Complement Factor H Is a Serum-binding Protein for Adrenomedullin, and the Resulting Complex Modulates the Bioactivities of Both Partners*

Adrenomedullin (AM) is an important regulatory peptide involved in both physiological and pathological states. We have previously demonstrated the existence of a specific AM-binding protein (AMBP-1) in human plasma. In the present study, we developed a nonradioactive ligand blotting assay, which, together with high pressure liquid chromatography/SDS-polyacrylamide gel electrophoresis purification techniques, allowed us to isolate AMBP-1 to homogeneity. The purified protein was identified as human complement factor H. We show that AM/factor H interaction interferes with the established methodology for quantification of circulating AM. Our data suggest that this routine procedure does not take into account the AM bound to its binding protein. In addition, we show that factor H affects AM in vitro functions. It enhances AM-mediated induction of cAMP in fibroblasts, augments the AM-mediated growth of a cancer cell line, and suppresses the bactericidal capability of AM on Escherichia coli. Reciprocally, AM influences the complement regulatory function of factor H by enhancing the cleavage of C3b via factor I. In summary, we report on a potentially new regulatory mechanism of AM biology, the influence of factor H on radioimmunoassay quantification of AM, and the possible involvement of AM as a regulator of the complement cascade.

Human adrenomedullin (AM) 1 is a 52-amino acid peptide originally isolated from a human pheochromocytoma and identified as a molecule capable of elevating rat platelet cAMP (1). AM belongs to the calcitonin gene peptide superfamily based on its slight homology with calcitonin gene-related peptide (CGRP) and amylin (1). The human mRNA is 1.6 kilobases long and encodes for a predicted 185-amino acid precursor from which two amidated peptides are generated: AM and a second peptide denoted as proadrenomedullin N-terminal 20 peptide (PAMP) (2). The expression of AM has been demonstrated in many tissues and biological fluids such as plasma (3), cerebrospinal fluid (4), sweat (5), amniotic fluid (6), urine (7), and milk (8). AM has been implicated in the modulation of numerous physiological functions, which include cardiovascular tone, central brain activity, bronchodilation, renal function, hormone secretion, cell growth, differentiation, and immune response (9).
Recently, we have demonstrated that plasma proteins from several species can specifically bind AM (10). The existence of these binding proteins was established using a radioligand blotting technique based on a method originally described for the detection of insulin-like growth factor-binding proteins (11). Most of the species analyzed, including humans, had an AM-binding protein (AMBP) with a M r of 120,000 under nonreducing conditions. Interestingly, the plasma from ruminant species (calf, goat, and sheep) had an additional band of M r 140,000. Whether these proteins are different or represent two glycosylation patterns from the same protein remains to be determined. An analysis of plasma from calves undergoing an acute phase response to a parasitic infection revealed a decrease in the expression of AMBP as compared with uninfected calves (10), whereas AM levels increased (12). The presence of a protein that specifically binds AM and the regulation of its expression in pathological situations may have a critical impact on AM physiology.
In this study, we isolate and characterize the human AMBP (AMBP-1) from plasma as complement factor H. We also describe how factor H may interfere with the quantification of AM by conventional radioimmunoassay (RIA) and how both binding partner proteins may modulate their respective biological activities.
with an equal volume of 0.1% alkali-treated casein (ATC) in Trisbuffered saline (TBS), pH 7.4, and extracted using reverse phase Sep-Pak C18 cartridges (Waters, Milford, MA) with 80% acidic isopropyl alcohol as the elution buffer. The extract was lyophilized and reconstituted in 1 ml of TBS containing 0.1% ATC, 0.1% Tween 20, and 0.05% Triton X-100, pH 7.4. Fluorescein-labeled AM was stored at 4°C for as long as 3 months without significant loss of activity.
Nonradioactive Ligand Blotting Assay-Proteins from human plasma (2 l) were electrophoretically fractionated on 3-8% Tris acetate gels (Novex, San Diego, CA) under nonreducing conditions and transferred to a 0.2-m nitrocellulose membrane. The membrane was washed with 1% Nonidet P-40, blocked with 0.1% ATC in TBS, and incubated with 50 nM fluorescein-labeled AM for 16 h at 4°C in blocking buffer containing 0.1% Tween 20. Binding was detected with a primary rabbit anti-fluorescein IgG (1:1000; Molecular Probes, Inc.), a secondary antibody coupled to alkaline phosphatase (1:2000; Dako, Carpinteria, CA), and nitro blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (Roche Molecular Biochemicals) as the color-substrate solution. For competition studies with unlabeled peptides, the membrane was preincubated with 5 M unlabeled ligands at 4°C for 6 h. Then labeled AM was added, and the membrane was incubated for 16 h at 4°C. AM fragments, PAMP, and CGRP were purchased from Phoenix Pharmaceuticals (Belmont, CA).
Reverse-phase HPLC-Preparative reverse phase HPLC was performed using a Delta Pak C18 300-Å column (30 mm ϫ 30 cm; Waters, Tokyo, Japan) and the "System Gold" modular system (Beckman Instruments Inc., Fullerton, CA). 2.5 ml of human plasma were mixed with an equal volume of 10% acetonitrile with 0.2% trifluoroacetic acid, processed through a 0.2-m filter, and loaded onto the column. After 5 min with 0.1% trifluoroacetic acid in 5% acetonitrile, the column was eluted with a linear gradient of acetonitrile containing 0.075% trifluoroacetic acid from 5 to 60% at a flow rate of 12 ml/min over 60 min. Each fraction (12 ml) was collected, freeze-dried, and dissolved in 0.3 ml of TBS, 0.1% Tween 20. Fractions were tested for the presence of AMBP-1 using the nonradioactive ligand blotting technique.
Amino Acid Analysis-Amino acid analysis was performed by The Protein/DNA Technology Center at the Rockefeller University, New York. HPLC (NovaPak C18 30-cm column) with the Waters PicoTag Work station and a two-pump gradient system (model 510) equipped with a model 490 UV multiwavelength detector were used as previously described (13).
Edman Degradation-The N-terminal sequence analysis was performed by the Biotechnology Resource Laboratory, Protein Sequencing and Peptide Synthesis Facility (Medical University of South Carolina, Charleston, SC). The sample was subjected to automated Edman degradation using a PE Biosystems Procise 494 Protein Sequencer and a PE Biosystems cLC Microblotter 173, using standard cycles and reagents (14, 15).
Mass Spectrometry-After fractionation by SDS-polyacrylamide gel electrophoresis under reducing conditions (5% ␤-mercaptoethanol), the gel was stained with Coomassie Blue, and the AMBP-1 band was excised. In-gel protein digestion and peptide extraction were performed as previously described (16). One-tenth of the extracted protein digest was analyzed by matrix-assisted laser desorption/ionization time-offlight mass spectrometry on a PerSeptive Voyager-DE STR (PE Biosystems, Foster City, CA) prior to liquid chromatography/mass spectrometry. The instrument was operated in reflector mode with the accelerating voltage set to 20,000, the laser energy to 2350, the guide wire voltage to 0.05%, and the grid voltage to 95%. The mirror ratio was set to 1:110. The remainder of the extracted protein digest was injected onto a 0.3 ϫ 100-mm, 5-m BetaBasic C18 column (Keystone Scientific, Bellafonte, PA), which had been equilibrated with 10% buffer B in buffer A (A: water with 0.1% formic acid; B: acetonitrile with 10% 1-propanol and 0.1% formic acid). Peptide elution was carried out using a linear gradient progressing from 10 to 60% buffer B over 60 min (Shimadzu Scientific Instruments LC10AD/VP pumps and LC10A controller). The eluting peptides were detected by a Finnigan LCQ mass spectrometer (ThermoQuest; Finnigan MAT Division, Piscataway, NJ). Peptide sequence data were obtained from the eluting peptides by MS/MS on those ions exceeding a preset threshold of 5 ϫ 10 4 ions. The operating parameters were as follows: sheath gas flow, ϭ 32, auxiliary gas flow ϭ 1, spray voltage ϭ 4.5 kV, capillary temperature ϭ 200°C, capillary voltage ϭ 8.0 V, and tube lens offset ϭ Ϫ20 V.
Western Blot-Proteins were electrophoretically fractionated on a 3-8% Tris acetate (for factor H) or a 4 -12% Bis-Tris gel (for AM) under nonreducing conditions, transferred to a 0.2-m nitrocellulose membrane, and blocked with 5% nonfat dry milk in phosphate-buffered saline (PBS). Afterward, the membrane was incubated with 1:2000 anti-factor H rabbit antibody (Quidel, San Diego, CA) or 1:4000 anti-AM rabbit antibody (17) and developed using the ECL Plus Western Blot Detection System (Amersham Pharmacia Biotech).
Binding Assay-A 96-well polyvinyl chloride plate was coated with factor H (5 ng/well; Sigma). The plate was blocked (TBS, 0.1% ATC, 0.1% Tween 20) and incubated with the unlabeled peptides for 2 h. Fluorescein-labeled AM (50 nM) was added, and after a 2-h incubation the assay was developed using an anti-fluorescein polyclonal antibody (1:1000, Molecular Probes, Inc.) and 125 I-Protein A (Amersham Pharmacia Biotech). The radioactivity was determined in a ␥-counter.
AM Immunoprecipitation-3 ml of sample were mixed with 1 ml of Protein A-agarose (Life Technologies, Inc.) containing a 1 M final concentration of each of the following protease inhibitors: pefabloc (Centerchem Inc., Stamford, CT), bestatin, and phosphoramidon (Sigma). After 1 h at 4°C, the sample was centrifuged, and the supernatant was divided and further incubated with 80 l of rabbit anti-AM (17) or rabbit preimmune serum for 1 h at 4°C. Protein A-agarose (80 l) was then added to the mix. After a 30-min incubation, the immunoprecipitate was collected by centrifugation, and the pellet was extensively washed with TBS, 0.1% Triton X-100. The final pellet was resuspended in 100 l of LDS sample buffer (Novex) and boiled before the Western blot analysis.
Extraction of Plasma and AM Radioimmunoassay-Extraction was performed using reverse-phase Sep-Pak C18 cartridges (Waters) as previously reported (18,19). Briefly, cartridges were activated with 80% methanol and washed with 0.9% NaCl. Plasma samples were mixed with an equal volume of PBS containing 0.1% ATC and 0.1% Triton X-100, pH 7.4. Samples were applied to the columns, and, after washing twice with 0.9% NaCl, AM was eluted with 80% isopropyl alcohol containing 125 mM HCl. Extracts were freeze-dried to remove the organic solvent. Concentrations of AM in the extracts were measured by radioimmunoassay as previously described (19).
cAMP Assay-Rat-2 fibroblasts were grown in RPMI 1640 containing 10% fetal bovine serum (Life Technologies). Cells were seeded into 24-well plates at 2 ϫ 10 4 cells/well and incubated 48 h at 37°C in 5% CO 2 . Before the assay, cells were incubated in TIS medium (RPMI 1640 plus 10 g/ml transferrin, 10 g/ml insulin, and 50 nM sodium selenite) for 15 min. Then cells were treated for 5 min with AM (Bachem, King of Prussia, PA) and/or factor H (Sigma) in 250 l of TIS medium containing 1% BSA, 1 mg/ml bacitracin, and 100 M isobutylmethylxanthine. The reaction was terminated by adding an equal volume of ice-cold ethanol. cAMP was measured using the Biotrac cAMP RIA (Amersham Pharmacia Biotech).
Receptor Binding Assay-Binding of rat 125 I-AM (Phoenix Pharmaceuticals) to Rat-2 cells was determined as previously reported (20). In brief, Rat-2 cells were plated out at 2 ϫ 10 4 cells/well in poly-D-lysinecoated 24-well plates (Becton Dickinson, Bedford, MA). Confluent cells were incubated for 60 min at 4°C in 0.5 ml of binding buffer (20 mM HEPES, pH 7.4, 5 mM MgCl 2 , 10 mM NaCl, 4 mM KCl, 1 mM EDTA, 1 M phosphoramidon, 0.25 mg/ml bacitracin, 0.3% bovine serum albumin) containing 100 pM rat 125 I-AM. After incubation, cells were washed twice with ice-cold binding buffer and then dissolved in 1 M NaOH for counting. Nonspecific binding was measured by incubating the cells with a 1000-fold excess of unlabeled AM.
Proliferation Assay-The breast cancer cell line T-47D (ATCC, Manassas, VA) was maintained in Dulbecco's modified Eagle's medium (Life Technologies) supplemented with 10% fetal bovine serum (Life Technologies). The MTT Proliferation Assay (Promega, Madison, WI) was carried out in serum-free conditions as previously reported (21). Briefly, cells were seeded in 96-well plates at 1-2 ϫ 10 4 cells/well, and appropriate concentrations of the indicated compounds were added. After 5 days of incubation at 37°C and 5% CO 2 in a humid incubator, the MTT colorimetric assay was carried out following the instructions from the manufacturer. The plate was read at a wavelength of 540 nm. Eight independent wells per treatment were averaged.
Antimicrobial Activity Assay-The antimicrobial activity was measured using Escherichia coli (ATCC 35218, Gaithersburg, MD) and a radial diffusion assay as previously described (22). Briefly, bacteria were incorporated into a thin underlay gel that contained 1% agarose, 2 mM HEPES, pH 7.2, and 0.3 mg/ml of trypticase soy broth powder. After polymerization, small wells of 10 l of capacity were carved in the agar. Test substances were added and allowed to diffuse for 3 h at 37°C. A 10-ml overlay gel composed of 1% agarose and 6% trypticase soy broth powder was poured on top of the previous gel, and the plates were incubated for 16 h at 37°C. The diameters of the inhibition halos were measured to the nearest 0.1 mm and, after subtracting the diameter of the well, were expressed in inhibition units (10 units ϭ 1 mm). We estimated the minimal inhibitory concentration (MIC) by performing a linear regression and determining the x intercepts.
Cofactor Activity of Factor H-C3b (28 pmol) was incubated with factor I (0.16 pmol) and factor H (0.16 pmol) in the presence or absence of AM and related peptides for 24 h at 37°C in a final volume of 50 l of PBS. Samples were analyzed by SDS-polyacrylamide gel electrophoresis using 4 -12% Tris-Bis gels (Novex) under reducing conditions and Coomassie Blue staining. C3b and factor I were purchased from Advanced Research Technologies (San Diego, CA).
Statistical Analysis-The MTT assay values and the MIC values were analyzed by the Student's t test. cAMP values were analyzed with a one-way analysis of variance and Tukey's test. p Ͻ 0.05 was considered significant.

Development of a Novel Nonradioactive Ligand Blotting
Assay for AMBP Detection-Using the radioligand blotting technique originally described by Hossenlopp et al. (11), we have previously demonstrated that human plasma contains at least one adrenomedullin-binding protein (AMBP-1) with M r 120,000 under nonreducing conditions (10). In the present study, we have developed a nonradioactive ligand blotting assay. Our initial approach included the labeling of AM with three different reporters: biotin, fluorescein, or dinitrophenyl. With all of these tracers, we were able to detect AMBP-1 in human plasma (data not shown). We discarded the use of the biotinylated AM reagent due to the possible interference with avidin-or biotin-like proteins present in plasma; this problem has already been described in the development of a nonradioactive ligand blotting assay for insulin-like growth factor-binding proteins (IGFBPs) using biotinylated insulin-like growth factor I (23). The procedure using fluorescein-labeled AM gave a better signal/noise ratio than the dinitrophenyl tag and was used for the evaluation of AMBP-1 expression. Fig. 1A compares the radioligand blotting using 125 I-AM or fluoresceinlabeled AM. One band with M r 120,000 was visualized in both systems; however, the nonradioactive technique resulted in sharper band definition. To demonstrate the specificity of the assay, fluorescein-labeled AM was incubated with 5 M AM, the gene-related peptide PAMP, and the structurally related peptide CGRP (Fig. 1B). Only AM was able to displace the nonradioactive tracer. When the competition was carried out with different fragments of AM, only the intact AM molecule reduced the binding (Fig. 1C).
Isolation and Characterization of AMBP-1-Human plasma (2.5 ml) was fractionated by reverse phase HPLC ( Fig. 2A). Each fraction was tested for the presence of AMBP-1 using the nonradioactive ligand blotting technique. AMBP-1 was revealed in fractions 48 -51 (Fig. 2C). Glycoprotein staining of AMBP-1 on SDS-polyacrylamide gels revealed that AMBP-1 was glycosylated (Fig. 3A). Isoelectric focusing showed a pI 6 for the protein (Fig. 3B). For the identification of purified AMBP-1, we used three different techniques: total amino acid analysis, Edman degradation, and mass spectrometry. Table I shows the amino acid composition of AMBP-1. A data base search revealed complement factor H as the protein with the highest degree of similarity to this composition profile. The amino acid composition of factor H is also shown in Table I. The percentage of methionine was lower than expected; however, the recovery of methionine in the case of bovine serum albumin, used as a control protein, was approximately half of the expected value. The analysis of the N-terminal amino acid sequence of AMBP-1 yielded a mix of at least two main Nterminal sequences; the data base search gave us again factor H as the protein with highest homology. The sequence of the 15 amino acids analyzed was identical to the N-terminal sequence of factor H with the exception of the threonine in position 12. Furthermore, a new search with the amino acids obtained from the segment peptide that did not correspond to the N terminus of factor H showed a 65% homology with an internal sequence of factor H (residues 578 -592), suggesting that this secondary sequence could correspond to factor H fragmentation. Finally, the peptide masses obtained by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry after tryptic digestion of AMBP-1 corresponded unequivocally to the tryptic digestion of factor H, with the detected peptides covering 21.8% of the total factor H sequence. MS/MS applied to the ion with a molecular mass of 1396.62 Da yielded the sequence, confirmed by both the b and the y series, of 10 out of the 12 amino acids from the fragment 737-748 in the factor H molecule.
To further confirm that AMBP-1 was in fact factor H, we analyzed the ability of anti-factor H antibody to recognize AMBP-1 by Western blot detection. AMBP-1 was immunoreactive for this antibody, giving a band of M r 120,000 under nonreducing conditions, an identical size to that of commercially available human factor H (Fig. 3C). Moreover, with nonradioactive ligand blotting using fluorescein-labeled AM, we demonstrated that factor H binds to AM (Fig. 3D). Under reducing conditions, AMBP-1 had a M r of 150,000 (Fig. 3A), which corresponds to the M r reported for factor H in such conditions (24). The reduction of the disulfur bonds prevented the binding of AMBP-1 to the fluorescein-labeled AM (Fig. 3E).
Finally, we carried out binding assays of fluorescein-labeled AM to immobilized factor H. Fig. 3F shows the competitive displacement with unlabeled AM that could not be achieved with either CGRP or PAMP.
Interference of Factor H with the AM Radioimmunoassay-We previously reported that the C18 reverse-phase separation technique used to prepare plasma for AM RIA analysis effectively eliminates AMBP-1 from the extract (10). In the present work, we confirmed this observation by analyzing the presence of factor H in plasma before and after the C18 extraction. When plasma is processed through the C18 column, factor H is obtained in the unbound portion of the sample and not in the fraction used for AM quantification (Fig. 4A). Based on this observation, we tested the possibility that a certain amount of AM may pass through the column bound to factor H. For that purpose, 1 ml of human plasma was processed through the column, and both the bound and unbound extracts were recovered. We immunoprecipitated AM from the extracts and deter-mined its presence by Western blotting. AM was detected in both the unbound and the bound fractions (Fig. 4B), suggesting that the traditional procedure used for peptide purification does not recover the total amount of AM present in plasma. Western blot after immunoprecipitation in the absence of extract did not yield any band, excluding the possibility that AM comes from the rabbit anti-AM serum (data not shown).
Disruption of AM/factor H interaction demonstrated a second source of AM in the plasma that was not routinely accounted for by traditional C18 purification procedures prior to RIA determination. A ligand blotting was performed with purified AMBP-1. After incubation with fluorescein-labeled AM, we tested several conditions for the dissociation of the binding between AM and factor H. For that purpose, the membrane was cut in strips and washed six times for 10 min each under the different conditions. Finally, the strips were equilibrated again in the assay buffer, and the ligand blotting was developed (Fig. 5A). Extreme conditions such as acidic pH and high salt concentration did not dissociate the interaction of factor H with AM. However, the incubation of the blot with labeled AM in those conditions effectively avoided the interaction (data not shown). One of the treatments that disrupted the binding was basic pH; however, further experiments indicated that this   effect could in fact correspond to an artifact, since this extreme pH alters the structure of the fluorescein and doing so negates its affinity for the anti-fluorescein antibody (data not shown). Using the binding assay in a 96-well plate, we further analyzed the dissociating ability of the chaotropic agent sodium thiocyanate (NaSCN). After incubation of solid-phased factor H (50 ng) with fluorescein-labeled AM (50 nM) and prior to the development of the assay, wells were incubated during different periods of time with PBS, 3 M NaSCN, pH 7.4 (Fig. 5B). The displacement curve suggests two dissociation events. By 15 min, the chaotropic salt had already displaced 50% of the binding; however, after that point the dissociation rate decreased.
Plasma samples of three healthy donors were quantified by RIA following either the method previously described (18,19) or the following modification. 1 ml of plasma was preincubated with an equal volume of 6 M NaSCN in PBS, 0.1% ATC, 0.1% Triton X-100, pH 7.4, for 10 min at room temperature. After that incubation, plasma was extracted through the C18 cartridges and quantified. The detected levels with the new protocol were 2-fold higher than those obtained with the standard technique (mean and S.D. values of the three donors were 23.0 Ϯ 4.8 versus 54.3 Ϯ 8.6 pg/ml). The same results were obtained when a longer preincubation with the chaotropic agent (16 h at 4°C) was used. Using the new protocol with NaSCN, recovery of unlabeled AM (200 pg) added to human plasma was 93.9 Ϯ 18.7% (n ϭ 3), whereas recovery of 125 I-AM was 82.7% Ϯ 4.4% (n ϭ 6). The dilution curve in the RIA obtained from a fourth plasma sample was parallel to that of the synthetic human AM used in the standard curve and to the curve generated with the same plasma extracted under the traditional conditions (Fig. 6). The parallel curves confirm that the increase in AM immunoreactivity generated by NaSCN treatment was not artifactual.
Effect of Factor H in AM-mediated cAMP Induction-AM was initially identified as a peptide capable of elevating cAMP (1). Recently, it has been reported that Rat-2 fibroblast cells express a specific AM receptor coupled to adenylate cyclase that produces a dose-dependent increase in cAMP upon exposure to AM (20). Using this cell line as a model, we studied the effect that factor H could have on AM-mediated cAMP response. Treatment of Rat-2 fibroblasts with 100 nM AM produced a 2-fold increase in cAMP (Fig. 7A). When the same concentration of AM was combined with increasing concentrations of factor H (50, 100 and 200 nM), we observed a significant and dose-dependent augmentation in cAMP production. The highest factor H concentration (200 nM) used alone had no effect on cAMP levels, confirming that the observed increase was in all cases due to the presence of AM (Fig. 7A). On the other hand, the presence of factor H did not apparently modify the kinetics of the binding between rat AM and its receptor. Neither the association rate nor the competitive binding of human AM was altered by the presence of factor H (Fig. 7, B and C).

Effect of Factor H in AM-induced T-47D
Growth-A human breast cancer cell line, T-47D, was used to investigate the effect of AM and factor H interaction on tumor cell growth. We have previously demonstrated that AM can act as an autocrine/ paracrine growth factor in several cancer cell lines (25). In the present work, we used the MTT proliferation assay to examine the growth-promoting activity of AM in the presence or absence of factor H. In a serum-free medium, AM had a growth-promoting activity on the cell line T-47D (data not shown). The presence of factor H further induced the proliferation of T-47D in a dose-dependent manner. Factor H in the absence of AM had no effect on growth (Fig. 8).
Modulation by Factor H of the Antimicrobial Activity of AM-A radial diffusion assay was used to characterize the influence of human complement factor H on the antimicrobial activity of AM (Fig. 9). Factor H by itself did not have any antimicrobial effect on E. coli. On the other hand, AM had an intense inhibitory impact on the bacterial growth. When AM and factor H were added together, a significant reduction in the inhibitory effect of AM was observed (p Ͻ 0.001, n ϭ 8), suggesting that factor H is able to hinder the antimicrobial activity of AM. The MIC for AM by itself was 18.4 Ϯ 1.3 g/ml, and it became 35.4 Ϯ 1.1 g/ml when factor H (50 g/ml) was added.
Modulation by AM of the Cofactor Activity of Factor H-We finally tested whether AM affects the cofactor activity of factor H in the factor I-mediated cleavage of C3b (26,27). Treatment of C3b with factor H and factor I caused cleavage of C3b (Fig.  10). When AM was added to the reaction, an increase in the cleavage of C3b was observed (Fig. 10A). Note that as the levels of AM are increased in the reaction mixture, there is a parallel increase in the split product formation with a reciprocal reduction in the 104-kDa band. AM (10 M) had no activity in the absence of factor H. Neither CGRP nor PAMP (10 M) had any effect on the cofactor activity (Fig. 10B). DISCUSSION We demonstrate here that AMBP-1 circulating in human plasma corresponds to complement factor H. The purification of AMBP-1 from plasma was greatly facilitated by a novel nonradioactive assay for the detection of AM binding proteins. This new technique gave us some distinct advantages over our previously described radioactive method (10); its development required a shorter period of time, and the use of the fluorescein-labeled AM had better reproducibility and sharper band formation than the use of 125 I-AM. In addition, it simplified handling procedures and extended the half-life of the tracer. Since both methods revealed a protein with the same mo-lecular weight and the binding with the labeled AM could only be totally displaced by the intact unlabeled AM, we conclude that the protein detected with the nonradioactive method corresponds to the protein previously detected with the radioactive ligand (10).
By a combination of HPLC, electrophoretic fractionation, and the nonradioactive detection system, we have been able to isolate AMBP-1 to homogeneity and complete its biochemical identification. Several different analytical techniques led to the unequivocal conclusion that the purified protein corresponds to complement factor H. Factor H is a single chain glycoprotein consisting of 20 subunits called short consensus repeats (28). Factor H binds to C3b, displacing Bb from the C3 convertase. It also acts as a cofactor for the factor I-mediated proteolytic cleavage of the ␣Ј chain of C3b. The final result of these activities is the inhibition of the alternative pathway of the complement (24,26,27). Additional roles have been identified for factor H; it binds to the integrin Mac-1 (C11b/CD18) enhancing the activation response of human neutrophils (29), is a ligand for L-selectin (30), induces the secretion of interleukin 1␤ by human monocytes (31), and acts as a chemotactic protein for monocytes (32). Finally, factor H binds to cell surface components of several pathogens (33)(34)(35)(36)(37). This binding apparently inhibits the activation of the complement, thus enhancing the pathogenicity of these microorganisms.
The discovery of the interaction between AM and factor H raises many questions about its biological implications. The presence of a binding protein can limit the transport of a peptide to the interstitial space and the access to its specific receptors; it can protect a peptide from metabolic clearance events, thereby prolonging its half-life in circulation; and it can modulate its biological activity. Here we show some of the consequences of the binding between AM and factor H.
One of the most immediate aspects was the interference of factor H with the routine AM quantification assay. The procedure for AM determination requires an extraction step to avoid interferences in the RIA. This process eliminates factor H from the extract to be analyzed, suggesting that factor H may be responsible for the interferences in the nonextracted samples. Although factor H is more hydrophobic than AM and therefore is better retained by the reverse-phase matrix, 2 we have demonstrated that in the established protocol for AM extraction, factor H is not retained by the Sep-Pak C18 cartridges. This fact could be related to the physical characteristics of the cartridges. Factor H is a molecule with a contour length of 495 Å and a cross-sectional diameter of 34 Å that folds on itself, reducing the length of the protein and increasing its width (38).
Factor H is not retained in the C18 matrix, probably because it is too big to penetrate through the particle pores (the pore size is 125 Å based on the manufacturer's specifications). The way factor H circulates through the column suggests that the AM bound to factor H will not be retained and therefore the extraction protocol would recover only the free AM in plasma. We have confirmed this by demonstrating the presence of a significant amount of AM in the unbound fraction after the extraction. Furthermore, treatment with a chaotropic agent (NaSCN) seems to dissociate, at least partially, the binding between factor H and AM, allowing the detection of higher levels of AM. Plasma levels of AM are elevated in several pathological conditions (39), and although AM seems to act in an autocrine/ paracrine manner, a physiological role for circulating AM remains possible (40). Therefore, we believe that determining the total AM concentration in plasma (versus the free AM currently measured) may be important to better understand the role of AM in the physiological and pathological conditions in which it is implicated. In addition, the variations in the levels of AMBP previously observed in infected animals (10) suggest that changes in circulating AM may be also dependent on modifications of its binding protein expression.
Factor H is present in plasma and has also been detected in extravascular compartments such as the synovial fluid (41,42). The liver is considered to be the main source of factor H, although it is also synthesized by extrahepatic cells such as mononuclear phagocytes, fibroblasts, endothelial cells, mesangial cells, astrocytes, oligodendrocytes, and neurons (43). This suggests that the presence of factor H in tissues may affect the autocrine/paracrine actions of AM. We describe here preliminary insights into the effect of factor H on AM activity. An increase in the cAMP induction mediated by AM was observed when Rat-2 fibroblasts were incubated with AM in the presence of factor H. On the other hand, factor H did not affect the binding between AM and its receptor. Factor H was also able to augment the growth-promoting activity of AM on the human breast cancer cell line T-47D. The exact mechanism by which the factor H-AM complex augments AM activity remains to be clarified; however some observations may shed some light on this issue. Factor H is able to bind to cell surfaces through at least three glycosaminoglycan binding sites present in its structure (44 -46). It has also been reported that factor H binds to human neutrophils through the integrin Mac-1 (C11b/CD18) (29), and it is a ligand for L-selectin (30). Hypothetically, the augmentation in cAMP production and cell growth may involve enhanced cell surface attachment, presenting AM in closer proximity to its membrane receptor. Factor H would act as a carrier and a reservoir of AM, which could provide high local levels of AM to stimulate its receptor. In this way, factor H would increase the AM effectiveness without modifying the affinity for its receptor. Other binding proteins enhance the biological activity of their ligands; the case of the well characterized IGFBPs is a good example. IGFBPs can either inhibit or augment the IGF actions (47). How IGFBPs enhance IGF function is not well understood, although it is known that many IGFBPs associate with cell surfaces and that the enhancing activity is probably mediated by this binding (47). A similar example of this phenomenon can be seen with the latent transforming growth factor-␤ (TGF-␤)-binding protein (LTBP), which seems to play an important role in the activation of latent TGF-␤, probably through targeting the latent TGF-␤ complex to the cell surface (48). The presence of an Arg-Gly-Asp (RGD) sequence may account for the cell attachment properties of some IGFBPs (47) as well as the LTBP (49). This sequence is present in several matrix proteins, and it is the essential structure recognized by the integrin superfamily of receptors (50). Interestingly, factor H also possesses an RGD cell adhesion sequence in its structure (28). Finally, it would also be interesting to determine whether the enhancing effect of factor H on Rat-2 cAMP production and on T-47D growth may be due to a protective effect of factor H on AM degradation.
We have also been able to demonstrate that factor H downregulates the antimicrobial activity of AM. It has been postulated that AM exerts its bactericidal effect by forming membrane pores, which ultimately cause pathogen lysis (51). Factor H inhibition of AM antimicrobial activity could be mediated by decreasing the concentration of available AM in the microenvironment, thus limiting its access to the pathogen's outer membrane. Although the physiological relevance of AM's antimicrobial activity still has to be addressed, the inhibition by factor H of this AM activity is in line with the fact that the binding of factor H by certain microorganisms seems to protect them from complement-mediated host defense (33). It has been suggested that this resistance could be due to the degradation of C3b by the membrane-bound factor H. The inactivation by factor H of the host-produced AM, a molecule with antimicrobial activity, could now be considered as an additional mechanism of microorganism resistance.
Another important aspect of this study is the identification of a role for AM in C3b degradation mediated by factor H/factor I interaction. We have demonstrated that AM accelerates this process. We can speculate that AM may induce conformational changes in the structure of factor H, increasing its affinity for C3b, similarly to what has been reported to occur in the presence of soluble polyanions (52). Since plasma AM is increased during sepsis and after endotoxin challenge (12,(53)(54)(55)(56), AM induction of factor H/factor I-mediated C3b degradation may have important implications in the immune response associated with these processes.
Although we have addressed some of the consequences of the factor H/AM interaction, many important questions remain. Do factor H and AM affect the production of their binding partner? How does factor H affect the biological actions of AM in vivo? Does factor H prolong the plasma half-life of AM? In regard to this last question, it is interesting to note that the recovery of endogenous AM from plasma is not affected by several freezethaw cycles; however, recovery of exogenous AM is markedly reduced after a single cycle (18). The authors of this work suggested that a protein may be present in circulation that could bind AM aiding in its stability in plasma. It would not be surprising if that protein was factor H.
In summary, we present here the interaction between factor H and AM and some of the resulting consequences it may have on the activities of both protein partners. We have shown the influence that factor H/AM binding has on established AM quantitative techniques traditionally used in the field. These observations challenge our understanding of the real significance of free AM in biological fluids. Factor H has been shown to affect AM biology in a diametric manner, depending on the experimental system examined. The exact mechanisms by which these actions occur remain to be determined. In addition, AM can modulate factor H activity during degradation of C3b, thereby implicating this peptide hormone as a new regulatory component of the complement cascade. Thus, our initial observation on the existence of an AM plasma binding protein has now opened diverse avenues for future studies in both AM and factor H biology.