Báo cáo khoa học: Cell surface adhesion of pregnancy-associated plasma protein-A is mediated by four clusters of basic residues located in its third and fourth CCP module

doi:10.1111/j.1432-1033.2004.04061.x

Eur. J. Biochem. 271, 1525–1535 (2004) (cid:1) FEBS 2004

Cell surface adhesion of pregnancy-associated plasma protein-A is mediated by four clusters of basic residues located in its third and fourth CCP module

Kathrin Weyer1, Michael T. Overgaard1, Lisbeth S. Laursen1, Claus G. Nielsen1, Alexander Schmitz1, Michael Christiansen2, Lars Sottrup-Jensen1, Linda C. Giudice3 and Claus Oxvig1 1Department of Molecular Biology, Science Park, University of Aarhus, Denmark; 2Department of Clinical Biochemistry, Statens Serum Institut, Copenhagen S, Denmark; 3Department of Gynecology and Obstetrics, Stanford University, California, USA

generally have a modest effect on PAPP-A heparin binding assessed by chromatography, but cell surface adhesion was critically reduced by several of these substitutions, empha- sizing the relevance of analysis by flow cytometry. The contributions of positively charged residues located in CCP4 were all minor when analyzed by heparin affinity chroma- tography. However, the mutation of CCP4 residues Arg1459 and Lys1460 to Ala almost abrogated cell surface adhesion. Furthermore, when acidic residues of the homologous pro- teinase PAPP-A2 (Asp1547, Glu1555 and Glu1567) were introduced into the corresponding positions in the sequence of PAPP-A, located in each of the three basic clusters of CCP3, binding to heparin was strongly impaired and cell surface binding was abrogated. This explains, at least in part, why PAPP-A2 lacks the ability of cell surface adhesion, and further emphasizes the role of the basic clusters defined in PAPP-A.

Keywords: cell surface adhesion; complement control pro- tein module; glycosaminoglycan; heparin binding; metallo- proteinase. The metalloproteinase pregnancy-associated plasma pro- tein-A (PAPP-A) cleaves a subset of insulin-like growth factor binding proteins (IGFBP), which inhibit the activities of insulin-like growth factor (IGF). Through this proteolytic activity, PAPP-A is believed to regulate IGF bioavailability in several biological systems, including the human repro- ductive system and the cardiovascular system. PAPP-A adheres to mammalian cells by interactions with glycos- aminoglycan (GAG), thus targeting the proteolytic activity of PAPP-A to the cell surface. Based on site-directed muta- genesis, we here delineate the PAPP-A GAG-binding site in the C-terminal modules CCP3 and CCP4. Using heparin affinity chromatography, commonly employed in such studies, we define three clusters of arginines and lysines of CCP3, which are important for the interaction of PAPP-A with heparin. In a model of PAPP-A CCP3-CCP4, basic residues of these sequence clusters form a contiguous patch located on one side of the structure. Binding to the unknown, natural cell surface receptor of PAPP-A, assessed by flow cytometry, also depends on residues of these three basic clusters. However, single or double residue substitutions

The metalloproteinase pregnancy-associated plasma pro- tein-A (PAPP-A) is responsible for cleavage of insulin-like growth factor (IGF)-binding protein (IGFBP)-4 [1,2]. IGFBP-4 is one of six homologous IGF-binding proteins

Correspondence to C. Oxvig, Department of Molecular Biology, University of Aarhus, Science Park, Gustav Wieds Vej 10C, DK-8000 Aarhus C, Denmark. Fax: +45 89425068, E-mail: co@mb.au.dk Abbreviations: b2GPI, b2-glycoprotein I; CCP, complement control protein; GAG, glycosaminoglycan; ELISA, enzyme-linked immuno- sorbent assay; IGF, insulin-like growth factor; IGFBP, insulin-like growth factor binding protein; mAb, monoclonal antibody; MMP, metalloproteinase; NaCl/Pi, phosphate-buffered saline; PAGE, polyacrylamide gel electrophoresis; PAPP-A, pregnancy-associated plasma protein-A; PAPP-A2, pregnancy-associated plasma protein- A2; proMBP, the proform of eosinophil major basic protein; SCR, short consensus repeat. Enzyme: pregnancy-associated plasma protein-A (PAPP-A, pappalysin-1) (EC 3.4.24.79) (Received 9 January 2004, revised 25 February 2004, accepted 27 February 2004)

(IGFBP-1–6), which bind IGF-I and -II with high affinities [3,4]. The IGFs are established regulators of cell differentiation and proliferation [5]. IGF bound to IGFBP is restricted from interacting with its cognate cell surface receptor, but proteolytic cleavage of IGFBP results in the release of bioactive IGF [6]. Thereby, PAPP-A functions as a regulator of local IGF bioavailability in several including the human reproductive biological systems, system [7–10] and the cardiovascular system [11,12]. Furthermore, it has been found that also IGFBP-2 [13] and -5 [14] are PAPP-A substrates, but to what extent PAPP-A functions as an IGFBP-2 and -5 proteinase in vivo is currently not known. Interestingly, cleavage of IGFBP-4 and -2 is greatly enhanced (about 20-fold) by IGF [14]. IGF is believed to enhance cleavage by binding to the IGFBP, as IGF is unable to bind PAPP-A directly [14].

PAPP-A is classified as a metalloproteinase of the metzincin superfamily [15], which is a diverse group of zinc endopeptidases, including the astacins, the reprolysins, the serralysins, and the matrix metalloproteinases (MMPs) [16,17]. As PAPP-A and the recently discovered homologue

1526 K. Weyer et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

receptor at the cell surface using flow cytometry. Based on our experimental data and a homology model of PAPP-A CCP3-CCP4, we propose that a patch of basic residues on one side of the surface of CCP3 forms the binding site responsible for the interactions between PAPP-A and GAGs.

Experimental procedures

Fig. 1. Schematic representation of the 1547-residue PAPP-A subunit. The localization of the approximately 300-residue proteolytic domain and the five CCP modules are indicated. Zn indicates the binding of a zinc ion to a sequence stretch of PAPP-A that conforms to the elon- gated zinc-binding motif (HEXXHXXGXXH) of the metzincins.

Plasmid construction and mutagenesis

PAPP-A2 [18] cannot be grouped into any of these families, they constitute the founding members of the pappalysins, a fifth family [15]. PAPP-A and -A2 are both synthesized as preproproteins of 1627 [19] and 1791 [18] residues, respect- ively. Mature PAPP-A2 (1558 residues), which shares 46% of its residues with mature PAPP-A (1547 residues), cleaves IGFBP-5 (and -3), but not IGFBP-4 [18].

In addition to the proteolytic domain, the pappalysin subunits contain three lin-notch modules (LNR1-3), two of which are located in the proteolytic domain, and five complement control protein modules (CCP1–5, each of 57–77 residues) located in the C-terminus of the protein [18] (Fig. 1). The CCP module, also known as the short consensus repeat (SCR), is frequently found in complement proteins [20], but within the metzincins it is unique to the pappalysins.

An expression plasmid encoding human PAPP-A, pcDNA3.1-PAPP-A (pPA), containing cDNA encoding residues 1–1547 of the mature polypeptide cloned into the HindIII/XbaI sites of pcDNA3.1+ (Invitrogen) was made previously [2]. [Note that the numbering of preproPAPP-A (AAC50543) [19] and preproPAPP-A2 (AF311940) [18] is used in this paper. Glu1 of the mature PAPP-A polypeptide is at position 81 of preproPAPP-A]. Mutagenesis was carried out by overlap extension PCR [26] using the PAPP- A expression plasmid pPA-BspEI [15] as template. All polymerase chain reactions were carried out with Pfu DNA polymerase (Promega). Outer primers were 5¢-CATCA TCGGACAGCCAGCAGCATC-3¢ (corresponding to pre- proPAPP-A nucleotides 3123–3147), and 5¢-GCAAAC AACAGATGGCTGGCAACTAG-3¢ (corresponding to nucleotides 1036–1061 of pcDNA3.1+). Internal primers with an overlap of 26–33 bp were used to generate mutated fragments. The resulting PCR products were swapped into the KpnI/XbaI sites of pPA-BspEI. All plasmid constructs were verified by sequence analysis.

Cell culture and expression of recombinant proteins

PAPP-A is secreted as a dimer of 400 kDa [21], but circulates during human pregnancy as a disulfide-bound, 2 : 2 complex with the proform of eosinophil major basic protein (proMBP) [22]. In the PAPP-A–proMBP complex, proMBP functions as an inhibitor of PAPP-A enzymatic activity [2]. Interestingly, the 222-residue preproMBP is substituted with a heparan sulfate glycosaminoglycan (GAG) at Ser62 [23].

Human embryonic kidney 293T cells (293tsA1609neo) [27] were maintained in high-glucose Dulbecco’s modified Eagle’s medium supplemented with 10% (v/v) fetal bovine serum, 2 mM glutamine, nonessential amino acids, and gentamicin (Invitrogen). Cells were plated onto 6 cm tissue culture dishes and transfected 18 h later by calcium phosphate coprecipitation using 10 lg of plasmid DNA prepared by QIAprep Spin Kit (Qiagen). Plasmid transfection were pPA-BspEI constructs used for and mutated pPA-BspEI constructs. After a further 48 h, supernatants were harvested and cleared by centri- fugation.

Western blotting

In recent studies, PAPP-A was found to be associated with the cell membrane [24], and was shown to adhere to the cell surface by interactions with heparan sulfate GAG [25]. It was shown that cell surface binding of PAPP-A is abrogated by heparinase treatment of the cells, and that heparin or heparan sulfate effectively competes for PAPP-A binding. As PAPP-A remains proteolytically active when bound to the cell surface, a model was proposed in which heparan sulfate proteoglycans on the cell surface function as extracellular docking molecules for PAPP-A, positioning PAPP-A favorably for directed proteolytic activity in the vicinity of the IGF receptor. Furthermore, it was shown that proMBP not only inhibits PAPP-A proteolytic activity, but also prevents PAPP-A adhesion to the cell surface. This effect was reversed by treatment of the PAPP-A–proMBP complex with hepari- nase, suggesting that, within the complex, the GAG chain of proMBP interacts with the PAPP-A GAG binding site, thereby preventing PAPP-A from interacting with heparan sulfates on the cell surface.

Herein, we determined the localization of the GAG binding site in the C-terminal of PAPP-A by site-directed mutagenesis of basic residues located in CCP3 and CCP4. The relative GAG binding affinities of the mutants were evaluated by heparin affinity chromatography and by analysis of binding to the natural, GAG-substituted For immunovisualization, culture supernatants containing recombinant wild-type or mutant protein were separated by nonreducing SDS/PAGE (5–15% Tris-glycine gels) and blotted onto a poly(vinylidene difluoride) membrane (Mil- lipore). The blots were dried and incubated with primary antibody (PAPP-A mAb 234-2 [28]) diluted in 10 mM sodium phosphate, 150 mM sodium chloride, 0.01% Tween 20, pH 7.2 (NaCl/Pi-T) containing 2% (w/v) skimmed milk powder for 1 h at room temperature. Blots were washed in NaCl/Pi-T and incubated with secondary peroxidase-con- jugated antimouse IgG (P260, Dako) diluted in NaCl/Pi-T with 2% (w/v) skimmed milk powder for 30 min at room temperature. Blots were washed again in NaCl/Pi-T and

Glycosaminoglycan binding of PAPP-A (Eur. J. Biochem. 271) 1527

(cid:1) FEBS 2004

Results

proteins were detected by enhanced chemiluminescence (ECL, Amersham).

Heparin and cell surface binding of PAPP-A is mediated primarily through CCP3 Heparin affinity chromatography

Culture supernatants containing recombinant wild-type or mutant protein (at about 10 lgÆmL)1) were diluted (1 : 4) in buffer A (50 mM Tris, 2 mM CaCl2, pH 7.5), and 600 lL of diluted supernatant was loaded onto a 6.5 mL column (1.2 · 5.7 cm) packed with heparin-Sepharose (CL-6B, Amersham Biosciences) previously equilibrated with buf- fer A. Bound proteins were eluted with a linear NaCl gradient, formed by increasing the amount of buffer B (1 M NaCl, 50 mM Tris, 2 mM CaCl2, pH 7.5) at a flow rate of 0.5 mLÆmin)1 (Pharmacia LKB Gradient Pump). Fractions of 0.5 mL were collected and analyzed by ELISA. The position of the gradient, relative to the fractions collected, was controlled by measuring the conductivity (cdm3, Radiometer Copenhagen).

Enzyme-linked immunosorbent assay (ELISA)

The levels of recombinant PAPP-A or PAPP-A mutants eluted from the column were measured by a standard sandwich enzyme-linked immunosorbent assay. From the collected fractions, 100 lL were transferred to 96 well plates (Nunc) coated with antibodies against PAPP-A–proMBP [21]. A monoclonal antibody (PAPP-A mAb 234-4 [28]), followed by peroxidase-conjugated antimouse IgG (P260, Dako), was used for detection. PAPP-A–proMBP purified from pregnancy serum [21] was used to establish standard curves.

Flow cytometry

As PAPP-A, but not PAPP-A2, binds to the surface of mammalian cells, previous analysis of PAPP-A/PAPP-A2 chimeras by flow cytometry of transfected cells allowed preliminary mapping of the binding site to CCP3 and CCP4 of PAPP-A [25]. Both CCP3 and CCP4 appeared to be involved in adhesion of PAPP-A. PAPP-A, in which CCP3 was replaced with PAPP-A2 sequence, did not show cell surface binding, and PAPP-A with CCP4 of PAPP-A2 showed only weak binding. Furthermore, PAPP-A2 con- taining both CCP3 and CCP4 of PAPP-A showed cell surface adhesion similar to wild-type PAPP-A. To compare the interactions of PAPP-A with the unknown, natural heparan sulfate-substituted receptor and the interactions with isolated heparin, we analyzed the PAPP-A/PAPP-A2 chimeras by affinity chromatography on heparin-Seph- arose (Fig. 2), a method commonly used to determine the relative affinities of proteins for heparan sulfate-like GAGs [31–36]. Recombinant wild-type PAPP-A, contained in culture medium, eluted from the column at 0.68 M NaCl. PAPP-A containing CCP3 of PAPP-A2 did not show any heparin binding, as it eluted before the gradient reached 0.1 M NaCl, consistent with the results obtained by flow cytometry. In contrast, the PAPP-A construct carrying CCP4 of PAPP-A2 eluted at 0.58 M NaCl, showing that the binding was only moderately reduced compared to wild-type PAPP-A. As expected, PAPP-A2 with CCP3 and CCP4 of PAPP-A eluted at the position of wild-type PAPP-A. These results confirm that the adhesion site of PAPP-A is located within CCP3 and CCP4, and they validate the analysis of PAPP-A GAG binding by affinity chromatography using gradient elution as a complement- ary method of analysis. To quantitatively assess the effects of single amino acid substitutions, analysis of further mutants was carried out by heparin affinity chromato- graphy, and cell surface adhesion was analyzed by flow cytometry.

Chromatographic analysis of PAPP-A in which basic residues of CCP3 and CCP4 are substituted

Transfected 293T cells were detached with phosphate- buffered saline (NaCl/Pi) containing 5 mM EDTA 48 h post-transfection, washed with cold L15 medium (Sigma) containing 2% fetal bovine serum (L15/FBS) and incubated on ice with primary monoclonal antibody (PAPP-A mAb 234–5 [28]) for 1 h. After washing two times in L15/FBS, the cells were incubated with fluorescein isothiocyanate goat antimouse IgG (Zymed Laboratories Inc.) diluted 1 : 50 for 30 min. After two washes, the cells were suspended in NaCl/Pi containing 2% paraformaldehyde and analyzed on a flow cytometer (BD Biosciences). At least 5000 cells were analyzed. To allow comparison of adhesion, geometric mean fluorescence values were calculated, and compared after background (mock signal) subtraction.

Molecular modeling

CCP3 of PAPP-A2 has only six positively charged residues, whereas 13 of the 65 residues of PAPP-A CCP3 are positively charged. The latter are located in three sequence clusters, which we denote A, B, and C (Fig. 3A). We therefore mutated single residues and combinations of residues within these three clusters to alanine. Likewise, CCP4 of PAPP-A has five basic residues of which only one is conserved in PAPP-A2 (Fig. 3A). These five basic residues of CCP4 were also substituted with alanine. A total of 17 CCP mutants (Fig. 3A) were expressed in human embryonic kidney 293T cells. All mutants expressed at the level of wild-type PAPP-A as determined by ELISA (not shown), and they migrated in SDS/PAGE as 400 kDa dimeric proteins (Fig. 4).

Homology models of CCP3-CCP4 of PAPP-A and PAPP- A2 were built using the SEQMOD program of Genemine [29] (release 3.5.2 for Linux). b2-Glycoprotein I (b2GPI) [30] (PDB code 1C1Z) was used as a template after manual alignment to CCP1-CCP2 of b2GPI. All images were made using SWISS-PDBVIEWER version 3.7 (SP5). Secondary struc- ture assignments and calculation of surfaces were carried out with the built-in tools of SWISS-PDBVIEWER. Figures were saved from SWISS-PDBVIEWER AND rendered using POV-RAY for Windows (version 3.5). Mutants of cluster A showed moderately reduced hep- arin binding compared to wild-type PAPP-A (Fig. 5A). Mutants A1 and A2 eluted at 0.58 M and 0.61 M NaCl,

1528 K. Weyer et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 2. Analysis of PAPP-A/PAPP-A2 chimeras by flow cytometry and heparin affinity chromatography. (A) Human embryonic kidney 293T cells were transfected with empty vector (mock), vector containing wild-type PAPP-A cDNA, or with cDNA encoding PAPP-A/PAPP-A2 chimeras, in which selected sequence stretches were replaced with sequence of PAPP-A2, as indicated above the histograms [25]. The cells were detached 48 h post-transfection and incubated with monoclonal PAPP-A antibodies (mAb 234-5). After washing, the cells were incubated with secondary fluorescein isothiocyanate-labeled antibodies and then analyzed by flow cytometry to visualize surface-bound antigen. The histograms show the number of cells (Events) at a given fluorescence intensity. Both CCP3 and CCP4 are required for adhesion to the cell surface. (B) The same samples were also eluted with a linear gradient of increasing sodium chloride from a heparin-Sepharose column, as shown below the corresponding histograms. The amount of PAPP-A as measured by ELISA is plotted relative to the amount in the peak fraction. Both wild-type PAPP-A and the mutant in which CCP3 and CCP4 are derived from PAPP-A eluted at 0.68 M NaCl. A substantial loss of retention was observed with PAPP-A containing PAPP-A2 CCP3 (elution < 0.10 M NaCl), but only a moderate loss of retention was seen with PAPP-A containing CCP4 of PAPP-A2 (elution at 0.58 M NaCl), in contrast to the results obtained by flow cytometry.

Fig. 3. Mutated residues of CCP3 and CCP4 of PAPP-A. (A) The sequence of PAPP-A (PA) CCP3 and CCP4 are aligned with the corresponding sequence of PAPP-A2 (PA2). Basic residues of PAPP-A mutated to alanine are shown in bold. Their combinations in mutants are indicated below the alignment. Clusters A, B and C are boxed, and cysteine residues, all conserved, are shaded. (B) The sequence of PAPP-A CCP3 is shown. Basic residues substituted with acidic residues, as found in the corresponding positions of PAPP-A2, are shown in bold. Their combinations in mutants are indicated below the sequence.

respectively. Mutant A3, in which all four basic residues of cluster A are substituted, eluted at 0.45 M NaCl. In contrast, PAPP-A binding to heparin was markedly reduced by substitutions in clusters B and C. Mutants of cluster B showed less binding the more residues were mutated: Mutant B3, in which all three basic residues are

Glycosaminoglycan binding of PAPP-A (Eur. J. Biochem. 271) 1529

(cid:1) FEBS 2004

Cell surface binding of mutants with basic residues substituted

The series of CCP-mutants were also analyzed for cell surface binding trough the natural, GAG-substituted, PAPP-A receptor using flow cytometry. Surprisingly, only CCP4 mutant 4–1 bound the cell surface as wild-type PAPP-A (Fig. 6). Mutants A1, A2, C1, and C2 showed substantially reduced adhesion, whereas the binding to the cell surface was almost or completely abrogated for mutants A3, B2, B3, C3–6, and B3/C6 (Fig. 6). In general, mutants eluting from the heparin-Sepharose column at salt concen- trations lower than 0.51 M were unable to bind the cell surface. Interestingly, however, the CCP4 mutant 4–4, with Arg1459 and Lys1460 mutated to alanine, eluted from the column at 0.58 M NaCl, but exhibited almost no cell surface binding. Thus, Arg1459 and Lys1460 of CCP4 appear to be responsible for the differentiation (Fig. 2) in PAPP-A binding to heparin and to the natural cell surface receptor. In contrast to these results, the CCP3 residues defined as critical for heparin binding were found to be equally important for cell surface adhesion of PAPP-A.

lane 8, mutant C6;

Fig. 4. Western analysis of mutated proteins. Nonreducing Western blots of culture supernatants from transfected 293T cells using monoclonal PAPP-A antibodies (mAb 234-2), verifying that all mutants expressed as 400 kDa dimeric proteins like wild-type PAPP- A. (A) CCP3 mutants of clusters A and B: Lane 1, medium from mock-transfected cells; lane 2, wild-type PAPP-A; lane 3, mutant A1; lane 4, mutant A2; lane 5, mutant A3; lane 6, mutant B1; lane 7, mutant B2; lane 8, mutant B3. (B) CCP3 mutants of cluster C: Lane 1, medium from mock-transfected cells; lane 2, wild-type PAPP-A; lane 3, mutant C1; lane 4, mutant C2; lane 5, mutant C3; lane 6, mutant C4; lane 7, mutant C5; lane 9, mutant B3/C6. (C) Mutants of CCP4: Lane 1, medium from mock-transfected cells; lane 2, wild-type PAPP-A; lane 3, mutant 4–1; lane 4, mutant 4–2; lane 5, mutant 4–3; lane 6, mutant 4–4.

The introduction of PAPP-A2 CCP3 acidic residues into PAPP-A causes reduced heparin and cell surface binding

Several of the PAPP-A residues identified by mutagenesis as important for both the interactions with heparin-Sepharose and the cell surface ligand are also found in PAPP-A2 (Fig. 3A). In particular, basic residues of clusters B and C are conserved, but the basic residues of cluster A are absent from the PAPP-A2 sequence. Of particular interest, however, one basic residue in each of the three clusters A, B, and C have a counterpart in PAPP-A2, which is an acidic residue. We speculated that these three residues might contribute signi- ficantly to the inability of PAPP-A2 to interact with GAGs, and therefore introduced the same acidic residues in various combinations into the PAPP-A sequence (Fig. 3B). All six mutants expressed at the level of wild-type PAPP-A, and they migrated in SDS/PAGE as 400 kDa dimeric proteins (not shown). Reduced binding to heparin was observed with all single-residue mutants (Ac1, Ac2, and Ac3; eluting at 0.45–0.58 M NaCl), and when combined (mutants Ac4, Ac5, and Ac6; eluting at 0.22–0.38 M NaCl), the heparin binding was markedly reduced (Fig. 7A).

substituted, eluted at 0.32 M NaCl, less than half of the concentration required to elute wild-type PAPP-A (Fig. 5B). The effect of mutations in cluster C was similar to B. The greatest effect was seen with mutant C6, which eluted at 0.25 M NaCl (Fig. 5C). Mutant B3/C6, combi- ning the substitutions of B3 and C6, eluted at 0.11 M NaCl, below physiological ionic strength, suggesting that basic residues in clusters B and C make a major contribution to the formation of the GAG-binding site of PAPP-A.

the presence of

All mutants of this series (Ac1 to Ac6) displayed reduced binding compared to wild-type PAPP-A when analyzed by flow cytometry (Fig. 7B). Interestingly, of the mutants carrying a single amino acid substitution, Ac1 and Ac3 were still able to interact weakly with the cell surface, whereas Ac2 did not show any cell surface binding. Thus, a single acidic residue introduced into cluster B, replacing Lys1376, is sufficient to abrogate cell surface adhesion of PAPP-A, suggesting that these acidic residues, combined with missing basic residues, renders PAPP-A2 unable to bind to the surface of cells.

Comparison of flow cytometry and heparin affinity chromatography

Only weak effects were seen in the analysis of basic residue-mutants of CCP4 (Fig. 5D). The greatest reduction was seen with mutant 4–4, which eluted at 0.58 M NaCl, whereas mutant 4–1 showed the same heparin binding as wild-type PAPP-A. Single residue mutants of CCP4 eluted from 0.61 to 0.68 M NaCl, whereas, in comparison, single residue mutants of CCP3 eluted from 0.45 to 0.58 M NaCl. In conclusion, our data indicate that PAPP-A heparin binding involves two basic clusters within the CCP3 module (B and C) and that an additional neighboring cluster (A) and, to a lesser extent, basic residues in CCP4 contribute to heparin binding. We generally observe a good correlation between data obtained by flow cytometry and heparin affinity

1530 K. Weyer et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 5. Heparin binding affinity of wild-type PAPP-A and CCP3 and CCP4 mutants with basic residues substituted. PAPP-A and PAPP-A mutants expressed in 293T cells were analyzed by heparin-Sepharose chromatography, as detailed in the legend of Fig. 2B. The identities of the mutants analyzed are given in Fig. 3, and the insets list the peak fraction numbers (Fr) and the corresponding concentration of NaCl. (A) Mutants of CCP3 cluster A. (B) Mutants of CCP3 cluster B. (C) Mutants of CCP3 cluster C. (D) Mutants of CCP4.

Homology model of CCP3 and CCP4 modules of PAPP-A and PAPP-A2

Fig. 6. Cell surface adhesion of CCP3 and CCP4 mutants with basic residues substituted. The abilities of wild-type PAPP-A and PAPP-A mutants (Fig. 3) to adhere to the surface of 293T cells were assessed by flow cytometry. Mean absolute fluorescence intensities of all proteins analyzed are expressed relative to the fluorescence intensity of wild- type PAPP-A. The mutant B3/C6 is depicted with an asterisk (*). All values are averages of three independent experiments, and standard deviations are indicated with bars.

A model of the adjoining CCP3 and CCP4 modules of PAPP-A was built using SEGMOD of Genemine and CCP modules 1 and 2 of b2GPI [30] as a template (Fig. 9). The identity between the aligned sequence stretched is relatively low, but the alignment emphasizes conserved key residues, including four cysteines (Fig. 9C). The modeled domains (Fig. 9A) conform to the CCP structure known from several proteins [20,30]. The globular, elongated CCP module consists of a hydrophobic core surrounded by b-strands that have their N- and C-termini at opposite ends of a longitudinal axis. The b-strands run approximately parallel and antiparallel to this axis, with the first and third turn located at the C-terminal end, and the second and fourth turn at the N-terminal end.

Mutated, basic CCP3 residues found to be important for binding to heparin and/or adhesion to cells are located in a large cluster on one side in the C-terminal end of the CCP3 module. Important CCP4 residues are located on the edge of this domain, close to CCP3 (Fig. 9A). In a view of the molecular surface, the majority of CCP3 residues engaged in binding to GAG define a basic, circular patch with a diameter of about 15 A˚ , occupying about half of the side of the CCP3 module (Fig. 9B).

A similar model of PAPP-A2 CCP3-CCP4, also using b2GPI as a template, was prepared (Fig. 9B–C). As expected, the overall structures of the two models were very similar, but interestingly, the molecular surface of CCP3 of PAPP-A and PAPP-A2 are very different (Fig. 9B). The basic, circular patch defined in PAPP-A chromatography (Fig. 8). Mutant Ac3, however, showed stronger binding to the cell surface than could be expected from its elution profile from heparin-Sepharose. In contrast, mutant 4–4 of CCP4 bound less to the cell surface than expected from its interaction with heparin. Interestingly, there appears to be a threshold value for cell surface adhesion, as cell surface binding was seen only with PAPP- A mutants eluting at > 0.51 M NaCl in heparin affinity chromatography. These observations emphasize the rele- vance of analyzing binding to the natural ligand, in this case only possible by using flow cytometry.

Glycosaminoglycan binding of PAPP-A (Eur. J. Biochem. 271) 1531

(cid:1) FEBS 2004

Fig. 8. Comparison of flow cytometry and heparin affinity chromato- graphy. The absolute fluorescence intensity of cells transfected with PAPP-A mutants containing basic residue substitutions (Fig. 3A) and mutants with acidic residues introduced (Fig. 3B) are plotted as function of the NaCl concentration in peak fractions when eluted from a column of heparin-Sepharose. The dotted line shows the background level of fluorescence of mock-transfected cells.

Fig. 7. Analysis of PAPP-A mutants with acidic residues of PAPP-A2 substituted into CCP3. (A) Wild-type PAPP-A and mutants Ac1–6 were analyzed by heparin-Sepharose chromatography (see legend of Fig. 2B). The identities of the mutants analyzed are given in Fig. 3, and the insets list the peak fraction numbers (Fr) and the corres- ponding concentration of NaCl. (B) The same set of wild-type protein and mutants was also analyzed by flow cytometry. Mean absolute fluorescence intensities of all proteins analyzed are expressed relative to the fluorescence intensity of wild-type PAPP-A. All values are averages of three independent experiments, and standard deviations are indi- cated with bars.

interaction with heparin, assessed by heparin affinity chromatography, but the effects of substitutions in clusters B and C were the most pronounced (Fig. 5A–C). In all cases, the effect of substituting a single residue or two residues together (A1, A2, B1, B2, C1, C2, and C3) was readily measurable. Compared to wild-type PAPP-A, eluting at 0.68 M NaCl, mutants of this subset all eluted at 0.45–0.61 M NaCl (Fig. 5A–C). The substitution of more residues together further decreased the ability of PAPP-A to bind heparin; mutants, in which three or more residues were substituted, eluted in the interval from 0.11 to 0.45 M NaCl (Fig. 5A–C). Only one mutant, the combination of B3 and C6, in which a total of seven basic residues from clusters B and C are substituted with alanine, eluted below physiologic ionic strength (0.11 M NaCl). All other mutants eluted at 0.25 M NaCl or above.

CCP3 is absent from CCP3 of PAPP-A2. In general, the basic character of this face of PAPP-A2 CCP3 is much less pronounced and furthermore contains interspersed acidic residues, in accordance with its inability to bind GAG.

Discussion

The ability of all mutants to bind to the natural, unknown GAG-substituted PAPP-A receptor on the surface of mammalian cells was also measured using flow cytometry (Fig. 6). Surprisingly, CCP3 mutants, in which only one or two residues are substituted, showed either substantially reduced or almost no cell surface binding. All mutants with three or more residues substituted showed very little (A3) or no (B3, C4, C5, C6) surface binding at all (Fig. 6). We thus conclude that several basic residues of CCP3 are important for the interaction between PAPP-A and GAGs. PAPP-A appears to bind stronger to heparin than to the GAG of the physiological cell surface receptor, as efficient adhesion was only observed with the wild-type protein. This emphasizes the importance of analyzing binding to the natural ligand by flow cytometry. What appears in heparin chromatography to be a minor effect of substitution can in fact be dramatic, assessed by analysis of binding to the physiological ligand on the cell surface. However, as the discriminatory potential clearly is greater using heparin chromatography, the two methods are complementary. We recently proposed a model, in which cell surface targeted proteolysis of PAPP-A plays a key role in the regulation of IGF activity [25]. The aim of the present study, to delineate and characterize the adhesion site in CCP3 and CCP4 of PAPP-A, was pursued by the analysis of mutated PAPP-A variants, in which basic residues were substituted.

In a homology model of PAPP-A CCP3-CCP4 (Fig. 9), prepared using b2-glycoprotein I (b2GPI) [30] as a template, all basic residues of CCP3 are located on one side of the molecule (Fig. 9A). A surface view of this model, shows that By substituting all basic residues of CCP3 into alanine, we have defined three sequence clusters of PAPP-A CCP3, containing four (cluster A), three (cluster B), and six (cluster C) basic residues, respectively. They all contribute to the

1532 K. Weyer et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 9. Models of PAPP-A and PAPP-A2 CCP3-CCP4. (A) A model of PAPP-A CCP3-CCP4 was made by homology modeling using Genemine [29] and the structure of CCP1-CCP2 of b2-glycoprotein I (b2GPI) [30] (PDB code 1C1Z) as a template. Annotated side chains of selected, basic residues and three disulfide bonds (yellow) in each domain are shown. (B) A similar model was made of PAPP-A2 CCP3-CCP4 for comparison with the modeled domains of PAPP-A. The molecular surfaces of the two model structures are displayed in similar orientations, and the positions of selected residues are annotated. Negatively charged side chains are red, positively charged side chains are shown in blue, polar residues in yellow and nonpolar in gray. Boxed residues, substituted in mutants Ac1 to Ac6, have acidic counterparts in PAPP-A2. (C) Alignment of the CCP3-CCP4 sequence of PAPP-A and PAPP-A2 with the sequence of b2GPI (GenBank P02749), on which the models are based. Known secondary structures (b-strands only; named according to Schwarzenbacher et al. [30]) are indicated with bars above the b2GPI sequence, and conserved residues are shaded. The known arrangement of disulfide bonds of PAPP-A (C1–C5, C2–C3, C4–C6) [42] are indicated with lines. Disulfide bonds, including the additional disulfide bond contained in each module of both PAPP-A and PAPP-A2 (C2–C3), were formed in both models.

residues of clusters B and C form a circular, all-basic patch of about 15 A˚ , occupying at least half of this side (Fig. 9B). Most residues of cluster A, found to be less important for heparin binding, are located next to this patch, closer to the CCP2 domain.

(90%) decrease in cell surface binding, corresponding closely to the decrease seen when the entire CCP module is swapped with CCP4 of PAPP-A2 (Fig. 2A). Hence, Arg1459 and Lys1460 are responsible for the contribution of CCP4 to PAPP-A surface binding. This result substan- tiates that CCP4 contributes to GAG binding: Swapping entire domains of CCP containing proteins may disturb interactions between neighboring domains, as intron/exon junctions do not colocalize with domain boundaries, indicating that the modules are not independent domains [20]. This, in turn, may disturb interaction with ligands depending on contact with two domains, or an interdomain A similar analysis was carried out with CCP4. The entire CCP4 sequence of PAPP-A contains only five basic residues that were all substituted into alanine, alone or in combina- tions (Fig. 3). A modest effect of alanine substitution was seen on both heparin (Fig. 5D) and cell surface binding (Fig. 6), except for mutant 4–4, in which Arg1459 and Lys1460 are substituted. This mutant showed a substantial

Glycosaminoglycan binding of PAPP-A (Eur. J. Biochem. 271) 1533

(cid:1) FEBS 2004

weaker heparin binding than all other mutants with basic residues substituted, except for mutant B3/C6, in which seven basic residues are replaced (Fig. 7A). Importantly, any combination of two of the three acidic residue mutants caused PAPP-A to loose its ability to bind to the cell surface (Fig. 7B). Based on this, the presence of these three acidic residues of PAPP-A2, in particular Glu1555, corresponding to Lys1376 of PAPP-A, substituted alone in Ac2, most likely explain the inability of PAPP-A2 to bind GAG.

region. Therefore, our interpretation of domain swapping experiments (Fig. 2) is likely not in error, as the two-residue substitution made is gentler in terms of disturbance of the domain interphase. In our model of PAPP-A CCP3-CCP4, residues Arg1459 and Lys1460 are located in the N-terminal end of CCP4, close to the circular, basic patch of CCP3 (Fig. 9A–B). The largest linear distance between positive charges of residues likely involved in binding is close to 42 A˚ (Lys1366–Arg1459), indicating that a stretch of at least 5 disaccharide units may be involved in the interaction with PAPP-A.

In conclusion, we propose that the GAG binding domain of PAPP-A is composed of four linear, contiguous, basic clusters brought spatially close to one another in the three- dimensional structure of the protein, and primarily located in the CCP3 module. This conclusion is supported experi- mentally, and also by a theoretical model structure of PAPP-A CCP3-CCP4. Furthermore, the inability of PAPP- A2 to bind GAG is also in agreement with a model structure of PAPP-A2 CCP3-CCP4, and with experimental data. Interestingly, although no single or double residue substi- tution has a dramatic effect in the binding of PAPP-A to heparin, cell surface binding was critically reduced by several of these substitutions.

Acknowledgements

This work was supported by grants from the Novo Nordic Foundation, and the National Institute of Health (HD31579-07).

References

1. Lawrence, J.B., Oxvig, C., Overgaard, M.T., Sottrup-Jensen, L., Gleich, G.J., Hays, L.G., Yates, J.R. 3rd & Conover, C.A. (1999) The insulin-like growth factor (IGF)-dependent IGF binding protein-4 protease secreted by human fibroblasts is pregnancy- associated plasma protein-A. Proc. Natl Acad. Sci. USA 96, 3149–3153.

2. Overgaard, M.T., Haaning, J., Boldt, H.B., Olsen, I.M., Laursen, L.S., Christiansen, M., Gleich, G.J., Sottrup-Jensen, L., Conover, C.A. & Oxvig, C. (2000) Expression of recombinant human pregnancy-associated plasma protein-A and identification of the proform of eosinophil major basic protein as its physiological inhibitor. J. Biol. Chem. 275, 31128–31133.

3. Jones, J.I. & Clemmons, D.R. (1995) Insulin-like growth factors and their binding proteins: biological actions. Endocr. Rev. 16, 3–34.

4. Hwa, V., Oh, Y. & Rosenfeld, R.G. (1999) The insulin-like growth factor-binding protein (IGFBP) superfamily. Endocr. Rev. 20, 761–787.

5. Daughaday, W.H. & Rotwein, P. (1989) Insulin-like growth fac- tors I and II. Peptide, messenger ribonucleic acid and gene struc- tures, serum, and tissue concentrations. Endocr. Rev. 10, 68–91. 6. Clay Bunn, R. & Fowlkes, J.L. (2003) Insulin-like growth factor binding protein proteolysis. Trends Endocrinol. Metab. 14, 176–181.

7. Conover, C.A., Faessen, G.F., Ilg, K.E., Chandrasekher, Y.A., Christiansen, M., Overgaard, M.T., Oxvig, C. & Giudice, L.C. (2001) Pregnancy-associated plasma protein-A is the insulin-like growth factor binding protein-4 protease secreted by human ovarian granulosa cells and is a marker of dominant follicle selection and the corpus luteum. Endocrinology 142, 2155–2158. 8. Mazerbourg, S., Overgaard, M.T., Oxvig, C., Christiansen, M., Conover, C.A., Laurendeau, I., Vidaud, M., Tosser-Klopp, G., Zapf, J. & Monget, P. (2001) Pregnancy-associated plasma pro- tein-A (PAPP-A) in ovine, bovine, porcine, and equine ovarian

Within the entire sequence stretch of CCP3 and CCP4, only one region contains a sequence motif identified in other GAG binding proteins. The sequence SKKRAFKT-1397 of cluster C in CCP3 conforms to the consensus sequence XBBBXXBX, in which X is a hydrophatic amino acid [37,38]. The motif is generally believed to adopt an a-helical conformation [37,39], and may also do so in PAPP-A CCP3, as it is located in a long loop to which no template structure corresponds in the current alignment, and which appears to be unique among CCP modules (Fig. 9A,C). In the modeled structure, the motif occupies half of the circular, basic patch. However, our data show that three other regions are also involved in the binding, even though they cannot be recognized as standard GAG binding motifs. GAG-binding sites of other proteins range from single, linear motifs to higher order motifs, as proposed for antithrombin III [38]. Likewise, GAG-binding sites of other CCP modules have been found to depend on basic sequence clusters. For example, the CCP module SCR 7 of factor H has been suggested to contain a GAG-binding site com- posed of three basic sequence clusters separated in the linear sequence [40]. Furthermore, the crystal structure of the four CCP modules of vaccinia virus complement control protein reveals two putative heparin binding sites, each composed of several clusters of basic residues brought together in the three-dimensional structure [41].

To address our finding that PAPP-A2 is unable to bind to GAGs, a model of PAPP-A2 CCP3-CCP4, also based on the b2GPI template structure, was prepared (Fig. 9B). Striking features distinguish the model of PAPP-A2 from the PAPP-A model. First, in the same orientation, PAPP- A2 CCP3 does not display the basic, circular patch of PAPP-A, and in general displays much fewer basic residues. Second, the sequence motif found in cluster C of PAPP-A is incomplete in PAPP-A2, with one basic residue being replaced by an asparagine (Fig. 3A). Third, several acidic residues are scattered over the surface of CCP3 of PAPP-A2 (Fig. 9B). Of interest, three of these acidic CCP3 residues of PAPP-A2 are located in each of the clusters A, B, and C at positions occupied by basic residues in PAPP-A (Fig. 3); they may therefore be responsible for the lack of PAPP-A2 GAG binding. We thus analyzed a series of PAPP-A mutants, in which residues of these positions were replaced by the corresponding acidic residues of PAPP-A2. A greater loss in both heparin binding and cell surface adhesion was observed. This is illustrated by a comparison of mutant B1 and mutant Ac2, in which Lys1376 is substituted with alanine and glutamic acid, respectively: Mutant Ac2 eluted earlier in heparin affinity chromatography (0.45 vs. 0.51 M), and also showed weaker cell surface binding. Ac6, in which the three acidic residues are substituted together, showed

1534 K. Weyer et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

involvement

and proform of eosinophil major basic protein. Biochim. Biophys. Acta 1201, 415–423.

in IGF binding protein-4 proteolytic follicles: degradation and mRNA expression during follicular develop- ment. Endocrinology 142, 5243–5253.

22. Oxvig, C., Sand, O., Kristensen, T., Gleich, G.J. & Sottrup-Jensen, L. (1993) Circulating human pregnancy-associated plasma pro- tein-A is disulfide-bridged to the proform of eosinophil major basic protein. J. Biol. Chem. 268, 12243–12246.

23. Oxvig, C., Haaning, J., Hojrup, P. & Sottrup-Jensen, L. (1994) Location and nature of carbohydrate groups in proform of human major basic protein isolated from pregnancy serum. Biochem. Mol. Biol. Int. 33, 329–336.

24. Sun, I.Y., Overgaard, M.T., Oxvig, C. & Giudice, L.C. (2002) Pregnancy-associated plasma protein A proteolytic activity is associated with the human placental trophoblast cell membrane. J. Clin. Endocrinol. Metab. 87, 5235–5240.

9. Giudice, L.C., Conover, C.A., Bale, L., Faessen, G.H., Ilg, K., Sun, I., Imani, B., Suen, L.F., Irwin, J.C., Christiansen, M., Overgaard, M.T. & Oxvig, C. (2002) Identification and regulation of the IGFBP-4 protease and its physiological inhibitor in human trophoblasts and endometrial stroma: evidence for paracrine regulation of IGF-II bioavailability in the placental bed during human implantation. J. Clin. Endocrinol. Metab. 87, 2359–2366. 10. Hourvitz, A., Kuwahara, A., Hennebold, J.D., Tavares, A.B., Negishi, H., Lee, T.H., Erickson, G.F. & Adashi, E.Y. (2002) The regulated expression of the pregnancy-associated plasma protein- A in the rodent ovary: a proposed role in the development of dominant follicles and of corpora lutea. Endocrinology 143, 1833–1844.

25. Laursen, L.S., Overgaard, M.T., Weyer, K., Boldt, H.B., Ebbesen, P., Christiansen, M., Sottrup-Jensen, L., Giudice, L.C. & Oxvig, C. (2002) Cell surface targeting of pregnancy-associated plasma protein A proteolytic activity. Reversible adhesion is mediated by two neighboring short consensus repeats. J. Biol. Chem. 277, 47225–47234.

11. Bayes-Genis, A., Schwartz, R.S., Lewis, D.A., Overgaard, M.T., Christiansen, M., Oxvig, C., Ashai, K., Holmes, D.R. Jr & Con- over, C.A. (2001) Insulin-like growth factor binding protein-4 protease produced by smooth muscle cells increases in the cor- onary artery after angioplasty. Arterioscler. Thromb. Vasc. Biol. 21, 335–341.

12. Bayes-Genis, A., Conover, C.A., Overgaard, M.T., Bailey, K.R., Christiansen, M., Holmes, D.R. Jr, Virmani, R., Oxvig, C. & Schwartz, R.S. (2001) Pregnancy-associated plasma protein A as a marker of acute coronary syndromes. N. Engl. J. Med. 345, 1022–1029.

26. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K. & Pease, L.R. (1989) Site-directed mutagenesis by overlap extension using the polymerase chain reaction [see comments]. Gene 77, 51–59. 27. DuBridge, R.B., Tang, P., Hsia, H.C., Leong, P.M., Miller, J.H. & Calos, M.P. (1987) Analysis of mutation in human cells by using an Epstein–Barr virus shuttle system. Mol. Cell Biol. 7, 379–387. 28. Qin, Q.P., Christiansen, M., Oxvig, C., Pettersson, K., Sottrup- Jensen, L., Koch, C. & Norgaard-Pedersen, B. (1997) Double- monoclonal immunofluorometric assays for pregnancy-associated plasma protein A/proeosinophil major basic protein (PAPP-A/ proMBP) complex in first-trimester maternal serum screening for Down syndrome. Clin. Chem. 43, 2323–2332.

13. Monget, P., Mazerbourg, S., Delpuech, T., Maurel, M.C., Maniere, S., Zapf, J., Lalmanach, G., Oxvig, C. & Overgaard, M.T. (2003) Pregnancy-associated plasma protein-A is involved in insulin-like growth factor binding protein-2 (IGFBP-2) proteolytic degradation in bovine and porcine preovulatory follicles: identi- fication of cleavage site and characterization of IGFBP-2 degradation. Biol. Reprod. 68, 77–86.

29. Lee, C. & Irizarry, K. (2001) The GeneMine system for genome/ proteome annotation and collaborative data mining. IBM Syst. J. 40, 592–603.

30. Schwarzenbacher, R., Zeth, K., Diederichs, K., Gries, A., Kostner, G.M., Laggner, P. & Prassl, R. (1999) Crystal structure of human beta2-glycoprotein I: implications for phospholipid binding and the antiphospholipid syndrome. EMBO J. 18, 6228–6239.

14. Laursen, L.S., Overgaard, M.T., Søe, R., Boldt, H.B., Sottrup- Jensen, L., Giudice, L.C., Conover, C.A. & Oxvig, C. (2001) Pregnancy-associated plasma protein-A (PAPP-A) cleaves insulin- like growth factor binding protein (IGFBP)-5 independent of IGF: implications for the mechanism of IGFBP-4 proteolysis by PAPP-A. FEBS Lett. 504, 36–40.

31. Kishore, R., Samuel, M., Khan, M.Y., Hand, J., Frenz, D.A. & Newman, S.A. (1997) Interaction of the NH2-terminal domain of fibronectin with heparin: role of the omega-loops of the type I modules. J. Biol. Chem. 272, 17078–17085.

32. Kinosaki, M., Yamaguchi, K., Murakami, A., Ueda, M., Morinaga, T. & Higashio, K. (1998) Identification of heparin- binding stretches of a naturally occurring deleted variant of hepatocyte growth factor (dHGF). Biochim. Biophys. Acta 1384, 93–102.

15. Boldt, H.B., Overgaard, M.T., Laursen, L.S., Weyer, K., Sottrup- Jensen, L. & Oxvig, C. (2001) Mutational analysis of the pro- teolytic domain of pregnancy-associated plasma protein-A (PAPP-A): classification as a metzincin. Biochem. J. 358, 359–367. 16. Sto¨ cker, W., Grams, F., Baumann, U., Reinemer, P., Gomis- Ru¨ th, F.X., McKay, D.B. & Bode, W. (1995) The metzincins – topological and sequential relations between the astacins, adamalysins, serralysins, and matrixins (collagenases) define a superfamily of zinc-peptidases. Protein Sci. 4, 823–840.

17. Gomis-Ru¨ th, F. (2003) Structural aspects of the metzincin clan of

metalloendopeptidases. Mol. Biotechnol. 24, 157–202.

33. Delacoux, F., Fichard, A., Cogne, S., Garrone, R. & Ruggiero, F. (2000) Unraveling the amino acid sequence crucial for heparin binding to collagen V. J. Biol. Chem. 275, 29377–29382.

18. Overgaard, M.T., Boldt, H.B., Laursen, L.S., Sottrup-Jensen, L., Conover, C.A. & Oxvig, C. (2001) Pregnancy-associated plasma protein-A2 (PAPP-A2), a novel insulin-like growth factor-binding protein-5 proteinase. J. Biol. Chem. 276, 21849–21853.

34. Pucci, M., Fibbi, G., Magnelli, L. & Del Rosso, M. (2001) Reg- ulation of urokinase/urokinase receptor interaction by heparin- like glycosaminoglycans. J. Biol. Chem. 276, 4756–4765.

35. Campanella, G.S., Lee, E.M., Sun, J. & Luster, A.D. (2003) CXCR3 and heparin binding sites of the chemokine IP-10 (CXCL10). J. Biol. Chem. 278, 17066–17074.

19. Haaning, J., Oxvig, C., Overgaard, M.T., Ebbesen, P., Kristensen, T. & Sottrup-Jensen, L. (1996) Complete cDNA sequence of the preproform of human pregnancy-associated plasma protein-A: evidence for expression in the brain and induction by cAMP. Eur. J. Biochem. 237, 159–163.

20. Kirkitadze, M.D. & Barlow, P.N. (2001) Structure and flexibility the multiple domain proteins that regulate complement

36. Deprez, P., Inestrosa, N.C. & Krejci, E. (2003) Two different heparin-binding domains in the triple-helical domain of ColQ, the collagen tail subunit of synaptic acetylcholinesterase. J. Biol. Chem. 278, 23233–23242.

of activation. Immunol. Rev. 180, 146–161.

21. Oxvig, C., Sand, O., Kristensen, T., Kristensen, L. & Sottrup- Jensen, L. (1994) Isolation and characterization of circulating complex between human pregnancy-associated plasma protein-A

37. Cardin, A.D. & Weintraub, H.J. (1989) Molecular modeling of protein–glycosaminoglycan interactions. Arteriosclerosis 9, 21–32. 38. Hileman, R.E., Fromm, J.R., Weiler, J.M. & Linhardt, R.J. (1998) Glycosaminoglycan–protein interactions: definition of consensus

Glycosaminoglycan binding of PAPP-A (Eur. J. Biochem. 271) 1535

(cid:1) FEBS 2004

in glycosaminoglycan binding proteins. Bioessays 20,

sites 156–167.

41. Murthy, K.H., Smith, S.A., Ganesh, V.K., Judge, K.W., Mullin, N., Barlow, P.N., Ogata, C.M. & Kotwal, G.J. (2001) Crystal structure of a complement control protein that regulates both pathways of complement activation and binds heparan sulfate proteoglycans. Cell 104, 301–311.

39. Dong, J., Peters-Libeu, C.A., Weisgraber, K.H., Segelke, B.W., Rupp, B., Capila, I., Hernaiz, M.J., LeBrun, L.A. & Linhardt, R.J. (2001) Interaction of the N-terminal domain of apolipopro- tein E4 with heparin. Biochemistry 40, 2826–2834.

42. Overgaard, M.T., Sorensen, E.S., Stachowiak, D., Boldt, H.B., Kristensen, L., Sottrup-Jensen, L. & Oxvig, C. (2003) Complex of pregnancy-associated plasma protein-A and the proform of eosi- nophil major basic protein: disulfide structure and carbohydrate attachment. J. Biol. Chem. 278, 2106–2117.

40. Giannakis, E., Jokiranta, T.S., Male, D.A., Ranganathan, S., Ormsby, R.J., Fischetti, V.A., Mold, C. & Gordon, D.L. (2003) A common site within factor H SCR 7 responsible for binding heparin, C-reactive protein and streptococcal M protein. Eur. J. Immunol. 33, 962–969.