|Home | About | Journals | Submit | Contact Us | Français|
Herein we report the bacterial expression, purification, and enzymatic characterization of the human asparaginase-like protein 1 (hASRGL1). We present evidence that hASRGL1 exhibits β-aspartyl peptidase activity consistent with enzymes designated as plant-type asparaginases, which had thus far only been found in plants and bacteria. Similar to non-mammalian plant-type asparaginases, hASRGL1 is shown to be an Ntn hydrolase for which Thr168 serves as the essential N-terminal nucleophile for intramolecular processing and catalysis, corroborated in part by abolishment of both activities through the point-mutation Thr168Ala. In light of the activity profile reported here, ASRGL1s may act synergistically with protein L-isoaspartyl methyl transferase to relieve accumulation of potentially toxic isoaspartyl peptides in mammalian brain and other tissues.
l-asparaginases (l-asparagine amidohydrolases, EC 18.104.22.168) catalyze the hydrolysis of l-asparagine to l-aspartic acid and ammonia. Enzymes with l-asparaginase activity are classified into either the bacterial-type or plant-type subfamilies based on sequence homology, structural features, and other biochemical properties. Considerable interest has been devoted to bacterial-type asparaginases for over 40 years, in part because of their antineoplastic properties (1-4). In contrast, plant-type asparaginases have not been studied as extensively and less is known about their structural and kinetic properties. Plant-type asparaginases share a high degree of amino acid sequence similarity (60-70%) with the aspartylglucosaminidase (AGA) family (EC 22.214.171.124) (5), with both enzyme groups belonging to the N-terminal nucleophile (Ntn) hydrolase protein superfamily (6-11). Enzymes in this superfamily (12) are translated as inactive precursors which undergo an autocatalytic intramolecular activation step that exposes the N-terminal nucleophile (Thr, Ser, or Cys) at the N-terminus of the newly generated β subunit. The N-terminal nucleophile also acts as the catalytic residue during the activation step. The core folding pattern shared by Ntn-hydrolases consists of a conserved αββα structure consisting of two antiparallel β-sheets between flanking α-helical layers (13, 14).
Plant-type asparaginases from plant (Lupinus luteus (10) and Arabidopsis thaliana (11)), cyanobacteria (11) (Synechocystis sp. PCC 6803 and Anabaena sp. PCC 7120), and bacteria (10, 11) (E. coli) display no hydrolytic activity towards N4-(β-N-acetyl-D-glucosaminyl)-l-asparagine (GlcNAc-l-Asn), despite their high sequence similarity to AGAs. In addition to the hydrolysis of l-asparagine, they also display significant and often higher activity towards β-aspartyl peptides, which has led to the suggestion that these enzymes be more accurately classified as β-aspartyl peptidases (EC 126.96.36.199) (10, 11, 15). Formation of isoaspartyl peptide bonds is one of the most common sources of non-enzymatic protein damage under physiological conditions, as it introduces a kink in the protein backbone that can disrupt normal folding, leading to altered susceptibility to proteolysis or loss of function (16). The β-aspartyl peptidases speculatively function to degrade these detrimental isoaspartyl peptides in the cell as they would otherwise go unnoticed by α-peptide bond specific peptidases. So far, 13 enzymes have been verified or putatively designated as β-aspartyl peptidases, but none have been identified from mammals.
A putative l-asparaginase alternatively designated asparaginase-like protein 1 (ASRGL1), Glial asparaginase (GLIAP), or CRASH (17-20), was previously cloned from rat and human cDNA libraries (17, 19, 20). Owing to its high sequence homology to a variety of asparaginases and AGAs, ASRGL1 was classified as an asparaginase though direct experimental evidence of its ability to turn over l-asparagine has been lacking. Herein, we describe the bacterial expression and characterization of the human ASRGL1 (hASRGL1) and demonstrate that this enzyme is an Ntn hydrolase that displays an activity profile consistent with other previously studied β-aspartyl peptidases, thus revealing hASRGL1 as the first mammalian enzyme of the β-aspartyl peptidase family.
Oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). Restriction enzymes, Vent DNA polymerase, T4 DNA ligase, and dNTPs were from New England Biolabs (Ipswich, MA). o-phthalaldehyde (OPA) reagent was from Agilent Technologies (Santa Clara, CA). Difco 2xYT growth medium was from Becton Dickinson. (Franklin Lakes, NJ) The β-aspartyl peptides, β-l-Asp-l-Phe, β-l-Asp-l-Ala, β-l-Asp-l-Leu, and β-l-Asp-l-Lys, as well as l-aspartic acid β-(7-amido-4-methylcoumarin), l-Asp β-methyl ester, and GlcNAc-l-Asn were purchased from Bachem (Torrance, CA). The β-aspartyl peptide β-l-Asp-l-Phe methyl ester was from Sigma-Aldrich (St. Louis, MO). All other reagents were from Sigma-Aldrich (St. Louis, MO) unless otherwise noted.
A gene encoding the human asparaginase-like protein 1 (hASRGL1) (927 bp) with an N-terminal hexahistidine affinity tag was assembled synthetically using a set of 36 codon-optimized overlapping oligonucleotides designed by the program DNAWorks (21). NcoI and EcoRI restriction sites were incorporated into the outermost 5′ and 3′ oligonucleotides respectively. The gene assembly PCR reaction consisted of the oligonucleotide mix, ThermoPol buffer, dNTPs, and Vent DNA polymerase and was carried out at an initial 95°C for 2 min, followed by 30 cycles at 95°C for 1 min, 60°C for 1 min, and 72°C for 2 min, followed by a 72°C polishing step for 10 min. After a subsequent amplification reaction with the outermost oligonucleotides, the resulting DNA product was gel purified (Qiagen), digested with NcoI/EcoRI and ligated into pET28-a (Novagen) The gene insert in the resulting plasmid, pASRGL1, was sequenced and then ultimately transformed into E. coli BL21(DE3) for subsequent expression. In addition, a hASRGL1-Thr168Ala variant was constructed by overlap- extension PCR using the pASRGL1 plasmid as template and the primer pairs shown in Supplementary Table I. The point-mutant gene was cloned analogously to hASRGL1, resulting in the plasmid pASRGL1-T168A.
E. coli BL21(DE3) cells containing pASRGL1 plasmid was cultured overnight at 37°C in 2xYT medium supplemented with 30 μg/mL kanamycin and used to inoculate fresh medium (1:100 dilution). When the absorbance at 600 nm (A600) reached 0.5-0.7, the cells were transferred to a 25°C and allowed to equilibrate for 20 min, at which point the culture was supplemented with IPTG to a final concentration of 1 mM to induce protein expression. After 16 hr of incubation at 25°C, the cells were harvested by centrifugation at 10,000 xg for 10 min. The cell pellet was resuspended in binding buffer (50mM Tris-HCl, 100 mM NaCl, 10 mM imidazole, pH 8) and placed on ice. The cells were then lysed by three passes through a French pressure cell and subsequently pelleted at 40,000 xg for 45 min. The resulting supernatant (soluble fraction) was decanted, diluted 1:1 in binding buffer and mixed with 1 mL of pre-equilibrated nickel-nitrilotriacetic acid (Ni2+-NTA) resin. After incubation for 90 min at 4°C with gentle rotation, the solution was applied to a 5 mL polypropylene column (Pierce). The resin was washed with 25 bed volumes binding buffer and 25 bed volumes wash buffer (50 mM Tris-HCl, 100 mM NaCl, 25 mM imidazole, pH 8) before the resin was incubated with 4 mL elution buffer (50 mM Tris-HCl, 100 mM NaCl, 250 mM imidazole, pH 8) for 10 min and collected drop-wise. All purification steps were performed at 4°C. The elution fraction was applied to an Amicon Ultra 10K MWCO filter, buffer exchanged against activity buffer (50 mM HEPES, 100 mM NaCl, pH 7.4), mixed with glycerol (final 10% v/v), and finally snap-frozen with liquid nitrogen and stored at −80°C. An identical protocol was followed for expression of hASRGL1-T168A. Protein concentrations were determined using a calculated extinction coefficient of 22,190 M−1cm−1(22). The expression of hASRGL1 and hASRGL1-T168A were confirmed by Western Blotting as described previously (23) using mouse monoclonal anti-polyhistidine peroxidase.
To evaluate the autocatalytic processing of hASRGL1, aliquots of both the purified wild-type and Thr168Ala enzyme variants were incubated at 37°C (24-27) and at various times, aliquots were withdrawn and analyzed by SDS/PAGE on a 4-20% precast Tris-Glycine gel (NuSep Ltd) run under reducing conditions and stained with GelCode Blue (Thermo Scientific). To quantitatively assess the effect of 37°C incubation on intramolecular processing, gel band intensities were measured using a densitometry imaging program (Quantity 1, Bio-Rad).
Catalytic activity was qualitatively determined using the fluorometric substrate laspartic-acid-β-7-amido-4- methylcoumarin (AspAMC). Briefly, aliquots of soluble crude cell lysate fractions were normalized to equal pre-lysis A600 and then diluted tenfold to 100 μL in 50 mM HEPES, 100 mM NaCl, pH 7.4. 1μL of 10 mM AspAMC in DMSO was added and mixed by pipetting in a 96 well plate (Nunc). The increase in fluorescence was monitored using a 360/40nm excitation filter and 460/40 emission filter (Synergy HT Fluorescent Platereader, BioTek) for 10 min at 25°C.
The kinetics of hASRGL1 hydrolysis were determined with freshly purified enzyme that was first incubated at 37°C for 48 hr and then stored at 4°C until needed. The formation of L-aspartic acid was determined following o-phthalaldehyde (OPA)- derivatization and HPLC analysis essentially as described by Agilent Technologies (28). Reactions of hASRGL1 (2-4 μM total enzyme) with substrate (concentrations from 0-5x KM) were carried out at 37°C in 50 mM HEPES, 100 mM NaCl, pH 7.4, to a total volume of 100 μL, and were subsequently quenched with 5 μL 12% (w/v) trichloroacetic acid. An aliquot of the quenched reaction was then mixed with a molar excess (relative to substrate) of OPA reagent and brought to a final volume of 100 μL with borate buffer. The resulting solutions were analyzed by HPLC using an Agilent ZORBAX Eclipse AAA Column (C18 reverse phase, 5μm, 4.6 x 150 mm). All reactions were done at least in triplicate and the observed rates were fit to the Michaelis-Menten equation using the program Kaleidagraph (Synergy).
The human ASRGL1 gene (GenBank accession no: BC093070) contains a number of rare E. coli codons such as the l-arginine codons AGA and AGG, whose presence has been shown to be detrimental to the expression of several recombinant proteins (29). To circumvent expression problems due to rare codons, a synthetic hASRGL1 fused to a 5′ sequence encoding a His6 affinity tag was constructed by PCR gene assembly using codon-optimized oligonucleotides. The optimized hASRGL1 gene was expressed in E. coli BL21(DE3) and protein synthesis was confirmed by Western blot analysis with an anti-His tag antibody (Figure 1). A protein band of the expected M.W. for the full-length protein (~33 kDa) was detected in both the soluble and insoluble fractions. In addition, a lower M.W. band (~18 kDa) was observed in the soluble fraction of the wild-type enzyme only. Activity assays using the fluorometric substrate AspAMC indicated the presence of l-asparagine hydrolytic activity in cells expressing hASRGL1, but not in cells expressing hASRGL1-T168A (data not shown).
hASRGL1 purification by immobilized metal ion affinity chromatography (IMAC) yielded 30 mg/L of protein with a purity > 90% as determined by SDS/PAGE. Incubation of purified wild-type enzyme at 37°C over time resulted in a gradual decrease of intensity in the band observed at ~33 kDa concomitant with an increase of intensity in the bands observed at ~18 kDa and ~15 kDa suggesting either specific proteolysis or intramolecular processing (Figure 2). Amino acid sequence alignment with other characterized β-aspartyl peptidases (Figure 3) identified Thr168 as the putative hASRGL1 N-terminal nucleophile requisite for the intramolecular processing characteristic of Ntn hydrolases. Consistent with this hypothesis, the hASRGL1-T168A variant was not processed to the lower M.W. bands as seen in both crude cell lysate and purified enzyme samples (Figures (Figures1,1, ,2).2). Approximately equivalent intensities of precursor and processed subunit gel bands were observed following incubation of hASRGL1 at 37°C for 48 hr, however; longer incubation times did not further enhance processing. Similarly, the rate of AspAMC hydrolysis by hASRGL1 increased over time following incubation of the enzyme at 37°C up to 48 hr, but the rate did not increase appreciably at later times (Figure 4).
The mass spectrum of the protein incubated at 37°C for 48 hr contained three prominent peaks of molecular masses 33,023 ± 11, 14547 ± 5 , and 18495 ± 6 Da corresponding to the predicted M.W.s of the precursor enzyme minus the N-terminal methionine residue (33,023 Da) and the α (expected mass 18,499 Da) and β (expected mass 14,551Da) subunits likely to be generated following intramolecular processing of the precursor at Thr168 cleavage site.
We determined the kinetic parameters of hydrolysis against a variety of potential hASRGL1 substrates including l-Asn; the β-aspartyl dipeptides: β-l-Asp-l-Phe, β-l-Asp-l-Phe-methyl ester, β-l-Asp-l-Ala, β-l-Asp-l-Leu, and β-l-Asp-l-Lys; l-Asp-β-methyl ester, and GlcNAc-l-Asn (Table 1). The calculation of kcat was based on the observation that 50% of the total enzyme used in a given reaction was in the processed active form based on gel densitometry analysis.
The activity profile of hASRGL1 was observed to be very similar to that of enzymes designated as isoaspartyl aminopeptidases with secondary l-asparaginase activity (10, 11) – classified as the β-aspartyl peptidase family (EC 188.8.131.52). The set of characterized enzymes in this family display a millimolar KM for l-asparagine, but a lower KM for a variety of isoaspartyl dipeptides. Moreover, these enzymes are unable to hydrolyze GlcNAc-l-Asn and l-Gln, consistent with the results seen for hASRGL1. In contrast, bacterial type II l-asparaginases such as the well studied enzymes from E. coli and Erwinia chrysanthemi, possess a micromolar affinity for l-asparagine and relatively little to no activity towards β-aspartyl peptides (11, 30). In addition, hASRGL1 was able to hydrolyze l-Asp β-methyl ester with a specificity constant greater than those for all other substrates tested.
The biochemical and structural features of plant-type asparaginases have only begun to be elucidated over the last decade. Many of these enzymes have been shown, or strongly suggested, to belong to the N-terminal nucleophile (Ntn) hydrolase superfamily (10, 11). ASRGL1 has been cloned from both rat and human cDNA libraries (17, 19, 20), and on account of its high sequence homology to a variety of asparaginases and AGAs, it was classified as an l-asparaginase, though direct experimental evidence of its ability to hydrolyze l-asparagine had not been demonstrated. Previously, ASRGL1 was shown to be concentrated in the cytosol and abundantly expressed in the brain, testis, and liver (19, 20).
Western blot, mass spectrometry and mutagenesis analyses revealed that, as its sequence homology to plant-type asparaginases implicated, hASRGL1 belongs to the Ntn hydrolase family for which Thr168 serves as the critical residue for intramolecular processing and catalytic activity. Kinetic analysis revealed that the enzyme displays both l-asparaginase and β-aspartyl peptidase activities, however, fails to hydrolyze either l-Gln or GlcNAc-l-Asn, consistent with other characterized plant-type asparaginases (10, 11). The kcat/KM values for β-aspartyl peptides containing a hydrophobic amino acid were to 2-4-fold higher relative to l-Asn, whereas substitution of a basic amino acid into the β-aspartyl position resulted in a 2-fold reduction in kcat/KM, again relative to l-Asn. We also found that an l-Asp β-methyl ester is hydrolyzed by this enzyme with a kcat/KM an order of magnitude higher than that for l-Asn, possibly because the methyl-ester is a superior leaving group relative to the ammonia of l-Asn. Further, we found that the KM value for β -l-Asp -l-Phe methyl ester (aspartame) was 4-fold higher relative to that for β-l-Asp-l-Phe, suggesting the enzyme prefers β-aspartyl dipeptides relative to longer β-aspartyl peptides, which may be related to the fact that such dipeptides are released following digestion of peptides containing a β-aspartyl linkage by certain carboxypeptidases (31). Given that the kcat/KM values for all the β-aspartyl dipeptides assayed were within 8-fold of each other, it appears reasonable to suggest that hASRGL1 is a general β-aspartyl dipeptidase, though it is possible that the optimal substrate for this enzyme has not yet been identified.
The human lysosomal AGA is the only other known mammalian enzyme for which β-aspartyl peptidase activity had been previously observed (32), however; the reported kcat/KM values of AGA for a variety of such substrates are much lower relative to those for hASRGL1.
In contrast of course, AGA displays activity towards GlcNAc-l-Asn unlike hASRGL1 and other β-aspartyl peptidases. This difference in catalytic specificity is likely a result of the divergence within the active site residues located near the l-asparagine side chain moiety (11, 20) A model of the hASRGL1 structure based on E. coli isoaspartyl aminopeptidase/l-asparaginase was created from Phyre (33) and aligned to human AGA (PDB: 1APZ; with bound l-aspartate) using PyMol (34). (Figure 5). Thus, while a near exact superimposition is observed in residues that bind the common l-aspartyl terminus of the substrate, differences are apparent in residues near the product side chain. In this region, whereas hASRGL1 predominantly contains Gly residues, human AGA contains Trp34 and Phe301 which are crucial for the binding of a sugar moiety.
The physiological function of β-aspartyl peptidases from plant, cyanobacteria, and bacteria has been speculated to be related to the hydrolysis of isoaspartyl peptides (10, 11). Formation of isoaspartyl peptide bonds is one of the most common sources of non-enzymatic protein damage under physiological conditions, as it introduces a kink in the protein backbone that can disrupt normal folding, leading to altered susceptibility to proteolysis, loss of function, or potential to elicit autoimmunity (16). Interestingly, the enzyme Protein l-isoaspartyl Methyltransferase (PIMT, EC 184.108.40.206) was found to have a high degree of specificity for l-isoaspartyl residues (35, 36). PIMT catalyzes the S-adenosyl-l-methionine (AdoMet)-dependent methylation of the α-carboxyl of an l-isoaspartyl site. Enzymatic methylation is followed by spontaneous ester hydrolysis ultimately resulting in a mixture of isoaspartyl (~70%) and aspartyl (~30%) linkages. This isoaspartyl product can then re-enter the methylation/demethylation cycle so that eventually there is a predominant shift towards the aspartyl linkage product (16). The action of PIMT has fostered the idea that the enzyme serves an important intracellular repair function to keep isoaspartyl levels low. However, the methyltransferase activity is highly influenced by both the local sequence and solution environment around the isoaspartyl modification (37, 38) and thus in some cases, a PIMT-catalyzed isoaspartyl repair cannot take place. Presumably, in these situations, the isoaspartyl-containing proteins must be degraded, though the proteolytic quality control machinery is comprised of α-peptide bond proteases. Therefore, elimination of the remaining β-aspartyl peptides may be carried out by specialized isoaspartyl peptidases – the previously postulated role for plant-type asparaginases (10).
Though PIMT is present in all mammalian tissues examined to date, levels of the enzyme are notably higher in the brain and testis similar to the distribution of hASRGL1 (19, 20, 39). While PIMT has been proposed to be important for the repair of damaged proteins within mature spermatozoa in the testes, its activity in the brain has proven to be crucial through the investigation of PIMT-deficient mice, which demonstrated the link between isoaspartate accumulation and neurological abnormalities (40-42). Mammalian β-aspartyl peptidases may play a synergistic role with PIMT in these particular tissues, which perhaps possess a critical need for the repair and/or degradation of isoaspartyl-damaged proteins. Furthermore, evidence for the existence of mammalian β-aspartyl peptidases may corroborate a proposed mechanism through which such enzymes could account for the steady state level of isoaspartyl-damaged proteins observed in PIMT-deficient mice despite the continuous and spontaneous generation of such proteins (43).
The authors thank Dr. Brent Iverson for useful discussions. This work was supported by the NIH (CA 139059) and Texas Institute for Drug and Diagnostic Development (TI3D). J.R.C also acknowledges the U.S. Department of Homeland Security (DHS) for a Graduate Fellowship under the DHS Scholarship and Fellowship Program. L.C. was supported by a fellowship from the Arnold & Mabel Beckman Foundation.
SUPPORTING INFORMATION AVAILABLE
A table of the primers used for construction of the hASRGL1-Thr168Ala variant. This material is available free of charge via the Internet at http://pubs.acs.org.