- Email: [email protected]
Characteristics of rat pancreatic regenerating protein
Characteristics of rat pancreatic regenerating protein Michael E. Zenilman, MD, Jian Chen, MD, Babak Danesh, MD, BA, and Qing-hu Zheng, MS, Bronx, NY
Background. The pancreatic regenerating (reg) gene is an acinar cell product involved in islet formation and maintenence. Human reg protein is mitogenic to pancreatic beta and ductular cells, and its amino acid sequence predicts it to be a calcium-dependent lectin. Methods. We studied the biologic activity and cellular localization of rat reg I isolated from the acinar cell line AR42J and the lectin properties of reg from AR42J and pancreatic juice. Bioactivity was assayed by mitogenesis on the ductular cell line ARIP. Cellular localization was determined by differential centrifugation. Lectin properties were assessed by affinity chromatography. Results. Reg protein from crude AR42J cellular lysate and purified reg protein from AR42J induced thymidine incorporation to ARIP. Conditioned medium from AR42J and co-culture of AR42J with morphologically distinguishable ARIP, however, failed to induce mitogenesis. Reg protein was localized within the vesicle fraction of the cell and was not membrane bound. Affinity chromotography revealed that reg protein did not bind to mannose or galactose in the presence or absence of calcium. In pancreatic juice a previously undescribed mannose-binding protein was discovered at 25,000 to 30,000 daltons. Conclusions. We conclude that reg produced in the acinar cell line AR42J is biologically active but not efficiently secreted, even though it localized within the cellular vesicles. Despite predictions based on its amino acid sequence, it does not appear to be a calcium-dependent lectin. (Surgery 1998;124:855-63.) From the Department of Surgery, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY
PANCREATIC REGENERATING PROTEIN (reg I) is an acinar cell product involved in islet regeneration. We have shown the human homolog to be mitogenic to rat pancreatic-derived cells: the beta cell line RIN, the duct cell line ARIP, and primary cultures of duct cells.1,2 Reg I protein was originally isolated from pancreatic juice, and we have recently shown that reg I gene is expressed in the rat acinar cell line AR42J. Its genetic expression can be modulated with dexamethasone.3 Western analysis has revealed that AR42J also produces reg I protein,1 and we have recently isolated the protein from cellular lysates.4 AR42J may therefore be a useful cell line for the study of rat reg I expression, translation, and secreSupported by an American College of Surgeons Faculty Fellowship, the Diabetes Research and Training Institute of Albert Einstein College of Medicine, and National Institutes of Health grant No. RO1 DK54511-01. Presented in part at the Annual Meeting of the American Gastroenterological Association, May 20, 1996, San Diego, Calif. Accepted for publication March 27, 1998. Reprint requests: Michael E. Zenilman, MD, Department of Surgery, Albert Einstein College of Medicine, Room 2S-5, 1825 Eastchester Rd, Bronx, NY 10461. Copyright © 1998 by Mosby, Inc. 0039-6060/98/$5.00 + 0
11/56/90945
tion, and it is important to know whether the protein produced is biologically active. The basic function and mechanism of action of reg I protein is presently unknown. The amino acid sequence analysis predicts it to be a calciumdependent lectin. The protein contains the Drickamer domains found in many such lectins,5 and its size is similar to that of the mannose-binding proteins and mannose-binding regions of the asialoglycoprotein receptor complex.6-8 The reg I amino acid sequence has significant homology to the mannose-binding regions of these domains, as described previously.9 Lectins are involved in formation of extracellular matrix, antigenic recognition between cells, cell-to-cell adhesion, and cellular differentiation. They may also have mitogenic properties. To date, however, no experimental data exist showing that reg I is in fact a calciumdependent lectin. In this study we examined the characteristics of rat reg I protein and the cellular biology of reg I in AR42J. First we studied the mitogenic effect of reg I protein found in AR42J, in both crude and pure isolates. For these studies, mitogenesis to the ductal cell line ARIP was used as a bioassay for reg I activity. We then studied whether reg I protein is secreted in a biologically active form from AR42J, using conditioned medium and co-culture techniques. Cellular SURGERY 855
856 Zenilman et al
localization studies were then performed on AR42J cells to determine which cellular compartment contains reg I protein. Specifically, if it were associated with the asialoglycoprotein complex, it would localize to the membrane portion of AR42J. Finally, we performed carbohydrate-binding studies of reg I isolated from AR42J and pancreatic juice. As a lectin, we predicted that the protein would bind carbohydrate columns in a calcium-dependent fashion.8 MATERIAL AND METHODS Cell culture. ARIP, a rat ductular cell line, and AR42J, a rat acinar cell line, were purchased from American Type Culture Collection (Rockville, Md). Cells were cultured in Dulbecco’s modified Eagle’s medium with 5% fetal calf serum. Isolation of rat reg I. Rat reg I was isolated from AR42J cellular lysates and pancreatic juice. Crude preparations of reg I protein were obtained from protein lysates of AR42J by sonication of 105 cells in 1 mL, centrifugation at 10,000g, and resuspension in 1 mL saline solution with 1 µg/mL aprotinin, for a final concentration of 5 µg/mL. Purified reg I protein from AR42J was isolated from 2 g AR42J, from the third 70% ammonium sulfate precipitation as described for pancreatic thread protein.10 We have previously shown that such isolates yield pure preparations of reg I protein.4 Pancreatic juice was isolated from 350 g breeder rats (Harlan Sprague Dawley, Indianapolis, Ind) as follows. After 50 mg/kg intraperitoneal anesthetic with sodium pentobarbital (Nembutal), a midline incision was performed. The common bile duct was identified and cannulated in an antegrade direction with PE-10 tubing (Fisher Scientific, Pittsburgh, Pa) such that the proximal end of the tube was beyond the choledochotomy and the distal end passed the ampulla of Vater in the duodenum. The bile duct was then ligated to prevent flow of bile, the duodenum was opened, and the distal tube was used for collection of pancreatic juice. Pancreatic flow was stimulated by intravenous injection of 1 unit/kg secretin (Sigma Chemical Co, St Louis, Mo), and juice was collected in the presence of aprotinin (Sigma Chemical Co), with a final concentration of 1 µg/mL. From each animal, 200 µL was typically collected for 2 hours. Mitogenesis. Bioactivity of reg I derived from AR42J was assayed by mitogenesis on the ductular cell line ARIP. One to 2 × 105 cells were plated in Dulbecco’s modified Eagle’s medium with fetal calf serum. Thymidine incorporation was assayed by scintillation 48 hours after treatment as described previously.1 Each dose was performed in triplicate and each experiment was performed at least 3 times.
Surgery November 1998
To determine if reg I was secreted from AR42J, conditioned medium from AR42J was obtained by incubating 106 cells in 3% bovine serum albumin for 24 hours at 37 o C. ARIP cells were then exposed to conditioned medium and subjected to thymidine incorporation. Co-culture of AR42J with ARIP was also performed. For these experiments, 106 ARIP was cultured with or without 104 AR42J for 48 hours in a 75-mm Petri dish. Cells were then treated with bromodeoxyuridine (1:1000 dilution; 10 µmol/L final concentration) (Amersham Corp, Arlington Heights, Ill) for 2 hours and fixed in 70% ethanol. Cells were then stained with monoclonal antibodies to bromodeoxyuridine (Amersham Corp), washed with phosphate-buffered saline solution (pH 7.4 + 0.5% bovine serum albumin + 0.5% Tween), and labeled with goat antimouse immunoglobulin G conjugated to fluorescein-5-isothiocyanate (FITC) (Sigma Chemical Co). All nuclei were then stained with 1 µg/mL propridium iodide (Sigma Chemical Co).11 When viewed under an Olympus IMT-2 inverted microscope and subjected to excitation with an EY455-nm filter, the propidium iodide–labeled nuclei fluoresced red and the FITC-labeled bromodeoxyuridine green. Under phase contrast, ARIP cells are morphometrically distinct from AR42J. The percentage of ARIP nuclei positive for bromodeoxyuridine was calculated from at least 50 nuclei within at least 5 separate colonies per experiment, according to point-counting techniques.12 These values were compared with control experiments in which ARIP cells were cultured alone. Each experiment was performed at least 3 times per dose along with zero controls. Cellular localization. Analysis of cellular fractions was performed by differential centrifugation according to a variation of the methods described13,14; 106 AR42J cells were harvested and washed with phosphate-buffered saline solution (0.1 mol/L sodium phosphate and 0.15 sodium chloride; pH 7.2) twice and then lysed in 1 mL hypotonic lysis buffer (1 mmol/L NaHCO3, 5 mmol/L MgCl2, 100 µmol/L phenylmethylsulfonyl fluoride, 10 µg/mL leupeptin, and 10 µg/mL soybean trypsin inhibitor) at 4°C for 2 minutes. The cell lysate was sonicated for 5 minutes and transferred into a microphage tube. Nuclei and residual cells were spun down at 500g for 5 minutes at 4°C. The supernatant, containing cellular vesicles and residual membranes, was centrifuged at 15,000g at 4°C for 1 hour. The supernatant, containing cell membranes, was saved. The pellet, containing vesicles (and residual membranes), was washed with
Surgery Volume 124, Number 5
phosphate-buffered saline solution 3 times and resuspended in 500 µL 1% sodium dodecylsulfate and heated at 50°C for 1⁄2 hour. This was frozen and thawed 3 times to break up the vesicles and then spun at 100,000g for 20 minutes at 4°C (Ultra-centrifuge; Beckman, Fullerton, Calif) to separate vesicular membranes (and residual cellular membranes) from vesicular contents. Samples were concentrated 5 times by dialysis against polyethylene glycol, and 20 µL was then run on a 15% polyacrylamide minigel (Biorad, Hercules, Calif). Protein characteristics. If reg I is a calciumdependent lectin, it should selectively bind a carbohydrate-affinity column.8 To determine the calcium-binding lectin characteristics of reg I protein, a 0.5 mL column of mannose or N-acetyl-D-galactosamine cross-linked to 4% agarose beads (Sigma Chemical Co) was used at 4o C. After washing with 10 × vol of 10 mmol/L Tris (pH 7.4), 15 mmol/L CaCl2, 1.25 mol/L NaCl, 0.02% NaN3, and 0.5% bovine serum albumin, 100 µL pancreatic juice or reg I isolated from AR42J was added to the column and allowed to bind overnight, and the column was washed. The column was eluted with 10 mmol/L Tris (pH 7.4), 2 mmol/L ethylenediaminetetraacetic acid (EDTA), 1.25 mol/L NaCl, 0.02% NaN3, and 0.5% bovine serum albumin. A 2-mol/L galactose or mannose solution was also used after the EDTA elution. The eluates were concentrated 10 times by ultrafiltration, and 10 µL aliquots from each run were subjected to polyacrylamide gel electrophoresis (15%) and silver staining. Reg I antibody. We produced a polyclonal antibody in guinea pigs (Harlan Sprague Dawley) to a 10 amino acid peptide portion of rat reg I protein linked to a keyhole limpet hemocyanin backbone (sequence YFMEDHLSWA, residues 48 to 57 from the N-terminus of the first methionine amino acid or residues 26 to 35 from the N-terminal methionine of the secreted protein) (Research Genetics, Inc, Huntsville, Ala). The sequence is unique to rat reg I (50% homology to human reg I, 10% homology to rat reg III, and 50% homology to rat reg II). The antibody was characterized by Western analysis to rat pancreatic juice and reg I purified from AR42J, described below. Western analysis. After confirmation of protein in the gel by Coomassie Brilliant Blue staining or Silver staining (Biorad), protein was transferred to nitrocellulose electrophoretically (0.45 µm; Micron Separations, Inc, Westboro, Mass) at 100 V for 1 hour. The transfer membrane was incubated in block solution (10% bovine serum albumin in 0.01 sodium phosphate, 150 mmol/L NaCl, and 0.1% thimerosal; pH 7.4) (Pierce, Rockford, Ill) at
Zenilman et al 857
4°C overnight. The membrane was then washed with Tris-buffered saline solution (TBS) for 5 minutes 3 times and incubated with polyclonal guinea pig antirat-reg I antibody, produced in our laboratory, at 1:1000 dilution at room temperature for 1 hour. The membrane was washed in TBS for 5 minutes, 3 times, and incubated with the second antibody goat anti-guinea pig immunoglobulin G alkaline phosphatase (Sigma Chemical Co) at 1:15,000 dilution at room temperature for 30 minutes. The membrane was washed with TBS for 5 minutes 3 times. The blot was developed according to the 1step technique with nitroblue tetrazolium/BCIP substrate (Pierce). RESULTS Mitogenic effect of reg I protein isolated from AR42J. Reg I protein was isolated from AR42J as described above, and purity was confirmed by polyacrylamide gel electrophoresis.4 In initial experiments, crude and pure isolates of reg I isolated from AR42J were used in experiments on ARIP cells. Mitogenesis was assessed by tritiated thymidine incorporation. Results from typical experiments are shown in Fig 1. A dose-related effect of both cellular lysates of AR42J (which contain reg I) and reg I purified from AR42J on ARIP thymidine incorporation was noted. Secretion of reg I from AR42J. Secretion of reg I from AR42J was studied with ARIP mitogenesis used as a bioassay. In the first series of experiments, ARIP cells were subjected to conditioned medium from AR42J for 48 hours and then pulsed with tritiated thymidine. No effect was noted with ARIP thymidine incorporation. In a typical experiment, treatment of ARIP with control medium followed by thymidine yielded a value of 5094 ± 365 cpm, 25% conditioned medium yielded 5618 ± 488 cpm, treatment with 75% conditioned medium yielded 5228 ± 1330 cpm, and treatment with 100% conditioned medium yielded 5162 ± 470 cpm (difference not significant). One hundred percent conditioned medium was then concentrated 10 times by dialysis and subjected to electrophoretic analysis (15% polyacrylamide gel electrophoresis and visualization by silver staining), which revealed a weak band consistent with amylase (molecular weight about 50,000 daltons) but no band at 16,000 or 18,000 daltons, indicating that significant quantities of reg I were not present (data not shown). A series of experiments was then performed to determine whether co-culture of AR42J cells with ARIP cells induced mitogenesis in ARIP. We surmised that local concentrations of secreted reg I might be higher in the region surrounding
858 Zenilman et al
Surgery November 1998
Fig 1. A, Effect of cellular lysate of AR42J on ARIP thymidine incorporation. Dose-related effect is seen with increasing concentrations of lysate, whose concentration was adjusted to 5 µg/mL (+1 µg/mL aprotinin). B, Purified reg I protein from AR42J was mitogenic to ARIP cells in dose-related fashion. *P < .05 compared with control, analysis of variance.
Fig 2. Co-culture of ARIP and AR42J. Under phase contrast, ARIPs are typically flat and discernable from AR42J because latter grow in clumps. After 48 hours of co-culture, cells were pulsed with bromodeoxyuridine (BrdU). Cells were stained with FITC-labeled antibromodeoxyuridine antibody, and all nuclei were stained with propidium iodide. Nuclei of AR42J are seen as clusters, surrounding light gray nuclei are all ARIP, and black single nuclei are ARIP positive for bromodeoxyuridine. Percent ARIP staining positive for bromodeoxyuridine was calculated by point counting and compared with pure ARIP cultures (color photos available on request). (Original magnification × 200.)
AR42J, and this should be a more reliable method of showing a mitogenic effect on the ARIP cell. Under phase contrast microscopy, AR42J cells are distinguishable from ARIP, and differential calculation of bromodeoxyuridine
incorporation by each group of cells is possible. Results from a typical experiment are shown in Fig 2. No difference was found in bromodeoxyuridine uptake between plates containing ARIP cells alone (5.49% ± 0.30%) and plates con-
Surgery Volume 124, Number 5
Zenilman et al 859
Fig 3. A, Western analysis of pancreatic juice (PJ) and AR42J lysate with polyclonal antibody to 10 amino acid peptide fragment of rat reg I. Band is seen at 16,000 daltons, consistent with reg I. B, Inhibition of Western blot by reg I 10 amino acid peptide. Pure reg I isolated from AR42J was subjected to standard Western analysis. When 2 mg/mL peptide is present in solution containing polyclonal antibody, band on Western is inhibited. Similar data were observed if reg I protein was used to block experiment.
taining both ARIP and AR42J (4.51% ± 0.38%; difference not significant). These data strongly suggest that reg I is not secreted by AR42J cells in a concentration high enough to affect ARIP in our bioassay, and maybe not at all. Antibody characterization. A polyclonal antibody was produced in guinea pigs to a unique 10 amino acid fragment of reg I. Western analysis revealed that the antibody isolated reacted as a single band at 16,000 daltons in pancreatic juice, AR42J cellular lysate, and purified reg I from AR42J, all consistent with rat reg I (Fig 3). It did not react with human reg I (not shown). Specificity of the antibody to rat reg I was shown by blocking the antibody during Western analysis with either pure peptide (Fig 3, B) or pure reg I protein isolated from AR42J (not shown). A competitive enzyme-linked immunosorbent assay was constructed that easily detected reg I in pancreatic juice, rat serum, and AR42J lysate. Interestingly, very little to no reg I (above background) was detected in conditioned medium from AR42J. This was consistent with our hypothesis that the protein is not secreted from this cell line. Cellular localization studies in AR42J. To determine where reg I protein localized within the AR42J cell, cytoplasmic, vesicular, and membrane compartments were isolated as described in Material and Methods. Isolates at each step were subjected
to sodium dodecylsulfate–polyacrylamide gel electrophoresis and then transferred to nitrocellulose and analyzed by Western analysis for reg I. Fig 4 shows the 15% polyacrylamide gel and Western analysis of the cytoplasmic, whole, and disrupted vesicle portions of the analysis. From this analysis it appears as if reg I localized to the lanes that contained whole vesicles and their contents, not the cytoplasmic or membrane portions. Western analysis of a similar gel that used a polyclonal antibody to amylase showed that amylase localized to the same isolate as reg I (data not shown) Lectin properties of reg I. Reg I protein was assayed for calcium-dependent carbohydrate binding by affinity chromatography. Solutions containing reg I were loaded on mannose-sepharose or galactose-sepharose columns in the presence of calcium and subsequently eluted in a calcium-free buffer (with EDTA), followed by elution in a buffer with 2 mol/L carbohydrate. Two separate preparations of reg I were used: pancreatic juice and the protein isolate from AR42J cells. Fig 5 shows the result of affinity chromatography of pancreatic juice on a galactose and mannose column. No binding was noted on the galactose column. During the elution of the mannose-affinity column, a band between 25,000 and 30,000 daltons was liberated. Although this is in the region of glycoproteins 2 and 3, chymotrypsin and proelastase,15 to date such a lectin in pancreatic juice has
860 Zenilman et al
Surgery November 1998
Fig 4. A, Fifteen percent polyacrylamide gel electrophoresis (PAGE) stained with Coomassie blue of AR42J celluar compartments. B, Western analysis of same gel with polyclonal reg I antibody. Lanes a and b depict product of separation of cytoplasmic contents (a) from vesicles and residual membranes (b). Lane c is vesicle membrane fraction after freeze thawing and ultracentrifugation, and lane d is vesicle content fraction. Reg I antibody hybridized at approximately 16,000 daltons in vesicular and vesicular lysate fractions alone. (MW, Molecular weight marker.)
not been described. Nothing was liberated in the 15,000- to 18,000-dalton range, the size of reg I. When reg I protein isolated from AR42J was subjected to galactose-sepharose or mannose-sepharose affinity chromatography in a manner similar to that above, no binding was noted (not shown). Examination of the carbohydrate-linked sepharose beads after treatment with both preparations of reg I under phase contrast microscopy followed by treatment with anti-reg I antibody and viewed with FITC-labeled second antibody also failed to show any binding to the beads (not shown). It was found, however, in the unbound fraction by both Western analysis and competitive enzyme-linked immunosorbent assay with the polyclonal antibody described above (not shown). These results indicate that reg I had no affinity to either mannose or galactose. DISCUSSION This is the first study that describes both protein analysis and cellular physiology of rat reg I protein. Although the reg I gene was originally described in 1988,16,17 its amino acid sequence has been shown to be identical to that of pancreatic stone protein (lithostatine), a pancreatic secretory product described originally in the 1970s.18 Lithostatine is believed to be important in inhibiting calcium carbonate precipitation,16,18-20 and decreased concentrations in pan-
creatic juice are believed to be associated with the development of chronic calcific pancreatitis. The reg I gene is induced by beta cell regeneration after 95% pancreatectomy followed by nicotinamide administration and during beta cell hyperplasia after pancreatic wrapping.21,22 Its mRNA expression is inhibited during beta cell suppression.23 We have recently shown that reg I mRNA and protein is found in the rat acinar cell line AR42J, and the protein can be isolated from AR42J lysates.4 The mechanism of action of this exocrine product on the pancreatic endocrine tissue is unknown. Reg I protein is mitogenic to pancreatic-derived cells. We have shown that the human homolog of reg I induces thymidine incorporation into a rat ductular and beta cell line1 and causes increased incorporation of bromodeoxyuridine in isolated rat duct cells.2,4 Other investigators have shown that exogenous administration of recombinant reg I protein can reverse diabetes after pancreatic resection, and it is mitogenic to only beta cells within the islet. 24 We believe that the protein acts as a stimulus for growth of both the native beta cell population and ductal precursor cells of the pancreas and may even induce the latter’s differentiation into beta cells. The mechanism of how the protein induces mitogenesis is unknown. Although much has been reported about reg I
Surgery Volume 124, Number 5
Zenilman et al 861
Fig 5. Affinity chromotography of pancreatic juice (PJ) followed by 15% polyacrylamide gel electrophoresis and silver staining. A, Results from mannose-sepharose column. B, Results from galactose-sepharose. Results revealed that reg I protein did not bind to mannose or galactose in presence of calcium, indicating that in mature form reg I is not calcium-dependent lectin. Previously undescribed mannose-binding lectin in pancreatic juice is noted at 25,000 to 30,000 daltons of panel A. (MW, Molecular weight marker; run, runthrough after loading; wash, last wash with buffer; EDTA, 2 mmol/L EDTA elution; man or gal, 2 mol/L mannose or galactose elution.)
protein in experiments focusing on pancreatic stone formation, other basic biochemical and cellular studies are lacking. The amino acid sequence analysis of reg I predicts it to be a calcium-dependent lectin, and its size is similar to the mannose-binding proteins and mannose-binding domains of the asialoglycoprotein receptor complex. 6-8 Reg I contains the Drickamer domains found in many such lectins.5 Specifically, comparison of the rat reg I protein sequence with rat mannose–binding domains9 has shown conservation of most of the 32 conserved sequences, including the important WIGL amino acid sequence (aromatic-aliphatic-glycine-aliphatic), a hydrophobic packing common to all carbohydrate-recognition domains (unpublished observations). 9 Animal lectins are important in cellular physiology; they are involved in formation of the extracellular matrix, antigenic recognition between cells, cellto-cell adhesion, and cellular differentiation. They may also have mitogenic properties, as evidenced by relationships with the epidermal growth factor receptor.8,9 If reg I is a lectin, this would shed light on the mechanism by which it is mitogenic to the pancreatic-derived cells. The first series of experiments focused on reg I protein produced in the rat acinar cell line AR42J. Because AR42J is able to produce reg I and has been shown to be a useful model for studying reg I genetic expression, is the reg I produced in AR42J biologically active and, if so, is it secreted? Our results show that the reg I protein produced in AR42J is biologically active, in both crude cellular lysates of AR42J and pure reg I preparations. We showed this with a bioassay of mitogenesis in the duc-
tular cell line ARIP, a cell line that we have shown is stimulated in this fashion by human reg I.1 There seems to be a problem, however, with secretion of reg I from AR42J. Conditioned medium from AR42J failed to induce thymidine incorporation into ARIP, and electrophoresis of concentrated conditioned medium, as well as analysis by enzymelinked immunosorbent assay, did not reveal reg I. We have observed similar negative results when RIN 1046-38 cells are used (unpublished observations), a beta cell line also sensitive to reg I.1,25 Finally, direct co-culture of ARIP with AR42J did not induce ARIP cell growth, further indicating that, although reg I is produced in AR42J, it is not actively secreted into the surrounding medium. This is a surprising finding because reg I is normally secreted from the pancreatic acinar cells into pancreatic juice, in high concentrations.19,20 We next performed cellular-localization studies of reg I protein in AR42J. Amino acid sequence homologies raised the possibility that reg I is a member of, or a portion of, the asialoglycoprotein receptor family. Comparison of the reg I amino acid sequence to the 15,000-dalton carbohydrate-recognition domains of this giant receptor have shown similarities.6-8 Although reg I is found in pancreatic juice at high concentrations, could it be a product of a membraneassociated glycoprotein, or is it found in a secretory vesicle? Our cellular localization studies by differential centrifugation and Western analysis revealed that reg I localized within the vesicular component of the cell and is not membrane bound. Further Western analysis has shown that reg I resides in the same fraction as amylase, which is secreted from AR42J.26,27 It is therefore likely to be a normally secreted product
862 Zenilman et al
of the cell, not part of the asialoglycoprotein receptor or any other receptor complex. Although reg I is located within cellular vesicles, it is not secreted efficiently. This may be the result of a defect in the secretory mechanism of AR42J, a defect in the amino acid sequence of the signal peptide of AR42J reg I that directs it to a secretory vesicle, or that the protein is secreted in quantities too dilute to be detected by our assays. The latter is possible because even amylase is secreted in relatively small quantities by AR42J under basal conditions.27 Finally, we studied the lectin properties of reg I found in both pancreatic juice and AR42J. Our findings with affinity chromatography with galactoseand mannose-sepharose columns indicate that reg I does not bind carbohydrate. Reg I protein is not, in its mature form, a calcium-dependent lectin. This is in agreement with Iovanna et al,28 who found similar results with pancreatitis-associated protein, which is reg III,5 a homolog of reg I. These data suggest that, despite predictions from their amino acid sequence, the reg family is not a family of calciumdependent lectins. Alternatively, although it is known that human reg I is heavily glycosylated,29 it is possible that the carbohydrate-recognition domains were previously occupied in both samples tested. This should not be the case with rats, however, because some believe that rat reg I is not glycosylated.30 To address this question, we are producing recombinant reg I in a bacterial expression system and will subject the product to similar tests as described above to determine the lectin properties. An unexpected finding from the affinity-chromatography experiments is that of a previously undescribed mannose-binding lectin at 25,000 to 30,000 daltons, in the region of glycoproteins 2 and 3, chymotrypsin and proelastase. We are currently isolating this protein to determine its sequence. In summary, we have performed preliminary experiments characterizing rat reg I protein. We have shown that reg I produced in the acinar cell line AR42J is biologically active but not actively secreted. It is found in cellular vesicles and is not membrane bound. Despite predictions based on its amino acid sequence, it does not appear to be a calcium-dependent lectin. We acknowledge the technical help of Akiva Marcus, Ferrel Motlow, and Benjamin Gelman and thank Phil Stahl, PhD, of Washington University for helpful conversations regarding lectin properties and the mannose receptor.
Surgery November 1998
2.
3.
4. 5.
6.
7.
8. 9.
10.
11.
12. 13.
14.
15.
16.
17.
18.
19.
20.
REFERENCES 1. Zenilman ME, Magnuson TH, Swinson K, Egan J, Perfetti R,
Shuldiner AR. Pancreatic thread protein is mitogenic to pancreatic-derived cells in culture. Gastroenterology 1996;110:1208-14. Zenilman ME, Chen J, Magnuson TH, Shuldiner AR. Human PTP/reg is mitogenic to primary culture of pancreatic ductal cells [abstract]. Gastroenterology 1995;108:A402. Zenilman ME, Magnuson TH, Perfetti R, Swinson K, Shuldiner AR. Reg gene expression is inhibited during differentiation of cultured rat acinar cells. Ann Surg 1997;225:327-32. Zenilman ME, Chen J, Magnuson TH. Effect of reg protein on pancreatic ductal cells. Pancreas In press. Unno M, Yonekura H, Nakagawara K, Watanabe T, Miyashita H, Moriizumi S, et al. Structure, chromosomal localization and expression of mouse reg genes reg I and reg II. J Biol Chem 1993;268:15974-82. Drickamer K, McCreary V. Exon structure of a mannosebinding protein gene reflects its evolutionary relationship to the asialoglycoprotein receptor and nonfibrillar collagens. J Biol Chem 1987;262:2582-9. Taylor ME, Conary JT, Lennartz MR, Stahl PD, Drickamer K. Primary structure of the mannose receptor contains multiple motifs resembling carbohydrate-recognition domains. J Biol Chem 1990;265:12156-62. Drickamer K. Two distinct classes of carbohydrate-recognition domains in animal lectins. J Biol Chem 1988;263:9557-60. Weis WI, Kahn R, Fourme R, Drickamer K, Hendrickson. Structure of the calcium-dependant lectin domain from a rat mannose–binding protein determined by MAD phasing. Science 1991;254:1608-15. Gross J, Carlson RI, Brauer AW, Margolies MN, Warshaw AL, Wands JR. Isolation, characterization, and distribution of an unusual pancreatic human secretory protein. J Clin Invest 1985;76:2115-26. Mozdziak PE, Fassel T, Gregory R, Schultz E, Greaser ML, Cassens RG. Quantitation of satellite cell proliferation in vivo using image analysis. Biotech Histochem 1994;69:249-52. Weibel ER. Stereological principles morphometry in electron microscopic cytology. Int Rev Cytol 1969;26:235-302. Leng L, Yu F, Dong L, Busquets X, Osada S, Richon VM, et al. Differential modulation of protein kinase C isoforms in erythroleukemia during induced differentiation. Cancer Res 1993;53:5554-8. Special considerations for glycoconjugates and their purification. In: Ausubel FM, editor. Current protocols in molecular biology, vol 2. New York: John Wiley & Sons; 1994. Keim V, Iovanna JL, Orell B, Verdier JM, Busing M, Hopt U, et al. A novel exocrine protein associated with pancreas transplantation in humans. Gastroenterology 1991;103:248-54. Terazono K, Yamamoto H, Takasawa S, Shiga K, Yonekura Y, Tochino Y, et al. A novel gene activated in regenerating islets. J Biol Chem 1988;263:2111-4. Terazono K, Uchiyama Y, Ide M, Yatanbe T, Yonekura H, Yamamoto H, et al. Expression of reg gene in rat regenerating islets and its co-localization with insulin in the beta cell secretory granules. Diabetologia 1990;33:250-2. de Caro A, Lohse J, Sarles H. Characterization of a protein isolated from pancreatic calculi of men suffering from chronic calcifying pancreatitis. Biochem Biophys Res Commum 1979;87:1176-82. Rouquier S, Verdier JM, Iovanna J, Dagorn JC, Giorgi D. Rat pancreatic stone protein in mRNA. J Biol Chem 1991;266:786-91. Giorgi D, Bernard JP, Rouquier S, Iovanna J, Sarles H, Dagorn JC. Secretory pancreatic stone protein messenger RNA. J Clin Invest 1989;84:100-6.
Zenilman et al 863
Surgery Volume 124, Number 5 21. Zenilman ME, Perfetti R, Swinson K, Magnuson TH, Shuldiner AR. Pancreatic regeneration (reg) gene expression in a rat model of islet hyperplasia. Surgery 1996;119:576-84. 22. Rafeloff R, Barlow SW, Rosenberg L, Vinik AI. Expression of reg gene in the Syrian golden hamster pancreatic islet regeneration model. Diabetologia 1995;38:906-13. 23. Miyaura C, Chen L, Appel M, Alam T, Inman L, Hughes SD, et al. Expression of reg/PSP, a pancreatic exocrine gene: relationship to changes in islet b-cell mass. Mol Endocrinol 1991;5:226-34. 24. Watanabe T, Yonemura Y, Yonekura H, Suzuki Y, Miyashita H, Sugiyama K, et al. Pancreatic beta-cell replication and amelioration of surgical diabetes by reg protein. Proc Natl Acad Sci USA 1994;91:3589-92. 25. Montrose-Rafizadeh C, Egan JM, Roth J. Incretin hormones regulate glucose-dependent insulin secretion in RIN 1046-38 cells: mechanisms of action. Endocrinology 1994;135:589-94.
26. Logsdon CD, Moessner J, Williams JA, Goldfine ID. Glucocorticoids increase amylase mRNA levels, secretory organelles and secretion in pancreatic acinar AR42J cells. J Cell Biol 1985;100:1200-8. 27. Logsdon CD. Glucocorticoids increase cholecystokinin receptors and amylase secretion in pancreatic acinar AR42J cells. J Biol Chem 1986;261:2096-101. 28. Iovanna J, Orelle B, Keim V, Dagorn JC. Messenger RNA sequence and expression of rat pancreatitis-associated protein, a lectin-related protein overexpressed during acute experimental pancreatitis. J Biol Chem 1991;266:24664-9. 29. De Reggi M, Capon C, Gharib B, Wieruszeski JM, Michel R, Fournet B. The glycan moiety of human pancreatic lithostatine: structure characterization and possible pathophysiological implications. Eur J Biochem 1995;230:503-10. 30. Bimmler D, Graf R, Scheele GA, Frick TW. Lithostatine: a physiologically relevant inhibitor of pancreatic stone formation? [abstract] Gastroenterology 1996;110:A378.
Surgery is abstracted and/or indexed in Index Medicus, Science Citation Index, Current Contents/ Clinical Medicine, Current Contents/Life Sciences, and MEDLINE. This Journal has been registered with Copyright Clearance Center, Inc., 222 Rosewood Dr., Danvers, MA 01923. Consent is given for the copying of articles for personal or internal use of specific clients. This consent is given on the condition that the copier pay directly to the Center the per-copy fee stated on the first page of each article for copying beyond that permitted by U.S. Copyright Law. This consent does not extend to other kinds of copying, such as for general distribution, resale, advertising and promotional purposes, or for creating new collective works. All inquiries regarding copyrighted material from this publication other than those that can be handled through Copyright Clearance Center should be directed to Associate Journal Publisher, Permissions, Mosby, Inc., 11830 Westline Industrial Dr., St. Louis, MO 63146; phone (314)453-4129; fax (314)432-1380.
Background. The pancreatic regenerating (reg) gene is an acinar cell product involved in islet formation and maintenence. Human reg protein is mitogenic to pancreatic beta and ductular cells, and its amino acid sequence predicts it to be a calcium-dependent lectin. Methods. We studied the biologic activity and cellular localization of rat reg I isolated from the acinar cell line AR42J and the lectin properties of reg from AR42J and pancreatic juice. Bioactivity was assayed by mitogenesis on the ductular cell line ARIP. Cellular localization was determined by differential centrifugation. Lectin properties were assessed by affinity chromatography. Results. Reg protein from crude AR42J cellular lysate and purified reg protein from AR42J induced thymidine incorporation to ARIP. Conditioned medium from AR42J and co-culture of AR42J with morphologically distinguishable ARIP, however, failed to induce mitogenesis. Reg protein was localized within the vesicle fraction of the cell and was not membrane bound. Affinity chromotography revealed that reg protein did not bind to mannose or galactose in the presence or absence of calcium. In pancreatic juice a previously undescribed mannose-binding protein was discovered at 25,000 to 30,000 daltons. Conclusions. We conclude that reg produced in the acinar cell line AR42J is biologically active but not efficiently secreted, even though it localized within the cellular vesicles. Despite predictions based on its amino acid sequence, it does not appear to be a calcium-dependent lectin. (Surgery 1998;124:855-63.) From the Department of Surgery, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY
PANCREATIC REGENERATING PROTEIN (reg I) is an acinar cell product involved in islet regeneration. We have shown the human homolog to be mitogenic to rat pancreatic-derived cells: the beta cell line RIN, the duct cell line ARIP, and primary cultures of duct cells.1,2 Reg I protein was originally isolated from pancreatic juice, and we have recently shown that reg I gene is expressed in the rat acinar cell line AR42J. Its genetic expression can be modulated with dexamethasone.3 Western analysis has revealed that AR42J also produces reg I protein,1 and we have recently isolated the protein from cellular lysates.4 AR42J may therefore be a useful cell line for the study of rat reg I expression, translation, and secreSupported by an American College of Surgeons Faculty Fellowship, the Diabetes Research and Training Institute of Albert Einstein College of Medicine, and National Institutes of Health grant No. RO1 DK54511-01. Presented in part at the Annual Meeting of the American Gastroenterological Association, May 20, 1996, San Diego, Calif. Accepted for publication March 27, 1998. Reprint requests: Michael E. Zenilman, MD, Department of Surgery, Albert Einstein College of Medicine, Room 2S-5, 1825 Eastchester Rd, Bronx, NY 10461. Copyright © 1998 by Mosby, Inc. 0039-6060/98/$5.00 + 0
11/56/90945
tion, and it is important to know whether the protein produced is biologically active. The basic function and mechanism of action of reg I protein is presently unknown. The amino acid sequence analysis predicts it to be a calciumdependent lectin. The protein contains the Drickamer domains found in many such lectins,5 and its size is similar to that of the mannose-binding proteins and mannose-binding regions of the asialoglycoprotein receptor complex.6-8 The reg I amino acid sequence has significant homology to the mannose-binding regions of these domains, as described previously.9 Lectins are involved in formation of extracellular matrix, antigenic recognition between cells, cell-to-cell adhesion, and cellular differentiation. They may also have mitogenic properties. To date, however, no experimental data exist showing that reg I is in fact a calciumdependent lectin. In this study we examined the characteristics of rat reg I protein and the cellular biology of reg I in AR42J. First we studied the mitogenic effect of reg I protein found in AR42J, in both crude and pure isolates. For these studies, mitogenesis to the ductal cell line ARIP was used as a bioassay for reg I activity. We then studied whether reg I protein is secreted in a biologically active form from AR42J, using conditioned medium and co-culture techniques. Cellular SURGERY 855
856 Zenilman et al
localization studies were then performed on AR42J cells to determine which cellular compartment contains reg I protein. Specifically, if it were associated with the asialoglycoprotein complex, it would localize to the membrane portion of AR42J. Finally, we performed carbohydrate-binding studies of reg I isolated from AR42J and pancreatic juice. As a lectin, we predicted that the protein would bind carbohydrate columns in a calcium-dependent fashion.8 MATERIAL AND METHODS Cell culture. ARIP, a rat ductular cell line, and AR42J, a rat acinar cell line, were purchased from American Type Culture Collection (Rockville, Md). Cells were cultured in Dulbecco’s modified Eagle’s medium with 5% fetal calf serum. Isolation of rat reg I. Rat reg I was isolated from AR42J cellular lysates and pancreatic juice. Crude preparations of reg I protein were obtained from protein lysates of AR42J by sonication of 105 cells in 1 mL, centrifugation at 10,000g, and resuspension in 1 mL saline solution with 1 µg/mL aprotinin, for a final concentration of 5 µg/mL. Purified reg I protein from AR42J was isolated from 2 g AR42J, from the third 70% ammonium sulfate precipitation as described for pancreatic thread protein.10 We have previously shown that such isolates yield pure preparations of reg I protein.4 Pancreatic juice was isolated from 350 g breeder rats (Harlan Sprague Dawley, Indianapolis, Ind) as follows. After 50 mg/kg intraperitoneal anesthetic with sodium pentobarbital (Nembutal), a midline incision was performed. The common bile duct was identified and cannulated in an antegrade direction with PE-10 tubing (Fisher Scientific, Pittsburgh, Pa) such that the proximal end of the tube was beyond the choledochotomy and the distal end passed the ampulla of Vater in the duodenum. The bile duct was then ligated to prevent flow of bile, the duodenum was opened, and the distal tube was used for collection of pancreatic juice. Pancreatic flow was stimulated by intravenous injection of 1 unit/kg secretin (Sigma Chemical Co, St Louis, Mo), and juice was collected in the presence of aprotinin (Sigma Chemical Co), with a final concentration of 1 µg/mL. From each animal, 200 µL was typically collected for 2 hours. Mitogenesis. Bioactivity of reg I derived from AR42J was assayed by mitogenesis on the ductular cell line ARIP. One to 2 × 105 cells were plated in Dulbecco’s modified Eagle’s medium with fetal calf serum. Thymidine incorporation was assayed by scintillation 48 hours after treatment as described previously.1 Each dose was performed in triplicate and each experiment was performed at least 3 times.
Surgery November 1998
To determine if reg I was secreted from AR42J, conditioned medium from AR42J was obtained by incubating 106 cells in 3% bovine serum albumin for 24 hours at 37 o C. ARIP cells were then exposed to conditioned medium and subjected to thymidine incorporation. Co-culture of AR42J with ARIP was also performed. For these experiments, 106 ARIP was cultured with or without 104 AR42J for 48 hours in a 75-mm Petri dish. Cells were then treated with bromodeoxyuridine (1:1000 dilution; 10 µmol/L final concentration) (Amersham Corp, Arlington Heights, Ill) for 2 hours and fixed in 70% ethanol. Cells were then stained with monoclonal antibodies to bromodeoxyuridine (Amersham Corp), washed with phosphate-buffered saline solution (pH 7.4 + 0.5% bovine serum albumin + 0.5% Tween), and labeled with goat antimouse immunoglobulin G conjugated to fluorescein-5-isothiocyanate (FITC) (Sigma Chemical Co). All nuclei were then stained with 1 µg/mL propridium iodide (Sigma Chemical Co).11 When viewed under an Olympus IMT-2 inverted microscope and subjected to excitation with an EY455-nm filter, the propidium iodide–labeled nuclei fluoresced red and the FITC-labeled bromodeoxyuridine green. Under phase contrast, ARIP cells are morphometrically distinct from AR42J. The percentage of ARIP nuclei positive for bromodeoxyuridine was calculated from at least 50 nuclei within at least 5 separate colonies per experiment, according to point-counting techniques.12 These values were compared with control experiments in which ARIP cells were cultured alone. Each experiment was performed at least 3 times per dose along with zero controls. Cellular localization. Analysis of cellular fractions was performed by differential centrifugation according to a variation of the methods described13,14; 106 AR42J cells were harvested and washed with phosphate-buffered saline solution (0.1 mol/L sodium phosphate and 0.15 sodium chloride; pH 7.2) twice and then lysed in 1 mL hypotonic lysis buffer (1 mmol/L NaHCO3, 5 mmol/L MgCl2, 100 µmol/L phenylmethylsulfonyl fluoride, 10 µg/mL leupeptin, and 10 µg/mL soybean trypsin inhibitor) at 4°C for 2 minutes. The cell lysate was sonicated for 5 minutes and transferred into a microphage tube. Nuclei and residual cells were spun down at 500g for 5 minutes at 4°C. The supernatant, containing cellular vesicles and residual membranes, was centrifuged at 15,000g at 4°C for 1 hour. The supernatant, containing cell membranes, was saved. The pellet, containing vesicles (and residual membranes), was washed with
Surgery Volume 124, Number 5
phosphate-buffered saline solution 3 times and resuspended in 500 µL 1% sodium dodecylsulfate and heated at 50°C for 1⁄2 hour. This was frozen and thawed 3 times to break up the vesicles and then spun at 100,000g for 20 minutes at 4°C (Ultra-centrifuge; Beckman, Fullerton, Calif) to separate vesicular membranes (and residual cellular membranes) from vesicular contents. Samples were concentrated 5 times by dialysis against polyethylene glycol, and 20 µL was then run on a 15% polyacrylamide minigel (Biorad, Hercules, Calif). Protein characteristics. If reg I is a calciumdependent lectin, it should selectively bind a carbohydrate-affinity column.8 To determine the calcium-binding lectin characteristics of reg I protein, a 0.5 mL column of mannose or N-acetyl-D-galactosamine cross-linked to 4% agarose beads (Sigma Chemical Co) was used at 4o C. After washing with 10 × vol of 10 mmol/L Tris (pH 7.4), 15 mmol/L CaCl2, 1.25 mol/L NaCl, 0.02% NaN3, and 0.5% bovine serum albumin, 100 µL pancreatic juice or reg I isolated from AR42J was added to the column and allowed to bind overnight, and the column was washed. The column was eluted with 10 mmol/L Tris (pH 7.4), 2 mmol/L ethylenediaminetetraacetic acid (EDTA), 1.25 mol/L NaCl, 0.02% NaN3, and 0.5% bovine serum albumin. A 2-mol/L galactose or mannose solution was also used after the EDTA elution. The eluates were concentrated 10 times by ultrafiltration, and 10 µL aliquots from each run were subjected to polyacrylamide gel electrophoresis (15%) and silver staining. Reg I antibody. We produced a polyclonal antibody in guinea pigs (Harlan Sprague Dawley) to a 10 amino acid peptide portion of rat reg I protein linked to a keyhole limpet hemocyanin backbone (sequence YFMEDHLSWA, residues 48 to 57 from the N-terminus of the first methionine amino acid or residues 26 to 35 from the N-terminal methionine of the secreted protein) (Research Genetics, Inc, Huntsville, Ala). The sequence is unique to rat reg I (50% homology to human reg I, 10% homology to rat reg III, and 50% homology to rat reg II). The antibody was characterized by Western analysis to rat pancreatic juice and reg I purified from AR42J, described below. Western analysis. After confirmation of protein in the gel by Coomassie Brilliant Blue staining or Silver staining (Biorad), protein was transferred to nitrocellulose electrophoretically (0.45 µm; Micron Separations, Inc, Westboro, Mass) at 100 V for 1 hour. The transfer membrane was incubated in block solution (10% bovine serum albumin in 0.01 sodium phosphate, 150 mmol/L NaCl, and 0.1% thimerosal; pH 7.4) (Pierce, Rockford, Ill) at
Zenilman et al 857
4°C overnight. The membrane was then washed with Tris-buffered saline solution (TBS) for 5 minutes 3 times and incubated with polyclonal guinea pig antirat-reg I antibody, produced in our laboratory, at 1:1000 dilution at room temperature for 1 hour. The membrane was washed in TBS for 5 minutes, 3 times, and incubated with the second antibody goat anti-guinea pig immunoglobulin G alkaline phosphatase (Sigma Chemical Co) at 1:15,000 dilution at room temperature for 30 minutes. The membrane was washed with TBS for 5 minutes 3 times. The blot was developed according to the 1step technique with nitroblue tetrazolium/BCIP substrate (Pierce). RESULTS Mitogenic effect of reg I protein isolated from AR42J. Reg I protein was isolated from AR42J as described above, and purity was confirmed by polyacrylamide gel electrophoresis.4 In initial experiments, crude and pure isolates of reg I isolated from AR42J were used in experiments on ARIP cells. Mitogenesis was assessed by tritiated thymidine incorporation. Results from typical experiments are shown in Fig 1. A dose-related effect of both cellular lysates of AR42J (which contain reg I) and reg I purified from AR42J on ARIP thymidine incorporation was noted. Secretion of reg I from AR42J. Secretion of reg I from AR42J was studied with ARIP mitogenesis used as a bioassay. In the first series of experiments, ARIP cells were subjected to conditioned medium from AR42J for 48 hours and then pulsed with tritiated thymidine. No effect was noted with ARIP thymidine incorporation. In a typical experiment, treatment of ARIP with control medium followed by thymidine yielded a value of 5094 ± 365 cpm, 25% conditioned medium yielded 5618 ± 488 cpm, treatment with 75% conditioned medium yielded 5228 ± 1330 cpm, and treatment with 100% conditioned medium yielded 5162 ± 470 cpm (difference not significant). One hundred percent conditioned medium was then concentrated 10 times by dialysis and subjected to electrophoretic analysis (15% polyacrylamide gel electrophoresis and visualization by silver staining), which revealed a weak band consistent with amylase (molecular weight about 50,000 daltons) but no band at 16,000 or 18,000 daltons, indicating that significant quantities of reg I were not present (data not shown). A series of experiments was then performed to determine whether co-culture of AR42J cells with ARIP cells induced mitogenesis in ARIP. We surmised that local concentrations of secreted reg I might be higher in the region surrounding
858 Zenilman et al
Surgery November 1998
Fig 1. A, Effect of cellular lysate of AR42J on ARIP thymidine incorporation. Dose-related effect is seen with increasing concentrations of lysate, whose concentration was adjusted to 5 µg/mL (+1 µg/mL aprotinin). B, Purified reg I protein from AR42J was mitogenic to ARIP cells in dose-related fashion. *P < .05 compared with control, analysis of variance.
Fig 2. Co-culture of ARIP and AR42J. Under phase contrast, ARIPs are typically flat and discernable from AR42J because latter grow in clumps. After 48 hours of co-culture, cells were pulsed with bromodeoxyuridine (BrdU). Cells were stained with FITC-labeled antibromodeoxyuridine antibody, and all nuclei were stained with propidium iodide. Nuclei of AR42J are seen as clusters, surrounding light gray nuclei are all ARIP, and black single nuclei are ARIP positive for bromodeoxyuridine. Percent ARIP staining positive for bromodeoxyuridine was calculated by point counting and compared with pure ARIP cultures (color photos available on request). (Original magnification × 200.)
AR42J, and this should be a more reliable method of showing a mitogenic effect on the ARIP cell. Under phase contrast microscopy, AR42J cells are distinguishable from ARIP, and differential calculation of bromodeoxyuridine
incorporation by each group of cells is possible. Results from a typical experiment are shown in Fig 2. No difference was found in bromodeoxyuridine uptake between plates containing ARIP cells alone (5.49% ± 0.30%) and plates con-
Surgery Volume 124, Number 5
Zenilman et al 859
Fig 3. A, Western analysis of pancreatic juice (PJ) and AR42J lysate with polyclonal antibody to 10 amino acid peptide fragment of rat reg I. Band is seen at 16,000 daltons, consistent with reg I. B, Inhibition of Western blot by reg I 10 amino acid peptide. Pure reg I isolated from AR42J was subjected to standard Western analysis. When 2 mg/mL peptide is present in solution containing polyclonal antibody, band on Western is inhibited. Similar data were observed if reg I protein was used to block experiment.
taining both ARIP and AR42J (4.51% ± 0.38%; difference not significant). These data strongly suggest that reg I is not secreted by AR42J cells in a concentration high enough to affect ARIP in our bioassay, and maybe not at all. Antibody characterization. A polyclonal antibody was produced in guinea pigs to a unique 10 amino acid fragment of reg I. Western analysis revealed that the antibody isolated reacted as a single band at 16,000 daltons in pancreatic juice, AR42J cellular lysate, and purified reg I from AR42J, all consistent with rat reg I (Fig 3). It did not react with human reg I (not shown). Specificity of the antibody to rat reg I was shown by blocking the antibody during Western analysis with either pure peptide (Fig 3, B) or pure reg I protein isolated from AR42J (not shown). A competitive enzyme-linked immunosorbent assay was constructed that easily detected reg I in pancreatic juice, rat serum, and AR42J lysate. Interestingly, very little to no reg I (above background) was detected in conditioned medium from AR42J. This was consistent with our hypothesis that the protein is not secreted from this cell line. Cellular localization studies in AR42J. To determine where reg I protein localized within the AR42J cell, cytoplasmic, vesicular, and membrane compartments were isolated as described in Material and Methods. Isolates at each step were subjected
to sodium dodecylsulfate–polyacrylamide gel electrophoresis and then transferred to nitrocellulose and analyzed by Western analysis for reg I. Fig 4 shows the 15% polyacrylamide gel and Western analysis of the cytoplasmic, whole, and disrupted vesicle portions of the analysis. From this analysis it appears as if reg I localized to the lanes that contained whole vesicles and their contents, not the cytoplasmic or membrane portions. Western analysis of a similar gel that used a polyclonal antibody to amylase showed that amylase localized to the same isolate as reg I (data not shown) Lectin properties of reg I. Reg I protein was assayed for calcium-dependent carbohydrate binding by affinity chromatography. Solutions containing reg I were loaded on mannose-sepharose or galactose-sepharose columns in the presence of calcium and subsequently eluted in a calcium-free buffer (with EDTA), followed by elution in a buffer with 2 mol/L carbohydrate. Two separate preparations of reg I were used: pancreatic juice and the protein isolate from AR42J cells. Fig 5 shows the result of affinity chromatography of pancreatic juice on a galactose and mannose column. No binding was noted on the galactose column. During the elution of the mannose-affinity column, a band between 25,000 and 30,000 daltons was liberated. Although this is in the region of glycoproteins 2 and 3, chymotrypsin and proelastase,15 to date such a lectin in pancreatic juice has
860 Zenilman et al
Surgery November 1998
Fig 4. A, Fifteen percent polyacrylamide gel electrophoresis (PAGE) stained with Coomassie blue of AR42J celluar compartments. B, Western analysis of same gel with polyclonal reg I antibody. Lanes a and b depict product of separation of cytoplasmic contents (a) from vesicles and residual membranes (b). Lane c is vesicle membrane fraction after freeze thawing and ultracentrifugation, and lane d is vesicle content fraction. Reg I antibody hybridized at approximately 16,000 daltons in vesicular and vesicular lysate fractions alone. (MW, Molecular weight marker.)
not been described. Nothing was liberated in the 15,000- to 18,000-dalton range, the size of reg I. When reg I protein isolated from AR42J was subjected to galactose-sepharose or mannose-sepharose affinity chromatography in a manner similar to that above, no binding was noted (not shown). Examination of the carbohydrate-linked sepharose beads after treatment with both preparations of reg I under phase contrast microscopy followed by treatment with anti-reg I antibody and viewed with FITC-labeled second antibody also failed to show any binding to the beads (not shown). It was found, however, in the unbound fraction by both Western analysis and competitive enzyme-linked immunosorbent assay with the polyclonal antibody described above (not shown). These results indicate that reg I had no affinity to either mannose or galactose. DISCUSSION This is the first study that describes both protein analysis and cellular physiology of rat reg I protein. Although the reg I gene was originally described in 1988,16,17 its amino acid sequence has been shown to be identical to that of pancreatic stone protein (lithostatine), a pancreatic secretory product described originally in the 1970s.18 Lithostatine is believed to be important in inhibiting calcium carbonate precipitation,16,18-20 and decreased concentrations in pan-
creatic juice are believed to be associated with the development of chronic calcific pancreatitis. The reg I gene is induced by beta cell regeneration after 95% pancreatectomy followed by nicotinamide administration and during beta cell hyperplasia after pancreatic wrapping.21,22 Its mRNA expression is inhibited during beta cell suppression.23 We have recently shown that reg I mRNA and protein is found in the rat acinar cell line AR42J, and the protein can be isolated from AR42J lysates.4 The mechanism of action of this exocrine product on the pancreatic endocrine tissue is unknown. Reg I protein is mitogenic to pancreatic-derived cells. We have shown that the human homolog of reg I induces thymidine incorporation into a rat ductular and beta cell line1 and causes increased incorporation of bromodeoxyuridine in isolated rat duct cells.2,4 Other investigators have shown that exogenous administration of recombinant reg I protein can reverse diabetes after pancreatic resection, and it is mitogenic to only beta cells within the islet. 24 We believe that the protein acts as a stimulus for growth of both the native beta cell population and ductal precursor cells of the pancreas and may even induce the latter’s differentiation into beta cells. The mechanism of how the protein induces mitogenesis is unknown. Although much has been reported about reg I
Surgery Volume 124, Number 5
Zenilman et al 861
Fig 5. Affinity chromotography of pancreatic juice (PJ) followed by 15% polyacrylamide gel electrophoresis and silver staining. A, Results from mannose-sepharose column. B, Results from galactose-sepharose. Results revealed that reg I protein did not bind to mannose or galactose in presence of calcium, indicating that in mature form reg I is not calcium-dependent lectin. Previously undescribed mannose-binding lectin in pancreatic juice is noted at 25,000 to 30,000 daltons of panel A. (MW, Molecular weight marker; run, runthrough after loading; wash, last wash with buffer; EDTA, 2 mmol/L EDTA elution; man or gal, 2 mol/L mannose or galactose elution.)
protein in experiments focusing on pancreatic stone formation, other basic biochemical and cellular studies are lacking. The amino acid sequence analysis of reg I predicts it to be a calcium-dependent lectin, and its size is similar to the mannose-binding proteins and mannose-binding domains of the asialoglycoprotein receptor complex. 6-8 Reg I contains the Drickamer domains found in many such lectins.5 Specifically, comparison of the rat reg I protein sequence with rat mannose–binding domains9 has shown conservation of most of the 32 conserved sequences, including the important WIGL amino acid sequence (aromatic-aliphatic-glycine-aliphatic), a hydrophobic packing common to all carbohydrate-recognition domains (unpublished observations). 9 Animal lectins are important in cellular physiology; they are involved in formation of the extracellular matrix, antigenic recognition between cells, cellto-cell adhesion, and cellular differentiation. They may also have mitogenic properties, as evidenced by relationships with the epidermal growth factor receptor.8,9 If reg I is a lectin, this would shed light on the mechanism by which it is mitogenic to the pancreatic-derived cells. The first series of experiments focused on reg I protein produced in the rat acinar cell line AR42J. Because AR42J is able to produce reg I and has been shown to be a useful model for studying reg I genetic expression, is the reg I produced in AR42J biologically active and, if so, is it secreted? Our results show that the reg I protein produced in AR42J is biologically active, in both crude cellular lysates of AR42J and pure reg I preparations. We showed this with a bioassay of mitogenesis in the duc-
tular cell line ARIP, a cell line that we have shown is stimulated in this fashion by human reg I.1 There seems to be a problem, however, with secretion of reg I from AR42J. Conditioned medium from AR42J failed to induce thymidine incorporation into ARIP, and electrophoresis of concentrated conditioned medium, as well as analysis by enzymelinked immunosorbent assay, did not reveal reg I. We have observed similar negative results when RIN 1046-38 cells are used (unpublished observations), a beta cell line also sensitive to reg I.1,25 Finally, direct co-culture of ARIP with AR42J did not induce ARIP cell growth, further indicating that, although reg I is produced in AR42J, it is not actively secreted into the surrounding medium. This is a surprising finding because reg I is normally secreted from the pancreatic acinar cells into pancreatic juice, in high concentrations.19,20 We next performed cellular-localization studies of reg I protein in AR42J. Amino acid sequence homologies raised the possibility that reg I is a member of, or a portion of, the asialoglycoprotein receptor family. Comparison of the reg I amino acid sequence to the 15,000-dalton carbohydrate-recognition domains of this giant receptor have shown similarities.6-8 Although reg I is found in pancreatic juice at high concentrations, could it be a product of a membraneassociated glycoprotein, or is it found in a secretory vesicle? Our cellular localization studies by differential centrifugation and Western analysis revealed that reg I localized within the vesicular component of the cell and is not membrane bound. Further Western analysis has shown that reg I resides in the same fraction as amylase, which is secreted from AR42J.26,27 It is therefore likely to be a normally secreted product
862 Zenilman et al
of the cell, not part of the asialoglycoprotein receptor or any other receptor complex. Although reg I is located within cellular vesicles, it is not secreted efficiently. This may be the result of a defect in the secretory mechanism of AR42J, a defect in the amino acid sequence of the signal peptide of AR42J reg I that directs it to a secretory vesicle, or that the protein is secreted in quantities too dilute to be detected by our assays. The latter is possible because even amylase is secreted in relatively small quantities by AR42J under basal conditions.27 Finally, we studied the lectin properties of reg I found in both pancreatic juice and AR42J. Our findings with affinity chromatography with galactoseand mannose-sepharose columns indicate that reg I does not bind carbohydrate. Reg I protein is not, in its mature form, a calcium-dependent lectin. This is in agreement with Iovanna et al,28 who found similar results with pancreatitis-associated protein, which is reg III,5 a homolog of reg I. These data suggest that, despite predictions from their amino acid sequence, the reg family is not a family of calciumdependent lectins. Alternatively, although it is known that human reg I is heavily glycosylated,29 it is possible that the carbohydrate-recognition domains were previously occupied in both samples tested. This should not be the case with rats, however, because some believe that rat reg I is not glycosylated.30 To address this question, we are producing recombinant reg I in a bacterial expression system and will subject the product to similar tests as described above to determine the lectin properties. An unexpected finding from the affinity-chromatography experiments is that of a previously undescribed mannose-binding lectin at 25,000 to 30,000 daltons, in the region of glycoproteins 2 and 3, chymotrypsin and proelastase. We are currently isolating this protein to determine its sequence. In summary, we have performed preliminary experiments characterizing rat reg I protein. We have shown that reg I produced in the acinar cell line AR42J is biologically active but not actively secreted. It is found in cellular vesicles and is not membrane bound. Despite predictions based on its amino acid sequence, it does not appear to be a calcium-dependent lectin. We acknowledge the technical help of Akiva Marcus, Ferrel Motlow, and Benjamin Gelman and thank Phil Stahl, PhD, of Washington University for helpful conversations regarding lectin properties and the mannose receptor.
Surgery November 1998
2.
3.
4. 5.
6.
7.
8. 9.
10.
11.
12. 13.
14.
15.
16.
17.
18.
19.
20.
REFERENCES 1. Zenilman ME, Magnuson TH, Swinson K, Egan J, Perfetti R,
Shuldiner AR. Pancreatic thread protein is mitogenic to pancreatic-derived cells in culture. Gastroenterology 1996;110:1208-14. Zenilman ME, Chen J, Magnuson TH, Shuldiner AR. Human PTP/reg is mitogenic to primary culture of pancreatic ductal cells [abstract]. Gastroenterology 1995;108:A402. Zenilman ME, Magnuson TH, Perfetti R, Swinson K, Shuldiner AR. Reg gene expression is inhibited during differentiation of cultured rat acinar cells. Ann Surg 1997;225:327-32. Zenilman ME, Chen J, Magnuson TH. Effect of reg protein on pancreatic ductal cells. Pancreas In press. Unno M, Yonekura H, Nakagawara K, Watanabe T, Miyashita H, Moriizumi S, et al. Structure, chromosomal localization and expression of mouse reg genes reg I and reg II. J Biol Chem 1993;268:15974-82. Drickamer K, McCreary V. Exon structure of a mannosebinding protein gene reflects its evolutionary relationship to the asialoglycoprotein receptor and nonfibrillar collagens. J Biol Chem 1987;262:2582-9. Taylor ME, Conary JT, Lennartz MR, Stahl PD, Drickamer K. Primary structure of the mannose receptor contains multiple motifs resembling carbohydrate-recognition domains. J Biol Chem 1990;265:12156-62. Drickamer K. Two distinct classes of carbohydrate-recognition domains in animal lectins. J Biol Chem 1988;263:9557-60. Weis WI, Kahn R, Fourme R, Drickamer K, Hendrickson. Structure of the calcium-dependant lectin domain from a rat mannose–binding protein determined by MAD phasing. Science 1991;254:1608-15. Gross J, Carlson RI, Brauer AW, Margolies MN, Warshaw AL, Wands JR. Isolation, characterization, and distribution of an unusual pancreatic human secretory protein. J Clin Invest 1985;76:2115-26. Mozdziak PE, Fassel T, Gregory R, Schultz E, Greaser ML, Cassens RG. Quantitation of satellite cell proliferation in vivo using image analysis. Biotech Histochem 1994;69:249-52. Weibel ER. Stereological principles morphometry in electron microscopic cytology. Int Rev Cytol 1969;26:235-302. Leng L, Yu F, Dong L, Busquets X, Osada S, Richon VM, et al. Differential modulation of protein kinase C isoforms in erythroleukemia during induced differentiation. Cancer Res 1993;53:5554-8. Special considerations for glycoconjugates and their purification. In: Ausubel FM, editor. Current protocols in molecular biology, vol 2. New York: John Wiley & Sons; 1994. Keim V, Iovanna JL, Orell B, Verdier JM, Busing M, Hopt U, et al. A novel exocrine protein associated with pancreas transplantation in humans. Gastroenterology 1991;103:248-54. Terazono K, Yamamoto H, Takasawa S, Shiga K, Yonekura Y, Tochino Y, et al. A novel gene activated in regenerating islets. J Biol Chem 1988;263:2111-4. Terazono K, Uchiyama Y, Ide M, Yatanbe T, Yonekura H, Yamamoto H, et al. Expression of reg gene in rat regenerating islets and its co-localization with insulin in the beta cell secretory granules. Diabetologia 1990;33:250-2. de Caro A, Lohse J, Sarles H. Characterization of a protein isolated from pancreatic calculi of men suffering from chronic calcifying pancreatitis. Biochem Biophys Res Commum 1979;87:1176-82. Rouquier S, Verdier JM, Iovanna J, Dagorn JC, Giorgi D. Rat pancreatic stone protein in mRNA. J Biol Chem 1991;266:786-91. Giorgi D, Bernard JP, Rouquier S, Iovanna J, Sarles H, Dagorn JC. Secretory pancreatic stone protein messenger RNA. J Clin Invest 1989;84:100-6.
Zenilman et al 863
Surgery Volume 124, Number 5 21. Zenilman ME, Perfetti R, Swinson K, Magnuson TH, Shuldiner AR. Pancreatic regeneration (reg) gene expression in a rat model of islet hyperplasia. Surgery 1996;119:576-84. 22. Rafeloff R, Barlow SW, Rosenberg L, Vinik AI. Expression of reg gene in the Syrian golden hamster pancreatic islet regeneration model. Diabetologia 1995;38:906-13. 23. Miyaura C, Chen L, Appel M, Alam T, Inman L, Hughes SD, et al. Expression of reg/PSP, a pancreatic exocrine gene: relationship to changes in islet b-cell mass. Mol Endocrinol 1991;5:226-34. 24. Watanabe T, Yonemura Y, Yonekura H, Suzuki Y, Miyashita H, Sugiyama K, et al. Pancreatic beta-cell replication and amelioration of surgical diabetes by reg protein. Proc Natl Acad Sci USA 1994;91:3589-92. 25. Montrose-Rafizadeh C, Egan JM, Roth J. Incretin hormones regulate glucose-dependent insulin secretion in RIN 1046-38 cells: mechanisms of action. Endocrinology 1994;135:589-94.
26. Logsdon CD, Moessner J, Williams JA, Goldfine ID. Glucocorticoids increase amylase mRNA levels, secretory organelles and secretion in pancreatic acinar AR42J cells. J Cell Biol 1985;100:1200-8. 27. Logsdon CD. Glucocorticoids increase cholecystokinin receptors and amylase secretion in pancreatic acinar AR42J cells. J Biol Chem 1986;261:2096-101. 28. Iovanna J, Orelle B, Keim V, Dagorn JC. Messenger RNA sequence and expression of rat pancreatitis-associated protein, a lectin-related protein overexpressed during acute experimental pancreatitis. J Biol Chem 1991;266:24664-9. 29. De Reggi M, Capon C, Gharib B, Wieruszeski JM, Michel R, Fournet B. The glycan moiety of human pancreatic lithostatine: structure characterization and possible pathophysiological implications. Eur J Biochem 1995;230:503-10. 30. Bimmler D, Graf R, Scheele GA, Frick TW. Lithostatine: a physiologically relevant inhibitor of pancreatic stone formation? [abstract] Gastroenterology 1996;110:A378.
Surgery is abstracted and/or indexed in Index Medicus, Science Citation Index, Current Contents/ Clinical Medicine, Current Contents/Life Sciences, and MEDLINE. This Journal has been registered with Copyright Clearance Center, Inc., 222 Rosewood Dr., Danvers, MA 01923. Consent is given for the copying of articles for personal or internal use of specific clients. This consent is given on the condition that the copier pay directly to the Center the per-copy fee stated on the first page of each article for copying beyond that permitted by U.S. Copyright Law. This consent does not extend to other kinds of copying, such as for general distribution, resale, advertising and promotional purposes, or for creating new collective works. All inquiries regarding copyrighted material from this publication other than those that can be handled through Copyright Clearance Center should be directed to Associate Journal Publisher, Permissions, Mosby, Inc., 11830 Westline Industrial Dr., St. Louis, MO 63146; phone (314)453-4129; fax (314)432-1380.