European Journal of Clinical Investigation (1992) 22, 283-29 1
Glucagon-like peptide-I cells in the gastrointestinal tract and pancreas of rat, pig and man R. EISSELE, R. GOKE, S. WILLEMER, H.-P. HARTHUS*, H. VERMEER*, R. ARNOLD & B. GOKE Department of Internal Medicine, Division of Gastroenterology and Metabolism, Philipps University of Marburg and Research Laboratories of Behringwerke AG*, Marburg, Germany Received 21 June 1991 and in revised form 2 October 1991; accepted 22 October 1991
Abstract. A highly specific monoclonal antibody directed against the C-terminal part of glucagon-like peptide-I (GLP- I ) was raised to immunohistochemically evaluate the distribution of GLP-I containing cells in the entire gastrointestinal tract including pancreas of rat, pig and man. In the pancreas GLP-Iimmunoreactive cells were found variously shaped and predominantly located in the periphery of the islets. Ultrastructurally, GLP- 1 was co-localized with glucagon in the a-granula of A-cells and was mainly restricted to the electrondense core. In the intestine open type cells reaching the lumen via a slender apical process were stained with the GLP-I antibody. They occurred in all parts of the crypts but predominantly in the basal portion. The density of GLP-I immunoreactive cells varied between species in a characteristic order: rat > pig > man. In pig and human gut a large number of cells occurred in the distal jejunum and ileum. A continuous increase of cell densities was found from the proximal to the distal colon resulting in highest numbers in the rectum. In rats the highest cell density occurred in the ileum. Again, a continuous increase of GLP-I-positive cell numbers was evident from the proximal to the distal portion of small and large bowel. GLP-I was partly co-localized with PYY. The GLP- 1 positive cells appeared electronmicroscopically as L-cells with the typical large granula. This morphological data indicates that GLP- I-releasing cells in the small intestine are appropriately positioned in the distal part to sense and respond to the presence of nutrients that have escaped the absorptive surface of the upper small intestine. Abbreviations GLP-I : Gucagon-like peptide-I, L-cell: endocrine cell in the intestine with large granules, A-cell: ‘glucagon cell’ in the islet of Langerhans, PYY: Peptide YY; GIP: Gastric inhibitory polypeptide, VIP: Vasoactive intestinal peptide; GRPP: Glicentin-related pancreatic peptide. Correspondence: Dr R.Coke, Laboratory of Molecular Endocrinology, Division of Gastroenterology, Department of Internal Medicine, Philipps University of Marburg. Baldingerstrape, W-3550 Marburg. Germany.
Introduction
Gut glucagon-like immunoreactants (gut GLIs) are located in the L-cell of the intestinal mucosa [I]. The greatest concentration of L-cells is detected in the distal small intestine and colon [2]. In recent years cloning and sequence analysis of cDNA and DNA fragments from genomic libraries has led to a dramatic increase in understanding of glucagon-related peptides [3]. The complex structural connections between the multiple molecular forms of the glucagon-like peptides in tissues have been determined [4,5]. The preproglucagon mRNAs from pancreas and intestine are identical, but the post-translational processing of the primary transcript differs markedly in the two tissues [4,5]. In the pancreas proglucagon is predominantly processed to proglucagon 1-30 (glicentin related pancreatic peptide; GRPP), glucagon and a carboxy-terminal fragment containing the sequences of glucagon-like peptide-I (GLP- I ) and glucagon-like peptide-2 (GLP2). In the gut proglucagon is mainly processed to proglucagon 1-69 (glicentin) and two smaller peptides, glucagon-like peptide-I (GLP-I) and glucagon-like peptide-2 (GLP-2) [4,5]. GLP-I is present in at least two forms (1-37 and 7-37). The truncated peptide is aamidated at the C-terminal end (6). GLP-1 (7-36) amide, which is secreted by the intestine, strongly stimulates insulin secretion, suppresses glucagon release, and inhibits gastric secretion [7- 171. Using polyclonal antibodies against GLP- 1 (1- 1 9) and GLP-1 (7-37), GLP-1-like immunoreactivity was identified in colorectal enteroglucagon cells [I, 181. Furthermore, GLP-I-like IR was described in pancreas and gut utilizing polyclonal antibodies antiGLP-1 (1-19) [1,19], and anti-GLP-l (I-37)/(1-36) amide [1,19,20,21]. Most of these studies were hampered by the circumstance that the antibodies used were only poorly characterized. Cross-reactivity with other important proglucagon-derived peptides or other members of the secretin family, to which GLP-I also belongs, was not studied. Therefore, in this study a specific monoclonal antibody against synthetic GLP- 1 (7-36) amide was developed and carefully character283
284
R. EISSELE et al.
ized to evaluate the distribution of GLP-I (7-36) amide containing cells in the gastrointestinal tract. In contrast to earlier reports, our study was extended to the entire gastrointestinal tract and pancreas of three different species, rat. pig, and man. Hereby, light and electron microscopic immunohistochemistry was employed. Furthermore, it was studied whether previous suggestions hold true that proglucagon-derived peptides are stored in separate compartments of the secretory granule [ 1,221. Data concerning the tissue-specific and/or species specific distribution of GLP-I-like IR could contribute to understand the regulation of GLP-I release and its physiological implications. We searched for a possible co-localization of GLP- 1 with other regulatory peptides. Such findings could help to understand tissuespecific processing of peptides and could give some insight into the specificity of stimulus-dependent peptide release from endocrine intestinal cells. Materials and methods Antisrra Preparation of monoclonal antibodies. GLP- 1 peptide/KLH conjugate. S-acetylmercaptosuccinic anhydrid (SAMSA) was used to introduce thiol groups into the peptide [23]. Glucagon-like peptide 1 (7-36) NH2 was obtained from Peninsula Laboratories (California, USA). Keyhole limpet hemocyanin (KLH) obtained from Calbiochem (La Jolla, USA) was dissolved in 0.1 M Li-phosphate containing 10% dioxan (pH 8.5) at a concentration of 10 mgml-I. To 40 mg of KLH was added 3.5 mg of gammamaleimidobutyric-succinimide (GMBS from Calbiochem, La Jolla, USA) dissolved in 180 p1 dioxan. After 30 min reaction at room temperature the reaction mixture was desalted on Sephadex G-25 into 0.1 M phosphate containing 5 mM EDTA (pH 6.0). The SH-peptide was reacted with the GMBS-KLH for 2 h at room temperature. After this time residual free SH-groups were blocked by reaction with 0.1 volumes of 0.1 M N-ethylmaleimide. The obtained conjugate was dialysed against physiological NaCl and used without further purification for immunization. Immunization. For the development of monoclonal antibodies against GLP-I a GLP-1/KLH-conjugate was used. Balb/c mice, 4-6 weeks old, were immunized subcutaneously and/or intraperitoneally with 20 pg of the GLP- 1 /KLH-conjugate emulsified in Freund’s complete adjuvant. Four weeks later, mice were boosted with 10 pg in incomplete Freund’s adjuvant followed by a third immunization 6 weeks later. Prior to fusion, mice were boosted intravenously four times at daily intervals. Spleen cells were isolated aseptically and were fused with the myeloma cell line SP 2/0 using polyethylene glycol (PEG) according to standard methods. The final cell pellet was resuspended in DMEM (Dulbecco’s Modified Eagle Medium) con-
taining 20% calf serum and HAT (0-1 mM hypoxanthin, 0.004 mM aminopterin, 0.0 16 mM thymidine) and added to 24-well culture plates (Nunc). About 14 days later, individual cell clones were picked out and were transferred to a new well. Three days later, supernatants were tested for antibody content as well as for the presence of GLP-I-specific antibodies. Positive cell cultures were grown up and were frozen in liquid nitrogen. In parallel, positive cell lines were cloned using a single cell manipulator. Screening of monoclonal antibodies (mabs) . Supernatants were screened using GLP-I-coated microtitration plates. Mabs bound to the solid phase were demonstrated by a second incubation with horseradish peroxidase (HRP) labelled rabbit anti-mouse antibodies. Relative affinity of anti-GLP-l mabs was determined by incubating defined concentrations of different antibody containing supernatants on GLP- 1 coated microtitration plates. Product ion and purijicat ion of monoclonul antibodies. Cell lines producing anti-GLP-1 antibodies were propagated in mass culture. The monoclonal antibody was purified by ammonium sulfate precipitation followed by affinity chromatography using Protein Asepharose (Pharmacia/LKB, Freiburg, Germany) according to the manufacturer. Purity was monitored by HPLC and SDS-PAGE (Pharmacia Phast System). Protein concentration was determined by optical density measurement at 280 nm. Class and subclass isotypes were determined by double diffusion techniques (Miles). The Protein A-purified antibody was stored at a protein concentration of 11.13 mg ml-’. Specijicity tests The newly raised monoclonal antibody to synthetic GLP-I (7-36) amide was tested for specificity and cross-reaction by immunocytochemistry and Dot-blot analysis. Immunocytochemical controls included the omission of the first antibody, the second antibody, the avidin-biotin complex and replacement of the first antibody by non-immune mouse antiserum. Increasing dilutions of the first antibody were tested ( 1 :50 to 1 :20000). In the preabsorption test the antibody was incubated with a 10,100 and 1000 fold excess of GLP-1 (7-36) amide (Peninsula Laboratories, Europe Ltd, St Helens, Merseyside, UK) and GLP-I (1-36) amide (Peninsula Laboratories, UK), glucagon, oxyntomodulin (Peninsula Laboratories, UK), glicentin-related pancreatic peptide (GRPP; kindly donated by L. Thim, Copenhagen, Denmark), glucagon-like peptide2 (GLP-2), gastric inhibitory peptide (GIP), vasoactive intestinal peptide (VIP), peptide YY (PYY) and peptide histidine isoleucine (PHI) (all peptides from Peninsula Laboratories, UK). Five hundred pl of GLP-1 (7-36) amide antiserum in the final dilution were incubated for 24 h by 4°C with the same volume
GLP-I CELLS
285
Table 1. List of antibodies used, working dilutions and sources Working dil. Antigen
Antibody
LM
GLP-I (7-36) amide glicentin glucagon insulin somatostatin PP PYY neurotensin serotonin VIP
monoclonal (mouse) monoclonal (mouse) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit)
I :8000 I :8000 1 : 4000 I : 4000 I : 4000 1 :6000 I :3000 1 :3000 1 :8000 I :5000
EM
Source
(a) Behringwerke AG, Marburg, Germany. (b) M. Gregor, Berlin, Germany. (c) DAKO Corporation, Caprinteria, USA. (d) Peninsula Laboratories, Europe Ltd.. St Helens, Merseyside, UK. (e) INCSTAR, Stillwater, USA. (f) MILAB, Malmo, Sweden.
of peptide dissolved in phosphate buffer. Dot-blot analysis was carried out according to Towbin and Gordon [24].
immersion in Bouin’s solution (24 h) for light microscopy or in Karnovsky’s fluid (2 h) for electron microscopy.
Other antisera
Light microscopy
The glicentin antibody was generously provided by Professor M . Gregor, Klinikum Steglitz, Berlin. The immunocytochemical characterization of this antibody has been reported previously [25]. The other antibodies used for the co-localization studies were commercially available and well characterized. Working dilutions and sources of all antisera are listed in Table 1.
After fixation the tissue was embedded in paraplast. Five pm sections were collected onto poly L-lysine coated glass slides, deparaffinized, rehydrated, incubated overnight with the specific antisera and then processed according to the avidin-biotin-complex method [26]. For co-localization studies of peptides the mirror image mounting procedure was used [22]. Briefly, serial sections were mounted on glass slides in a way that the second sections were turned around and identical but specular surfaces could be stained for immunocytochemistry by two different antisera. Light microscopic sections were viewed by bright field illumination and pictures were taken by a Zeiss photomicroscope I1 (Zeiss, Oberkochem, Germany).
Tissue preparation Animal tissue. Tissue specimens of 5 female Sprague Dawley rats (body. wt. 200-250 g) were removed from exactly defined locations in the distal esophagus, antrum, oxyntic mucosa, duodenum, proximal and distal jejunum, ileum, coecum, colon ascendens and transversum, rectum and pancreas. For quantitative measurements the gut tissue was pinned on a cork plate without distension. Tissue from the same locations as in rats was saved from five pigs which was supplied by a local slaughterhouse. Porcine and rat tissue was fixed for light microscopy by immersion in Bouin’s solution for 24 h. Human tissue. Tissue specimens from the distal esophagus, corpus/fundus region and antrum of the stomach, duodenum, proximal and distal jejunum, ileium, coecum, colon transversum, rectum and pancreas were obtained from patients with carcinoma or Crohn’s disease after surgery. Special care was taken to investigate only normal appearing tissue specimens without any pathological changes. For morphometry specimens of five individuals were analysed from each location, respectively. The tissue was either fixed by
Electron microscopy
Tissue specimens from human gut and pancreas were postfixed in osmium 1% for 1 h and embedded in Epon. The ultrathin sections were pretreated for 60 min with a saturated aqueous solution of sodium metaperiodate [27] before incubation with the antihormone sera. Immunoelectromicroscopic studies were performed by an indirect postembedding colloidal gold technique. The sections were incubated with the primary antibodies GLP 1 (7-36) amide, glicentin, glucagon and PYY for 2 h at room temperature (dilutions see Table 1).
Species specific colloidal gold (15 nm) coated secondary antibodies were purchased from Janssen Pharmaceutics, Belgium. After staining, the sections were contrasted with uranyl acetate and lead citrate and viewed by a Zeiss EM 109 electron microscope.
286
R. EISSELE et al.
Table 2. Specificity tests of the GLP-I antibody BW 53/018 by immunocytochemistry (immunoreaction after preabsorption with the respective antigen) and Dot-blot analysis
Antigen
lmmunocytochemistry
Dot blot
GLP-I (7 36) amide GLP-I ( I -36) amide GLP-2 glucagon oxyntomodulin GRPP PYY VIP PHI GIP
negative negative
+++ +++
+++ +++ +++ +++ +++ +++ +++ +++
negative negative negative negative negative negative negative negative
For ultrastructural co-localization of peptides ultrathin serial sections were used. Furthermore, the above described indirect colloidal method was applied as a double labelling technique when the primary antibodies were raised in different species. The grids were incubated with two different antisera and secondary antibodies with 5 nm and 15 nm particle seizes of the gold probes were used.
Figure 1. GLP-I immunoreactivity in the human pancreas. The GLP-I positive cells are mainly located in the periphery of the islets. Single cells are also distributed among the exocrine acini ( ). x 600. Bar: 25 pm.
Morphometrical analysis
Quantitative evaluations of GLP- 1 and glicentin immunoreactivities were performed in esophagus, stomach, jejunum, ileum and colon of all three species. For estimation of cell density only sections were used which were cut vertically to the mucosal surface. Sectioning was repeated if no vertical cut was achieved. A grid was placed parallel to the lamina muscularis mucosa and only cells with visible nuclei were counted in the area underlying the grid with a Zeiss-Photo Microscope I1 (Zeiss, Oberkochem, Germany). The length of a mucosa cylinder was 0.251 mm (magnification 8 x 1.6 x 25 = 320). The area of the corresponding mucosa cylinder was measured by a Zeiss Morphomat M 30. At least 25 visual fields were counted in rat and porcine tissue and 15 fields in human specimens. Data of the cell density measurements were expressed as cell number per mm2 mucosal area.
Figure 2. Co-localization of GLP- I (a) and glucagon (b) in an islet of rat pancreas. By the mirror image technique immuno-reactions of the two antisera on specular surfaces demonstrate that both peptides are located in the same cells in the periphery of the islet. x 680. Bar: 25 pm.
Results Characterization of the GLP-1 antibody
The monoclonal antibody BW 53/018 (IgG I ) directed against the C-terminal part of GLP-1 proved to give best staining results for immunocytochemistry. The optimal dilution was 1 : 8000 although after dilutions up to 1 : 20 000 still specific staining occurred. Positive immunoreaction was achieved both on Bouin (light microscopy) and Karnovsky (electron microscopy) fixed tissue. Furthermore, osmification did not impair the immunogold reaction. Preabsoption of the GLP-1 antibody by GLP- 1 (7-36) amide and GLP- 1 ( I -36) amide resulted in a negative reaction which indicated
Figure 3. Ultrathin section of an A-cell in human pancreas. The immunogold reaction with the GLP-I antiserum shows that thegold particles are mainly located above the dense core of the a-granule. x 3 I 400. Bar: 500 nm.
GLP-I CELLS Table 3. Cell densities of GLP-l and glicentin in the gut of man (A). pig ( B ) and rat (C)
GLP-I oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens C . transversum rectum
oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens c. transversum rectum
oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens c. transversum rectum
Glicentin
S.C.
S.C.
-
~~
S.C. 1.20f0.20 4.70 f I .34 4.55 2. I6 0.79 f0.16 2.05 0.62 8.06 5.03
S.C.
2.23 & 1 .OO 6.45 f I .33 5.60 f I .70 148f 1.10 3.70 I .08 8.68 f5.00
+ + +
GLP- I
Glicentin
S.C.
S.C.
-
-
S.C. 0.25f0.14 9.76 f4.77 8.9 I f0.53 3.68 f0.59 4.73 k0.46 6.04 f I .30
S.C.
0.25 f0.07 10.34f 5.60 1144f1.61 4.77 f0.85 8.34 f I .oo 13.98& 2.00
GLP-I
Glicentin
~
~
S.C. -
0. I6 0.09 0.20 f0.16 10.30f 1.25 I 2.78 f I 4 7 12.75 +2.16 15.49 f4.64 5.99 & I .24
S.C. ~
0,26+0.05 0.29 k0.05 4.97 f 0 . 9 9 I I .76 f2.75 10.00f2.45 13.93 f3.47 9.57 f2.19
specificity of the antibody. In addition, preabsorption studies with GLP-2, glucagon, oxyntomodulin, GRPP or any of the other tested peptides did not influence the specific immunoreaction in immunocytochemistry. In the Dot-blot analysis the antibody showed a positive reaction with GLP-I (7-36) amide and GLP-I (1-36) amide but did not cross-react with the other peptides tested (Table 2). Pancreas
In the pancreas of all three species the GLP-I antibody revealed variously shaped cells predominantly located in the periphery of the islets of Langerhans (Figs 1,2a). However, the pattern in human pancreatic islets differed markedly from the rat islets. It consisted of small lobules of insulin cells surrounded by A-cells. Single positive endocrine cells were also distributed among the exocrine acini and within ductular walls (Fig. 1). GLP-I was co-localized with glucagon in these cells
287
(Fig. 2). However, GLP-I was never found in insulin (B-), somatostatin (D-) or pancreatic polypeptide (PP-) positive cells. Ultrastructurally, GLP- 1 was localized by the immunogold technique in the agranula of the A-cells. The gold particles were largely restricted to the electrodense core of the a-granula (Fig. 3). GLP-1-IR and glucagon-IR showed the same distribution pattern in serial sections. Intestine
Cells stained with the GLP-1 antibody occurred in all part of the crypts and in the villous epithelium with a predominance in the basal portion in the intestine of all three species (Figs 4a, 5). They were of the open type and seemed to reach the lumen via a slender apical process (Fig. 4b). No GLP- 1 immunoreactive cells could be seen in the oesophagus and antral part of the stomach. In the corpus/fundus very few cells were stained in man and pig. More positive cells were found in the oxyntic mucosa of the rat. In the duodenum no immunostaining occurred in human and porcine tissue, whereas in rats single GLP1 positive cells could be seen. The density of GLP-I positive cells varied among the different locations of the small and large bowel and among the three species tested (Fig. 6; Table 3). Most cells were found in rats, followed by porcine and human tissue. In human and porcine gut the distribution of GLP-I positive cells was very similar. A large number of cells occurred in the distal jejunum and in the ileum. There was a continuous increase of cell density from the proximal to the distal colon and the number of GLP-I immunoreactive cells was even higher in the rectum than in the ileum. In rats cell density was comparatively high in the ileum and colon. However, on the contrary to man and pig, the number of GLP-I positive cells decreased in the rectum. Comparison of GLP- 1 and glicentin immunoreactive cells showed an almost identical distribution (Figs 6,7). In man and rats more cells were positive for GLP- I than for glicentin, whereas in porcine tissue this ratio was the other way around. The only exception was the rectum of rats: there was considerably more glicentin than GLP-I positive cells. Doubling labelling studies revealed that most cells stained with the GLP-I antibody were also positive for glicentin. However, a minority of cells were only labelled by either GLP- 1 or glicentin. Co-localization studies with GLP-I and PYY showed a number ofcells positive for both peptides but there were cells stained only either by the GLP-I or by the PYY antibody. No co-localization was found in the intestine with neurotensin, somatostatin or VIP. By electron microscopy GLP- I immunoreactive cells appeared to be L-cells with the typical large granules. The distribution of gold particles within the granula was homogenous for GLP1 (Figs 8a, 9), glicentin and PYY (Figs 8b, 9). Interestingly, PYY immunoreaction showed that only
288
R. EISSELE et al.
Figure 4. GLP-I immunoreactive cells in the human rectum. The cells occur on all parts of the crypts with a predominance in the basal part (a). They reach the lumen via a slender apical process (b). (a) x 410 and (b) x 680. Bar 25 pm.
0human Figure 5. GLP- I immunoreactive cells in porcine rectum. x 520. Bar: 25 pn.
PlQ
=
rat
.,
Figure Density of glicentinimmunoreactive cells in various part of the human, porcine and rat gastrointestinal tract. Data show means f s. Cell density is defined as cells per mm2 mucosal area.
Discussion
human
PlQ
=
rat
Figure 6. Density of GLP- I immunoreactive cells in various parts of the human, porcine and rat gastrointestinal tract. Data show means & s. Cell density is defined as cells per mm2 mucosal area.
a part of the granula within a cell was labelled, whereas nearly all granula were labelled for GLP-1 and glicentin.
In this study, the monoclonal antibody BW 53/0 18 bound to GLP-I (7-36) amide and GLP-I (1-36) amide but did not cross-react with other proglucagonderived peptides like glucagon, GLP-2, oxyntomodulin and GRPP. Furthermore, we observed no crossreaction with other structurally related peptides like VIP, PHI, GIP and PYY. This proves the high specificity of our monoclonal antibody. Previously, it was demonstrated that the amino acid sequence of GLP-1 is identical in several mammals including rat, pig and human [5]. Therefore, this specific antibody is a suitable tool to study the distribution of GLP-1-like immunoreactivity in the gastrointestinal tract and pancreas of rat, pig and human. We identified GLP-I in cells which were identical with A-cells. This was true for all three investigated species. Furthermore, we found single GLP-1-IR positive cells among the exocrine acini and within ductular walls. However, exocrine cells and duct cells remained unstained. GLP-1-IR and glucagon-IR were localized in the
GLP-I CELLS
289
Figure 8. Ultrathin section of a L-cell in the human ileum. The immunogold reaction for GLP-I (a) and PYY (b) shows a homogenous distribution of gold particles above thegranula. However, only part of thegranula is labelled for PYY (b) whereas nearly all granules are labelled for GLP-I (a). ( a ) x 31 400 and (b) x 18800. Bar: 500 nm.
Figure 9. Ultrathin section of a L-cell of the human ileum. Double labelling by the immunogold reaction: GLP-I ( I S nm particle size) and PYY (5 nm particle size). Both peptides are homogenous distributed in the granule. x 220000. Bar: 500 nm.
granula of the A-cells. Previous studies suggested that proglucagon-derived peptides are stored in separate compartments of such secretory granules [ 1,221. Glucagon-IR was identified in the electrondense core and glicentin and/or GRPP-like-IR in the electronlucent halo of the secretory granule. In support of these results, we found a corresponding distribution of GLP-I-IR in the granula. This pattern was not different for glucagon-IR and GLP-I-IR in serial sections. Thus, our findings support the hypothesis that in secretory granules of the A-cells proglucagonderived peptides are stored in different compartments. In the gastrointestinal tract GLP-I-IR was present in A-like cells in the corpus/fundus region and in Lcells in the small and large bowel. However, the regional amounts of GLP- I -1R in the gastrointestinal tract differed among the three investigated species. Most cells containing GLP-I-IR were found in rats followed by porcine and human tissue. In the corpus/ fundus region of the stomach more GLP- I -1R containing cells were found in rat than in human and pig.
Similarly, only in rat but not in human or porcine duodenum single GLP- I -IR positive cells were identified. Within the gut of all three species highest concentrations of GLP- I -IR were found in the lower intestine, particularly in ileum and colon. Similar results have been recently reported from canine tissue extracts (28). Interestingly, in contrast to rat and porcine rectum, in the human rectum the number of GLP- 1 -1R containing cells was significantly higher than in the ileum. The first communication to demonstrate ‘glucagon immunoreactivity’ in the human rectum was published more than 15 years ago [34]. However, the biological meaning of these observations still has to be explained. GLP- 1 -1R was co-localized with glicentin- and PYY-IR in the L- cells. In contrast to pancreatic A-cell granula there was no evidence that proglucagonderived peptides are stored in different compartments within the granula of L-cells. Furthermore, peptides like neurotensin, somatostatin or VIP were not found to be co-localized with GLP-I-IR in L-cells. I t has been demonstrated that the direct effects of nutrients and their metabolites on the pancreas cannot account adequately for regulation of the endocrine secretions of the gland. I t is clear that absorption of nutrients from the gut is accompanied by patterns of secretion of pancreatic hormones that differ from those resulting from parenteral alimentation. Such entero-insular effects are most obvious in the response to glucose ingestion [29]. GLP-I (7-36) amide, in addition to G I P [29], is now considered to be an important hormonal signal (‘incretin’) in this mechanism since it is released in response to oral glucose and potently stimulates insulin secretion (7-1 3). GIP-IR is mainly localized in the duodenum and proximal jejunum which allows us to easily understand its prompt release after glucose ingestion. In contrast hereto, highest concentrations of GLP-I -IR were shown in the lower intestine and colon which is somewhat puzzling considering the glucose-dependent
290
R. EISSELE et al.
secretion of GLP-I. This might indicate that glucose itself is not a direct stimulator of GLP-I-release. One possibility is that the release of GLP-I from the lower intestine is mediated by nerval reflexes in response to oral glucose. On the other hand, the GLP-I-releasing cells in the small intestine are ideally positioned in the distal gut to sense and respond to the presence of nutrients that have escaped the absorptive surface of the upper small intestine. It seems that further experiments are necessary to fully understand the mechanisms leading to the release of GLP-1 in response to glucose. In any case, the accumulation of GLP-Isecreting cells in the lower small intestine could at least have important pathophysiological implications. Miholic ef al. [30] reported very recently data from patients with an early dumping syndrome which suggested that an exaggerated GLP-I -release, induced by rapid transit of nutrients to the distal small bowel is a potent stimulus for hyperinsulinemia and contributes to reactive hypoglycemia after total gastrectomy. However, the pattern of localization could explain some findings on the effects of GLP-I on insulin secretion and gene regulation in the pancreatic B-cell. There is solid evidence that several intestinal hormones are involved in the regulation of the enteroinsulinar-axis [29]. A combination of GLP-1 (7-36) amide and GIP exerts an additive synergistic effect on insulin secretion which was only observed when submaximal effective hormone concentrations were utilized (31). It seems possible to speculate that an early release of GIP modulates the first phase of the postprandial insulin answer which is then consecutively followed by an effect which results out of the combination of already declining GIP-levels and increasing GLP-I-levels. The more need exists for a prolonged or stronger insulin answer in the second phase the more GLP-I would be released from the lower small intestine. Furthermore, since GLP-1 (736) amide was reported to increase insulin mRNA transcript in islet tissue (32) it is possible to believe that in the late phase of the insulin response under GLP-1 stimulation a regulation of gene transcription or translation efficiency occurs which facilitates an adequate re-equilibration of the endocrine pancreas. However, these latter ideas await further proof. Acknowledgments
This study was supported by DFG grants Go 429/2-1 and Go 417/3-1. We thank Professor M. Gregor (Berlin, Germany) for generously supplying the glicentin antibody. The skilful technical assistance of Mrs B. Kranz is appreciated. References 1 Varndell IM, Bishop AE, Sikri KL, Uttenthal LO, Bloom SR,
Polak JM. Localization ofglucagon-likepeptide-I (GLP) immunoreactants in human gut and pancreas using light and electron
microscopic immunocytochemistry. J Histochem Cytochem 1985;33:1080-6. 2 Moody AJ. Gut glucagon-like immunoreactivity.Clin Gastroenterol 1980;9:699-709. 3 Drucker DJ. Glucagon and the glucagon-like peptides. Pancreas 19903484-8. 4 Mojsov S, Heinrich G, Wilson IB, Ravazzola M, Orci L, Habener J Preproglucagon gene expression in pancreas and intestine diversifiesat the level of post-translational processing. J Biol Chem 1986;261:1180-9. 5 Goke R, Fehmann HC, Goke B. Glucagon-like peptide-l (7-36) amide is a new incretin/enterogastrone candidate. Eur J Clin Invest 1991;21:135-44. 6 Orskov C, Bersani M, Johnsen AH, Hojrup P, Holst JJ. Complete sequences of glucagon-like peptide- I from human and pig small intestine. J Biol Chem 1989;264-12826-9, 7 Holst JJ, Orskov C, Vagn Nielsen 0, Schwartz TW. Truncated glucagon-like peptide- I , an insulin releasing hormone from the distal gut. FEBS Lett 1987;211:169-74. 8 Mojsov S, Weir G, Habener JF. Insulinotropin: glucagon-like peptide l(7-37) co-encoded in the glucagon gene is a potent stimulator of insulin release in the perfused rat pancreas. J Clin Invest I987;79:616-9. 9 Kreymann B. Ghatei MA, Williams G, Bloom SR. GlUgdgOnlike peptide-I (7- 36): A physiological incretin in man. Lancet 1987; ii: 1300-4. 10 Orskov C, Holst JJ, Vagn Nielsen 0. Effect of truncated glucagon-like peptide-I (proglucagon-(78- 107)amide) on endocrine secretion from pig pancreas, antrum, and non-antral stomach. Endocrinology 1988;123:2009-13. I I Shima K, Hirota M, Ohboshi C. Effect of glucagon-like peptide1 on insulin secretion. Regul Pep 1988;22:245-52. 12 Weir GC, Mojsov S, Hendrick GK, Habener JF. Glucagon-like peptide I(7-37) actions on endocrine pancreas. Diabetes 38: 1989:338-42. 13 Goke R, Fehmann HC, Richter G, Trautmann M, Goke B. Interaction of glucagon-like peptide (7-36) amide and somatostatin-I4 in RINm5F cells and in the perfused rat pancreas. Pancreas 1989;4:668-73. 14 Komatsu R, Matsuyama T, Namba M, Watanabe N, ltoh H, Kono N, Tarui S. Glucagonostatic and insulinotropic action of ghcagon-like peptide I-(7-36)-amide. Diabetes 1989:38:902-5. 15 Kawai K, Suzuki S, Ohashi S, Mukai H, Ohmori H, Murayama Y, Yamashita K. Comparison of the effects of glucagon-like peptide-l(l 37) and -(7-37) and glucagon on islet hormone release from isolated perfused canine and rat pancreas. Endocrinology 1989;124:1768-73. 16 Hansen AB, Gespach CP, Rosselin GE, Holst JJ. Effect of truncated glucagon-like peptide 1 on CAMPin rat gastric glands and HGT-I human gastric cancer cells. FEBS Lett 1988;236:119-22. 17 Schjoldager BTG, Mortensen PE, Christiansen J, Orskov C, Holst JJ. GLP-I (ghcagon-like peptide I ) and truncated GLP-I. fragments of human proglucagon, inhibit gastric acid secretion in humans. Dig Dis Sci 1989;34:703 8. 18 Kauth T, Metz J. Immunohistochemical localization of glucagon-like peptide I . Use of poly- and monoclonal antibodies. Histochemistry 1987;86:509-15. 19 Uttenthal LO, Ghiglione M, George SK, Bishop AE, Polak JM. Bloom SR. Molecular forms of glucagon-like peptide-I in human pancreas and glucagonomas. J Endocrinol Metabol 1985;61:772-9. 20 George SK, Uttenthal LO, Ghiglione M, Bloom SR. Molecular forms of glucagon-like peptides in man. FEBS Lett 1985;192:275-8. 21 Orskov C, Holst JJ, Knuthsen S,Baldissera FGA, Poulssen SS. Vagn Nielsen 0. Glucagon-like peptides GLP-I and GLP-2, predicted products of the glucagon gene, are secreted separately from pig small intestine but not pancreas. Endocrinology 1986;119:1467-75. 22 Ali-Rachedi A, Varndell IM, Adrian TE, Gapp DA, Van Noorden S, Bloom SR, Polak JM. Peptide YY (PYY) immunoreactivity is costored with glucagon-rlated immunoreactants in
GLP-1 CELLS
23
24 25 26
27 28
endocrine cells of the gut and pancreas. Histochemistry 1984;80:487-91. Ishikawa E. lmagawa M, Hashida S, Yoshitake S. Hamaguchi Y, Ueno T. Enzyme-labeling of antibodies and their fragments for enzyme immunoassay and immunohistochemical staining. J lmmunoassay 1983;4:209-327. Towbin H, Gordon J. lmmunoblotting and dot immunoblottiny-Current status and outlook. J Immunological Methods 1984;72:313 40. Gregor M, Riecken EO. Production and characterization of Nterminally and C-terminally directed monoclonal antibodies against pancreatic glucagon. Gastroenterology 1985;89:571-80. Hsu SM, Raine L. Fanger H. Use of avidin-biotin complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 1981;29:577 80. Bendayan M. Zollingr M. Ultrastructural localization of antigenic sites on osmium fixed tissues applying the protein-A gold technique. J Histochem Cytochem 1983;31:101-9. Namba M, ltoh H, Watanabe N. Kono N, Komatsu R, Matsuyama T. Hirota M. Shima K. Tarui S. Multiple forms of glucagon-like peptide- I and glucagon-like immunoreactivities in canine gaslroinlestinal tract and their release into circulation. Bioined Res 1990;11:247-54.
291
29 Creutzfeldt W. The incretin concept today. Diabetologia 1979;16:75-85. 30 Miholic J, Orskov C.. Meyer H-J, Holst JJ. Glucagon-like peptide- I , early dumping, and hypoglycemia after total gastrectomy. Gastroenterology 1991;100:A156. 31 Fehmann HC, Coke B, C o k e R, Trautmann ME, Richter G. Arnold R. Synergistic stimulatory effect of glucagon-like peptide-1 (7-36) amide and glucose-dependent insulin-releasing polypeptide on the endocrine rat pancreas. FEBS Lett 1989;252:109-12. 32 Drucker DJ, Philippe J, Mojsov S, Chick WL, Habener JF. Glucagon-like peptide- I stimulates insulin gene expression and increases cyclic A M P levels in a rat islet cell line. Proc Natl Acad Sci USA 1987;84:3434-38. 33 Ravazzola M, Orci L. Glucagon and glicentin immunoreactivity are topologically segregated in the alpha granule of the human pancreatic A cell. Nature 1980;284:66-7. 34 Knudsen JB, Holst JJ, Asnas S, Johansen A. Identification of cells with pancreatic-type and gut-type glucagon immunoreactivity in the human colon. Acta Path microbiol Scand 1975:83:741-3.
Glucagon-like peptide-I cells in the gastrointestinal tract and pancreas of rat, pig and man R. EISSELE, R. GOKE, S. WILLEMER, H.-P. HARTHUS*, H. VERMEER*, R. ARNOLD & B. GOKE Department of Internal Medicine, Division of Gastroenterology and Metabolism, Philipps University of Marburg and Research Laboratories of Behringwerke AG*, Marburg, Germany Received 21 June 1991 and in revised form 2 October 1991; accepted 22 October 1991
Abstract. A highly specific monoclonal antibody directed against the C-terminal part of glucagon-like peptide-I (GLP- I ) was raised to immunohistochemically evaluate the distribution of GLP-I containing cells in the entire gastrointestinal tract including pancreas of rat, pig and man. In the pancreas GLP-Iimmunoreactive cells were found variously shaped and predominantly located in the periphery of the islets. Ultrastructurally, GLP- 1 was co-localized with glucagon in the a-granula of A-cells and was mainly restricted to the electrondense core. In the intestine open type cells reaching the lumen via a slender apical process were stained with the GLP-I antibody. They occurred in all parts of the crypts but predominantly in the basal portion. The density of GLP-I immunoreactive cells varied between species in a characteristic order: rat > pig > man. In pig and human gut a large number of cells occurred in the distal jejunum and ileum. A continuous increase of cell densities was found from the proximal to the distal colon resulting in highest numbers in the rectum. In rats the highest cell density occurred in the ileum. Again, a continuous increase of GLP-I-positive cell numbers was evident from the proximal to the distal portion of small and large bowel. GLP-I was partly co-localized with PYY. The GLP- 1 positive cells appeared electronmicroscopically as L-cells with the typical large granula. This morphological data indicates that GLP- I-releasing cells in the small intestine are appropriately positioned in the distal part to sense and respond to the presence of nutrients that have escaped the absorptive surface of the upper small intestine. Abbreviations GLP-I : Gucagon-like peptide-I, L-cell: endocrine cell in the intestine with large granules, A-cell: ‘glucagon cell’ in the islet of Langerhans, PYY: Peptide YY; GIP: Gastric inhibitory polypeptide, VIP: Vasoactive intestinal peptide; GRPP: Glicentin-related pancreatic peptide. Correspondence: Dr R.Coke, Laboratory of Molecular Endocrinology, Division of Gastroenterology, Department of Internal Medicine, Philipps University of Marburg. Baldingerstrape, W-3550 Marburg. Germany.
Introduction
Gut glucagon-like immunoreactants (gut GLIs) are located in the L-cell of the intestinal mucosa [I]. The greatest concentration of L-cells is detected in the distal small intestine and colon [2]. In recent years cloning and sequence analysis of cDNA and DNA fragments from genomic libraries has led to a dramatic increase in understanding of glucagon-related peptides [3]. The complex structural connections between the multiple molecular forms of the glucagon-like peptides in tissues have been determined [4,5]. The preproglucagon mRNAs from pancreas and intestine are identical, but the post-translational processing of the primary transcript differs markedly in the two tissues [4,5]. In the pancreas proglucagon is predominantly processed to proglucagon 1-30 (glicentin related pancreatic peptide; GRPP), glucagon and a carboxy-terminal fragment containing the sequences of glucagon-like peptide-I (GLP- I ) and glucagon-like peptide-2 (GLP2). In the gut proglucagon is mainly processed to proglucagon 1-69 (glicentin) and two smaller peptides, glucagon-like peptide-I (GLP-I) and glucagon-like peptide-2 (GLP-2) [4,5]. GLP-I is present in at least two forms (1-37 and 7-37). The truncated peptide is aamidated at the C-terminal end (6). GLP-1 (7-36) amide, which is secreted by the intestine, strongly stimulates insulin secretion, suppresses glucagon release, and inhibits gastric secretion [7- 171. Using polyclonal antibodies against GLP- 1 (1- 1 9) and GLP-1 (7-37), GLP-1-like immunoreactivity was identified in colorectal enteroglucagon cells [I, 181. Furthermore, GLP-I-like IR was described in pancreas and gut utilizing polyclonal antibodies antiGLP-1 (1-19) [1,19], and anti-GLP-l (I-37)/(1-36) amide [1,19,20,21]. Most of these studies were hampered by the circumstance that the antibodies used were only poorly characterized. Cross-reactivity with other important proglucagon-derived peptides or other members of the secretin family, to which GLP-I also belongs, was not studied. Therefore, in this study a specific monoclonal antibody against synthetic GLP- 1 (7-36) amide was developed and carefully character283
284
R. EISSELE et al.
ized to evaluate the distribution of GLP-I (7-36) amide containing cells in the gastrointestinal tract. In contrast to earlier reports, our study was extended to the entire gastrointestinal tract and pancreas of three different species, rat. pig, and man. Hereby, light and electron microscopic immunohistochemistry was employed. Furthermore, it was studied whether previous suggestions hold true that proglucagon-derived peptides are stored in separate compartments of the secretory granule [ 1,221. Data concerning the tissue-specific and/or species specific distribution of GLP-I-like IR could contribute to understand the regulation of GLP-I release and its physiological implications. We searched for a possible co-localization of GLP- 1 with other regulatory peptides. Such findings could help to understand tissuespecific processing of peptides and could give some insight into the specificity of stimulus-dependent peptide release from endocrine intestinal cells. Materials and methods Antisrra Preparation of monoclonal antibodies. GLP- 1 peptide/KLH conjugate. S-acetylmercaptosuccinic anhydrid (SAMSA) was used to introduce thiol groups into the peptide [23]. Glucagon-like peptide 1 (7-36) NH2 was obtained from Peninsula Laboratories (California, USA). Keyhole limpet hemocyanin (KLH) obtained from Calbiochem (La Jolla, USA) was dissolved in 0.1 M Li-phosphate containing 10% dioxan (pH 8.5) at a concentration of 10 mgml-I. To 40 mg of KLH was added 3.5 mg of gammamaleimidobutyric-succinimide (GMBS from Calbiochem, La Jolla, USA) dissolved in 180 p1 dioxan. After 30 min reaction at room temperature the reaction mixture was desalted on Sephadex G-25 into 0.1 M phosphate containing 5 mM EDTA (pH 6.0). The SH-peptide was reacted with the GMBS-KLH for 2 h at room temperature. After this time residual free SH-groups were blocked by reaction with 0.1 volumes of 0.1 M N-ethylmaleimide. The obtained conjugate was dialysed against physiological NaCl and used without further purification for immunization. Immunization. For the development of monoclonal antibodies against GLP-I a GLP-1/KLH-conjugate was used. Balb/c mice, 4-6 weeks old, were immunized subcutaneously and/or intraperitoneally with 20 pg of the GLP- 1 /KLH-conjugate emulsified in Freund’s complete adjuvant. Four weeks later, mice were boosted with 10 pg in incomplete Freund’s adjuvant followed by a third immunization 6 weeks later. Prior to fusion, mice were boosted intravenously four times at daily intervals. Spleen cells were isolated aseptically and were fused with the myeloma cell line SP 2/0 using polyethylene glycol (PEG) according to standard methods. The final cell pellet was resuspended in DMEM (Dulbecco’s Modified Eagle Medium) con-
taining 20% calf serum and HAT (0-1 mM hypoxanthin, 0.004 mM aminopterin, 0.0 16 mM thymidine) and added to 24-well culture plates (Nunc). About 14 days later, individual cell clones were picked out and were transferred to a new well. Three days later, supernatants were tested for antibody content as well as for the presence of GLP-I-specific antibodies. Positive cell cultures were grown up and were frozen in liquid nitrogen. In parallel, positive cell lines were cloned using a single cell manipulator. Screening of monoclonal antibodies (mabs) . Supernatants were screened using GLP-I-coated microtitration plates. Mabs bound to the solid phase were demonstrated by a second incubation with horseradish peroxidase (HRP) labelled rabbit anti-mouse antibodies. Relative affinity of anti-GLP-l mabs was determined by incubating defined concentrations of different antibody containing supernatants on GLP- 1 coated microtitration plates. Product ion and purijicat ion of monoclonul antibodies. Cell lines producing anti-GLP-1 antibodies were propagated in mass culture. The monoclonal antibody was purified by ammonium sulfate precipitation followed by affinity chromatography using Protein Asepharose (Pharmacia/LKB, Freiburg, Germany) according to the manufacturer. Purity was monitored by HPLC and SDS-PAGE (Pharmacia Phast System). Protein concentration was determined by optical density measurement at 280 nm. Class and subclass isotypes were determined by double diffusion techniques (Miles). The Protein A-purified antibody was stored at a protein concentration of 11.13 mg ml-’. Specijicity tests The newly raised monoclonal antibody to synthetic GLP-I (7-36) amide was tested for specificity and cross-reaction by immunocytochemistry and Dot-blot analysis. Immunocytochemical controls included the omission of the first antibody, the second antibody, the avidin-biotin complex and replacement of the first antibody by non-immune mouse antiserum. Increasing dilutions of the first antibody were tested ( 1 :50 to 1 :20000). In the preabsorption test the antibody was incubated with a 10,100 and 1000 fold excess of GLP-1 (7-36) amide (Peninsula Laboratories, Europe Ltd, St Helens, Merseyside, UK) and GLP-I (1-36) amide (Peninsula Laboratories, UK), glucagon, oxyntomodulin (Peninsula Laboratories, UK), glicentin-related pancreatic peptide (GRPP; kindly donated by L. Thim, Copenhagen, Denmark), glucagon-like peptide2 (GLP-2), gastric inhibitory peptide (GIP), vasoactive intestinal peptide (VIP), peptide YY (PYY) and peptide histidine isoleucine (PHI) (all peptides from Peninsula Laboratories, UK). Five hundred pl of GLP-1 (7-36) amide antiserum in the final dilution were incubated for 24 h by 4°C with the same volume
GLP-I CELLS
285
Table 1. List of antibodies used, working dilutions and sources Working dil. Antigen
Antibody
LM
GLP-I (7-36) amide glicentin glucagon insulin somatostatin PP PYY neurotensin serotonin VIP
monoclonal (mouse) monoclonal (mouse) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit) polyclonal (rabbit)
I :8000 I :8000 1 : 4000 I : 4000 I : 4000 1 :6000 I :3000 1 :3000 1 :8000 I :5000
EM
Source
(a) Behringwerke AG, Marburg, Germany. (b) M. Gregor, Berlin, Germany. (c) DAKO Corporation, Caprinteria, USA. (d) Peninsula Laboratories, Europe Ltd.. St Helens, Merseyside, UK. (e) INCSTAR, Stillwater, USA. (f) MILAB, Malmo, Sweden.
of peptide dissolved in phosphate buffer. Dot-blot analysis was carried out according to Towbin and Gordon [24].
immersion in Bouin’s solution (24 h) for light microscopy or in Karnovsky’s fluid (2 h) for electron microscopy.
Other antisera
Light microscopy
The glicentin antibody was generously provided by Professor M . Gregor, Klinikum Steglitz, Berlin. The immunocytochemical characterization of this antibody has been reported previously [25]. The other antibodies used for the co-localization studies were commercially available and well characterized. Working dilutions and sources of all antisera are listed in Table 1.
After fixation the tissue was embedded in paraplast. Five pm sections were collected onto poly L-lysine coated glass slides, deparaffinized, rehydrated, incubated overnight with the specific antisera and then processed according to the avidin-biotin-complex method [26]. For co-localization studies of peptides the mirror image mounting procedure was used [22]. Briefly, serial sections were mounted on glass slides in a way that the second sections were turned around and identical but specular surfaces could be stained for immunocytochemistry by two different antisera. Light microscopic sections were viewed by bright field illumination and pictures were taken by a Zeiss photomicroscope I1 (Zeiss, Oberkochem, Germany).
Tissue preparation Animal tissue. Tissue specimens of 5 female Sprague Dawley rats (body. wt. 200-250 g) were removed from exactly defined locations in the distal esophagus, antrum, oxyntic mucosa, duodenum, proximal and distal jejunum, ileum, coecum, colon ascendens and transversum, rectum and pancreas. For quantitative measurements the gut tissue was pinned on a cork plate without distension. Tissue from the same locations as in rats was saved from five pigs which was supplied by a local slaughterhouse. Porcine and rat tissue was fixed for light microscopy by immersion in Bouin’s solution for 24 h. Human tissue. Tissue specimens from the distal esophagus, corpus/fundus region and antrum of the stomach, duodenum, proximal and distal jejunum, ileium, coecum, colon transversum, rectum and pancreas were obtained from patients with carcinoma or Crohn’s disease after surgery. Special care was taken to investigate only normal appearing tissue specimens without any pathological changes. For morphometry specimens of five individuals were analysed from each location, respectively. The tissue was either fixed by
Electron microscopy
Tissue specimens from human gut and pancreas were postfixed in osmium 1% for 1 h and embedded in Epon. The ultrathin sections were pretreated for 60 min with a saturated aqueous solution of sodium metaperiodate [27] before incubation with the antihormone sera. Immunoelectromicroscopic studies were performed by an indirect postembedding colloidal gold technique. The sections were incubated with the primary antibodies GLP 1 (7-36) amide, glicentin, glucagon and PYY for 2 h at room temperature (dilutions see Table 1).
Species specific colloidal gold (15 nm) coated secondary antibodies were purchased from Janssen Pharmaceutics, Belgium. After staining, the sections were contrasted with uranyl acetate and lead citrate and viewed by a Zeiss EM 109 electron microscope.
286
R. EISSELE et al.
Table 2. Specificity tests of the GLP-I antibody BW 53/018 by immunocytochemistry (immunoreaction after preabsorption with the respective antigen) and Dot-blot analysis
Antigen
lmmunocytochemistry
Dot blot
GLP-I (7 36) amide GLP-I ( I -36) amide GLP-2 glucagon oxyntomodulin GRPP PYY VIP PHI GIP
negative negative
+++ +++
+++ +++ +++ +++ +++ +++ +++ +++
negative negative negative negative negative negative negative negative
For ultrastructural co-localization of peptides ultrathin serial sections were used. Furthermore, the above described indirect colloidal method was applied as a double labelling technique when the primary antibodies were raised in different species. The grids were incubated with two different antisera and secondary antibodies with 5 nm and 15 nm particle seizes of the gold probes were used.
Figure 1. GLP-I immunoreactivity in the human pancreas. The GLP-I positive cells are mainly located in the periphery of the islets. Single cells are also distributed among the exocrine acini ( ). x 600. Bar: 25 pm.
Morphometrical analysis
Quantitative evaluations of GLP- 1 and glicentin immunoreactivities were performed in esophagus, stomach, jejunum, ileum and colon of all three species. For estimation of cell density only sections were used which were cut vertically to the mucosal surface. Sectioning was repeated if no vertical cut was achieved. A grid was placed parallel to the lamina muscularis mucosa and only cells with visible nuclei were counted in the area underlying the grid with a Zeiss-Photo Microscope I1 (Zeiss, Oberkochem, Germany). The length of a mucosa cylinder was 0.251 mm (magnification 8 x 1.6 x 25 = 320). The area of the corresponding mucosa cylinder was measured by a Zeiss Morphomat M 30. At least 25 visual fields were counted in rat and porcine tissue and 15 fields in human specimens. Data of the cell density measurements were expressed as cell number per mm2 mucosal area.
Figure 2. Co-localization of GLP- I (a) and glucagon (b) in an islet of rat pancreas. By the mirror image technique immuno-reactions of the two antisera on specular surfaces demonstrate that both peptides are located in the same cells in the periphery of the islet. x 680. Bar: 25 pm.
Results Characterization of the GLP-1 antibody
The monoclonal antibody BW 53/018 (IgG I ) directed against the C-terminal part of GLP-1 proved to give best staining results for immunocytochemistry. The optimal dilution was 1 : 8000 although after dilutions up to 1 : 20 000 still specific staining occurred. Positive immunoreaction was achieved both on Bouin (light microscopy) and Karnovsky (electron microscopy) fixed tissue. Furthermore, osmification did not impair the immunogold reaction. Preabsoption of the GLP-1 antibody by GLP- 1 (7-36) amide and GLP- 1 ( I -36) amide resulted in a negative reaction which indicated
Figure 3. Ultrathin section of an A-cell in human pancreas. The immunogold reaction with the GLP-I antiserum shows that thegold particles are mainly located above the dense core of the a-granule. x 3 I 400. Bar: 500 nm.
GLP-I CELLS Table 3. Cell densities of GLP-l and glicentin in the gut of man (A). pig ( B ) and rat (C)
GLP-I oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens C . transversum rectum
oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens c. transversum rectum
oesophagus corpus antrum duodenum prox. jejunum dist. jejunum ileum c. ascendens c. transversum rectum
Glicentin
S.C.
S.C.
-
~~
S.C. 1.20f0.20 4.70 f I .34 4.55 2. I6 0.79 f0.16 2.05 0.62 8.06 5.03
S.C.
2.23 & 1 .OO 6.45 f I .33 5.60 f I .70 148f 1.10 3.70 I .08 8.68 f5.00
+ + +
GLP- I
Glicentin
S.C.
S.C.
-
-
S.C. 0.25f0.14 9.76 f4.77 8.9 I f0.53 3.68 f0.59 4.73 k0.46 6.04 f I .30
S.C.
0.25 f0.07 10.34f 5.60 1144f1.61 4.77 f0.85 8.34 f I .oo 13.98& 2.00
GLP-I
Glicentin
~
~
S.C. -
0. I6 0.09 0.20 f0.16 10.30f 1.25 I 2.78 f I 4 7 12.75 +2.16 15.49 f4.64 5.99 & I .24
S.C. ~
0,26+0.05 0.29 k0.05 4.97 f 0 . 9 9 I I .76 f2.75 10.00f2.45 13.93 f3.47 9.57 f2.19
specificity of the antibody. In addition, preabsorption studies with GLP-2, glucagon, oxyntomodulin, GRPP or any of the other tested peptides did not influence the specific immunoreaction in immunocytochemistry. In the Dot-blot analysis the antibody showed a positive reaction with GLP-I (7-36) amide and GLP-I (1-36) amide but did not cross-react with the other peptides tested (Table 2). Pancreas
In the pancreas of all three species the GLP-I antibody revealed variously shaped cells predominantly located in the periphery of the islets of Langerhans (Figs 1,2a). However, the pattern in human pancreatic islets differed markedly from the rat islets. It consisted of small lobules of insulin cells surrounded by A-cells. Single positive endocrine cells were also distributed among the exocrine acini and within ductular walls (Fig. 1). GLP-I was co-localized with glucagon in these cells
287
(Fig. 2). However, GLP-I was never found in insulin (B-), somatostatin (D-) or pancreatic polypeptide (PP-) positive cells. Ultrastructurally, GLP- 1 was localized by the immunogold technique in the agranula of the A-cells. The gold particles were largely restricted to the electrodense core of the a-granula (Fig. 3). GLP-1-IR and glucagon-IR showed the same distribution pattern in serial sections. Intestine
Cells stained with the GLP-1 antibody occurred in all part of the crypts and in the villous epithelium with a predominance in the basal portion in the intestine of all three species (Figs 4a, 5). They were of the open type and seemed to reach the lumen via a slender apical process (Fig. 4b). No GLP- 1 immunoreactive cells could be seen in the oesophagus and antral part of the stomach. In the corpus/fundus very few cells were stained in man and pig. More positive cells were found in the oxyntic mucosa of the rat. In the duodenum no immunostaining occurred in human and porcine tissue, whereas in rats single GLP1 positive cells could be seen. The density of GLP-I positive cells varied among the different locations of the small and large bowel and among the three species tested (Fig. 6; Table 3). Most cells were found in rats, followed by porcine and human tissue. In human and porcine gut the distribution of GLP-I positive cells was very similar. A large number of cells occurred in the distal jejunum and in the ileum. There was a continuous increase of cell density from the proximal to the distal colon and the number of GLP-I immunoreactive cells was even higher in the rectum than in the ileum. In rats cell density was comparatively high in the ileum and colon. However, on the contrary to man and pig, the number of GLP-I positive cells decreased in the rectum. Comparison of GLP- 1 and glicentin immunoreactive cells showed an almost identical distribution (Figs 6,7). In man and rats more cells were positive for GLP- I than for glicentin, whereas in porcine tissue this ratio was the other way around. The only exception was the rectum of rats: there was considerably more glicentin than GLP-I positive cells. Doubling labelling studies revealed that most cells stained with the GLP-I antibody were also positive for glicentin. However, a minority of cells were only labelled by either GLP- 1 or glicentin. Co-localization studies with GLP-I and PYY showed a number ofcells positive for both peptides but there were cells stained only either by the GLP-I or by the PYY antibody. No co-localization was found in the intestine with neurotensin, somatostatin or VIP. By electron microscopy GLP- I immunoreactive cells appeared to be L-cells with the typical large granules. The distribution of gold particles within the granula was homogenous for GLP1 (Figs 8a, 9), glicentin and PYY (Figs 8b, 9). Interestingly, PYY immunoreaction showed that only
288
R. EISSELE et al.
Figure 4. GLP-I immunoreactive cells in the human rectum. The cells occur on all parts of the crypts with a predominance in the basal part (a). They reach the lumen via a slender apical process (b). (a) x 410 and (b) x 680. Bar 25 pm.
0human Figure 5. GLP- I immunoreactive cells in porcine rectum. x 520. Bar: 25 pn.
PlQ
=
rat
.,
Figure Density of glicentinimmunoreactive cells in various part of the human, porcine and rat gastrointestinal tract. Data show means f s. Cell density is defined as cells per mm2 mucosal area.
Discussion
human
PlQ
=
rat
Figure 6. Density of GLP- I immunoreactive cells in various parts of the human, porcine and rat gastrointestinal tract. Data show means & s. Cell density is defined as cells per mm2 mucosal area.
a part of the granula within a cell was labelled, whereas nearly all granula were labelled for GLP-1 and glicentin.
In this study, the monoclonal antibody BW 53/0 18 bound to GLP-I (7-36) amide and GLP-I (1-36) amide but did not cross-react with other proglucagonderived peptides like glucagon, GLP-2, oxyntomodulin and GRPP. Furthermore, we observed no crossreaction with other structurally related peptides like VIP, PHI, GIP and PYY. This proves the high specificity of our monoclonal antibody. Previously, it was demonstrated that the amino acid sequence of GLP-1 is identical in several mammals including rat, pig and human [5]. Therefore, this specific antibody is a suitable tool to study the distribution of GLP-1-like immunoreactivity in the gastrointestinal tract and pancreas of rat, pig and human. We identified GLP-I in cells which were identical with A-cells. This was true for all three investigated species. Furthermore, we found single GLP-1-IR positive cells among the exocrine acini and within ductular walls. However, exocrine cells and duct cells remained unstained. GLP-1-IR and glucagon-IR were localized in the
GLP-I CELLS
289
Figure 8. Ultrathin section of a L-cell in the human ileum. The immunogold reaction for GLP-I (a) and PYY (b) shows a homogenous distribution of gold particles above thegranula. However, only part of thegranula is labelled for PYY (b) whereas nearly all granules are labelled for GLP-I (a). ( a ) x 31 400 and (b) x 18800. Bar: 500 nm.
Figure 9. Ultrathin section of a L-cell of the human ileum. Double labelling by the immunogold reaction: GLP-I ( I S nm particle size) and PYY (5 nm particle size). Both peptides are homogenous distributed in the granule. x 220000. Bar: 500 nm.
granula of the A-cells. Previous studies suggested that proglucagon-derived peptides are stored in separate compartments of such secretory granules [ 1,221. Glucagon-IR was identified in the electrondense core and glicentin and/or GRPP-like-IR in the electronlucent halo of the secretory granule. In support of these results, we found a corresponding distribution of GLP-I-IR in the granula. This pattern was not different for glucagon-IR and GLP-I-IR in serial sections. Thus, our findings support the hypothesis that in secretory granules of the A-cells proglucagonderived peptides are stored in different compartments. In the gastrointestinal tract GLP-I-IR was present in A-like cells in the corpus/fundus region and in Lcells in the small and large bowel. However, the regional amounts of GLP- I -1R in the gastrointestinal tract differed among the three investigated species. Most cells containing GLP-I-IR were found in rats followed by porcine and human tissue. In the corpus/ fundus region of the stomach more GLP- I -1R containing cells were found in rat than in human and pig.
Similarly, only in rat but not in human or porcine duodenum single GLP- I -IR positive cells were identified. Within the gut of all three species highest concentrations of GLP- I -IR were found in the lower intestine, particularly in ileum and colon. Similar results have been recently reported from canine tissue extracts (28). Interestingly, in contrast to rat and porcine rectum, in the human rectum the number of GLP- 1 -1R containing cells was significantly higher than in the ileum. The first communication to demonstrate ‘glucagon immunoreactivity’ in the human rectum was published more than 15 years ago [34]. However, the biological meaning of these observations still has to be explained. GLP- 1 -1R was co-localized with glicentin- and PYY-IR in the L- cells. In contrast to pancreatic A-cell granula there was no evidence that proglucagonderived peptides are stored in different compartments within the granula of L-cells. Furthermore, peptides like neurotensin, somatostatin or VIP were not found to be co-localized with GLP-I-IR in L-cells. I t has been demonstrated that the direct effects of nutrients and their metabolites on the pancreas cannot account adequately for regulation of the endocrine secretions of the gland. I t is clear that absorption of nutrients from the gut is accompanied by patterns of secretion of pancreatic hormones that differ from those resulting from parenteral alimentation. Such entero-insular effects are most obvious in the response to glucose ingestion [29]. GLP-I (7-36) amide, in addition to G I P [29], is now considered to be an important hormonal signal (‘incretin’) in this mechanism since it is released in response to oral glucose and potently stimulates insulin secretion (7-1 3). GIP-IR is mainly localized in the duodenum and proximal jejunum which allows us to easily understand its prompt release after glucose ingestion. In contrast hereto, highest concentrations of GLP-I -IR were shown in the lower intestine and colon which is somewhat puzzling considering the glucose-dependent
290
R. EISSELE et al.
secretion of GLP-I. This might indicate that glucose itself is not a direct stimulator of GLP-I-release. One possibility is that the release of GLP-I from the lower intestine is mediated by nerval reflexes in response to oral glucose. On the other hand, the GLP-I-releasing cells in the small intestine are ideally positioned in the distal gut to sense and respond to the presence of nutrients that have escaped the absorptive surface of the upper small intestine. It seems that further experiments are necessary to fully understand the mechanisms leading to the release of GLP-1 in response to glucose. In any case, the accumulation of GLP-Isecreting cells in the lower small intestine could at least have important pathophysiological implications. Miholic ef al. [30] reported very recently data from patients with an early dumping syndrome which suggested that an exaggerated GLP-I -release, induced by rapid transit of nutrients to the distal small bowel is a potent stimulus for hyperinsulinemia and contributes to reactive hypoglycemia after total gastrectomy. However, the pattern of localization could explain some findings on the effects of GLP-I on insulin secretion and gene regulation in the pancreatic B-cell. There is solid evidence that several intestinal hormones are involved in the regulation of the enteroinsulinar-axis [29]. A combination of GLP-1 (7-36) amide and GIP exerts an additive synergistic effect on insulin secretion which was only observed when submaximal effective hormone concentrations were utilized (31). It seems possible to speculate that an early release of GIP modulates the first phase of the postprandial insulin answer which is then consecutively followed by an effect which results out of the combination of already declining GIP-levels and increasing GLP-I-levels. The more need exists for a prolonged or stronger insulin answer in the second phase the more GLP-I would be released from the lower small intestine. Furthermore, since GLP-1 (736) amide was reported to increase insulin mRNA transcript in islet tissue (32) it is possible to believe that in the late phase of the insulin response under GLP-1 stimulation a regulation of gene transcription or translation efficiency occurs which facilitates an adequate re-equilibration of the endocrine pancreas. However, these latter ideas await further proof. Acknowledgments
This study was supported by DFG grants Go 429/2-1 and Go 417/3-1. We thank Professor M. Gregor (Berlin, Germany) for generously supplying the glicentin antibody. The skilful technical assistance of Mrs B. Kranz is appreciated. References 1 Varndell IM, Bishop AE, Sikri KL, Uttenthal LO, Bloom SR,
Polak JM. Localization ofglucagon-likepeptide-I (GLP) immunoreactants in human gut and pancreas using light and electron
microscopic immunocytochemistry. J Histochem Cytochem 1985;33:1080-6. 2 Moody AJ. Gut glucagon-like immunoreactivity.Clin Gastroenterol 1980;9:699-709. 3 Drucker DJ. Glucagon and the glucagon-like peptides. Pancreas 19903484-8. 4 Mojsov S, Heinrich G, Wilson IB, Ravazzola M, Orci L, Habener J Preproglucagon gene expression in pancreas and intestine diversifiesat the level of post-translational processing. J Biol Chem 1986;261:1180-9. 5 Goke R, Fehmann HC, Goke B. Glucagon-like peptide-l (7-36) amide is a new incretin/enterogastrone candidate. Eur J Clin Invest 1991;21:135-44. 6 Orskov C, Bersani M, Johnsen AH, Hojrup P, Holst JJ. Complete sequences of glucagon-like peptide- I from human and pig small intestine. J Biol Chem 1989;264-12826-9, 7 Holst JJ, Orskov C, Vagn Nielsen 0, Schwartz TW. Truncated glucagon-like peptide- I , an insulin releasing hormone from the distal gut. FEBS Lett 1987;211:169-74. 8 Mojsov S, Weir G, Habener JF. Insulinotropin: glucagon-like peptide l(7-37) co-encoded in the glucagon gene is a potent stimulator of insulin release in the perfused rat pancreas. J Clin Invest I987;79:616-9. 9 Kreymann B. Ghatei MA, Williams G, Bloom SR. GlUgdgOnlike peptide-I (7- 36): A physiological incretin in man. Lancet 1987; ii: 1300-4. 10 Orskov C, Holst JJ, Vagn Nielsen 0. Effect of truncated glucagon-like peptide-I (proglucagon-(78- 107)amide) on endocrine secretion from pig pancreas, antrum, and non-antral stomach. Endocrinology 1988;123:2009-13. I I Shima K, Hirota M, Ohboshi C. Effect of glucagon-like peptide1 on insulin secretion. Regul Pep 1988;22:245-52. 12 Weir GC, Mojsov S, Hendrick GK, Habener JF. Glucagon-like peptide I(7-37) actions on endocrine pancreas. Diabetes 38: 1989:338-42. 13 Goke R, Fehmann HC, Richter G, Trautmann M, Goke B. Interaction of glucagon-like peptide (7-36) amide and somatostatin-I4 in RINm5F cells and in the perfused rat pancreas. Pancreas 1989;4:668-73. 14 Komatsu R, Matsuyama T, Namba M, Watanabe N, ltoh H, Kono N, Tarui S. Glucagonostatic and insulinotropic action of ghcagon-like peptide I-(7-36)-amide. Diabetes 1989:38:902-5. 15 Kawai K, Suzuki S, Ohashi S, Mukai H, Ohmori H, Murayama Y, Yamashita K. Comparison of the effects of glucagon-like peptide-l(l 37) and -(7-37) and glucagon on islet hormone release from isolated perfused canine and rat pancreas. Endocrinology 1989;124:1768-73. 16 Hansen AB, Gespach CP, Rosselin GE, Holst JJ. Effect of truncated glucagon-like peptide 1 on CAMPin rat gastric glands and HGT-I human gastric cancer cells. FEBS Lett 1988;236:119-22. 17 Schjoldager BTG, Mortensen PE, Christiansen J, Orskov C, Holst JJ. GLP-I (ghcagon-like peptide I ) and truncated GLP-I. fragments of human proglucagon, inhibit gastric acid secretion in humans. Dig Dis Sci 1989;34:703 8. 18 Kauth T, Metz J. Immunohistochemical localization of glucagon-like peptide I . Use of poly- and monoclonal antibodies. Histochemistry 1987;86:509-15. 19 Uttenthal LO, Ghiglione M, George SK, Bishop AE, Polak JM. Bloom SR. Molecular forms of glucagon-like peptide-I in human pancreas and glucagonomas. J Endocrinol Metabol 1985;61:772-9. 20 George SK, Uttenthal LO, Ghiglione M, Bloom SR. Molecular forms of glucagon-like peptides in man. FEBS Lett 1985;192:275-8. 21 Orskov C, Holst JJ, Knuthsen S,Baldissera FGA, Poulssen SS. Vagn Nielsen 0. Glucagon-like peptides GLP-I and GLP-2, predicted products of the glucagon gene, are secreted separately from pig small intestine but not pancreas. Endocrinology 1986;119:1467-75. 22 Ali-Rachedi A, Varndell IM, Adrian TE, Gapp DA, Van Noorden S, Bloom SR, Polak JM. Peptide YY (PYY) immunoreactivity is costored with glucagon-rlated immunoreactants in
GLP-1 CELLS
23
24 25 26
27 28
endocrine cells of the gut and pancreas. Histochemistry 1984;80:487-91. Ishikawa E. lmagawa M, Hashida S, Yoshitake S. Hamaguchi Y, Ueno T. Enzyme-labeling of antibodies and their fragments for enzyme immunoassay and immunohistochemical staining. J lmmunoassay 1983;4:209-327. Towbin H, Gordon J. lmmunoblotting and dot immunoblottiny-Current status and outlook. J Immunological Methods 1984;72:313 40. Gregor M, Riecken EO. Production and characterization of Nterminally and C-terminally directed monoclonal antibodies against pancreatic glucagon. Gastroenterology 1985;89:571-80. Hsu SM, Raine L. Fanger H. Use of avidin-biotin complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 1981;29:577 80. Bendayan M. Zollingr M. Ultrastructural localization of antigenic sites on osmium fixed tissues applying the protein-A gold technique. J Histochem Cytochem 1983;31:101-9. Namba M, ltoh H, Watanabe N. Kono N, Komatsu R, Matsuyama T. Hirota M. Shima K. Tarui S. Multiple forms of glucagon-like peptide- I and glucagon-like immunoreactivities in canine gaslroinlestinal tract and their release into circulation. Bioined Res 1990;11:247-54.
291
29 Creutzfeldt W. The incretin concept today. Diabetologia 1979;16:75-85. 30 Miholic J, Orskov C.. Meyer H-J, Holst JJ. Glucagon-like peptide- I , early dumping, and hypoglycemia after total gastrectomy. Gastroenterology 1991;100:A156. 31 Fehmann HC, Coke B, C o k e R, Trautmann ME, Richter G. Arnold R. Synergistic stimulatory effect of glucagon-like peptide-1 (7-36) amide and glucose-dependent insulin-releasing polypeptide on the endocrine rat pancreas. FEBS Lett 1989;252:109-12. 32 Drucker DJ, Philippe J, Mojsov S, Chick WL, Habener JF. Glucagon-like peptide- I stimulates insulin gene expression and increases cyclic A M P levels in a rat islet cell line. Proc Natl Acad Sci USA 1987;84:3434-38. 33 Ravazzola M, Orci L. Glucagon and glicentin immunoreactivity are topologically segregated in the alpha granule of the human pancreatic A cell. Nature 1980;284:66-7. 34 Knudsen JB, Holst JJ, Asnas S, Johansen A. Identification of cells with pancreatic-type and gut-type glucagon immunoreactivity in the human colon. Acta Path microbiol Scand 1975:83:741-3.
