The human CLN2 protein/tripeptidyl-peptidase I is a serine protease that autoactivates at acidic pH.

The CLN2 gene mutated in the fatal hereditary neurodegenerative disease late infantile neuronal ceroid lipofuscinosis encodes a lysosomal protease with tripeptidyl-peptidase I activity. To understand the enzymological properties of the protein, we purified and characterized C-terminal hexahistidine-tagged human CLN2p/tripeptidyl-peptidase I produced from insect cells transfected with a baculovirus vector. The N terminus of the secreted 66-kDa protein corresponds to residue 20 of the primary CLN2 gene translation product, indicating removal of a 19-residue signal peptide. The purified protein is enzymatically inactive; however, upon acidification, it is proteolytically processed and concomitantly acquires enzymatic activity. The N terminus of the final 46-kDa processed form (Leu196) corresponds to that of mature CLN2p/tripeptidyl-peptidase I purified from human brain. The activity of the mature enzyme is irreversibly inhibited by the serine esterase inhibitor diisopropyl fluorophosphate, which specifically and stoichiometrically reacts with CLN2p/tripeptidyl-peptidase I at Ser475, demonstrating that this residue represents the active site nucleophile. Expression of wild type and mutant proteins in CHO cells indicate that Ser475, Asp360, Asp517, but not His236 are essential for activity. These data indicate that the CLN2 gene product is synthesized as an inactive proenzyme that is autocatalytically converted to an active serine protease.

Late infantile neuronal ceroid lipofuscinosis (LINCL, OMIM 204500) 1 is a recessive fatal neurological disease characterized by lysosomal accumulation of proteinaceous autofluorescent storage material in neurons and other cell types (reviewed in Ref. 1). The CLN2 gene product was first identified as a normally abundant 46-kDa mannose 6-phosphorylated glycoprotein that was absent in brain autopsy specimens from LINCL patients, leading to the molecular characterization of the disease gene (2). The encoded protein has significant sequence similarities to two previously characterized bacterial endopro-teases from Xanthomonas and Pseudomonas (3,4). These prokaryotic enzymes have been named bacterial pepstatin-insensitive carboxyl peptidases (BPICPs) based on a series of studies that suggest that, like classic aspartyl proteases, their catalytic mechanism involves a pair of amino acids with carboxyl side chains that catalyze peptide bond hydrolysis at acidic pH but, unlike the classic aspartyl proteases, they are not inhibited by pepstatin. Based on these sequence relationships, we demonstrated that the CLN2 protein was a lysosomal protease that could degrade hemoglobin at acidic pH in the presence of aspartyl and cysteine protease inhibitors (2,5). More recently, the CLN2 protein has been shown to have tripeptidyl exopeptidase activity. Peptide sequencing of tripeptidyl-peptidase I (TPP-I) purified from rat spleen demonstrated that it is the rodent ortholog of human CLN2p (6), and TPPI-I activity is absent in LINCL specimens (7). Recently, Ezaki and colleagues (8) have shown that purified CLN2p has both TPP-I and endopeptidase activity.
Mechanistic studies on purified CLN2p/TPP-I by a number of groups indicate that enzymatic activity is not inhibited by standard serine, cysteine, metallo, or aspartyl protease inhibitors (6,8,9). Based on this, it has been proposed that the CLN2p/TPP-I is a pepstatin-insensitive carboxyl peptidase. However, Rawlings and Barrett (10) critically reviewed existing evidence for classification of the BPICPs and concluded that the catalytic mechanism for these enzymes has not been rigorously established. These investigators alertly pointed out that a conserved Gly-Xaa-Ser sequence in CLN2p/TPP-I and the BPICPs is characteristic of the active site motif of many serine peptidases. Interestingly, data base searches using the CLN2p/TPP-I sequence with different BLAST tools (CD-Search and PSI-BLAST) retrieve members of the S8 subtilisin family of serine peptidases (data not shown).
A thorough understanding of the biochemical properties of the CLN2 gene product may provide valuable clues toward understanding and developing therapies for LINCL. In this study, we have characterized the enzymatic properties of CLN2p/TPP-I. We find that the protein is synthesized as an inactive zymogen that is autocatalytically converted to an active serine protease at acidic pH.

EXPERIMENTAL PROCEDURES
Hexahistidine-tagged CLN2p/TPP-I-Two different human CLN2 cDNAs encoding the His 175 and Arg 175 variants (GenBank TM accession number AF017456) were subcloned into pBACgus-1 (Novagen). The constructs encode full-length CLN2p/TPP-I (residues 1-563) with a C-terminal hexahistidine tag (residues 564 -569). Both constructs were expressed at Kemp Biotechnologies (Fredrick, MD) using the Sf9 cell baculovirus system. Although the Arg 175 variant has only been observed in one of several cDNAs sequenced to date, it was used for large scale production purposes because it gave significantly higher yield. Aliquots of conditioned medium (1 liter) were stored at Ϫ80°C until use. All purification steps were conducted at 4°C, and TPP-I activity was monitored after preincubating samples at pH 3.5 to activate the proenzyme (see below). Thawed medium was adjusted to contain 2 M NaCl/20 mM Tris, pH 8.0, centrifuged at 16,000 ϫ g for 1 h, and the supernatant was applied to a 53-cm 3 butyl-Sepharose 4 Fast Flow column (Amersham Pharmacia Biotech). The column was eluted with 20 mM Tris, pH 8.0, and the CLN2p/TPP-I precursor peak was applied to a 10-cm 3 TALON immobilized cobalt column (CLONTECH Laboratories). The column was washed with 50 mM sodium chloride, 20 mM Tris, pH 8.0, and eluted with the same buffer containing 100 mM imidazole. The hexahistidine-tagged CLN2p/TPP-I precursor was further purified by anion exchange chromatography on either a Mono Q HR 5/5 (Amersham Pharmacia Biotech) or an UNO Q-12 (Bio-Rad) column using a 50 -500 mM sodium chloride gradient in 20 mM Tris, pH 8.0. The purified protein eluted at ϳ170 mM sodium chloride and appeared homogeneous by SDS-PAGE and Coomassie Blue staining. Protein (1-2 mg/ml) was divided into aliquots and stored at Ϫ80°C until use. The total yield of purified protein was typically 6 mg/liter of culture medium.
Untagged CLN2p/TPP-I-Mature CLN2p/TPP-I was purified from frozen human brain autopsy specimens obtained from National Disease Research Interchange (Philadelphia, PA). Brain mannose 6-phosphorylated glycoproteins were isolated by chromatography on immobilized cation-independent mannose 6-phosphate receptor as described previously (11) except that the column was sequentially eluted with 5 mM mannose 6-phosphate and then with 100 mM glycine, pH 2.5. The glycine eluate contained the bulk of the mannose 6-phosphorylated CLN2p/TPP-I and was applied to a Mono S HR 5/5 column (Amersham Pharmacia Biotech). The column was eluted using a 0 -1 M sodium chloride gradient in 0.1% Tween 20, 20 mM sodium citrate, pH 5. Fractions containing TPP-I activity eluted at ϳ0.5 M sodium chloride and were further purified by gel filtration on a 1.0 ϫ 30-cm Superose 12 (Amersham Pharmacia Biotech) column eluted using 0.1% Tween 20, 50 mM sodium chloride, 20 mM sodium formate, pH 4.0. The purified 46-kDa protein appeared homogeneous by SDS-PAGE and Coomassie Blue staining. Total yield of purified protein was typically 20 g/100 g of human brain. Production of untagged human CLN2p/TPP-I precursor in CHO cells will be described elsewhere. 2 Enzyme and Inhibitor Assays-Standard conditions were as follows unless otherwise noted. All incubation steps contained 150 mM sodium chloride and 0.1% Triton X-100 (saline/Triton) buffered to the indicated pH as stated in the figure legends. The proenzyme was activated by dilution into 50 mM formate, pH 3.5, buffered saline/Triton and incubation at 22-37°C for at least 10 min. For incubations with inhibitors, the concentration of the enzyme was typically ϳ50 nM, and the final concentration of solvent used to prepare the inhibitor stock was Ͻ1%. TPP-I activity was assayed using the kinetic assay described previously (12) except that the temperature of the fluorescent plate reader was adjusted so that the input samples were at thermal equilibrium to eliminate the initial lag period. The final concentration of enzyme was typically ϳ5 nM, and the Ala-Ala-Phe-aminomethylcoumarin (AAF-AMC) substrate concentration was 200 M.
Protein Chemistry-For peptide mapping, the activation and DFP reaction conditions were as described above except that volatile buffers (150 mM ammonium acetate, pH 4.0) were used to facilitate downstream processing. Samples (typically 100 g) were evaporated to dryness in a vacuum centrifuge, resuspended in 100 l of 6 M guanidine hydrochloride, 2 mM EDTA, 375 mM Tris, pH 8.1, and incubated for 1 h at 50°C. Samples were reduced by adding 4 l of 1 M dithiothreitol (incubation, 1.5 h at 50°C) and alkylated by adding 16 l of 0.5 M iodoacetamide (incubation in the dark for 30 min at 25°C). The mixtures were desalted on a Superdex Peptide column (Amersham Pharmacia Biotech) equilibrated with 20% acetonitrile, 0.1% trifluoroacetic acid. The peak fractions were pooled, evaporated to dryness in a vac-2 L. Lin and P. Lobel, manuscript in preparation.
FIG. 1. The CLN2p/TPP-I precursor is autocatalytically processed. CLN2p/TPP-I was incubated at the indicated pH for 10 min at 30°C. Reactions were initiated by diluting hexahistidine-tagged precursor protein (1.65 mg/ml protein in sodium chloride, 20 mM Tris, pH 8, ion exchange buffer) into 7 volumes of saline/Triton buffer (pH 2.5, 100 mM sodium citrate; pH 3.0 -5.5, 100 mM sodium acetate) and terminated by addition of reducing SDS-PAGE sample buffer. Samples (1 g) were fractionated by SDS-PAGE using a 12% Tris-glycine gel (Novex) and visualized by sequential staining with Coomassie blue and silver (Novex). The lane labeled pH 8.0 represents starting material that was diluted directly into SDS-PAGE sample buffer.
a Activated enzyme was preincubated with 2.5 times the indicated concentration of inhibitor for 1 h at 22°C, diluted 2.5-fold into substrate solution to give the indicated concentration of inhibitor, and then assayed for TPP-I activity. Preincubation and assay buffers were saline/ Triton with 100 mM sodium citrate, pH 4.25.
uum centrifuge, resuspended in 100 l of 150 mM Tris, pH 8.0, and incubated at 37°C for 4 h after adding 2.5 g of trypsin (Worthington, TRSEQZ grade). The reactions were stopped by addition of trifluoroacetic acid to a final concentration of ϳ0.1%. Digests were analyzed on a PE Biosystems Voyager Pro mass spectrometer using sinapinic acid matrix before and after fractionation by HPLC on a C18 Vydac protein/ peptide column eluted using an acetonitrile gradient in 0.1% trifluoroacetic acid. Peptides of interest were further analyzed by Edman degradation using an Applied Biosystems 477A sequencer. CLN2p/TPP-I Mutants-Polymerase chain reaction-based mutagenesis was used to introduce the S475A (TCG to GCG), D360A (GAC to GCC), D517A (GAT to GCT), and H236A (CAT to GCT) mutations into the human CLN2p/TPP-I cDNA sequence. Briefly, for each construct, two complementary oligonucleotide primers that contained the appropriate nucleotide changes were used separately with outside primers flanking either the 5Ј or 3Ј coding sequence to generate two polymerase chain reaction products. These were then gel purified, mixed, and used with the outside primers in a second round of polymerase chain reaction. The resulting product containing the full-length CLN2 coding sequence was subcloned into the expression vector pSFFV-neo (13), and the entire insert was sequenced to confirm the presence of the desired mutations and absence of unwanted changes. DNA was introduced into CHO cells using LipofectAMINE (Life Technologies, Inc.), and resistant clones were selected using G418. For activity assays and Western blotting, cells were seeded into 6-well plates and cultured in Dulbecco's modified Eagle's medium/Ham's F-12 medium, 10% fetal bovine serum until monolayers reached confluence. Cells were washed and cultured in 2 ml of serum-free medium for 24 h, washed with phosphate-buffered saline, and lysed with a solution (0.5 ml) of 1% Nonidet P-40, 150 mM sodium chloride, 10 mM Tris, pH 7.5. Samples were clarified by centrifugation at 12,000 ϫ g for 5 min (medium) or 20 min (lysates) and analyzed for TPP-I activity using the end point assay (12) after preactivation at pH 3.5 for 30 min at 37°C. Clarified lysates contained between 1.2 and 1.5 mg/ml protein as determined by the Lowry assay (14). Samples (lysates, 8 g/lane; medium, 15.6 l/lane) were fractionated by SDS-PAGE on a 10% NuPAGE Bis-Tris gel (Novex). CLN2p/ TPP-I was visualized by Western blotting using a rabbit antiserum prepared against hexahistidine-tagged CLN2p/TPP-I (antiserum R72-5, contract production by Cocalico Biologicals) and enhanced chemiluminescence (Renaissance, PerkinElmer Life Sciences).

RESULTS
Autocatalytic Processing of CLN2p/TPP-I-We purified a C-terminal hexahistidine-tagged version of the CLN2p/TPP-I secreted by an insect cell production system as described under "Experimental Procedures." When maintained at nearly neutral pH, the protein had an apparent size of 66 kDa by SDS-PAGE (Fig. 1). Edman degradation revealed that the N terminus of the 66-kDa species (SYSPE . . . ) corresponds to residue 20 of the predicted CLN2p precursor. Differences in the predicted and actual molecular mass of the His-tagged CLN2p/ TPP-I (residue 20 -569; 60,118 Da) are probably due to N-linked glycosylation (2,9). In acidic conditions (pH Ͻ4.5), the protein is converted to lower molecular mass forms (Fig. 1). Note that in addition to the major species of ϳ46 kDa, faster migrating bands also appear (ϳ20 kDa), suggesting that the processing occurs through an endoproteolytic mechanism.
More detailed analysis of the conversion at pH 4.0 indicates that the 66-kDa species is rapidly (t1 ⁄2 ϭ 7 min) converted to a 46-kDa species that is stable upon prolonged (24 h) incubation (Fig. 2). Note the transient appearance of lower molecular mass species near the bottom of the gel that appear to decrease in size and eventually disappear at long time points (Fig. 2, lane  14), consistent with endoproteolytic liberation and eventual degradation of the propeptide. Also, at early time points there appears to be a ϳ49-kDa species that transiently appears (Fig.  2, lanes 2 and 3), suggesting that the initial endoproteolytic cleavage may be upstream of the terminally processed 46-kDa mature form. Edman degradation revealed that the N terminus of the 46-kDa species (LHLGV) corresponds to residue 196 of the predicted CLN2p precursor, identical to that of human brain CLN2p (2). Enzyme activity measurements indicate that the 66-kDa form has very low TPP-I activity and that the time course for acquisition of activity (t1 ⁄2 ϭ 6 min) is nearly identical to that of proteolytic processing (Fig. 2). Similar findings were obtained using untagged proenzyme purified from CHO cells (data not shown). Taken together, these data indicate that the CLN2 protein is synthesized as an inactive proenzyme that upon acidification undergoes autocatalytic conversion to an enzymatically active species.
Inhibition of CLN2p/TPP-I Activity-To investigate the catalytic properties of the CLN2 protein, we incubated the processed protein at pH 4.25 in the presence of different groupspecific protease inhibitors and then measured TPP-I activity.
Essentially identical results were obtained using purified Histagged CLN2p/TPP-I produced in insect cells, untagged CLN2p/TPP-I purified from a CHO cell expression system, mature CLN2p/TPP-I purified from human brain, and crude human brain homogenates (Table I). Consistent with previous studies, there was essentially no effect on activity by the metalloproteinase inhibitor EDTA, by the cysteine protease inhibitor E-64, by the aspartyl protease inhibitor pepstatin, or by the serine protease inhibitor phenylmethylsulfonyl fluoride. A relatively minor effect was seen with the metalloproteinase inhibitor 1,10-phenanthroline. Also, confirming the results of others (6, 15-17), a tripeptide substrate-based chloromethylketone (AAF-CMK) inhibited activity (Table I) in a competitive manner (data not shown). Unexpectedly, TPP-I activity was inhib- ited by the serine protease inhibitor DFP (Table I). Similar results were obtained when activity was assayed using fluorescein isothiocyanate-hemoglobin substrate (data not shown). In addition, the serine protease inhibitor 3,4-dichloroisocoumarin (DCI) partially inhibited activity (Table I). Note that at the acidic pH used for these experiments, DCI formed an insoluble precipitate at millimolar or higher concentrations, thus complicating analysis of the concentration dependence of this inhibitor.
We conducted a series of dilution experiments to ascertain the stability of the enzyme-inhibitor complex. Activated enzyme was preincubated with inhibitor for 1 h, and then the initial TPP-I activity was measured immediately after diluting the preparation into a substrate solution adjusted to contain the same or a 10-fold lower final concentration of inhibitor (Table II). This was compared with the activity of a parallel sample where low concentration of inhibitor was present in both the preincubation and assay mixture. These experiments revealed that the inhibition by AAF-CMK was readily reversible, indicating that the enzyme-inhibitor complex was in rapid equilibrium (Table II, compare treatments 2 and 4). The inhibition of activity by high concentrations of DCI was partially reversed following rapid dilution (Table II, compare treatments  5 and 7), and the reversal was nearly complete after 3 h (data not shown). In contrast, the inhibition by DFP was essentially irreversible (Table II, compare treatments 8 and 10), even when measured 24 h after dilution (data not shown). The DFP inhibition could be prevented by preincubating the enzyme with the competitive inhibitor AAF-CMK (Table II, compare treatment 9 with treatments 11 and 4) or by the AAF-AMC substrate (data not shown). Taken together, these data indicate that a serine in the active site of CLN2p/TPP-I plays a key role in catalysis.
We investigated different parameters that affected inactivation of CLN2p/TPP-I by DFP. Inactivation was dependent on time and DFP concentration (Fig. 3, top panel). Plots of the log of the residual activity versus time yielded a series of straight lines for different DFP concentrations, indicating that the inactivation reaction followed pseudo-first order kinetics (Fig. 3, middle panel). A double reciprocal plot of the inverse apparent first order rate constant versus inverse DFP concentration yielded a straight line passing near the origin, suggesting that DFP does not form a Michaelis complex with CLN2p/TPP-I (Fig. 3, bottom panel), and the second order rate constant (k app /[DFP]) was ϳ30 M Ϫ1 min Ϫ1 at 25°C and pH 4.25 (Fig. 3,  bottom panel, inset). Note that this analysis ignores hydrolysis of DFP during the reaction period and is likely to slightly underestimate the true second order rate constant (see Fig. 3  legend).
The rate of TPP-I/CLN2p inactivation by DFP was most rapid at pH ϳ4.5 (Fig. 4, top panel). This pH dependence was similar to that of enzymatic activity assayed using the AAF-AMC TPP-I substrate (Fig. 4, bottom panel). Experiments using nonactivated precursor and preactivated enzyme indicated that the proenzyme reacted very slowly if at all with DFP (data not shown). Given that the DFP reaction occurs by nucleophilic attack of an activated serine (18), these data indicate that both proteolytic processing and acidic pH are required for activation and/or accessibility of the catalytic serine to external substrates.
Identification of the Active Site Serine-To determine the site and stoichiometry of labeling, we compared tryptic digests of unmodified and DFP-inactivated CLN2p/TPP-I. MALDI-TOF mass spectrometry revealed that the patterns were essentially identical except that a 2835.1 m/z fragment disappeared and a new 3000.6 m/z peak appeared after DFP treatment (Fig. 5,  upper panel). The masses of these peaks indicate that they represent the same tryptic fragment Val 466 -Lys 492 with or without the diisopropyl phosphate modification (theoretical m/z, 3000.4 and 2836.2, respectively). Therefore, modification of CLN2p/TPP-I on a single residue within Val 466 -Lys 492 is responsible for inhibition.
HPLC analysis of tryptic digests also demonstrates that one residue within fragment Val 466 -Lys 492 is stoichiometrically labeled by DFP. For instance, in Fig. 5 (lower panel), the two peaks labeled A and AЈ present in a tryptic digest of unmodified CLN2p/TPP-I disappear after DFP inactivation with the concomitant appearance of the two peaks B and BЈ. Chemical sequencing and MALDI-TOF analysis reveal that peak A represents tryptic fragment Val 466 -Lys 492 , and AЈ represents an incompletely digested tryptic fragment encompassing Ala 448 -Lys 492 , whereas B and BЈ represent the corresponding diisopropyl phosphate-labeled fragments. The lack of any other discernable differences in the modified and unmodified digests also supports our conclusion that DFP-labeling of a single residue within Val 466 -Lys 492 is responsible for inhibition.
The Val 466 -Lys 492 CLN2p/TPP-I peptide and the corresponding regions of the BPICPs align to the active site serine of subtilisin (Fig. 6). Edman degradation of the purified DFPmodified tryptic peptide was complicated by poor retention of the peptide on the polyvinylidene difluoride membrane during sequencing (data not shown). However, in analysis of three different preparations, for cycles 1-16 we could clearly assign all residues to that predicted from the cDNA sequence except for cycle 10, which was predicted to be Ser 475 . Inspection of the repetitive yield of serine showed that serine clearly was present at cycles 7, 12, and 21 but absent at cycle 10. In contrast, serine was clearly present at cycle 10 in sequencing of the corresponding unmodified tryptic peptide. Finally, Ser 475 represents the only conserved serine among CLN2p/TPP-I and the BPICPs in the tryptic peptide. Taken together, these data indicate that Ser 475 represents the active site nucleophile of CLN2p/TPP-I and, by extension, the BPICPs.
Expression of Wild Type and Mutant CLN2p/TPP-I-To directly investigate the function of Ser 475 we expressed wild type CLN2p/TPP-I and a S475A mutant in CHO cells. The cells expressing wild type human CLN2p/TPP-I had elevated TPP-I activity and Western blotting revealed elevated levels of both precursor and processed protein (Fig. 7). The cells expressing the S475A mutant had activity similar to the neo-transfected control, indicating that the mutant was inactive compared with the wild type CLN2p/TPP-I protein. Western blotting indicated that the majority of the protein was in the precursor form, suggesting that processing was impaired (Fig. 7). However, the increased amount of ϳ46-kDa protein compare with the neotransfected control indicates that other proteases, possibly including endogenous CHO cell CLN2p/TPP-I, are capable of cleaving the catalytically inactive mutant. We also analyzed two constructs that had alanine substitutions at Asp 360 and Asp 517 , which align to residues important for function of the bacterial pepstatin-insensitive proteases (19,20). The two aspartate mutants were similar to the S475A mutation in terms of lack of enzyme activity and processing (Fig. 7). Finally, in a preliminary attempt to find other catalytically important residues, we also analyzed a H236A mutant and found that it resembled the wild type construct in regard to processing and activity (Fig. 7). DISCUSSION In this study, we find that CLN2p/TPP-I is synthesized as a catalytically inactive protein that upon acidification is autocatalytically processed to an active protease. More detailed kinetic studies indicate that activation can occur both by intermolecular and intramolecular events, with the latter route predominating at low proenzyme concentrations. 3 Based on the sequence of the precursor and mature protein, residues 20 -195 (M r ϭ 19,523) are removed following acidification. Processing is likely to entail an endoproteolytic cleavage of the CLN2p/TPP-I as shown by the appearance of small (Ͻ20 kDa) fragments that transiently appear after acidification. Consistent with this, CLN2p/TPP-I has recently been shown to have intrinsic endoproteolytic as well as tripeptidyl exopeptidase activity (8). The appearance of a transient 49-kDa species suggests that the initial endoproteolytic cleavage may occur upstream of the Leu 196 N terminus of the mature form. However, identification of processing intermediates is hindered by the appearance of TPP-I activity that can further trim various species including the remaining proenzyme, liberated propeptide, and the processing intermediates themselves. Although the precise details remain to be elucidated, our in vitro studies indicate that the CLN2p/TPP-I can undergo autocatalytic processing at conditions that mimic the acidic environment of the lysosome. This property should be useful when considering possible enzyme 3 I. Sohar and P. Lobel, manuscript in preparation.

FIG. 4. pH dependence of DFP inactivation and TPP-I activity.
Top panel, DFP inactivation. Preactivated CLN2p/TPP-I was incubated at 25°C in the absence and presence of 0.5 mM DFP in 100 mM formatebuffered saline/Triton of the indicated pH. At various times (0.17, 1, 2, 3, 4, and 5 h) aliquots were diluted into 250 mM formate-buffered Triton/saline, pH 4.0, containing AAF-AMC and assayed for TPP-I activity at 30°C. Second order rate constants (k app /[DFP]) were obtained as described in the legend to Fig. 3. Bottom panel, TPP-I activity. Preactivated CLN2p/TPP-I was diluted into 100 mM formate-buffered saline/Triton of the indicated pH containing AAF-AMC and assayed for TPP-I activity at 30°C. replacement therapy for LINCL patients, where the recombinant proenzyme could be administered as an inactive prodrug that is converted to active enzyme after proper delivery to its site of action.
Proteases are typically classified by catalytic type based on their susceptibility to different types of inhibitors, with DCI and DFP being diagnostic for serine proteases (21). We find that DCI inhibits CLN2p/TPP-I in a reversible manner. Although DCI is generally thought to be an irreversible inhibitor, some bona fide serine proteases do regain activity following DCI inactivation (22). Our clear finding that DFP inactivates CLN2p/TPP-I is somewhat unexpected given the reported insensitivity of TPP-I (6,8,17) and a bacterial pepstatin-insensitive protease (23) to this inhibitor. However, other studies have reported partial to complete inhibition of TPP-I activity with high concentrations of DFP (15,16). These discrepancies may reflect the kinetics and pH dependence of the DFP reaction. Even at its optimal pH, DFP inhibition of CLN2p/TPP-I is slow compared with that of other serine esterases (24), so it is possible that the experimental conditions were not optimized to detect the reaction.
Serine proteases typically contain a catalytic triad of aspartate, histidine, and serine residues, with the serine hydroxyl group being activated by the histidine, and the aspartate stabilizing the histidine imidazole ring as it gains and loses a proton (25). Although most serine proteases are inactive at acidic pH, this is not true for carboxypeptidase II and cathepsin A. In both of these enzymes, the environment of the active site includes a cluster of acid groups that may lower the histidine pK a (26,27) and thus allow it to participate in both general acid and base catalysis at acidic pH (25). There are also examples of serine proteases that lack the classic catalytic triad such as members of the SF and SG clan, which have a catalytic dyad consisting of a lysine and serine (28).
The results presented in this study, as well as our previous detection of a S475L mutation in a LINCL patient (29), demonstrate that Ser 475 represents the active site nucleophile of CLN2p/TPP-I. Previous studies reported that diethlypyrocarbonate inhibited TPP-I activity (6,16), suggesting that a histidine may be involved in catalysis. Although His 236 is excluded  (lanes 1 and 7), wild type human CLN2p/TPP-I (lanes 2 and 8), S475A mutant (lanes 3 and 9), D360A mutant (lanes 4 and 10), D517A mutant (lanes 5 and 11), and H236A mutant (lanes 6 and 12) were analyzed directly for CLN2p/ TPP-I expression by Western blotting. TPP-I activity was measured after incubating the samples at pH 3.5 to convert proenzyme to the active form as described under "Experimental Procedures." Data are representative of at least two independent clones for each construct.
Enzymological characterization of CLN2p/TPP-I from the catalytic triad based on our mutagenesis data, it is possible that some other histidine activates the serine. If so, the alignments among CLN2p/TPP-I, the BPICPs, and other apparently related proteins (29) do not reveal an obvious candidate histidine.
The lack of activity of the Asp 360 and Asp 517 mutants are consistent with a role for these residues in catalysis. One possibility is that one aspartate participates in a classical catalytic triad and the other helps decrease the pK a of the histidine. Alternatively, if a histidine is not important for catalysis, one or both of the aspartates may assume a novel role in directly activating the catalytic serine. Protein engineering combined with structural approaches will be required for detailed insights into the catalytic process.