Characterization of the Heparin Binding Properties of Annexin II Tetramer*

In this report, we have characterized the interaction of heparin with the Ca2+- and phospholipid-binding protein annexin II tetramer (AIIt). Analysis of the circular dichroism spectra demonstrated that the Ca2+-dependent binding of AIIt to heparin caused a large decrease in the α-helical content of AIIt from ∼44 to 31%, a small decrease in the β-sheet content from ∼27 to 24%, and an increase in the unordered structure from 20 to 29%. The binding of heparin also decreased the Ca2+ concentration required for a half-maximal conformational change in AIIt from 360 to 84 μm. AIIt bound to heparin with an apparentK d of 32 ± 6 nm (mean ± S.D., n = 3) and a stoichiometry of 11 ± 0.9 mol of AIIt/mol of heparin (mean ± S.D., n = 3). The binding of heparin to AIIt was specific as other sulfated polysaccharides did not elicit a conformational change in AIIt. A region of the p36 subunit of AIIt (Phe306–Ser313) was found to contain a Cardin-Weintraub consensus sequence for glycosaminoglycan recognition. A peptide to this region underwent a conformational change upon heparin binding. Other annexins contained the Cardin-Weintraub consensus sequence, but did not undergo a substantial conformational change upon heparin binding.

In this report, we have characterized the interaction of heparin with the Ca 2؉ -and phospholipid-binding protein annexin II tetramer (AIIt). Analysis of the circular dichroism spectra demonstrated that the Ca 2؉ -dependent binding of AIIt to heparin caused a large decrease in the ␣-helical content of AIIt from ϳ44 to 31%, a small decrease in the ␤-sheet content from ϳ27 to 24%, and an increase in the unordered structure from 20 to 29%. The binding of heparin also decreased the Ca 2؉ concentration required for a half-maximal conformational change in AIIt from 360 to 84 M. AIIt bound to heparin with an apparent K d of 32 ؎ 6 nM (mean ؎ S.D., n ‫؍‬ 3) and a stoichiometry of 11 ؎ 0.9 mol of AIIt/mol of heparin (mean ؎ S.D., n ‫؍‬ 3). The binding of heparin to AIIt was specific as other sulfated polysaccharides did not elicit a conformational change in AIIt. A region of the p36 subunit of AIIt (Phe 306 -Ser 313 ) was found to contain a Cardin-Weintraub consensus sequence for glycosaminoglycan recognition. A peptide to this region underwent a conformational change upon heparin binding. Other annexins contained the Cardin-Weintraub consensus sequence, but did not undergo a substantial conformational change upon heparin binding.
The annexins are a family of ϳ13 proteins that bind to acidic phospholipids and biological membranes in a Ca 2ϩ -dependent manner (see Refs. 1-3 for reviews). These proteins are expressed in a wide range of organisms such as slime molds, higher plants, invertebrates, and vertebrates. Studies of the amino acid sequence of the annexins have established the homology of these proteins. All annexins contain four repeats (eight repeats in the case of annexin VI) of ϳ70 amino acids that are highly homologous. In contrast, the N terminus of each of the annexins is unique and displays the greatest variation in sequence and length. The crystal structure of several of the annexins has been reported (4 -6) and has established that the annexins are composed of two distinct sides. The convex side faces the biological membrane and contains the Ca 2ϩ -and phospholipid-binding sites The concave side faces the cytosol and contains the N and C termini.
Annexin II (p36) contains three distinct functional regions, the N-terminal region, the C-terminal region, and the core region. The core region of p36 contains the Ca 2ϩ -and phospho-lipid-binding sites, whereas the C-terminal region contains the 14-3-3 homology domain (7) and the plasminogen-binding domain (8). The N terminus of annexin II (p36) contains two important regulatory domains, the L and P domains. The L domain consists of the first 14 residues of the N terminus and contains a high affinity binding site for the p11 protein (reviewed in Ref. 9). The P domain of p36 contains the phosphorylation sites for protein kinase C (Ser 25 ) and pp60 src (Tyr 23 ). The N-terminal L and P domains play regulatory roles; activation of the phosphorylation sites of annexin II tetramer results in an increase in the A 0.5 (Ca 2ϩ ) for chromaffin granule aggregation and F-actin binding, whereas binding of the p11 subunit decreases the A 0.5 (Ca 2ϩ ) for these activities. The heterotetrameric complex (p36 2 ⅐p11 2 ) formed by the binding of p11 to p36, referred to as annexin II tetramer (AIIt), 1

is the predominant form in most cells (reviewed in Ref. 9).
AIIt has been shown to be present at both the cytosolic and extracellular surfaces of the plasma membrane of many cells (9). Extracellular AIIt has been proposed to function as a cell adhesion factor (10, 11), a receptor for plasminogen and tissue plasminogen activator (8,12), and a receptor for tenascin-C (13,14).
In a previous study (15), we reported that AIIt bound to a heparin affinity column and that the phosphorylation of AIIt on tyrosine residues blocked the heparin-binding activity of the protein. In this report, we have characterized the interaction of AIIt with heparin. Our results identify AIIt as a specific, high affinity heparin-binding protein. Furthermore, we show that the Ca 2ϩ -dependent binding of heparin to AIIt causes a dramatic conformational change in the protein. Last, we show that the p36 subunit of AIIt contains a Cardin-Weintraub glycosaminoglycan recognition site (16) and that a peptide to this region of AIIt binds heparin.

EXPERIMENTAL PROCEDURES
Materials-Annexin II tetramer and annexin II monomer were prepared from bovine lung as described (17) and stored at Ϫ70°C in 40 mM Tris-HCl, pH 7.5, 1.0 mM DTT, 0.1 mM EGTA, and 150 mM NaCl. Carbohydrates were obtained from Sigma and were the purest grade available. Heparin (bovine lung) was obtained from Calbiochem and had an average molecular mass of 17 kDa and an activity of 149 USP units/mg. Porcine intestinal mucosa heparin (6 kDa, 70 USP units/mg), bovine intestinal mucosa heparin (3 kDa, 50 USP units/mg), and bovine kidney heparan sulfate (7.5 kDa) were obtained from Sigma.
Measurement of Protein Secondary Structure by Circular Dichroism-CD measurements were performed with a Jasco J-715 spectropolarimeter. The spectropolarimeter was calibrated with an aqueous solution of recrystallized ammonium camphor sulfonate-d 10  1.0 mM CaCl 2 ) or buffer B (10 mM Tris-HCl, pH 7.5, 0.1 mM DTT, 150 mM NaCl, and 0.5 mM EGTA) in the presence or absence of ligand (0.5-50 M) for 20 min at room temperature. Samples (0.1 ml) were scanned in a quartz cuvette (0.5-mm path length) from 178 to 260 nm at a rate of 10 nm/s, using a bandwidth of 1 nm and a response time of 4 s. CD spectra of proteins were obtained by averaging four wavelength scans and were corrected by subtracting buffer scans or, where appropriate, scans of ligand in buffer. The measurements at 222 nm were obtained using the time scan mode of the software, and each point was the average of a minimum of 20 ϫ 1-s measurements (at a bandwidth of 1 nm). Results are expressed as ⌬⑀ (M Ϫ1 cm Ϫ1 ).
Analysis of CD Spectra-CD spectra of AIIt were analyzed by the self-consistent method of Sreerama and Woody (18). The analysis was performed with the computer program SELCON, which was a generous gift from Dr. Robert W. Woody (Department of Biochemistry and Molecular Biology, Colorado State University, Fort Collins, CO). Typically, scans from six individual experiments were analyzed, and these values were averaged. The root mean square difference between the predicted CD data and the actual CD data was always Ͻ0. 15.
AIIt Affinity Chromatography-Approximately 1.4 g of CNBr-activated Sepharose 4B was washed with 400 ml of 1 mM HCl, followed by extensive washing in coupling buffer (0.1 M NaHCO 3 , pH 8.3, and 0.5 M NaCl). AIIt (10 mg) was extensively dialyzed against coupling buffer, mixed with 5.0 ml of washed CNBr-activated Sepharose 4B resin in a final volume of 20 ml, and incubated overnight at room temperature. The resin was then washed with coupling buffer; incubated with 1 M ethanolamine, pH 8.5, in a final volume of 50 ml for 2 h; and equilibrated with buffer C (20 mM Tris-HCl, pH 7.5, 1 mM DTT, and 1 mM CaCl 2 ) containing 25 mM NaCl.
Heparin was dissolved in buffer C containing 25 mM NaCl and applied to a 5-ml AIIt affinity column. The column was washed with buffer C containing 25 mM NaCl, followed by step elution with buffer C containing 150 and 500 mM NaCl. The heparin concentration of these fractions was determined by the uronic acid/carbazole reaction (19).
Heparin Binding Assays-Heparin binding experiments were performed in a total volume of 200 l in a reaction mixture containing 25 mM Hepes, pH 7.5, 1 mM DTT, 150 mM NaCl, 2.0 M AIIt, and variable concentrations of heparin. The reaction mixture was incubated at room temperature for 20 min and then centrifuged at 400,000 ϫ g for 30 min. The pellet was extracted with Laemmli SDS disruption buffer (20) and resolved by SDS-polyacrylamide gel electrophoresis. Protein bands were visualized with Coomassie Brilliant Blue R-250, and the p36 subunit protein band (36 kDa) was cut from each lane of the polyacrylamide gel, extracted overnight with pyridine (25%), and then quantified spectrophotometrically by measuring the absorbance at 605 nm as described (21).
Data Analysis-Nonlinear least-squares fitting was performed with the computer program SigmaPlot (Jandel Scientific). Titration data were analyzed as detailed by the SigmaPlot reference manual with the four-parameter logistic equation f ϭ (a Ϫ d)/(1 ϩ (x/c) b ) ϩ d, where a ϭ asymptotic maximum, b ϭ slope parameter, c ϭ value at inflection point (A 0.5 ), and d ϭ asymptotic minimum. The nonlinear least-squares curve fitting was then iterated by allowing the four fitting parameters to float while utilizing the Marquardt method for the minimization of the sum of the squared residuals. The K d and n values for the binding of heparin to AIIt were derived from plots of the fraction of AIIt bound versus the total heparin concentration ([HEP] T ) using the quadratic equilibrium binding equation (Equation 1), where ␣ ϭ fractional binding (AIIt pelleted/total AIIt), n ϭ mol of heparin/mol of AIIt, and K d ϭ dissociation constant. The nonlinear least-squares curve fitting was then iterated by allowing the K d and n parameters to float while utilizing the Marquardt method for the minimization of the sum of the squared residuals. For this analysis, it was assumed that there was only a single class of heparin-binding sites on AIIt.
Miscellaneous Techniques-AIIt concentration was determined spectrophotometrically using an extinction coefficient of A 280 nm ϭ 0.68 for 1 mg/ml AIIt. Ca 2ϩ /nitrilotriacetic acid buffers were prepared according to Ref. 22. The Ca 2ϩ concentration of CaCl 2 stock solutions was determined by atomic absorption. The final Ca 2ϩ concentrations of Ca 2ϩ / nitrilotriacetic acid buffers were verified by Ca 2ϩ electrode and FURA-2 measurement. Peptides were synthesized by the University of Calgary Peptide Facility and were purified by high pressure liquid chromatog-raphy. Phospholipid liposomes were prepared from mixtures of 50 l of 20 mg/ml phosphatidylserine, phosphatidylethanolamine, and cholesterol in chloroform, which were dried in an N 2 stream and hydrated with 1 ml of buffer containing 30 mM Hepes, pH 7.5, and 2 mM MgCl 2 . The mixture was sonicated (4 ϫ 15-s bursts at 75 watts with a Braun probe sonicator), and the phospholipid liposomes were stored at room temperature prior to their use.

CD Spectra of the AIIt-Heparin
Complex-Consistent with a previous report (23), we also found that the binding of Ca 2ϩ to AIIt caused a 10% reduction in intensity of the CD spectrum of AIIt (Fig. 1A). In contrast, the binding of heparin to AIIt in the presence of Ca 2ϩ resulted in a decrease in the intensity of the CD spectrum at 222 nm of ϳ40%. Furthermore, the CD spectrum of AIIt lost intensity and changed shape as the protein bound heparin in the presence of Ca 2ϩ (Fig. 1B). In the absence of Ca 2ϩ , heparin did not produce a dramatic change in the intensity and shape of the CD spectrum of AIIt. We also observed that in the absence of Ca 2ϩ , the CD spectrum of AIIt measured in the presence of heparin was slightly less intense (a decrease of ϳ4% at 222 nm) than the CD spectrum of AIIt measured in the absence of both heparin and Ca 2ϩ .
Quantitative assessment of the secondary structure components of the CD spectra of AIIt was made as described (18). The results are summarized in Table I. In the presence of Ca 2ϩ , the binding of heparin to AIIt resulted in a decrease in the ␣-helical content of AIIt from ϳ44 to 31%, a decrease in the ␤-sheet content from ϳ27 to 24%, an increase in the content of ␤-turns from ϳ12 to 14%, and an increase in the unordered structure from ϳ20 to 29%.
In the absence of Ca 2ϩ , heparin induced a more moderate change in the conformation of AIIt. Of interest was the heparin-induced increase in the ␤-sheet from ϳ21 to 27% and decrease in unordered structure from 22 to 18%.
These results establish that in the presence of Ca 2ϩ , heparin induces a large conformational change in AIIt, resulting in a substantial change in the conformation of the protein. However, heparin can also interact with AIIt in the absence of Ca 2ϩ and, to a much smaller degree, affect the conformation of the protein.
Dose Dependence of the Heparin-induced Conformational Change in AIIt- Fig. 2 presents the dose dependence of the heparin-induced conformational change in AIIt as determined by the changes in the CD spectrum at 222 nm. Increasing the heparin concentration decreased the intensity of the CD spectrum of AIIt until a maximum was reached at ϳ0.59 M heparin. At the optimal concentration of heparin, the CD intensity decreased by ϳ36%. The heparin concentration required for a half-maximal decrease in the CD intensity of 2.2 M AIIt was ϳ0.10 M. Increasing the heparin concentration in excess of 0.59 M resulted in an increase in the CD intensity.
Ca 2ϩ Dependence of the Heparin-induced Conformational Change in AIIt-AIIt binds Ca 2ϩ with a K d(Ca 2ϩ) of ϳ0.5 mM. When AIIt is bound to chromaffin granules, phospholipid liposomes, or F-actin, the K d(Ca 2ϩ) is reduced to M values (9). The decrease in the K d(Ca 2ϩ) upon phospholipid binding is due to a conformational change in the protein resulting from the binding of the phospholipid to the protein (6). We therefore used CD to examine the Ca 2ϩ requirement for the changes in the conformation of AIIt observed upon heparin binding. As shown in Fig. 3, when assayed in the presence of saturating heparin, the half-maximal change in the CD spectrum of AIIt at 222 nm required 0.084 Ϯ 0.008 M Ca 2ϩ (mean Ϯ S.D., n ϭ 3). In contrast, the half-maximal change in the CD spectrum of AIIt at 222 nm in the absence of heparin required 0.361 Ϯ 0.149 M Ca 2ϩ (mean Ϯ S.D., n ϭ 3). This result therefore establishes that the binding of heparin to AIIt increases the affinity of the protein for Ca 2ϩ .
Specificity of the Heparin-induced Conformational Change in AIIt-We have also examined the effect of a variety of charged polysaccharides on the CD spectra of AIIt at 222 nm. As shown in Table II, charged polysaccharides such as heparan sulfate, chondroitin sulfate, and dextran sulfate failed to induce a large conformational change in AIIt. Similarly, other amino monosaccharides such as N-acetyl-D-glucosamine, N-acetyl-D-galactosamine, and N-acetylneuraminic acid or other monosaccharides such as glucose, mannose, fucose, and galactose failed to induce a substantial change in the conformation of AIIt. We also examined the possibility that other carbohydrates could bind to the heparin-binding site of AIIt without producing a conformational change. This possibility was ruled out by our observation that 50 M heparan sulfate or 50 M N-acetylglucosamine failed to block the conformational change in AIIt induced by 0.5 M heparin.
We have also observed that smaller molecular mass forms of heparin can induce a conformational change in AIIt. As shown in Table II, both 3-and 6-kDa forms of heparin could induce a conformational change in AIIt. However, we have been unable to detect a conformational change in AIIt in the presence of disaccharide heparin derivatives (data not shown). These results therefore suggest that the heparin-induced conformational change in AIIt requires the interaction of AIIt with a region of heparin larger than a disaccharide repeating unit.
Binding of Heparin to an AIIt Affinity Column-Heparin is  an acidic polysaccharide characterized by a disaccharide repeating unit of hexosamine and uronic acid (L-iduronic or Dglucuronic acid) connected through 1-4-linkages. Commercially prepared heparin is heterogeneous due to the varying degree of modification of the functional groups in the disaccharide unit and also to variations in the size of the polysaccharide chains (24). This presented the possibility that the heterogeneous heparin preparation might contain subpopulations of heparin that might not bind to AIIt. We therefore examined the binding of the heparin preparation to an AIIt affinity column. As shown in Fig. 4, ϳ3% of the heparin applied to the AIIt affinity column flowed through the column, whereas ϳ85% was recovered when the AIIt column was washed with 150 mM NaCl. Washing the column with 1 M NaCl failed to further elute any heparin.
Characterization of the Binding of AIIt to Heparin-We observed that in the presence of Ca 2ϩ , the AIIt-heparin complex could be pelleted by centrifugation at 400,000 ϫ g. This simple pelleting assay allowed an alternative estimation of the affinity and capacity of the binding of AIIt to heparin. As shown in Fig.  5, half-maximal binding of AIIt to heparin required 0.115 Ϯ 0.021 M heparin (mean Ϯ S.D., n ϭ 3). The asymptotic maximum of the plot was 1.904 M AIIt, which suggested that ϳ97% of the total AIIt could be recovered in the pellet. Analysis of the binding data (Fig. 5, inset) estimated that AIIt bound to heparin with an apparent K d of 32 Ϯ 6 nM (mean Ϯ S.D., n ϭ 3) and a stoichiometry of 11 Ϯ 0.9 mol of AIIt/mol of heparin (mean Ϯ S.D., n ϭ 3).
We also examined the Ca 2ϩ dependence of AIIt-heparin complex formation. Fig. 6 presents a typical result. Half-maximal formation of the AIIt-heparin complex required pCa 2ϩ ϭ 4.259 Ϯ 0.035. This value (55 M Ca 2ϩ ) was similar to the value of 84 M Ca 2ϩ determined for the half-maximal Ca 2ϩ dependence of the heparin-induced conformational change in AIIt.
As discussed under "Experimental Procedures," AIIt can be purified by NaCl-dependent elution from a heparin affinity column. In the presence of Ca 2ϩ , a 500 mM NaCl wash was sufficient to elute the AIIt from the heparin affinity column. We therefore examined the dependence of the formation of the AIIt-heparin complex on ionic strength. As shown in Fig. 7, half-maximal dissociation of the AIIt-heparin complex required ϳ380 Ϯ 99 mM NaCl. This result is consistent with the formation of AIIt-heparin complexes under physiological conditions.
Identification of a Heparin-binding Site in AIIt-It has been suggested that many heparin-binding proteins contain consensus sequences for glycosaminoglycan recognition (16). Two consensus sequences, referred to as Cardin-Weintraub sequences, that have been identified are XBBBXXBX and XBBXBX, where B has the probability of a basic residue and X is a hydropathic residue. We have found that the p36 subunit of AIIt contains a Cardin-Weintraub sequence of the XBBBXXBX type. This region of the p36 subunit comprises 306 FKKKYGKS 314 . As shown in Table III, in contrast to many heparin-binding proteins that contain the Cardin-Weintraub consensus sequence, this region  4. Binding of heparin to an AIIt affinity column. Heparin (0.5 mg) was dissolved in buffer C containing 25 mM NaCl and applied to an AIIt affinity column equilibrated against the same buffer. The column was washed with buffer C containing 50 mM NaCl and eluted with buffer C containing 150 mM NaCl. Fractions were analyzed for the presence of heparin with the uronic acid/carbazole reaction (see "Experimental Procedures").

FIG. 5. Binding isotherm for the titration of AIIt with heparin.
AIIt (1.971 M) was incubated with 25 mM Hepes, pH 7.5, 1 mM DTT, 150 mM NaCl, 1 mM CaCl 2 , and variable concentrations of heparin at room temperature for 20 min. The reaction mixture was centrifuged at 400,000 ϫ g for 30 min, and the amount of AIIt in the pellet was determined as described under "Experimental Procedures." The line through the points is a nonlinear least-squares curve fit of the data points calculated from computer modeling of data using the logistic equation and the following values: a ϭ 1.904 M (asymptotic maximum), b ϭ 1.592 (slope parameter), c ϭ 0.115 M (value at inflection point, A 0.5 ), and d ϭ 0 (asymptotic minimum). Inset, the fractional saturation (mol of AIIt in pellet/total mol of AIIt) is plotted against the total heparin concentration. The line through the points is a nonlinear least-squares curve fit of the data points calculated from computer modeling of data to the quadratic binding equation (Equation 1 under "Experimental Procedures") with the assumption of a single class of heparin-binding sites on AIIt and two fitting parameters, K d and n, which were allowed to float during the computer iterations. The convergent best fit for these experiments was determined for K d ϭ 32 nM and n ϭ 0.091 mol of heparin/mol of AIIt. of the p36 subunit conforms exactly to the Cardin-Weintraub consensus sequence.
To determine if the Cardin-Weintraub consensus sequence present in the p36 subunit of AIIt participates in heparin binding, we synthesized a peptide to this region of the p36 subunit and examined the potential interaction of the peptide with heparin. As shown in Fig. 8A, the CD spectrum of the peptide comprising 300 LKIRSEFKKKYGKSLYY 316 demonstrated a significant heparin-dependent conformational change. In contrast, only a slight change in conformation was observed when the consensus sequence peptide was incubated with N-acetylglucosamine. We also examined the interaction of two other peptides with heparin. Although heparin produced a conformational change in a peptide modeled to the actin-bundling region of the p36 subunit ( 286 VLIRIMVSR 294 ) (25), a similar conformational change was also induced by the interaction of this peptide with N-acetylglucosamine (Fig. 8B). Similarly, a peptide modeled to the phosphorylation sites of the p36 subunit (15), but containing an additional two lysines at the N terminus (KK 9 KLSLEGDHSTPPSAYGSVKAYT 30 ), demonstrated conformational changes in the presence of both heparin and N-acetylglucosamine (Fig. 8C). Therefore, these results indicate that the region of the p36 subunit of AIIt that contains the Cardin-Weintraub consensus sequence probably participates in heparin binding.
Heparin-dependent Conformational Changes in Other Annexins-Since the p36 subunit of AIIt contains the heparinbinding domain, it was reasonable to suspect that both AIIt and the isolated p36 subunit (annexin II) would undergo conformational changes upon heparin binding. This was not the case, however; and the p36 subunit did not undergo a significant conformational change in the presence of heparin (Table IV). Furthermore, of the six monomeric annexins examined, none demonstrated a specific conformational change in the presence of heparin. This was surprising because all these annexins were purified by heparin affinity chromatography. This observation therefore suggested that the specific heparin-induced conformational change in AIIt did not occur in other annexins.
Affect of Heparin on Membrane-bound AIIt-AIIt is a phospholipid-binding protein that has been shown to be located on both the intracellular and extracellular surfaces of the plasma membrane (9). Since the binding of heparin to AIIt has such a profound effect on the conformation of the protein, it was important to investigate the possibility that the interaction of AIIt with heparin might disrupt the binding of AIIt with membranes. However, as shown in Fig. 9, heparin did not block the interaction of AIIt with phospholipid liposomes.

DISCUSSION
Previous work from our laboratory established that AIIt is a Ca 2ϩ -dependent heparin-binding protein (15). The interaction of AIIt with heparin was also shown to be inhibited by tyrosine phosphorylation of AIIt (15). Since the role that heparin binding plays in the structure or function of AIIt is unknown, the current study was aimed at the characterization of the interaction of heparin with AIIt. Analysis of the CD spectra of AIIt showed that the binding of heparin to AIIt resulted in a pro-  (n ϭ 3). The line through the points is a nonlinear least-squares curve fit of the data points calculated from computer modeling of data to the four-parameter logistic equation (see "Experimental Procedures") and four fitting parameters, a (asymptotic maximum), b (slope parameter), c (value at inflection point, A 0.5 ), and d (asymptotic minimum), which were allowed to float during the computer iterations.

TABLE III Comparison of heparin-binding domains of heparin-binding proteins
Sequences for comparisons were obtained from Ref. 16. The Cardin-Weintraub consensus sequence (X-B-B-B-X-B-X) was derived from the usage frequency of basic (B) and hydropathic (X) residues in a data base of heparin-binding proteins (16).
found change in the conformation of AIIt ( Fig. 1 and Table I).
We also found that in the absence of Ca 2ϩ , a small change in the conformation of AIIt occurred upon heparin binding.
Animal carbohydrate-binding proteins can be broadly classified into seven major groups. These include the C-type or Ca 2ϩdependent lectins, the S-type or Gal-binding galectins, P-type mannose 6-phosphate receptors, the I-type lectins, the pentrax-ins, the hyaluronan-binding proteins, and the heparin-binding proteins (26). The C-type lectins bind several carbohydrates including mannose and galactose and require Ca 2ϩ to form a coordination bond with the sugar ligand. The galectins bind only ␤-galactoside, whereas the P-type proteins bind only mannose 6-phosphate. The I-type lectins bind only sialic acid, whereas the pentraxins bind several carbohydrates such as heparin and sialic acid as well as phosphorylcholine. The hyaluronan-binding proteins bind only hyaluronan. The heparin-   5). Percent change at ⌬ 222 nm ϭ ⌬ 222 nm (AIIt) Ϫ ⌬ 222 nm (AIIt ϩ ligand)/⌬ 222 nm (AIIt) ϫ 100. Annexins (0.2 mg/ml) were incubated in buffer containing 10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.2 mM DTT, 1 mM CaCl 2 , and 5.0 or 0.5 M heparin (AIIt). Annexins used in this study were isolated as described (17) and further purified by Ca 2ϩ -dependent binding to a heparin affinity column as described (15  binding proteins generally demonstrate Ca 2ϩ -independent binding of both heparin and heparan sulfate. Since AIIt binds heparin (Table II) but not phosphorylcholine (15), AIIt is most likely a member of the heparin-binding family of proteins. However, AIIt appears to be unique among heparin-binding protein members in that the binding of AIIt to heparin is stimulated by Ca 2ϩ . Furthermore, AIIt appears to be a unique member of the heparin-binding proteins because AIIt can discriminate between heparin and heparan sulfate ligands.
Several consensus sequences have been identified among members of the heparin-binding family of proteins. For example, the heparin-binding sequence of the C-terminal region of fibronectin has been identified as WQPPRARI (27). In contrast, a region of thrombospondin containing the sequence WSPW has been identified as the heparin-binding region of the protein (28,29). Analysis of several heparin-binding proteins has suggested the potential existence of two consensus sequences referred to as Cardin-Weintraub heparin-binding consensus sequences (16,30). Site-directed mutagenesis and binding studies with synthetic or isolated peptides from several of these proteins have confirmed that this consensus region is often involved in binding heparin (30 -34, 36). Other studies have shown that the orientation of the Cardin-Weintraub consensus sequence within the protein is critical and may determine if the consensus sequence participates in heparin binding (33,37). As shown in Table III, the p36 subunit of AIIt contains a Cardin-Weintraub heparin-binding consensus sequence. Furthermore, a peptide to this region of the p36 subunit of AIIt ( 300 LKIRSEFKKKYGKSLYY 316 ) undergoes a conformational change upon heparin binding (Fig. 8). These results therefore suggest that residues 300 -316 of the p36 subunit of AIIt are involved in heparin binding.
Although the monomeric annexins I-VI bind to a heparin affinity column in the presence of Ca 2ϩ , a heparin-dependent conformational change was not observed for these proteins (Table IV). The p36 subunit of AIIt can exist as a monomer or as a heterotetramer. Heterotetrameric AIIt is composed of two p36 subunits and two p11 subunits. Considering that the p36 subunit (annexin II) binds to a heparin affinity column and contains the Cardin-Weintraub consensus sequence, it was surprising that the p36 subunit did not undergo a conformational change upon heparin binding. This suggests that the heparinbinding site of the p36 subunit and other monomeric annexins is preformed and does not require the recruitment of residues from other regions of the protein. This is consistent with the observation that carbohydrate-binding proteins undergo few if any changes in conformation upon carbohydrate binding (26). The p11 subunit of AIIt does not bind heparin and does not contain any heparin-binding consensus sequences. It is therefore unlikely that the heparin-dependent conformational change in AIIt was due to the coordinated binding of heparin by both the p36 and p11 subunits of AIIt. We cannot, however, rule out the possibility that the binding of the p36 subunit to the p11 subunit induces a conformational change in the p11 subunit that results in exposure of a novel heparin-binding domain. The simplest explanation for the large conformational change in AIIt upon heparin binding is that the orientation of the p36 subunits in AIIt is not optimal for heparin binding. Therefore, the binding of heparin to AIIt results in the realignment of the p36 subunits.
Of particular interest was our observation that the Ca 2ϩ -dependent conformational change in AIIt was induced by heparin, but not by other negatively charged glycosaminoglycans such as heparan sulfate, chondroitin sulfate, and dextran sulfate. Heparan sulfates are structurally related glycosaminoglycans that are found on cell surfaces and in the extracellular matrix, where they form the chains of heparan sulfate proteoglycans and bear only short stretches of dense sulfation. In contrast, heparin is the glycosaminoglycan that is secreted by mast cells and other hematopoietic cells and therefore may serve as a signaling molecule (38,39). To date, a heparinbinding protein capable of distinguishing between heparin and heparan sulfate has not been described. Recently, annexin IV was shown to bind heparin, but the binding of heparin to this protein was inhibited by a variety of carbohydrates including glucose, N-acetylneuraminic acid, heparan sulfate, and chondroitin sulfate (40). In contrast, we have found that heparan sulfate or other glycosaminoglycans do not induce a conformational change in AIIt (Table II). Furthermore, high concentrations of heparan sulfate (50 M) do not inhibit the conformational change in AIIt elicited by 0.5 M heparin (Table II), therefore suggesting that heparan sulfate does not bind to AIIt. However, considering the heterogeneity of the cell-surface heparan sulfate proteoglycan (38), it is possible that AIIt may interact with other heparan sulfate proteoglycans.
We also observed that AIIt formed a large complex with heparin and that this complex was pelleted by centrifugation at 400,000 ϫ g. Analysis of the binding isotherm suggested that AIIt bound heparin with an apparent K d of 32 Ϯ 6 nM (mean Ϯ S.D., n ϭ 3) and a stoichiometry of 11 Ϯ 0.9 mol of AIIt/mol of heparin (mean Ϯ S.D., n ϭ 3). This K d for the binding of heparin to AIIt is slightly lower than the K d reported for the binding of heparin to heparinase (60 nM), acidic fibroblast growth factor (50 -140 nM), or fibronectin (34 nM), but higher than that reported for the binding of heparin to basic fibroblast growth factor (2.2 nM) or antithrombin III (11 nM) (32,41,42). AIIt does not bind to disaccharides of heparin, but does bind to 3-kDa heparin that contains ϳ10 monosaccharides (Table II). The binding of ϳ11 molecules of AIIt to a single 17-kDa heparin strand that contains ϳ50 monosaccharide units (Fig. 5) suggests that AIIt requires ϳ4 -5 monosaccharide units for binding.
The physiological significance of the binding of heparin to AIIt is unclear. Heparin has been shown to interact with enzymes of the clotting and fibrinolysis systems (24), protect proteins from inactivation, play an essential role in the interaction of growth factors with their receptors, directly activate growth factor receptors, and serve as an essential cofactor in cell-cell recognition and cell-matrix adhesion processes (27,35,(43)(44)(45)(46)(47). AIIt is the major cellular receptor for tenascin-C and plasminogen (8,14). It is therefore possible that heparin might be involved in the regulation of the interaction of AIIt with these ligands.