Conditional Alternative Protein Splicing Promoted by Inteins from Haloquadratum walsbyi

Abstract

Protein splicing is a post-translational process by which an intervening protein, or intein, catalyzes its own excision from flanking polypeptides, or exteins, coupled to extein ligation. Four inteins interrupt the MCM helicase of the halophile Haloquadratum walsbyi, two of which are mini-inteins that lack a homing endonuclease. Both inteins can be over-expressed in E. coli and purified as unspliced precursors; splicing can be induced in vitro on incubation with salt. However, one intein can splice at 0.5 M NaCl in vitro, whereas the other splices efficiently only above 2 M NaCl; the organism also requires high salt to grow, with the standard growth media containing over 3 M NaCl and about 0.75 M magnesium salts. Consistent with this difference in salt-dependent activity, an intein-containing precursor protein with both inteins promotes conditional alternative protein splicing (CAPS) to yield different spliced products dependent on the salt concentration. Native Trp fluorescence of the inteins suggests that the difference in activity may be due to partial unfolding of the inteins at lower salt concentrations. This differential salt sensitivity of intein activity may provide a useful mechanism for halophiles to respond to environmental changes.

Introduction

Inteins mediate the process of protein splicing, a post-translational event by which the intervening protein, or intein, catalyzes its own excision from the flanking polypeptides, or exteins, concomitant with extein ligation¹ (Figure 1A). Splicing is a four-step process (Figure S1): (1) an amide to thioester rearrangement of the peptide bond linking the N-extein to the first residue of the intein (Cys for the inteins described here), (2) a transesterification reaction that results in the transfer of the N-extein from the side chain of the first residue of the intein to the side chain of the first residue of the C-extein (Ser for these inteins) to produce a branched ester intermediate, (3) cleavage of the intein from the exteins via cyclization of the C-terminal residue of the intein (Asn) coupled to peptide bond cleavage, and (4) reversion of the ester bond linking the ligated exteins to the native peptide bond.¹

Figure 1. Schematic of protein splicing and the H. walsbyi cdc21 precursor.

A. Protein splicing results in the excision of an intein from flanking polypeptides (the N- and C-exteins) concomitant to extein ligation to produce a functional extein protein. B. The insertion sites of the four inteins that interrupt the cdc21 protein in H. walsbyi. Mini-inteins (A and D) are in orange, and homing endonuclease-containing inteins are in blue (B and C). The codes above the schematic indicate the amino acid residue that is the first residue of the intein (orange or green bar), and those below indicate the first residue of the extein segment (blue). The insertion sites were named in the order of their discovery and are therefore out of alphabetical order, and the schematic is approximately to scale.

Improper coordination of the steps of splicing can lead to unproductive side reactions. The thioester or ester formed in steps 1 and 2 can be subject to hydrolysis if step 3 is prevented or delayed. This side reaction is called N-terminal cleavage and results in the production of the N-extein separate from the intein-C-extein fusion protein. Alternatively, if step 3 happens before step 1 or 2, C-terminal cleavage can result, with the N-extein and intein fused and the C-extein liberated.¹

Many inteins are interrupted by nested homing endonuclease (HEN) domains similar to those found in mobile introns.² Other inteins lack this domain and are called mini-inteins. The HEN domain makes a double-stranded break in an intein-less allele, facilitating intein homing by homologous recombination. This parasitic nature of inteins has been described as a homing cycle, by which an intein that splices poorly may exert a fitness cost on the host organism and loss of the intein by a random genetic event may result from this negative selective pressure. However, the HEN domain permits reintegration of the intein as long as an intein-containing allele is present either in a polyploid cell or in a population capable of mating.^{3, 4} Selective pressure against inefficient splicing also could result in the molecular evolution of a more efficient intein that splices co-translationally and exerts no significant burden on the host. In this case, the selective pressure to retain the HEN domain is lost, and the domain may be degraded by mutation or may be lost to form a mini-intein. Subsequent genetic deletion of these mini-inteins would result in an intein-less allele that could be subject to reinvasion by a HEN-containing intein. The genetic removal of an intein must be precise as most inteins invade near the active sites of essential enzymes involved in DNA replication, repair, and recombination.^{5, 6} It is likely that a dynamic equilibrium exists in the archaeal ecosystem between intein-containing and intein-lacking organisms.⁷

In contrast to this parasitic view of inteins, it is possible that some inteins may play a role for the host and be maintained by positive selective pressure. If so, such inteins likely would splice conditionally, subject to an environmental cue that might signal for the production of active extein segments from an inactive intein-interrupted precursor. For example, split inteins have been engineered to splice conditionally when induced with light or the binding of small molecules.^8–17 Full-length inteins have been engineered to be triggered to splice by temperature, light, pH, or small molecules.^18–20 Inteins from thermophiles may require elevated temperatures to splice in vitro,^21–26 and inteins may require reduction of native disulfide bonds involving active site Cys residues to allow splicing.^27–30 No direct evidence of conditional splicing playing a beneficial role in a native system has been shown. However, the activity of the Methanoculleus marisnigri PolII intein is regulated by the redox state of an internal disulfide bond, and its splicing activity is sensitive to redox state in E. coli.³¹ The splicing of the Pyrococcus horikoshii RadA intein, when interrupting the native recombinase, is activated by single-stranded DNA, which suggests that splicing may be triggered when ssDNA signals the need for recombinase activity.^32–35 The Mycobacterium tuberculosis SufB intein is sensitive to oxidative and nitrosative stress that may link such stressors to Iron-Sulfur cluster assembly.³⁶ The splicing of a Mycobacterium smegmatis DnaB intein is sensitive to oxidative stress in M. smegmatis, suggesting that pausing DNA replication may be connected to a stress response.³⁷ These instances suggest that splicing-mediated gene regulation might serve a role for host cells.

Some genes contain multiple sites for intein insertion in different species, while in some cases multiple inteins are present in a single gene.^37–40 If these inteins were to splice differently under environmentally relevant conditions, they could produce alternatively spliced products, which we will call conditional alternative protein splicing, or CAPS. For instance, the two DnaB inteins in M. smegmatis are differentially sensitive to oxidative stress and splice at different rates.³⁷ Here, we describe another potential example in the extreme halophile Haloquadratum walsbyi,^{41, 42} where the differential dependence on salt concentration can lead to CAPS.

Inteins are broadly distributed in the haloarchaea, with at least 24 different genes invaded by inteins.⁴⁰ Four inteins from halophiles have been purified as unspliced precursor fusion proteins in E. coli and induced to splice efficiently in vitro with incubation at high salt concentrations, another example of conditional protein splicing.^43–45 Three of these inteins are mini-inteins (Halobacterium salinarum PolII and cdc21 and Halorhabdus utahensis MCM2), with the H. utahensis intein splicing at lower salt concentrations than the inteins from H. salinarum.^{43, 44} The other is a full-length intein with an intervening homing endonuclease domain (Haloferax volcanii PolB-c).⁴⁵ Haloquadratum walsbyi has 15 inteins, with four separate inteins invading the cdc21 mini-chromosome maintenance (MCM) gene, a DNA helicase key to replication.³⁹ Two of these inteins have nested homing endonuclease domains, and the other two, studied here, are mini-inteins.

Below, we show that the two mini-inteins in the H. walsbyi cdc21 protein can be expressed as unspliced fusion proteins in E. coli. The inteins splice at different salt concentrations in vitro, both when expressed separately and also when expressed as part of the same fusion protein. Analysis of their global structural stability by native Trp fluorescence shows that they fold at different salt concentrations, which likely contributes to the differential dependence of intein activity on salt concentration.

Materials and Methods

Plasmid preparation

To create E. coli expression plasmids for the H. walsbyi inteins, we obtained H. walsbyi from the Leibniz Institute DSMZ German Collection of Microorganisms and Cell Culture (DSM 16790, strain HBSQ001) and cultured the cells unshaken at 37°C in medium 1091.⁴¹ We harvested the cells by centrifugation and purified the genomic DNA using a Wizard Genomic DNA purification kit (Promega) by the manufacturer’s instructions. The plasmids described below are listed in Table S1.

To create plasmids pMIH-A and pMIH-D, which express intein precursor proteins MIH-A and MIH-D, we used PCR to amplify the intein genes from genomic DNA. For intein A, we used primers MIHAU and MIHAL, and, for intein D, we used primers MIHDU and MIHDL, digested the PCR products with StuI and EcoRI, and inserted them between those same sites in plasmid pMIHMma, described previously.³¹ (See Table S2 for oligonucleotide sequences.) Plasmid pMIH-A expresses a 397-residue N-extein with N-terminal MBP followed by a short linker and the nine C-terminal residues of the native N-extein, fused in-frame to the 209 residue intein and 23 residue C-extein, which has 11 native C-extein residues and a His-tag. Plasmid pMIH-D expresses a 398-residue N-extein with N-terminal MBP followed by a short linker and the 10 C-terminal residues of the native N-extein, fused in-frame to the 211-residue intein and a 23-residue C-extein, which has 11 native C-extein residues and a His-tag.

To create plasmid pMIHAD, which encodes the fusion protein M-(I^A)-E-(I^D)-H, we purchased a synthetic gene from Genscript in pET28-b(+) as an in-frame fusion of the gene for the E. coli maltose binding protein (MBP), a short poly-N linker, the 22 C-terminal residues of the N-extein, the 209 residue intein A, the following 66 extein residues (27 between inteins A and B and 39 between inteins B and D), the 211 residue intein D, and the 14 extein residues between inteins D and C, followed by a short linker and a His-tag. In addition, we obtained site-directed mutations of the plasmid pMIHAD from Genscript to create pMIHADInAX, for which Cys1, Asn209, and Ser+1 are all changed to Ala for intein A to make fusion protein M-(mI^A)-E-(I^D)-H, and pMIHADInDX, for which Cys1, Asn211, and Ser+1 are all changed to Ala for intein D to make fusion protein M-(I^A)-E-(mI^D)-H.

To create plasmids pNIH-A and pNIH-D to express proteins for native Trp fluorescence experiments, we used Gibson assembly with linearized pET21-b(+) as the vector. We used pMIH-A or pMIH-D as templates for PCR for the insert, using the InFusion HD cloning kit per the manufacturer’s instructions (Takara Bio USA). To make pNIH-A, we used PCR primers AGibF and AGibR and for pNIH-D we used PCR primers DGibF and DGibR. Plasmid pNIH-A expresses NIH-A, a nine residue N-extein fused in-frame to the 209 residue intein and a 24-residue C-extein including a C-terminal His-tag. Plasmid pNIH-D expresses NIH-D, a 10 residue N-extein fused in-frame to the 211 residue intein and a 24-residue C-extein including a C-terminal His-tag. To prevent splicing, site-directed mutagenesis was performed by Genscript to make plasmids pNIH-AX and pNIH-DX, which have mutations to Ala for residues Cys1, Ser+1, and the final Asn residue for each intein. Plasmids pNIH-AX and pNIH-A expressed poorly. Therefore, we inserted a GB1 tag using annealed and phosphorylated oligonucleotides GBU and GBL into the NdeI site of pNIH-A and pNIH-D to make pGBAI and pGBDI. Plasmid pGBAI expresses GbA1, the 56-residue GB1 tag fused in-frame to a 6 residue linker, 7 residues of the native N-extein, the 209 residue intein and a 24-residue C-extein including a C-terminal His-tag. Plasmid pGBD1 expresses GbD1, the 56-residue GB1 tag fused in-frame to a 6-residue linker, 7 residues of the native N-extein, the 211 residue intein and a 24-residue C-extein including a C-terminal His-tag. Genscript then made site-directed mutants that resulted in the conversion of Trp43 of the GB1 tag to Phe, Cys1 and Ser+1 of each intein to Ala, and Asn209 (GBAI) or Asn211 (GBDI) to Ala to make plasmids pGBA and pGBD to express proteins GBA and GBD.

Protein expression and purification

In order to over-express intein fusion proteins, the plasmids described above were transformed into E. coli BL21(DE3). A 1:50 inoculum in LB media was grown to mid-log phase, induced with isopropyl-β-d-1-thiogalactopyranoside (IPTG) to 1 mM, and incubated with shaking at 20°C for 3 h. Cells were pelleted by centrifugation and resuspended in buffer A (20 mM HEPES, pH 7.5, 500 mM NaCl) supplemented with 0.1 mM phenylmethylsulfonyl fluoride, 10 units of benzonase nuclease, and an EDTA-free protease inhibitor cocktail. Cells were disrupted by passage through a French-pressure cell, and the lysate was clarified by centrifugation at 16 000 × g for 30 min at 4°C. Proteins were purified using the C-terminal His-tag by immobilized metal affinity chromatography using Talon metal affinity resin (Takara Bio USA) per manufacturer’s instructions. Elution fractions were exchanged into buffer A using an EMD-Millipore Amicon Centrifugal Filer with a 3000 MWCO, and the protein concentration was determined by a Bradford assay.⁴⁶

Protein splicing activity was assayed by incubating about 0.2 mg/ml protein in a reaction cocktail that has 20 mM HEPES, pH 7.5, 2 mM Tris(2-carboxyethyl)phosphine (TCEP), 5 mM ethylenediaminetetraacetic acid (EDTA), and sodium chloride at the concentrations indicated in the figure legends at 20°C. The protein reaction was diluted 1:1 with water before SDS-PAGE to prevent salt precipitation on addition of the SDS-PAGE loading buffer.

The extent of protein splicing was estimated by SDS-PAGE stained with QC Colloidal Coomassie Stain (Bio-Rad). For MIH-A, the precursor protein MIH has a M_r of 68,356.42, and protein splicing would yield MH (M_r 46,176.05) and I (M_r 22,198.38). For MIH-D, the precursor protein MIH has a M_r of 69,116.90, and protein splicing would yield MH (M_r 46,432.20) and I (M_r 22,702.72).

For M-(I^A)-E-(I^D)-H, the precursor (M_r 100,502.63) can splice into MEH (M_r 55,637.56), I^A (M_r 22,198.38) and I^D (M_r 22,702.72). Thus, splicing of intein I^D first would yield M-(I^A)-E-H (M_r 77,817.53) and splicing of intein I^A first would yield M-E-(I^D)-H (M_r 78,322.26). The percentage of splicing was estimated using densitometry via ImageJ,⁴⁷ using the formula % splicing = 100* (spliced product) / ((precursor) + (spliced product)), which each densitometry reading corrected for relative molar mass.

His-tagged proteins were detected using InVision His-tag in-gel stain (Life Technologies) and visualized with a BioRad Gel Doc Imager with a UV tray. For mass spectrometry, high resolution mass spectra were obtained with electrospray ionization on an Orbitrap Fusion mass spectrometer following an in-line C4 desalting column by the University of Massachusetts, Amherst.

Native Trp fluorescence

We measured native Trp fluorescence using a Perkin Elmer LS55 fluorescence spectrometer, with a constant temperature of 25°C maintained by a Peltier control system. An excitation wavelength of 295 nm was used to excite only the Trp residues in the intein rather than Tyr residues.⁴⁸ Emission was monitored from 310 nm to 380 nm at a scan rate of 10 nm/min, with 2.5 nm slit widths for excitation and emission on a medium gain setting. Protein Gb1A was measured at 10 μM and protein Gb1D was measured at 8 μM in 20 mM HEPES, pH 7.5, at varying concentrations of NaCl.

Results and Discussion

The cdc21 protein from H. walsbyi is interrupted by four inteins, as shown schematically in Figure 1B. Inteins A and D are mini-inteins, and inteins B and C have nested HEN domains. (The inteins are named by insertion site, which were named as they were discovered and are not in alphabetical order.) To study the splicing of the inteins separately, we elected to focus on the mini-inteins, as inteins with nested HEN domains can be challenging to express in E. coli.³⁸ An alignment of the intein sequences is given in Figure S2. The inteins have 32% sequence identity and 50% similarity.

We created expression vectors to over-express inteins A and D separately as fusion proteins with an N-terminal maltose binding protein (M) and a C-terminal His-tag (H) serving as the exteins, creating fusion proteins MIH-A and MIH-D. To avoid any effects on splicing by non-native flanking residues, we designed the expression vectors such that the M is followed by a flexible linker and nine (intein A) or ten (intein D) native N-extein residues, and 11 native C-extein residues precede the His-tag for both inteins.

We were able to separately purify the intein fusion proteins MIH-A or MIH-D from E. coli via immobilized metal affinity chromatography. The intein fusion protein was incubated in vitro with EDTA to prevent any potential inhibition of splicing by divalent cations,^49–52 with TCEP to reduce a potential disulfide bond between Cys1 and the internal Cys residues,^{27–29, 31} and with varying concentrations of sodium chloride.

Intein A promoted protein splicing at 20°C and 2.0 M NaCl, as seen by conversion of the precursor band MIH to the splicing products MH and I in Figure 3A. The identity of the SDS-PAGE bands were confirmed by InVision staining, which identifies bands that contain a His-tag (MIH and MH, but not I) (Figure 3B). This confirms that we are observing splicing and not N-terminal cleavage, which would produce M and IH. We also confirmed the identity of the protein products by LC-MS (Table S3). These confirmations are important as halophile proteins can have aberrant migration patterns on SDS-PAGE, likely due to their highly negative surface charge.^{53, 54} For the MS peaks corresponding to the excised inteins, the M_r values are about 18 Da smaller than expected, which may be due to the intein retaining the C-terminal aminosuccinimide from the third step of splicing rather than undergoing subsequent hydrolysis to either Asn or iso-Asn.

Figure 3. In vitro protein splicing activity of H. walsbyi inteins as separate fusion proteins.

SDS-PAGE analysis of in vitro protein splicing of MIH, stained with Coomassie blue. Splicing of precursor MIH (68 kDa) results in MH (46 kDa) and I (22 kDa). In panels A and C, purified precursor fusion protein MIH-A or MIH-D (0.2 mg/mL) was incubated for 2 h or 6 h at 20°C at the total NaCl concentrations indicated above the lanes in 20 mM HEPES buffer, pH 7.5, with 2 mM TCEP and 5 mM EDTA, as described in the Materials and Methods. The percentage of splicing was estimated by densitometry using ImageJ and represents the average of two independent trials.⁴⁷ In panels B and D, the gels were probed with InVision stain, which detects proteins with a His-tag. Hence, we would expect to see bands corresponding to MIH, MH, or IH, but not MI or I. UR is unreacted protein, nd is not determined.

Likewise, we were able to show that intein D also promotes splicing at 20°C (Figure 3C) and confirmed this splicing with InVision staining and LC-MS (Figure 3D and Table S3). However, efficient protein splicing occurs at lower salt concentrations. Although splicing efficiency improves at higher salt concentrations, we observe splicing at 0.5 M NaCl, and observe some splicing in E. coli by the appearance of the MH band in the purified and unreacted fraction. This suggests that intein D does not have the same requirement for high salt as intein A. At 2 M NaCl, the inteins splice over a range of temperatures from 15°C to 30°C to a similar extent (Figure S3). Splicing of intein A increases slightly over 6 h, whereas splicing of intein D is mostly complete after 2 h of incubation (Figure 3).

Although the inteins display significant differences in salt dependence, there is a high degree of similarity in their amino acid sequences (Figure S2). These two intein sequences are 32% identical and 50% similar, as measured by the EMBOSS Water tool.⁵⁵ Within the conserved intein sequence blocks the similarity is even higher, with 80% or higher similarity in blocks A, F, and G, which comprise the N-terminal (A) and C-terminal (F and G) sections of the active site. The sequences are only 50% similar in conserved block B, but the highly conserved Thr and His residues are present. As calculated by the ExPasy pI/Mw tool, the isoelectric point (pI) values of the two inteins are both low and similar, at 4.48 and 4.43, respectively.⁵⁶ Both inteins have similar numbers of potentially charged residues, with intein A having 33 acidic and 21 basic residues and intein D having 34 acidic and 22 basic residues.

Interestingly, despite relatively high amino acid similarity, the inteins show differential activity as a function of salt concentration. In contrast, the three inteins that interrupt the Pyrococcus abyssi ribonucleotide reductase all splice efficiently in E. coli, despite having little sequence similarity and being flanked at their C-termini by different nucleophiles to promote step 2.³⁸ On the other hand, the two inteins that interrupt the DnaB from M. smegmatis splice differently.³⁷ The second M. smegmatis intein splices relatively quickly and its splicing is not sensitive to oxidative or nitrosative stress. However, the first intein splices more slowly and its splicing is sensitive to oxidative stress, both in vitro and in M. smegmatis. The first intein is also susceptible to inhibition by zinc and reactive chlorine species.^{57, 58} These two inteins also can facilitate conditional alternative protein splicing (CAPS) as a function of time when expressed as part of the same precursor. The M. smegmatis inteins differ considerably, as the first intein is a non-canonical class three intein;³⁷ class three inteins splice by an alternative mechanism utilizing an internal nucleophile in place of the N-terminal Cys or Ser.^{59, 60} In contrast, the second intein is a traditional class one intein that splices by the canonical mechanism, so it is not surprising that the two inteins might promote CAPS from the same precursor. We were interested in whether the two more similar H. walsbyi mini-inteins can also promote CAPS.

To test whether the inteins retained their differential salt dependence when interrupting the same protein, we created the fusion protein M-(I^A)-E-(I^D)-H, which encodes for N-terminal MBP, followed by a flexible linker, the 22 residues that flank the N-terminus of intein A, intein A, the next 66 extein residues (Exteins-2 and -3 in Figure 1B) that would result from the splicing of intein B, then intein D, the 14 extein residues between inteins D and C, and a His-tag (Figure 1B). We anticipated that at low salt concentrations, intein D would splice, producing M-(I^A)-E-H and I^D. We expected both inteins to splice at higher salt concentrations, producing M-E-H, I^A and I^D (Figure 2). These predictions are supported by the results in Figure 4. In Figure 4A, product MI^AEH is produced at 1.0 M salt, indicating that intein D has been excised. At 2.0 M salt, the fully spliced product MEH is produced, resulting from the splicing of intein A from MI^AEH. The extent of splicing at 2 h and 6 h is also consistent with the extent of splicing of the individual inteins at those times (Figures 2 and 3). The identity of the bands is supported by InVision staining (Figure S4), which indicates which bands have a C-terminal His-tag. Given the similarity in size between inteins A and D, protein MI^AEH (from splicing of intein D) and MEI^DH (splicing of intein A) would be similar in size on SDS-PAGE. Therefore, we also analyzed the splicing via LC-MS (Table S3). For a reaction run for 6 h at 20°C at 2 M salt for a fusion protein with both active inteins, we observed a peak consistent with the MI^AEH intermediate but not the MEI^DH intermediate, as expected if the reaction follows the scheme in Figure 2 (Table S3).

Figure 2. Schematic of conditional alternative protein splicing (CAPS).

If intein D splices at lower salt concentrations than intein A, it would be removed first to produce an alternative precursor fusion protein. Intein A would then splice on incubation at a higher salt concentration. The codes to the right of the cartoon schematic correspond to the expressed protein products observed in Figure 4.

Figure 4. In vitro protein splicing activity of H. walsbyi inteins as part of one fusion protein.

SDS-PAGE analysis of in vitro protein splicing of M-(I^A)-E-(I^D)-H, stained with Coomassie blue. Splicing of precursor M-(I^A)-E-(I^D)-H (68 kDa) results in MI^AEH (78 kDa) or MEI^DH after splicing of one intein, MEH (56 kDa) after splicing of both inteins, as well as inteins I^A (22 kDa) or I^D (23 kDa). In panel A, purified precursor fusion protein M-(I^A)-E-(I^D)-H (0.2 mg/mL) was incubated for 2 h or 6 h at 20°C at the total NaCl concentrations indicated above the lanes in 20 mM HEPES buffer, pH 7.5, with 2 mM TCEP and 5 mM EDTA, as described in the Materials and Methods. In panels B and C, the same reaction was run, but with inactivating mutations to intein A (M-(mI^A)-E-(I^D)-H) or to intein D (M-(I^A)-E-(mI^D)-H). The percentages reflect the average of two independent trials.

Given the similarity in sizes between inteins A and D, we confirmed the splicing via mutational analysis and via LC-MS. To inactivate one intein, we made mutations to key catalytic residues, Cys1, Ser+1, and C-terminal Asn, to Ala to prevent splicing activity. We made one fusion protein with mutated intein A (M-(mI^A)-E-(I^D)-H), and another with mutated intein D (M-(I^A)-E-(mI^D)-H). As seen in Figure 4B, when intein A is mutated we observe splicing at a salt concentration similar to that of intein D alone in Figure 3. When intein D is mutated as seen in Figure 4C, we observe splicing at a salt concentration similar to that of intein A in Figure 3. We confirmed the identity of the products of the splicing products for the two intein-inactivated mutants via both Invision staining (Figure S4) and LC/MS (Table S3).

In order to determine if the difference in intein activity is correlated to protein folding, we used native Trp fluorescence to measure the folding of the intein as a function of salt concentration (Figure 5). We attempted to express the inteins from separate plasmids as a fusion protein with an N-terminal His-tag rather than MBP. To prevent splicing, we mutated Cys1, the C-terminal Asn, and Ser+1 all to Ala. However, protein expression was poor, particularly with intein A. This is not surprising, given that the excised intein bands in Figures 2 and 3 are faint relative to the MBP-containing spliced product, suggesting compromised solubility for the excised intein. Therefore, we expressed the intein genes such that they were separately expressed as fusion proteins with an N-terminal Gb1 tag.⁶¹ We mutated the Trp in the Gb1 sequence to Phe such that we could monitor the three Trp residues in intein A and the two Trp residues in intein D, using an excitation wavelength of 295 nm to specifically excite the Trp fluorescence.⁴⁸ The fluorescence of intein A at salt concentrations below 1.5 M NaCl is low, suggesting that the protein folds and hence buries the Trp residues as the salt concentration increases to values near those that permit splicing (Figure 5). In comparison, intein D, which can splice at lower salt concentration, also appears to be well folded at 1 M salt. The blue shift of the maximum emission wavelength also depends on salt concentration (Figure S5). These results suggest that the splicing dependence of the intein activity is due to substantive unfolding or disorder at sub-optimal ionic strength.

Figure 5. Salt-dependence of intein folding as measured by native Trp fluorescence.

Native Trp fluorescence was measured at 340 nm via excitation at 295 nm using a Perkin Elmer LS55 fluorescence spectrometer. Gain was set to medium with emission and excitation slit widths of 2.5 nm. Temperature was maintained at 25°C with a Peltier system. Protein GBA was at 10 μM, and protein GBD at 8 μM. Samples were measured in 20 mM HEPES buffer at pH 7.5 at the indicated salt concentrations on the abscissa. Data reported are from two independent trials.

Proteins from archaeal extremophiles have various strategies to prevent unfolding or a transition to disorder under extreme conditions.⁶² Many archaeal halophiles use a “salt-in” strategy to avoid osmotic stress and have proteins with very low pI values and hence very negatively charged surfaces, with a slightly lower number of bulky hydrophobic residues.^63–65 Although both inteins A and D have similarly low pI values and hence are likely to be similarly negatively charged, it has been suggested that more subtle structural differences at the surface, such as the choice of less bulky charged side chains, may affect the interaction surface area of halophilic proteins and hence their structural stability in the low water activity conditions of a high salt environment.⁶⁶ Inteins are highly mobile genetic elements, and the inconsistency of inteins being present at the same insertion site in proteins across the haloarchaea suggests transfer and loss of inteins within the family.⁴⁰ It is possible that the salt dependence of halophile inteins may be influenced by the salt requirements of the species they have invaded during their mobile evolutionary history.

H. walsbyi is the dominant archaeal species at very high aqueous salinity, despite its slow growth in laboratory-based culture.^{41, 67} The degree of species representation of H. walsbyi in saline pools can differ based on salinity.⁶⁸ On extreme dilution from over 30% salt to about 12% salt, it has been shown that although the ecotype of H. walsbyi changes and the percentage abundance of the population declines, that the dominant ecotype of H. walsbyi is still present and recovers after evaporation results in an increase in salinity.⁶⁹ H. walsbyi abundance can also vary seasonally, likely due to the balance of ions, particularly magnesium.⁷⁰ These observations of variable growth under physiological cues of ionic strength and/or water activity are consistent with the organism needing biochemical triggers that induce cell division under permissive environmental conditions. It is possible that CAPS activity could serve this purpose by linking post-translational activation of intein-containing genes related to DNA replication and cell-cycle progression to successful protein splicing of an interrupted precursor. One could speculate that in times of flooding, H. walsbyi may express its MCM in an intein-interrupted, inactivated format, ready for immediate activation on a return to salt concentrations conducive to growth and cell division.

For the inteins described here, the ligated exteins constitute the cdc21 (MCM) protein, the DNA helicase important in initiation of replication.⁷¹ Comparing the sequence of the H. walsbyi MCM to the well characterized MCM from the archaea Sulfolobus solfataricus,^{72, 73} the core AAA+ sequences are highly similar. The four inteins in H. walsbyi interrupt this core AAA+ region. Figure S6 shows a sequence alignment between the ligated H. walsbyi exteins and the uninterrupted S. solfataricus MCM using BLAST.⁷⁴ The sequences have 39% identity and 60% sequence similarity, with more conservation in the AAA+ region. The conserved regions of the MCM are noted under the S. solfataricus sequence in Figure S6.⁷⁵ The Ser+1 residue of intein A is at the C-terminus of the Walker A motif, and the Ser+1 of intein D (an Asp residue in S. solfataricus) immediately follows the Walker B motif. This suggests that the inteins interrupt the motifs responsible for nucleotide and magnesium ion binding. Mutation of the Lys that would precede intein A to Ala has been shown to compromise both helicase and ATPase activity in the S. solfataricus MCM, suggesting this region is vital to MCM activity.⁷³ Mutations of residues Ala+6 and Ala+10 for intein D to Arg result in a S. solfataricus MCM that cannot form hexamer and lacks helicase activity,⁷⁶ suggesting that a region immediately C-terminal of intein D may be at an interface, and that intein D might therefore interfere with hexamer formation. Interestingly, the change of Asp for Ser at the end of the Walker B motif in the H. walsbyi might compromise the conserved Walker B sequence to allow for intein D insertion, although a Glu residue that follows could serve in the role of the conserved Asp. Although not studied here, intein C is at a region that may be trans-acting with adjacent monomers in the MCM hexamer, suggesting that intein C could interfere with hexamer formation.⁷⁵ The residues six and eight positions downstream of intein B have been shown to be essential for DNA binding in the S. solfataricus MCM,⁷⁷ suggesting the importance of these region. Figure S7 uses the 296 C-terminal residues of the S. solfataricus MCM crystal structure (PDB code 3F9V)⁷⁶ to represent the MCM in the same orientation as the structure of the Natronomonas moolapensis MCM used by Belfort and coworkers to compare the intein insertion sites of bacterial DnaB and archaeal MCM exteins.⁵

Interestingly, in the S. solfataricus MCM, the AAA+ region can function as an ATPase and helicase on its own, and the helicase activity can be enhanced by addition of the N-terminal domain in trans. This suggests that the MCM protein may have different properties when expressed in fragments. One could speculate that differential protein splicing (or intein-mediated cleavage at splicing boundaries) could result in H. walsbyi MCM variants that also have different properties. Perhaps an intein remaining in the precursor might disrupt helicase activity by interfering with an interface between two monomers while still permitting ATPase activity.

It is interesting that two examples of CAPS are from M. smegmatis DnaB and H. walsbyi MCM. DnaB is the most common known intein host in bacteria, and MCM is the most common known host in the archaea.⁵ Both serve as replicative helicases, but are not closely related in sequence or structure.⁵ Given that these proteins are essential to the initiation of DNA replication and cell division, an intein-mediated trigger in these proteins could be useful to the host organism by linking cell division to appropriate environmental conditions for growth.

In addition, halophile inteins are useful in biotechnology applications including protein cyclization and semi-synthesis.⁴³ Adding these two inteins that are easily purified as unspliced precursors to the intein toolkit could allow for further development of interesting applications, including biosensors that could detect salt concentration differences by facilitating the splicing of fluorescent proteins under selective conditions.^78–80

ABBREVIATIONS

CAPS

Conditional Alternative Protein Splicing

DTT

dithiothreitol

EDTA

ethylenediaminetetraacetic acid

HEN

homing endonuclease

H. walsbyi

Haloquadratum walsbyi

IPTG

isopropyl-1-β-d-thiogalactopyranoside

MBP

E. coli maltose binding protein

MCM

mini-chromosome maintenance

TCEP

tris(2-carboxyethyl)phosphine