Identification of Functional Domains of Rat Intestinal Phospholipase B/Lipase

A cDNA encoding a rat intestinal Ca2+-independent phospholipase B/lipase (PLB/LIP) was cloned from an ileac mucosa cDNA library using a probe amplified by polymerase chain reaction based on the purified enzyme’s sequence. PLB/LIP consists of an NH2-terminal signal peptide, four tandem repeats of about 350 amino acids each, and a hydrophobic domain near the COOH terminus. The enzyme purified previously was found to be derived from the second repeat part. To examine the function of each domain, the full-length PLB/LIP, individual repeats, and a protein lacking the COOH-terminal hydrophobic stretch were expressed in COS-7 cells. The results showed that the second repeat, but not the other repeats, had all the activities (phospholipase A2, lysophospholipase, and lipase) found in the purified natural and expressed full-length enzymes, suggesting repeat 2 is a catalytic domain. The full-length enzyme was mainly present in membrane fractions and efficiently solubilized by treatment with 1% Triton X-100, but not with phosphatidylinositol-specific phospholipase C. Deletion of the COOH-terminal hydrophobic stretch caused the secretion of >90% of synthesized PLB/LIP into culture media. These results suggest the hydrophobic domain is not replaced by a glycosylphosphatidylinositol anchor but serves as a membrane anchor directly. A message of the full-length PLB/LIP was abundantly expressed in the ileum and also, in a smaller, but significant amount, in the esophagus and testis. Immunohistochemistry showed that PLB/LIP is localized in brush border membranes of the absorptive cells, Paneth cells, and acrosomes of spermatid, suggesting its roles related and unrelated to intestinal digestion.

Digestion of phospholipids and triacylglycerol in gastrointestinal tract involves several hydrolysis reactions. A variety of lipases, including acid lipase in lingual gland or stomach, and pancreatic lipase (which is of primary importance in luminal digestion of fats), hydrolyze triacylglycerol to produce monoacylglycerols and free fatty acids (see Refs. 1-3 for reviews)).
Pancreatic phospholipases A 2 (PLA 2 ) 1 hydrolyze ester bonds at the sn-2 position of glycerophospholipids and produce fatty acids and lysophospholipids. These steps are prerequisites for lipid absorption by intestinal epithelium cells (4). Lysophospholipid can be directly absorbed, or hydrolyzed by pancreatic lysophospholipase and converted to glycerol 3-phosphate esters and fatty acids.
Until recently, all those processes were believed to proceed in the lumen of alimentary tracts by the action of secretory enzymes mentioned above. However, recent studies suggest the presence of a lipid-hydrolyzing enzyme associated with intestinal brush border membranes (5,6), named phospholipase B/lipase (PLB/LIP) because this enzyme displayed broad lipolytic activities (PLA 2 , lysophospholipase, and lipase activities) (7). PLB/LIP might participate in terminal digestion, or membrane digestion, of dietary lipids and biliary phospholipids, like well established glycosidases and peptidases in brush border membranes (8). In the preceding paper, we detailed the purification and characterization of the catalytic domain of PLB/ LIP after its solubilization from the membrane by autolysis. The purified enzyme consisted of a 14-kDa peptide and a 21-kDa glycosylated peptide and catalyzed PLA 2 , lysophospholipase, and lipase reactions in a single active site (9). To understand the molecular basis of such a broad enzyme specificity, the structure of the nascent enzyme, the membrane anchoring mode, and its physiological significance, we cloned and sequenced the full-length PLB/LIP cDNA based on information of the NH 2 -terminal amino acid sequences of the two peptides derived from the purified enzyme. PLB/LIP is translated as a large single peptide of 1450 amino acids containing a signal peptide, four tandem repeats, and the COOH-terminal hydrophobic stretch. The second repeat has the stretch coding the amino acid sequences found in the previously purified enzyme and has full enzymatic activities, as confirmed by the expression of individual repeats in COS-7 cells. Furthermore, to determine whether or not the hydrophobic stretch is responsible for a membrane anchor, a deleted mutant was constructed and expressed in COS-7 cells.

EXPERIMENTAL PROCEDURES
Materials-1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) was obtained from Avanti Polar Lipids, Inc. Trioleoylglycerol (TOG), 1-palmitoyl-sn-glycero-3-phosphocholine (1-palmitoyl-GPC), and diiso-* This work was supported in part by a grant from JCR Pharmaceuticals Co, Ltd. and by Research Grant 08670173 from the Ministry of Education 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 "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) D63648.
Amplification of cDNA Fragments by Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Surgically removed rat intestines were washed with distilled water. The ileal mucosa was scraped off and stored at Ϫ80°C until use. The frozen tissues (about 1.0 g) were homogenized in 10 ml of 4 M guanidine thiocyanate containing 50 mM Tris-HCl (pH 7.6), 1 mM EDTA, 2% sodium lauryl sarcosinate, and 2% 2-mercaptoethanol. The homogenate was extracted with phenol/chloroform, and RNA in the aqueous phase separated by centrifugation was precipitated with ethanol. One g of total RNA was reverse-transcribed with 200 units of SuperScript RT (Life Technologies, Inc.) in the presence of 150 ng random hexanucleotides in a 20-l reaction mixture according to the protocol recommended by the manufacturer. The products were heated at 70°C for 5 min, and then a 10-l aliquot of the reaction mixture was added to a 90-l PCR mixture containing 100 pmol of primers. The primers used were as follows: primer KF, ccgga-tccAA(A/G)TT(C/T)ACITG(C/T)CCIGA(C/T)AA(A/G)GA(C/T)CC, corresponding to the previously determined peptide, KFTCPDKDP, followed by BamHI adaptor; primer QT, GCIGT(C/T)TTIAT(A/G)TT(A/G)TCIG-T(A/G)AANGT(C/T)TG, corresponding to the peptide QTFTDNIKTA (9), where the letter I in the oligonucleotide sequences means inosine. The PCR was conducted in 30 cycles of reactions at 94°C for 40 s, 54°C for 1 min, and 72°C for 1 min. The amplified DNA fragments were visualized by ethidium bromide staining as a single band of 540 bp on a 5% polyacrylamide gel. The products were directly cloned into pCRII plasmid vector (Invitrogen) without further purification. Six clones containing the PCR products were isolated. Double-stranded DNA sequencing of each clone was performed by the dideoxy method (10) using an Applied Biosystems model 373A automatic DNA sequencer according to the manufacturer's instructions. Predicted amino acid sequences at both termini of these clones coincided with those of the primers. One clone named pCR-RIPLB was used as the probe DNA for RNA blot analysis and screening of a phage library.
Construction and Screening of cDNA Library-A rat ileum mucosa cDNA library was prepared based on the method of Gubler and Hoffman (11) using a MosElox cloning vector (Amersham). Six g of poly(A) ϩ RNA, an oligo(dT) primer, and avian myeloblastosis virus reverse transcriptase were used for the first strand synthesis. The second strand was then synthesized by treatment with RNase H and DNA polymerase I, and then the generated cDNA was blunt-ended with T4 DNA polymerase, ligated to EcoRI adaptor (Amersham) by T4 ligase, and further treated with polynucleotide kinase. The cDNA greater than 500 bp in length were selected on a spin column. The resultant cDNAs were ligated into an EcoRI site of MosElox vector, and the vectors were packaged into 3 ϫ 10 6 phages. The library was screened without amplification.
For screening, 10 5 phages were plated and replicas were made from the resultant plaques (12). Positively hybridizable plaques were screened with a 32 P-labeled BamHI cDNA fragment of pCR-RIPLB and 50 strongly hybridized clones were selected and stored at 4°C. Randomly chosen 20 clones of the 50 were subjected to the second screening, and 19 positive clones were isolated. To recover the cloned cDNAs as plasmids, an Escherichia coli strain BM25.8 (Amersham) was infected with the phage, and plasmid was subcloned by a cre (site-direct endonuclease)-mediated plasmid self-subclone system (13). Plasmids prepared from ampicillin-resistant colonies were digested with EcoRI, and the fragments were separated on an agarose gel. Four clones revealed an identical pattern composed of three fragments of 1.4 kb, 3.1 kb, and 4.0 kb. Since the 4.0-kb fragment was derived from the vector, we concluded that each of these four clones contained a 4.5-kb insert with a single EcoRI site, which was the same size as the PLB/LIP transcript as judged by Northern blot analysis. One of the four clones, named pMOS-RIPLB3, was used for further analysis. To determine its complete nucleotide sequence, the EcoRI-digested cDNA fragments were independently subcloned into pBluescript KS(Ϫ) (Stratagene). A plasmid containing a 1.4-kb cDNA fragment (NH 2 -terminal portion) and that containing a 3.1-kb fragment (COOH-terminal portion) were named pBS-3S and pBS-3L, respectively. Clones having various sizes of the insert were obtained from each plasmid by ExoIII nuclease and mung bean digestion system (Takara, Kyoto, Japan).
Preparation of Antibody and Immunoblot Analysis-Polyclonal antibodies were raised against a PLB/LIP fragment (amino acids 450 -1450). The antigen was produced in E. coli as a fusion protein using T7 RNA polymerase expression system (14). An EcoRI insert of pBS-3L encoding the peptide was subcloned into EcoRI-cleaved pET28a(ϩ) vector (Novagen), and it was transfected into E. coli strain, BL21(DE3) (Novagen). The expressed proteins were solubilized from a particulate fraction of cell homogenates with an SDS sample buffer containing 1% SDS and 10% mercaptoethanol, and subjected to SDS-PAGE (15, 16). The fusion protein (about 1.5 mg) was electrophoretically extracted from the gel slices (12). Its aliquot (about 0.4 mg) was emulsified with 2 volumes of complete Freund's adjuvant and used to immunize Japanese white rabbits (females, body weight of 2.0 kg). Immunoglobulin G (IgG) was purified from 15 ml of antiserum by an IgG purification kit (Amersham).
The freshly prepared intestinal mucosa was homogenized in 5 volumes of HBS buffer (0.25 M sucrose, 1 mM EDTA, 30 mM Tris-HCl (pH 7.4), 2 mM MgCl 2 ). Testis tissues and sperm collected from epididymis were sonicated in 10 volumes of HBS buffer, and the homogenates were centrifuged at 1,000 ϫ g for 10 min at 4°C. Supernatants were further centrifuged at 100,000 ϫ g for 45 min at 4°C, and pellets (membrane fraction) were taken and suspended in 0.1 volume of HBS buffer. All the samples were stored at Ϫ40°C until use. The proteins (20 g each) were subjected to SDS-PAGE (on 6 or 10% gels), and immunoblotting was performed as described (12). The antigen-antibody complexes on the membranes were further reacted with goat anti-rabbit IgG-horseradish peroxidase conjugate (Cappel). The peroxidase conjugate was developed with a Konica immunostaining kit (Konica, Japan).
Expression of PLB/LIP in COS-7 Cells-To construct the expression vector, pMOS-RIPLB3 was partially digested by BamHI and the fulllength 4.5-kb cDNA fragment having adaptor links on both sides was isolated by agarose gel electrophoresis and cloned into the BamHI site of a pSVL vector (Pharmacia Biotech Inc.). This vector, named pSVL-RIPLB3, had an extra ATG at NcoI site in the adaptor sequence preceding the initiating ATG of RIPLB cDNA. This was removed by replacing an XhoI fragment of pSVL-RIPLB3 with a fragment from the initiator ATG to XhoI site (nucleotide position 333) that was amplified by PCR using the pMOS-RIPLB3 as a template and primers, Nfor ( Expression plasmids for repeats 1, 2, 3, or 4, named pSVL#1, pSVL#2, pSVL#3, or pSVL#4, respectively, were constructed by linking a cDNA fragment encoding the NH 2 -terminal peptide of PLB/LIP including the signal peptide with a cDNA encoding each repeat to meet requirements for secretable proteins. We first constructed pSVL-Nhead that contains a cDNA encoding amino acids 1-40 of PLB/LIP and a synthetic polylinker consisting of NheI and DraI sites, a stop codon, and a MluI site in this order. The insert was amplified by PCR using primers Nfor and Nrev (Fig. 3), and pSVL-RIPLB as template, and subcloned directly into a pMOS-Blue vector (Amersham). The ApaI-BamHI fragment excised from this vector was substituted for an ApaI-BamHI fragment of pSVL-RIPLB, generating pSVL-Nhead. pSVL#1 containing amino acids 1-357 of PLB/LIP was generated by replacing an ApaI-DraI fragment of pSVL-Nhead with an ApaI-PvuII fragment of pSVL-RIPLB (Fig. 3). Expression vectors for the other repeats were constructed as follows. Fragments encoding repeats 2, 3, and 4 were first amplified by PCR using the pSVL-RIPLB template, sense primers (F2, F3, and F4, respectively) with a 5Ј-terminal NheI site, and antisense primers (R2, R3, and R4, respectively) with a 5Јterminal MluI and stop codon. Then they were digested by NheI and MluI, and the resultant fragments were finally cloned into pSVL-Nhead that had been previously digested by the same restriction endonucleases. An expression vector lacking the COOH-terminal hydrophobic stretch (pSVL-⌬C) was constructed by replacing the 3Ј-side BamHI fragment of pSVL-RIPLB (nucleotides 3701-3Ј end) with that of pSVL#4 (nucleotides 3701-4259). COS-7 cells (4 ϫ 10 6 cells) suspended in 0.4 ml of saline G were transfected with the plasmids (50 g) in the presence of 250 g of carrier DNA by electroporation using a single pulse (125 microfarads at 400 V) as reported (16). The efficiency of transfection estimated by enzyme activity recovered medium and cell fractions depended on cell density (16); the cell density of 60 -70% confluence was optimal. The transfected cells were plated on two 9-cm dishes containing Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. The dishes were incubated at 37°C under an atmosphere of 5% CO 2 , 95% air. The medium was changed once every 24 h. The medium was recovered for enzyme assay, and the cells were harvested for immunoblot and Northern blot analyses after 72 h.
Fractionation of COS-7 Cells-Cells were sonicated in 600 l of HBS buffer, and the sonicate was centrifuged at 1,000 ϫ g at 4°C for 10 min. The supernatant was further centrifuged at 100,000 ϫ g at 4°C for 45 min to obtain the cytosol and membrane fractions. The pellets were resuspended in 200 l of HBS buffer. All the fractions were stored at Ϫ40°C until use.
Lipolytic Activity Measurements-PLA 2 , lysophospholipase, and lipase activities were measured using POPC, 1-palmitoyl-GPC, and FIG. 1 triolein as substrates, respectively, as described in the preceding paper (9). The lipolytic activities of PLB/LIP toward exogenous substrates were expressed as the difference between the activities in the presence and absence of the substrates and as nanomoles of free fatty acid released/1 min. Protein concentration was measured by a Bio-Rad protein assay system.
RT-PCR Analysis-Total RNA (1 g) prepared from the ileum and poly(A)-rich RNA (1 g) prepared from esophagus or testis were reverse-transcribed as described above. Transcripts (4 l) were mixed with a pair of primers (50 pmol each), Nfor and R2 or F3 and R4, in 5 l of 10 ϫ Taq polymerase buffer supplied by company (Takara), 4 l of 10 mM dNTP solution, 5 l of dimethyl sulfoxide, and H 2 O to give the final volume of reaction mixture, 50 l. The mixture was overlaid with 30 l of mineral oil to prevent evaporation. Taq polymerase (2.5 units) was added to the reaction mixture, and the chain reaction was conducted as follows: denaturation at 94°C for 40 s, annealing at 60°C for 1 min, and polymerization at 72°C for 4 min with 30 cycles.
Immunohistochemical Analysis-Rat ileum and testis were fixed in 4% paraformaldehyde/PBS (pH 7.4) for 2 h. Frozen sections (6 m) prepared on glass slides were stained immunohistochemically (17). Prior to incubation with the primary antiserum, the sections were treated with Gomori's oxidation mixture (0.5% potassium permanganate, 0.25% sulfuric acid) (18). The primary antiserum was diluted 1:300, and the second antibody (fluorescein isothiocyanate-labeled goat anti-rabbit IgG) was diluted 1:300. For the controls, the preimmune rabbit serum and the primary antiserum absorbed by the E. coli extract containing overexpressed antigen (PLB/LIP) were used. The E. coli containing PLB/LIP was solubilized in 1% SDS-PBS, diluted 20 times with PBS, and incubated with the antiserum for 12 h at 4°C. The specimens were observed under a fluorescence microscope (Carl Zeiss, Oberkochen, Germany). For testes, photographs were taken in combination with phase-contrast transmission light.

RESULTS
Isolation of cDNA Encoding PLB/LIP-A set of oligonucleotide primers designed on the basis of previously determined partial amino acid sequences of rat intestinal PLB/LIP (9) was used for amplification of cDNA fragments from total RNA of intestinal mucosa by RT-PCR. The PCR products were visualized as a clear 540-bp single band on agarose or polyacrylamide gel. The cDNA was sequenced, and parts of the predicted amino acid sequence coincided with those of peptides derived from the purified enzyme. The cDNA was used as a probe for screening the cDNA library of ileal mucosa. Four independent clones were isolated, all having the identical fragment pattern when digested by restriction enzymes. A cDNA insert of one of the clones was cleaved in two fragments by EcoRI digestion, each of which was subcloned into plasmid and partially sequenced. A longer fragment had a poly(A) tail and the sequence corresponding to the large peptide of purified PLB/LIP, and the other had a TGA-opal codon just six bases upstream of the first ATG starting codon and the sequence corresponding to its small peptide. From these results, we concluded that this clone contained a full open reading frame of PLB/LIP.
Nucleotide Sequence and Predicted Amino Acid Sequence-As shown in Fig. 1, the full-length 4590-bp cDNA of PLB/LIP contained an open reading frame encoding a 1450amino acid protein with a calculated molecular mass of 161 kDa and a potential polyadenylation sequence (AATAAn) (19) just upstream of its poly(A) tail. Two hydrophobic regions were identified in the protein: the first was a putative 30-amino acid signal peptide at the NH 2 terminus (20), and the second a stretch from residue 1421 to 1443, functioning as a transmembrane anchor as described below. Notably, the PLB/LIP contains four tandem repeats in a putative extracellular domain of the protein, each repeat being composed of about 350 amino acid residues (Fig. 2). These repeats began shortly after the signal sequence and occupied 93% of the entire protein. Amino acid sequences corresponding to the PCR-amplified cDNA probe were found in the second repeat.
A Swissprot data base search revealed that the amino acid sequence of PLB/LIP was 67.9% identical to that of AdRab-B, the cDNA of which had been isolated as an adult-specific clone of rabbit intestine (21). Both proteins shared functionally important domains, i.e. a putative NH 2 -terminal signal sequence, four tandem repeats each of which was composed of about 350 amino acids and occupied more than 90% of the entire protein, and a short hydrophobic stretch followed by a short hydrophilic one at the COOH-terminal end. There are 14 potential Nglycosylation sites, of which one is in the NH 2 -terminal sequence preceding repeat 1, two in repeat 1, one in repeat 2, six in repeat 3, and four in repeat 4, when weak sites in an Asn-Pro-Ser(Thr) sequence were not counted.
Expression of PLB/LIP in COS-7 Cells-To confirm that the rat intestinal PLB/LIP cDNA determined here encodes an active enzyme, we expressed the full-length cDNA in COS-7 cells. The cells and culture media were separately recovered and the former was further fractionated into cytosol and membrane fractions as described under "Experimental Procedures." Expression was assessed in these fractions by enzyme assay and immunoblot analyses. The cells transfected with the recombinant plasmid revealed greatly enhanced PLA 2 , lysophospholipase and lipase activities, compared with cells transfected with the pSVL vector alone. This confirmed that the full-length PLB/LIP exhibits the same substrate specificity as the enzyme purified from rat intestine (9) as a single polypeptide protein.
These activities were mainly localized in membrane fractions as verified by immunoblot analysis (Fig. 4, a and b), confirming PLB/LIP is a membrane-bound protein (9). The enzyme produced by COS-7 cells apparently migrated as two bands on a 6% gel, and their estimated molecular masses (ϳ200 and 220 kDa) were larger than that calculated from the predicted amino acid sequence data (161 kDa). This may be due in part to the difference in the degree of glycosylation.
Membrane Anchoring Mode of PLB/LIP-The COOH-terminal segment, residues 1421-1443, is the only stretch that is hydrophobic and long enough to meet requirement for membrane anchoring. To determine whether the stretch plays a role in binding of the enzyme to membrane, an expression vector (pSVL-⌬C) that lacked the COOH-terminal 42 amino acid residues (Fig. 3c) was constructed and introduced into COS-7 cells. Immunoblot analysis indicated that about 90% of the expressed truncated enzyme was recovered in culture media (Fig. 4a), in contrast with membrane localization of the full-length PLB/ LIP. In that culture medium, PLA 2 and lysophospholipase activities were unexpectedly not observed, but a substantial activity of lipase was detectable. The truncated protein, however, restored phospholipid hydrolyzing activities to the level as comparable as lipase activity, when dithiothreitol (DTT) was phobic stretch can serve as a direct membrane anchor or as a signal for attaching a glycosylphosphatidylinositol (GPI) moiety to the COOH-terminal end after the cleavage of that stretch (22). As in the case of PLB/LIP in brush border membranes (9), phospholipase C treatment (25 and 100 munits/ml) did not solubilized PLA 2 activity in membrane fractions of full-length PLB/LIP-expressing COS-7 cells, but treatment with 1% Triton X-100, a poor solubilizer for GPI-anchored proteins (23), did solubilize 72% of the activity. This suggests that the hydrophobic stretch functions as a direct membrane anchor.
Location of the Catalytic Domain-The enzyme purified from rat intestine had an apparent molecular weight of 35,000 with PLA 2 , lysophospholipase and lipase activities. It consisted of the 14-kDa and 21-kDa peptides, and their amino acid sequences were derived from the second repeat of PLB/LIP, suggesting the second repeat is a catalytic domain. To determine which domains are responsible for the activities observed in the expressed full-length and purified natural PLB/LIP, we constructed four expression vectors, each of which contained a nucleotide sequence encoding a NH 2 -terminal signal peptide and the respective repeats (Fig. 3c), and transfected them into COS-7 cells. The antibody used in this study (raised against the PLB/LIP peptide, which lacked the first repeat) immunoreacted with the second and third repeats well, but did not with the first or the fourth repeat. We estimated the level of expression of each repeat in the cells by Northern blot analysis using the full-length PLB/LIP cDNA as a probe. As shown in Fig. 5a, transcripts of 2.0 kb were found at comparable levels in the cells transfected with each vector. The culture media and harvested cells were examined for the enzyme activities (Fig. 5b). All the activities found in the purified enzyme were present in the culture media of repeat 2-expressing cells, but not in the media of cells transfected with the other constructs. No activity was detected in the cell homogenates except for a trace amount of enzyme activities in repeat 2-expressing cell fractions (data not shown). This may preclude impaired secretion of expressed proteins. Furthermore, dithiothreitol had no effect on the enzyme activities of all culture media (data not shown). DFP is a potent inhibitor that inhibited PLA 2 , lysophospholipase, and lipase activities of the purified PLB/LIP to a comparable extent (9). It similarly inhibited these three activities of the expressed full-length PLB/LIP by 99, 88, and 94%, respectively, and of the repeat 2 enzyme by 99, 97, and 91%, respectively, compared with those in the absence of inhibitors. These results suggested that repeat 2 is the catalytic domain with a single active site responsible for all the enzymatic activities tested.
Distribution of PLB/LIP-The distribution of PLB/LIP among various rat tissues was determined by Northern analysis using rat PLB/LIP cDNA as a probe. Fig. 6a showed that two kinds of transcripts of about 4.6 kb and larger than 6.0 kb in length were readily detected in the ileum. The shorter transcript was also present in esophagus and testis in a small amount. Further examination of intestinal segments showed that a large amount of message existed in the ileum and a trace amount in jejunum, but not in duodenum (Fig. 6b). No difference was found in the levels of the message between upper and lower portions of esophagus. The message was exclusively expressed in the esophageal mucosa (Fig. 6c).
To ascertain that the 4.6-kb transcripts present in esophagus and testis were identical to that in intestine, the size of transcripts in these tissues was analyzed by RT-PCR. Because the full-length mRNA was too large to be amplified by PCR, we amplified two fragments from the message separately; one corresponded to repeats 1 and 2, and the other to repeats 3 and 4 (Fig. 7) as described under "Experimental Procedures." Since the amount of messages in esophagus and testis seemed to be less than 2% of that in the ileum, poly(A)-rich fractions prepared from the former two tissues were used for reverse transcription. As shown in Fig. 7 (a and b), single bands of the expected sizes (2164 or 2109 bp, respectively) in all three tissues were detected at the identical position. To eliminate a possibility that these PCR bands were amplified from genomic DNA contaminated into the RNA fractions, a negative control reaction was performed with 1 g of rat genomic DNA. No specific DNA fragment was amplified from genomic DNA under these conditions (data not shown). The PCR products were cloned into plasmid vectors, and their nucleotide sequences were determined up to about 300 bp from both the termini. The cDNAs of esophagus and testis were found to have sequences identical to that of the intestine cDNA. Furthermore, digestion of the PCR products of all tissues by restriction enzymes gave the identical fragment patterns on agarose gel electrophoresis (Fig. 7). These results strongly suggest that the 4.6-kb transcripts expressed in esophagus and testis are identical to the intestinal PLB/LIP mRNA.
Next, we examined levels of protein expression in the mRNA-expressing tissues by immunoblot analysis using SDS-PAGE under the reducing conditions. As shown in Fig. 8, the immunoblot of freshly prepared ileac mucosa preparations revealed at least three strongly stained bands corresponding to the molecular masses of ϳ200, 130, and 90 kDa. PLB/LIP was not detectable in testis homogenate, but a faint band of ϳ200 kDa, which is the same as the largest one of the three bands found in ileum homogenate, was detectable in membrane fractions of testis (Fig. 8). When membrane fractions of sperm collected from epididymis were used for the analysis, the single band appeared more clearly. This band was not detectable using preimmune serum (not shown), suggesting that the PLB/ LIP protein is expressed in the membrane fractions of sperm (and its precursor). Comparison of the sizes of PLB/LIP in freshly prepared extracts of the ileum, sperm membrane, and COS-7 cells suggested that proteolytic cleavage was considerably limited in the latter two extracts.
We investigated the localization of PLB/LIP in the ileum and testis by light microscopy using anti-PLB/LIP IgG. In the ileum, absorptive cells located in the apical region of villi were diffusely but strongly stained by anti-PLB/LIP IgG (Fig. 9a). Examination on higher magnification showed that the brush border membranes were more intensely stained than the cytoplasm (Fig. 9c). No immunopositive reactions were observed in FIG. 3. Constructions used in expression of rat intestinal PLB/LIP. a, oligonucleotide primers used for PCR (hemi-arrows), and restriction enzyme sites are indicated along an outlined structure of pSVL-RIPLB, and procedures of preparing pSVL-Nhead are summarized: box with thick diagonal stripes, the signal peptide; box with thin diagonal stripes, the hydrophobic stretch. F and R indicate sense and antisense primers, respectively. b, sequences of the primers. The restriction enzyme sites connected with the 5Ј-end of each primer are underlined. c, expression vectors were prepared using pSVL-Nhead and pSVL-RIPLB as described under "Experimental Procedures." control sections stained with the antibody preabsorbed with E. coli extracts containing the PLB/LIP fragment as antigen (Fig.  9b). Secretory granules of Paneth cells were also stained (Fig.  9d). In testis, spermatocytes and spermatids were stained with the antibody, being visualized as round and crescent bodies, respectively (Fig. 9e), but not with preimmune serum (Fig. 9f). The combination of the immunofluorescent and phase-contrast figures showed that PLB/LIP proteins may be localized in the acrosome.

DISCUSSION
Using a probe amplified by RT-PCR based on the knowledge of partial amino acid sequences of the previously purified PLB/ LIP (9), we isolated and sequenced its full-length cDNA. The NH 2 -terminal amino acid sequences of the small and large fragments of purified PLB/LIP were found in complete accordance with the second repeat part of the deduced amino acid sequence. COS-7 cells transfected with the full-length cDNA produced the active enzyme that exhibited the same substrate specificity, i.e. PLA 2 , lysophospholipase, and lipase activities, as the purified enzyme (9). These results demonstrated that the isolated cDNA indeed encodes the functional rat intestinal PLB/LIP.
The amino acid sequence of PLB/LIP showed 67.9% similarity to that of rabbit AdRab-B protein that had been isolated from an adult intestine specific cDNA library (21). Both proteins had the primary structural arrangement similar to lactose-phlorizin hydrolase, a well characterized hydrolase associated with brush border membranes: a signal peptide, four internal repeats, and a COOH-terminal transmembrane anchor. This suggests AdRab-B is a rabbit counterpart of rat PLB/LIP, although their substrate specificities were different from each other (9). Pind and Kuksis (24) reported a purifica-  5. The second repeat is a catalytic domain. a, Northern blot analysis of PLB/LIP mRNA extracted from cells transfected with the same constructs as in Fig. 3c. Upper panel, total RNA (20 g) was electrophoresed, blotted onto a nylon membrane, and hybridized with the [ 32 P]dCTP-labeled PLB/LIP cDNA fragment that had been prepared by PCR amplification from the pMOS-RIPLB3 plasmid with Nfor and R4 primers (corresponding to almost full-length RIPLB, see Fig. 3a). Lower panel, ribosomal RNAs were stained by ethidium bromide. b, lipolytic activities in each culture medium were tested as in Fig. 4b. As a positive control, we used the culture medium of COS-7 cells transfected with pSVL-⌬C, because the expressed full-length enzyme was not secreted into the medium. Only the lipase activity was detected in this medium in the absence of DTT as in Fig. 4b.   FIG. 6. Distribution of PLB/LIP mRNA in various rat tissues. Northern blot analysis was performed using a BamHI fragment of pCR-RIPLB (upper panels) or an actin cDNA fragment (a, lower panel) as a probe. In a and b, each lane contained 30 g of total RNA, and in c 20 g of total RNA. b, the small intestine was divided into four parts: duodenum, jejunum, and upper and lower ileum. c, the esophagus was cut into the upper and lower halves, and total RNA was extracted from the whole tissue. In some cases, the mucosal layer was scraped off with forceps after the tissue was inverted, and total RNA was extracted from the scrapings and the remaining part containing muscle and adventitia. tion of detergent solubilized PLA 2 /lysophospholipase from rat intestinal brush border membranes. The use of SDS-PAGE at the final purification step gave an inactive enzyme, which renatured under appropriate conditions. That enzyme may be identical to PLB/LIP, because the amino acid composition reported was similar to that calculated from the amino acid sequence deduced in this study.
The putative signal sequences of rat PLB/LIP and AdRab-B were similar to each other. The method of von Heijne predicted cleavage between Gly 19 -Ala 20 for AdRabB (21); in rat PLB/LIP, the corresponding cleavage at Gly 21 -Pro 22 generates a new NH 2 -terminal Pro, which may make this cleavage unlikely, because the occurrence of Pro at the NH 2 terminus is extremely rare. Other candidates for a signal sequence cleavage site are the peptide linkage between residues 28 and 29 or that between residues 29 and 30. It, however, remains to be clarified which After purification by phenol extraction, aliquots of the former products were digested with EcoRI or BclI, while those of the latter were digested with XbaI or BamHI (see Fig. 3). The digests were separated on a 1.2% agarose gel. All digestions gave two fragments of the following lengths: 796 and 1369 bp for EcoRI digestion, 582 and 1582 bp for BclI digestion, 870 and 1239 bp for XbaI digestion, 570 and 1541 bp for BamHI digestion. The sizes of markers are: 23.13, 9.24, 6.56, 4.36, 2.32, 2.02, and 0.56 kb for /HindIII; 1419, 517, 396, 214, and 75 bp for pUC19/ HinfI. Asterisk denotes a nonspecific product, which was also amplified from rat genomic DNA.
FIG. 8. Immunoblot analysis of PLB/LIP in the ileum, testis, and sperm. Ileum (3 g) and testis (20 g) homogenates and sperm membrane fractions (20 g) were subjected to SDS-PAGE on a 12% gel, blotted to a PVDF membrane, and detected with 500-fold diluted anti-PLB/LIP serum. Prestained proteins (New England Biolabs) were used as size markers. The results presented in this and the preceding paper (9) demonstrated that the repeat 2 is the catalytic domain that catalyzes PLA 2 , lysophospholipase, and lipase activities. Comparable inhibition of these activities of purified (9), expressed full-length and repeat 2 enzymes by DFP suggested the involvement of a single active site with broad substrate specificity. This raises further questions about the functional roles of the other repeats. The primary structural similarity between repeat 2 and the other repeats, especially repeats 3 and 4 (Fig.  2c), implies the presence of another active sites with different specificity in these repeats, like lactase-phlorizin hydrase, which has the same structural arrangement as PLB/LIP and two distinct active sites in homologous repeat 3 and 4 (25). However, we have not yet obtained direct evidence for this. Alternatively, repeats other than repeat 2 may play roles in regulating the enzyme action of repeat 2. To address these problems, we will need the structural and functional identification of active site catalytic residues.
In rabbit lactase-phlorizin hydrolase, repeats 1 and 2 serve as parts of the propeptide that is intracellularly removed to give the mature brush border membrane form. Its precursor protein is not a zymogen but by itself is as active as the mature enzyme. In rat PLB/LIP, the intestinal enzyme was partially degraded into proteins in size ranging from 90 to 200 kDa, but the testis enzyme has the same size as the full-length PLB/LIP expressed in COS-7 cells, suggesting proteolytic cleavage did not occur in that tissue. The similar cleavage has been reported in rabbit intestinal AdRab-B (21). It is unknown at present whether this proteolysis merely reflects greater abundance of proteases in the intestinal mucosa than in the testis or whether the difference in the degree of degradation is relevant to tissuespecific roles of this enzyme. The full-length PLB/LIP is again not a zymogen because it exhibited as high a lipolytic activity as the enzyme purified from rat intestine when enzyme concentrations were estimated by immunoblot analysis (Fig. 4a).
We purified the 35-kDa catalytic domain of PLB/LIP with a nick at the peptide linkage between Arg 528 and Phe 529 that was released from brush border membrane by the action of endogenous protease(s). Although the NH 2 -terminal amino acid sequences of its small and large fragments were included in the second repeat, its exact COOH-terminal processing site remained to be clarified. PLB/LIP has 14 potential N-glycosylation sites; repeat 2 has a single site at Asn 696 near its COOHterminal end. The results of concanavalin A-peroxidase stain on PVDF membrane consistently indicated that the COOHterminal large peptide of the purified PLB/LIP was indeed glycosylated (9). The apparent molecular mass of the large peptide was about 21 kDa, which agreed with that calculated from amino acid sequence data (20.7 kDa), assuming that the processing occurs at the peptide bond just after the Lys 710 that is present in the COOH-terminal K(R)NS sequence conserved among all repeats (Fig. 2a). These considerations suggested that the processing site of the purified enzyme is present at the COOH-terminal side of Asn 696 .
In a previous study on rabbit AdRab-B, it was inferred by analogy with lactose-phlorizin hydrolase that the COOH-terminal hydrophobic domain is responsible for its membrane anchoring (21). To address the problem experimentally, we created a PLB/LIP mutant that lacks the COOH-terminal 42 amino acids including the hydrophobic stretch. The wild-type PLB/LIP was mainly found in the membrane fractions of COS-7 cells, whereas the truncated mutant was released into the culture media. Triton X-100 treatment, but not phosphatidylinositol-specific phospholipase C treatment, efficiently solubilized PLB/LIP from COS-7 cell membrane and brush border membrane fractions. These results provide evidence that the COOH-terminal hydrophobic domain does not serve as a signal to direct glycophospholipid attachment, but directly as a membrane anchor. Since there exist phospholipase C-resistant GPI-anchored proteins (26), conclusive proof for the enzyme's anchoring mechanism must, however, await structural determination of detergent-solubilized enzyme.
Surprisingly, virtually no PLA 2 and lysophospholipase activities were detectable in the culture medium of the cells expressing the deleted mutant-enzyme, despite the fact that both lipase activity and immunoreactivity were appreciably detectable. DTT treatment of the mutant, however, restored the lost activities. This can be explained as follows. The interaction between repeat 2 and another repeat(s), presumably via disulfide bridge formation, might cause the truncated mutant to undergo a conformational change, leading to selectively decreasing the binding affinity for and/or catalytic efficiency to phospholipids, of which the binding mode to the enzyme (9) and the surface quality (1) appear to differ from those of triacylglycerol. It is clear that further detailed kinetic and structural studies are required to solve this problem.
A recent study indicated that the COOH-terminal tail of lactase-phlorizin hydrolase contains the consensus sequence for the phosphorylation site by protein kinase A, and this site was indeed phosphorylated by protein kinase A in vitro and in tissue culture (27). This confirmed the intracellular disposition of this tail and suggested its possible physiological significance. In rat PLB/LIP, the corresponding tail does not contain the consensus sequence for a protein kinase A site. The intracellular regulation mechanism mediated by protein kinase A may not operate in the rat PLB/LIP. Northern blot and RT-PCR analyses suggested that PLB/LIP mRNA was expressed not only in ileal mucosa but also in esophageal mucosa and testis (Figs. 6 and 7). However, rabbit AdRab-B mRNA was not detected in the testis. This may be due to species difference or due to the difference in sensitivity of the method used. The RT-PCR products of these tissues were found to be identical to one another by partial sequencing and restriction fragment length analysis. Immunoblot and immumohistochemical analyses provided further evidence for the localization of PLB/LIP protein (Figs. 8 and 9). In the ileum, a large amount of PLB/LIP existed in the brush border membrane on the apical side of villi; in contrast, secretory PLA 2 s are synthesized in and secreted from cells in the bottom regions of gastrointestinal mucosa: gastric chief cells for group I PLA 2 (28) and Paneth cells for group II PLA 2 (29). A similar localization of rabbit AdRab-B has been reported (21). These results support the idea that PLB/LIP acts on the brush border membrane to facilitate the absorption of digested lipids.
Finally, immunohistochemistry showed that, in addition to the absorptive cells, Paneth cells and acrosomes of spermatids were immunopositive, but to a lesser extent, to anti-PLB/LIP antibody, suggesting PLB/LIP's roles other than intestinal digestion. Paneth cells contain a variety of antibacterial materials including group II PLA 2 , which can kill certain strains of intestinal bacteria directly or with the aid of neutrophil bactericidal/permeability-increasing protein (29 -31). PLB/LIP with broad specificity toward polar head groups of phospholipids actively hydrolyzes phosphatidylglycerol, the major component of bacterial cell membrane phospholipids, like group II PLA 2 (32,33). It is interesting to examine whether PLB/LIP in Paneth cells participates in an antibacterial defense mechanism and where it is destined for after synthesis, since Paneth cells lack brush border membranes.