Article in press - uncorrected proof

Biol. Chem., Vol. 389, pp. 91–98, January 2008 • Copyright by Walter de Gruyter • Berlin • New York. DOI 10.1515/BC.2008.006

Cloning, expression and characterization of -degrading from tomato (Solanum lycopersicum)

Yoann Huet1,a, Jochen Strassner2 and Andreas named inverzincins (Ding et al., 1992). It was demonstrat- Schaller1,* ed by site-directed mutagenesis that the two histidine residues are involved in the coordination of the zinc ion, 1 Institute of Plant Physiology and Biotechnology, while the glutamate is required for catalysis (Perlman et University of Hohenheim, D-70593 Stuttgart, Germany al., 1993; Becker and Roth, 1995). 2 Altana Pharma AG, D-78467 Konstanz, Germany Inverzincins generally accept a wide variety of peptide * Corresponding author substrates and appear to lack selectivity for specific e-mail: [email protected] sequence motives in the primary structure of their sub- strates. Indeed, IDE was shown to cleave many different peptides, including insulin and glucagon (Kirschner and Abstract Goldberg, 1983), natriuretic peptides (Muller et al., 1992), transforming growth factor a (Garcia et al., 1989; Gehm A cDNA encoding insulin-degrading enzyme (IDE) was and Rosner, 1991), b-endorphin and dynorphins (Safavi cloned from tomato (Solanum lycopersicum) and et al., 1996), growth hormone releasing factor (Garcia expressed in Escherichia coli in N-terminal fusion with et al., 1989), amylin (Bennett et al., 2000) and amyloid glutathione S-. GST-SlIDE was characterized b peptide (Qiu et al., 1998; Vekrellis et al., 2000) at unre- as a neutral thiol-dependent metallopeptidase with insu- lated sites. However, despite the degeneracy of cleavage linase activity: the recombinant enzyme cleaved the oxi- sites, these peptide substrates are recognized by IDE dized insulin B chain at eight peptide bonds, six of which and cleaved with high affinity, while other biologically are also targets of human IDE. Despite a certain prefer- important peptides are not cleaved at all. IDE can there- ence for proline in the vicinity of the cleavage site, syn- fore not be considered as a general peptidase for non- thetic peptides were cleaved at apparently stochastic specific peptide turnover (Duckworth, 1990; Safavi et al., positions indicating that SlIDE, similar to IDEs from other 1996). Similarly, mitochondrial processing peptidase organisms, does not recognize any particular amino acid (MPP) and plant stromal processing peptidase (SPP) are motif in the primary structure of its substrates. Under specific inverzincins, which process mitochondrial and " steady-state conditions, an apparent Km of 62 7 mM and plastid precursor proteins to release the N-terminal tar- " -1 -1 a catalytic efficiency (kcat/Km)of62 15 mM s were geting peptides after import into the respective organelle determined for Abz-SKRDPPKMQTDLY(NO3)-NH2 as the (Vander Vere et al., 1995; Richter and Lamppa, 1998; substrate. GST-SlIDE was effectively inhibited by ATP Gakh et al., 2002). The targeting peptides are variable in at physiological concentrations, suggesting regulation of length and primary structure, and there is little sequence its activity in response to the energy status of the cell. conservation at MPP cleavage sites in yeast and plant While mammalian and plant IDEs share many of their bio- systems (Branda and Isaya, 1995; Glaser et al., 1998). chemical properties, this similarity does not extend to Pre-sequence (PrePs), which also belong to their function in vivo, because insulin and the b-amyloid the M16 family and lack specificity towards particular peptide, well-established substrates of mammalian IDEs, peptide bonds are responsible for the further degradation as well as insulin-related signaling appear to be absent of targeting peptides after release from protein precur- from plant systems. sors (Sta˚ hl et al., 2002; Bhushan et al., 2003; Moberg et al., 2003). A recent report on the structure of human Keywords: insulinase; insulysin; metalloprotease; plant; IDE in complex with several peptide substrates (Shen et proteolysis. al., 2006) sheds light on the mechanisms underlying sub- strate recognition by M16 peptidases. The four domains of the enzyme fold to enclose a chamber, just large Introduction enough to accommodate peptides of up to about 50 amino acids. Additional selectivity is provided by favor- Insulin-degrading enzyme (IDE) is an evolutionary well able interactions between the enzyme and the N-termi- conserved zinc-dependent of the nus as well as the cleavage sites within the substrate pitrilysin family and belongs to the M16A family of pep- peptides. Moreover, the charge distribution in the cata- tidases, according to the MEROPS classification (Raw- lytic cavity explains why peptides with positive charges lings et al., 2004). M16 peptidases are characterized by in proximity of their C-termini are usually poor substrates the conserved HXXEH motif for zinc binding, which is a (Shen et al., 2006). functional inversion of the more common HEXXH motif IDE, as the name suggests, has first been implicated of other metalloproteases, and they have thus been in the regulation of insulin signaling, and experimental evidence confirming its role in the intracellular degrada- a Present address: Plant Biology and Pest Control, Jules Verne tion of this hormone continues to cumulate since its University of Picardie, F-80000 Amiens, France. activity was first described 50 years ago (Duckworth

2008/236 Article in press - uncorrected proof

92 Y. Huet et al. et al., 1988, and references therein). More recently, atten- present in the purified fraction and likely represent deg- tion has focused on the capacity of IDE to degrade b- radation or arrested translation products as they immu- amyloid peptides involved in the neurodegenerative noreact with an antibody directed against the N-terminus Alzheimer disease (Kurochkin, 2001; and references of the fusion protein, including the GST moiety and the therein). Indeed, IDE-deficient rodents suffer not only first 142 N-terminal residues of SlIDE (not shown). The from hyperinsulinemia and glucose intolerance, but also presence of such low-molecular weight fragments during accumulate high levels of amyloid peptides in the brain bacterial expression appears to be a recurring problem, (Farris et al., 2003; Miller et al., 2003). A broader role for as it was also reported for the recombinant human IDE IDE is also supported by the fact that IDE cleaves many (Chesneau and Rosner, 2000). biologically important peptides other than insulin, by its GST-SlIDE was confirmed as a bona fide IDE using localization in cell types that do not bind or internalize MALDI-TOF mass spectrometry to characterize its activ- insulin (Duckworth, 1988). Moreover, IDE homologs are ity in vitro. The oxidized insulin B chain was rapidly pro- found in organisms lacking insulin-related peptides and cessed at the Glu13-Ala14, Ala14-Leu15 and Tyr16-Leu17 pep- insulin signaling, including plants. We report here the bio- tide bonds (complete processing in less than 5 min under chemical characterization of IDE from tomato which is the chosen reaction conditions), while cleavage at His10- the first report on a M16A peptidase from any plant Leu11, Leu15-Tyr16, Phe24-Phe25, Phe25-Tyr26 and Pro28- source. Lys29 occurred at a slower rate (complete processing required more than 30 min; Figure 3A). Six of these cleav- age sites are identical with those generated by human Results IDE (Duckworth et al., 1989; Figure 3A). In contrast, E. coli pitrilysin, also an inverzincin from the M16A family, 16 17 A partial cDNA showing similarity to human IDE was cleaves the insulin B chain exclusively at the Tyr -Leu cloned from a tomato shoot cDNA library in a functional bond (Anastasi and Barrett, 1995; Figure 3A). To rule out screen for proteases cleaving the peptide AVQSKPPSKR the possibility that the observed activity is due not to DPPKMQTD (systemin), and its 59end was completed by SlIDE but rather to a contaminating E. coli , a RACE-PCR. The full-length cDNA of 3358 bp (accession mock purification was performed using a bacterial culture number AJ308542) encompassed an open reading frame expressing the anti-sense SlIDE cDNA, which resulted in of 2913 bp coding for a protein of 971 amino acids with a protein preparation devoid of any proteolytic activity a calculated molecular mass of 112 kDa. The tomato (not shown). (Solanum lycopersicum) sequence, called SlIDE, shared Further characterization of SlIDE activity using syn- the highest degree of amino acid identity with plant thetic peptide substrates revealed a certain preference orthologs, including two Arabidopsis sequences (68 and for proline one or two residues amino-terminal of the w 62% identity with At2g41790 and At3g57470, respect- hydrolyzed bond i.e., the P1 and P2 sites, according to x ively), and is 39% identical to human IDE (Figure 1). The the terminology of Schechter and Berger (1967) . How- predicted domain organization of SlIDE resembles that of ever, this is not an absolute requirement as glucagon, a human IDE. In both , the catalytic M16 domain 29-amino acid peptide lacking proline residues, was (Pfam domain PF00675; http://www.sanger.ac.uk/Soft- nevertheless cleaved at two unrelated sites (Figure 3B). ware/Pfam/) is located close to the amino-terminus. The Consistent with the identification of SlIDE in a screen for M16 domain (amino acids 34–172 in SlIDE) includes the proteases active against systemin, this peptide was conserved HXXEH motif, which is characteristic for the processed efficiently at the Lys14-Met15 bond (Figure 3B). inverzincin family of metalloproteases (Figure 1). It is fol- There are no common sequence elements at the cleav- lowed by two M16-C domains (PF05193; amino acids age sites in different synthetic peptide substrates (Figure 197–378 and 665–853, respectively), which have been 3B), and therefore SlIDE does not seem to require any implicated not in catalysis but rather in substrate binding specific amino acid motif for substrate recognition. Con- and/or the interaction with other proteins (Taylor et al., sistently, the individual replacement of each of the amino 2001). Similar to human IDE, the primary structure of acids of the systemin peptide with alanine did not pre- SlIDE comprises a potential peroxisomal targeting signal vent recognition by SlIDE. However, it did affect the rate at its extreme carboxy-terminus (VRL). of hydrolysis. The stability of alanine-substituted syste- For functional expression in Escherichia coli, the open min analogs in the presence of SlIDE was compared in reading frame of SlIDE was cloned into the expression MALDI-TOF MS experiments (data not shown). The rate vector pGEX-G to obtain the recombinant protein in of hydrolysis was increased when residues in positions 4 N-terminal fusion with glutathione S-transferase (GST). (Ser), 12 (Pro), 13 (Pro) and 17 (Thr) were substituted with The bulk of the expressed protein accumulated in inclu- alanine. All other substitutions rendered the peptide more sion bodies. However, a fraction could be purified from stable with respect to degradation by SlIDE. The data bacterial extracts by affinity chromatography on immo- suggest that SlIDE does not recognize any amino acid bilized glutathione. The progress of purification, yielding sequence per se, but rather higher-order structural fea- approximately 100 mg of GST-SlIDE per liter of culture, tures of its substrates, a feature that is shared with is shown in Figure 2. The apparent molecular mass after human IDE (Shen et al., 2006) and other M16 proteases SDS-PAGE analysis of the purified protein (approximately (Moberg et al., 2003). 150 kDa) is consistent with the mass expected for the A continuous fluorescence assay was used to further fusion protein (26 kDa for GST and 112 kDa for SlIDE). characterize the activity of GST-SlIDE. The internally

Small amounts of lower molecular weight species are quenched substrate Abz-SKRDPPKMQTDLY(NO3)-NH2 Article in press - uncorrected proof

Insulin-degrading enzyme from tomato 93

Figure 1 Comparison of human and tomato IDE sequences. The amino acid sequence deduced from the SlIDE cDNA is compared to human IDE (HsIDE, accession no. BC096339). Amino acid sequences were aligned using ClustalW (http://www.ch.embnet.org). Identical amino acids and conservative replacements are shown in black and gray shading, respectively (‘boxshade’ at http//www.ch.embnet.org/). Amino acids involved in Zn2q-binding are indicated by asterisks. Open and closed circles indicate residues corresponding to the cytosolic fatty acid proteins signature (Prosite access no. PS00214) present in HsIDE, and the cysteine residue previously suggested as responsible for NEM sensitivity of animal IDEs, respectively. Potential peroxisomal targeting signals at the C-termini are underlined. is derived from the systemin sequence and includes determined for the fluorigenic peptide substrate and the 14 15 " the Lys -Met peptide bond processed by GST-SlIDE. catalytic efficiency (kcat/Km) was found to be 62 15 Highest activity was observed at neutral pH, and 50% of mM-1 s-1 (Figure 4C). Consistent with SlIDE being a zinc- the activity were retained at pH 6.5 and 8 (Figure 4B). dependent metalloprotease, its activity was inhibited by Enzymatic activity was stimulated by KCl (or NaCl) and 52% of the control in the presence of 1 mM EDTA, while full activity was observed at rather high concentrations PMSF as a serine protease inhibitor was ineffective (Fig- of )800 mM (Figure 4A). Divalent cations at mM concen- ure 4D). Resembling mammalian IDEs, but unlike E. coli trations (Ca2q,Zn2q,Mg2q and Fe2q) failed to stimulate pitrilysin, the tomato enzyme was efficiently inhibited by GST-SlIDE activity in the absence of KCl. 1mM NEM (51% of control activity; Figure 4D). A more Under optimum steady-state reaction conditions physiological inhibitor and also the most potent in our " (pH 7.0; 800 mM KCl), an apparent Km of 62 7 mM was assay is ATP which, at 1 mM, reduced the activity of GST- Article in press - uncorrected proof

94 Y. Huet et al.

Figure 2 Expression and purification of GST-SlIDE. SlIDE was expressed in E. coli as an N-terminal fusion protein with glutathione S-transferase (GST) and purified from bacterial extracts. A Coomassie Blue-stained SDS-PAGE gel is shown with crude extracts from control and induced cultures in lanes 1 and 2, respectively. The bulk of the GST-SlIDE fusion protein was insoluble (3). However, some of it was present also in the soluble fraction (4) and was purified by affinity chromatography on immobilized glutathione (5).

Figure 3 Processing of the oxidized insulin B chain and other unrelated peptides by GST-SlIDE. Degradation of synthetic peptide substrates was monitored, and cleavage products were identified by MALDI-TOF mass spectrometry. Peptide substrates (2.5 nmol) were incubated with Figure 4 Catalytic properties of GST-SlIDE. 1–2 pmol GST-SlIDE in a 50 ml reaction volume at room tem- Unless otherwise indicated, GST-SlIDE activity was assayed in perature, and aliquots of the reaction were taken at several time 20 mM potassium phosphate buffer at pH 7.0 in the presence intervals for MALDI-TOF analysis. (A) GST-SlIDE rapidly cleaves of 800 mM KCl using Abz-SKRDPPKMQTDLY(NO3)-NH2 (20 mM) insulin B chain at three sites in less than 5 min (line 1, long as the substrate. (A) GST-SlIDE activity is stimulated by KCl. arrows), while five other peptide bonds are cleaved at a slower Proteolytic activity is expressed in arbitrary fluorescence units rate, requiring 30 min to 2 h for complete processing (line 1, and was measured in the presence of increasing concentration " short arrows). Arrows in lines 2 and 3 indicate the cleavage pat- of KCl up to 1.2 M. Data represent the mean SD of three inde- terns previously determined for human IDE (42) and E. coli pitri- pendent experiments. (B) GST-SlIDE activity as a function of pH. lysin (43), respectively. (B) GST-SlIDE cleaves (arrows) several GST-SlIDE activity was assayed in 20 mM MES, pH 5.0 to 6.5, unrelated peptides without preference for any particular amino or 20 mM Tris-HCl, pH 7.0 to 10.0, in the presence of 800 mM acid sequence motif. KCl. The activity is expressed in percent of the maximum activity observed at pH 7.0. Data represent the mean"SD of three experiments. (C) Substrate dependence of GST-SlIDE activity. SlIDE down to 5% of the control. In contrast to rat IDE, GST-SlIDE activity was assayed with increasing concentrations

of Abz-SKRDPPKMQTDLY(NO3)-NH2 and apparent catalytic which is inhibited by long chain fatty acids with an IC50 constants were derived from a double-reciprocal plot of the ranging from 9 mM for linoleic acid (18:2) to 44 mM for data. The values represent the mean"SD of two replicates from palmitic acid (16:0) (Hamel et al., 2003), GST-SlIDE activ- three independent experiments. (D) Residual activity was deter- ity was not affected by linoleic acid between 25 mM and mined in the presence of 1 mM EDTA, NEM, PMSF, ATP, GTP or 2.5 mM (Figure 4D). Consistent with the apparent insen- 25 mM linoleic acid. GST-SlIDE activity is expressed in percent sitivity to fatty acid inhibition, the consensus sequence of the control. Data represent the mean"SD of three of cytosolic fatty acid-binding proteins (Prosite access experiments. Article in press - uncorrected proof

Insulin-degrading enzyme from tomato 95 no. PS00214) previously detected in mammalian IDEs recognized in the protein sequences of mouse and dog (Hamel et al., 2003) is not present in the tomato IDEs, but not in Arabidopsis and Drosophila homologs, sequence. This signature sequence is defined by nine suggesting that the regulation by long chain fatty acids residues, seven of which are conserved in human IDE, may occur in mammals specifically. but not in tomato (Figure 1). In addition to fatty acids, ATP appears to contribute to the metabolic control of IDE activity in mammals. The inhibition of insulin degradation by ATP was first Discussion observed in vivo (Hashimoto et al., 1987), and physiolog- ical concentrations of ATP were later shown to inhibit the degradation of insulin by purified IDE in vitro (Camberos We have reported here the first cloning and biochemical et al., 2001). In contrast, Song and coworkers reported characterization of a plant IDE, a metalloproteinase from an enhanced rate of degradation for small fluorigenic the M16A subgroup in the inverzincin family of pepti- peptide substrates in the presence of ATP, due to allo- dases (http://merops.sanger.ac.uk; Rawlings et al., steric activation of IDE (Song et al., 2004, 2005; Yao and 2004). The tomato (Solanum lycopersicum) cDNA for Hersh, 2006). SlIDE was found to be inhibited by ATP at SlIDE revealed 39% identity at the amino acid level to physiological concentrations, suggesting maximum the well-characterized human IDE, which is involved in SlIDE activity in planta under conditions of low cellular the intracellular degradation of insulin (reviewed by Duck- ATP concentration. Depletion of the cytosolic ATP pool worth et al., 1998). Expression in E. coli and biochemical occurs when oxygen availability limits the rate of respi- characterization of the recombinant tomato protein con- ratory ATP production. IDE activity would thus increase firmed SlIDE as a true insulinase. Indeed, the tomato in response to hypoxia, and a role for SlIDE in stress enzyme cleaves the oxidized insulin B chain with a frag- mentation pattern nearly identical to the one generated adaptation and the regulation of energy homeostasis by animal IDEs (Figure 3A; Duckworth et al., 1988). More- under low-oxygen conditions may be envisaged. In this over, its ability to cleave several unrelated short peptides context, it is of particular interest to note that the expres- in vitro at apparently stochastic sites (Figure 3B) is a sion of one of the two IDE paralogs in Arabidopsis characteristic feature of IDE (Duckworth et al., 1998), (At3g57470) is induced by more than two-fold following pitrilysin (Cornista et al., 2004) and other M16 peptidases anoxia treatment as the only one out of many different of animal (Mzhavia et al., 1999; Falkevall et al., 2006) and stress regimes (Zimmermann et al., 2004; https:// plant (Moberg et al., 2003) origin. www.genevestigator.ethz.ch). SlIDE behaves like a neutral thiol-dependent metallo- SlIDE activity was found to increase with increasing protease, with a pH optimum around 7 and ca. 50% of salt concentrations well into the molar range (Figure 4A). its activity being inhibited by 1 mM EDTA or 1 mM NEM. In previous studies, the activity of mammalian IDE was Sensitivity to thiol reagents (NEM) discriminates proka- shown to depend on the state of oligomerization (Song ryotic from higher eukaryotic IDE activities. While human et al., 2003; Li et al., 2006), and high salt concentrations (Ding et al., 1992), Drosophila (Garcia et al., 1988) and used in our experiments were suspected to alter the plant IDEs (this work) are efficiently inhibited, the E. coli equilibrium between the monomer and the more active enzyme pitrilysin is not. Sensitivity to the thiol-modifying oligomeric forms. However, gel filtration analysis in the compound has previously been attributed to the pres- presence of low (150 mM) or high (2 M) potassium chlo- ence of a cysteine residue adjacent to the zinc-binding ride did not reveal any significant changes in the quater- histidines, which is conserved in human, rat and Dro- nary structure of SlIDE. The GST-SlIDE monomer was sophila IDEs (HXCI EH), but absent from pitrilysin (Ding et undetectable, likely due to the fact that GST forms homo- al., 1992; Perlman et al., 1993). In SlIDE, leucine is found dimers by itself. Three oligomeric forms of approximately in the respective position which argues against a contri- 280, 420 and 600 kDa were observed, which likely rep- bution of this residue to NEM sensitivity. This conclusion resent the dimer, trimer and tetramer at an estimated is in agreement with the work of Perlman et al. (1993) ratio of 1:2:3. The relative abundance of oligomers was showing that substitution of the cysteine in human IDE the same at low and high salt concentrations (data not with either serine or glycine affected neither catalysis nor shown). inhibitor sensitivity. SlIDE cleaves several proline-rich oligopeptides, Physiological regulators of IDE include nucleotide tri- including systemin, substance P and bradykinin in vitro, phosphates (Camberos et al., 2001; Song et al., 2004) and a proline residue was frequently observed proximal and long chain fatty acids (Hamel et al., 2003) allowing to the hydrolyzed bond. Proline plays an important for the metabolic control of enzyme activity. Linoleic acid physiological role in protecting biologically active pep- (18:2) was shown to be the most potent fatty acid inhib- tides against enzymatic degradation and, indeed, many itor with an IC50 of 9 mM for rat IDE (Hamel et al., 2003). proteases are unable to cleave peptide bonds in the Consistent with fatty acid inhibition, mammalian IDEs vicinity of this imino acid. Hence, a stabilizing proline res- contain an almost perfect match to the consensus of the idue is frequently found and conserved in peptide hor- ‘cytosolic fatty-acid binding proteins signature’ sequence mones, neuropeptides and growth factors in animals (Figure 1; Prosite access no. PS00214). In contrast, SlIDE (Mentlein, 1988; Yaron and Naider, 1993; Vanhoof et al., was neither sensitive to linoleic acid inhibition (Figure 4D) 1995) and also in plant peptides with signaling function up to a concentration of 2.5 mM, nor does it contain (Ryan and Pearce, 2003; Ito et al., 2006). The stability of the aforementioned sequence motif. Interestingly, the such peptides is thus controlled by peptidases evolved ‘cytosolic fatty acid binding proteins signature’ can be to specifically hydrolyze proline-containing substrates Article in press - uncorrected proof

96 Y. Huet et al.

(Cunningham and O’Connor, 1997), and such a role may cloned into the StuI/SmaI sites of pGEX-G (Go¨ rlach and Schmid, be taken by SlIDE in planta. 1996), a derivative of pGEX-3x (GE Healthcare, Uppsala, Despite the low degree of sequence conservation, IDE Sweden), to yield pGEX-IDE. This construct allows expression homologs in plants and in mammals appear to share of SlIDE in N-terminal fusion with GST in E. coli. The identity of all PCR-generated clones was confirmed by sequence analysis many of their biochemical features. However, the simi- of at least three independent PCR products using fluorescent larity does not seem to extend to physiological function: dideoxy chain terminators in the cycle sequencing reaction insulin and b-amyloid peptide, two well-established sub- (Perkin Elmer, Foster City, CA, USA) and the Applied Biosystems strates of human IDE, do not exist in plants. SlIDE was model 373A DNA sequencer (Foster City, CA, USA). identified in a functional screen for proteases able to cleave the plant wound hormone systemin. While cleav- Expression and purification of SlIDE age of systemin was confirmed for the recombinant enzyme in vitro, this peptide is not likely to be a physio- For expression of the GST-SlIDE fusion protein in E. coli DH5a, logically relevant substrate of SlIDE in vivo: SlIDE does two 500-ml cultures were grown at 378CtoanOD600 of 0.8. not co-localize with systemin in the leaf apoplast, plants Isopropyl-1-thio-b-D-galactopyranoside was added to a final silenced for SlIDE expression were not affected in sys- concentration of 0.1 mM, and cells were grown for another temin-mediated wound signaling, nor did they show any 4 days at 48C when they were harvested by centrifugation obvious growth defects under greenhouse conditions (6000 g for 15 min at 48C). The cells were resuspended in 100 ml of buffer A (50 mM Tris-HCl, pH 8.0, 10 mM NaCl, 1 mM EDTA) (unpublished results). Therefore, the role of insulinases in containing 0.1 mg/ml DNAseI and 1 mg/ml lysozyme. After plants and their physiological substrates remain to be 20 min at room temperature, cells were lysed by sonication. The characterized. cell debris was removed by centrifugation (10,000 g for 30 min at 48C), and the supernatant was subjected to affinity chroma- tography on glutathione-sepharose 4B (GE Healthcare), accord- Materials and methods ing to the manufacturer’s instructions. The affinity matrix was washed with buffer A and buffer B (50 mM Tris-HCl, pH 8.0), and the fusion protein was eluted with 10 mM glutathione in buffer Cloning of the SlIDE gene B. The progress of purification was monitored by SDS-PAGE using the Laemmli buffer system (Laemmli, 1970), and the puri- A partial SlIDE cDNA was isolated in a genetic screen for pro- fied protein was stored at -808C. teases able to cleave the plant wound-signaling peptide syste- min (AVQSKPPSKRDPPKMQTD). The screening procedure was based on the loss of reporter gene expression in the yeast, Sac- MALDI-TOF MS assay of SlIDE specificity charomyces cerevisiae, as a result of the proteolytic inactiv- ation of the modified GAL4 transcription factor and was adapted The degradation of synthetic peptide substrates (obtained from from a method developed by B. Kohorn (Smith and Kohorn, Sigma or Enzyme Systems Products, Livermore, CA, USA) by 1991; Kohorn et al., 1992; Hauser et al., 2001). Briefly, a plasmid SlIDE was assayed using MALDI-TOF mass spectrometry (MS) carrying the yeast GAL4 gene was engineered for the expression to detect and identify the generated peptide fragments. In a total in yeast of a modified transcription factor, in which the systemin volume of 50 ml of reaction buffer (50 mM Tris-HCl, pH 7.5), the sequence was inserted between its two functional domains. Pro- assay contained 1–2 pmol of the protease and 2.5 nmol of the teolytic cleavage of systemin would then result in the separation respective peptide substrate. The reaction mixture was incubat- of the two domains and in the inactivation of GAL4. The plasmid ed at room temperature, 0.8 ml aliquots were taken at intervals was introduced into a S. cerevisiae reporter strain carrying the and mixed on the MALDI-TOF MS sample plate with an equal lacZ gene for b-galactosidase under control of the GAL1/10 volume of the crystallization matrix w2 parts of saturated 2,5- promoter. A tomato cDNA library was constructed in pYES2 dihydroxybenzoic acid in acetone, 1 part 0.1% (v/v) trifluoroa- (Invitrogen, Groningen, Netherlands) with poly(A)q RNA isolated cetic acidx. Crystals were washed repeatedly with cold from tomato (Solanum lycopersicum, cv. Castlemart II) shoot tis- de-ionized water and air-dried before recording of the mass sue (complexity of 3.5=106 independent cDNAs with an average spectra with a Voyager Elite mass spectrometer (Applied size of 1.4 kb) and used to transform the reporter strain. The Biosystems). screening for cDNA clones conferring loss of GAL4 function was performed as described previously (Smith and Kohorn, 1991; Steady-state kinetic analyses Kohorn et al., 1992). To obtain the full-length SlIDE cDNA, RACE (rapid amplifica- For the continuous assay of SlIDE activity, Abz- tion of cDNA ends)-PCR was performed using the SMART RACE SKRDPPKMQTDLY(NO3)-NH2 (Jerini AG, Berlin, Germany) was cDNA amplification kit (Clontech, Palo Alto, CA, USA), according used as an internally quenched fluorogenic substrate. In a total to the manufacturer’s instructions. In a first step, single stranded volume of 50 ml, the standard assay contained 2 pmol of the cDNA was synthesized from total RNA of tomato leaves using GST-IDE fusion protein, 20 mM of the substrate and 800 mM KCl M-MLV reverse transcriptase (Promega, Madison, WI, USA) and in 20 mM potassium phosphate buffer at pH 7.0. The fluores- oligo(dT) as the primer. Subsequently, the full-length SlIDE cDNA cence was recorded continuously in a Cary Eclipse fluorimeter was amplified using a gene-specific primer (59-ACC TGT TGA (Varian, Darmstadt, Germany; lex: 320 nm, lem: 420 nm) at 258C. TAT GGC TAT TGG AAC TTG A-39) and the universal primer To convert relative fluorescence units into molar quantities of provided with the kit. Three independent PCR products were reaction product, defined quantities of the substrate peptide gel-purified, cloned into the pCR-XL-TOPO vector (Invitrogen) were digested to completion using an excess of enzyme and the and sequenced (accession no. of the full-length SlIDE cDNA is resulting fluorescence recorded. The inhibitor profile of SlIDE AJ308542). The open reading frame of SlIDE was then amplified was determined by preincubating the enzyme for 5 min with by PCR using Pfu DNA polymerase (Stratagene, La Jolla, CA, 1mM EDTA (ethylenediaminetetraacetic acid), NEM (N-ethyl- USA) and synthetic oligonucleotide primers (forward primer: maleimide), PMSF (phenylmethylsulfonyl fluoride), ATP, GTP, or 59-ATG GCA GTT GGT AAA AAG GA-39; and reverse primer: 59- 25 mM linoleic acid. Control reactions were performed in parallel GGG GGG CTT CTC GAG ATG AT-39). The PCR product was and contained an equivalent volume of the solvents of the Article in press - uncorrected proof

Insulin-degrading enzyme from tomato 97

respective inhibitors, i.e., water for EDTA and the nucleo- levels of insulin, amyloid b-protein, and the b-amyloid pre- tides, and ethanol for NEM, PMSF and linoleic acid. cursor protein intracellular domain in vivo. Proc. Natl. Acad. Sci. USA 100, 4162–4167. Gakh, E., Cavadini, P., and Isaya, G. (2002). Mitochondrial pro- Acknowledgments cessing peptidases. Biochim. Biophys. Acta Mol. Cell Res. 1592, 63–77. We greatly appreciate valuable advice and the gift of plasmids Garcia, J.V., Fenton, B.W., and Rosner, M.R. (1988). Isolation and yeast strains from Dr. Bruce Kohorn (Duke University, NC, and characterization of an insulin-degrading enzyme from USA) allowing us to perform the functional screen for peptidases Drosophila melanogaster. Biochemistry 27, 4237–4244. in yeast. Garcia, J.V., Gehm, B.D., and Rosner, M.R. (1989). An evolution- arily conserved enzyme degrades transforming growth factor-a as well as insulin. J. Cell Biol. 109, 1301–1307. Gehm, B.D. and Rosner, M.R. (1991). Regulation of insulin, References epidermal growth factor, and transforming growth factor-a levels by growth factor-degrading enzymes. Endocrinology Anastasi, A. and Barrett, A.J. (1995). Pitrilysin. Methods Enzy- 128, 1603–1610. mol. 248, 684–693. Glaser, E., Sjoling, S., Tanudji, M., and Whelan, J. (1998). Mito- Becker, A.B. and Roth, R.A. (1995). Insulysin and pitrilysin: chondrial protein import in plants – signals, sorting, targeting, insulin-degrading enzymes of mammals and bacteria. Meth- processing and regulation. Plant Mol. Biol. 38, 311–338. ods Enzymol. 248, 693–704. Go¨ rlach, J. and Schmid, J. (1996). Introducing StuI sites Bennett, R.G., Duckworth, W.C., and Hamel, F.G. (2000). Deg- improves vectors for the expression of fusion proteins with radation of amylin by insulin-degrading enzyme. J. Biol. factor Xa cleavage sites. Gene 170, 145–146. Chem. 275, 36621–36625. Hamel, F.G., Upward, J.L., and Bennett, R.G. (2003). In vitro inhi- Bhushan, S., Lefebvre, B., Stahl, A., Wright, S.J., Bruce, B.D., bition of insulin-degrading enzyme by long-chain fatty acids Boutry, M., and Glaser, E. (2003). Dual targeting and function and their coenzyme A thioesters. Endocrinology 144, 2404– of a protease in mitochondria and chloroplasts. EMBO Rep. 2408. 4, 1073–1078. Hashimoto, N., Robinson, F.W., Shibata, Y., Flanagan, J.E., and Branda, S.S. and Isaya, G. (1995). Prediction and identification Kono, T. (1987). Diversity in the effects of extracellular ATP of new natural substrates of the yeast mitochondrial inter- and adenosine on the cellular processing and physiologic mediate peptidase. J. Biol. Chem. 270, 27366–27373. actions of insulin in rat adipocytes. J. Biol. Chem. 262, Camberos, M.C., Perez, A.A., Udrisar, D.P., Wanderley, M.I., and 15026–15032. Cresto, J.C. (2001). ATP inhibits insulin-degrading enzyme Hauser, F., Strassner, J., and Schaller, A. (2001). Cloning, expres- activity. Exp. Biol. Med. 226, 334–341. sion, and characterization of tomato (Lycopersicon esculen- Chesneau, V. and Rosner, M.R. (2000). Functional human insulin- tum) aminopeptidase P. J. Biol. Chem. 276, 31732–31737. degrading enzyme can be expressed in bacteria. Protein Ito, Y., Nakanomyo, I., Motose, H., Iwamoto, K., Sawa, S., Expr. Pur. 19, 91–98. Dohmae, N., and Fukuda, H. (2006). Dodeca-CLE peptides Cornista, J., Ikeuchi, S., Haruki, M., Kohara, A., Takano, K., Mori- as suppressors of plant stem cell differentiation. Science kawa, M., and Kanaya, S. (2004). Cleavage of various pep- 313, 842–845. tides with pitrilysin from Escherichia coli: kinetic analyses Kirschner, R.J. and Goldberg, A.L. (1983). A high molecular using b-endorphin and its derivatives. Biosci. Biotech. Bio- weight metalloendoprotease from the cytosol of mammalian chem. 68, 2128–2137. cells. J. Biol. Chem. 258, 967–976. Cunningham, D.F. and O’Connor, B. (1997). Proline-specific pep- Kohorn, B.D., Lane, S., and Smith, T.A. (1992). An Arabidopsis tidases. Biochim. Biophys. Acta 1343, 160–186. serine/threonine kinase homologue with an epidermal growth Ding, L., Becker, A., Suzuki, A., and Roth, R. (1992). Comparison factor repeat selected in yeast for its specificity for a thyla- of the enzymatic and biochemical properties of human insu- koid membrane protein. Proc. Natl. Acad. Sci. USA 89, lin-degrading enzyme and Escherichia coli protease III. J. 10989–10992. Biol. Chem. 267, 2414–2420. Kurochkin, I.V. (2001). Insulin-degrading enzyme: embarking on Duckworth, W.C. (1988). Insulin degradation: mechanisms, pro- amyloid destruction. Trends Biochem. Sci. 26, 421–425. ducts, and significance. Endocrine Rev. 9, 319–345. Laemmli, U.K. (1970). Cleavage of structural proteins during Duckworth, W.C. (1990). Insulin. In: Handbook of Experimental assembly of the head of bacteriophage T4. Nature 227, Pharmacology: Insulin, P. Cuatrecasas and S. Jacobs, eds. 680–685. (Berlin, Germany: Springer), pp. 143–165. Li, P., Kuo, W.L., Yousef, M., Rosner, M.R., and Tang, W.J. (2006). Duckworth, W., Hamel, F., Peavy, D., Liepnieks, J., Ryan, M., The C-terminal domain of human insulin degrading enzyme Hermodson, M., and Frank, B. (1988). Degradation products is required for dimerization and substrate recognition. Bio- of insulin generated by hepatocytes and by insulin protease. chem. Biophys. Res. Comm. 343, 1032–1037. J. Biol. Chem. 263, 1826–1833. Mentlein, R. (1988). Proline residues in the maturation and deg- Duckworth, W.C., Garcia, J.V., Liepnieks, J.J., Hamel, F.G., radation of peptide hormones and neuropeptides. FEBS Lett. Hermodson, M.A., Frank, B.H., and Rosner, M.R. (1989). 234, 251–256. Drosophila insulin-degrading enzyme and rat skeletal muscle Miller, B.C., Eckman, E.A., Sambamurti, K., Dobbs, N., Chow, insulin protease cleave insulin at similar sites. Biochemistry K.M., Eckman, C.B., Hersh, L.B., and Thiele, D.L. (2003). 28, 2471–2477. Amyloid-b peptide levels in brain are inversely correlated Duckworth, W.C., Bennett, R.G., and Hamel, F.G. (1998). Insulin with insulysin activity levels in vivo. Proc. Natl. Acad. Sci. degradation: progress and potential. Endocrine Rev. 19, USA 100, 6221–6226. 608–624. Moberg, P., Stahl, A., Bhushan, S., Wright, S.J., Eriksson, A., Falkevall, A., Alikhani, N., Bhushan, S., Pavlov, P.F., Busch, K., Bruce, B.D., and Glaser, E. (2003). Characterization of a novel Johnson, K.A., Eneqvist, T., Tjernberg, L., Ankarcrona, M., zinc metalloprotease involved in degrading targeting pep- and Glaser, E. (2006). Degradation of the amyloid b-protein tides in mitochondria and chloroplasts. Plant J. 36, 616–628. by the novel mitochondrial peptidasome, preP.J. Biol. Chem. Muller, D., Schulze, C., Baumeister, H., Buck, F., and Richter, D. 281, 29096–29104. (1992). Rat insulin-degrading enzyme: cleavage pattern of Farris, W., Mansourian, S., Chang, Y., Lindsley, L., Eckman, E.A., the natriuretic peptide hormones ANP, BNP, and CNP Frosch, M.P., Eckman, C.B., Tanzi, R.E., Selkoe, D.J., and revealed by HPLC and mass spectrometry. Biochemistry 31, Guenette, S. (2003). Insulin-degrading enzyme regulates the 11138–11143. Article in press - uncorrected proof

98 Y. Huet et al.

Mzhavia, N., Berman, Y.L., Qian, Y., Yan, L., and Devi, L.A. enzyme are mediated primarily through its triphosphate (1999). Cloning, expression, and characterization of human moiety. J. Biol. Chem. 279, 54216–54220. metalloprotease 1: a novel member of the pitrilysin family of Song, E.S., Daily, A., Fried, M.G., Juliano, M.A., Juliano, L., and metalloendoproteases. DNA Cell Biol. 18, 369–380. Hersh, L.B. (2005). Mutation of active site residues of insulin- Perlman, R., Gehm, B., Kuo, W., and Rosner, M. (1993). Func- degrading enzyme alters allosteric interactions. J. Biol. tional analysis of conserved residues in the active site of Chem. 280, 17701–17706. insulin-degrading enzyme. J. Biol. Chem. 268, 21538–21544. Sta˚ hl, A., Moberg, P., Ytterberg, J., Panfilov, O., Brockenhuus Qiu, W.Q., Walsh, D.M., Ye, Z., Vekrellis, K., Zhang, J.M., Pod- von Lo¨ wenhielm, H., Nilsson, F., and Glaser, E. (2002). Iso- lisny, M.B., Rosner, M.R., Safavi, A., Hersh, L.B., and Selkoe, lation and identification of a novel mitochondrial metallopro- D.J. (1998). Insulin-degrading enzyme regulates extracellular tease (preP) that degrades targeting presequences in plants. levels of amyloid b-protein by degradation. J. Biol. Chem. J. Biol. Chem 277, 41931–41939. 273, 32730–32738. Taylor, A.B., Smith, B.S., Kitada, S., Kojima, K., Miyaura, H., Rawlings, N.D., Tolle, D.P., and Barrett, A.J. (2004). MEROPS: Otwinowski, Z., Ito, A., and Deisenhofer, J. (2001). Crystal the peptidase database. Nucleic Acids Res. 32, D160–D164. structures of mitochondrial processing peptidase reveal the Richter, S. and Lamppa, G.K. (1998). A chloroplast processing mode for specific cleavage of import signal sequences. enzyme functions as the general stromal processing pepti- Structure 9, 615–625. dase. Proc. Natl. Acad. Sci. USA 95, 7463–7468. Vander Vere, P.S., Bennett, T.M., Oblong, J.E., and Lamppa, G.K. Ryan, C.A. and Pearce, G. (2003). Systemins: a functionally (1995). A chloroplast processing enzyme involved in precur- defined family of peptide signal that regulate defensive sor maturation shares a zinc-binding motif with a recently genes in Solanaceae species. Proc. Natl. Acad. Sci. USA recognized family of . Proc. Natl. 100, 14577–14580. Acad. Sci. USA 92, 7177–7181. Safavi, A., Miller, B.C., Cottam, L., and Hersh, L.B. (1996). Vanhoof, G., Goossens, F., Demeester, I., Hendriks, D., and Identification of g-endorphin-generating enzyme as insulin- Scharpe, S. (1995). Proline motifs in peptides and their bio- degrading enzyme. Biochemistry 35, 14318–14325. logical processing. FASEB J. 9, 736–744. Schechter, I. and Berger, A. (1967). On the size of the active site Vekrellis, K., Ye, Z., Qiu, W.Q., Walsh, D., Hartley, D., Chesneau, in proteases. I. Papain. Biochem. Biophys. Res. Commun. V., Rosner, M.R., and Selkoe, D.J. (2000). Neurons regulate 27, 157–162. extracellular levels of amyloid b-protein via proteolysis by Shen, Y.Q., Joachimiak, A., Rosner, M.R., and Tang, W.J. (2006). insulin-degrading enzyme. J. Neurosci. 20, 1657–1665. Structures of human insulin-degrading enzyme reveal a new Yao, H.B. and Hersh, L.B. (2006). Characterization of the binding substrate recognition mechanism. Nature 443, 870–874. of the fluorescent ATP analog TNP-ATP to insulysin. Arch. Smith, T.A. and Kohorn, B.D. (1991). Direct selection for Biochem. Biophys. 451, 175–181. sequences encoding proteases of known specificity. Proc. Yaron, A. and Naider, F. (1993). Proline-dependent structural and Natl. Acad. Sci. USA 88, 5159–5162. biological properties of peptides and proteins. Crit. Rev. Bio- Song, E.S., Juliano, M.A., Juliano, L., and Hersh, L.B. (2003). chem. Mol. Biol. 28, 31–81. Substrate activation of insulin-degrading enzyme (insulysin) Zimmermann, P., Hirsch-Hoffmann, M., Hennig, L., and Gruis- – a potential target for drug development. J. Biol. Chem. 278, sem, W. (2004). Genevestigator. Arabidopsis microarray data- 49789–49794. base and analysis toolbox. Plant Physiol. 136, 2621–2632. Song, E.S., Juliano, M.A., Juliano, L., Fried, M.G., Wagner, S.L., and Hersh, L.B. (2004). ATP effects on insulin-degrading Received June 17, 2007; accepted September 20, 2007