Deamidation of labile asparagine residues in the autoregulatory sequence of human phenylalanine hydroxylase Structural and functional implications Therese Solstad 1 , Raquel N. Carvalho 1 , Ole A. Andersen 2 , Dietmar Waidelich 3 and Torgeir Flatmark 1 1 Department of Biochemistry and Molecular Biology and the Proteomic Unit, University of Bergen, Norway; 2 Department of Chemistry, University of Tromsø, Norway; 3 Applied Biosystems, Applera Deutschland GmbH, Langen, Germany Two dimensional electrophoresis has revealed a micro- heterogeneity in the recombinant human phenylalanine hydroxylase (hPAH) protomer, that is the result of sponta- neous nonenzymatic deamidations of labile asparagine (Asn) residues [Solstad, T. and Flatmark, T. (2000) Eur. J. Biochem. 267, 6302–6310]. Using of a computer algorithm, the relative deamidation rates of all Asn residues in hPAH have been predicted, and we here verify that Asn32, followed by a glycine residue, as well as Asn28 and Asn30 in a loop region of the N-terminal autoregulatory sequence (residues 19–33) of wt-hPAH, are among the susceptible residues. First, on MALDI-TOF mass spectrometry of the 24 h expressed enzyme, the E. coli 28-residue peptide, L15–K42 (containing three Asn residues), was recovered with four monoisotopic mass numbers (i.e., m/z 1 of 3106.455, 3107.470, 3108.474 and 3109.476, of decreasing intensity) that differed by 1 Da. Secondly, by reverse-phase chromatography, isoaspartyl (isoAsp) was demonstrated in this 28-residue peptide by its methylation by protein- L -isoaspartic acid O-methyltransferase (PIMT; EC 2.1.1.77). Thirdly, on incubation at pH 7.0 and 37 °C of the phosphorylated form (at Ser16) of this 28-residue peptide, a time-dependent mobility shift from t R  34 min to  31 min (i.e., to a more hydrophilic position) was observed on reverse-phase chro- matography, and the recovery of the t R  34 min species decreased with a biphasic time-course with t 0.5 -values of 1.9 and 6.2 days. The fastest rate is compatible with the rate determined for the sequence-controlled deamidation of Asn32 (in a pentapeptide without 3D structural interfer- ence), i.e., a deamidation half-time of  1.5 days in 150 m M Tris/HCl, pH 7.0 at 37 °C. Asn32 is located in a cluster of three Asn residues (Asn28, Asn30 and Asn32) of a loop structure stabilized by a hydrogen-bond network. Deami- dation of Asn32 introduces a negative charge and a partial b-isomerization (isoAsp), which is predicted to result in a change in the backbone conformation of the loop structure and a repositioning of the autoregulatory sequence and thus affect its regulatory properties. The functional implications of this deamidation was further studied by site-directed mutagenesis, and the mutant form (Asn32fiAsp) revealed a 1.7-fold increase in the catalytic efficiency, an increased affinity and positive cooperativity of L-Phe binding as well as substrate inhibition. Keywords: phenylalanine hydroxylase; microheterogeneity; deamidation; asparagine; structure and function. The irreversible, spontaneous, nonenzymatic deamidation of asparagine (Asn) residues is a common post-trans- lational modification known to occur in a large number of mammalian proteins [1], and it represents an important source of protein instability at biologically relevant conditions [2,3]. The deamidation of Asn at neutral pH has been reported to proceed primarily by a succinimide mechanism involving the formation of a succinimide intermediate via nucleophilic attack on the amide carbonyl of Asn by the nitrogen of the peptide group linking the Asn to the following residue [4–6] 2 . As hydrolysis of this intermediate may occur on either side of the imide nitrogen, the Asp residue produced by the deamidation reaction will be linked to the subsequent residue by a normal 3 a-aspartyl (Asp) or by a b-aspartyl (or isoaspartic acid – isoAsp) bond. In the latter case, the b-carbon is part of the polypeptide backbone, and the a-carboxyl group is present as an atypical one carbon carboxylic acid side-chain available for methylation by isoaspartyl O-methyltransferase [7]. In general, Asn residues deami- date faster and more frequently than do glutamine residues, due to a more energetically favourable formation of a cyclic intermediate (reviewed in [8,9]). The rate of Asn deamidation in proteins has been shown to depend primarily on their nearest (to the Asn [10]) neighbour amino acid C-terminal, their localization in the 3D Correspondence to: T. Flatmark, Department of Biochemistry and Molecular Biology, University of Bergen, A ˚ rstadveien 19, N-5009 Bergen, Norway. Fax: +47 5558600, Tel.: +47 55586428, E-mail: torgeir.flatmark@ibmb.uib.no Abbreviations: hPAH, human phenylalanine hydroxylase; rPAH, rat phenylalanine hydroxylase; H 4 , biopterin (6R)- L -erythro-5,6,7,8- tetrahydrobiopterin; PIMT, protein- L -isoaspartate O-methyltrans- ferase; MALDI-TOF, matrix-assisted desorption/ionization time of flight; IPTG, isopropyl-thio-a- D -galactoside; L -Phe, L -phenylalanine; MBP, maltose binding protein; wt, wild-type. Enzyme: Phenylalanine 4-monooxygenase or phenylalanine hydroxylase (EC 1.14.16.1). (Received 9 October 2002, revised 23 December 2002, accepted 8 January 2003) Eur. J. Biochem. 270, 929–938 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03455.x structure [9,10] as well as on environmental factors such as pH, temperature and ionic strength [8,11] and including some specific ion effects [12]. Two dimensional electro- phoresis of the 50-kDa subunit of purified monkey, human and recombinant human PAH (hPAH) has revealed a marked microheterogeneity in which the individual molecular forms of the protomer have the same apparent molecular mass ( 50 kDa), but differ in their pI by about 0.1 pH unit [13] 4 . The microheterogeneity was proven to be the result of progressive, spontaneous, nonenzymatic deamidations of labile amide containing amino acid residues [13]. Based on the specific deamida- tion rates in a cellular system (expression in E. coli)and the experimental conditions in vitro required for the deamidation reactions to occur, the labile amide groups were concluded to represent Asn residues [13]. Interest- ingly, a comparison of the catalytic properties of non- deamidated and highly deamidated enzyme revealed that the catalytic efficiency (k cat /[S] 0.5 )wasalmostthreefold higher for the tetramer (as isolated by size-exclusion chromatography) with multiple deamidated protomers (generated by 24 h expression in E. coli at 28 °C) than for the essentially nondeamidated form (generated by 2 h expression) [13]. Therefore, the unambiguous identification of the Asn residues susceptible to deamidation at biologi- cally relevant conditions represents a major challenge for the characterization and understanding of the catalytic, regula- tory and stability properties of this homotetrameric enzyme 5 . Using a recently developed computer algorithm [9,10], that accurately ( 95%) predicts the relative deamidation rates of Asn residues within a single protein, when its 3D structure is known, several candidate Asn residues have been predicted in hPAH and rPAH [14]. Interestingly, based on this predictive algorithm and the 2D electrophoretic patterns of wt-hPAH and the truncated form DN(1–102)/ DC(429–452)-hPAH, two of the labile Asn residues have been located in the catalytic domain structure. In addition, three residues in the designated N-terminal autoregulatory sequence (residues 19–33), extending over the active site pocket as a ÔlidÕ [15], were also predicted to be susceptible to deamidation. This sequence contains a cluster of three Asn residues (Asn28, Asn30 and Asn32) as well as the phosphorylation site for PKA (Ser16) in close proximity, and we here present experimental data verifying the computational prediction of Asn32 as the most susceptible residue to undergo deamidation to Asp/isoAsp. Materials and methods Materials The restriction protease, enterokinase, was delivered by Invitrogen (the Netherlands). The IsoQuantÒ protein isoaspartic acid detection kit was purchased from Promega. The Sigma Chemical Co. delivered TPCK-treated trypsin and soybean trypsin inhibitor. The catalytic subunit of cAMP-dependent protein kinase (PKA) was purified to homogeneity from bovine heart and was a generous gift from S. O. Døskeland, Department of Anatomy and Cell Biology, University of Bergen. A synthetic 28-residue peptide representing the N-terminal tryptic peptide, LSD FGQETSYIEDNCNQNGAISLIFSLK (L15–K42), was synthesized by Research Genetics, AL, 6 USA. [c- 32 P]ATP and S-adenosyl- L -[methyl- 3 H]methionine were obtained from Amersham Pharmacia Biotech, U.K. Other chemicals were of the highest purity available, and some specific chemicals are referred to in the text. Site-specific mutagenesis The Asn32fiAsp mutation was introduced into the pMAL-hPAH expression system containing the entero- kinase cleavage site (D 4 K) (New England Biolabs) using the QuikChange TM site-directed mutagenesis kit (Stratagene). The following primers (provided by MWG-Biotech AG) were used for mutagenesis: (forward) 5¢-AAGACAACT GCAATCAAGATGGTGCCATATCACTGATC-3¢ and (reverse) 5¢-GATCAGTGATATGGCACCATCTTGATT GCAGTTGTCT T-3¢ (the mismatched nucleotides are shown in boldtype). The authenticity of the mutagenesis was verified by DNA sequencing in an ABI Prism TM 377 DNA Sequencer (Perkin Elmer) using the oligonucleotides malE, 13B [16] and A 674 [17] and the Big Dye TM Terminator Ready Reaction Mix (Perkin Elmer Applied Biosystems). The analysis of the electropherograms was carried out with the programs CHROMAS 1.6 (Technelysium Pty, Ltd) and CLUSTAL X (1.8). The DNA was introduced into E. coli TB1 cells by electroporation using a Gene PulserÒ II (Bio-Rad). Expression and purification of recombinant hPAH The pMAL expression system was used for the production of the wild-type fusion protein MBP-(D 4 K) ek -hPAH [18] with maltose binding protein as the fusion partner. Cells were grown at 37 °C, and expression was induced at 15 °C or 28 °C by the addition of 1 m M isopropyl thio-b- D -galactoside (IPTG); the cells were harvested after 2 h or 24 h of induction. The fusion protein was cleaved by enterokinase for 5 h at 4 °C using 4 U protease per mg fusion protein, and the tetrameric forms were isolated to homogeneity by size-exclusion chromatography [18]. Protein measurements Purified enzyme was measured by the absorbance at 280 nm, using the absorption coefficient A 280 (1 mgÆmL )1 cm )1 ) ¼ 1.63 for the fusion protein MBP-(D 4 K) ek -hPAH and 1.0 for the isolated hPAH protein [18]. Phosphorylation of hPAH The enzyme was phosphorylated by PKA as described previously [19]. The standard reaction mixture contained 15 m M Na/Hepes (pH 7.0), 0.1 m M ethylene glycol bis- (a-amino ether)-N,N,N¢,N¢-tetraacetic acid, 0.03 m M EDTA, 1 m M dithiothreitol, 10 m M MgAc 2 ,[c- 32 P]ATP, 60 l M ATP, 100 n M of the catalytic subunit of PKA and 20 l M of hPAH; 30 °C for 30 min. Trypsination of hPAH Tryptic proteolysis of hPAH was performed in 20 m M Na/Hepes buffer, pH 7.0 at 30 °C for 2 h at a trypsin to 930 T. Solstad et al. (Eur. J. Biochem. 270) Ó FEBS 2003 substrate ratio of 1 : 10 (by mass). Soybean trypsin inhibitor was added at the end with a protease to inhibitor ratio of 1 : 1.5 (by mass) for the analyses of peptides by reverse- phase chromatography. Methylation of tryptic peptides Therepairenzymeprotein- L -isoaspartate O-methyltrans- ferase (PIMT; EC 2.1.1.77) catalyses the methylation of the a-carboxyl in isoAsp residues with S-adenosyl- L -methionine as the methyl donor. The products of this reaction are the formation of peptide/protein L -isoaspartartyl methyl ester and S-adenosyl- L -homocysteine. The reaction was per- formed as described in the manual for the IsoQuantÒ protein isoaspartic acid detection kit with S-adenosyl- L -[methyl- 3 H]methionine as the methyl donor. Reverse-phase chromatography Reverse-phase chromatography of the tryptic peptides was performed using a ConstaMetric Gradient System (Labor- atory Data Control) and a 4.6 mm · 10 cm Hypersil ODS C18 column (Hewlett Packard, USA) fitted with a 2-cm guard column of the same material. Solvent A was 50 m M ammonium acetate (pH 8.0) and solvent B, 50 m M ammonium acetate in 70% (v/v) acetonitrile (pH 8.0), and a linear gradient of 10–50% solvent B was used at a flow rate of 1 mLÆmin )1 for 60 min. Samples were collected every 15 s and the elution pattern of 32 P-labelled and 3 H-labelled peptides was analysed and resolved into individual compo- nents, assuming a Gaussian distribution of each peptide, using the PEAKFIT software program (SPSS Inc., IL, USA); the ÔAutoFit-peak II-ResidualsÕ was used with the confid- ence level set at ‡ 95%. 2D electrophoresis Isoelectric focusing (IEF) was performed as described [13]. SDS/PAGE was performed at 200 V for 3–4 h on 10% (w/v) polyacrylamide gels [20]. The gels were stained by 0.5% (w/v) Coomassie Brilliant Blue R250 (Bio-Rad) in 30% (v/v) ethanol and 10% (v/v) acetic acid. The apparatus used for IEF and electrophoresis (EPS 3500XL power supply, Protean Xi 2D Cell and Tube cell) were from Bio-Rad. Finally, the 2D gels were dried on a slab gel dryer (Bio-Rad model 443) and scanned on a Hewlett Packard Scan Jet 4C/T. Mass spectrometry Matrix-assisted desorption/ionization time of flight (MALDI-TOF) spectra and tandem (MS/MS) mass spec- trometry of target peptides were acquired on a 4700 Proteomic Analyser (Applied Biosystems) in the reflectron positive-ion mode, and the spectra were mass calibrated externally. The tryptic peptides were diluted (1 : 100 or 1 : 50) in 25% (v/v) acetonitrile and 0.1% trifluoroacetic acid with a-cyano-4-hydroxycinnamic acid (Aldrich) as the matrix. Samples were spotted on the sample plate and allowed to crystallize at room temperature. The instrument was supplied with a software tool that uses a scanning algorithm to isotopically deconvolute the mass spectra; for theoretical consideration see [21]. The deconvolution method is particularly useful to detect labile Asn residues in peptides as deamidation of Asn to Asp/isoAsp increases the monoisotopic mass ([M + H] + ) of the peptide by only 1Da. Assay of hPAH activity The hPAH activity was assayed as described [18], the catalase concentration was 0.1 lgÆlL )1 and the enzyme was activated by prior incubation (5 min) with L -Phe. The enzyme source was the isolated tetrameric forms and the reaction time was 1 min 0.5% (w/v) BSA was included in the reaction mixture to stabilize the diluted, purified enzyme. The steady-state kinetic data were analysed by nonlinear regression analysis using SIGMA PLOT (Jandel Scientific Software) and the modified Hill equation of LiCata and Allewell [22] for cooperative substrate binding as well as substrate inhibition [13,23]. Results Expression, purification and 2D electrophoresis of recombinant hPAH The expression of wild-type hPAH and its Asn32fiAsp mutant form in the E. coli pMAL-system resulted in the expected high yields of the recombinant fusion protein MBP- (D 4 K) ek -hPAH [18] using an IPTG induction period of 2–24 h at 28 °C or 24 h at 15 °C. After cleavage of the affinity purified fusion proteins with the restriction protease enterokinase, the tetrameric forms were isolated by size- exclusion chromatography. Whereas the wild-type protomer gave a single band on 1D SDS/PAGE, this was not the case when subjected to 2D electrophoresis. As described previ- ously [13,24], recombinant wt-hPAH expressed as MBP- (D 4 K) ek -hPAH fusion protein for 24 h at 28 °C revealed multiple ( 5) molecular forms of the protomer (Fig. 1) that differed in their isoelectric point by about 0.1 pH unit, but shared the same apparent molecular mass ( 50 kDa). Labile asparagine residues in the regulatory domain Based on the microheterogeneity of the protomer in a double truncated form of hPAH (DN(1–102)/DC(428–452)- Fig. 1. 2D-electrophoresis pattern of full-length wt-hPAH obtained as a fusion protein after 24 h of induction in E. coli at 28 °C. Approximately 30 lg of enterokinase cleaved fusion protein MBP-(D 4 K) ek -hPAH was subjected to 2D electrophoresis and stained with Commassie Brilliant Blue. The multiple molecular forms of the protomer (denoted hPAH I-IV [13]) differed in pI by  0.1 pH unit, but shared the same apparent molecular mass of  50 kDa. Ó FEBS 2003 Deamidation of labile asparagine residues (Eur. J. Biochem. 270) 931 hPAH), expressed in E. coli, it has been concluded that at least two of the labile Asn residues are located in the catalytic domain of the protomer [13]. Furthermore, on the basis of a nearest neighbour amino acid analysis of all the Asn residues in wt-hPAH and taking into account the contribution of the 3D structure (PDB accession numbers, 1PAH and 1PHZ) to the instability of Asn residues [10], three residues in the N-terminal regulatory domain are predicted to deamidate nonenzymatically at biologically relevant conditions ([14] and Table 1). Notably, Asn32 in hPAH and rPAH (and in addition, Asn8 in rPAH) is thus predicted to be a very labile residue with a theoretical deamidation coefficient (C D ) of 0.5 and a theoretical first- order half-time of  1.5 days in 150 m M Tris/HCl, pH 7.5 at 37 °C ([14] and Table 1). Asn32 is located in a small cluster of Asn residues, including Asn28 and Asn30, that have predicted higher deamidation coefficients (7.9 and 5.0) and longer deamidation half-times (54 and 60 days). From Fig.2itisseenthatthisclusterofAsnresiduesisinaloop structure and that Asn30 Od1 is stabilized by a hydrogen bond to Gln134 Ne2. The conformation of the loop structure is further stabilized by hydrogen bonds as shown in Fig. 2. As can be seen from Table 1, none of the other Asn residues in the N-terminal regulatory domain are predicted to contribute to the microheterogeneity of the wild-type protomer. Asn58, which is located at the end of an a-helix (Ra1) [15], is particularly unlikely to undergo nonenzymatic deamidation [25]. Table 1. Asn residues in the N-terminal domain, their nearest neighbour amino acids, secondary structure position and their theoretical half-times for nonenzymatic deamidation/deamidation coefficients. N.D., not determined. Asn residue Sequence Secondary structure position First-order deamidation half-times (t 0.5 ) a (days) Deamidation coefficient (C D ) b Asn8 ENP N.D. >500 N.D. Asn28 DNC Loop 54.1 7.9 Asn30 CNQ Loop 60 5.0 Asn32 QNG Loop 1.45 0.5 Asn58 END End of Ra1 helix 32 8.4 Asn61 VNL b-Turn between Ra1 and Ra2 291 130 Asn93 TNI Middle of Ra2 helix 271 >500 a The values were obtained from the estimated half-times of penta-peptides containing the sequence GlyXxxAsnYyyGly [1]. b The values were predicted by the computer algorithm developed by Robinson and Robinson [10] based on the crystal structure data obtained for the truncated dimeric rat PAH (PDF id. code, 1PHZ) containing the regulatory and catalytic domains [15]. Fig. 2. Stereo picture of the three Asn residues in the N-terminal autoregulatory sequence of rPAH. Figure based on the crystal structure of the ligand- free, phosphorylated dimeric DC(428–452)-rPAH (PDB id. code, 1PHZ), which contains both the regulatory and the catalytic domains (residues 1–427). The N-terminal autoregulatory sequence is highly homologous in rPAH and hPAH, including the conserved Asn residues at positions 28, 30 and 32 [14]. The Asn residues and the surrounding residues are shown by ball-and-stick representation. The figure was produced using MOLSCRIPT [38]. 932 T. Solstad et al. (Eur. J. Biochem. 270) Ó FEBS 2003 Mass spectrometry To identify the labile Asn residues in the N-terminal autoregulatory sequence, tryptic peptides of wt-hPAH (Fig. 3) were analysed by MALDI-TOF using a scanning algorithm to isotopically deconvolute the mass spectra. The theoretical isotopic distribution was calculated on the basis of the elemental composition (C135 H209 N34 O48 S1) for the peptide L15–K42 (Fig. 4A,D). This acidic and hydro- phobic 28-residue tryptic peptide (containing Asn28, Asn30 and the predicted most labile residue, Asn32) revealed an isotopic mass spectrum identical to the theoretical spectrum, and on deconvolution, a single monoisotopic mass peak ([M + H] + )ofm/z 3106.514 Da when obtained from wt-hPAH isolated after 2 h at 28 °C of induction with IPTG (Fig. 4B,C). Tandem (MS/MS) mass spectrometry of this peptide revealed a fragmentation pattern which iden- tified Asn residues in positions 28, 30 and 32. By contrast, the same tryptic peptide obtained from wt-hPAH isolated after 24 h induction at 28 °C, revealed three additional monoisotopic mass peaks (m/z 3107.470, 3108.474 and 3109.476) of decreasing intensity on deconvolution of the mass spectrum (Fig. 4E,F) 7 . This increase of 1 Da for each of the additional peaks corresponds to the increase in mass as expected from deamidation of 1, 2 or 3 Asn residues, respectively. The apparent intensity ratio of 3106.455 to S (3107.470, 3108.474 and 3109.476) was  4:3andmay give an estimate of the residual amount of nondeamidated peptide (m/z 3106.455); note, however, the lower S/N ratio of the spectra in Fig. 4E than in 4B that has an effect on the accuracy of the deconvoluted spectrum. Lability of the Asn residues in the autoregulatory sequence The 28-residue tryptic peptide, L15–K42 with the cluster of Asn residues also contains the phosphorylation site (Ser16) which is a preferred substrate for PKA [26]. 32 P-labelled tetrameric hPAH (obtained after 24 h of induction at 15 °C Fig. 3. Tryptic peptides of wt-hPAH containing Asn28, Asn30 and Asn32. The alternative cleavage sites for trypsin are indicated by arrows, which may contribute to the heterogeneity of phosphopeptides after reverse-phase chromatography (Fig. 5). Fig. 4. MALDI-TOF mass spectra of the N-terminal 28-residue tryptic peptide L15–K42 obtained from wt-hPAH isolated after 2 and 24 h of induction at 28 °CinE. coli. Panels A and D represent the theoretical isotopic mass spectrum of the 28-residue peptide L15–K42 (elemental composition C135 H209 N3 O48 S1). Panels B and E represent the mass spectra of this peptide obtained from the 2 h and 24 h expressed wt-hPAH at 28 °C, respectively, and panels C and F represent the corresponding deconvoluted monoisotopic mass spectra ([M + H] + ). Note that the S/N ratio is slightly better in the isotopic mass spectrum of B than in E. 17 Ó FEBS 2003 Deamidation of labile asparagine residues (Eur. J. Biochem. 270) 933 in E. coli), that had been fully phosphorylated by PKA, was digested with trypsin and subjected to reverse-phase chro- matography that resolved one major (t R  34 min) and some minor phosphopeptides (Fig. 5, inset). The minor components represent peptides generated by alternative tryptic cleavage (see Fig. 3) and/or the presence of a mixture of peptides with either 8 Asn, Asp and isoAsp at position 32 (and eventually at positions 28 and/or 30). In order to study further the stability of the three Asn residues, the phospho- peptides were incubated at 37 °C and pH 7 (15 m M Na/ Hepes containing 1 m M dithiothreitol), and at timed intervals, aliquots were subjected to reverse-phase chroma- tography. The main phosphopeptide (with t R  34 min) revealed a time-dependent mobility shift to a more hydro- philic position with t R  31 min. By making corrections for the decay of 32 P-radioactivity and any loss of peptide, the amount of phosphopeptide with t R  34 min was found to decrease with a biphasic time-course and with calculated half-times of 1.9 days (r ¼ 0.98) and 6.2 days (r ¼ 0.92), assuming pseudo first-order kinetics for the deamidation [1,13,27]. Demonstration of isoaspartate in tryptic peptides of wt-hPAH Theenzymeprotein- L -isoaspartate O-methyltransferase (PIMT) specifically methylates isoAsp residues in peptides and a broad range of proteins [7] at substoichiometric levels [28,29,30].Thus,PIMTcanbeusedtoidentifyisoAsp formed due to nonenzymatic deamidation or spontaneous isomerization of Asp [30]. In order to detect the presence of isoAsp residues in hPAH, the wt-hPAH (obtained after 24 h induction with IPTG at 28 °C) was digested with trypsin, and the peptides methylated by PIMT with S-adenosyl- L -[methyl- 3 H]methionine as the methyl donor and analysed by reverse-phase chromatography. The 3 H-labelled peptides were resolved into several components with retention times in the range of 20–60 min (Fig. 6A), with major peaks at t R  28 min, t R  34 min, t R  36 min and t R  51 min. Amajor 3 H-labelled peptide was eluted at the end of the gradient, this was also the case for the synthetic 28-residue peptide L15–K42 (Fig. 6B). In the latter case, multiple, closely spaced, but nonresolved peaks of 3 H-radioactivity were observed and indeed expected from the MALDI-TOF mass spectrum of the peptide as isolated (Fig. 6B, inset), including the full-length form with a monoisotopic m/z of 3106.396 (theoretical monoisotopic m/z of 3106.467); such a heterogeneity was expected on the synthesis of this 28-residue peptide and possibly also as a result of partial deamidation to Asp/isoAsp during the synthesis procedure. However, the pattern of nonresolved methylated peptides eluted between 50 and 60 min (Fig. 6B) demonstrated the presence of isoAsp residues that increased markedly upon its incubation in 150 m M Tris/HCl buffer, pH 7.5 at 37 °C; i.e., at the standard incubation conditions previously selected for deamidation of model peptides [1]. In order to demonstrate that the phosphopeptides of wt-hPAH (Fig. 5) also contained isoAsp, the 32 P-labelled tryptic peptides of wt-hPAH were incubated for 1 week at 37 °C, pH 7.5 and then subjected to methylation by PIMT and reverse-phase chromatography. The elution pattern of the tryptic peptides revealed distinctly different profiles for 32 P (phosphopeptides) and 3 H (methylated peptides), but with some major overlapping peaks at t R  29 min,  36 min and  39 min. In addition, several minor over- lapping peaks were also observed, thus proving that the presence of one or more isoAsp residues in the N-terminal tryptic peptide of wt-hPAH are generated because of nonenzymatic dedamidation in vitro. Steady-state kinetic analysis of tetrameric wt-hPAH and its Asn32fiAsp mutant form The progressive deamidation events observed on 24 h expression of hPAH in E. coli have been shown to alter the catalytic properties of the tetrameric form when isolated by size-exclusion chromatography [13]. Partially deamida- ted enzyme (24 h induction at 28 °CinE. coli) revealed a 3-fold higher catalytic efficiency (k cat /[S] 0.5 ), resulting from a higher V max and an increased affinity for the substrate L -Phe, a lower affinity for its pterin cofactor and a Fig. 5. Time-course for the spontaneous deamidation of the major tryptic phosphopeptide of wt-hPAH containing the residues Asn28, Asn30 and Asn32. Full-length wt-hPAH expressed for 24 h at 28 °C was phosphorylated by PKA and subjected to digestion with trypsin. (Inset) The elution pattern of the phosphopeptides separated by reverse-phase chromatography on a Hypersil C18 column; the column was equilibrated with 50 m M ammonium acetate (pH 8.0) and the peptides eluted with a linear gradient of 10–50% (v/v) of 50 m M ammonium acetate in 70% (v/v) acetonitrile (pH 8.0) at a flow-rate of 1mLÆmin )1 . Detection of the 32 P-radioactivity revealed a heteroge- neity of the phosphopeptides that may be related to alternative cleavage sites for trypsin (see Fig. 4) and partial deamidation(s) of AsnfiAsp and AsnfiisoAsp; main peak at t R  34 min. (Main figure) The phosphopeptides were further incubated in phosphorylation medium at pH 7.0 and 37 °C, and at timed intervals, aliquots ( 150 lgpep- tide) were subjected to reverse-phase chromatography. Fractions (250 lL) were collected every 15 s followed by scintillation counting and analysis of the data by the PEAKFIT software program. Each data point represents the average value obtained in three separate experi- ments, and the two lines were calculated by linear regression analysis with the correlation coefficients r 1 ¼ 0.98 and r 2 ¼ 0.92. The time- course for the remaining 32 P-radioactivity in the t R  34 min peak (log d.p.m. t R  34 min) was thus resolved into two deamidation half-times of 1.9 and 6.2 days, assuming pseudo first-order kinetics. 934 T. Solstad et al. (Eur. J. Biochem. 270) Ó FEBS 2003 pronounced substrate ( L -Phe) inhibition when compared to the nondeamidated tetramer (2 h induction at 28 °C). Steady-state kinetic analysis of the Asn32fiAsp mutant form (2 h induction at 28 °C), demonstrated kinetic properties that where comparable qualitatively to those observed for highly deamidated wt-hPAH with a 1.7-fold increase in its catalytic efficiency, a 34% decrease in the [S] 0.5 -value for L -Phe, increased cooperativity on L -Phe binding and substrate inhibition (Fig. 7, Table 2). Discussion The observed microheterogeneity of recombinant wt-hPAH on isoelectric focusing and 2D electrophoresis (Fig. 1) has been demonstrated to be the result of spontaneous non- enzymatic deamidation of labile amide containing amino acid residues and has implications both for the catalytic efficiency and stability properties of the enzyme [13]. Based on the rate of the multiple deamidation reactions, their pH and temperature dependence, the labile amide residues were concluded to be Asn, and at least two of them were found to be present in the double truncated form DN(1–102)/ DC(428–452)-hPAH including the catalytic core domain of the enzyme [13]. Deamidation of Asn residues can occur by several alternative mechanisms, but in proteins and peptides the most common is a nonenzymatic deamidation via the b-aspartyl shift mechanism (see Introduction), that proceeds with a high frequency in proteins at neutral to basic pH [31]. Asn, followed by Gly, Ser, Thr or Lys residues are most commonly subjected to deamidation [9,31], particularly when the Asn residue is located in a flexible segment of the protein [32]. Under physiological conditions (with respect to temperature, pH and ion composition) the ratio of Asp to isoAspisreportedtobe 1 : 2 for Asn–Gly or Asn–Ser sequences [33], and ratios of  1 : 3 have been reported in model pentapeptides [6,34], however, conformational con- straints in the protein may have an effect on this ratio. Labile Asn residue(s) in the N-terminal regulatory domain of wt-hPAH On the basis of nearest neighbour analyses of the Asn residue in wt-hPAH and the recently solved 3D crystal structures of rPAH and hPAH, a computer algorithm has predicted that the most labile Asn residue in rPAH/hPAH is Asn32, together with Asn8 in rPAH ([14] and Table 1). Its nearest neighbour amino acid (Gly33), in a loop structure, favours the deamidation of Asn32 with the formation of Asp and isoAsp at a ratio of  1 : 2 [33]. MALDI-TOF mass spectrometry analyses have confirmed this prediction. Thus, the 28-residue tryptic peptide {residues 15–42 with a theoretical monoisotopic ([M + H] + ) m/z of 3106.467, see Fig. 4A or D} obtained from wt-hPAH isolated after 24 h induction at 28 °C revealed on deconvolution of the mass spectrum four monoisotopic peaks (Fig. 4F) at m/z values of decreasing intensity (3106.455, 3107.474, 3108.474 and 3109.476; Fig. 4F) 9 .Thesem/z values correspond to that expected for nondeamidated (m/z 3106.455), mono-deami- dated (+ 1 Da), double-deamidated (+ 2 Da) and triple- deamidated (+ 3 Da) forms of the peptide. This conclusion is further supported by the existence of only one monoiso- topic peak (i.e., at m/z 3106.514) for the peptide obtained from the 2 h expressed enzyme (Fig. 4C). That this peptide represents the nondeamidated form, with Asn residues at positions 28, 30 and 32, was confirmed by the fragmentation pattern obtained by tandem (MS/MS) mass spectrometry. Thus, the MALDI-TOF spectra in Figs 4E, F are compa- tible with a progressive deamidation of the Asn residues at positions 28, 30 and 32. However, attempts to further confirm this conclusion by the MS/MS fragmentation pattern of the selected precursor ion were not successful. As Fig. 6. Reverse-phase chromatography of 3 H-labelled methylated tryptic peptides of wt-hPAH and the synthetic N-terminal peptide, L15–K42. (A) The pattern of 3 H-labelled methylated tryptic peptides obtained from wt-hPAH isolated after induction for 24 h at 28 °C. Peptides from  2.5 mg of enzyme were subjected to methylation and reverse- phase chromatography as described in the legend to Fig. 5. Fractions (250 lL) were collected every 15 s and the 3 H-radioactivity was counted. (B) Approximately 100 lg of the 28-residue synthetic peptide (L15–K42 of hPAH) was subjected to methylation by protein iso- aspartyl methyltransferase and then to reverse-phase chromatography. The bottom trace (thin line) represents the radioactivity profile of the peptide as isolated and the upper trace (thick line) the profile obtained after its incubation for 7 days at 37 °C in 150 m M Tris/HCl, pH 7.5. (Inset) The MALDI-TOF mass spectrum of the synthetic peptide isolated with a main component of m/z 3106.396 (theoretical m/z-value for the peptide is 3106.467) and several minor components, including peptides with nonreleased blocking groups (in the high-molecular- mass region). Ó FEBS 2003 Deamidation of labile asparagine residues (Eur. J. Biochem. 270) 935 a further approach to study the multiple deamidations in this 28-residue tryptic peptide, the wt-hPAH was labelled with 32 P-phosphate (at Ser16) and the phosphopeptides separated by reverse-phase chromatography (Fig. 6, inset). When the phosphopeptides were further incubated at 37 °C and pH 7.0 (15 m M Na/Hepes), a half-time of 1.9 days (Fig. 6) was estimated for the nonenzymatic deamidation of the major phosphopeptide (as estimated by its time- dependent change in retention time from t R  34 min to a more hydrophilic position (with t R  31 min). This is close to the half-time estimated for an Asn in a pentapeptide containing the identical nearest neighbour amino acids xQNGx as in hPAH, i.e.,  1.5 days when incubated at 37 °C and 150 m M Tris buffer, pH 7.5 [1] 10 . As there is no other Asn–Gly sequence in the human enzyme, our data (Fig. 6) support the conclusion that Asn32 is the most labile Asn residue in hPAH. The longer t 0.5 –value (6.2 days) calculated from the second slope in Fig. 6 most likely reflects the deamidation of Asn28/Asn30 which are both predicted to be more stable than Asn32 (Table 1), notably Asn30 which is stabilized by a hydrogen bond to Gln134 (Fig. 2). The deamidations of these Asn residues were further substantiated by the chromatography 11 of peptides containing 32 P-labelled S16 and 3 H-labelled (methylated) isoAsp of wt-hPAH on reverse-phase chromatography and the demonstration of isoAsp in a synthetic peptide contain- ing both the full-length species, L15–K42 – of the expected monoisotopic mass 3106.467 – and a number of lower molecular mass peptides, recovered as a mixture of peptides on reverse-phase chromatography. Deamidation of Asn32 during isoelectric focusing and 2D electrophoresis The rapid rate of deamidation of Asn32 in wt-hPAH observed in vitro in the present study (Fig. 6) explains why we have not been able to obtain a single component on isoelectric focusing and 2D electrophoresis of the 2 h (28 °C) expressed enzyme [13] as expected from its MALDI-TOF mass spectrum. Thus, no microheterogeneity was observed for the Asn containing peptides on MALDI- TOF and tandem (MS/MS) mass spectrometry of tryptic peptide obtained from wt-hPAH (2 h induction at 28 °C), in particular not for the sequence L15–K42 (containing the Fig. 7. Effect of L-Phe (A and C) and H 4 biopterin (B and D) concentration on the catalytic activity of recombinant wt-hPAH and hPAH(Asn32fiAsp) obtained after 2 h of induction in E. coli at 28 °C. Assay conditions and nonlinear regression ana- lysis are described in the Methods section. In each graph, both the experimental (closed circle) and fitted (open circle) data are shown. The kinetic constants obtained are presented in Table 2. Table 2. Steady-state kinetic parameters for recombinant wt-hPAH and hPAH(Asn32fiAsp). Wt-hPAH was obtained after 2 h and 24 h of induction with IPTG in E. coli at 28 °C. The Asn32fiAsp mutant form was obtained after 2 h at 28 °C. The PAH activity was assayed and the apparent kinetic constants were calculated by nonlinear regression analysis as described in the Materials and methods section. The substrate concentrations were 1 m ML -Phe (H 4 biopterin variable) and 75 l M H 4 biopterin ( L -Phe variable). Values are shown as means ± SEM, n ¼ 3. 18 Enzyme tetramer L -Phe H 4 biopterin V max a (nmolÆmin )1 Æmg )1 ) ½S 0:5 a (l M ) k cat /[S] 0.5 (l M Æmin )1 ) h Substrate inhibition V max (nmolÆmin )1 Æmg )1 ) K m b (l M ) Wt (2 h) 3484 ± 162 238 ± 18 0.72 1.5 ± 0.1 ‚ 5325 ± 137 31 ± 3 Wt (24 h) 4946 ± 281 135 ± 11 1.83 1.9 ± 0.1 + 6392 ± 281 40 ± 5 N32D 3736 ± 470 159 ± 30 1.20 2.2 ± 0.1 + 4725 ± 133 34 ± 2 a The kinetic parameters were calculated from a modified Hill equation as described in the Materials and methods section. b The kinetic parameters were calculated from the Michaelis–Menton equation. 936 T. Solstad et al. (Eur. J. Biochem. 270) Ó FEBS 2003 predicted most labile residue Asn32; Fig. 4B,C). hPAH requires  20 h to reach the equilibrium position in the pH gradient at  20 °C, i.e., conditions that favour deamida- tion of at least Asn32 and, therefore, represents a major contributor to the observed microheterogeneity of hPAH isolated after 2 h of induction with IPTG in E. coli [13]. Thus, the double spot seen, ) e.g., for the most basic component of the protomers on 2D electrophoresis (denoted hPAH I [13]) for the 24 h enzyme – (Fig. 1) is most likely explained by a minor partial deamidation of Asn32 during isoelectric focusing for 20 h (8 M urea,  20 °C) and represents deamidated protomers where Asp/isoAsp have not yet reached the equilibrium position in the pH gradient. 12 Structural and functional implications of Asn32 deamidation In the reported crystal structure of DC(429–452)-rPAH (PDB id. code, 1PHZ) [15], Asn32 is located in the loop structure including residues D27–G33 (Fig. 2), preceding the first b-sheet (Rb1) of the regulatory domain [15], and is stabilized by several hydrogen bonds, including a hydrogen bond between Asn30 Od1 and Gln134 Ne2(seeFig.2). Consequently, the deamidation of Asn32 is likely to affect the higher order structure of the NH 2 terminus in several ways. First, on deamidation of Asn32fiAsp/isoAsp, the resulting charge shift will lead to a repulsive electrostatic interaction between the negatively charged carboxyl groups of Asp/isoAsp32 and Asp84. Moreover, the presence of isoAsp at position 32 in one of the generated isoforms will also change the backbone conformation by adding an extra carbon (methylene group) to the polypeptide chain [35]. Interestingly, in the reported crystal structure of rPAH, the sequence around Asn32 revealed high displace- mentfactorsofupto105A ˚ 2 [15] reflecting poor electron density. High displacement factors are a consequence of dynamic disorder (vibrations of the atoms) in the crystal as well as static disorder caused by discrepancies between the different unit cells in the crystal. Combining the crystal structure information with the results of this study, we conclude that the recombinant enzyme used in the crystal- lographic study of rPAH (PDB id. code, 1PHZ) most likely contained a mixture of isomeric dimer forms with either Asn, Asp and isoAsp at position 32 in the protomers. Thus, the conserved residue Asn32 is expected to be deamidated to the same extent in hPAH as in rPAH, as its nearest neighbour amino acids and the loop structure are the same in the two proteins. The perturbation of the overall conformation of the loop structure as a result of deamidation of Asn32 to Asp/isoAsp have immediate implications for the function of the homotetrameric enzyme as they occur in a conformationally sensitive part of the protein. Thus, the loop structure including Asn32 is part of the designated autoregulatory sequence (residues 19–33) [15] extending like a ÔlidÕ over the active site pocket and thus may control the access of substrate and pterin cofactor to this site [15,36]. Phosphory- lation of hPAH at Ser16 by PKA results in a 1.4-fold increase in the basal activity and a 1.7-fold increase in catalytic efficiency [37]. Molecular modelling based on the crystal structure of the recombinant rat enzyme has revealed that phosphorylation induces a local conformational change that is in agreement with the observed increased accessibility of the substrate to the active site [37]. The functional effect of deamidation of Asn32fiAsp/isoAsp is rather similar and can be explained by a related mechanism. Thus, the substitution of Asn32fiAsp by site-directed mutagenesis results in a 1.7-fold increase in its catalytic efficiency, a 34% decrease in the [S] 0.5 -value for L -Phe, an increased Hill coefficient of substrate binding as well as substrate inhibi- tion (Fig. 7, Table 2). 13 These changes in kinetic properties are all characteristics observed because of multiple deami- dations of wt-hPAH on 24 h expression of the recombinant enzyme in E. coli [13]. It should be noted, however, that the main-chain loop conformation may be slightly differ- ent in the Asn32fiAsp/isoAsp deamidated form and the Asn32fiAsp mutant form, which may account for the quantitative differences observed between the kinetic pro- perties of the two events. That Asn32 has a regulatory function in hPAH is further supported by experiments demonstrating that both deamidation of Asn32 and muta- tion of Asn32fiAspareaccompaniedbya 20% increase in the initial rate of phosphorylation at Ser16 by PKA [39]. Thus, it appears that phosphorylation of Ser16 and deamidation of Asn32 act synergistically in the regulation of hPAH. Acknowledgements This work was supported by grants from the Research Council (NFR), the Norwegian Council on Cardiovascular Diseases, Rebergs legat, L. Meltzers Høyskolefond, the Novo Nordisk Foundation, the European Commission and the Fundac¸ a ˜ oparaaCieˆ ncia e Tecnologia, Portugal (SFRH/BD/1100/2000). We thank Ali Sepulveda Mun ˜ oz for expert technical assistance in preparing the recombinant hPAH fusion proteins. References 1. Robinson, N.E. & Robinson, A.B. (2001) Molecular clocks. Proc. Natl Acad. Sci. 98, 944–949.11. 2. Robinson, A.B., McKerrow, J.H. & Cary, P. (1970) Controlled deamidation of peptides and proteins: an experimental hazard and a possible biological timer. Proc.NatlAcad.Sci.66, 753–757. 3. McKerrow, J.H. & Robinson, A.B. (1971) Deamidation of asparaginyl residues as a hazard in experimental protein and peptide procedures. Anal. Biochem. 42, 565–568. 4. Bornstein, P. & Balian, G. (1970) The specific nonenzymatic cleavage of bovine ribonuclease with hydroxylamine. J. Biol. Chem. 245, 4854–4856. 5. 19 Meinwald, Y.C., Stimson, E.R. & Scheraga, H.A. (1986) Deami- dation of the asparaginyl-glycyl sequence. Int. J. Pept. Protein Res. 28, 79–84. 6. Geiger, T. & Clarke, S. (1987) Deamidation, isomerization, and racemization at asparaginyl and aspartyl residues in peptides. Succinimide-linked reactions that contribute to protein degrada- tion. J. Biol. Chem. 262, 785–794. 7. Aswad, D.W. (1984) Stoichiometric methylation of porcine adrenocorticotropin by protein carboxyl methyltransferase requires deamidation of asparagine. Evidence for methylation at the alpha-carboxyl group of atypical 1-isoaspartyl residues. J. Biol. Chem. 259, 10714–10721. 8. Wright, H.T. (1991) Nonenzymatic deamidation of asparaginyl and glutaminyl residues in proteins. Crit. Rev. Biochem. Mol. Biol. 26, 1–52. Ó FEBS 2003 Deamidation of labile asparagine residues (Eur. J. Biochem. 270) 937 9. Robinson, N.E. (2002) Protein deamidation. Proc. Natl Acad. Sci. 99, 5283–5288. 10. Robinson, N.E. & Robinson, A.B. (2001) Prediction of protein deamidation rates from primary and three-dimensional structure. Proc.NatlAcad.Sci.98, 4367–4372. 11. Flatmark, T. (1967) Multiple molecular forms of bovine heart cytochrome c. V. A comparative study of their physicochemical properties and their reactions in biological systems. J. Biol. Chem. 242, 2454–2459. 12. Robinson, N.E. & Robinson, A.B. (2001) Deamidation of human proteins. Proc. Natl Acad. Sci. 98, 12409–12413. 13. Solstad, T. & Flatmark, T. (2000) Microheterogeneity of recombinant human phenylalanine hydroxylase as a result of nonenzymatic deamidations of labile amide containing amino acids. Effects on catalytic and stability properties. Eur. J. Biochem. 267, 6302–6310. 14. Carvalho, R.M.N., Solstad, T., Robinson, N.E., Robinson, A.B. & Flatmark, T. (2002) Possible contributions of labile asparqine residues to differences in regulatory properties of human and rat phenylalanine hydroxylase. In Chemistry and Biology of Pteridines and Folates (Milstien, S., Kapatos, G., Levine, R.A. & Shane, B., eds), pp. 103–107, Kluwer Academic Publishers, Boston. 15. Kobe, B., Jennings, I.G., House, C.M., Michell, B.J., Goodwill, K.E., Santarsiero, B.D., Stevens, R.C., Cotton, R.G. & Kemp, B.E. (1999) Structural basis of autoregulation of phenylalanine hydroxylase. Nat. Struct. Biol. 6, 442–448. 16. Eiken, H.G., Knappskog, P.M., Apold, J. & Flatmark, T. (1996) PKU mutation G46S is associated with increased aggregation and degradation of the phenylalanine hydroxylase enzyme. Hum. Mutat. 7, 228–238. 17. Erlandsen, H., Bjørgo, E., Flatmark, T. & Stevens, R.C. (2000) Crystal structure and site-specific mutagenesis of pterin-bound human phenylalanine hydroxylase. Biochemistry 39, 2208–2217. 18. Martı ´ nez, A., Knappskog, P.M., Olafsdottir, S., Døskeland, A.P., Eiken, H.G., Svebak, R.M., Bozzini, M., Apold, J. & Flatmark, T. (1995) Expression of recombinant human phenylalanine hydro- xylase as fusion protein in Escherichia coli circumvents proteolytic degradation by host cell proteases. Isolation and characterization of the wild-type enzyme. Biochem. J. 306, 589–597. 19. Døskeland, A.P., Martı ´ nez, A., Knappskog, P.M. & Flatmark, T. (1996) Phosphorylation of recombinant human phenylalanine hydroxylase: effect on catalytic activity, substrate activation and protection against non-specific cleavage of the fusion protein by restriction protease. Biochem. J. 313, 409–414. 20. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 21. Wehofsky, M., Hoffmann, R. & Hubert, M. (2001) Isotopic deconvolution of matrix-assisted laser desorption/ionization mass spectra for substance-class specific analysis of complex samples. Eur. J. Mass Spectrom. 7, 39–46. 22. LiCata, V.J. & Allewell, N.M. (1997) Is substrate inhibition a consequence of allostery in aspartate transcarbamylase? Biophys. Chem. 64, 225–234. 23. Bjørgo, E., de Carvalho, R.M. & Flatmark, T. (2001) A com- parison of kinetic and regulatory properties of the tetrameric and dimeric forms of wild-type and Thr427fiPro mutant human phenylalanine hydroxylase: contribution of the flexible hinge region Asp425-Gln429 to the tetramerization and cooperative substrate binding. Eur. J. Biochem. 268, 997–1005. 24. Døskeland, A.P. & Flatmark, T. (1996) Recombinant human phenylalanine hydroxylase is a substrate for the ubiquitin- conjugating enzyme system. Biochem. J. 319, 941–945. 25. Kosky, A.A., Razzaq, U.O., Treuheit, M.J. & Brems, D.N. (1999) The effects of alpha-helix on the stability of Asn residues: deami- dation rates in peptides of varying helicity. Protein Sci. 8, 2519– 2523. 26. Døskeland, A.P., Schworer, C.M., Døskeland, S.O., Chrisman, T.D., Soderling, T.R., Corbin, J.D. & Flatmark, T. (1984) Some aspects of the phosphorylation of phenylalanine 4-mono- oxygenase by a calcium-dependent and calmodulin-dependent protein kinase. Eur. J. Biochem. 145, 31–37. 27. Flatmark, T. (1966) On the heterogeneity of beef heart cytochrome c. A kinetic study of the non-enzymatic deamidation of the main subfractions (Cy I–Cy III). Acta. Chem. Scand. 20, 1487–1496. 28. Freitag, C. & Clarke, S. (1981) Reversible methylation of cyto- skeletal and membrane proteins in intact human erythrocytes. J. Biol. Chem. 256, 6102–6108. 29. O’Connor, C.M. & Clarke, S. (1984) Carboxyl methylation of cytosolic proteins in intact human erythrocytes. Identification of numerous methyl-accepting proteins including hemoglobin and carbonic anhydrase. J. Biol. Chem. 259, 2570–2578. 30. 20 Johnson, B.A. & Aswad, D.W. (1991) Optimal conditions for the use of protein 1-isoaspartyl methyltransferase in assessing the isoaspartate content of peptides and proteins. Anal. Biochem. 192, 384–391. 31. Wright, H.T. (1991) Sequence and structure determinants of the nonenzymatic deamidation of asparagine and glutamine residues in proteins. Protein Eng. 4, 283–294. 32. Clarke, S. (1987) Propensity for spontaneous succinimide forma- tion from aspartyl and asparaginyl residues in cellular proteins. Int. J. Pept. Protein. Res. 30, 808–821. 33. Aswad, D.W., Paranandi, M.V. & Schurter, B.T. (2000) Iso- aspartate in peptides and proteins: formation, significance, and analysis. J. Pharm. Biomed. Anal. 21, 1129–1136. 34. Tyler-Cross, R. & Schirch, V. (1991) Effects of amino acid sequence, buffers, and ionic strength on the rate and mechanism of deamidation of asparagine residues in small peptides. J. Biol. Chem. 266, 22549–22556. 15 35. Johnson, B.A., Langmack, E.L. & Aswad, D.W. (1987) Partial repair of deamidation-damaged calmodulin by protein carboxyl methyltransferase. J. Bio. Chem. 262, 12283–12287. 36. Jennings, I.G., Cotton, R.G. & Kobe, B. (2000) Functional ana- lysis, using in vitro mutagenesis,ofaminoacidslocatedinthe phenylalanine hydroxylase active site. Arch. Biochem. Biophys. 384, 238–244. 37. Miranda, F.F., Teigen, K., Tho ´ ro ´ lfsson, M., Svebak, R.M., Knappskog, P.M., Flatmark, T. & Martı ´ nez, A. (2002) Phos- phorylation and mutations of Ser16 in human phenylalanine hydroxylase. Kinetic and structural effects. J. Biol. Chem. 277,40937–40943. 16 38. Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crys- tallog. 24, 946–950. 39. Carvalho, R.N., Solstad, T., Bjørgo, E., Barroso, J.F. & Flatmark, T. (2003) Deamidations in recombinant human phenylalanine hydroxylase. Identification of labile asparagine residues and functional characterization of AsnfiAsp mutant forms. J. Biol. Chem., in press. 938 T. Solstad et al. (Eur. J. Biochem. 270) Ó FEBS 2003 . Deamidation of labile asparagine residues in the autoregulatory sequence of human phenylalanine hydroxylase Structural and functional implications Therese. micro- heterogeneity in the recombinant human phenylalanine hydroxylase (hPAH) protomer, that is the result of sponta- neous nonenzymatic deamidations of labile asparagine (Asn)