed alongside parental C-Ia-Y (P) as in Fig. 4. The ≈10-fold difference between the percentage of precursor in lanes 6 and 8 was reproducible over 3 independent experiments.

Replacing the DnaE intein's native N-extein residues, Glu-Tyr, with Asn-Asp strongly accelerated N-terminal cleavage. Similar effects were observed when native C-extein residues Phe-Asn were mutated to Ser-Cys. With either mutant extein, the enhanced reactivity could be attributed to hydrolysis of the linear ester intermediate, to the branched ester intermediate (Fig. 5A), or to nonintein-mediated proteolysis. To distinguish between these possibilities, Cys-to-Ala mutations were made separately at the C1 position and at the branch-forming C+1 position of both the n-ND and c-SC variants (Fig. 5B, lanes 3, 4, 7, and 8). For both n-ND and c-SC, the C1A mutation abolished N-terminal cleavage (Fig. 5B, lanes 3 and 7). Dependence on a catalytic residue for accelerated cleavage diminishes the likelihood that the observed effects were caused by nonintein-mediated proteolysis or hydrolysis.

In contrast, N-terminal cleavage was apparent in the C+1A mutants of n-ND and c-SC, albeit to differing extents. In the n-ND C+1A mutant, N-terminal cleavage remained at the same enhanced level as was observed for n-ND (Fig. 5B, compare lanes 2 and 4). This result suggests that intein-catalyzed formation of the linear ester intermediate, but not the branched ester, is necessary and sufficient for the Asp effect. However, the rate of in vivo N-terminal cleavage was reduced with the same C+1A mutation in the c-SC background (Fig. 5B, compare lanes 6 and 8). This result is consistent with the idea that interactions responsible for the Cys effect at C+3 require intein-catalyzed formation of the linear intermediate and the branched intermediate. Although speculative, a substituted Cys at C+3 could be acting as a surrogate nucleophile. Regardless of mechanism, these observations suggest that in the present system divergent pathways are responsible for enhanced N-terminal cleavage, depending on whether the N-extein residue or C-extein residue mediates the effect. Thus, the mutant N-extein residues act strictly on the linear ester intermediate, whereas mutant C-extein residues act also on the branched ester intermediate. However, this inference does not preclude that more complex interactions may have emerged in the course of intein evolution in the natural extein environment (e.g., see ref. 16).

Extein Effects on Splicing in Addition to Cleavage.

All experiments to this point used an intein harboring an N159A substitution that blocked splicing. It might be thought that an intein's native tendency to splice would be sufficiently strong to counteract any alteration in thioester intermediate stability that is mediated by a substrate extein. To test that hypothesis, we restored the splicing pathway in selected variants by reverting N159A to the wild-type Asn. In the final catalytic step of intein splicing, N159 cyclizes, releasing the intein from the exteins (Fig. 6A, step 3). An isomerization step yields an unmodified host protein (Fig. 6A, step 4) (15). Measurements of the extent of in vivo splicing, and calculation of the relative abundance of splice product and N-terminal cleavage product, were carried out for 6 mutant exteins: accelerators n-ND, c-RA, and c-SC; and attenuators n-TV, n-LE, and n-RP (Fig. 6).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Splicing is altered by N- and C-extein residues. (A) Complete intein splicing pathway (described in Extein Effects on Splicing in Addition to Cleavage). (B) Commassie-stained gel of Ni²⁺-NTA-purified products from cultures expressing splicing competent C-I-Y (WT) and its extein mutants, along with Ni²⁺-NTA purified MC1061 lysate (BG). (C) Data summary. The relative abundance of precursor (C-I-Y), spliced product (C-Y), and N-terminal cleavage product (I-Y) apparent from B.

The signature feature of accelerator mutants (n-ND, c-RA, and c-SC) was a reduction in the yield of splice product (C-Y), compared with the native-extein control (Fig. 6 B and C, lanes 2–4). In n-ND, there was roughly two-thirds the amount of splice product relative to the wild-type level. Compromised splicing efficiency was most apparent for the C-extein mutants c-RA and c-SC, where ligated exteins were absent. Unlike the accelerators, splice product was formed in all attenuators examined (n-TV, n-LE, and n-RP), but with lower efficiency, as apparent from the accumulation of precursor (Fig. 6 B and C, lanes 5–7). Suppression is most severe in n-RP, where only trace amounts of products formed. Although the overall reaction was inhibited, it is noteworthy that the splice product was made in greater proportion than the cleavage product (C-Y > I-Y) for n-TV, n-LE, and n-RP (Fig. 6 B and C, lane 5–7). Thus, in all 6 cases examined, extein mutation resulted in deviations in splicing.