Elucidating the Specificity Determinants of the AtxE2 Lasso Peptide Isopeptidase*

Lasso peptide isopeptidase is an enzyme that specifically hydrolyzes the isopeptide bond of lasso peptides, rendering these peptides linear. To carry out a detailed structure-activity analysis of the lasso peptide isopeptidase AtxE2 from Asticcacaulis excentricus, we solved NMR structures of its substrates astexin-2 and astexin-3. Using in vitro enzyme assays, we show that the C-terminal tail portion of these peptides is dispensable with regards to isopeptidase activity. A collection of astexin-2 and astexin-3 variants with alanine substitutions at each position within the ring and the loop was constructed, and we showed that all of these peptides except for one were cleaved by the isopeptidase. Thus, much like the lasso peptide biosynthetic enzymes, lasso peptide isopeptidase has broad substrate specificity. Quantitative analysis of the cleavage reactions indicated that alanine substitutions in loop positions of these peptides led to reduced cleavage, suggesting that the loop is serving as a recognition element for the isopeptidase.

Ribosomally synthesized and post-translationally modified peptides (RiPPs) 3 (1) are a diverse set of natural products that are formed by the action of maturation enzymes on a linear peptide substrate. An emerging theme in the biosynthesis of RiPPs is that the maturation enzymes tend to have broad substrate specificity (2)(3)(4)(5)(6)(7)(8). Lasso peptides are a class of RiPPs, characterized by a threaded structure resembling a slipknot (9 -11). The internal macrocycle is realized via a single isopeptide bond installed post-translationally by two maturation enzymes (12). While there is a quickly expanding literature about the lasso peptide maturation enzymes (12)(13)(14), there is little known about the catabolism of these molecules. Recently, we reported an enzyme, lasso peptide isopeptidase, that specifically hydrolyzes the isopeptide bond of lasso peptides (Fig. 1A) (15). This enzyme, related to prolyl oligopeptidases, was found in the vicinity of a lasso peptide gene cluster in the freshwater ␣-proteobacterium Asticcacaulis excentricus. This organism has two separate lasso peptide gene clusters, each with an associated isopeptidase gene (15,16). There have been a large number of lasso peptides discovered in proteobacteria (11,17), and an isopeptidase is commonly associated with such clusters (9).
We named the two isopeptidases in A. excentricus AtxE1 and AtxE2. The gene for AtxE1 is found next to the gene cluster that encodes for the biosynthesis of astexin-1, while the AtxE2 gene is located next to the gene cluster encoding astexins-2 and -3 ( Fig. 1B) (15). We have previously characterized AtxE2 in vitro (15). This enzyme cleaves astexin-2 and astexin-3, but no crossreactivity was observed between AtxE2 and astexin-1. Whereas astexin-2 and astexin-3 share relatively high sequence homology (identity at 13/24 positions), the astexin-1 sequence is more divergent (Fig. 1C). In addition, we have shown that, at least for astexin-2, the thermally unthreaded variant of the peptide is not a substrate for AtxE2. This suggests that the lasso peptide isopeptidase enzyme recognizes the overall 3-dimensional fold of the peptide rather than a simple linear sequence epitope. Given these observations, we were interested in the substrate tolerance of AtxE2, and more broadly whether RiPP catabolic enzymes have the same promiscuity as RiPP maturation enzymes.
Here we have solved the NMR structure of a variant of astexin-2 with its four C-terminal amino acids removed (astexin-2 ⌬C4) in water and solved a new structure of fulllength astexin-3 in water. The structure of astexin-3 was previously determined in DMSO (15). In addition, we determined that the C-terminal tail portions of astexins-2 and -3 are dispensable with regards to isopeptidase activity. Finally, we carried out extensive alanine scanning mutagenesis on astexin-2 and -3 to determine the effect of these substitutions on the activity of both the maturation enzymes and the isopeptidase AtxE2. We show that both the maturation enzymes and catabolic enzyme AtxE2 are highly tolerant of different sequences.

Experimental Procedures
Materials-All cloning was performed using XL-1 Blue Escherichia coli. All expressions were performed using BL21 E. coli. Restriction enzymes were obtained from New England Biolabs. PicoMaxx DNA polymerase used for PCR was obtained from Agilent. Oligonucleotide primers used for cloning and mutagenesis were ordered from IDT.
Plasmid Construction-Previously we constructed plasmids pMM39 and pMM40 for the heterologous expression of * This work was supported by National Institutes of Health Grants R01 GM107036 and T32 GM7388. The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The atomic coordinates and structure factors (codes 2N6U and 2N6V) have been deposited in the Protein Data Bank (http://wwpdb.org/). □ S This article contains supplemental Table S1 and Figs. S1-S9 . 1 Both authors contributed equally to this work. 2 To whom correspondence should be addressed: 207 Hoyt Laboratory, Princeton, NJ 08544. E-mail: ajlink@princeton.edu. 3 The abbreviations used are: RiPP, ribosomally synthesized and post-translationally modified peptide; AtxE2, A. excentricus lasso peptide isopeptidase 2; TOCSY, total correlation spectroscopy; HSQC, heteronuclear single quantum correlation; aTc, anhydrotetracycline; AUC, area under the curve; aa, amino acids; RMSD, root-mean-square-deviation.
astexin-3 and astexin-2, respectively (15). These plasmids are derived from pASK-75, which includes an anhydrotetracyclineinducible promoter (18). To simplify the cloning of precursor variants, we redesigned these plasmids to allow for easy swapping of the precursor gene (atxA21 and atxA22 for astexin-2 and astexin-3, respectively). To make the swappable astexin-2 plasmid, pMM79, the primer A21rep75-1 forward and primer A21rep75-overlap reverse (supplemental Table S1) were used to amplify a fragment including the atxA21 gene from pMM40. These primers ablate the BglII site 5Ј of the atxA21 start codon and introduce a new, unique BglII site 3Ј of the atxA21 stop codon. Another PCR fragment spanning from the intergenic region between atxA21 to the EcoRI site of atxB2 was generated from pMM40 using the primers A21rep75-overlap forward and A21rep75-1 reverse. The two fragments were joined using overlap PCR, cleaved with XbaI and EcoRI, and inserted back into similarly cleaved pMM40 to generate pMM79. The swappable astexin-3 plasmid pMM65 was constructed in the same fashion using primers A22rep75-1 forward and reverse and A22rep75-overlap forward and reverse (supplemental Table  S1). Mutants of the atxA21 and atxA22 genes encoding either alanine substitutions or tail truncations were generated by overlap PCR using the primers listed in supplemental Table S1. The overlap PCR reactions were digested with XbaI and BglII and inserted into similarly digested pMM79 for astexin-2 vari-ants or pMM65 for astexin-3 variants. All constructs were confirmed by sequencing (Genewiz). Large Scale Expression of Astexin-2 ⌬C4 and Astexin-3 for NMR Studies-The expression and purification of astexins-2 and -3 have been previously described (15). To maximize production of full-length astexin-3, it was expressed at room temperature, which essentially eliminates any tail truncation. Room temperature expression of astexin-2 results in a mixture of full-length, ⌬C3, and ⌬C4 products. Thus we expressed astexin-2 at 37°C, giving primarily the ⌬C3 and ⌬C4 variants. Full-length astexin-3 and astexin-2 ⌬C4 were purified to homogeneity by HPLC. The final yield of HPLC-purified astexin-2 ⌬C4 used for NMR was 470 g from 18 liters of culture. Astexin-3 is produced at much higher levels, and the final yield of astexin-3 was 1.35 mg from 1.8 liters of culture.
NMR Studies-The purified astexin-2 and astexin-3 samples were dissolved in 95:5 H 2 O/D 2 O with the astexin-2 ⌬C4 sample at a final concentration of 2.7 mg/ml (1.28 mM) and the astexin-3 sample at 6.4 mg/ml (2.54 mM). COSY, TOCSY, NOESY, 13 C-HSQC, and 15 N-HSQC spectra were acquired for each peptide. Following a full proton resonance assignment based on intra and inter residue connectivity observed in the COSY, TOCSY, NOESY, and HSQC spectra, cross peak volumes from the 100 ms NOESY spectrum were integrated and calibrated using the r 1/6 relation. Pseudoatoms were introduced in cases when a stereospecific assignment could not be made. The volumes associated with such peaks were scaled by the number of two-proton interactions that they represent. The average distance between prochiral ␤-methylene hydrogens was set to 2.11 Å as a reference using a modified version of the caliba macro in CYANA in which the r 1/6 relation was enforced for all restraint types. Calibrated volumes were converted to upper distance restraints using CYANA and used as input to seven iterative cycles of NOE assignment and structure calculation (19,20) allowing ambiguous assignment to vary. Ambiguous assignments were manually refined and the structure reannealed from 200 random starting conformers using a finalized set of restraints. The ensemble of structures was then aligned using the averaged coordinates of C␣ carbons from residues 1 through 17 as template. The structures were energyminimized in explicit solvent in GROMACS using a procedure described by Spronk et al. (21)(22)(23)(24). Briefly, the peptide was placed in the simulation box and subsequently solvated with tip3p water. The system was simulated for 25 ps with cooling from 300 to 50 K.
Expression of AtxE2-AtxE2 used in the isopeptidase assays was expressed and purified as described previously (15). Briefly, BL21 cells harboring a pQE derivative with a C-terminally Histagged gene for AtxE2 were grown and induced with IPTG at 37°C. The cells were lysed via sonication, and the protein was purified on Ni-NTA resin per the manufacturer's suggestions.
Heterologous Expression and Crude Purification of Astexin-2 and -3 Variants-For astexin-2 variants, cultures of cells containing plasmids for each variant were grown overnight in LB supplemented with 100 mg/liter ampicillin and used to inoculate 500 ml of M9 medium (3 g/liter Na 2 HPO 4 , 1.5 g/liter KH 2 PO 4 , 0.5 g/liter NH 4 Cl, 0.25 g/liter NaCl, 2 g/liter glucose, 1 mM MgSO 4 , and 500 g/liter thiamine, supplemented with the FIGURE 1. Lasso peptide isopeptidase AtxE2 and its substrates. A, schematic of lasso peptide hydrolysis by lasso peptide isopeptidase. The isopeptide bond that defines the lasso fold is cleaved by the isopeptidase to give a linear peptide. The ring, loop, and tail sections of the lasso peptide are labeled. Gly-1 and Asp-9, which form the isopeptide bond, are colored red. B, genome regions in Asticcacaulis excentricus harboring the astexin-1 and astexin-2/3 gene clusters. The A, B, and C genes encode the biosynthesis of the lasso peptides. The isopeptidase E gene is divergently transcribed. Other conserved genes are also shown. C, sequences of the astexin lasso peptides. Astexins-2 and -3 are substrates for AtxE2 and share Ͼ50% sequence identity while astexin-1 is not a substrate and has a more divergent sequence. 20 amino acids (0.04 g/liter each) along with 100 mg/liter ampicillin) at OD 600 0.02. The cultures were grown at 37°C before inducing at OD 600 0.2-0.3 with 200 g/liter anhydrotetracycline (aTc) and incubation was continued at 37°C. After 20 h, the cultures were spun down at 9,800 ϫ g, and the pellets were washed with cold 1ϫ PBS. Cells were then resuspended in 5 ml of methanol and lysed by vortexing with glass beads. After reaching homogeneity, the mixture was spun down at 22,700 ϫ g for 10 min. The supernatant was collected and dried using a speed-vac. After the samples were dried, they were resuspended in 10 ml of ultrapure water and spun down at 22,700 ϫ g to remove insoluble material. The remaining liquid was passed through a 0.2 m filter and 9 ml were applied to Strata 1000 mg/6 ml C 8 columns (Phenomenex), followed by a 20-ml wash with ultrapure water and elution with 10 ml of methanol. Samples were dried again using a speed-vac and resuspended in 100 l of 50% acetonitrile in water (5000ϫ concentrated relative to the initial culture volume) for additional studies. In the case of the astexin-2 tail truncation variants, the dried material after C 8 extraction was resuspended in 500 l to give an extract 1000ϫ concentrated relative to the initial culture.
For the astexin-3 variants, a similar protocol was followed but with two changes. First, instead of growing at 37°C after induction with aTc, the cells were grown at room temperature for the same amount of time to maximize the full-length form of the peptide. Second, the final extracts were resuspended in 500 l of acetonitrile/water to make an extract 1000ϫ concentrated relative to the initial culture.
The identity of each peptide was confirmed using MALDI mass spectrometry (MALDI-MS). The concentrated extract samples were diluted 40-fold in 50:50 acetonitrile/water and mixed 1:1 with a 2.5 mg/ml solution of ␣-cyano-4-hydroxycinnamic acid matrix prior to being spotted onto an Applied Biosystems (ABI) 384 Opti-TOF 123 mm ϫ 81 mm SS plate. The samples were analyzed using an ABI 4800 MALDI-TOF-MS in positive ion mode to verify the correct product was present. For MALDI-MS on tail truncation variants, samples were diluted 80-fold, instead of 40-fold, prior to being mixed with 2.5 mg/ml solution of ␣-cyano-4-hydroxycinnamic acid matrix.
Measuring Production of Astexin-2 and -3 Variants in Extracts-For each of the alanine variants along with the wildtype for each peptide, 15 l of the concentrated extract samples were injected onto a Zorbax 300-SB C 18 analytical column (3.0 ϫ 150 mm) and the following gradient at 0.75 ml/minute was applied: 10% acetonitrile for 1 min, 10 -50% ACN over 19 min, 50 -90% ACN over 5 min, 90% acetonitrile for 5 min, and 90 -10% acetonitrile over 2 min. The traces were analyzed, and the area under the curve (AUC) corresponding to each of the variants was calculated. For astexin-2, the AUC was calculated for both the ⌬C3 and ⌬C4 peptide variants. For astexin-3, truncations are not observed, so the AUC was calculated for the full-length variant. Wild-type astexin-3, however, also contains a methionine at position 5, which becomes partially oxidized in the course of preparing the extract (15). Thus the AUC for peaks corresponding to the methionine sulfoxide species was also determined. Three separate cultures were processed for each variant to give error bars in the production level. For engineered astexin-3 tail truncation variants, the same procedure as above was followed with the following variations: 20 l of the concentrated extract samples were injected onto the Zorbax 300-SB C 18 analytical column.
In Vitro Isopeptidase Assays-For assays to assess cleavage by the isopeptidase AtxE2, varying amounts of concentrated extract samples of astexin-2 and astexin-3 variants in 50:50 acetonitrile/water were dried using a speed-vac. The dried material was reconstituted in 50 l of 1ϫ PBS pH 7.5 such that the peptide AUC was constant, resulting in concentrations of astexin-2 varying from 98.2 to 100.7 M and concentrations of 84.1-86.7 M for astexin-3. Purified AtxE2 enzyme was added to a final concentration of 233 nM. Reactions were incubated at 24°C in a thermocycler (Bio-Rad DNAEngine) for 40 min before stopping the reaction by the addition of 950 l water and application of the sample to a Strata 100 mg/1 ml C 8 column (Phenomenex), washing with 2 ml of water, and eluting with 2 ml of methanol. The methanol extracts were dried and resuspended in 33 l of 50% acetonitrile/water, followed by injection of 15 l onto a Zorbax 300-SB C 18 analytical column using the same gradient as described above to determine the extent of digestion by integrating the peaks corresponding to uncleaved (lasso) and cleaved (linear) peptide. The lasso and linear forms of the peptide often differ in retention time by at least 1 min under these conditions. To confirm cleavage, samples were diluted 40 -200-fold, mixed with a 2.5 mg/ml solution of ␣-cyano-4-hydroxycinnamic acid matrix prior to being analyzed using MALDI-MS as described above. For assays on engineered astexin-2 and astexin-3 truncations, a variant of the above procedure was followed: 20 l of the C 8 -purified extract was incubated with 620 nM AtxE2 for 3 h at 25°C in a thermocycler before stopping and purifying the reaction in the same manner as the other variants.

Results
Heterologous Expression and Purification of Astexins-2 and -3-We have previously reported plasmids for the heterologous expression of astexins-2 and -3 in E. coli (15). In our previous studies, the peptides were produced at 37°C in E. coli, leading to different patterns of C-terminal truncation. At 37°C, astexin-3 is produced primarily as a full-length peptide, with some ⌬C1 and ⌬C2 variants at longer expression times. Astexin-2 is expressed at lower levels than astexin-3 and is isolated primarily as the ⌬C3 and ⌬C4 variants with little fulllength product.
Since we were interested in maximizing peptide yield for NMR studies, we examined the expression of astexins-2 and -3 at room temperature. Cells that are shifted to room temperature after induction produce more full-length peptide than those that continue to grow at 37°C. Expression at room temperature allows for nearly exclusive production of full-length astexin-3. Truncations of astexin-2 are observed even at room temperature, but a nearly equal amount of full-length astexin-2 is also produced. To obtain sufficient material for NMR analysis we expressed astexin-3 at room temperature and HPLC-purified the full-length species (supplemental Fig. S1A). We chose to express astexin-2 at 37°C to maximize production of the ⌬C3 and ⌬C4 variants, and solved the structure of the ⌬C4 variant (supplemental Fig. S1B).

Structural Analysis of Astexins-2 and -3 and Comparison to
Astexin-1-NMR-derived models for astexin-1 in DMSO (16), astexin-1 ⌬C4 in water (25), and astexin-3 in DMSO (15) have previously been published. We have also recently revisited the structure of full-length astexin-1 in aqueous solution, and this new structure (PDB file 2N68) is in good agreement with the structure previously published by Marahiel et al. (25). To date, the structure of astexin-2 has not been determined. A sample of astexin-2 ⌬C4 was prepared in 95% H 2 O/D 2 O and COSY, TOCSY, NOESY, and C 13 -HSQC and N 15 -HSQC spectra were acquired. Well-dispersed signals in the amide region of the TOCSY and NOESY spectra allowed for a successful residue assignment. To permit a fair comparison of this structure to that of astexin-1 and astexin-3, a sample of astexin-3 was prepared in 95% H 2 O/D 2 O, and the same spectral set was acquired. Residue assignments for astexin-3 were similarly obtained. Supplemental Figs. S2-S5 show the TOCSY and NOESY spectra of the astexin-2 ⌬C4 and astexin-3 peptides. Simulated annealing of the structures was carried out using CYANA (26), and energy minimization of the structures was realized using GROMACS with explicit water (21)(22)(23)(24).
Ensembles consisting of the 20 lowest energy structures are presented in Fig. 2A for astexin-2 ⌬C4 and Fig. 2B for astexin-3. The newly resolved astexin-2 ⌬C4 model shows that the steric lock residues are amino acids 15 and 16 just as in astexin-3 (Fig.  2C). However, the steric lock residues of astexin-2 are Phe-15 and Arg-16 in contrast to astexin-3, in which the steric lock residues are Tyr-15 and Trp-16 (Fig. 2D) (15). Both astexin-2 and -3 have identical 5-residue loops with a conserved SVSGQ sequence. We calculated a backbone RMSD of 0.74 Å when aligning the first 17 residues of astexin-2 and astexin-3, showing that the overall fold of these two peptides is highly similar.
An overlay of the lowest energy structures of astexin-1, -2, and -3 from two different angles is shown in Fig. 3. Ring portions were used for alignment and are rendered as semi-transparent spheres to aid visualization. The comparison is striking in that it highlights the similarity between astexin-2 and -3 structures and their difference with astexin-1. The backbone of the rings of astexin-2 (blue) and -3 (green) are nearly planar, while the ring of astexin-1 (orange) has a kink at Pro-8 immediately preceding the isopeptide branch point at Asp-9. This sets the loop of astexin-1 in a different orientation, characterized by a sharp vertical rise that is nearly perpendicular to the plane of the ring. The loops of astexin-2 and -3 bend down first before rising and heading into the ring opening. This difference in loop structure can be attributed to the difference in loop size (4 aa versus 5 aa) and, to a lesser extent, the proline in position 8, the rigidity of which might help prime the ascent of the loop.
Effect of Astexin-2 and -3 Tail on Isopeptidase Activity-The presence of a lasso peptide isopeptidase specificity determinant in the tail portion of the astexins is not likely given that AtxE2 readily cleaves naturally occurring truncations of astexin-2 and -3 (15). To determine whether astexin-2 or -3 tail length had any effect on isopeptidase activity, we constructed genetically truncated variants of astexin-2 and astexin-3. For astexin-3, the ⌬C1, ⌬C2, ⌬C4, ⌬C5, and ⌬C6 variants were all produced as judged by MALDI-MS (supplemental Fig. S6). However, HPLC analysis of cellular extracts containing these peptides revealed the expression levels of the peptides decreased with decreasing tail length. For the ⌬C4, ⌬C5, and ⌬C6 constructs, no peptide signal was found on the HPLC using a UV detector (supplemental Fig. S7). For astexin-2, the ⌬C3, ⌬C4, ⌬C5, and ⌬C6 constructs were all produced as judged by MALDI-MS (supplemental Fig. S6). Though several of these truncated peptides were produced at levels too low to observe by HPLC, we could nonetheless determine whether these peptides were substrates for the isopeptidase using MALDI-MS. Under conditions in which the full-length astexin-2 and -3 peptides are completely   Overlays of astexins-1, -2, and -3. Astexin-1 structure is from PDB file 2N68. A, side view relative to the loop segments show that the shorter, 4 aa loop of astexin-1 (orange) is structurally divergent from the highly-similar 5 aa astexin-2 (blue) and astexin-3 (green) loops. B, view rotated 90°relative to panel A. This view shows that the tail of astexin-1 exits the ring at a different angle than does the tail of either astexin-2 or astexin-3. In all three peptides, the first 20 aa of the peptide are shown to allow for comparison. DECEMBER 25, 2015 • VOLUME 290 • NUMBER 52 cleaved (620 nM AtxE2 for 3 h), AtxE2 hydrolyzed each of the tail truncated peptides (supplemental Fig. S6). These data indicate that the tail regions of astexins-2 and -3 are dispensable with regards to the isopeptidase. However, the deletion of the tail regions greatly reduces the amount of lasso peptide produced suggesting that the tail likely plays a role in the maturation of astexins-2 and -3.

Lasso Peptide Isopeptidase Specificity
Astexins-2 and -3 Are Tolerant of Amino Acid Substitutions-To further probe the specificity determinants of AtxE2, we systematically mutagenized residues 2-14 in astexin-2 and -3 to the amino acid alanine. The residues forming the macrocycle (G1 and D9) and the steric lock residues (residues 15 and 16) were not altered. Peptides with alanine substitutions at each of the ring positions (aa 2-8) were successfully produced for both astexin-2 and astexin-3 as judged by MALDI-MS (supplemental Figs. S8 and S9). In addition, we quantified the production of each of the alanine variants by HPLC (Fig. 4). Full-length, wildtype astexin-3 is produced with a yield of 1.83 mg/liter in our heterologous expression system. All of the astexin-3 variants were produced at levels of 20 -90% of wild-type production (Fig. 4A), with the exception of the L8A variant, which was not detected in HPLC analysis. In contrast to the high yield of astexin-3, wild-type astexin-2 is produced at only 70 g/liter of culture (including both the ⌬C3 and ⌬C4 variants), 26-fold less than the expression level of astexin-3. Alanine substitutions in the ring of astexin-2 were well-tolerated with yields of these variants ranging from 75-400% of wild-type production as judged by HPLC (Fig. 4B). The exception again is the L8A variant, which was detected by mass spectrometry, but not visible on the HPLC. It is interesting to note that the wild-type astexin-2 sequence does not appear to be an optimal substrate for the maturation enzymes as the L2A, Q4A, and Q6A variants are produced at higher levels than the wild-type peptide. This may be due to the fact that positions 2, 4, and 6 within the astexin-2 ring are on the less sterically congested side of the lasso peptide, away from the isopeptide bond (Fig. 4B). Astexins-2 and -3 share the same 5 aa loop sequence, SVSGQ. The V11A, G13A, and Q14A substitutions are deleterious for production of both astexin-2 and astexin-3. This finding suggests that the lasso maturation process involves recognition of specific amino acids in the loop region of the peptide. Like other lasso peptides (6,25,(27)(28)(29)(30), astexins-2 and -3 are highly tolerant of single amino acid substitutions. Our results show that alteration of the residue immediately preceding the isopeptide bond (Leu-8 in astexins-2 and -3) is strongly deleterious for production.
Lasso Peptide Isopeptidase Has Broad Substrate Specificity-With the collection of alanine variants of astexins-2 and -3 in hand, we next turned our attention to the substrate specificity of the lasso peptide isopeptidase. We first carried out isopeptidase cleavage assays directly with cellular extracts while keeping the enzyme concentration (233 nM) and incubation time (3 h) constant. We analyzed these assays using MALDI-MS, and found evidence of hydrolysis for every variant of astexin-2 and astexin-3 except for the V11A variant of astexin-3 (supplemental Figs. S8 and S9). These data indicate that, just like the maturation enzymes, lasso peptide isopeptidase is highly tolerant of single amino acid substitutions in the peptide substrate. To get a more quantitative view of the landscape of isopeptidase activity, we next normalized the peptide concentration in assays and analyzed them using HPLC to determine the extent of cleavage. Assay conditions were set up such that cleavage of the wild-type FIGURE 4. Quantitative assessment of astexin-2 and astexin-3 variant production by HPLC. A, astexin-3 variant production levels. The error bars in the bar graphs represent the standard deviation of three biological replicates. The relative production data is mapped onto a schematic of the ring and loop portions of astexins-3 in the accompanying scheme. Circles represent residues and are colored in a gradient from red (0 -10% WT production) to green (Ͼ90% WT production). Line thickness of the circles represents absolute production level from 0 to 1.83 mg/liter of culture. Production of the L8A variant of astexin-3 was not detected by HPLC, but was observed by MALDI-MS (see supplemental Fig. S9). B, as in panel A, but for astexin-2 variants. Production of the L8A, V11A, and G13A variants of astexin-2 was too low to be observed by HPLC.
astexin-2 and -3 was incomplete at the end of the assay so as to be able to detect substitutions that either decreased or increased cleavage. The HPLC area under the curve of each astexin-2 or -3 variant was kept constant, resulting in astexin-3 variant concentrations ranging from 84.1-86.7 M and astexin-2 variants ranging from 98.2 to 100.7 M. These substrate concentrations are below the K m we had previously measured for astexin-3 cleavage by AtxE2, which is 133 M (15). AtxE2 was added at 233 nM and the enzyme reactions were incubated at room temperature for 40 min. The low production levels of the L8A variants of astexin-2 and astexin-3 and the V11A, G13A, and Q14A variants of astexin-2 precluded analysis by HPLC.
Despite the fact that astexin-2 is produced at levels ϳ25-fold lower than astexin-3, it is a better substrate for the isopeptidase than astexin-3 (Fig. 5). Since astexin-2 is produced at lower levels relative to astexin-3, the improved isopeptidase recognition of astexin-2 may ensure that both peptides are cleaved efficiently in the cell. In general, substitutions to the ring of astexin-2 and astexin-3 were well-tolerated, with the extent of hydrolysis of these variants being equivalent or better than that of the wild-type peptides (Fig. 5). The exception to this trend is the P4A variant of astexin-3, which is cleaved more poorly than wild-type astexin-3 under these assay conditions (Fig. 5A). In contrast, four of the five loop variants of astexin-3 (S10A, V11A, S12A, and Q14A) were cleaved to a lesser extent than the wild-type peptide (Fig. 5A). The one exception to this trend was the G13A variant of astexin-3, which is a conservative substitution that adds a single methyl group to the astexin-3 loop. This data points to the loop region of astexin-3 as being potentially important for isopeptidase recognition. We have previously demonstrated that the unthreaded variant of astexin-2 is not a substrate for AtxE2. Given that the loop segment of a lasso peptide is transformed into a linear epitope upon unthreading, it is possible that the loop region of threaded astexin-2 and astexin-3 is serving as a recognition element for the isopeptidase.

Discussion
In this study we set out to determine the sequence and structural characteristics that dictate the substrate specificity of a novel enzyme, lasso peptide isopeptidase. In order to generate a panel of substrates for these tests, we also investigated the tolerance of the lasso peptides astexin-2 and astexin-3 to alanine substitutions. As is the case with several lasso peptides, substitutions of the constituent amino acids with alanine were welltolerated overall. The locations within astexins-2 and -3 that were least tolerant to alanine substitutions include the Leu-8 residue immediately preceding the isopeptide bond and three positions within the loop of the peptide. Since the lasso peptide loop (residues 10 -14, SVSGQ for both astexin-2 and astexin-3) is converted from a linear epitope into the loop structure during lasso peptide maturation, it is conceivable that loop residues have an important role in the maturation process.
Lasso peptide isopeptidase is the first example of an enzyme that serves to catabolize a RiPP. Our work here indicates that this catabolic enzyme has broad substrate tolerance, a hallmark of RiPP biosynthetic enzymes. Analogous to what was observed with the maturation enzymes, alanine substitutions in the loop of astexin-3 led to decreased extents of cleavage of the peptide substrate. This finding suggests that the loop of the peptide may serve as a recognition element for the isopeptidase, an assertion that agrees with our previous data showing that unthreaded lasso peptides are not cleaved by lasso peptide isopeptidase.
The broad specificity observed in RiPP biosynthetic enzymes and now in a RiPP catabolic enzyme may be a necessary feature of these enzymes. In an enzyme with a small molecule substrate, only the enzyme is subject to amino acid substitutions arising from mutations to the gene or translation errors. In the case of RiPPs, both the substrate and the enzyme are subject to such errors. Thus it is sensible that the enzymes that assemble (and disassemble) RiPPs have a degree of substrate promiscuity. In the case of lasso peptide isopeptidase, the biological role of the enzyme remains unknown. Based on the genomic context of the astexin biosynthesis genes and the isopeptidase gene, we have previously hypothesized that the astexins may be binding a cargo and that the isopeptidase releases that cargo. Another possibility is that the astexins are acting as a signaling molecule, and the isopeptidase allows for tight control of the levels of the astexins within cells. In either scenario, some substrate promiscuity in lasso peptide isopeptidase allows for proper function of the enzyme in the case that the substrate genes are mutated or mistranslated. Work to test this theory and more fully under- FIGURE 5. Extent of cleavage of astexin-2 and astexin-3 variants by lasso peptide isopeptidase AtxE2. A, relative cleavage of full-length astexin-3 alanine variants as judged by HPLC. The error bars represent the standard deviation of three biological replicates. The V11A variant is the only one not cleaved as judged by MALDI-MS (see supplemental Fig. S9). The L8A variant is not displayed because of its low production level (see Fig. 4). B, cleavage of the ⌬C3 and ⌬C4 astexin-2 alanine variants. As in part A, the error bars reflect the standard deviation of three measurements. The asterisks for the V11A, G13A, and Q14A variants indicate that the production level of these peptides was too low for quantitative analysis by HPLC. All of the astexin-2 variants shown here are cleaved as judged by MALDI-MS.