Primary structure of rat liver dipeptidyl peptidase IV deduced from its cDNA and identification of the NH2-terminal signal sequence as the membrane-anchoring domain.

Two forms of dipeptidyl peptidase IV (DPP) were purified from rat liver plasma membranes: a membrane form (mDPP) extracted with Triton X-100 and a soluble form (sDPP) prepared by treatment with papain. Apparent molecular masses of mDPP and sDPP were 109 and 105 kDa, respectively, when determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The NH2-terminal sequences of the two forms were found to be completely different from each other. For further information on the molecular structure, we constructed a lambda gt11 liver cDNA library and isolated two cDNA clones for DPP, lambda cDP37 and lambda cD5. The 3.5-kilobase cDNA insert of lambda cDP37 contains an open reading frame that encodes a 767-residue polypeptide with a calculated size of 88,107 Da, which is in reasonable agreement with that of DPP (87 kDa) immunoprecipitated from cell-free translation products. Eight potential N-linked glycosylation sites were found in the molecule, accounting for the difference in mass between the precursor and mature forms. Of particular interest is that the deduced NH2-terminal sequence with a characteristic signal peptide is completely identical to that determined for mDPP. In addition, the NH2-terminal sequence of sDPP is identified in the predicted sequence starting at the 35th position from the NH2 terminus. These results indicate that the signal peptide of DPP is not cleaved off during biosynthesis but functions as the membrane-anchoring domain even in the mature form. It is also found that the primary structure thus predicted has striking homology to that of gp 110, a bile canaliculus domain-specific membrane glycoprotein (Hong, W., and Doyle, D. (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 7962-7966).

Primary Structure of Rat Liver Dipeptidyl Peptidase IV Deduced from Its cDNA and Identification of the NHz-terminal Signal Sequence as the Membrane-anchoring Domain* (Received for publication, September 19,1988) Shigenori Ogata, Yoshio Misumi, and Yukio IkeharaS Two forms of dipeptidyl peptidase IV (DPP) were purified from rat liver plasma membranes: a membrane form (mDPP) extracted with Triton X-100 and a soluble form (sDPP) prepared by treatment with papain. Apparent molecular masses of mDPP and sDPP were 109 and 105 kDa, respectively, when determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The NH2-terminal sequences of the two forms were found to be completely different from each other.
For further information on the molecular structure, we constructed a hgt 11 liver cDNA library and isolated two cDNA clones for DPP, hcDP37 and XcD5. The 3.5kilobase cDNA insert of hcDP37 contains an open reading frame that encodes a 767-residue polypeptide with a calculated size of 88,107 Da, which is in reasonable agreement with that of DPP (87 kDa) immunoprecipitated from cell-free translation products. Eight potential N-linked glycosylation sites were found in the molecule, accounting for the difference in mass between the precursor and mature forms. Of particular interest is that the deduced NHz-terminal sequence with a characteristic signal peptide is completely identical to that determined for mDPP. In addition, the NH2-terminal sequence of sDPP is identified in the predicted sequence starting at the 35th position from the NH2 terminus. These results indicate that the signal peptide of DPP is not cleaved off during biosynthesis but functions as the membrane-anchoring domain even in the mature form. It is also found that the primary structure thus predicted has striking homology to that of gpl10, a bile canaliculus domain-specific membrane glycoprotein (Hong, W., and Doyle, D. (1987) Proc. Natl. Acad. Sei. U. S. A. 84,7962-7966).
Dipeptidyl peptidase IV (DPP)' (EC 3.4.14.5) is a serine * This study was supported in part by a Grant-in-Aid for scientific research from the Ministry of Education, Science, and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
to the GenBankTM/EMBL Data Bank with accession number(s) T h e nucleotide sequence(s) reported in this paper has been submitted
The abbreviations used are: DPP, dipeptidd peptidase IV; mDPP and sDPP, membrane and soluble forms, respectively, of DPP; HPLC, high performance liquid chromatography; PAGE, polyacrylamide gel electrophoresis; SDS, sodium dodecyl sulfate, kb, kilobase(s). peptidase that cleaves dipeptides from the NHz-terminal end of peptide chains provided that the penultimate residue is proline (1). The enzyme is an intrinsic membrane glycoprotein, and the highest activity is found in the kidney and the intestinal brush-border membrane (2,3). Although its activity is relatively low in liver, DPP is also localized in the apical domain, bile canalicular membrane of hepatocytes (4)(5)(6). The purified enzyme is found to be dimeric, comprising two identical subunits of 110-130 kDa (4, 7) which are variable depending on species and tissue, possibly due to the extent of glycosylation (4,(6)(7)(8)(9).
Several hydrolases in the kidney and the intestinal brushborder membrane have been studied in detail for their orientation in the membrane (2,3). These enzymes, including DPP, aminopeptidases A and N, neutral endopeptidase (EC 3.4.24.11), y-glutamyl transpeptidase, and sucrase-isomaltase, have a common feature in their membrane topology. Most of their protein mass, including the catalytic site, protrudes on the extracellular side. Comparisons of the NHz-terminal sequences of the detergent-and proteinase-released forms suggest that their anchoring domain is located in the NH2terminal region (2,(10)(11)(12). Recently this mode of anchoring has been confirmed by cloning and sequencing of the cDNAs for y-glutamyl transpeptidase (13,14), sucrase-isomaltase (151, and neutral endopeptidase (EC 3.4.24.11) (16) but not yet for DPP.
To study in more detail the membrane anchoring and primary structure of DPP, we have cloned and sequenced its cDNA. The deduced amino acid sequence was compared with the NHZ-terminal sequence of the detergent-and papainsolubilized forms, demonstrating that DPP has an uncleaved signal sequence at its NHz terminus functioning as the membrane-anchoring domain.  Purification of Two Forms of DPP-Plasma membranes were isolated from Donryu rat livers as described previously (17,181. A soluble form of DPP (sDPP) was purified as follows. Plasma membranes were suspended in 20 mM Tris-HCl (pH 7.5) containing 5 mM Lcysteine (protein concentration, about 20 mg/ml). Papain (final concentration, 1.5 mg/ml) was added to the membrane suspension, and the mixture was stirred at 37 "C for 3 h followed by centrifugation at 105,000 X g for 1 h. The resulting supernatant was subjected to (NH&SO, precipitation. Proteins precipitated between 60 and 90% saturated (NH4)2S04 were dissolved in 3 ml of 20 mM Tris-HC1 (pH 7.5) containing 0.2 M NaCl and subjected to gel filtration through a Sephacryl 5-300 column (2.5 X 100 cm) equilibrated with the above buffer. Fractions with DPP activity were pooled and then applied to a wheat germ lectin-Sepharose column (2 X 15 cm) (4). After the column was washed with 200 ml of 20 mM Tris-HCI (pH 7.5) containing 0.5 M NaCl, adsorbed proteins including DPP were eluted with 0.2 M GlcNAc in the same buffer. Fractions with DPP activity were pooled and concentrated to a small volume of 20 mM Tris-HC1 (pH 7.5) in an Amicon Corp. ultrafiltration cell with an XM-50 membrane. Aliquots of the sample were subjected to preparative polyacrylamide gel electrophoresis (PAGE) at pH 8.5 followed by staining the gels for enzyme activity (19). Stained areas of the gels were cut out and homogenized in 25 ml of 20 mM Tris-HC1 (pH 7.5). The homogenates were stirred at 4 "C for 2-3 h and then centrifuged at 30,000 X g for 30 min. Samples obtained by repeating PAGE/ extraction were combined and concentrated to about 2 ml. After these purification steps were repeated several times, we finally obtained 3.2 mg of the purified sDPP from 1,079 mg of plasma membranes used.
A membrane form of DPP (mDPP) was purified as follows. The plasma membrane suspension was adjusted to contain 0.5% Triton X-100 and 20 mM Tris-HC1 (pH 7.5) and stirred at 4 "C for 30 min followed by centrifugation at 105,000 X g for 1 h. The resulting supernatant was applied to an Affi-Gel Blue column (3 X 30 cm). When the column was subjected to stepwise elutions with 0.6, 1.0, and 1.5 M NaCl in 20 mM Tris-HC1 (pH 7.5) containing 0.2 mM MgC12 and 0.5% Triton X-100, most of the DPP activity was eluted with 1.0 M NaCl. Fractions of the activity peak were pooled and subjected successively to wheat germ lectin-Sepharose chromatography and PAGE in the presence of Triton X-100 as described for the sDPP preparation. Finally, 3.1 mg of the purified mDPP was obtained from 916 mg of plasma membranes under these conditions.
Isolation of CNBr Cleavage Peptides-The purified sDPP was cleaved with 2.5 M CNBr in 70% (v/v) formic acid for 24 h at room temperature (20). The sample thus treated was subjected to high performance liquid chromatography (HPLC) through a Superose 12 column (1.0 X 30 cm) with 70% formic acid resulting in separation of eight major peptide peaks when monitored by absorbance at 280 nm. Fractions of each peak were pooled and freeze-dried. Each sample was dissolved in 0.1% trifluoroacetic acid and then subjected to HPLC on a pBondapack CIS column (0.4 X 30 cm) with a linear gradient from 0 to 90% acetonitrile in 0.1% trifluoroacetic acid (20). Fractions of each major peptide peak were freeze-dried and used for chemical analysis.
Determination of Amino Acid Sequences-Purified samples of mDPP and sDPP (30 pg each) and CNBr cleavage peptides (about 5 pg each) were sequenced on an Applied Biosystems Inc. model 477A Gas-Phase Sequencer with an on-line model 120A phenylthiohydantoin derivative analyzer using the manufacturer's program (20,21).
Poly(A)+ RNA Extraction and Fractionation-Total poly(A)+ RNA was prepared from rat livers and fractionated by centrifugation on a sucrose density gradient (5-20%) as described previously (20). Identification of fractions containing the DPP mRNA was carried out by translation in vitro of the fractionated poly(A)' RNA in the reticulocyte-lysate system (22). Each fractionated RNA (2 pg/assay) was incubated in the reaction mixture at 25 "C for 90 min with 50 pCi of [35S]methionine. 35S-Labeled DPP was immunoprecipitated from the translation products and analyzed by SDS-PAGE (9.0% gels) followed by fluorography as described (23).
cDNA Library Construction and Screening-The poly(A)+ RNA fraction enriched with the DPP mRNA was used for construction of the following cDNA library. Double-stranded cDNA was prepared according to the method described by Gubler and Hoffman (24) and methylated with EcoRI methylase. EcoRI linkers were ligated to the cDNA and digested with EcoRI. Linked cDNA was purified by chromatography through a Bio-Gel A-50 column and ligated to hgtll (25). The ligated cDNA was packaged into bacteriophages by using the Gigapack Gold kit followed by plating onto dishes with appropriate bacterial hosts (26). The resultant library of 1.4 X IO6 recombinant phages was divided into two portions, 8 X 10' and 6 X lo5 independent clones. One portion of the cDNA clones (8 X lo5) was amplified as the cDNA library stock. The other portion (6 X 10' clones), without prior amplification, was screened with anti-rat liver DPP IgG in combination with peroxidase-conjugated second antibodies (27). Plaques of positive clones were isolated, and bacteriophage DNAs were digested with EcoRI. EcoRI-excised cDNA inserts were subcloned into plasmid vector pUC118 (28) and characterized by restriction endonuclease mapping (20). A clone (hcD5) with the longest cDNA insert (3.6 kb) was obtained. An EcoRI-BglII fragment with 891 base pairs was prepared from the XcD5 clone, oligolabeled, and used as a probe for further screening the 8 X lo5 clones of the cDNA library stock. Out of 40 positive clones obtained, another cDNA clone (hcDP37) with the longest insert (3.5 kb) was isolated.
DNA Sequencing-Restriction and exonuclease I11 fragments of cDNAs were subcloned into pUC118 or pUC119 (28). After singlestranded DNAs were isolated with the aid of helper phages (M13K07), both strands of all regions were sequenced by the dideoxynucleotide chain termination method (29) using the SequenaseTM DNA-sequencing kit.
Computer Analysis of cDNA and Protein-Nucleotide and protein sequences were analyzed by using the GENAS System at Kyushu University Computer Center (30). Hydropathy analysis was performed in accordance with Kyte and Doolittle (31).
Other Methods-DPP activity was assayed with Gly-Pro-p-nitroaniline as a substrate (34). Protein was determined by the method of Lowry et al. (35) with bovine serum albumin as a standard. Anti-(rat liver DPP) antiserum was prepared as follows. 0.5 ml of the purified sDPP (1 mg/ml) was emulsified with an equal volume of Freund's complete adjuvant and injected intramuscularly into multiple sites on the back of a rabbit. At 2-week intervals the rabbit received three injections of the antigen in the same dose in complete adjuvant. Two weeks after the last injection, 0.2 mg of the antigen in 0.2 ml of saline was injected intravenously three times every other day. A week after the last booster injection, the rabbit was bled for collection of the antiserum. The monospecific anti-DPP IgG was purified from the antiserum by affinity chromatography on Sepharose 4B coupled with the purified sDPP (20).

RESULTS
Purification and NH2-terrninal Sequencing of mDPP and sDPP-Two forms of DPP, the Triton X-100-solubilized membrane form and the papain-released soluble form, were purified from rat liver plasma membranes. When the purified samples were subjected to PAGE in the absence of detergents, mDPP did not completely migrate into the gel (Fig. 1A, lane  1 ), in contrast to the mobility of sDPP (lane 2 ) and that of mDPP after treatment with papain (lane 3). The retardation in mobility of mDPP may be due mainly to aggregation in the gels. The aggregation of mDPP in the absence of detergents was confirmed by gel filtration through a Sephacryl S-300 column in which most of the mDPP was eluted at the void volume of the column, whereas sDPP was at a position corresponding to 250 kDa (data not shown). The difference in the two forms was also observed in their mobility on SDS-PAGE (Fig. 1B). Apparent sizes of mDPP and sDPP were estimated to be 109 (lane 1 ) and 105 kDa (lane 2 ) , respectively. The purified mDPP was converted by papain treatment to a form ( h n e 3 ) with the same molecular mass as that of sDPP.
The purified mDPP and sDPP were then subjected to sequencing. Their sequences determined to position 20 were completely different from each other (see Fig. 4).  stretch of hydrophobic amino acids from position 7, in contrast to that of sDPP. Taken together, these results suggest that mDPP as compared with sDPP has a hydrophobic NH2terminal extension of about 4 kDa which, when removed by papain, results in the conversion of mDPP to sDPP. I n Vitro Translation and Fractionation of D P P mRNA-Poly(A)+ was prepared from rat liver and translated in vitro with (:''S]methionine. A single component of 87 kDa (Fig. 2, lane 1 ) was obtained by immunoprecipitation with anti-DPP IgG from the products translated in the absence of dog pancreas microsomes, whereas a larger component of 103 kDa (lane 2) was obtained from the products translated in the presence of the microsomes. The latter form was converted to an 88-kDa form when treated with endo-P-N-acetylglucosaminidase H (data not shown). The results indicate that the 87-kDa precursor of DPP is glycosylated to the 103-kDa form in the presence of the microsomes. The difference in mass between the 103-kDa form and purified mDPP (109 kDa) may reflect further processing of its oligosaccharides from the high mannose to the complex type. For enrichment of the DPP mRNA, the poly(A)+ RNA fraction from rat liver was subjected to centrifugation on a sucrose density gradient and fractionated (data not shown). Cell-free translation of each RNA fraction identified the location of the DPP mRNA, a fraction of which was used for construction of a cDNA library.
Isolation of Two Overlapping cDNA Clones for DPP-One portion of the Xgtll cDNA library (6 x lo5 clones), without prior amplification, was screened with anti-DPP IgG. Of seven positive recombinant phages obtained, clone XcD5 had the longest cDNA insert (3.6 kb). This insert was found to contain an open reading frame coding for all six CNBr fragments of DPP but not for the NH2-terminal sequences of mDPP and sDPP (Figs. 3 and 4), indicating that clone XcD5 does not contain the total mRNA sequence. For further screening of the cDNA library, we prepared an EcoRI-BglII fragment of 891 base pairs of XcD5 as a probe. Screening of 8 X IO5 clones with the new probe yielded 40 positive plaques, of which clone XcDP37 had the longest insert, with 3.5 kb. The cDNA inserts of the two clones thus obtained were subcloned into the plasmid vector pUC118, and the relationship among these plasmid inserts was analyzed by restriction endonuclease mapping (Fig. 3). The clones XcDP37 and XcD5 contained overlapping cDNA inserts which together spanned a stretch of 4.9 kb of DNA.
Nucleotide and Deduced Amino Acid Sequence of DPP-The two cDNA inserts were sequenced from multiple restriction sites and on both DNA strands according to the strategy outlined in Fig. 3. The combined nucleotide sequence includes the complete coding and 3"noncoding region, as well as part of the 5"noncoding region (Fig. 4). XcDP37 comprises 3512 base pairs corresponding to nucleotides from -88 to 3424, in which an in-phase TGA stop codon is found 54 nucleotides upstream from the initiator ATG. XcD5 comprises 3576 base pairs corresponding to nucleotides from 1271 to the 3'-end at 4846, in which the polyadenylation signal AATAAA is found a t two sites. The appearance of more than one polyadenylation signal may imply the occurrence of polymorphism in the 3"noncoding region of mRNA.  The open reading frame encodes a protein of 767 residues with a calculated mass of 88,107 Da (Fig. 4). The sequence starts with a putative signal peptide which has a hydrophobic core domain (indicated by a broken line in Fig. 4) preceded by the positively charged lysine. The predicted NHz-terminal sequence, however, is found to be completely identical to that of mDPP determined by Edman degradation. In addition, the NHn-terminal sequence for sDPP is identified in a sequence starting at the 35th position, alanine, from the NH, terminus.
Thus, it is evident that the putative signal peptide cannot be cleaved during biosynthesis but remains as the potential membrane-spanning domain of DPP. The deduced structure also contains all the other peptide sequences determined by Edman degradation. The entire primary structure contains 8 asparagine residues that are part of the consensus sequence Asn-X-Ser/Thr for N-glycosylation sites. Glycosylation of some of these asparagine residues could account for the difference between the molecular mass calculated from the amino acid composition (88,107 Da) and that estimated by SDS-PAGE for the mature protein (109 kDa). Fig. 5 shows a hydropathy plot of the cDNA-derived DPP sequence. Regions of about 20 amino acids with a hydropathy average greater than 1.6 are likely to be associated with the lipid bilayer, spanning the membrane in a helical conformation (31,36). The only potential hydrophobic anchor segment is the sequence between residues 7 and 28 at the NH2 terminus (Figs. 4 and 5). The absence of a sufficiently hydrophobic segment in any other portion is in agreement with the fact that papain treatment releases a completely soluble form of  identified (2,4).
The primary structure of DPP thus obtained also demonstrates that five of the eight potential N-glycosylation sites are located in the NHp-terminal region, where no cysteine residues are found at all (Fig. 5). Most of the cysteine residues are characteristically observed in the middle portion of the molecule.
of Dipeptidyl Peptidase IV Homology to g p l l 0-The predicted amino acid sequence of DPP was compared for homology with the known sequences of several other membrane-bound enzymes and glycoproteins in rat liver. Striking homology was found between DPP and gpll0, a bile canaliculus domain-specific membrane glycoprotein (37). As shown in Fig. 6, gpll0 has the entire sequence of 792 residues in contrast to the 767 in DPP. However, only 5 amino acid residues are different from each other in the corresponding positions when compared up to position 767. The difference of 4 residues in positions 183, 394, 562, and 624 is caused by single nucleotide substitutions, and that in position 767 results from a single nucleotide insertion (indicated by an arrow in Fig. 6). The frame shift by this insertion also causes an extension of the open reading frame, resulting in the longer sequence in gpll0.

DISCUSSION
Clones XcDP37 and hcD5 obtained in this study contain overlapping DNA inserts which together span a stretch of 4.9 kb of DNA. Complete sequence analysis of the clones demonstrates an open reading frame of 2301 nucleotides starting at the first ATG codon encountered from the 5'-end. We believe that this open reading frame encodes the total primary structure of DPP. This is based on the findings that the NH2terminal sequence of the mDPP and several other sequences determined by Edman degradation are all identified in the protein sequence deduced from the cDNA and that an inphase TGA stop codon is found 54 nucleotides upstream from the initiator ATG (Fig. 4).
An attempt to identify the membrane-anchoring domain of DPP was made previously by Macnair and Kenny (2). They purified the detergent and autolysis forms of DPP from pig kidney microvillar membrane and analyzed them for their NH2-terminal sequences. The sequences determined for the detergent (8 residues) and autolysis forms (16 residues) correspond to those starting at the 12th and 37th positions, respectively, from the NHz terminus in the primary structure predicted here, although there are differences in some positions due to the species difference. All the data presented in this study support the conclusion that DPP is anchored in the membrane by a single hydrophobic segment located at the NH, terminus: (i) the NHz-terminal sequence of the detergent form is completely identical to that of the predicted structure; (ii) the sequence between residues 7 and 28 at the NH, terminus is the only domain that fulfills the criteria proposed by Kyte and Doolittle (31) for membrane-spanning domains in transmembrane proteins; (iii) the soluble form released by papain has an NHn-terminal sequence starting at position 35 in the predicted sequence; (iv) the difference in molecular mass between mDPP and sDPP (4 kDa) is in good agreement with the mass of the predicted fragment (3566 Da of residues 1-34) remaining in the membrane after papain cleavage. The NHz-terminal segment of DPP is also the most probable candidate for a signal sequence to target the protein to the endoplasmic reticulum and initiate its translocation across the membrane. Thus this highly hydrophobic stretch is proposed to have dual roles as both a translocation signal and membrane anchor. An uncleaved signal sequence functioning as a membrane anchor at the NHz terminus has been shown for influenza neuraminidase (38), y-glutamyl transpeptidase (13,14), sucrase-isomaltase (15), and neutral endopeptidase (EC 3.4.24.11) (16). There is no explanation available at present for why the NHz-terminal signal peptide of these membrane proteins is uncleaved, although mechanisms for protein insertion into membranes have been proposed by many investigators (for review see Refs. 39 and 40). A cleavable NHz-terminal signal sequence found on most secretory proteins and many transmembrane glycoproteins has three structurally and possibly functionally distinct regions: a basic NH2-terminal region, a hydrophobic core region, and a more polar COOH-terminal region (41)(42)(43). In the last region of the three, positions -3 and -1 relative to the cleavage site are considered to be the most important for recognition by signal peptidase (41,42). Three regions (residues 1-6, 7-28, and 29-35) of the DPP sequence reasonably satisfy the criteria proposed for the three domains of the cleavable signal. Nevertheless, cleavage of the signal sequence does not occur in DPP, suggesting that there must be some difference between the cleavable and uncleavable signals. A careful comparison of the available sequences reveals that the hydrophobic core region of all the uncleavable signals consists of at least 22 amino acid residues (13)(14)(15)(16)38, and this study), significantly different from that of the cleavable ones with the average of 13 residues (variable from the shortest 7 to the longest 16) (41)(42)(43). Thus, it is likely that the length of the hydrophobic core region is an additional important factor for actual cleavage by signal peptidase. The predicted DPP sequence has no cysteine residue in the NHp-terminal half of the molecule followed by a middle portion containing a relatively high content of cysteine residues (Figs. 4 and 5 ) . This may be favorable for a "stalked" structure proposed for DPP (2,3), the stalk of which is easily attacked by proteinases such as papain. Although a Ser/Thr-rich sequence, which could be potential sites for 0-glycosylation, is characteristically observed in the stalked domain of sucraseisomaltase (3,15) and the low density lipoprotein receptor (44), the DPP structure contains no such sequence. In fact, no galactosamine was identified in purified DPP when analyzed for its chemical composition.' The sequence, instead, contains eight potential N-glycosylation sites, glycosylation of which accounts for the difference in molecular mass between the predicted precursor and mature forms. It is, however, unlikely that all of the eight sites are glycosylated since detailed analysis of its carbohydrate moiety suggests the presence of at most six oligosaccharide chains/molecule2 (9).
Striking homology in the predicted sequences of DPP and gpll0 (37) is of particular interest (Fig. 6). The gpll0 glycoprotein was isolated from rat liver plasma membranes, the same source as our DPP preparation, and identified by an immunocytochemical technique to localize specifically in the bile canalicular domain (45) as has been shown for DPP (5, * S. Ogata, T. Fujiwara, and Y . Ikehara, manuscript in preparation. Primary Structure of L 6). Our preparation of DPP was always monitored by the enzyme activity, and the polyclonal antibodies against the purified enzyme were found to be monospecific for it. In contrast, gpll0 has no assigned enzymatic activity because it was purified as a denatured form from gels after SDS-PAGE (45). The anti-gpll0 antibodies, however, were reported not to cross-react with DPP (45). It was speculated that a hydrophobic domain at the COOH terminus of a l l 0 (residues 624-648) is its transmembrane domain (37), in contrast to our proposal.