ARCHIVES
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 289, No. 1, August 15, pp. 137-144, 1991
Carbohydrate-Binding Specificity of Calcyclin and Its Expression in Human Tissues and Leukemic Cells Fu-Yue
Zeng and Hans-Joachim
Max-Planck-Institut
fiir experimentelle
Gabiusl Medizin,
Abteilung
Chemie, Hermann-Rein-Str.
3, D-3400 Giittingen,
Germany
Received December 27, 1990, and in revised form April 12, 1991
Binding of biotinylated fetuin in a solid-phase assay served as activity assay for purification of calcyclin, the product of a cell growth-related cDNA with homologies to Ca2+-binding proteins. Asialofetuin failed to bind to calcyclin, emphasizing the importance of sialic acids. Binding of fetuin was most effectively reduced by N-glycolylneuraminic acid within a panel of mostly negatively charged sugars. Bovine submaxillary mucin and the ganglioside GM1, but not asialo-GM,, proved more effective than neoglycoproteins, carrying negatively charged carbohydrate moieties. Extension of N-acetylneuraminic acid to its 1acto;syl derivative increased its inhibitory potency. Among charge-free carbohydrate residues, only N-acetylglucosamine, lactose, and mannose, but not fucose, melibiose, or N-acetylgalactosamine affected fetuin binding, substantiating the inherent selectivity. Chemical modification with group-specific reagents revealed that lysine and arginine residues appear to be involved in ligand binding that is optimal in the presence of Ca’+, but not Zn”+ and stable up to 1 M NaCl. Biotinylation of calcyclin by modification of carboxyl groups facilitated performance of solid-phase assays with calcyclin in solution, yielding similar results with (neo)glycoproteins in relation to assays with immobilized calcyclin, thereby excluding an impact of binding to nitrocellulose on calcyclin’s specificity. Subcellular fractionation disclosed the presence of fetuin-binding activity in all fractions, the specific activity decreasing from the nuclear to the particulate cytoplasmic fraction and the cytoplasmic supernatant. Affinity-purified antibodies were employed to detect high levels of calcyclin expression in acute lymphoblastic, myelogenous, and monocytic leukemia cell lines, but not in myeloma or lymphoblastoid cells. In comparison, most cells were nearly devoid of an 0-acetylsialic acid-specific protein that is more abundant 0 1991 Academic in various tissue types than calcyclin. Press,
Inc.
1 To whom correspondence
should be addressed.
0003.9861/91$3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
The search for genes that undergo growth-related changes of expression and their subsequent analysis is a rational approach to obtain information on growth regulation. It has led to the establishment of a series of cDNA clones due to the marked increases of levels of certain types of mRNA concomitant to growth stimulation and cell cycle progression of quiescent fibroblasts by serum (1, 2). Sequence comparison of a cDNA clone, termed 2A9, revealed a notable extent of homology to Ca’+-binding proteins like S-100, prompting the putative protein product to be referred to as calcyclin (3). Its sequence was later found to be identical to a prolactin receptor-associated protein (4). Evidently, elucidation of the function of this gene product requires its purification. This has independently been achieved with the search for further Ca’+-binding proteins as well as for proteins with affinity to sialic acid residues (5,6). The latter property may enable calcyclin to selectively interact with sialic acids of carbohydrate chains of cellular glycoconjugates, contributing to the determination of its position in the cell. It was thus of interest to assess calcyclin’s carbohydratebinding specificity. Similarly, the availability of specific antibodies allowed the measurement of its distribution in human tissues and in leukemic cells, where overexpression has been disclosed by measuring the amounts of 2A9specific mRNA in blot hybridization assays (7,8). EXPERIMENTAL
PROCEDURES
Purifiication of caky&z. Fetuin (Biomol, Hamburg, FRG) was coupled to divinylsulfone-activated Sepharose 4B with a coupling yield of 25 mg/ml column resin. Processing of 10 g of bovine heart started routinely with homogenization in 50 ml 20 mM Tris-buffered saline (TBS), pH 7.8, containing 20 mM CaClz, 4 mM @-mercaptoethanol, and 0.01 mM benzylsulfonyl fluoride (buffer A). Following centrifugation at 20,OOOg for 45 min, the resulting supernatant was passed through a fetuin-Sepharose 4B column (20 X 1.5 cm) at a flow rate of 5 ml/h that subsequently was washed extensively with buffer A containing 1 M NaCl to remove contaminants. The fetuin-binding activity was eluted with buffer A, substituting 20 mM CaCl, with 4 mM EDTA. The eluate was
* Abbreviations fered saline.
used: TBS, Tris-buffered
saline; PBS, phosphate-buf137
138
ZENG
AND
concentrated by ult~~ltration using a membranous YM-5 filter, dialyzed against buffer A in tubing impermeable to molecules with a molecular weight above 6 kDa and fractionated on a Sephadex G-75 column (160 X 0.5 cm) at a flow rate of 2.5 ml/h. Alternatively, the supernatant after homogenization and centrifugation was fractionated by (NH&SOI precipitation. The precipitate from 50-100% saturation, wherein the pH had been adjusted to 4.0 with acetic acid, was dissolved in 10 mM NHHCO%, pH 8.0, and dialyzed against this solution. Gel filtration on a Sephadex G-75 column (100 X 2.0 cm) at a flow rate of 20 ml/h, concentration of activity-containing fractions by ultrafiltration, dialysis against buffer A, and chromatography on a fetuin-Sepharose 4B column (20 X 1.2 cm) followed. Calcyclin was also obtained, when the precipitated protein was dissolved in 10 mM Trie-buffered saline, pH 8.3, containing 2 mM EDTA and 2 mM @-mercaptoethanol. The clear solution was applied to a DEAE-Sephadex A-50 column (25 X 2.0 cm) and protein was eluted by application of a linear gradient of NaCl up to 0.2 M salt. Activity-containing fractions were pooled, concentrated, dialyzed against 10 mM NH,HCOe, pH 8.3, and further purified by gel filtration on a Sephadex G-75 column (100 X 2.0 cm). Analytical procedures. Protein was determined by the dye-binding assay, adapted for microtiter plates, with bovine serum albumin as standard (9). One- and two-~mensional gei el~trophoretic analyses under reductive and denaturing conditions with a 12.5 or 15% running gel and a 3% stacking gel combined with isoelectric focusing by nonequilibrium pH gel electrophoresis and a highly sensitive silver staining procedure were carried out, as described (10). The isoelectric point (pl) of calcyclm was calculated on the basis of the deduced amino acid sequence of the cDNA clone (3). The calculation was performed as described (11). Analytical gel filtration was performed on a Sephadex G-50 column (100 X 0.5 cm) equilibrated with 20 mM Na~HPO~/KH~PO~ buffer, pH 7.2. The applied protein was eluted at a flow rate of 2.5 ml/h. Solid-phase activity assay. Fetuin was biotinylated using biotinylN-hydroxysuccinimide ester, as described (12). Spotting detection of activity was based on the binding of biotinylated fetuin to calcyclin, immobilized on nitroceilulose squares (1 X 1 cm; 0.2 gm from Schleicher & Schuell, Dassel, FRG), and the subsequent, highly sensitive detection of the Iabel biotin, as analogously performed for other carbohydratebinding proteins (10,13). In detail, following spotting of 1~1 of calcyclincontaining solution (1 pg/pl), the nitrocellulose squares were allowed to dry for 30 min. The squares, placed in separate wells of tissue culture plates, were incubated for 1 h at room temperature with a 3% (w/v) solution of bovine sernm albumin in 20 mM Tris-HCl buffer, pH 7.5, containing 0.9% NaCi to block residual protein-binding sites. After thorough washing with this buffer containing 0.2% bovine serum albumin and 2 mM CaClx, incubation with 20 pg/mI biotinylated fetuin in the absence or presence of inhibitors was carried out for 2 h at room temperature. To determine the dependence of binding on ionic strength and divalent cations, the buffer was supplemented by increasing concentrations of NaCl up to 4 M or by 10 mM of the respective cation. To assess the influence of pH value on binding, various buffer systems (pH 3-6: 50 mM sodium acetate, pH 6.5-9.0: 50 mM Tris-HCI, pH 9-12.5: 50 mM glycine/NaOH) were used in this step. Similarly, the inhibitory potency of a panel of commercially available glycosubstances as well as neoglycoproteins, prepared by diazo coupling of p-aminophenyl glycosides to carbohydrate-free bovine serum albumin as carrier with an incorporation level of 10 + 2 carbohydrate moieties per carrier molecule (12), was measured. Unbound ligand was carefully removed by repetitively incubating the nitrocellulose squares with fresh buffer solution (20 mM Tris-HCl, pH 7.5, containing 0.9% NaCI, 0.2% bovine serum albumin, and 2 mM CaCI,) for 5 min. For quantitative evaluation of bound biotinylated fetuin as a measure of calcyclin presence, a solution of the avidin-peroxidase conjugate (10 pg/ml; Sigma, Munich, FRG) was applied for 90 min at room temperature. Extensive washing followed prior to the visualization of bound probe with 4-~hloro-l-naphthol/H~O~ solution as substrate and quantitation, as performed recently (10,13). The activity is expressed as titer-‘. The titer is the highest dilution of calcyclin-containing solution that yielded a definitely positive reaction in the assay.
GABIUS V~~izat~n of fetu~n-bi~i~ after e~ct~b~tting of ca~cl~n. Transfer of purified calcyclin (10 pg) after separation on a 15% running gel under denaturing and reductive conditions was carried out with the semi-dry blot system Sartoblott II (Sartorius, GBttingen, FRG). The two horizontal graphite plates enclosed a sandwich comprising (from the anodic plate): two layers of Whatman No. 1 filter paper soaked in 0.3 M Tris-HCl/BO% (v/v) methanol (pH 10.4), 1 layer of filter paper soaked in 25 mM his-HCl/20% (v/v) methanol (pH lo.@, the nitrocellulose membrane as well as the polyacrylamide gel, both soaked in 25 mM Tris-HCl/20% (v/v) methanol (pH 10.4), and three layers of filter paper soaked in freshly prepared 40 mM 6-amino-n-hexanoic acid/ 25 mM Tris-HCl/20% (v/v) methanol (pH 9.4). Transfer was performed at room temperature for 30 min at 400 mA. Following transfer, the membrane was washed twice for 5 min in buffer (20 mM Tris-HCl, pH 7.8, containing 0.2% bovine serum albumin and 0.9% NaCl) and the remaining binding sites on the nitroceliulose were blocked by incubation in 50 mM Tris-HCI (pH 7.8) containing 5% bovine serum albumin and 0.9% NaCl for 30 min at room temperature. After extensive washing the blots were incubated with 20 pg/ml biotinylated fetuin in TBS contaming 20 mM CaCl*, 0.9% NaCl, and 0.5% bovine serum albumin in the absence or presence of inhibitors (fetuin, asialofetuin in 500-fold excess). Visualization of bound probe was achieved by applying the same protocol, as described for the solid-phase assay. Chemical modification of amino acid residues. In each reaction 7.2 nmol calcyclin was chemically modified by adding reagents that exhibit specificity to certain types of functional groups under conditions given in detail elsewhere (13, 14). Namely, 0-methylisourea, citraconic anhydride, cyclohexane-1,2dione, phenytglyoxai, N-acetyhmidazole, diethy~pyr~arbonate, and glycine methyl es~r/l-ethyl-3-{3-~methylaminopropyl]-carb~iimide were employed. In all cases excess reagents were removed by dialysis against 20 mu TBS, pH 7.8, prior to performing the solid-phase assay with biotinylated probe. To exclude any further influence of the conditions, under which modification is carried out, on the activity, respective controls were performed under identical conditions in the absence of modifying reagent. ~iotiny~~n of caky&n. Purified calcyclin (1.0 mgf was biotinyla~ using biotinamidocaproyl hydrazide (34.2 mg; Sigma, Munich, FRG) in the presence of 19.2 mg of the coupling reagent I-ethyl-3-(3-dimethylaminopropyl)-carbodiimide under mild conditions, optimized for the @nerve growth factor (15). The solid-phase assay was performed under identical conditions, as given in detail above, with 20 rg/ml biotinylated calcyciin as probe. Subcel~u~r fract~o~tion and correlation of fetu~n-bided activity to the different fractions. All individual steps were performed at 4”C, as given in detail elsewhere (14). Fresh bovine heart was homogenized in 5 vol 0.3 M sucrose, 4 mM P-mercaptoethanol and 3 mM MgCl,. The extract was centrifuged at 12,000g for 10 min, resulting in the cytoplasmic supernatant fraction. The pellet was resuspended in 2 M sucrose containing 4 mM P-mercaptoethanol and 1 mM MgC!l,. The top 3 ml of each sample after ~entri~gation in a swinging bucket rotor at 40,OOOgfor 1 h was referred to as the cytoplasmic particulate fraction. The pellet was resuspended in a Potter-Elvehjem homogenizer with 25 mM Tris-HCl buffer, pH 7.8, containing 200 mM NaCI, 0.1 mM benzylsulfonyl fluoride, 1% Triton X-100, and 1 mM dithiothreitol. After centrifugation at 12,000g a DNA-rich pellet and the nuclear supernatant were obtained. Assessment of binding activity by solid-phase assays was based on specimens from each fraction, diluted to a similar concentration with TBS and promptly monitored for activity. Preparation of the antibody. To raise specific polyclonal antibodies in rabbits, matrix-immobilized antigen was used, as described for an endogenous lectin (16). Calcyclin was transferred after electrophoretic separation under reductive and denaturing conditions from polyacrylamide gels onto nitrocellulose membranes and stained with red Ponceau solution. Nitrocellulose strips with immobilized calcyclin were destained in water, minced with a mortar and pestle under liquid nitrogen, and the resulting nitrocellulose powder containing approximately 100 fig calcyclin was injected intradermally in a rabbit in 1 ml phosphate-buffered saline. The IgG-fraction after at least three injections was purified
BINDING
OF CARBOHYDRATES TABLE
139
TO CALCYCLIN
I
Purification of Calcyclin by Three Different Methods Total protein (md
Step 1
Specific activity”
400
~ Crude extract
5
2000
100
Purification (-fold) 1
Fetuin-Sepharose Sephadex G-75
2b/c 3b 4b
(NH,)zSO,-precipitation Sephadex G-75 Fetuin-Sepharose
75 2.8 0.08
10 150 300
750 420 24
37.5 21 1.2
2 30 60
3c 4c
DEAE-Sephadex Sephadex G-75
10 0.98
50 300
500 294
25 14.7
10 60
A-50
24.0 (120) 22.8 (114)
Recovery (X)
2a 3a
0.12 (0.60)’ 0.076 (0.38)
200 300
Total activity”
’ Specific activity is expressed as titer/mg protein/ml. b Total activity is expressed as specific activity X mg (titer X ml). ’ Values in parentheses denote results that are obtained with elution of the column with 0.1 M NH,OH
by protein A-Sepharose chromatography (Pharmacia, Freiburg, FRG). Further purification was facilitated by affinity chromatography on calcyclin-Sepharose 4B, obtained by coupling of calcyclm to divinylsulfoneactivated Sepharose with a yield of 760 pg calcyclin/ml. Elution was pH 2.3, and 100 mM subsequently performed with 100 mM glycine-HCl, triethylamine, pH 11.5. The eluants were quickly neutralized and extensively dialyzed against phosphate-buffered saline, pH 7.2. Western blots of S-100 A (01,p) from Sigma (Munich, FRG), of pll, kindly provided by Dr. V. Gerke (Goettingen, FRlG), and of calcyclin or tissue extract were probed with affinity-purified antibody (10 pg/ml) at room temperature overnight. The following individual steps were identical to the visualization of binding of biotinylated fetuin, given above. However, GAR/IgG (H + L)-peroxidase (5 pg/ml; Nordic, Bochum, FRG) substituted avidin-peroxidase. Quuntituttin of cc&&n in tissue and cell extracts. The various human tissues, obtained at autopsy, were homogenized in 20 mM TBS, pH 7.5, containing 4 mM EDTA. Cell pellets of various human tumor lines, listed in the footnotes to Table VII, were similarly processed. After centrifugation at 20,OOOgfor 45 mi:n each supernatant was used to inhibit antibody binding to immobilized calcyclin. Aliquots of 100 ~1 containing 2 pg/ml purified calcyclin in 0.2 M Na&Os, pH 9.2, had been used for coating polystyrene microtiter wells at 37°C for 2 h. Wells were washed saline, pH 7.2, containing four times with 10 mM phosphate-buffered
1.2 (6.0) 1.14 (5.7)
40 60
(pH 11.0) instead of 4 mM EDTA.
0.5% Tween 20 (PBS-Tween). The remaining free protein-binding sites were then blocked with 150 pl of PBS containing 5% bovine serum albumin at 37°C for 1 h followed by further washes. Affinity-purified antibody (100 pl with a concentration of 1.0 pg/ml) in PBS + 5% BSA was added. After 2 h at 37°C the wells were washed six times with PBSTween and then incubated with GAR/IgG (H + L)-peroxidase conjugate (5 pg/ml in PBS + 5% bovine serum albumin) at 37°C for 1 h. Wells were again washed six times with 150 ~1 PBS-Tween. Following the last wash, a 150-~1 aliquot of 0-phenylenediamine hydrochloride/H202 solution (0.4 mg OPD and l).d30% HzOz per milliliter of 75 mM sodium citrate buffer, pH 5.5) was added to each well. The reaction was allowed to proceed for 10 min in the dark and was stopped by the addition of 50 pl of 4 N sulfuric acid, and the absorbance at 490 nm was determined using a microplate reader equipped with a narrow bandpass filter for 490 nm. The extent of antibody binding was reduced in correlation to the tissue content of the antigen, when extracts, diluted in PBS containing 5% bovine serum albumin, were preincubated with the antibody for 1 h at 37°C. A standard curve for quantitative assessment of calcyclin was obtained by preincubation of antibody-containing solution with known amounts of purified antigen. Besides calcyclin, the tissue distribution of a human 0-acetylsialic acid-specific protein from human placenta, purified as described (13), was similarly determined.
RESULTS
V’
2
3
4
Molecular
4
6;ibio Weight
(x10-4
)
FIG. 1. Determination of molecular weight of calcyclin by analytical gel filtration on a Sephadex-G50 column (160 X 0.5 cm) that has been calibrated with standards.
Purification of calcyclin from bovine heart. Calcyclin was purified from bovine heart to apparent homogeneity by three different methods that yielded identical specific activities for binding of biotinylated fetuin (Table I). This assay was chosen as an indicator of carbohydrate binding. The combination of ion-exchange chromatography, where the protein eluted around 0.11 M NaCl in the gradient, and gel filtration proved most effective in terms of recovery. Analytical gel filtration with purified calcyclin revealed a molecular weight of approximately 11,000 (Fig. 1). The experimentally determined isoelectric point of the purified protein is identical within the limits of detection to the theoretically calculated pI value of 4.3 on the basis of the cDNA sequence, arguing against major chargeshifting post-translational modifications (Fig. 2). The activity of binding was nearly unaffected by pH changes within the range from 5 to 8. However, sharp decreases
140
ZENG
AND
GABIUS
PH
FIG. 2. Visualization of calcyclin by silver staining after two-dimensional gel electrophoresis involving isoelectric focusing in the first dimension and sodium dodecyl sulfate-polyacrylamide (15%) gel electrophoresis under reductive conditions in the second dimension. Positions of standards for molecular weight designation are indicated by arrowheads.
occurred outside of this range (Fig. 3). With respect to alterations of the ionic strength, the activity was rather stable, when NaCl molarities up to 1.5 M NaCl were present. Further increases led to diminution of binding of fetuin. Among the cations that had been tested at a concentration of 10 mM, Ca2+, Mg2+, Co2+ Ni2+, and Mn2+ enhanced binding of fetuin. The presence of Cd2+, Zn2+, and Cu2+ allowed only partial binding. A decrease of binding activity was noted below a concentration of 1 mM for Ca2+ and Mg2’. Binding of labeled (neojglycoproteins to calcyclin. Biotinylated fetuin had been used as ligand to assess the binding activity. Labeled asialofetuin failed to bind to calcyclin, emphasizing the importance of sialic acid moieties for binding. This result was independently obtained in solid-phase assays with biotinylated probes as well as in inhibition experiments on blots in the presence of fetuin and asialofetuin as inhibitors of binding of labeled fetuin (Fig. 4, Table II). To analyze the selectivity of calcyclin for carbohydrates more closely, several neoglycoproteins, carrying the same density of carbohydrate residues with the same spacer, were employed. Notably, the carrier protein exhibited no ligand properties (Table II). Immobilized
a
0
1
2
3
15
6
7
8
9
10
11
12
13
pH - value
FIG. 3. Determination ylated fetuin to calcyclin residues are indicated.
of pH-dependence of the binding of biotinin a solid-phase assay. pK, values of certain
C
negatively charged carbohydrate residues caused binding of the labeled carrier to calcyclin. Similarly, carrier-conjugated N-acetylglucosamine, mannose, and lactose, but not the a-galactoside melibiose, N-acetylgalactosamine, or fucose, led to a signal (Table II). The sialic acid a-2,6galNAc-a-Thr/Ser-rich bovine submaxillary mucin exhibited comparatively strong ligand properties.
TABLE
II
Binding Activity of Calcyclin to Biotinylated (Neo)glycoproteins in a Solid-Phase Assay
(Neo)glycoprotein
00
b
FIG. 4. Visualization of fetuin-binding activity of immobilized purified calcyclin (10 rg) after electrophoresis under reductive and denaturing conditions and transfer to nitrocellulose with biotinylated fetuin in the absence of further substances (a) as well as in the presence of asialofetuin (b) and fetuin (cl. Standards for molecular weight designation are indicated by arrowheads.
Fetuin Asialofetuin BSM” BSAb Sialic acid’ Glucuronic acid’ Mannose-&phosphate’ Galactose-6-phosphate’ @-glcNAc’ &galNAc” cu-Fucose’ Lactose’ Melibiose’ Mannose”
Binding activity (%) 100 0 90 0 20 60 60
50 40
0 0 40
0 40
’ Bovine submaxillary mucin. * Bovine serum albumin. ’ The carbohydrate part is given to define the neoglycoprotein.
BINDING
OF CARBOHYDRATES
TABLE
TABILE III Inhibition of Fetuin-Bind.ing of Calcyclin and Glycosubstances ,in a Solid-Phase
Carbohydrates and glycosubstances Neu5Ac Neu5Gc N-Acetylneuraminyllactose Glucuronic acid Mannose-6-phosphate Galactose-6-phosphate glcNAc Lactose Mannose GM1 Asialo-GM1 BSM” Sialic acid-BSA Mannose-g-phosphate BSA Galactose-6-phosphate BSA glcNAc-BSA lac-BSA man-BSA Heparin Fucoidan Dextran sulfate
141
TO CALCYCLIN
by Sugars Assay
Concentration required for 50% inhibition (mM) >lOO 3 25 50 (20% inhibition) 25 30 100 50 50 (20% inhibition) 3.0 5.0 (no inhibition) 0.1 0.15 0.12 0.12 0.15 (20% inhibition) 0.15 (25% inhibition) 0.12 (20% inhibition) 5 mg/ml (20% inhibition) 2 mg/ml 1.5 mg/ml
’ Denotes concentration in terms of sialic acid; 50 mM galNAc, 50 mM melibiose, and 50 mM fucose were ineffective.
The tested panel of sugar residues was enlarged, when the inhibition of binding of biotinylated fetuin in the presence of glycosubstances was monitored, as summarized in Table III. N-Glycolylneuraminic acid was more efficient as inhibitor than the acetylated derivative. However, the extension of its carbohydrate structure to the a-2,3/a-2,6-containing N-acetylneuraminyl-lactose improved the extent of inhibition considerably. Removal of the a-2,3-linked N-acetylneuraminic acid residue from the ganglioside GM1 markedly reduced its capacity as an inhibitor. The binding activity is yet not strictly dependent on the presence of sialic acids. Neoglycoproteins with carboxylated or phosphorylated sugars were rather similarly effective as inhibitors in this assay. As already noted with labeled neoglycoproteins, lactose, N-acetylglucosamine, and mannose, too, have an impact on binding of fetuin. Fetuin binding was also impaired by the presence of sulfated polysaccharides. Dextran sulfate and fucoidan reduced it more effectively than heparin. Chemical modification of amino acid residues withgroupspecific reagents. To infer the nature of molecular interactions between calcyclin and its carbohydrate ligands, we analyzed the effect of modification of certain amino acid residues by group-specific reagents. As summarized in Table IV, impairment of fetuin binding mainly occurred after modification of the functional groups of lysine and arginine. No dependence for binding activity on the in-
IV
Effect of Amino Acid Modification by Group-Specific Reagents
on Binding
of Biotinylated
Residues modified
Chemical treatment Native protein 0-Methylisourea Citraconic anhydride
Fetuin Binding activity (%) 100 40 40
Lysine Lysine, Hz-terminal amino group Arginine Arginine Tyrosine Histidine Carboxyl
Cyclohexane-l$dione Phenylglyoxal N-Acetylimidazole Diethylpyrocarbonate Ester-carbodiimide
40 60 100 100 100
tegrity of tyrosine, histidine, or carboxyl groups could be disclosed. Since binding activity was retained after carboxy1 esterification with glycine methyl ester, these groups can thus be biotinylated to lead to labeled calcyclin that should still exhibit activity for fetuin binding. This tool can be employed to answer the question whether immobilization to nitrocellulose that so far has commonly been employed for the determination of binding activity influences its carbohydrate specificity. Binding of labeled calcyclin to (neo)glycoproteins. Using immobilized (neo)glycoprotein binding of the biotinylated calcyclin from solution to the matrix was assessed. No significant changes in the ligand properties of the various (neo)glycoproteins could be discerned, as shown in Table V. In addition to exclusion of an effect of immobilization on binding activity the access to biotinylated calcyclin will eventually be of value to localize ligands in situ. It is likewise noteworthy that no information on the intracellular localization of calcyclin itself
TABLE Binding Activity (Neo)glycoproteins
(Neo)glycoprotein Fetuin Asialofetuin BSM Sialic acid” Glucuronic acid” Mannose-6-phosphate” Galactose-B-phosphate” @-glcNAc” @-galNAc” lac ’ man ’ ’ The carbohydrate
V
of Biotinylated Calcyclin to in a Solid-Phase Assay Binding activity (%) 100 0 100 40 50 70 70 50 10 25 20
part is given to define the neoglycoprotein.
142
ZENG TABLE Distribution
Fraction Whole cell extract Cytoplasmic supernatant Cytoplasmic particulate fraction Nuclear fraction
AND
GABIUS
VI
of Fetuin-Binding Activity in Subcellular Fractions of Bovine Heart
Volume (ml)
Total protein bd
40
400
5
2000
38.5
350
4
1400
17.5
20
2
50
350 100
Specific activity”
Total activity* a
15 10
a Specific activity is expressed as titer/mg/ml. * Total activity is expressed as specific activity
X mg.
is available. Thus, subcellular fractionation combined with activity assays was employed to address this issue. Subcellular localization of fetuin-binding activity. Three subcellular fractions, referred to as the nuclear fraction, to which the DNA was confined, a cytoplasmic particulate fraction, and a cytoplasmic supernatant fraction, were established. Activity was found to be distributed among each of the three parts, the specific activity being highest in the nuclear fraction (Table VI). Further studies on the precise localization of calcyclin require the availability of specific antibody. Consequently, we sought to provide this tool to check its specificity and to use it for the determination of expression of calcyclin in tissues and cells to enable focusing of immunohistochemical studies. Immunological determination of calcyclin in tissue and in cell extracts. Matrix-bound calcyclin was applied as antigen to raise specific polyclonal antibodies. The serum was processed by chromatography on protein A-Sepharose as well as on calcyclin-Sepharose. The level of selectivity of the affinity-purified antibodies was ascertained by immunoblotting with structurally related Ca2+-binding proteins and the tissue extract (Fig. 5). Their availability prompted the analysis of expression of calcyclin in tissues and cells. Due to the reported overexpression of calcyclin mRNA in human leukemic cells (7,8) we chose to include extracts of various cell lines into this series of measurements. A specific antibody to a human 0-acetylsialic acidbinding protein, purified recently (13), was included to determine whether these two proteins with affinity to sialic acids may be expressed at similar levels. A quality control for this antibody fraction is shown in Fig. 6. Tissue and cell extracts were employed to reduce binding of the antibody to the two purified proteins according to their antigen content. Quantitation was based on the respective calibration curves, established with the purified proteins (Fig. 7). Calcyclin was measurable in several organs, the levels being lower than for the lectin (Table
b
c
d
FIG. 5. Analysis of reactivity of calcyclin-specific antibodies to 20 pg purified S-100A ((u, 8) (a), 20 pg purified pll (b), 2 pg purified calcyclin (c), and extract of bovine heart (d) after gel electrophoretic separation under reductive and denaturing conditions and electrophoretic transfer to nitrocellulose. Positions of standards for molecular weight determination are indicated by arrowheads.
VII). Notably, it is present in acute lymphoblastic, myelogenous, and monocytic leukemia cell lines in comparatively high abundance relative to myeloma or lymphoblastoid cells. This result corroborates measurements on the level of the mRNA, warranting further immunohistochemical studies on these types of cell. The protein, however, is nearly undetectable in the three types of leukemia lines with high calcyclin levels. Expression of these two sialic acid-binding proteins is apparently not coregulated in leukemic cells. DISCUSSION
cDNA cloning of growth-regulated mRNAs from quiescent fibroblasts that had been stimulated by serum or certain mitogens disclosed preferential expression of certain gene products (1, 2). One putative protein was designated calcyclin on the basis of its cell cycle-dependent expression and its sequence homologies to charac-
FIG. 6. Analysis of reactivity of antibodies against the human Oacetylsialic acid-specific protein to extract proteins from human kidney after gel electrophoretic separation under reductive and denaturing conditions and electrophoretic transfer to nitrocellulose. Positions of standards for molecular weight designation are indicated by arrowheads.
BINDING
OF CARBOHYDRATES
143
TO CALCYCLIN
2A9 (ns)
53KDa
Lectin
(ng)
FIG. 7. Calibration curves with purified calcyclin (a) and 0-acetylsialic acid-specific protein (b), termed 53-kDa protein according to its molecular weight, for determination of the plresence of the respective proteins in tissue extracts by ELISA.
terized Ca2+-binding proteins (3). Calcyclin was later referred to as a prolactin receptor-associated protein, present in certain, but not a.11,prolactin receptor-positive human breast cancer cell lines (4). A similar heterogeneity of calcyclin expression for a distinct category of cells had been reported in the case of specimens from patients with acute myelogenous leukemia (7,8). Calcyclin is also present in a number of mouse and rat tissues, as indicated by Northern and Western blots (4, 17). When native calcyclin is employed as inhibitor for binding of specific antibody to immobilized extract proteins, the presence of calcyclin in various human tissue types is ascertained. Subcellular fractionation of heart extract indicates that its activity is not confined to one compartment. Notably, the monitoring of extracts detects high levels of calcyclin expression for acute myelogenous and monocytic leukemia cell lines and a pronounced reduction for lymphoblastoid or myeloma cells. To shed light on calcyclin’s physiological role, the two characteristics that have guided the purification certainly deserve attention. The binding of Ca2+ and of sialic acids may even be connected to each other. Ca2+, but not Zn2+ or Cd’+, supports optimal binding to carbohydrates. These two cations are incapable of competing with Ca2+ for its binding sites, whose occupation induces a conformational change (18). It is certainly :remarkable to find at least two properties, namely cation and carbohydrate binding, in a protein of a molecular weight of approximately 10 kDa. Coincidentally, sialic acid-binding lectins from frog eggs exhibit a similarly small molecular weight (19, 20). Calcyclin’s ability to bind the sialoglycoprotein fetuin strictly requires the presence of the sialic acid moieties. Its apparent affinity to other carboxylated or to phosphorylated sugars as well as to sulfated polysaccharides is reminiscent of binding properties of a number of proteins that are known to interact with neuraminic acids, like an agglutinin from rat uterus (21), the interferon antagonist sarcolectin (22), a lectin from kidney (23) and
from brain synaptic vesicle-enriched fractions (24), the murine lymphocyte homing receptor MEL-14 (15), and the platelet granular-membrane protein GMP-140 (26, 27). A lectin from stratum corneum even exhibits a markTABLE
VII
Distribution of Calcyclin and the 0-Acetylsialic Acid-Specific Protein (53 kDa) in Human Tissues and Leukemic Cell Lines
Tissues or cells Heart Tongue Small intestine Stomach Lungs Liver Kidney cortex Aorta Brain Spleen Lymph node Adrenal gland Thyroid gland Pancreas Testes KG-l” KG-lab K-562’ THP-ld RPMI-8226’ CCRF-CEM’ Croco BP Hs-294-Th
Calcyclin hdmd
53 kDa (w/md
375 300 290 200 300 145 110 1110 280 150 160 280 160
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 289, No. 1, August 15, pp. 137-144, 1991
Carbohydrate-Binding Specificity of Calcyclin and Its Expression in Human Tissues and Leukemic Cells Fu-Yue
Zeng and Hans-Joachim
Max-Planck-Institut
fiir experimentelle
Gabiusl Medizin,
Abteilung
Chemie, Hermann-Rein-Str.
3, D-3400 Giittingen,
Germany
Received December 27, 1990, and in revised form April 12, 1991
Binding of biotinylated fetuin in a solid-phase assay served as activity assay for purification of calcyclin, the product of a cell growth-related cDNA with homologies to Ca2+-binding proteins. Asialofetuin failed to bind to calcyclin, emphasizing the importance of sialic acids. Binding of fetuin was most effectively reduced by N-glycolylneuraminic acid within a panel of mostly negatively charged sugars. Bovine submaxillary mucin and the ganglioside GM1, but not asialo-GM,, proved more effective than neoglycoproteins, carrying negatively charged carbohydrate moieties. Extension of N-acetylneuraminic acid to its 1acto;syl derivative increased its inhibitory potency. Among charge-free carbohydrate residues, only N-acetylglucosamine, lactose, and mannose, but not fucose, melibiose, or N-acetylgalactosamine affected fetuin binding, substantiating the inherent selectivity. Chemical modification with group-specific reagents revealed that lysine and arginine residues appear to be involved in ligand binding that is optimal in the presence of Ca’+, but not Zn”+ and stable up to 1 M NaCl. Biotinylation of calcyclin by modification of carboxyl groups facilitated performance of solid-phase assays with calcyclin in solution, yielding similar results with (neo)glycoproteins in relation to assays with immobilized calcyclin, thereby excluding an impact of binding to nitrocellulose on calcyclin’s specificity. Subcellular fractionation disclosed the presence of fetuin-binding activity in all fractions, the specific activity decreasing from the nuclear to the particulate cytoplasmic fraction and the cytoplasmic supernatant. Affinity-purified antibodies were employed to detect high levels of calcyclin expression in acute lymphoblastic, myelogenous, and monocytic leukemia cell lines, but not in myeloma or lymphoblastoid cells. In comparison, most cells were nearly devoid of an 0-acetylsialic acid-specific protein that is more abundant 0 1991 Academic in various tissue types than calcyclin. Press,
Inc.
1 To whom correspondence
should be addressed.
0003.9861/91$3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
The search for genes that undergo growth-related changes of expression and their subsequent analysis is a rational approach to obtain information on growth regulation. It has led to the establishment of a series of cDNA clones due to the marked increases of levels of certain types of mRNA concomitant to growth stimulation and cell cycle progression of quiescent fibroblasts by serum (1, 2). Sequence comparison of a cDNA clone, termed 2A9, revealed a notable extent of homology to Ca’+-binding proteins like S-100, prompting the putative protein product to be referred to as calcyclin (3). Its sequence was later found to be identical to a prolactin receptor-associated protein (4). Evidently, elucidation of the function of this gene product requires its purification. This has independently been achieved with the search for further Ca’+-binding proteins as well as for proteins with affinity to sialic acid residues (5,6). The latter property may enable calcyclin to selectively interact with sialic acids of carbohydrate chains of cellular glycoconjugates, contributing to the determination of its position in the cell. It was thus of interest to assess calcyclin’s carbohydratebinding specificity. Similarly, the availability of specific antibodies allowed the measurement of its distribution in human tissues and in leukemic cells, where overexpression has been disclosed by measuring the amounts of 2A9specific mRNA in blot hybridization assays (7,8). EXPERIMENTAL
PROCEDURES
Purifiication of caky&z. Fetuin (Biomol, Hamburg, FRG) was coupled to divinylsulfone-activated Sepharose 4B with a coupling yield of 25 mg/ml column resin. Processing of 10 g of bovine heart started routinely with homogenization in 50 ml 20 mM Tris-buffered saline (TBS), pH 7.8, containing 20 mM CaClz, 4 mM @-mercaptoethanol, and 0.01 mM benzylsulfonyl fluoride (buffer A). Following centrifugation at 20,OOOg for 45 min, the resulting supernatant was passed through a fetuin-Sepharose 4B column (20 X 1.5 cm) at a flow rate of 5 ml/h that subsequently was washed extensively with buffer A containing 1 M NaCl to remove contaminants. The fetuin-binding activity was eluted with buffer A, substituting 20 mM CaCl, with 4 mM EDTA. The eluate was
* Abbreviations fered saline.
used: TBS, Tris-buffered
saline; PBS, phosphate-buf137
138
ZENG
AND
concentrated by ult~~ltration using a membranous YM-5 filter, dialyzed against buffer A in tubing impermeable to molecules with a molecular weight above 6 kDa and fractionated on a Sephadex G-75 column (160 X 0.5 cm) at a flow rate of 2.5 ml/h. Alternatively, the supernatant after homogenization and centrifugation was fractionated by (NH&SOI precipitation. The precipitate from 50-100% saturation, wherein the pH had been adjusted to 4.0 with acetic acid, was dissolved in 10 mM NHHCO%, pH 8.0, and dialyzed against this solution. Gel filtration on a Sephadex G-75 column (100 X 2.0 cm) at a flow rate of 20 ml/h, concentration of activity-containing fractions by ultrafiltration, dialysis against buffer A, and chromatography on a fetuin-Sepharose 4B column (20 X 1.2 cm) followed. Calcyclin was also obtained, when the precipitated protein was dissolved in 10 mM Trie-buffered saline, pH 8.3, containing 2 mM EDTA and 2 mM @-mercaptoethanol. The clear solution was applied to a DEAE-Sephadex A-50 column (25 X 2.0 cm) and protein was eluted by application of a linear gradient of NaCl up to 0.2 M salt. Activity-containing fractions were pooled, concentrated, dialyzed against 10 mM NH,HCOe, pH 8.3, and further purified by gel filtration on a Sephadex G-75 column (100 X 2.0 cm). Analytical procedures. Protein was determined by the dye-binding assay, adapted for microtiter plates, with bovine serum albumin as standard (9). One- and two-~mensional gei el~trophoretic analyses under reductive and denaturing conditions with a 12.5 or 15% running gel and a 3% stacking gel combined with isoelectric focusing by nonequilibrium pH gel electrophoresis and a highly sensitive silver staining procedure were carried out, as described (10). The isoelectric point (pl) of calcyclm was calculated on the basis of the deduced amino acid sequence of the cDNA clone (3). The calculation was performed as described (11). Analytical gel filtration was performed on a Sephadex G-50 column (100 X 0.5 cm) equilibrated with 20 mM Na~HPO~/KH~PO~ buffer, pH 7.2. The applied protein was eluted at a flow rate of 2.5 ml/h. Solid-phase activity assay. Fetuin was biotinylated using biotinylN-hydroxysuccinimide ester, as described (12). Spotting detection of activity was based on the binding of biotinylated fetuin to calcyclin, immobilized on nitroceilulose squares (1 X 1 cm; 0.2 gm from Schleicher & Schuell, Dassel, FRG), and the subsequent, highly sensitive detection of the Iabel biotin, as analogously performed for other carbohydratebinding proteins (10,13). In detail, following spotting of 1~1 of calcyclincontaining solution (1 pg/pl), the nitrocellulose squares were allowed to dry for 30 min. The squares, placed in separate wells of tissue culture plates, were incubated for 1 h at room temperature with a 3% (w/v) solution of bovine sernm albumin in 20 mM Tris-HCl buffer, pH 7.5, containing 0.9% NaCi to block residual protein-binding sites. After thorough washing with this buffer containing 0.2% bovine serum albumin and 2 mM CaClx, incubation with 20 pg/mI biotinylated fetuin in the absence or presence of inhibitors was carried out for 2 h at room temperature. To determine the dependence of binding on ionic strength and divalent cations, the buffer was supplemented by increasing concentrations of NaCl up to 4 M or by 10 mM of the respective cation. To assess the influence of pH value on binding, various buffer systems (pH 3-6: 50 mM sodium acetate, pH 6.5-9.0: 50 mM Tris-HCI, pH 9-12.5: 50 mM glycine/NaOH) were used in this step. Similarly, the inhibitory potency of a panel of commercially available glycosubstances as well as neoglycoproteins, prepared by diazo coupling of p-aminophenyl glycosides to carbohydrate-free bovine serum albumin as carrier with an incorporation level of 10 + 2 carbohydrate moieties per carrier molecule (12), was measured. Unbound ligand was carefully removed by repetitively incubating the nitrocellulose squares with fresh buffer solution (20 mM Tris-HCl, pH 7.5, containing 0.9% NaCI, 0.2% bovine serum albumin, and 2 mM CaCI,) for 5 min. For quantitative evaluation of bound biotinylated fetuin as a measure of calcyclin presence, a solution of the avidin-peroxidase conjugate (10 pg/ml; Sigma, Munich, FRG) was applied for 90 min at room temperature. Extensive washing followed prior to the visualization of bound probe with 4-~hloro-l-naphthol/H~O~ solution as substrate and quantitation, as performed recently (10,13). The activity is expressed as titer-‘. The titer is the highest dilution of calcyclin-containing solution that yielded a definitely positive reaction in the assay.
GABIUS V~~izat~n of fetu~n-bi~i~ after e~ct~b~tting of ca~cl~n. Transfer of purified calcyclin (10 pg) after separation on a 15% running gel under denaturing and reductive conditions was carried out with the semi-dry blot system Sartoblott II (Sartorius, GBttingen, FRG). The two horizontal graphite plates enclosed a sandwich comprising (from the anodic plate): two layers of Whatman No. 1 filter paper soaked in 0.3 M Tris-HCl/BO% (v/v) methanol (pH 10.4), 1 layer of filter paper soaked in 25 mM his-HCl/20% (v/v) methanol (pH lo.@, the nitrocellulose membrane as well as the polyacrylamide gel, both soaked in 25 mM Tris-HCl/20% (v/v) methanol (pH 10.4), and three layers of filter paper soaked in freshly prepared 40 mM 6-amino-n-hexanoic acid/ 25 mM Tris-HCl/20% (v/v) methanol (pH 9.4). Transfer was performed at room temperature for 30 min at 400 mA. Following transfer, the membrane was washed twice for 5 min in buffer (20 mM Tris-HCl, pH 7.8, containing 0.2% bovine serum albumin and 0.9% NaCl) and the remaining binding sites on the nitroceliulose were blocked by incubation in 50 mM Tris-HCI (pH 7.8) containing 5% bovine serum albumin and 0.9% NaCl for 30 min at room temperature. After extensive washing the blots were incubated with 20 pg/ml biotinylated fetuin in TBS contaming 20 mM CaCl*, 0.9% NaCl, and 0.5% bovine serum albumin in the absence or presence of inhibitors (fetuin, asialofetuin in 500-fold excess). Visualization of bound probe was achieved by applying the same protocol, as described for the solid-phase assay. Chemical modification of amino acid residues. In each reaction 7.2 nmol calcyclin was chemically modified by adding reagents that exhibit specificity to certain types of functional groups under conditions given in detail elsewhere (13, 14). Namely, 0-methylisourea, citraconic anhydride, cyclohexane-1,2dione, phenytglyoxai, N-acetyhmidazole, diethy~pyr~arbonate, and glycine methyl es~r/l-ethyl-3-{3-~methylaminopropyl]-carb~iimide were employed. In all cases excess reagents were removed by dialysis against 20 mu TBS, pH 7.8, prior to performing the solid-phase assay with biotinylated probe. To exclude any further influence of the conditions, under which modification is carried out, on the activity, respective controls were performed under identical conditions in the absence of modifying reagent. ~iotiny~~n of caky&n. Purified calcyclin (1.0 mgf was biotinyla~ using biotinamidocaproyl hydrazide (34.2 mg; Sigma, Munich, FRG) in the presence of 19.2 mg of the coupling reagent I-ethyl-3-(3-dimethylaminopropyl)-carbodiimide under mild conditions, optimized for the @nerve growth factor (15). The solid-phase assay was performed under identical conditions, as given in detail above, with 20 rg/ml biotinylated calcyciin as probe. Subcel~u~r fract~o~tion and correlation of fetu~n-bided activity to the different fractions. All individual steps were performed at 4”C, as given in detail elsewhere (14). Fresh bovine heart was homogenized in 5 vol 0.3 M sucrose, 4 mM P-mercaptoethanol and 3 mM MgCl,. The extract was centrifuged at 12,000g for 10 min, resulting in the cytoplasmic supernatant fraction. The pellet was resuspended in 2 M sucrose containing 4 mM P-mercaptoethanol and 1 mM MgC!l,. The top 3 ml of each sample after ~entri~gation in a swinging bucket rotor at 40,OOOgfor 1 h was referred to as the cytoplasmic particulate fraction. The pellet was resuspended in a Potter-Elvehjem homogenizer with 25 mM Tris-HCl buffer, pH 7.8, containing 200 mM NaCI, 0.1 mM benzylsulfonyl fluoride, 1% Triton X-100, and 1 mM dithiothreitol. After centrifugation at 12,000g a DNA-rich pellet and the nuclear supernatant were obtained. Assessment of binding activity by solid-phase assays was based on specimens from each fraction, diluted to a similar concentration with TBS and promptly monitored for activity. Preparation of the antibody. To raise specific polyclonal antibodies in rabbits, matrix-immobilized antigen was used, as described for an endogenous lectin (16). Calcyclin was transferred after electrophoretic separation under reductive and denaturing conditions from polyacrylamide gels onto nitrocellulose membranes and stained with red Ponceau solution. Nitrocellulose strips with immobilized calcyclin were destained in water, minced with a mortar and pestle under liquid nitrogen, and the resulting nitrocellulose powder containing approximately 100 fig calcyclin was injected intradermally in a rabbit in 1 ml phosphate-buffered saline. The IgG-fraction after at least three injections was purified
BINDING
OF CARBOHYDRATES TABLE
139
TO CALCYCLIN
I
Purification of Calcyclin by Three Different Methods Total protein (md
Step 1
Specific activity”
400
~ Crude extract
5
2000
100
Purification (-fold) 1
Fetuin-Sepharose Sephadex G-75
2b/c 3b 4b
(NH,)zSO,-precipitation Sephadex G-75 Fetuin-Sepharose
75 2.8 0.08
10 150 300
750 420 24
37.5 21 1.2
2 30 60
3c 4c
DEAE-Sephadex Sephadex G-75
10 0.98
50 300
500 294
25 14.7
10 60
A-50
24.0 (120) 22.8 (114)
Recovery (X)
2a 3a
0.12 (0.60)’ 0.076 (0.38)
200 300
Total activity”
’ Specific activity is expressed as titer/mg protein/ml. b Total activity is expressed as specific activity X mg (titer X ml). ’ Values in parentheses denote results that are obtained with elution of the column with 0.1 M NH,OH
by protein A-Sepharose chromatography (Pharmacia, Freiburg, FRG). Further purification was facilitated by affinity chromatography on calcyclin-Sepharose 4B, obtained by coupling of calcyclm to divinylsulfoneactivated Sepharose with a yield of 760 pg calcyclin/ml. Elution was pH 2.3, and 100 mM subsequently performed with 100 mM glycine-HCl, triethylamine, pH 11.5. The eluants were quickly neutralized and extensively dialyzed against phosphate-buffered saline, pH 7.2. Western blots of S-100 A (01,p) from Sigma (Munich, FRG), of pll, kindly provided by Dr. V. Gerke (Goettingen, FRlG), and of calcyclin or tissue extract were probed with affinity-purified antibody (10 pg/ml) at room temperature overnight. The following individual steps were identical to the visualization of binding of biotinylated fetuin, given above. However, GAR/IgG (H + L)-peroxidase (5 pg/ml; Nordic, Bochum, FRG) substituted avidin-peroxidase. Quuntituttin of cc&&n in tissue and cell extracts. The various human tissues, obtained at autopsy, were homogenized in 20 mM TBS, pH 7.5, containing 4 mM EDTA. Cell pellets of various human tumor lines, listed in the footnotes to Table VII, were similarly processed. After centrifugation at 20,OOOgfor 45 mi:n each supernatant was used to inhibit antibody binding to immobilized calcyclin. Aliquots of 100 ~1 containing 2 pg/ml purified calcyclin in 0.2 M Na&Os, pH 9.2, had been used for coating polystyrene microtiter wells at 37°C for 2 h. Wells were washed saline, pH 7.2, containing four times with 10 mM phosphate-buffered
1.2 (6.0) 1.14 (5.7)
40 60
(pH 11.0) instead of 4 mM EDTA.
0.5% Tween 20 (PBS-Tween). The remaining free protein-binding sites were then blocked with 150 pl of PBS containing 5% bovine serum albumin at 37°C for 1 h followed by further washes. Affinity-purified antibody (100 pl with a concentration of 1.0 pg/ml) in PBS + 5% BSA was added. After 2 h at 37°C the wells were washed six times with PBSTween and then incubated with GAR/IgG (H + L)-peroxidase conjugate (5 pg/ml in PBS + 5% bovine serum albumin) at 37°C for 1 h. Wells were again washed six times with 150 ~1 PBS-Tween. Following the last wash, a 150-~1 aliquot of 0-phenylenediamine hydrochloride/H202 solution (0.4 mg OPD and l).d30% HzOz per milliliter of 75 mM sodium citrate buffer, pH 5.5) was added to each well. The reaction was allowed to proceed for 10 min in the dark and was stopped by the addition of 50 pl of 4 N sulfuric acid, and the absorbance at 490 nm was determined using a microplate reader equipped with a narrow bandpass filter for 490 nm. The extent of antibody binding was reduced in correlation to the tissue content of the antigen, when extracts, diluted in PBS containing 5% bovine serum albumin, were preincubated with the antibody for 1 h at 37°C. A standard curve for quantitative assessment of calcyclin was obtained by preincubation of antibody-containing solution with known amounts of purified antigen. Besides calcyclin, the tissue distribution of a human 0-acetylsialic acid-specific protein from human placenta, purified as described (13), was similarly determined.
RESULTS
V’
2
3
4
Molecular
4
6;ibio Weight
(x10-4
)
FIG. 1. Determination of molecular weight of calcyclin by analytical gel filtration on a Sephadex-G50 column (160 X 0.5 cm) that has been calibrated with standards.
Purification of calcyclin from bovine heart. Calcyclin was purified from bovine heart to apparent homogeneity by three different methods that yielded identical specific activities for binding of biotinylated fetuin (Table I). This assay was chosen as an indicator of carbohydrate binding. The combination of ion-exchange chromatography, where the protein eluted around 0.11 M NaCl in the gradient, and gel filtration proved most effective in terms of recovery. Analytical gel filtration with purified calcyclin revealed a molecular weight of approximately 11,000 (Fig. 1). The experimentally determined isoelectric point of the purified protein is identical within the limits of detection to the theoretically calculated pI value of 4.3 on the basis of the cDNA sequence, arguing against major chargeshifting post-translational modifications (Fig. 2). The activity of binding was nearly unaffected by pH changes within the range from 5 to 8. However, sharp decreases
140
ZENG
AND
GABIUS
PH
FIG. 2. Visualization of calcyclin by silver staining after two-dimensional gel electrophoresis involving isoelectric focusing in the first dimension and sodium dodecyl sulfate-polyacrylamide (15%) gel electrophoresis under reductive conditions in the second dimension. Positions of standards for molecular weight designation are indicated by arrowheads.
occurred outside of this range (Fig. 3). With respect to alterations of the ionic strength, the activity was rather stable, when NaCl molarities up to 1.5 M NaCl were present. Further increases led to diminution of binding of fetuin. Among the cations that had been tested at a concentration of 10 mM, Ca2+, Mg2+, Co2+ Ni2+, and Mn2+ enhanced binding of fetuin. The presence of Cd2+, Zn2+, and Cu2+ allowed only partial binding. A decrease of binding activity was noted below a concentration of 1 mM for Ca2+ and Mg2’. Binding of labeled (neojglycoproteins to calcyclin. Biotinylated fetuin had been used as ligand to assess the binding activity. Labeled asialofetuin failed to bind to calcyclin, emphasizing the importance of sialic acid moieties for binding. This result was independently obtained in solid-phase assays with biotinylated probes as well as in inhibition experiments on blots in the presence of fetuin and asialofetuin as inhibitors of binding of labeled fetuin (Fig. 4, Table II). To analyze the selectivity of calcyclin for carbohydrates more closely, several neoglycoproteins, carrying the same density of carbohydrate residues with the same spacer, were employed. Notably, the carrier protein exhibited no ligand properties (Table II). Immobilized
a
0
1
2
3
15
6
7
8
9
10
11
12
13
pH - value
FIG. 3. Determination ylated fetuin to calcyclin residues are indicated.
of pH-dependence of the binding of biotinin a solid-phase assay. pK, values of certain
C
negatively charged carbohydrate residues caused binding of the labeled carrier to calcyclin. Similarly, carrier-conjugated N-acetylglucosamine, mannose, and lactose, but not the a-galactoside melibiose, N-acetylgalactosamine, or fucose, led to a signal (Table II). The sialic acid a-2,6galNAc-a-Thr/Ser-rich bovine submaxillary mucin exhibited comparatively strong ligand properties.
TABLE
II
Binding Activity of Calcyclin to Biotinylated (Neo)glycoproteins in a Solid-Phase Assay
(Neo)glycoprotein
00
b
FIG. 4. Visualization of fetuin-binding activity of immobilized purified calcyclin (10 rg) after electrophoresis under reductive and denaturing conditions and transfer to nitrocellulose with biotinylated fetuin in the absence of further substances (a) as well as in the presence of asialofetuin (b) and fetuin (cl. Standards for molecular weight designation are indicated by arrowheads.
Fetuin Asialofetuin BSM” BSAb Sialic acid’ Glucuronic acid’ Mannose-&phosphate’ Galactose-6-phosphate’ @-glcNAc’ &galNAc” cu-Fucose’ Lactose’ Melibiose’ Mannose”
Binding activity (%) 100 0 90 0 20 60 60
50 40
0 0 40
0 40
’ Bovine submaxillary mucin. * Bovine serum albumin. ’ The carbohydrate part is given to define the neoglycoprotein.
BINDING
OF CARBOHYDRATES
TABLE
TABILE III Inhibition of Fetuin-Bind.ing of Calcyclin and Glycosubstances ,in a Solid-Phase
Carbohydrates and glycosubstances Neu5Ac Neu5Gc N-Acetylneuraminyllactose Glucuronic acid Mannose-6-phosphate Galactose-6-phosphate glcNAc Lactose Mannose GM1 Asialo-GM1 BSM” Sialic acid-BSA Mannose-g-phosphate BSA Galactose-6-phosphate BSA glcNAc-BSA lac-BSA man-BSA Heparin Fucoidan Dextran sulfate
141
TO CALCYCLIN
by Sugars Assay
Concentration required for 50% inhibition (mM) >lOO 3 25 50 (20% inhibition) 25 30 100 50 50 (20% inhibition) 3.0 5.0 (no inhibition) 0.1 0.15 0.12 0.12 0.15 (20% inhibition) 0.15 (25% inhibition) 0.12 (20% inhibition) 5 mg/ml (20% inhibition) 2 mg/ml 1.5 mg/ml
’ Denotes concentration in terms of sialic acid; 50 mM galNAc, 50 mM melibiose, and 50 mM fucose were ineffective.
The tested panel of sugar residues was enlarged, when the inhibition of binding of biotinylated fetuin in the presence of glycosubstances was monitored, as summarized in Table III. N-Glycolylneuraminic acid was more efficient as inhibitor than the acetylated derivative. However, the extension of its carbohydrate structure to the a-2,3/a-2,6-containing N-acetylneuraminyl-lactose improved the extent of inhibition considerably. Removal of the a-2,3-linked N-acetylneuraminic acid residue from the ganglioside GM1 markedly reduced its capacity as an inhibitor. The binding activity is yet not strictly dependent on the presence of sialic acids. Neoglycoproteins with carboxylated or phosphorylated sugars were rather similarly effective as inhibitors in this assay. As already noted with labeled neoglycoproteins, lactose, N-acetylglucosamine, and mannose, too, have an impact on binding of fetuin. Fetuin binding was also impaired by the presence of sulfated polysaccharides. Dextran sulfate and fucoidan reduced it more effectively than heparin. Chemical modification of amino acid residues withgroupspecific reagents. To infer the nature of molecular interactions between calcyclin and its carbohydrate ligands, we analyzed the effect of modification of certain amino acid residues by group-specific reagents. As summarized in Table IV, impairment of fetuin binding mainly occurred after modification of the functional groups of lysine and arginine. No dependence for binding activity on the in-
IV
Effect of Amino Acid Modification by Group-Specific Reagents
on Binding
of Biotinylated
Residues modified
Chemical treatment Native protein 0-Methylisourea Citraconic anhydride
Fetuin Binding activity (%) 100 40 40
Lysine Lysine, Hz-terminal amino group Arginine Arginine Tyrosine Histidine Carboxyl
Cyclohexane-l$dione Phenylglyoxal N-Acetylimidazole Diethylpyrocarbonate Ester-carbodiimide
40 60 100 100 100
tegrity of tyrosine, histidine, or carboxyl groups could be disclosed. Since binding activity was retained after carboxy1 esterification with glycine methyl ester, these groups can thus be biotinylated to lead to labeled calcyclin that should still exhibit activity for fetuin binding. This tool can be employed to answer the question whether immobilization to nitrocellulose that so far has commonly been employed for the determination of binding activity influences its carbohydrate specificity. Binding of labeled calcyclin to (neo)glycoproteins. Using immobilized (neo)glycoprotein binding of the biotinylated calcyclin from solution to the matrix was assessed. No significant changes in the ligand properties of the various (neo)glycoproteins could be discerned, as shown in Table V. In addition to exclusion of an effect of immobilization on binding activity the access to biotinylated calcyclin will eventually be of value to localize ligands in situ. It is likewise noteworthy that no information on the intracellular localization of calcyclin itself
TABLE Binding Activity (Neo)glycoproteins
(Neo)glycoprotein Fetuin Asialofetuin BSM Sialic acid” Glucuronic acid” Mannose-6-phosphate” Galactose-B-phosphate” @-glcNAc” @-galNAc” lac ’ man ’ ’ The carbohydrate
V
of Biotinylated Calcyclin to in a Solid-Phase Assay Binding activity (%) 100 0 100 40 50 70 70 50 10 25 20
part is given to define the neoglycoprotein.
142
ZENG TABLE Distribution
Fraction Whole cell extract Cytoplasmic supernatant Cytoplasmic particulate fraction Nuclear fraction
AND
GABIUS
VI
of Fetuin-Binding Activity in Subcellular Fractions of Bovine Heart
Volume (ml)
Total protein bd
40
400
5
2000
38.5
350
4
1400
17.5
20
2
50
350 100
Specific activity”
Total activity* a
15 10
a Specific activity is expressed as titer/mg/ml. * Total activity is expressed as specific activity
X mg.
is available. Thus, subcellular fractionation combined with activity assays was employed to address this issue. Subcellular localization of fetuin-binding activity. Three subcellular fractions, referred to as the nuclear fraction, to which the DNA was confined, a cytoplasmic particulate fraction, and a cytoplasmic supernatant fraction, were established. Activity was found to be distributed among each of the three parts, the specific activity being highest in the nuclear fraction (Table VI). Further studies on the precise localization of calcyclin require the availability of specific antibody. Consequently, we sought to provide this tool to check its specificity and to use it for the determination of expression of calcyclin in tissues and cells to enable focusing of immunohistochemical studies. Immunological determination of calcyclin in tissue and in cell extracts. Matrix-bound calcyclin was applied as antigen to raise specific polyclonal antibodies. The serum was processed by chromatography on protein A-Sepharose as well as on calcyclin-Sepharose. The level of selectivity of the affinity-purified antibodies was ascertained by immunoblotting with structurally related Ca2+-binding proteins and the tissue extract (Fig. 5). Their availability prompted the analysis of expression of calcyclin in tissues and cells. Due to the reported overexpression of calcyclin mRNA in human leukemic cells (7,8) we chose to include extracts of various cell lines into this series of measurements. A specific antibody to a human 0-acetylsialic acidbinding protein, purified recently (13), was included to determine whether these two proteins with affinity to sialic acids may be expressed at similar levels. A quality control for this antibody fraction is shown in Fig. 6. Tissue and cell extracts were employed to reduce binding of the antibody to the two purified proteins according to their antigen content. Quantitation was based on the respective calibration curves, established with the purified proteins (Fig. 7). Calcyclin was measurable in several organs, the levels being lower than for the lectin (Table
b
c
d
FIG. 5. Analysis of reactivity of calcyclin-specific antibodies to 20 pg purified S-100A ((u, 8) (a), 20 pg purified pll (b), 2 pg purified calcyclin (c), and extract of bovine heart (d) after gel electrophoretic separation under reductive and denaturing conditions and electrophoretic transfer to nitrocellulose. Positions of standards for molecular weight determination are indicated by arrowheads.
VII). Notably, it is present in acute lymphoblastic, myelogenous, and monocytic leukemia cell lines in comparatively high abundance relative to myeloma or lymphoblastoid cells. This result corroborates measurements on the level of the mRNA, warranting further immunohistochemical studies on these types of cell. The protein, however, is nearly undetectable in the three types of leukemia lines with high calcyclin levels. Expression of these two sialic acid-binding proteins is apparently not coregulated in leukemic cells. DISCUSSION
cDNA cloning of growth-regulated mRNAs from quiescent fibroblasts that had been stimulated by serum or certain mitogens disclosed preferential expression of certain gene products (1, 2). One putative protein was designated calcyclin on the basis of its cell cycle-dependent expression and its sequence homologies to charac-
FIG. 6. Analysis of reactivity of antibodies against the human Oacetylsialic acid-specific protein to extract proteins from human kidney after gel electrophoretic separation under reductive and denaturing conditions and electrophoretic transfer to nitrocellulose. Positions of standards for molecular weight designation are indicated by arrowheads.
BINDING
OF CARBOHYDRATES
143
TO CALCYCLIN
2A9 (ns)
53KDa
Lectin
(ng)
FIG. 7. Calibration curves with purified calcyclin (a) and 0-acetylsialic acid-specific protein (b), termed 53-kDa protein according to its molecular weight, for determination of the plresence of the respective proteins in tissue extracts by ELISA.
terized Ca2+-binding proteins (3). Calcyclin was later referred to as a prolactin receptor-associated protein, present in certain, but not a.11,prolactin receptor-positive human breast cancer cell lines (4). A similar heterogeneity of calcyclin expression for a distinct category of cells had been reported in the case of specimens from patients with acute myelogenous leukemia (7,8). Calcyclin is also present in a number of mouse and rat tissues, as indicated by Northern and Western blots (4, 17). When native calcyclin is employed as inhibitor for binding of specific antibody to immobilized extract proteins, the presence of calcyclin in various human tissue types is ascertained. Subcellular fractionation of heart extract indicates that its activity is not confined to one compartment. Notably, the monitoring of extracts detects high levels of calcyclin expression for acute myelogenous and monocytic leukemia cell lines and a pronounced reduction for lymphoblastoid or myeloma cells. To shed light on calcyclin’s physiological role, the two characteristics that have guided the purification certainly deserve attention. The binding of Ca2+ and of sialic acids may even be connected to each other. Ca2+, but not Zn2+ or Cd’+, supports optimal binding to carbohydrates. These two cations are incapable of competing with Ca2+ for its binding sites, whose occupation induces a conformational change (18). It is certainly :remarkable to find at least two properties, namely cation and carbohydrate binding, in a protein of a molecular weight of approximately 10 kDa. Coincidentally, sialic acid-binding lectins from frog eggs exhibit a similarly small molecular weight (19, 20). Calcyclin’s ability to bind the sialoglycoprotein fetuin strictly requires the presence of the sialic acid moieties. Its apparent affinity to other carboxylated or to phosphorylated sugars as well as to sulfated polysaccharides is reminiscent of binding properties of a number of proteins that are known to interact with neuraminic acids, like an agglutinin from rat uterus (21), the interferon antagonist sarcolectin (22), a lectin from kidney (23) and
from brain synaptic vesicle-enriched fractions (24), the murine lymphocyte homing receptor MEL-14 (15), and the platelet granular-membrane protein GMP-140 (26, 27). A lectin from stratum corneum even exhibits a markTABLE
VII
Distribution of Calcyclin and the 0-Acetylsialic Acid-Specific Protein (53 kDa) in Human Tissues and Leukemic Cell Lines
Tissues or cells Heart Tongue Small intestine Stomach Lungs Liver Kidney cortex Aorta Brain Spleen Lymph node Adrenal gland Thyroid gland Pancreas Testes KG-l” KG-lab K-562’ THP-ld RPMI-8226’ CCRF-CEM’ Croco BP Hs-294-Th
Calcyclin hdmd
53 kDa (w/md
375 300 290 200 300 145 110 1110 280 150 160 280 160
