Characterization and Cloning of Tripeptidyl Peptidase II from the Fruit Fly, Drosophila melanogaster *

We describe the characterization, cloning, and genetic analysis of tripeptidyl peptidase II (TPP II) fromDrosophila melanogaster. Mammalian TPP II removes N-terminal tripeptides, has wide distribution, and has been identified as the cholecystokinin-degrading peptidase in rat brain. Size exclusion and ion exchange chromatography produced a 70-fold purification of dTPP II activity from Drosophila tissue extracts. The substrate specificity and the inhibitor sensitivity of dTPP II is comparable to that of the human enzyme. In particular, dTPP II is sensitive to butabindide, a specific inhibitor of the rat cholecystokinin-inactivating activity. We isolated a 4309-base pairdTPP II cDNA which predicts a 1354-amino acid protein. The deduced human and Drosophila TPP II proteins display 38% overall identity. The catalytic triad, its spacing, and the sequences that surround it are highly conserved; the C-terminal end of dTPP II contains a 100-amino acid insert not found in the mammalian proteins. Recombinant dTPP II displays the predicted activity following expression in HEK cells. TPP II maps to cytological position 49F4-7; animals deficient for this interval show reduced TPP II activity.

Intercellular communication depends critically on both the generation and termination of biological signals. In the case of neuropeptides and neurohormones, the signaling molecules may be cleaved by extracellular enzymes to produce new active peptides (e.g. by angiotensin-converting enzyme, ACE 1 (1) and endothelin-converting enzyme (2)) or cleaved by them to be inactivated (e.g. by enkephalinase (3)). These peptidases are found in the extracellular space or on the cell surface and typically have low specificity (for example substance P, neurokinins, neurotensin, and the endothelins can all be cleaved by enkephalinase, reviewed in Refs. 4 and 5). Although a large number of peptidases have been identified in various tissues, only a handful have been shown definitively to be ectoenzymes with neuropeptide-degrading capability (reviewed in Refs. 4 and 5). These observations have led to the current hypothesis that diverse neuropeptides are inactivated by a relatively small number of enzymes. Recently Rose et al. (6) identified a membrane-associated variant of the enzyme tripeptidyl peptidase II (TPP II) as the peptidase responsible for cleavage and inactivation of the mammalian neuropeptide CCK-8. Intravenous injection of the potent TPP II inhibitor, butabindide, has prosatiety effects on both behavior and gastric emptying (6). Based on this pharmacology, as well as its substrate specificity, and its correlated expression in CCK-responsive tissues, Rose et al. (6) propose that the neuropeptidase activity of TPP II is intimately associated with CCK signaling and not broadly active on diverse neuropeptides.
TPP II was previously isolated and characterized as an extra-lysosomal peptidase that could release N-terminal tripeptides from a wide range of larger substrates (7,8). TPP II is a serine protease with a subtilisin-like catalytic domain, but compared with other subtilases, it contains an extended Cterminal region (9). The native form of the enzyme has a remarkably high molecular mass (Ͼ1000 kDa) that suggests an oligomeric association of the ϳ138-kDa subunits (7,8). The cDNAs encoding the human (10,11), murine (12), and rat (6) enzymes have been cloned, and the genomic region encoding a putative homologue in Caenorhabditis elegans has been sequenced (GenBank TM accession number U23176). In both human and mouse cDNAs (12), an alternatively spliced exon encoding an additional 13 amino acids has been identified, which is involved in complex formation (13). Rose et al. (6) identified both cytoplasmic and membrane-associated forms of rat TPP II and suggested that an alternatively spliced TPP II mRNA was involved in the membrane association through a glycosylphosphatidylinositol anchor.
Neuropeptides are important signaling molecules in insects (for review see Refs. 14 and 15), and the enzymes involved in neuropeptide regulation appear to be highly conserved (16). Previous studies of neuropeptidases in Drosophila have defined enzyme activities resembling enkephalinase (17,18), angiotensin-converting enzyme (19,20), and aminopeptidase activity (21). In each case, the Drosophila activities displayed similar substrate specificities and inhibitor sensitivities to known mammalian enzymes. Furthermore, the gene encoding the Drosophila ACE, AnCE, demonstrates a high degree of sequence similarity to one form of mammalian ACE (20). We propose to use Drosophila genetics to investigate the extent to which TPP II enzyme function has been conserved. In the current study, we show that a TPP II-like activity is present in Drosophila extracts. Furthermore, we partially purify this enzyme activity and characterize it in comparison to the mammalian enzyme. We have cloned and expressed a dTPP II cDNA. Finally, we use genetics to demonstrate that this cloned gene is largely responsible for the observed TPP II activity in tissue extracts.
We defined 1 unit to equal a change in absorbance of 0.001/min, which under these conditions corresponds to the hydrolysis of 0.04 nmol of the substrate, using the molar absorbance of 9.6 mM Ϫ1 cm Ϫ1 (25). One unit defined as above corresponds to about 0.01 unit defined previously (11) using the substrate Arg-Arg-Ala-[ 32 P]Ser-Val-Ala, as determined by measuring of the same sample of purified human TPP II under the two assay conditions. Enzyme Purification-Drosophila TPP II enzyme was partially purified from embryos and from adult flies in a two-step purification process. Tissues were homogenized at 1 g/4 ml in 50 mM Tris-HCl, pH 7.5, using a probe tip homogenizer (Ultra-turrax T25) with 2 pulses (10 s each) followed by sonication (MSE Soniprep 150) with 3 pulses (5 s each). Extract samples were prepared by centrifugation at 14,000 ϫ g for 30 min and filtration of the supernatant through glass wool. The filtered extract was loaded onto a Sepharose CL-4B column (Fig. 1A). The active fractions were pooled and further purified by DEAE-cellulose chromatography (Fig. 1C). The column was eluted with a potassium gradient, and fractions were collected and analyzed by the standard assay. Active fractions were pooled for further characterization. The amount of protein in the samples was measured by the modified Bradford method (27,28) with bovine serum albumin as standard.
Western Blot Analysis-Samples were mixed with sample buffer to give final concentrations of 2.3% SDS, 5% ␤-mercaptoethanol, and 10% glycerol. The samples were heated for 5 min at 95°C before loading onto an 8% polyacrylamide gel. The SDS-polyacrylamide gel electrophoresis and Western blot analysis were performed essentially as described previously (29) using chicken anti-human TPP II antibody. Silver staining of the polyacrylamide gel was performed according to Morrisey (30).
Characterization of dTPP II Activity-Assays to determine substrate specificity were conducted according to the standard procedure, but various chromogenic substrates were substituted for the standard AAF-pNA, all at a concentration of 0.2 mM. During inhibitor sensitivity assays, 0.5 mg/ml bovine serum albumin was substituted for DTT in the enzyme dilution buffer to stabilize the enzyme because the presence of 1 mM DTT would hamper the effect of the thiol-reactive compounds. The final concentration of DTT in the incubation was therefore about 50 M.
In control experiments greater than 95% of the activity was retained under these conditions. Inhibitor solutions were prepared according to manufacturers' protocols, and none of the solvents used had any effect on dTPP II activity at the concentrations used. The enzyme was preincubated with the inhibitor for 30 min at 22°C before addition of the substrate.
For the determination of K m , 2-5 units of enzyme prepared from Drosophila according to Table I or purified recombinant dTPP II were incubated in triplicates with AAF-pNA at concentrations of 1.0, 0.5, 0.2, 0.1, 0.05, and 0.025 mM in 0.1 M potassium phosphate buffer, pH 7.5, that contained 15% glycerol and 2.5 mM DTT and 1% Me 2 SO. The K m was calculated from Lineweaver-Burk plots to 157 and 138 M using two different Drosophila enzyme preparations and to 107 and 103 M for two different preparations of recombinant dTPP II. For the determination of K i for butabindide, the enzyme was incubated with AAF-pNA at concentrations of 0.5, 0.1, or 0.025 mM in the absence or presence of butabindide (0.025, 0.1, or 0.4 M) at 37°C and pH 7.5, as described above. The K i was determined in two separate experiments to 0.12 and 0.15 M using two different Drosophila enzyme preparations and 0.11 and 0.14 M for two different preparations of recombinant dTPP II.
HPLC Analysis-For analysis of cleavage products, 6 units of enzyme were incubated with the nonapeptide GVLRRASVA (10 nmoles) in 0.1 M potassium phosphate buffer, pH 7.5, that contained 2 mM DTT and 3% glycerol, in a final volume of 100 l at 37°C. As a control the enzyme was incubated without the substrate. The reaction was interrupted by diluting the sample 10-fold with buffer A, and the sample was stored at Ϫ20°C. An aliquot (200 l) was loaded on a Nucleosil C 18 column that was equilibrated with a mixture containing 90% (v/v) buffer A (60 mM sodium phosphate buffer, pH 2.8, 15 mM sodium pentane sulfonate, and 17.7 mM triethylamine) and 10% buffer B (18 mM sodium phosphate buffer, pH 2.8, 15 mM sodium pentane sulfonate, 17.7 mM triethylamine, and 35% (v/v) ethanol). As per Rosén et al. (31), elution was performed at 1 ml/min, with a gradient of 10 -90% buffer B. The gradient was started at 10 min and finished at 40 min, and the column was thereafter eluted with 90% buffer B for 20 min.
Molecular Biology-A partial clone, PGA9, recovered from a 9-to 12-h embryonic cDNA library (K. Zinn, CalTech) showed high sequence homology to the 5Ј end of the mammalian TPP II. 2 The missing 3Ј portion of the dTPP II gene was cloned using two PCR steps. Nested, gene-specific oligonucleotides were made in the sense direction corresponding to positions ϩ2171 (CCAAGCAATCCTGTCGT) and ϩ2278 (TGCATTGCCAAGTACTG) of the PGA9 clone. In the first round of PCR, oligonucleotide ϩ2171 and the lambda gt11 reverse vector oligonucleotide were used to amplify templates from 5 l of an amplified stock of the Zinn library. Klentaque LA (Sigma) enzyme was added prior to a 5-min 95°C phage-denaturing step followed by 30 cycles of 20-s 96°C denaturing, 30-s 50°C annealing, and 3-min 68°C extension. 5% of the unpurified product of the first PCR reaction served as template for a second round of amplification with the nested gene-specific oligonucleotide ϩ2278 and the lambda gt11 reverse vector oligonucleotide. The same cycling conditions, without the initial denaturing step, were used and produced a distinct band. This product was cloned into the T/A vector according to manufacturer's recommendations (Promega, Madison, WI) with the exception that the Klentaque LA enzyme was used. A full-length clone, LD18681, containing EST sequence number AA538993, was obtained from Genome Systems (St. Louis, MO). Genomic clones containing dTPPII sequences were isolated from a Charon 4A library using the PGA9 clone and standard techniques (32). The nucleotide sequence was determined by automated sequencing (Applied Biosystems) using gene-and vector-specific oligonucleotides.
Low stringency Southern blot analysis was performed using a 50% formamide hybridization solution at 37°C. The filter was rinsed with 2ϫ SSC, 0.1% SDS at room temperature.
Expression and Purification of Recombinant dTPP II-The dTPP II cDNA clone LD18681 was amplified by PCR using the vector oligonucleotide T3 and a gene-specific oligonucleotide that included the stop codon and introduced a NotI restriction site for cloning into the expression vector pCDNA3 (Invitrogen, Carlsbad, CA). Clones were recovered in both the forward and reverse orientation. The mouse TPP II gene was cloned into the EcoRI site of the same vector (13). The constructs were introduced into HEK cells in culture by lipid transfection (TransIT TM ) according to manufacturer's protocol (Mirus Corp., Madison, WI) and incubated 40 -60 h. Cells were harvested, rinsed in PBS, lysed in 50 mM Tris, pH 7.5, with 1% Triton X-100 (100 l/10 6 cells), and centrifuged at 14,000 ϫ g for 30 min. The supernatant was diluted 10-fold in 100 mM potassium phosphate buffer that contained 30% glycerol and 1 mM DTT. 2 R. Nusse and F. Van Leeuwen, personal communication.
TPP II activity was assayed according to standard procedure.
For the preparation of stable transformants the constructs were introduced into HEK cells in culture by the calcium phosphate precipitation method, and stable clones were selected after growing the cells in medium containing Geneticin, as described previously (13). Approximately 30 ϫ 10 6 cells were harvested, lysed (10 l lysis buffer/10 6 cells), centrifuged, and diluted as described above. The diluted supernatant (2.6 ml) was loaded onto a Sepharose column for chromatography and analysis as described in Fig. 1A. Pooled fractions were loaded as a 10-ml sample onto a DEAE-cellulose column, and chromatography was performed as described in Fig. 1C. The peak fractions were pooled and used for further characterization.
Cytological Location-Two independent strategies were used to identify the cytological location of the dTPP II gene. In situ hybridization of the genomic phage clones to polytene chromosomes of salivary glands of third instar larvae was carried out using a biotin-avidin detection system (ENZO Detek) and standard techniques (33). As the second strategy, we screened a Drosophila P1 library (34) (Genome Systems) with a 1-kb 3Ј fragment of the dTPP II cDNA.

RESULTS
Partial Purification of Drosophila TPP II-To investigate the endogenous TPP II-like activity in Drosophila extracts we used the substrate AAF-pNA (8,13). The activity was measured in the presence of the inhibitor bestatin in order to protect the substrate from sequential aminopeptidase degradation. We found approximately 0.5 units/g protein in Drosophila extracts (Table I). To achieve a partial purification of the Drosophila TPP II activity, we used the purification procedure that was previously designed for partial purification of TPP II from rat liver (7). Sepharose CL-4B followed by DEAE-cellulose chromatography produced a 70-fold purification, and 48% total recovery, of TPP II-like activity from the extract ( Fig. 1 and Table I). In the Sepharose CL-4B chromatography, the enzyme eluted at a K av of 0.26, which is comparable to the value for the purified mammalian enzyme (7,8,13). This indicates that, like the mammalian enzyme, the native M r is above 10 6 for the Drosophila enzyme and suggests that oligomeric dTPP II-included complexes are formed (Fig. 1A). Unlike mammalian TPP II, the Drosophila enzyme was not retained by the DEAEcellulose column. Based upon the dTPP II gene sequence (see below) the deduced isoelectric point is 7.0 for the dTPP II enzyme, compared with 6.2 for the human enzyme. The prediction of a higher isoelectric point for dTPP II is consistent with the observation that it does not bind under these conditions (Fig. 1C). Nevertheless, a large portion of the contaminating proteins did bind the anion exchanger, and a significant purification factor was achieved (Table I) as illustrated by the silver-stained gel (Fig. 2).
We monitored the extent of dTPP II purification using an antibody raised against the human TPP II (29). Western blot analysis revealed a ϳ150,000-Da band co-eluting with the enzyme activity in the Sepharose chromatography ( Fig. 1B and  Fig. 2). Following the DEAE chromatography (Fig. 2), the active fractions retained the same ϳ150-kDa TPP II immunoreactive band. This size is consistent with that predicted by the dTPP II sequence (see below). A few additional immunoreactive bands of lower molecular weight also co-purified with the enzyme activity; these may represent partially degraded sub-units still incorporated in the high molecular weight complex and/or other cross-reacting proteins. Immunoreactive bands of lower molecular weight were also seen in the purification of human TPP II (8,22).
We found a higher specific activity in extracts from embryos (0.83 units/g) as compared with adult Drosophila. The same purification scheme applied to the embryonic extracts produced an overall purification of only 40-fold. In preliminary experiments, the enzyme preparation from the embryo and adult tissues behaved identically with respect to substrate specificity and inhibitor sensitivity (data not shown). Due to its greater purity, the enzyme preparation derived from adult tissue was A, adult Drosophila extract (2.0 ml, corresponding to 0.5 g wet weight), was prepared and loaded onto a Sepharose CL-4B column (1 ϫ 89 cm); the column had a void volume of 27.5 ml and a total volume of 75.3 ml. The column was equilibrated and eluted with 100 mM potassium phosphate buffer, pH 7.5, that contained 30% glycerol and 1 mM DTT. Fractions of 0.9 ml were collected at a flow rate of 5.4 ml/h and pooled according to activity as indicated by the bar. The dashed line indicates the absorbance at 280 nm; the solid line indicates the peptidase activity; the black arrowhead indicates the K av corresponding to 10 6 Da; the white arrowhead indicates the estimated elution point of monomeric TPP II (ϳ150 kDa). B, Western blot analysis of fractions from the Sepharose column following SDS-polyacrylamide gel electrophoresis separation; samples were probed with chicken anti-human TPP II antibodies. Arrow indicates the predicted ϳ150-kDa size of dTPP II. C, 8.8 ml of the pooled active fractions from A were loaded onto a DEAE-column (1.6 ϫ 13.6 cm) that was equilibrated with 100 mM potassium phosphate buffer, pH 7.5, containing 30% glycerol and 1 mM DTT. Fractions eluted at a flow rate of 6 ml/h with a linear gradient (100 ϩ 100 ml) of 100 -750 mM potassium phosphate, pH 7.5, that contained glycerol and DTT. Fractions of 1.3 ml were collected and analyzed by the standard assay. The arrow indicates where the gradient was started. Fractions were pooled as indicated by the bar and used for further characterization.
selected for further characterization.
Characterization of dTPP II-In order to characterize further the partially purified dTPP II, we investigated its pH optimum, substrate specificity, inhibitor sensitivity, and kinetics. The optimal pH for cleavage of the standard substrate AAF-pNA was between 7.5 and 7.8 (data not shown). For comparison, the mammalian enzyme has a pH optimum of 7.5 with the hexapeptide Arg-Arg-Ala-Ser(P)-Val-Ala as the substrate (7,8). This feature differentiates TPP II from the lysosomal enzyme TPP I that prefers an acidic environment (35,36).
The sequential cleavage of a larger substrate to form a series of tripeptides is the defining nature of a tripeptidyl peptidase. The tripeptidyl peptidase specificity of the partially purified dTPP II was confirmed by cleavage of the nonapeptide Gly-Val-Leu-Arg-Arg-Ala-Ser-Val-Ala. Fig. 3 shows that the major products corresponded to tripeptides cleaved sequentially from the N terminus of the nonapeptide. No additional cleavage products were detected.
To pursue the hypothesis that the activity contains a true tripeptidyl peptidase, we used the chromogenic substrates Ala-pNA, Phe-pNA, and Suc-Ala-Ala-Phe-pNA (Table II). None of these substrates were cleaved at a detectable rate by the partially purified dTPP II. Failure to cleave suc-Ala-Ala-Phe-pNA demonstrated the need for a free N terminus and showed that the observed cleavage of the standard substrate was not due to a chymotrypsin-like activity. Failure to cleave the aminopeptidase substrates Ala-pNA and Phe-pNA demonstrated that the cleavage of AAF-pNA was not due to sequential cleavage by contaminating aminopeptidase activity. We studied the substrate specificity of dTPP II compared with that of the human enzyme by use of five additional chromogenic tripeptidyl-pNA substrates (Table II). From these experiments we conclude that various peptide bonds are cleaved at different rates and that the enzymes from human and Drosophila show similar specificity. Ala-Ala-Ala-pNA and Phe-Pro-Ala-pNA were cleaved at an appreciable rate by human and Drosophila enzymes. The cleavage of Val-Leu-Lys-pNA could only be detected when a relatively high enzyme concentration of either enzyme was used (data not shown). Ala-Phe-Pro-pNA and Pro-Leu-Gly-pNA were not cleaved by either the Drosophila or the human enzyme. Table III describes the sensitivity of dTPP II to various inhibitors. As expected, the Drosophila enzyme was inhibited by the serine peptidase inhibitor phenylmethanesulfonyl fluoride but not by the aspartic peptidase inhibitor pepstatin, the metallopeptidase inhibitor EDTA, or the aminopeptidase inhibitor bestatin. Similar to the human enzyme (8), dTPP II was inhibited by some thiol-reactive compounds (p-chloromercuribenzoate and N-ethylmaleimide) but not by others (e.g. io-  doacetamide). The specific TPP II inhibitor butabindide is a competitive inhibitor designed against the CCK-8 degrading activity of TPP II activity in rat brain (6). It was an efficient inhibitor of the Drosophila TPP II enzyme. However, the K i for butabindide determined as described under "Experimental Procedures" was about 0.1 M, i.e. an order of magnitude lower for the Drosophila enzyme than for the rat enzyme (7 nM) (6).
Cloning of dTPP II-A 2.5-kb clone fortuitously isolated by Drs. F. Van Leeuwen and R. Nusse from an expression library provided the 5Ј portion of the Drosophila TPP II cDNA (PGA9). The Berkeley Drosophila Genome Project (BDGP) has described several dTPP II ESTs (clot 474) containing TPP II sequences, among these, the EST produced from cDNA clone LD18681 indicates additional 5Ј-untranslated region sequences. By use of PCR a complete cDNA was isolated as described under "Experimental Procedures." The complete cDNA sequence consists of 4309 bp that contains a single long open reading frame encoding 1354 aa (Fig. 4). The putative initiator methionine, deduced by alignment with the human protein, is preceded by the sequence GCAGG which does not correspond well with the consensus for Drosophila mRNAs (37). The cDNA has an untranslated 5Ј end of 126 bp, containing no other AUGs. The 107-bp 3Ј-untranslated region ends with 15 A bases and displays the consensus polyadenylation signal AATAAA 12 bp upstream.
TPP II was previously cloned from rat, human, and mouse species (6, 10 -12). In addition, a putative homologue has been identified in C. elegans by the genome sequencing project (38). The deduced amino acid sequence of the proposed Drosophila TPP II homologue was aligned with the corresponding proteins from human and C. elegans using ClustalW method (39) and manual adjustment (Fig. 4). dTPP II is more similar to the mammalian enzyme (38% identity) than it is to the C. elegans TPP II (28%). The high degree of conservation is notable in the subtilisin-like catalytic domain (aa 1-526) (44%). A special feature of TPP II, in comparison with other subtilases, is the large 200-aa insert between the Asp and His of the catalytic triad (12). Only one other subtilase, pyrolysin, also has a large insert in this area (40). This region (amino acids 50 -270) shows a high degree of identity (33%) to human TPP II, whereas the identity to pyrolysin is only 17%. Together these observations support the hypothesis that the cloned Drosophila cDNA is a true TPP II homologue and not a different subtilase gene.
The C-terminal half of the deduced dTPP II protein (amino acids 925-1354) is less well conserved with human TPP II than is the N-terminal. When comparing the human and murine enzymes, the entire subunit was well conserved (96% overall identity), with the exception of residues 1140 -1183 which only showed 70% identity (12). The region corresponding to Drosophila aa 1033-1195 differs from those of the aligned human or C. elegans proteins in both sequence and number of amino acids. The size of this region in the C. elegans protein is more similar to that of Drosophila.
By using the PGA9 cDNA clone, we recovered three independent phage clones that represent overlapping portions of the dTPP II genomic locus. By blot and partial sequence analysis, we found that the entire 4309-bp cDNA maps to a genomic region not greater than 6 kb. Sequence analysis of the 3Ј end of the gene (corresponding to deduced aa 740 -1354) revealed the presence of four introns. None of the introns are greater than 100 bp, and they lie at deduced amino acids 752, 789, 1121, and 1231 (Fig. 4). The positions of Drosophila and deduced C. elegans TPP II introns are not similar. Furthermore, there is no potential for additional amino acids in the Drosophila sequence at the position aligned to the alternatively spliced site of the mammalian enzyme.
Expression and Characterization of Recombinant dTPP II-To test the activity of the putative dTPP II cDNA, we used transient heterologous expression in HEK-293 cells. Transfection with the dTPP II sequence produced a 1.5-11-fold increase of AAF-pNA cleavage activity in cell extracts (Fig. 5). The mouse sequence produced a 2-7-fold increase, under these same conditions. Butabindide produced approximately 80% inhibition of the measured activity which is consistent with the induction of a TPP II-like activity. Transfection with the dTPP II clone in reverse orientation produced no increase in TPP II activity.
Cells with a stable expression of recombinant dTPP II were selected as described under "Experimental Procedures." The highest expressing clones had 6-and 8-fold higher AAF-pNA cleaving activity than the control cells transfected with dTPP II in the reverse orientation. Control experiments demonstrated that the majority of the activity was cytoplasmic and not membrane-bound or secreted from the cells. The two-step purification procedure (cf. Fig. 1) was used to purify recombinant dTPP II from the cells (Fig. 6). The recombinant enzyme eluted from the Sepharose column with a K av of 0.26 (i.e. the same as the purified Drosophila enzyme). The recombinant dTPP II did not bind to the anion exchanger and could therefore be separated by this purification step from the endogenous human TPP II expressed by the HEK-293 cells (Fig. 6). The presence of an immunoreactive band of lower molecular weight co-purifying with the active enzyme indicated that, similar to the purified Drosophila enzyme (Fig. 2), the recombinant dTPP II was partly degraded. Upon prolonged storage at ϩ4°C, the Drosophila enzyme could be completely converted into the lower molecular weight product without any loss of activity (data not shown), thus indicating that the partial degradation was not detrimental to the activity of the enzyme. The recombinant enzyme had the same substrate specificity as the purified dTPP II (Table II) and was as sensitive to the inhibitor butabindide, i.e. in the presence of 2 M butabindide only 15 Ϯ 1% of the activity remained (cf. Table III). The K i for this inhibition was 0.1 M, consistent with that for the partially purified enzyme. In addition, the K m for the cleavage of the chromogenic substrate AAF-pNA, determined as described under "Experimental Procedures," was approximately 0.1 mM similar for both the purified and recombinant dTPP II.
dTPP II Cytological Location and Genetic Analysis-In situ hybridization of a dTPP II genomic phage clone to polytene chromosomes produced a major hybridizing band at 49F (Fig.  7A) and a minor band at 95F. A dTPP II-specific probe, repre- senting the 3Ј portion of the gene, produced a strong hybridization signal on the genomic P1 clone array (34) corresponding to the clone DS01087. This P1 clone has been assigned to the Mp20 contig and maps to the chromosomal interval 49F14 -59A1. Low stringency Southern blot analysis of D. melanogaster genomic DNA gave a hybridization pattern corresponding to the recovered genomic clone and no additional bands (data not shown). Together these observations indicate the presence of a single dTPP II gene at position 49F of chromosome 2R. We attribute the weak hybridization at 95F to sequences in the genomic phage clone unrelated to dTPP II. We related the observed enzyme activity to the cloned dTPP II gene through the analysis of available mutations of the 49F region. Southern blot analysis of heterozygous deficiency lines using a dTPP II-specific probe in comparison to an unrelated control probe revealed reduced relative hybridization in three lines. This result reflects the heterozygous deletion of the dTPP II coding region in the lines Df(2R)vg 33, Df(2R)vg 56, and Df(2R)vg B but not in Df(2R)vg C (Fig. 7B). TPP II activity of extracts from the three dTPP II-deficient lines was reduced ϳ50 -60% compared with wild type and to the non-dTPP IIcompromised stock Df(2R)vg C (Fig. 7C). For each line, the cleavage rate of the control substrate, AFP-pNA, was not significantly different from that of wild type; this demonstrates that there is not a general reduction in peptidase activity. The ratios of AAF:AFP cleavage rates among these lines were significantly different (one-way analysis of variance: F(4,19) ϭ 21.06, p ϭ 0.0001). As compared with wild type, the ratio of AAF:AFP cleavage was significantly reduced for the lines Df(2R)vg 33, Df(2R)vg 56, and Df(2R)vg B but not for Df(2R)vg C (Scheff's S test p ϭ 0.0001, p ϭ 0.0001, p ϭ 0.0005, p ϭ 0.6134, respectively). The correlation between reduced TPP II activity and elimination of dTPP II DNA sequences suggests that the 49F dTPP II gene is largely responsible for the observed activity.
The deficiency lines are homozygous lethal, and display reduced vitality even as heterozygotes. 3 Within the region defined by deficiency analysis, there are at least 11 lethal complementation groups (23). This complexity precludes analysis of the deficiency lines for dTPP II-specific phenotypes. Two of the 11 lethal mutations have been correlated with specific genes (see "Discussion" for details); the remaining 9 complementation groups represent candidate mutations of the dTPP II locus. We analyzed dTPP II enzyme levels in one or more representative alleles from each of these complementation groups to determine if any represent mutations in the dTPP II locus. Two alleles of one complementation group, vr6, showed reduced AAF-pNA cleavage. However, we found no evidence for decreased dTPP II (AAF-pNA cleavage) as compared with control peptidase activity (AFP-pNA cleavage) in any of the stocks tested ( Fig. 8) (one-way analysis of variance: F(13,47) ϭ 0.231, p ϭ 0.996). Thus, none of these alleles are likely to be specific mutations in the dTPP II gene.

DISCUSSION
This study has assigned a TPP II identity to a Drosophila gene located at cytological position 49F. This assignment derives from results of biochemical, molecular, and genetic experiments. Antibodies against the human enzyme cross-react with a Drosophila protein of expected size that is enriched through the partial purification of activity. The partially purified activity shows similar pH optimum, substrate specificity, and inhibitor sensitivity compared with the mammalian TPP II. The Lacking an endogenous substrate, we used AAF-pNA to assay TPP II activity from Drosophila because this chromogenic peptide has been shown to be a good substrate (8). This assay revealed a high level of TPP II-like activity in both adult and embryonic Drosophila extracts, 0.5 and 0.85 units/g, respectively. For comparison, when using the substrate RRAS( 32 P)-VA, the TPP II activity in rat liver homogenate was 1.2 units/g (8); this corresponds to about 0.1 units/g under the present assay conditions (see "Experimental Procedures").
Our two-step chromatographic purification strategy follows that used for the initial characterization of rat TPP II (7). We have used the partially purified material to begin defining the properties of the insect enzyme in comparison to those of mammalian TPP II. The Drosophila enzyme, both partially purified from tissue and recombinant, cleaved the substrate AAF-pNA with a K m of 0.1 mM, whereas the K m for the human enzyme is 0.02 mM when measured under identical conditions. 4 This value corresponds to previously reported K m values for cleavage of a similar substrate AAF-7-amido-4-methylcoumarin by mammalian TPP II: 0.025 mM for rat TPP II (6) and 0.016 mM for human TPP II (8). Based on these results, we hypothesize that human TPP II cleaves AAF-pNA more efficiently than does the Drosophila enzyme. A deviating K m value (0.148 mM) has, however, been reported for human TPP II (41).
TPP II displays low substrate specificity in vitro, as evidenced by the release of tripeptides of no apparent similarity. However, the peptidase is not indiscriminate in that different peptide substrates are cleaved at different rates (7)(8)41). The cleavage rate is sensitive to the N-terminal tripeptide sequence and also to the C-terminal sequences and to the phosphorylation state of the substrate (7). However, the tripeptidyl nature of TPP II is absolute. Our results using a nonapeptide substrate (Fig. 3) as well as the chromogenic tripeptide substrates (Table II) demonstrate that the Drosophila enzyme activity adheres to the strict tripeptidyl peptidase characteristic. Furthermore, although the exact rate of cleavage may vary between the mammalian and the Drosophila enzymes, good substrates for one are also cleaved well by the other. Proline residues within the N-terminal tripeptide sequence affect the cleavage rate by the mammalian enzyme (8). Among the chromogenic tripeptide substrates tested, a proline is accepted only in the second position by both the Drosophila and mammalian enzymes (Table II), although a proline in the first position is accepted by the human enzyme in some peptide substrates (41). 4 B. Tomkinson, unpublished data. FIG. 7. Analysis of four genetic deficiencies of the 49F polytene region. A, in situ hybridization of a single genomic phage clone to polytene chromosomes produces a major band at 49F7-8 (black arrow) and a minor band at 95F (not shown). B, high stringency Southern blot analysis of five distinct deficiency stocks. For each lane, DNA of 10 flies was restricted with the enzyme PstI. The filter was hybridized with a dTPP II-specific probe (bp 1-913) and an unrelated, control probe (ϳ500 bp) from the 83E polytene region, each at a specific activity of ϳ10 6 cpm/ml. C, enzyme activity assay. Extracts of 10 flies from each of the five distinct, genetic stocks were prepared and assayed according to "Experimental Procedures." The cleavage rates for AAF-pNA (black bars) and AFP-pNA (gray bars) are expressed as a percentage of the wild type cleavage rate on that substrate. The data represent the mean of three experiments Ϯ S.D. For each experiment wild type activity levels were determined as the mean of four samples. D, schematic representation of genomic regions deleted in the various deficiency stocks. Hatched bar indicates minimal region of dTPP II gene as determined from data of B and C. Bracket marks the dTPP II localization by in situ hybridization (49F7-8, A). In conclusion, even though the general pattern of acceptable substrates is conserved between the enzyme from human and Drosophila, there are small differences in the relative rates of cleavage.
The results of the inhibitor sensitivity for dTPP II (Table III) are consistent with those observed previously for human TPP II (8) as well as with our own preliminary comparison to human TPP II. The inhibitor sensitivities define the enzyme activity as that of a serine protease. Furthermore, dTPP II is sensitive to the same thiol-reactive compounds as is the mammalian enzyme (8), suggesting a similarity in tertiary structure and accessibility to essential cysteine residues. This property is quite remarkable since only 3 out of 16 cysteine residues in the predicted Drosophila enzyme are conserved in the human enzyme (aa 207, 599, and 760). Of these three, only Cys-760 is also conserved in the predicted C. elegans sequence (Fig. 4).
The Drosophila EST project has produced 9 ESTs (clot 474) containing TPP II sequences, all of which derive from the dTPP-II gene we have mapped to 49F. The C. elegans genome project has identified 5 cDNAs which correspond to a predicted C. elegans TPP II gene. The longest C. elegans clone, yk15e7, contains sequence that predicts an initiator methionine 46 amino acids upstream from the methionine which begins the region of high homology. There is no evidence for alternate initiation sites for the mammalian or Drosophila enzyme.
The Sepharose column provides an estimation of the very high molecular weight of the native dTPP II enzyme (Fig. 1A). Thus similar to mammalian TPP II, the Drosophila enzyme is active in a large, oligomeric complex. In addition to this abundant active complex, mammalian TPP II can form an even larger complex, eluting at the void volume of the Sepharose column, as judged by Sepharose chromatography of human erythrocytes (8) and expression experiments of alternate splice forms (13). Since an activity shoulder can be seen in this position also for the Drosophila enzyme (Fig. 1A), it is possible that the Drosophila enzyme can form a similar larger complex. Two features of the mammalian sequence have been implicated in the formation of the oligomeric TPP II complex. The first is a KEKE domain (13); such domains are thought to mediate protein-protein interactions (42). There is no apparent KEKE motif in the Drosophila protein. The second domain implicated in oligomer formation is a 13-amino acid sequence (GQSAAKRQGKFKK, separated from the KEKE domain by 20 amino acids in the human sequence) (13). The Drosophila enzyme does not contain a sequence resembling the 13-amino acid domain at a position aligned to the mammalian splice site (aa 985). In addition, the possibility of an alternative exon at this position is precluded because the genomic sequence for this region corresponds directly to that of the cDNA without interruption. However, dTPP II does contain a sequence closer to the C terminus (GESADKQKEDQKK, aa 1175) that shows some similarity to the 13-amino acid domain. When expressed in a human cell line, the recombinant dTPP II, including the similar Drosophila 13-aa domain, did not give rise to the higher molecular weight complex that elutes in the void volume of the Sepharose column. This suggests that either this 13-amino acid region does not serve a similar complex-forming function in the Drosophila enzyme or that the complex requires additional Drosophila proteins not present in the transfected cell line. These differences in the behavior of the dTPP II enzyme indicate a need for further structure-function studies. Such studies will be facilitated by the ability to perform them in vivo through efficient generation of transgenic Drosophila. Furthermore, the hypothesis that TPP II is involved in CCK physiology (6) raises several questions that can best be addressed by genetic experiments.
Another advantage of using Drosophila is the ability to create and isolate mutants. Toward this goal we have identified three deficiency lines that display reduced TPP II activity due to loss of the gene. Embryos homozygous for the deficiencies Df(2R)vg 56, Df(2R)vg 33, or Df(2R)vg B display abnormal mitosis soon after fertilization (43), and development proceeds until the germ band retraction stage when the embryos die, presumably due to loss of multiple gene functions. Therefore, accurate analysis of a dTPP II null phenotype is not possible with these large deficiencies. These lines define a cytogenetic region that contains 11 known lethal complementation groups (23,44) 5 suggesting the presence of at least 11 vital gene functions. Two of these have been identified as follows: one, vr10, is the dDP transcription factor involved in cell cycle regulation (44), and the other, vr 14, is Su(z)2, a DNA-binding protein (45,46). We assayed the AAF-pNA cleavage activity for stocks representing one or more alleles from each of the (as yet) unidentified complementation groups. None of the lines showed significant reduction of TPP II-like activity, relative to the AFP-pNA cleavage activity. From this we conclude that the dTPP II gene is not mutated in any of these defined lethal stocks. Further analysis is required to determine if dTPP II is necessary for survival.
The genetic analysis initiated in this study provides the basis for future studies of TPP II function. TPP II is a single copy gene in Drosophila, and deficiencies of the dTPP II locus significantly reduce the enzyme activity. Our results now provide the means to identify single gene mutations of dTPP II and analyze its loss-of-function phenotypes. By using germ line transformation, it will also be possible to test partial restoration and gain-of-function TPP II phenotypes to address the hypothesis that this enzyme is involved in neuropeptide metabolism.