Planta (1983)159:322-328
P l a n t a 9 Springer-Verlag 1983
Localization of arabinogalactan proteins in the membrane system of etiolated hypocotyls of Phaseolus vulgaris L. Marieke R. Samson, Frans M. Klis, Corrien A.M. Sigon and Durk Stegwee Department of Plant Physiology, University of Amsterdam, Kruislaan 318, NL-1098 SM Amsterdam, The Netherlands
Abstract. The subcellular distribution of arabino-
galactan protein (AGP) in etiolated bean hypocotyls was studied by isopycnic density centrifugation on sucrose gradients at different Mg 2+ concentrations. The distribution of hydroxyproline (a major amino acid in AGP) in the membrane-containing fractions indicated that hydroxyproline-containing proteins were associated with rough endoplasmic reticulum, possibly with the Golgi apparatus, and with the plasma membrane. Non-specific binding of hydroxyproline-containing molecules to membranes could be excluded. To detect AGPs, fractions obtained after isopycnic density centrifugation were isoelectrofocused on polyacrylamide gels, and the gels were stained with fl-Gal-Yariv reagent. Bands appeared only at low pH values, where also most hydroxyproline was found. In the fractions at low densities (presumably membranefree), several bands were visible supporting the idea of the heterogeneous character of soluble AGP. The distribution of AGP in the membranous fractions strongly indicated that AGP was associated with the plasma membrane. Specific agglutination of protoplasts in the presence of fl-Gal-Yariv reagent indicated that AGP was exposed at the outside of the cell membrane. Key words: Arabinogalactan protein (localization)
- Hydroxyproline - Phaseolus (arabinogalactan protein) - Yariv reagent.
Introduction
teins; their protein moiety is rich in hydroxyproline, serine, and alanine, and the principal sugars are arabinose, galactose, and uronic acids, the latter probably being responsible for their isoelectric point at low pH. The high degree of glycosylation of AGPs presumably explains why they are usually soluble in 5% trichloroacetic acid. Many AGPs show fl-lectin activity, i.e. they specifically bind fl-D-glycosyl-Yariv reagent, a red coloured dye. Arabinogalactan proteins are distinct from the hydroxyproline-rich glycoprotein extensin which is found in the primary cell wall (Lamport 1969; Van Holst and Klis 1981) and is believed to be a structural component (Keegstra et al. 1973). At the moment, no function can be ascribed to AGPs. Although AGPs have been localized in a variety of plant tissues (for review see Clarke et al. 1979), detailed information on the subcellular distribution of AGP is lacking. Van Holst et al. (1981) have isolated AGP released by sonication from a crude organelle fraction of dark-grown hypocotyls of Phaseolus vulgaris L., but they did not identify the AGP-containing organelles. Larkin (1978) showed that plant protoplasts agglutinate in the presence of fl-glycosyl-Yariv reagent, suggesting that AGP is associated with the cell surface of protoplasts. We have investigated the subcellular distribution of hydroxyproline, a major amino acid in the protein moiety of AGP, and of AGP itself. We report here that hydroxyproline is found i n the rough-endoplasmic-reticulum fraction, the plasma membrane, and possibly in the Golgi fraction, and that AGP is associated with the plasma membrane.
Arabinogalactan proteins (AGPs) are widely distributed throughout the plant kingdom (Clarke et al. 1979; Jermyn and Yeow 1975). They are highly glycosylated (usually more than 80%) pro-
Plant material
Abbreviations: AGP = arabinogalactan protein
In most experiments we used 6-d-old seedlings of Phaseolus vulgaris L. cv. Prglude (Royal Sluis, The Netherlands) that had
Material and methods
M.R. Samson et al. : Localization of arabinogalactan protein been grown in the dark. Growth conditions were as follows: 75 g of beans were sown in a plastic tray (40 x 30 x 8 cm) filled with 5 1 Perlite (Mauritz & Zonen, Bussum, The Netherlands) wetted with 1.71 tap water; during growth the temperature was maintained at 25 ~ C, and the relative humidity at 65%.
Homogenization and fractionation Sections (3 cm long) of etiolated hypocotyls were excised directly below the hook and homogenized with a mechanical tissue chopper in homogenization buffer (40 mM 2-amino-2-(hydroxymethyl)-l,3-propane diol (Tris)-HC1, I mM dithiothr6itol, 1 m M ethylenediaminetetraacetic acid (EDTA), 3 m M or 0.1 m M MgC12, 15% (w/w) sucrose, pH 7.8; 1 ml buffer per g fresh weight). The homogenate was filtered through 10-~tm nylon gauze to remove tissue fragments and cell walls, and the filtrate was centrifuged for 5 min at 1,000 g. The supernarant was layered upon a 15-50% (w/w) linear sucrose gradient in homogenization buffer. The gradients were centrifuged (50,000 g) in a Beckman SW 20-1 rotor (Beckman Instr., Palo Alto, Calif., USA) at 4 ~ C for 16 h. One-ml fractions were collected. Sucrose concentrations were determined using a refractometer.
Marker enzymes Latent inosine-5'-diphosphatase (IDPase) (EC 3.6.1.6). Latent IDPase activity was determined after storing the sucrose gradient fractions at 4 ~ C for 3 d. The assay was a modification of the method of Galbraith and Northcote (1977), using the inorganic phosphate assay of Chandra Rajan and Klein (1976): a 25 gl sample was incubated with 225 gl substrate solution (3 m M IDP, 5 m M MgC1 z, 40 mM Tris-HC1, pH 7.5) at 25 ~ C for 40 rain. Then, 1.75 ml 10% (w/v) sodium dodecyl sulphate in 0.1 M sodium-acetate buffer (pH 4.0), 0.25 ml 2.5% (w/v) ammonium molybdate, and 0.25 ml 1% (w/v) ascorbic acid were added, and the extinction at 870 nm was determined after 20 rain.
Antimycin-A-insensitive NADPH-cytochrome-c reductase (EC 1.6.2.4). Determination of NADPH-cytochrome-c reductase was as follows: 800 gl 10 ~tM cytochrome c in 100 mM phosphate buffer containing 1 m M K C N (pH 7.7) was incubated with 100 gl sample at 25 ~ C. The increase in extinction per rain at 550 nm was determined; then 100 ~1 1 mM N A D P H was added and the increase in extinction per rain was measured again. The difference between these values gave the activity of NADPH-cytochrome-c reductase.
UDP-glueose-fl-gluean-fl-glucosyl transferase (glucan synthetase II) (EC 2.4.1.12). The glucan-synthetase-II assay was carried out essentially according to Ray (1979): 20 I~14 mM uridine 5'-diphosphate-(UDP)-glucose and 12.5 gl (1.18 kBq) [~4C]UDP-glucose (specific activity 9.08 GBq.mmo1-1) were incubated with 100-gl sample at 25 ~ C for 45 rain. The reaction was stopped by adding boiled total particles (total membrane fraction of homogenized bean hypocotyls boiled for 5 min; 20 ~tl suspension per assay) and 850 gl 80% (v/v) ethanol. The mixture was stored overnight at - 2 0 ~ C and then filtered on a Whatman G F / C glass filter (Springfield Mill, UK). The filters were washed three times with 2 ml 80% (v/v) ethanol, and radioactivity was measured.
323 HC1 at 120 ~ C for 3 h. They were then air-dried, and the residue was taken up in water. Polyacrylamide gels were hydrolysed as described by Martin et al. (1976). Hydroxyproline was determined by a modification of the method of Drozdz et al. (1976). The assay was as follows: 250 gl sample was mixed with 150 gl 3.5% chloramine T in citrate-acetate buffer (the buffer was prepared by dissolving 38 g sodium acetate.3 HzO and 25 g sodium citrate. 3 H 2 0 in 200 ml water, 255 ml isopropanol was added, and after adjusting the pH to 6.0 with acetic acid, water was added to a total volume of 1,000 ml). The mixture was kept at 3(~35~ C for exactly 10 min. Then 300 gl perchloric acid: 6% (w/v)p-dimethylaminobenzaldehyde in isopropanol (1:5, v/v) was added. After 20 rain at 60 ~ C, the extinction at 558 nm was measured.
Isoelectrofocusing on polyacrylamide gels Hydroxyproline-containing molecules were precipitated by adding five volumes of methanol to one volume of sample, and keeping these mixtures at 4 ~ for 2 h. After centrifuging at 1,500 g for 15 min, the pellets were taken up in a small volume of 1% (v/v) Triton X-100. Polyacrylamide slab gels (5 % polyacrylamide, thickness 0.7 ram) were made according to Laemmli (1970) with 2.4% (w/w) Ampholines, pH 2.5-4 (LKB, Bromma, Sweden). The gels were run on a LKB Multiphor Apparatus with 1 M glycine and I M H3PO 4 as cathode and anode buffer, respectively; the cooling temperature was 8 ~ C. Samples were applied on small paper filters placed on the middle of the gel. After 5 h at 1,000 V, 1-cm pieces were excised at one side of the gel to determine the pH, and the remainder of the gel was stained directly with fl-Gal-Yariv reagent (0.1 mg m1-1 10% (v/v) dimethylsulphoxide (DMSO) in water) overnight at 40 ~ C. The gel was destained with several changes of water.
Radioactive labeling of proteins Hypocotyl sections (3 cm long) were placed in a Petri dish on filter paper soaked in 2 gM indole acetic acid, and incubated at 30 ~ C for 1 h. They were then submerged in a [14C]proline solution (75.4 kBq m1-1) and vacuum-infiltrated for 3 min. After another 1 h at 30 ~ C in a Petri dish on wet filter paper, the sections were homogenized for 10 rain on ice with a razor blade in homogenization buffer (1 ml buffer per g fresh weight) and further treated as described in the section ,,Homogenization and fractionation".
1,3,5 Tris (4-fl-D-galactopyranosyl-oxyphenylazo) 2,4,6 trihydroxybenzene (fl-Gal- Yariv) fl-Gal-Yariv reagent was prepared by coupling diazotized paminophenyl-fl-D-galactopyranoside (Sigma) to phloroglucinol (Yariv et al. 1962).
fl-Galactosidase (EC 3.2.1.23) fl-Galactosidase activity was measured by incubating 100 ~1 sample with 100 gl 50 m M p-nitrophenyl-fl-D-galactopyranoside and 300 gl 40 mM acetate buffer, pH 4.0, at 37 ~ C. After 30 min, the reaction was stopped by adding 1 ml 200 mM glycine-NaOH, pH 10.6, and the amount of p-nitrophenol was determined by measuring the extinction at 405 nm.
Determination of hydroxyproline
Results
Sucrose gradient fractions were extensively dialyzed against water. The samples were hydrolysed in the presence of 6 M
Figure 1 A-C shows the distribution in isopycnic sucrose gradients of marker-enzyme activities for,
324
M.R. Samson et al. : Localization of arabinogalactan protein 9
i
i
i A
i
Bq 190
i
d
,
1.20
50
3
" .
%
2
1.16
1.16
1.12
t.12 25
1
1.0B
1.08
d
10
d
1.20
,&
1.16
3
1.12
-r~
2
".
t.20
9
1.16
~
1,12
d..-
1.08
I
10
20
I ~
ml
30
1
0
1.08
L 10
20
rnl
30
Fig. 1. A Distribution on isopycnic density gradients of NADPH-cytochrome-c reductase in the presence of 3 mM Mg 2+ ( o - - o ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 o, density (g cm-3). B Distribution on isopycnic density gradients of latent inosine-5'-diphosphatase in the presence of 3 mM Mg 2+ ( 0 - - 0 ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density. C Distribution on isopycnic density gradients of glucan synthetase II in the presence of 3 mM Mg 2+ ( o - - 9 or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density. D Distribution of hydroxyproline on isopycnic density gradients in the presence of 3 mM Mg ~+ ( 0 - - 0 ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density
respectively, endoplasmic reticulum (NADPH-cytochrome-c reductase), Golgi apparatus (latent inosine-5-diphosphatase; IDPase) and plasma membrane (glucan synthetase II). In the presence of 3 mM Mg 2+, membrane-bound NADPH-cytochrome-c reductase showed a peak at a density of 1.17; also, considerable activity was detected in the soluble fractions at low sucrose concentrations. When homogenization and further steps were performed at 0.1 mM Mg 2+, most membrane-bound activity equilibrated at a density of 1.11, although some activity could still be found at higher densities. These results indicated that in our system rough endoplasmic reticulum had a density of about 1.17 and that at low Mg 2+ concentration ribosomes were stripped from the membranes; the resulting smooth endosplasmic reticulum banded at a density of 1.11. Electron-micro-
scopic observations confirmed this (results not shown). Figure 1 B shows that the activity of latent IDPase, a Golgi-apparatus marker, peaked around densities of 1.15, independent of the Mg 2§ concentration present. In Fig. 1C we show the distribution profiles of glucan synthetase II in the presence of 3 mM or of 0.1 mM MgC12. Like Robinson et al. (1982) we found a density shift of the enzyme activity, from 1.17 at 3 mM M g 2+ to 1.15 at 0.1 mM Mg 2+, indicating that in bean hypocotyls also the density of plasma membrane vesicles is influenced by the Mg 2+ concentration. Mitochondria, as judged by the marker enzyme, cytochromec oxidase, banded at 1.18 irrespective of the Mg 2+ concentration used (data not shown). Arabinogalactan protein is a hydroxyprolinerich glycoprotein and therefore the distribution of
M.R. Samson et al. : Localization of arabinogalactan protein u
u
/'d
OO0
Bq 100
1.16
1.12 50 1.08
.............J 10
20
ml
3'0
Fig. 2. Distribution of radioactivity on isopycnic density gradients after homogenization of bean hypocotyl sections in the presence of soluble [14C]hydroxyproline-containing compounds. 9 9 9 e, density (g cm 3)
hydroxyproline in isopycnic sucrose gradients was determined, both in the presence of 3 mM Mg 2§ and of 0.1 mM Mg 2§ (Fig. 1 D). In both cases, most hydroxyproline was present in the presumably membrane-free fractions (density < 1.09). In the presence of 3 mM Mg 2+ the distribution profile of membrane-bound hydroxyproline showed a peak at densities about 1.17, and a plateau at somewhat lower densities. At densities between 1.13 and 1.09 hardly any hydroxyproline was detectable. The distribution of membrane-bound hydroxyproline in the presence of 0.1 mM Mg 2+ showed a different pattern. Only a negligible amount of hydroxyproline was present at densities higher than 1.17, but two peaks appeared at densities 1.15 and 1.10. As shown above, most hydroxyproline in the gradients was present in soluble compounds. To exclude the possibility that soluble hydroxyproline-containing molecules became nonspecifically attached to membranes during homogenization, thereby causing artefacts, the following control experiment was done: we labeled hypocotyl sections with [l~C]proline; the homogenate of these sections was fractioned on a linear sucrose gradient (15-50%, w/w), and 24 fractions were collected. The fractions with a density < 1.09 contained 14C-labeled proteins including a considerable amount of [14C]hydroxyproline-containing molecules (data not shown). We homogenized unlabeled bean hypocotyl sections in the presence of these fractions. This homogenate was also fractioned by isopycnic density centrifugation, and the
325
radioactivity was determined in all fractions. As shown in Fig. 2, virtually no radioactivity was present in membrane-containing fractions; nonspecific adhesion or occlusion of hydroxyprolinecontaining material could, therefore, be excluded. Hydroxyproline is not absolutely specific for AGP, because in bean hypocotyls hydroxyproline has also been found in extensin (Van Holst et al. 1980), a well known structural component of the cell wall. To discriminate between AGP and extensin, isoelectrofocusing was used. Extensin and its precursors probably have their isoelectric points at alkaline pH values (Stuart and Varner 1980), whereas AGP has an isoelectric point at low pH because of the large amount of uronic acids in its carbohydrate side-chains (Van Holst et al. 1981). The precursors of AGP probably have isoelectric points at neutral to acid pH values. After isopycnic density centrifugation, the macromolecules from each fraction were isoelectrofocused on polyacrylamide gels (pH2.5-4.5); to visualize AGP, the gels were stained with fl-Gal-Yariv reagent. Figure 3 shows the subcellular distribution of AGP in the presence of 3 mM and 0.1 mM Mg 2§ The presumably membrane-free fractions (density < 1.09) showed several intensely stained bands at pH 2-3 at both Mg 2+ concentrations. At higher densities a single, somewhat diffuse band appeared at pH 2.8; at 3 mM Mg 2§ this band was found in fractions 1-7, whereas at 0.1 mM Mg 2§ it was found in fractions 3-9. Furthermore, at 0.1 mM Mg 2+ this band stained consistently less than at 3 mM Mg 2+, possibly because the association between AGP and membrane is Mg 2§ dent. Comparison of Figs. 3 and 1 C shows that the subcellular distribution of membrane-bound AGP parallels that of glucan synthetase II, including the shift to lower densities at lower Mg 2§ concentration. No clear bands can be seen at other densities. Since fl-galactosidase activity might obscure the picture by hydrolysing the fl-Gal bond in the Yariv reagent and so generating an insoluble red compound, we assayed for fl-galactosidase activity with p-nitrophenyl-fl-D-galactoside as a substrate. No activity could be found at pH values where redcoloured band were visible in the gel (results not shown). Additional evidence that AGP was indeed responsible for the fl-Gal-Yariv binding in the polyacrylamide isoelectrofocusing gels was obtained by isofocusing both a - presumably - membrane-flee fraction (density __,
t_ "I3
i
x: 60
t
i
B
20 )H >~ 3.8'3.5'3.3'30'2.0' 2.L
Fig. 4A, B. Percentages of total hydroxyproline determined in polyacrylamide gels after isoelectrofocusing a membranous fraction (A) and a soluble fraction (B). The a r r o w s indicate the position of sample application on the gel
organelles or from the cytosol. The distribution of hydroxyproline in the membrane-containing fractions indicated that hydroxyproline-containing molecules are associated with rough endoplasmic reticulum, since a clear shift of hydroxyprolinecontaining molecules to lower densities is seen when 0.1 mM Mg 2 § is used instead of 3 mM; this shift coincides with a shift in NADPH-cytochrome-c-reductase activity and indicates a conversion of rough endoplasmic reticulum to smooth endoplasmic reticulum at low Mg 2+ concentration. A small amount of hydroxyproline-containing molecules is possibly associated with the Golgi apparatus. Association of hydroxyproline-containing molecules with the plasma membrane seems clear, since at 3 mM Mg 2§ hydroxyproline peaks at about 1.17, whereas at 0.1 mM Mg 2+ a new peak appears at 1.15; the same shift is shown by glucan-synthetase-II activity, a plasma-membrane marker enzyme. The presence of hydroxyprolinecontaining molecules in membranous fractions, however, does not give conclusive evidence for the presence of AGP, since hydroxyproline is also found in extensin and in precursor molecules. To further identify the subcellular fractions with which AGP was associated we used isopycnic density centrifugation followed by isoelectrofocusing on polyacrylamide gels. After isoelectrofocusing the gels were stained with fl-Gal-Yariv reagent, which is bound specifically by AGPs. Bands appeared only at low pH values, mostly between pH2.5 and 3.0. Arabinogalactan proteins are known to have their isoelectric point at low pH (Van Holst et al. 1981) because of the large amount
327
of uronic acids in their carbohydrate chains. Analysis of the gels for hydroxyproline showed that most hydroxyproline was in the very acid regions of the gels, where the/?-Gal-Yariv-stained bands appear, again indicating that these bands indeed represent AGPs. Interestingly, no clear indications for the presence of the other hydroxyproline-rich protein, i.e. extensin, or its precursors, were found. In the - presumably - soluble fractions (fractions 15-22, density
P l a n t a 9 Springer-Verlag 1983
Localization of arabinogalactan proteins in the membrane system of etiolated hypocotyls of Phaseolus vulgaris L. Marieke R. Samson, Frans M. Klis, Corrien A.M. Sigon and Durk Stegwee Department of Plant Physiology, University of Amsterdam, Kruislaan 318, NL-1098 SM Amsterdam, The Netherlands
Abstract. The subcellular distribution of arabino-
galactan protein (AGP) in etiolated bean hypocotyls was studied by isopycnic density centrifugation on sucrose gradients at different Mg 2+ concentrations. The distribution of hydroxyproline (a major amino acid in AGP) in the membrane-containing fractions indicated that hydroxyproline-containing proteins were associated with rough endoplasmic reticulum, possibly with the Golgi apparatus, and with the plasma membrane. Non-specific binding of hydroxyproline-containing molecules to membranes could be excluded. To detect AGPs, fractions obtained after isopycnic density centrifugation were isoelectrofocused on polyacrylamide gels, and the gels were stained with fl-Gal-Yariv reagent. Bands appeared only at low pH values, where also most hydroxyproline was found. In the fractions at low densities (presumably membranefree), several bands were visible supporting the idea of the heterogeneous character of soluble AGP. The distribution of AGP in the membranous fractions strongly indicated that AGP was associated with the plasma membrane. Specific agglutination of protoplasts in the presence of fl-Gal-Yariv reagent indicated that AGP was exposed at the outside of the cell membrane. Key words: Arabinogalactan protein (localization)
- Hydroxyproline - Phaseolus (arabinogalactan protein) - Yariv reagent.
Introduction
teins; their protein moiety is rich in hydroxyproline, serine, and alanine, and the principal sugars are arabinose, galactose, and uronic acids, the latter probably being responsible for their isoelectric point at low pH. The high degree of glycosylation of AGPs presumably explains why they are usually soluble in 5% trichloroacetic acid. Many AGPs show fl-lectin activity, i.e. they specifically bind fl-D-glycosyl-Yariv reagent, a red coloured dye. Arabinogalactan proteins are distinct from the hydroxyproline-rich glycoprotein extensin which is found in the primary cell wall (Lamport 1969; Van Holst and Klis 1981) and is believed to be a structural component (Keegstra et al. 1973). At the moment, no function can be ascribed to AGPs. Although AGPs have been localized in a variety of plant tissues (for review see Clarke et al. 1979), detailed information on the subcellular distribution of AGP is lacking. Van Holst et al. (1981) have isolated AGP released by sonication from a crude organelle fraction of dark-grown hypocotyls of Phaseolus vulgaris L., but they did not identify the AGP-containing organelles. Larkin (1978) showed that plant protoplasts agglutinate in the presence of fl-glycosyl-Yariv reagent, suggesting that AGP is associated with the cell surface of protoplasts. We have investigated the subcellular distribution of hydroxyproline, a major amino acid in the protein moiety of AGP, and of AGP itself. We report here that hydroxyproline is found i n the rough-endoplasmic-reticulum fraction, the plasma membrane, and possibly in the Golgi fraction, and that AGP is associated with the plasma membrane.
Arabinogalactan proteins (AGPs) are widely distributed throughout the plant kingdom (Clarke et al. 1979; Jermyn and Yeow 1975). They are highly glycosylated (usually more than 80%) pro-
Plant material
Abbreviations: AGP = arabinogalactan protein
In most experiments we used 6-d-old seedlings of Phaseolus vulgaris L. cv. Prglude (Royal Sluis, The Netherlands) that had
Material and methods
M.R. Samson et al. : Localization of arabinogalactan protein been grown in the dark. Growth conditions were as follows: 75 g of beans were sown in a plastic tray (40 x 30 x 8 cm) filled with 5 1 Perlite (Mauritz & Zonen, Bussum, The Netherlands) wetted with 1.71 tap water; during growth the temperature was maintained at 25 ~ C, and the relative humidity at 65%.
Homogenization and fractionation Sections (3 cm long) of etiolated hypocotyls were excised directly below the hook and homogenized with a mechanical tissue chopper in homogenization buffer (40 mM 2-amino-2-(hydroxymethyl)-l,3-propane diol (Tris)-HC1, I mM dithiothr6itol, 1 m M ethylenediaminetetraacetic acid (EDTA), 3 m M or 0.1 m M MgC12, 15% (w/w) sucrose, pH 7.8; 1 ml buffer per g fresh weight). The homogenate was filtered through 10-~tm nylon gauze to remove tissue fragments and cell walls, and the filtrate was centrifuged for 5 min at 1,000 g. The supernarant was layered upon a 15-50% (w/w) linear sucrose gradient in homogenization buffer. The gradients were centrifuged (50,000 g) in a Beckman SW 20-1 rotor (Beckman Instr., Palo Alto, Calif., USA) at 4 ~ C for 16 h. One-ml fractions were collected. Sucrose concentrations were determined using a refractometer.
Marker enzymes Latent inosine-5'-diphosphatase (IDPase) (EC 3.6.1.6). Latent IDPase activity was determined after storing the sucrose gradient fractions at 4 ~ C for 3 d. The assay was a modification of the method of Galbraith and Northcote (1977), using the inorganic phosphate assay of Chandra Rajan and Klein (1976): a 25 gl sample was incubated with 225 gl substrate solution (3 m M IDP, 5 m M MgC1 z, 40 mM Tris-HC1, pH 7.5) at 25 ~ C for 40 rain. Then, 1.75 ml 10% (w/v) sodium dodecyl sulphate in 0.1 M sodium-acetate buffer (pH 4.0), 0.25 ml 2.5% (w/v) ammonium molybdate, and 0.25 ml 1% (w/v) ascorbic acid were added, and the extinction at 870 nm was determined after 20 rain.
Antimycin-A-insensitive NADPH-cytochrome-c reductase (EC 1.6.2.4). Determination of NADPH-cytochrome-c reductase was as follows: 800 gl 10 ~tM cytochrome c in 100 mM phosphate buffer containing 1 m M K C N (pH 7.7) was incubated with 100 gl sample at 25 ~ C. The increase in extinction per rain at 550 nm was determined; then 100 ~1 1 mM N A D P H was added and the increase in extinction per rain was measured again. The difference between these values gave the activity of NADPH-cytochrome-c reductase.
UDP-glueose-fl-gluean-fl-glucosyl transferase (glucan synthetase II) (EC 2.4.1.12). The glucan-synthetase-II assay was carried out essentially according to Ray (1979): 20 I~14 mM uridine 5'-diphosphate-(UDP)-glucose and 12.5 gl (1.18 kBq) [~4C]UDP-glucose (specific activity 9.08 GBq.mmo1-1) were incubated with 100-gl sample at 25 ~ C for 45 rain. The reaction was stopped by adding boiled total particles (total membrane fraction of homogenized bean hypocotyls boiled for 5 min; 20 ~tl suspension per assay) and 850 gl 80% (v/v) ethanol. The mixture was stored overnight at - 2 0 ~ C and then filtered on a Whatman G F / C glass filter (Springfield Mill, UK). The filters were washed three times with 2 ml 80% (v/v) ethanol, and radioactivity was measured.
323 HC1 at 120 ~ C for 3 h. They were then air-dried, and the residue was taken up in water. Polyacrylamide gels were hydrolysed as described by Martin et al. (1976). Hydroxyproline was determined by a modification of the method of Drozdz et al. (1976). The assay was as follows: 250 gl sample was mixed with 150 gl 3.5% chloramine T in citrate-acetate buffer (the buffer was prepared by dissolving 38 g sodium acetate.3 HzO and 25 g sodium citrate. 3 H 2 0 in 200 ml water, 255 ml isopropanol was added, and after adjusting the pH to 6.0 with acetic acid, water was added to a total volume of 1,000 ml). The mixture was kept at 3(~35~ C for exactly 10 min. Then 300 gl perchloric acid: 6% (w/v)p-dimethylaminobenzaldehyde in isopropanol (1:5, v/v) was added. After 20 rain at 60 ~ C, the extinction at 558 nm was measured.
Isoelectrofocusing on polyacrylamide gels Hydroxyproline-containing molecules were precipitated by adding five volumes of methanol to one volume of sample, and keeping these mixtures at 4 ~ for 2 h. After centrifuging at 1,500 g for 15 min, the pellets were taken up in a small volume of 1% (v/v) Triton X-100. Polyacrylamide slab gels (5 % polyacrylamide, thickness 0.7 ram) were made according to Laemmli (1970) with 2.4% (w/w) Ampholines, pH 2.5-4 (LKB, Bromma, Sweden). The gels were run on a LKB Multiphor Apparatus with 1 M glycine and I M H3PO 4 as cathode and anode buffer, respectively; the cooling temperature was 8 ~ C. Samples were applied on small paper filters placed on the middle of the gel. After 5 h at 1,000 V, 1-cm pieces were excised at one side of the gel to determine the pH, and the remainder of the gel was stained directly with fl-Gal-Yariv reagent (0.1 mg m1-1 10% (v/v) dimethylsulphoxide (DMSO) in water) overnight at 40 ~ C. The gel was destained with several changes of water.
Radioactive labeling of proteins Hypocotyl sections (3 cm long) were placed in a Petri dish on filter paper soaked in 2 gM indole acetic acid, and incubated at 30 ~ C for 1 h. They were then submerged in a [14C]proline solution (75.4 kBq m1-1) and vacuum-infiltrated for 3 min. After another 1 h at 30 ~ C in a Petri dish on wet filter paper, the sections were homogenized for 10 rain on ice with a razor blade in homogenization buffer (1 ml buffer per g fresh weight) and further treated as described in the section ,,Homogenization and fractionation".
1,3,5 Tris (4-fl-D-galactopyranosyl-oxyphenylazo) 2,4,6 trihydroxybenzene (fl-Gal- Yariv) fl-Gal-Yariv reagent was prepared by coupling diazotized paminophenyl-fl-D-galactopyranoside (Sigma) to phloroglucinol (Yariv et al. 1962).
fl-Galactosidase (EC 3.2.1.23) fl-Galactosidase activity was measured by incubating 100 ~1 sample with 100 gl 50 m M p-nitrophenyl-fl-D-galactopyranoside and 300 gl 40 mM acetate buffer, pH 4.0, at 37 ~ C. After 30 min, the reaction was stopped by adding 1 ml 200 mM glycine-NaOH, pH 10.6, and the amount of p-nitrophenol was determined by measuring the extinction at 405 nm.
Determination of hydroxyproline
Results
Sucrose gradient fractions were extensively dialyzed against water. The samples were hydrolysed in the presence of 6 M
Figure 1 A-C shows the distribution in isopycnic sucrose gradients of marker-enzyme activities for,
324
M.R. Samson et al. : Localization of arabinogalactan protein 9
i
i
i A
i
Bq 190
i
d
,
1.20
50
3
" .
%
2
1.16
1.16
1.12
t.12 25
1
1.0B
1.08
d
10
d
1.20
,&
1.16
3
1.12
-r~
2
".
t.20
9
1.16
~
1,12
d..-
1.08
I
10
20
I ~
ml
30
1
0
1.08
L 10
20
rnl
30
Fig. 1. A Distribution on isopycnic density gradients of NADPH-cytochrome-c reductase in the presence of 3 mM Mg 2+ ( o - - o ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 o, density (g cm-3). B Distribution on isopycnic density gradients of latent inosine-5'-diphosphatase in the presence of 3 mM Mg 2+ ( 0 - - 0 ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density. C Distribution on isopycnic density gradients of glucan synthetase II in the presence of 3 mM Mg 2+ ( o - - 9 or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density. D Distribution of hydroxyproline on isopycnic density gradients in the presence of 3 mM Mg ~+ ( 0 - - 0 ) or 0.1 mM Mg 2+ ( o - - o ) . 9 9 9 0, density
respectively, endoplasmic reticulum (NADPH-cytochrome-c reductase), Golgi apparatus (latent inosine-5-diphosphatase; IDPase) and plasma membrane (glucan synthetase II). In the presence of 3 mM Mg 2+, membrane-bound NADPH-cytochrome-c reductase showed a peak at a density of 1.17; also, considerable activity was detected in the soluble fractions at low sucrose concentrations. When homogenization and further steps were performed at 0.1 mM Mg 2+, most membrane-bound activity equilibrated at a density of 1.11, although some activity could still be found at higher densities. These results indicated that in our system rough endoplasmic reticulum had a density of about 1.17 and that at low Mg 2+ concentration ribosomes were stripped from the membranes; the resulting smooth endosplasmic reticulum banded at a density of 1.11. Electron-micro-
scopic observations confirmed this (results not shown). Figure 1 B shows that the activity of latent IDPase, a Golgi-apparatus marker, peaked around densities of 1.15, independent of the Mg 2§ concentration present. In Fig. 1C we show the distribution profiles of glucan synthetase II in the presence of 3 mM or of 0.1 mM MgC12. Like Robinson et al. (1982) we found a density shift of the enzyme activity, from 1.17 at 3 mM M g 2+ to 1.15 at 0.1 mM Mg 2+, indicating that in bean hypocotyls also the density of plasma membrane vesicles is influenced by the Mg 2+ concentration. Mitochondria, as judged by the marker enzyme, cytochromec oxidase, banded at 1.18 irrespective of the Mg 2+ concentration used (data not shown). Arabinogalactan protein is a hydroxyprolinerich glycoprotein and therefore the distribution of
M.R. Samson et al. : Localization of arabinogalactan protein u
u
/'d
OO0
Bq 100
1.16
1.12 50 1.08
.............J 10
20
ml
3'0
Fig. 2. Distribution of radioactivity on isopycnic density gradients after homogenization of bean hypocotyl sections in the presence of soluble [14C]hydroxyproline-containing compounds. 9 9 9 e, density (g cm 3)
hydroxyproline in isopycnic sucrose gradients was determined, both in the presence of 3 mM Mg 2§ and of 0.1 mM Mg 2§ (Fig. 1 D). In both cases, most hydroxyproline was present in the presumably membrane-free fractions (density < 1.09). In the presence of 3 mM Mg 2+ the distribution profile of membrane-bound hydroxyproline showed a peak at densities about 1.17, and a plateau at somewhat lower densities. At densities between 1.13 and 1.09 hardly any hydroxyproline was detectable. The distribution of membrane-bound hydroxyproline in the presence of 0.1 mM Mg 2+ showed a different pattern. Only a negligible amount of hydroxyproline was present at densities higher than 1.17, but two peaks appeared at densities 1.15 and 1.10. As shown above, most hydroxyproline in the gradients was present in soluble compounds. To exclude the possibility that soluble hydroxyproline-containing molecules became nonspecifically attached to membranes during homogenization, thereby causing artefacts, the following control experiment was done: we labeled hypocotyl sections with [l~C]proline; the homogenate of these sections was fractioned on a linear sucrose gradient (15-50%, w/w), and 24 fractions were collected. The fractions with a density < 1.09 contained 14C-labeled proteins including a considerable amount of [14C]hydroxyproline-containing molecules (data not shown). We homogenized unlabeled bean hypocotyl sections in the presence of these fractions. This homogenate was also fractioned by isopycnic density centrifugation, and the
325
radioactivity was determined in all fractions. As shown in Fig. 2, virtually no radioactivity was present in membrane-containing fractions; nonspecific adhesion or occlusion of hydroxyprolinecontaining material could, therefore, be excluded. Hydroxyproline is not absolutely specific for AGP, because in bean hypocotyls hydroxyproline has also been found in extensin (Van Holst et al. 1980), a well known structural component of the cell wall. To discriminate between AGP and extensin, isoelectrofocusing was used. Extensin and its precursors probably have their isoelectric points at alkaline pH values (Stuart and Varner 1980), whereas AGP has an isoelectric point at low pH because of the large amount of uronic acids in its carbohydrate side-chains (Van Holst et al. 1981). The precursors of AGP probably have isoelectric points at neutral to acid pH values. After isopycnic density centrifugation, the macromolecules from each fraction were isoelectrofocused on polyacrylamide gels (pH2.5-4.5); to visualize AGP, the gels were stained with fl-Gal-Yariv reagent. Figure 3 shows the subcellular distribution of AGP in the presence of 3 mM and 0.1 mM Mg 2§ The presumably membrane-free fractions (density < 1.09) showed several intensely stained bands at pH 2-3 at both Mg 2+ concentrations. At higher densities a single, somewhat diffuse band appeared at pH 2.8; at 3 mM Mg 2§ this band was found in fractions 1-7, whereas at 0.1 mM Mg 2§ it was found in fractions 3-9. Furthermore, at 0.1 mM Mg 2+ this band stained consistently less than at 3 mM Mg 2+, possibly because the association between AGP and membrane is Mg 2§ dent. Comparison of Figs. 3 and 1 C shows that the subcellular distribution of membrane-bound AGP parallels that of glucan synthetase II, including the shift to lower densities at lower Mg 2§ concentration. No clear bands can be seen at other densities. Since fl-galactosidase activity might obscure the picture by hydrolysing the fl-Gal bond in the Yariv reagent and so generating an insoluble red compound, we assayed for fl-galactosidase activity with p-nitrophenyl-fl-D-galactoside as a substrate. No activity could be found at pH values where redcoloured band were visible in the gel (results not shown). Additional evidence that AGP was indeed responsible for the fl-Gal-Yariv binding in the polyacrylamide isoelectrofocusing gels was obtained by isofocusing both a - presumably - membrane-flee fraction (density __,
t_ "I3
i
x: 60
t
i
B
20 )H >~ 3.8'3.5'3.3'30'2.0' 2.L
Fig. 4A, B. Percentages of total hydroxyproline determined in polyacrylamide gels after isoelectrofocusing a membranous fraction (A) and a soluble fraction (B). The a r r o w s indicate the position of sample application on the gel
organelles or from the cytosol. The distribution of hydroxyproline in the membrane-containing fractions indicated that hydroxyproline-containing molecules are associated with rough endoplasmic reticulum, since a clear shift of hydroxyprolinecontaining molecules to lower densities is seen when 0.1 mM Mg 2 § is used instead of 3 mM; this shift coincides with a shift in NADPH-cytochrome-c-reductase activity and indicates a conversion of rough endoplasmic reticulum to smooth endoplasmic reticulum at low Mg 2+ concentration. A small amount of hydroxyproline-containing molecules is possibly associated with the Golgi apparatus. Association of hydroxyproline-containing molecules with the plasma membrane seems clear, since at 3 mM Mg 2§ hydroxyproline peaks at about 1.17, whereas at 0.1 mM Mg 2+ a new peak appears at 1.15; the same shift is shown by glucan-synthetase-II activity, a plasma-membrane marker enzyme. The presence of hydroxyprolinecontaining molecules in membranous fractions, however, does not give conclusive evidence for the presence of AGP, since hydroxyproline is also found in extensin and in precursor molecules. To further identify the subcellular fractions with which AGP was associated we used isopycnic density centrifugation followed by isoelectrofocusing on polyacrylamide gels. After isoelectrofocusing the gels were stained with fl-Gal-Yariv reagent, which is bound specifically by AGPs. Bands appeared only at low pH values, mostly between pH2.5 and 3.0. Arabinogalactan proteins are known to have their isoelectric point at low pH (Van Holst et al. 1981) because of the large amount
327
of uronic acids in their carbohydrate chains. Analysis of the gels for hydroxyproline showed that most hydroxyproline was in the very acid regions of the gels, where the/?-Gal-Yariv-stained bands appear, again indicating that these bands indeed represent AGPs. Interestingly, no clear indications for the presence of the other hydroxyproline-rich protein, i.e. extensin, or its precursors, were found. In the - presumably - soluble fractions (fractions 15-22, density
