Specific interactions of pancreatic amylase at acidic pH. Amylase and the major bind to immobilized or protein of the zymogen granule membrane (GP-2) polymerized amylase

'

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

M I C H ~ LJACOB, E JEANLAIN& AND DENISL E B E L ~ Centre de recherche sur les mkcanismes de secretion, Facult6 des sciences, Universitk de Sherbrooke, Sherbrooke (Quebec), Canada J I K 2RI Received February 19, 1992 JACOB,M., LAINE,J., and LEBEL.D. 1992. Specific interactions of pancreatic amylase at acidic pH. Amylase and the major protein of the zymogen granule membrane (GP-2) bind to immobilized or polymerized amylase. Biochem. Cell Biol. 70: 1105-1114. Regulated secretory proteins are thought to be sorted in the trans-Golgi network towards the secretory granule via acidic aggregation. In the exocrine pancreas, amylase is one of the major zymogens. It is a basic protein of pI 8.6 and does not precipitate in acidic conditions. To identify the mechanism by which amylase aggregates in the acidic cisternae of the pancreatic trans-Golgi network, we have developed an in vitro model in which amylase was fixed to plastic microtiter plates. The fixed amylase was probed with two ligands: amylase itself and GP-2, the major protein of the zymogen granule membrane. Biotinylated amylase bound to fixed amylase in a strict pH-dependent manner with optimal binding between pH 5.0 and 5.7. The affinity of binding was in the nanogram range (Kd = 20.0 ng/mL) at pH 5.5. Acid binding of amylase was not reversible by incubation at neutral pH, nor could it be displaced by native amylase. GP-2 binding to fixed amylase was also pH dependent with optimal binding between pH 5.0 and 5.7. As for amylase, it was not reversible by incubation at neutral pH. GP-2 binding sites on fixed amylase appeared to be different from those of biotinylated amylase. While native and biotinylated amylase did not bind to GP-2, polymerized amylase precipitated GP-2 at acidic pH. Taken together these data suggest that slight modifications are sufficient to reveal on the amylase molecule binding sites for GP-2 and for amylase itself. These new binding capacities acquired at acidic pH could be involved in the cascade of reactions that lead to the in vivo formation of the immature secretory granule. Key words: regulated secretion, sorting, granules, trans-Golgi network. JACOB,M., LAINE,J., et LEBEL,D. 1992. Specific interactions of pancreatic amylase at acidic pH. Amylase and the major protein of the zymogen granule membrane (GP-2) bind to immobilized or polymerized amylase. Biochem. Cell Biol. 70 : 1105-1 114. Les proteines de secretion de la voie contr6lk seraient trikes dans le rCseau trans-Golgi vers la formation d'un granule sCcrCteur par agrkgation acide. Dans le pancrCas exocrine, l'amylase est un des principaux zymogtnes. C'est une protCine basique de pI 8,6 et elle ne prtcipite pas dans des conditions acides. Dans le but d'identifier le mecanisme permettant l'agrkgation de I'amylase dans les citernes acides du reseau trans-Golgi pancrkatique, nous avons dkveloppk un modkle in vitro dans lequel l'amylase est fixCe A des plaques de plastique pour la microtitration. L'amylase fixke est examinee avec deux ligands : I'amylase elle-m&meet la GP-2, la principale proteine de la membrane des granules de zymogkne. L'amylase biotinylie se lie a l'amylase fixee d'une faqon strictement dependante du pH et la liaison est optimum entre pH 5,O et 5,7. L'affinite de liaison est dans la zone des nanogrammes (Kd = 20,O ng/mL) 1 pH 5,s. La liaison acide de I'amylase n'est pas reversible par incubation a pH neutre et elle n'est pas dkplacte par I'amylase native. La liaison de la GP-2 a I'amylase fixke est tgalement dependante du pH et elle est optimum entre pH 5,O et 5,7. Comme pour I'amylase, elle n'est pas reversible par incubation a pH neutre. Les sites de liaison de la GP-2 sur I'amylase fix& semblent diffkrents de ceux de l'amylase biotinylte. Alors que l'amylase native et biotinylke ne se lie pas a la GP-2, l'amylase polymCriste prkcipite la GP-2 a pH acide. Dans I'ensemble, ces resultats suggkrent que des modifications lkgtres sont suffisantes pour revtler les sites de liaison de la GP-2 et de l'amylase elle-mCme sur la moltcule d'amylase. Ces nouvelles capacites de liaison acquises a pH acide pourraient agir dans la cascade de rkactions qui conduisent a la formation in vivo du granule skcreteur immature. Mots cles : skcrktion contr61ke, triage, granules, rtseau trans-Golgi. [Traduit par la rtdaction]

Introduction In the events leading to the biogenesis of the secretion granule, sorting of proteins targeted to the regulated pathABBREVIATIONS: RER, rough endoplasmic reticulum; TGN, trans-Golgi network; BAM, biotinylated amylase; PBS, phosphatebuffered saline; GT, 1% fish gelatin and 0.05% Tween 20@;CPBS, citrate-supplemented phosphate-buffered saline; NSB, nonspecific binding; PEG, polyethylene glycol; ELISA, enzyme-linked immunosorbent assay; BCA, bicinchoninic acid; SDS-PAGE, sodium dodecyl sulfate - polyacrylamide gel electrophoresis; PVC, polyvinylchloride; M,, relative mass; OD, optical density. his is the 17th paper in a series "Elucidation of the mechanisms of cellular secretion." 2 ~ u t h oto r whom all correspondence should be addressed. Printed ~n Canada / Imorimb au Canada

way has been associated with their aggregation (Tooze et al. 1987). This aggregation leads to the state of high concentration of the secretory material that is observed in mature granules. In endocrine cells, regulated secretory proteins could be concentrated up to 200-fold between the RER and the mature granule. In pancreatic exocrine cells, the concentration factor is around 5-10 (Bendayan 1984). In parallel with their concentration, the secretory proteins are osmotically inactivated (Wong et al. 1991). The mechanism by which this phenomenon is accomplished is not entirely understood. Sulfated polyanions have been proposed to be involved in the concentration of zyrnogens in the exocrine pancreas, but owing to their very low concentration, are not believed to be accountable for the whole process (Reggio

BIOCHEM. CELL BIOL. VOL. 70,

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

1106

and Dagorn 1980). Another event taking place in the regulated secretory pathway between the TGN and the final condensation of the secretory material in specialized vacuoles is a vectorial acidification of the lumen of these cisterna: (Anderson and Orci 1988). The phenomenon is necessary for the normal operation of the regulated secretion (Burgess and Kelly 1987; De Lisle and Williams 1987) and would be involved in processes leading to the formation of mature secretory granules. In endocrine glands, such an acidification has been correlated with the in vivo aggregation of regulated secretory proteins in the TGN (Chanat and Huttner 1991). A direct relationship between acidification and protein aggregation has also been observed in vitro with pancreatic zymogens when granule content is reacidified (Rothman 1971). Most of the zymogens are found in these aggregates, including amylase. Amylase is the most abundant zymogen synthesized by the exocrine pancreas of the rat and pig. Under normal conditions, amylase could constitute up to 30% of the zymogens (Rausch et al. 1985). In this article we investigated a possible mechanism by which amylase may aggregate in vivo under conditions believed to exist from the TGN to the zymogen granule (Chanat and Huttner 1991; LeBel et al. 1988). We also address the question of how amylase aggregates and gets sorted to the secretory granule, since in vitro purified amylase does not precipitate at acidic pH. Indeed amylase is a basic protein of pI 8.6 (Reddy et al. 1987; Sanders and Rutter 1972; Vandermeers and Christophe 1968) and does not precipitate in acidic conditions. To determine the mechanism by which amylase might aggregate in the acidic cisterna: of the pancreatic secretory pathway, we have developed an in vitro model in which amylase was fixed to microtiter plates and probed with soluble ligands. Two presumptive ligands were tested, GP-2 and amylase itself. We show that when pure amylase is used as a binding matrix, BAM binds specifically to the fixed amylase in a pHdependent manner. Under similar conditions, GP-2, the major protein of the zymogen granule membrane, also binds to fixed amylase. However, soluble native amylase did not bind to fixed amylase nor to fixed GP-2. To allow binding of amylase or GP-2, amylase must be in a complex form (i.e., fixed to a substrate or as a polymer). These findings have provided us with clues on the mechanisms by which amylase could be sorted and packaged in the exocrine pancreas, and on a role played by GP-2 in the process. Materials and methods Purification of amylase from zymogen granule content Zymogen granule contents were prepared from rat and pig pancreas according to previously published procedures (LeBel and Beattie 1984; Paquette et al. 1986). Amylase was purified on Sephadexm G-75 by taking advantage of its affinity for the polymerized glucopyranose that lead to its retardation when eluted from the gel (Vandermeers and Christophe 1968). A 96 x 2.6 cm column equilibrated with 10 mM NaHPO, - 5 mM NaCl (pH 8.0) was eluted at a flow rate of 0.5 mL/min. Fractions containing pure amylase eluted at a volume larger than the total bed volume, an evidence for its interaction with the gel matrix. Fractions containing pure amylase were pooled and precipitated with 3.2 M ammonium sulfate final concentration. Kept at 4°C as a precipitate in 3.2 M (82% saturation) ammonium sulfate, amylase activty was stable for months. Biotinylation of amylase Biotin was coupled to purified amylase according to previously

1992

published procedures (Bayer and Wilchek 1980). Briefly, 10 pL of 10 mg N-hydroxysuccinirnide biotin/mL in dimethyl sulfoxide was added to 1 mL of 10 mg amylase/mL in PBS (pH 7.0) - 0.1 mM c a 2 + and incubated at room temperature for 4 h. Residual N-hydroxysuccinimide biotin was reacted by addition of 8 pL of 1 M NH,Cl. Labelled amylase was dialysed overnight at 4°C against PBS. Residual activity of the biotinylated amylase was 70.2 + 4.8% (n = 6).

Fixation of amylase on microwell plates and saturation of the remainder of binding sites with gelatin Except for the parameters tested to set up the assay as described in Results, fixation of amylase to plates was done according to the following procedure. Purified amylase (100 pL) was added to microwells at a concentration of 1.5 pg/mL in PBS (pH 6.9) and incubated at 37OC for 1 hand at room temperature for 3 h. Plates were then washed twice with PBS (pH 6.9) supplemented with 1% fish gelatin and 0.05% Tween 2@ (PBS + GT buffer). Plates were then saturated overnight at room temperature with 3% fish gelatin in PBS and after they were washed five times with PBS + GT (pH 6.9). Plates could be kept for 2 weeks at 4OC in plastic sealed bags in 100% humidity without loss of binding capacity. Binding to immobilized amylase Four proteins were tested for binding to amylase fixed on microplates (Titertech polystyrene flat bottom, Flow Laboratories): anti-amylase, purified amylase, biotinylated amylase, and GP-2. Binding assays were as foIlows. A nti-amylase One hundred microlitres of 1 : 20 000 dilutions of rabbit antiamylase were incubated for 1 h at room temperature. Plates were washed five times with PBS + GT (pH 6.9) and developed with 100 pL of peroxidase-labelled donkey F(abl), to rabbit Ig (Amersham NA 9340) diluted 1:2000 in PBS + GT. Plates were washed five times with PBS + GT before colour development with 0.01 Vo 3,3 ',5,5 '-tetramethylbenzidine as the chromogenic substrate in PBS + GT. Readings were done at 450 nm, as previously reported (Leblond et al. 1989). Amylase BAM binding was tested using 100 pL of 2 pg BAM/mL, in 1 mM citrate supplemented PBS + GT (CPBS + GT) adjusted to the desired pH from 4.0 to 7.0. The pH used in the first binding step was used in all further steps, up to the peroxidase colorimetric assay. All other conditions were identical to anti-amylase binding. Peroxidase-labelled streptavidin (Amersham RPN 1231) diluted 1:2000 was used for development. NSB was evaluated on wells saturated with fish gelatin, where no amylase was fixed. Binding of streptavidin to fixed amylase was insignificant. GP-2 GP-2 binding was tested by addition of 100 pL of 2 pg purified GP-2/mL in 0.005% octyl glucoside. Incubation was carried out for 1 h at room temperature, in CPBS + GT adjusted to the desired pH from 4.1 to 7.0. The pH used in the first step was used in all further steps. After five washes, 1 : 20 000 dilutions of rabbit antirat GP-2 were added and incubated 1 h at room temperature. The anti-GP-2 was developed with peroxidase-labelled anti-rabbit antibodies (NA 9340) for 30 min in buffer supplemented with 4% PEG 8000, using the above protocol. Polymerization of amylase Purified amylase was polymerized with benzoquinone according to previously published procedures (Porah and AxCn 1976). In short, 28 pL of a 6.7 mM benzoquinone in methanol was added to 5 mg/mL of amylase in PBS. After a 1-h incubation at room temperature under agitation in the dark, excess benzoquinone was reacted with 5 pL of 10% glycine for 1 h. After dialysis against PBS, polymerized soluble amylase was separated from insoluble polymers by centrifugation at 15 000 x g.

JACOB ET AL.

E

3

1.5

1.0

0 0.5

0.0

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

PVC

(A)

1

2

8.5

6.9

(B)

SUPPORT

3

(D)

POLYSTYRENE

3

PH

4

+

9.6

(c)

-

POLY-L-LYSINE

5

Amylase (pg/ m L)

Amylase (pg/ mL) FIG. 1. Parameters for fixation of pancreatic amylase on plastic microtiter plates. Optimal conditions for fixation of amylase to plastic plates have been systematically determined. (A) Two commonly used plastic mitrotiter plates were tested: polyvinylchloride and polystyrene plates. (B) The pH adopted for the fixation step was 6.9, the optimal pH for amylase activity. (C) Precoating of plates with highly charged poly-L-lysine did not increase the amount of amylase fixed to plastic. (D) The amount of amylase activity bound to plastic was assessed by binding of BAM. Binding was performed at pH 5.5 (o),5.8 (o),6.0 (A), and 8.0 (+). (E) The amount of amylase fixed to plastic was also monitored by its enzymatic activity in the microtiter plate with the a-amylase EPS C-system". The amount of bound amylase was monitored in A, B, and C by binding of anti-amylase antibodies and in D by binding of BAM. Both were developed calorimetrically at 450 nm with peroxidase-labelled secondary antibodies or peroxidase-labelled streptavidin, respectively.

Purification of GP-2 and precipitation by polymerized amylase GP-2 (anchored) was purified by affinity chromatography on Helix pomatia agarose packed in a HR 5/20, 200 x 5 mm glass column (Pharmacia), equilibrated in 20 mM Tris-HC1 (pH 8.0) - 150 mM NaCl - 0.8% octyl glucoside and run at 0.2 mL/min. Elution was carried out with 50 mM N-acetyl-Dgalactosamine. Aggregation of GP-2 and amylase was carried out in 1.5-mL EppendorP tubes with 5 pg of purified GP-2 a'd 500 pg of the different forms of amylase in a final volume of 200 pL of CPBS (pH 5.5). After a 30-min incubation at room temperature, tubes were centrifuged at 15 000 x g at 4OC for 15 min. The pellet was mechanically resuspended with a plastic pestle in CPBS (pH 5.5) for a second incubation of 30 min at room temperature. Tubes were centrifuged at 15 000 x g at 4OC. GP-2 of the final pellet and the two supernatants was assayed by ELISA (Leblond et al. 1989). Total recovery of GP-2 was greater than 90% for each condition. Other methods Amylase enzymatic activity in solutions or in microplates was assayed with the a-amylase EPS C-system" obtained from Boehringer Mannheim Ltd. (Laval, Que.). Proteins were estimated

by the BCA protein assay using bovine serum albumin as the standard (Smith et al. 1985). SDS-PAGE was performed on 6-15% acrylamide gradient gels using Laemmli's discontinuous buffer system (Laemmli 1970) under reducing conditions.

Results Making of amylase-coated microplates Conditions for fixation of amylase on plastic microplates were systematically studied. PVC plates were compared with polystyrene for their capacity to adsorb amylase. PCV did not fix significantly more protein than the routinely used polystyrene (Fig. 1A). Polystyrene was then adopted. Precoating of plates with poly-L-lysinedid not increase the amount of fixed amylase (Fig. 1C). Three pHs were tested for amylase fixation: pH 6.9, 8.5, and 9.6 (Fig. 1B). pH 6.9 was slightly more effective and had the advantage of being milder for amylase, whose enzymatic activity is optimal at pH 6.9. Binding to plastic was very stable. Extensive washing of the plates in slightly alkaline conditions did not release any bound amylase. By use of ethylidene-protected

BIOCHEM. CELL BIOL. VOL. 70, 1992

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

1108

0.00

0.0

Tween 20 ( W ) Citrate (mN) P E G (XI

0.05 10

-

0.05 1

-

0.5 1

-

0.05 1 4

Gel

Gel-Tw

BSA

BSA-Tw

FIG.2. Standardization of the conditions for the binding of BAM to amylase fixed on microtiter plates. (A) The concentration of citrate to add to PBS, the concentration of Tween 20Q, and the possibility of adding PEG to the assay were tested to determine their effects on BAM binding to fixed amylase at pH 5.5. A medium containing 0.05% Tween 20@without PEG and buffered with 1.0 mM citrate was adopted. (B) Two proteins with or without 0.05% Tween 20@were tested for blocking unspecific sites on plastic. BAM binding was done at pH 5.5

FIG. 3. pH dependence, saturation curves, and effect of KC1 on BAM binding to amylase fixed on microtiter plates. The effect of pH on the binding of rat (A) or pig (B) BAM to homologous amylase fixed on microtiter plates was tested. Curves represent two separate to experiments. (C) A saturation curve of BAM binding was performed with a range of concentrations of BAM from 1 X 1 x lo-' g/mL at pH 5.5 (w) or pH 7.0 (0). Approximate Kd values of 20 ng/mL and 0.6 pg/mL could be estimated at pH 5.5 and 7.0, respectively. (D) Effect of KC1 on BAM binding. Except for B, BAM and fixed amylase were from the rat. NSB consisted of the OD obtained in absence of BAM and were subtracted from the OD in presence of BAM.

4-nitrophenyl a,~-maltoheptaosideas substrate (Laine et al. 1993), the activity of the fixed amylase was confirmed (Fig. 1E). The optimal concentration of amylase chosen to saturate microplates was 1.5 pg/mL. This concentration was

obtained by monitoring fixation of BAM (Fig. ID) and of specific anti-amylase antibodies, or by the enzymatic activity of amylase (Fig. 1E). The choice of a protein and buffer conditions to saturate

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

JACOB ET AL.

plates, to reduce unspecific attachment sites to a minimum, were tested using a combination of fish gelatin or bovine serum albumin, with or without 0.05% Tween 20@. Figure 2B shows that fish gelatin supplemented with 0.05% Tween 20@was the most effective combination to saturate the plate and reduce nonspecific bindings of BAM to wells. On the other hand, we checked for a mixture having a reliable buffering capacity for the total range of pH used, and at the same time, the optimal concentration of Tween 20@. PBS supplemented with 1 mM citrate and 0.05% Tween 20@was ideal for these purposes (Fig. 2A). A high concentration of Tween 20@(0.5%) reduced BAM binding by more than 70% (Fig. 2B). This was an indication that hydrophobic interactions could be involved in the acidic association of amylase. The stability of the fixation of amylase to rnicroplates was confirmed by its total resistance to extensive washing (1 h) at the two currently used pHs of 5.5 and 6.9 (results not shown).

Binding of BAM to amylase-coated microplates When granule content is titrated back to acid pH (Rothman 1971), amylase coprecipitates with the zymogens. On the contrary, pure amylase does not precipitate at acid pH (results not shown). In search of pH-dependent factors that would be involved in amylase aggregation, amylasecoated microtiter plates were first probed with BAM. As mentioned in the previous section, biotinylation inactivated amylase only by 30%. BAM was therefore considered to be virtually native amylase. High binding of BAM was observed at acidic pH, with a plateau between pH 5.0 and 5.7 in the rat (Fig. 3A). A similar pattern was observed with pig amylase in an also homologous system, i.e., pig amylase probed with pig BAM (Fig. 3B). In both cases, a low level of binding was observed at neutral pH. Binding was saturable with 2 pg BAM/mL at pH 5.5, as well as at pH 7.0 (Fig. 3C). An approximate Kd of 20 ng/mL at pH 5.5 and 0.6 pg/mL at pH 7.0 could be estimated from these data. At pH 7.0, however, only 40% of the signal was obtained. These observations show that the effect of pH on binding is on the affinity, as well as on the number of functional binding sites on fixed amylase. Removal of KC1 from the buffer did not significantly increase binding (Fig. 3D), but results obtained suggest that more nonspecific ionic interactions occur at low ionic strength. Reversibility of binding at pH 5.2 was assessed by washing plates at pH 7.0 (Fig. 4A) or pH 8.0 (results not shown). No significant release of BAM was measured in either case. This shows that most BAM bound to fixed amylase is tightly bound and is stable to neutral media. Finally, competition for binding of BAM to fixed amylase was performed with purified amylase. Purified amylase (200 pg/mL) was added directly in the BAM solution or added subsequently to the plate after BAM binding. As shown in Fig. 4B, no significant differences in BAM binding were observed in either condition. Therefore, amylase does not compete with BAM for binding to fixed amylase. To verify if BAM binding was not due to the biotin adduct on BAM, 60 pM free biotin was added to BAM during incubations (Fig. 4C). No differences were observed, suggesting that BAM binding to fixed amylase is due to the amylase polypeptide and not to its biotin adduct. BAM binding to amylase fixed on microplates appeared to be of high affinity and pH dependent, irreversible, and saturable.

(A)

2,0

0.0

BINDING pH WASH pH

(B)

' e O

AMYLASE

(C)

-

~imult.

After

2.0

0.0

CONTROL

BIOTIN

FIG.4. Reversibility and specificity of BAM binding to amylase fixed on microtiter plates. (A) To test the reversibility of binding as a function of pH, BAM was first bound. Afterwards, wells were washed extensively for 1 h with medium of pH 7.0 or 5.2. (B) Specificity of BAM binding was tested by adding 200 pg/mL of native amylase directly to the 2 pg/mL BAM solution or subsequently to the plate after BAM was bound. (C) The possibility that BAM binding to fixed amylase could be through its biotin adduct was tested by the addition of 60 pM free biotin in the binding mixture to displace BAM from the fixed amylase. Binding in B and C was performed at pH 5.5. NSB consisted of the OD obtained in absence of BAM and were subtracted from the OD in presence of BAM.

However, native unmodified amylase could not compete for binding. These observations led us to infer that native amylase would not bind to fixed amylase, or if so, its affinity would be so low that it could not compete with BAM. Binding of

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

1110

BIOCHEM. CELL BIOL. VOL. 70, 1992

FIG. 5. Binding of native amylase and BAM to amylase fixed on microtiter plates as revealed by amylase enzymatic activity. Purified native amylase (A) or BAM (m) were used for binding to fixed amylase at pH 5.5, in concentrations of up to 100 pg/mL. The bound amylase activity measured colorimetrically at 405 nm was developed in the microtiter plate with the a-amylase EPS C-systemTM. The total activity was very low and required a 42-h incubation period.

,

native amylase was then assayed. Monitoring of amylase activity in the microplate was performed in both cases when native amylase and when BAM was used as the ligand. Figure 5 shows that binding of native amylase to fixed amylase was very low. Saturation was barely reached at 2 mg amylase/mL (results not shown) compared with 2 pg/mL for BAM (Fig. 3C). The latter activity was 1000 times less than with BAM, even if biotinylation inhibited 30% of BAM enzyme activity. Thus biotinylation appeared to confer to amylase the capacity to bind to fixed amylase. This point was confirmed with overbiotinylated amylase, when a concentration of biotin 15 times higher was used for labelling. Such overbiotinylated amylase was not denatured and retained 71 % of its activity. It bound to fixed amylase with the same affinity as normally biotinylated amylase (results not shown), but had totally lost is pH dependence for binding. Therefore, overbiotinylation, a procedure that did not affect the enzymatic activity, confirmed that subtle structural modifications to amylase are directly correlated with its binding properties. In conclusion, biotinylation alters amylase structure, conferring to amylase its capacity to bind to fixed amylase in a pH-dependent manner. In the same way, fixation of amylase to plastic has a comparable effect, conferring to fixed amylase the ability to act as a receptor for BAM.

Binding of GP-2 to pancreatic and intestinal proteins In search for a membrane-bound protein with the ability to interact in a pH-dependent manner with zymogens that could serve as membrane receptor for regulated secretory proteins, GP-2 was the most obvious candidate to study in the pancreas. GP-2 is indeed the major intrinsic membrane protein of zymogen granule membranes in numerous animal species. In addition, GP-2 is found in equivalent amounts in the membrane and in the granule content (Paquette et al. 1986). The latter observation made us suspect that GP-2 could interact with zymogens. Using this rationale, Western blots of pancreatic homogenate, zymogen granule content,

Z

H PANCREAS

M

H

M DUOD

H

M JEJUN

FIG. 6. pH-dependent binding of '25~-labelledGP-2 to pancreatic and intestinal proteins. An homogenate (H) and a crude membrane fraction (M) from pig pancreas, duodenum (DUOD), and jejunum (JEJUN) were run on SDS-PAGE under reducing conditions, along with a zymogen granule content fraction (Z). After transfer on nitrocellulose, blots were probed at pH 5.5 (A) and pH 7.2 (B) with pig '25~-labelledGP-2 with the BoltonHunter reagent. The amylase position corresponds to the two higher M, labelled bands shown by the arrow in A and B. A Coomassiestained gel of zymogen granule content with M, standards (94 000, 66 000, 45 000, 30 000, 21 000, and 14 000) is shown at the upper right (COOM).

and crude membrane preparations were probed with 1 2 5 ~ labelled GP-2 at the physiologic pH of 7.2 and at pH 5.5, to assess pH-dependent GP-2 bindings to endogenous pancreatic or intestinal proteins. Figure 6 shows that GP-2 bound almost exclusively to amylase in a pH-dependent manner. Two other bands, located just under amylase, also bound 125~-labelled GP-2 at acidic pH, but very faintly. At pH 7.2, 125~-labelled GP-2 bound weakly to a low M, zymogen (= 15 000) of the granule content. No significant pH-dependent binding of 125~-labelled GP-2 was detected with crude membranes from pancreas, duodenum, or jejunum as previously observed by Scheffer et al. (1980) using immunohistochemicalmeans. These results show that, among all the pancreatic zymogens, amylase is selectively recognized by GP-2 in an acidic environment. It seems to be the prime target of GP-2. On the other hand, the reverse situation, binding of soluble amylase (i.e., BAM) to GP-2, could not be achieved

1111

JACOB ET AL.

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

(A)

'.5

0.0 CONTROL

LIGAND ADDITION

0.0

Binding pH Wash pH

5.2

5.2

5.2 7.0

7.0 7.0

FIG. 7. Characterization of GP-2 binding to amylase fixed on microtiter plates. Binding of GP-2 was performed on amylasecoated microtiter plates using 2 pg purified GP-2/mL. The binding of GP-2 to the plate was detected using anti-GP-2 antibodies and developed with peroxidase-labelled secondary antibodies (see Materials and methods). All procedures were done at the same pH. (A) The intensity of coloration generated was tested for two of the pHs used. (B) Binding of GP-2 was performed at different pHs ranging from 4.0 to 7.0 in CPBS. Curves represent two separate experiments. (C) To test the reversibility of binding as a function of pH, GP-2 was first bound and then wells extensively washed for 1 h with media of pH 7.0 or 5.2. NSB consisted of the OD obtained in absence of GP-2 and was subtracted from the OD in presence of GP-2.

on blots. This observation was confirmed using the microplate system where GP-2 was fixed to plastic and probed with BAM. No binding of BAM could be obtained in spite of many attempts and the very efficient fixation of GP-2 to plastic. All attempts with soluble native amylase as the ligand for binding to fixed amylase or to GP-2 were unsuccessful. These observations correlate well with the inability of pure amylase to precipitate at acidic pH. The

BAM GP-2

ATP AMYLASE MALTOSE (O.lmM) (150mg/mL) (5mM)

-

BAM

GP-2 BAM

-

GP-2

FIG. 8. Effects of ATP, amylase, maltose, and BAM on binding of GP-2 to amylase fixed on microtiter plates. (A) ATP, purified native amylase, and maltose were added to the purified GP-2 solution (2 pg) to test their effect on binding of GP-2 to fixed amylase (see Results for justification). (B) Since BAM and GP-2 bind to fixed amylase, the effects of both ligands on each other's binding was assessed. Detection of BAM was done using peroxidase-labelled streptavidin and detection of GP-2 was done using anti-GP-2 as described in Fig. 7. NSB was as in Fig. 4 for BAM binding and as in Fig. 7 for GP-2 binding.

sum of these observations suggest that amylase must be in an insoluble aggregated form of some sort to allow any binding of GP-2. Fixation to plastic or nitrocellulose satisfies these requirements. Binding of GP-2 to amylase fixed on microplates The binding of GP-2 to fixed amylase on microplates was detected by the use of anti-GP-2 in the same conditions as for the assay of GP-2 by ELISA (Leblond et al. 1989). The effect of pH on binding of antibodies was tested since the whole procedure, including colour development, had to be performed at the same pH as the one at which binding was performed. This was done to minimize an eventual release owing to alkaline conditions. At pH 5.5, the resulting absorbance was 75% of the one obtained at pH 7.0. At pH 5.5 the amount of GP-2 bound to amylase precoated plates was, therefore, underestimated by about 26% relative to estima-

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

BIOCHEM. CELL

BIOL. VOL.

70, 1992

CONTROL

NATIVE

POLYMS

POLYW1

FIG. 9. Acid-dependent precipitation of soluble GP-2 by polymerized amylase. Interaction of purified GP-2 with polymerized amylase was performed using two types of amylase polymers: a soluble polymer and an insoluble one. (A) Soluble polymers (right track) consisted of a mixture of monomers, dimers, trimers, and tetramers (55 000 - 200 000 M,), while very high M, polymers that could not enter the separating or even the stacking gels were present in the insoluble mixture (left track). MI of standard proteins are indicated by bars on the left (200 000, 94 000, 66 000, and 45 000). (B) Five hundred micrograms of each polymer mixture was mixed with 5 pg of purified GP-2 in CPBS (pH 5.5). After centrifugation, GP-2 was determined by ELISA in pellets. POLYM,, insoluble amylase polymers; POLYM,, soluble polymers. tions done at pH 7.0 (Fig. 7A). Even under these nonideal conditions, GP-2 interaction with amylase-coated plates was clearly pH dependent. It was maximal at pH between 5.0 and 5.7 (Fig. 7B). As for amylase binding, removal of KC1 from the medium seemed to increase nonspecific ionic interactions, but not significantly (results not shown). Reversibility of the acid-dependent binding was assessed by exhaustive washing with pH 7.0 buffer. No significant release of GP-2 was observed (Fig. 7C). This shows that most GP-2 bound to fixed amylase is tightly bound and is fairly stable to neutral conditions. ATP (0.1 mM) or maltose (5 mM) did not have any effect on GP-2 binding to amylase (Fig. 8A). Results with maltose addition suggest that GP-2 binding is not mediated by carbohydrates on GP-2. Amylase is a glycohydrolase that cleaves a-(1-4)-glucosidic bonds in starch and has affinity for glycosidic residues. Amylase purification by retardation on Sephadex is a good example of this behaviour. ATP, a cofactor that GP-2 binds in acidic conditions with high affinity (Kd a 4.0 pM) (LeBel and Beattie 1986), was found to be without effect on GP-2 binding to fixed amylase. Soluble amylase (150 pg/mL), whether added after (results not shown) or simultaneously (Fig. 8A), in a concentration 75 times the one necessary to saturate microwells (see Fig. 3C), did not compete or displace GP-2 bound to the fixed amylase. Since both GP-2 and amylase bind to fixed amylase, interaction between these two ligands on binding to amylase was assessed by performing binding experiments with both GP-2 and BAM in the binding mixture, followed by specific development for each bound ligand. Figure 8B shows that addition of BAM increased GP-2 binding by 9.6%, and that conversely, GP-2 increased BAM binding by 7.7%. This is an indication that GP-2 and amylase have a genuine interaction and do not share the same binding site on amylase.

Acid-dependent aggregation of soluble GP-2 by polymerized amylase We have shown above that soluble BAM does not bind to fixed GP-2. Similarly, amylase has to be fixed to plastic to interact with BAM. This led us to conclude that amylase had to be in an immobilized form to interact with a ligand. Considering this particular feature, binding of GP-2 to amylase was further investigated by estimating how amylase, in various forms of complexation, could induce an interaction with GP-2 that would result in the precipitation of the complex. The monomeric state of purified amylase was first confirmed at pH 7.2, as well as at pH 5.5, by gel filtration on Bio-GelB P-300 (results not shown). Pure amylase was then polymerized with benzoquinone. The M, of polymerized amylase was estimated on SDS-PAGE to be between 110 000 and 200 000 for soluble polymers and more than 200 000 for insoluble polymers (Fig. 9A). Some of the higher M,s were trapped at the top of the separating gel, while others could not even enter the stacking gel. The system used to assess the interaction of GP-2 with polymerized amylase was through the precipitation of the complex in which GP-2 was assayed by ELISA. These assays showed that soluble polymerized amylase could precipitate 22% of the GP-2, while the higher molecular weight insoluble form of amylase could precipitate more than 50% of the GP-2 (Fig. 9B). Soluble monomeric amylase did not significantly precipitate GP-2. On the other hand, results with soluble polymers showed that dimers or tetramers at the most (Fig. 9A) are sufficient to induce the formation of complexes large enough to sediment into a pellet. Insoluble polymers were more potent for interaction with GP-2. One conclusion can be reached from these observations. As fixation to plastic, polymerization transforms amylase into a good substrate for GP-2 binding. According to results with soluble polymers, GP-2 should have at least two binding sites for polymerized amylase to allow the formation of

ET AL.

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

a sedimentable network. If GP-2 had only one binding site for amylase, no such pelletable network would be conceivable. Therefore, amylase has the capacity to bind GP-2 when polymerized or fixed to plastic. Discussion The aim of this article is to uncover the mechanism by which amylase aggregates in the TGN, so that it gets sorted to the secretion granule of the pancreatic acinar cell. There is now a consensus on the direct implication of the acidification of the secretory pathway that leads to the aggregation of regulated secretory proteins in the TGN and the ensuing sorting to the granule (Chanat and Huttner 1991; Tooze et al. 1987). In the case of the exocrine pancreas, when zymogen granule content is acidified in vitro, all zymogens aggregate, amylase included (Rothman 1971). On the other hand, when purified amylase is submitted to similar acidic conditions, it does not aggregate. Amylase is indeed a basic zymogen of pI 8.6 in the rat and in the pig (Sanders and Rutter 1972). The question then is how does amylase get packaged in the pancreatic granule if it does not respond to acidification by itself? We have raised the hypothesis that factors in the granule may be directly involved in its aggregation. To identify such factors, we have devised a system where purified amylase is fixed to microtiter plates. The fixation preserved the enzymatic activity of amylase. The first factor that we have found to bind to the fixed amylase was BAM. Biotinylation was chosen because it is a very mild procedure that in our hands left more than 70% of the enzymatic activity. BAM bound to fixed amylase with high affinity (Kd = 20 ng/mL) in a pH-dependent and a saturable manner. It could not be reverted by alkaline pH, nor removed by competition with native amylase. This was not surprising because native amylase stays monomeric at this pH. If amylase could bind to itself at acidic pH, polymerization and even aggregation would have been observed in the first place. Therefore, to bind to fixed amylase, the amylase polypeptide had to be modified. In the case presented here, biotinylation was sufficient to confer to amylase this novel capacity. Therefore, biotinylation altered in some manner the structure of amylase without significantly affecting the enzymatic activity. The observation that native amylase could not compete with biotinylated amylase is thus consistent. To apply this observation to the situation in vivo, it must first be realized that our in vitro system is a dissection of what exists in the TGN. All other factors and proteins normally present in the TGN have been removed. The question then is how the modification of amylase by biotinylation in vitro is reproduced in vivo in the TGN upon acidification. Knowing that biotinylation is on the €-amino group of lysine (Sanders and Rutter (1972) have determined 26 Lys residues in amylase) and that these groups are highly protonated in the pH range under consideration, a simple neutralization could be the mechanism that takes place in vivo. One could imagine that the binding of an acidic protein (a zymogen) or another negatively charged factor on amylase lysine residues could be sufficient to trigger a structural alteration, giving amylase its new binding capacity. As for the irreversibility of BAM binding, it could be explained by the permanent character of biotinylation.

1113

In vivo, pH-dependent reversible modifications are more likely. On the other hand, fixation to plastic is another way to give amylase the capacity to bind to itself (BAM) and to GP-2. In our system, such a fixed amylase was still active. Binding to plastic is considered an extremely mild procedure. This is why it is routinely used in all sorts of immunoassays. Although modifications of the protein structure generated in this situation are likely to be very discreet, they were sufficient to modify amylase. From these observations we believe that the modifications that amylase undergoes to acquire the capacity to interact with another molecule at acidic pH is so gentle that it could easily take place in vivo. As mentioned above, interactions with acidic zymogens that directly respond to acidification could easily trigger this process and lead to a cascade of aggregation of amylase. In addition to the autologous binding of amylase, we also report on the binding of GP-2 under similar acidic conditions. Amylase had to be fixed on plastic or nitrocellulose, or to be polymerized to allow GP-2 binding. Neither purified native amylase nor BAM could compete for binding of GP-2 to fixed amylase. GP-2 binding to fixed amylase is, therefore, through sites totally different from those found for BAM binding. As shown for amylase binding, GP-2 binding also required subtle modification of the amylase molecule. In the case of GP-2, tetramerization of amylase was sufficient. As for amylase binding, GP-2 binding to fixed amylase was not reversible. This behaviour is undoubtedly alarming, since solubilization of the secreted zymogens is critical for the normal flow of secretory proteins in pancreatic ducts. As mentioned for amylase binding, the system used in this report is a dissection of what is thought to exist in vivo. Only two components of the zymogen granule were examined. We believe that in vivo factors exist, in addition to those tested here, that are involved in the modification of amylase and that would help binding reactions to occur reversibly. GP-2 is the most abundant protein of the pancreatic zymogen granule membrane of numerous mammalian species. Owing to a glycosyl phosphatidylinositol membrane anchor (LeBel and Beattie 1988), GP-2 is present in equivalent amounts in a soluble form in the granule content (Geuze et al. 1981; Paquette et al. 1986). Its presence in the granule content is an indication that it could interact with zymogens. As shown on Western blots (Fig. 6), amylase is the major, if not the only zymogen, to which GP-2 binds in a pH-dependent manner. These observations argue for a role of GP-2 as membrane receptor for sorting secretory proteins to the granule. As evidenced from in vitro studies (Leblond et al. 1993), GP-2 also has the capacity to induce the acidic aggregation of pancreatic zymogens. Here we suggest that the protein with which GP-2 interacts to keep the mass of aggregated zymogens attached to the membrane of the TGN is quite likely amylase. These results suggest that, upon acidification of the TGN, the protonation of a factor, most probably a zymogen, would induce its binding to amylase. By binding to amylase, this factor would cause a slight modification of structure, allowing binding of other modified amylase molecules. Such an aggregate of amylase could then bind to GP-2 and thus render the aggregate membrane bound in the TGN. Finally budding of such membrane-bound aggregates could form immature granules.

1114

BIOCHEM. CELL BIOL. VOL. 70, 1992

The actual factor that triggers this aggregation cascade has yet to be identified. Acknowledgements We thank Ms. Marlyne Beattie for her expert technical assistance. M.J. was awarded a predoctoral studentship from the Canadian Cystic Fibrosis Foundation. D.L. is supported by the Natural Sciences and Engineering Research Council of Canada, the Canadian Cystic Fibrosis Foundation, and the Fonds pour la formation de chercheurs et l'aide a la recherche (Quebec).

Leblond, F.A., Talbot, B.G., Lauzon, I., and LeBel, D. 1989. A competition enzyme-linked immunosorbent assay @LISA) for the measurement of pancreatic GP-2 glycoprotein. J. Immunol. Methods, 124: 71-75. Leblond, F.A., Viau, G., Laint, J., and LeBel, D. 1993. In vitro reconstitution of the pH-dependent aggregation of pancreatic zymogens en route to the secretory granule: implication of GP-2. Biochem. J. 290. In press. Paquette, J., Leblond, F.A., Beattie, M., and LeBel, D. 1986. Reducing conditions induce a total degradation of the major zymogeigranule membrane protein in both its membranous and its soluble form. Immunochemicalauantitation of the two forms. Biochem. cell Biol. 64: 456-462. Porah, J., and Axtn, R. 1976. Immobilization of enzymes to agar, agarose, and Sephadex supports. Methods Enzymol. 44: 19-45. Rausch, U., Vasiloudes, P., Riidiger, K., and Kern, H.F. 1985. In vivo stimulation of rat pancreatic acinar cells by infusion of secretin. 11. Changes in individual rates of enzyme and isoenzyme biosynthesis. Cell Tissue Res. 242: 641-644. Reddy, M.K., Heda, G.D., and Reddy, J.K. 1987. Purification and characterization of a-amylase from rat pancreatic acinar carcinoma. Comparison with pancreatic a-amylase. Biochem. J. 242: 681-687. Reggio, H., and Dagorn, J.C. 1980. Packaging of pancreas secretory proteins in the condensing vacuoles of the Golgi complex. I n Biology of normal and cancerous exocrine pancreatic cells. Proceedings of an INSERM Symposium, No.15, Toulouse, May 7-9. Edited by A. Ribet, L. Pradayrol, and C. Susini. Elsevier, Amsterdam. pp. 229-244. Rothman, S.S. 1971. The behavior of isolated zymogen granules: pH-dependent release and reassociation of protein. Biochim. Biophys. Acta, 241: 567-577. Sanders, T.G., and Rutter, W.J. 1972. Molecular properties of rat pancreatic and parotid a-amylase. Biochemistry, 11: 130-136. Scheffer, R.C.T., Poort, C., and Slot, J.W. 1980. Fate of the major zymogen granule membrane-associated glycoproteins from rat pancreas. A biochemical and immunocytochemical study. Eur. J. Cell Biol. 23: 122-128. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B.J., and Klenk, D.C. 1985. Measurement of protein using bicinchoninic acid. Anal. Biochem. 150: 76-85. Tooze, J., Tooze, S.A., and Fuller, S.D. 1987. Sorting of progeny coronavirus from condensed secretory proteins at the exit from the trans-Golgi network of AtT2O cells. J. Cell Biol. 105: 1215-1226. Vandermeers. A., and Christophe, J. 1968. a-amylase et lipase du pancrtas de rat. Purification chromatographique, recherche du poids moltculaire et composition en acides m i n t s . Biochim. Biophys. Acta, 154: 110-129. Wong, J.G., Izutsu, K.T., Robinovitch, M.R., Iversen, J. M., Cantino, M.E., and Johnson, D.E. 1991. Microprobe analysis of maturation-related elemental changes in rat parotid secretory granules. Am. J. Physiol. 261: C1033-C1041. -

~ -- -

~

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by University of P.E.I. on 11/15/14 For personal use only.

A

Anderson, R.G.W., and Orci, L. 1988. A view of acidic intracellular compartments. J. Cell Biol. 106: 539-543. Bayer, E.A., and Wilchek, M. 1980. The use of avidin-biotin complex as a tool in molecular biology. Methods Biochem. Anal. 26: 1-45. Bendayan, M. 1984. Concentration of amylase along its secretory pathway in the pancreatic acinar cell as revealed by high resolution immunocytochemistry. Histochem. J. 16: 85-108. Burgess, T.L., and Kelly, R.B. 1987. Constitutive and regulated secretion of proteins. Annu. Rev. Cell Biol. 3: 243-293. Chanat, E., and Huttner, W .B. 1991. Milieu-induced, selective aggregation of regulated secretory proteins in the trans-Golgi network. J. Cell Biol. 115: 1505-1519. De Lisle, R.C., and Williams, J.A. 1987. Zymogen granule acidity is not required for stimulated pancreatic protein secretion. Am. J. Physiol. 253: G711-G719. Geuze, H.J., Slot J.W., van der Ley, P.A., and Scheffer, R.C.T. 1981. Use of colloidal gold particles in double-labelling immunoelectron microscopy of ultrathin frozen tissue sections. J. Cell Biol. 89: 653-665. Laemmli, U.K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London), 227: 680-685. Laint, J., Beattie, M., and LeBel, D. 1993. Simultaneous kinetic determinations of lipase, chymotrypsin, trypsin, elastase and amylase on the same microtiter plate. Pancreas, 8. In press. LeBel, D., and Beattie, M. 1984. The integral and peripheral proteins of the zymogen granule membrane. Biochim. Biophys. Acta, 769: 611-621. LeBel, D., and Beattie, M. 1986. Identification of the catalytic subunit of the ATP diphosphohydrolase by photoaffinity labeling of high-affinity ATP-binding sites of pancreatic zymogen granule membranes with ~ - ~ z ~ ~ o - [ ~ - ~ ' P Biochem. ] A T P . Cell Biol. 64: 13-20. LeBel, D., and Beattie, M. 1988. The major protein of pancreatic zymogen granule membranes (GP-2) is anchored via covalent bonds to phosphatidylinositol. Biochem. Biophys. Res. Commun. 154: 818-823. LeBel, D., Grondin, G., and Paquette, J. 1988. In vitro stability of pancreatic zymogen granules: roles of pH and calcium. Biol. Cell, 63: 343-353.