CD163 Binding to Haptoglobin-Hemoglobin Complexes Involves a Dual-point Electrostatic Receptor-Ligand Pairing*

Background: CD163 mediates endocytosis of haptoglobin-hemoglobin complexes formed upon intravascular hemolysis. Results: Acidic CD163 residues and basic haptoglobin residues mediate receptor-ligand binding in humans. Conclusion: A two-point electrostatic pairing between Ca2+-coordinated acidic clusters in receptors and basic ligand residues explains high-affinity CD163-(haptoglobin-hemoglobin) binding. Significance: The data add further evidence that Ca2+-dependent two-point electrostatic pairing is a common mechanism for coupling ligands to endocytic receptors. Formation of the haptoglobin (Hp)-hemoglobin (Hb) complex in human plasma leads to a high affinity recognition by the endocytic macrophage receptor CD163. A fast segregation of Hp-Hb from CD163 occurs at endosomal conditions (pH <6.5). The ligand binding site of CD163 has previously been shown to involve the scavenger receptor cysteine-rich (SRCR) domain 3. This domain and the adjacent SRCR domain 2 of CD163 contain a consensus motif for a calcium-coordinated acidic amino acid triad cluster as originally identified in the SRCR domain of the scavenger receptor MARCO. Here we show that site-directed mutagenesis in each of these acidic triads of SRCR domains 2 and 3 abrogates the high affinity binding of recombinant CD163 to Hp-Hb. In the ligand, Hp Arg-252 and Lys-262, both present in a previously identified CD163 binding loop of Hp, were revealed as essential residues for the high affinity receptor binding. These findings are in accordance with pairing of the calcium-coordinated acidic clusters in SRCR domains 2 and 3 with the two basic Arg/Lys residues in the Hp loop. Such a two-point electrostatic pairing is mechanistically similar to the pH-sensitive pairings disclosed in crystal structures of ligands in complex with tandem LDL receptor repeats or tandem CUB domains in other endocytic receptors.

In humans, the macrophage protein CD163 acts as the scavenger receptor for hemoglobin (Hb) 3 leaking into plasma during intravascular hemolysis (1). In the first step of this scavenging process, the released Hb is captured by haptoglobin (Hp), an acute-phase protein patrolling the bloodstream in search for free Hb, which it binds almost irreversibly (2,3). The binding of Hp to Hb not only protects against the toxicity of Hb and its heme group (3)(4)(5), it also triggers the second step of the scavenging process, i.e. recognition of Hp-Hb by CD163 and subsequent clearance of the entire complex by receptor-mediated endocytosis (1). This Hp-CD163-dependent clearance system protects against the potentially hazardous Hb, which, if left in the circulation, may cause oxidative tissue damage and inflammation (5,6). Inside the macrophage, the proinflammatory heme is converted into metabolites with antiinflammatory effect (for review, see Refs. [7][8][9]. In accordance with an overall antiinflammatory role, CD163 is up-regulated in antiinflammatory subtypes of macrophages (7).
Structurally, CD163 belongs to the scavenger receptor cysteine-rich (SRCR) family of proteins characterized by the presence of SRCR domains in the extracellular region (10). The SRCR domain is a conserved protein fold consisting of 100 -110 residues. In CD163 nine such consecutive SRCR domains are contained within the extracellular region, followed by a transmembrane segment and a short cytoplasmic tail (11). Hp is also a multidomain protein with a complement control protein (CCP domain) and a serine protease (SP) domain (12,13) as building blocks. A single disulfide bond connects a CCP subunit with an SP subunit to form the basic (CCP-SP) unit of Hp (14). In its simplest form (Hp1-1), corresponding to the version found in most mammals, Hp has a dimeric structure where two (CCP-SP) units are connected by a ␤-strand swap and a disulfide bond between the CCP subunits (3,14). Humans have two allelic forms of Hp (Hp1 and 2) owing to a duplication of the CCP region involved in dimerization. Individuals with the Hp2 gene in one or both of the Hp loci express multimeric Hp variants (Hp2-1 and Hp2-2) that also bind to CD163 with high affinity (1,(15)(16)(17). In addition to these variant forms of Hp, humans and other old world primates express an Hp-related protein (Hpr), which is 91% identical to Hp1 (15,17). Hpr binds Hb, but unlike Hp-Hb, Hpr-Hb is not recognized by CD163 (18).
The high affinity interaction between CD163 and Hp-Hb is Ca 2ϩ -dependent and critically relies on the amino-terminal third of the extracellular SRCR region in CD163 (1,19). Using a panel of truncated recombinant CD163 variants, we have previously shown that SRCR domains 1-5 bind Hp-Hb with an affinity similar to the entire extracellular region of CD163 (19). In the ligand, comparative binding analyses with human Hp, Hpr, and Hp/Hpr hybrids have shown that the so-called loop 3 of the Hp SP domain plays an essential function in CD163 recognition of Hp-Hb (20). The recently published crystal structure of porcine Hp-Hb reveals that this loop protrudes from the distal part of the Hp-Hb complex and small angle x-ray scattering measurements of human Hp-Hb in complex with recombinant CD163 SRCR domains 1-5 support the approximate zones of contact between the ligand and receptor (3). In contrast to the complex of Hp-Hb, Hb and Hp alone exhibit weak or no binding to CD163 in humans (1,20,21).
Mapping of the CD163-(Hp-Hb) interaction surfaces is so far restricted to the human system, but a recent study in mice has uncovered subtle and major evolutionary differences in the Hb scavenging mechanism. First, the binding of Hb alone to CD163 is of significantly higher affinity than the corresponding interaction in humans; and second, Hp fails to elicit high affinity receptor binding upon complex formation with Hb in the mouse system (22).
Ligand binding by the SRCR domain of the scavenger receptor MARCO also depends on Ca 2ϩ , and this domain has been shown to harbor a metal ion binding site composed of three negatively charged residues that are involved in ligand binding (23). A corresponding motif for Ca 2ϩ -coordinated clustering of acidic residues is present in SRCR domains 2, 3, 7, and 9 of CD163. Based on a series of binding experiments with mutated variants of both CD163 and Hp, we here present data pinpointing the contact region between CD163 and its high affinity ligand Hp-Hb and propose a common model for Ca 2ϩ -dependent coupling and uncoupling of ligand.
Cell Lines-Flp-In HEK293 cells (Invitrogen) were cultured in Dulbecco's modified Eagle's medium (Cambrex Bioscience, Verviers, Verviers, Belgium) supplemented with 10% fetal calf serum, 2 mM glutamine, and 100 g/ml Zeocin (Invitrogen). Transfection with Hp mutant cDNA was performed using FuGENE 6 transfection reagent (Roche Diagnostics), and stable transfectants were selected with 150 g/ml Hygromycin B (Invitrogen). Flp-In CHO cells (Invitrogen) were maintained in serum-free HyClone medium for CHO cells (Thermo Scientific). Stably transfected Flp-In CHO cells expressing CD163 variants were established by means of FuGENE 6 and subsequent selection with 750 g/ml Hygromycin B (Invitrogen). Hp and CD163 expression products were visualized by subjecting growth medium and cell lysate to SDS-PAGE and subsequent Western blotting using a rabbit polyclonal anti-human Hp antibody (DAKOCytomation, Glostrup, Denmark) and a rabbit polyclonal anti-CD163 antibody (1), respectively.
Purification of Recombinant Proteins-We purified Hp and Hp mutants from serum-free HEK293 cell culture medium (Invitrogen) by Hb (Sigma) affinity chromatography as detailed previously (20). Purification of human CD163 SRCR1-5 and mutated versions was performed by subjecting harvested cell culture medium containing secreted expression products to antibody affinity chromatography. The affinity column was generated by coupling 5 mg of the humanized anti-human CD163 antibody KN2/NRY (a gift from Cytoguide Aps, Aarhus, Denmark) to 1 ml of HiTrap NHS-activated HP (GE Healthcare) according to manufacturer's instructions. The harvested culture medium was concentrated and filtered before application on the column. After allowing the expression products to bind to the immobilized antibody, the column was washed with phosphate-buffered saline, pH 7.4, before elution with a solution containing 50 mM acetate and 500 mM NaCl, pH 5.0. Secondary elution was performed with 100 mM glycine HCl, 500 mM NaCl, pH 4.0. The collected fractions were neutralized by a Tris buffer, pH 8.0, and the proteins were stabilized by addition of a protease inhibitor mixture (Roche Applied Science). Fractions were dialyzed against phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 8.1 mM Na 2 HPO 4 , and 1.75 mM KH 2 PO 4 ), pH 7.4, at 4°C and analyzed by SDS-PAGE followed by silver staining.
Pulldown Assay-Bovine serum albumin (BSA) or complexes of human Hp (mixed phenotypes, Sigma) and human Hb A 0 (Sigma) were coupled to CNBr-activated Sepharose 4B beads (Amersham Biosciences) according to the manufacturer's instructions. To test for interaction with CD163 variants, beads were washed twice in MB buffer (2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM Hepes, and 140 mM NaCl, pH 7.8) prior to incubation with growth medium harvested from CD163-transfected Flp-In CHO cells. Incubation was performed overnight at 4°C, and the following morning beads were washed six times in MB buffer. Bound CD163 variant proteins were eluted by a lithium dodecyl sulfate-containing sample buffer and visualized by Western blotting using rabbit polyclonal anti-CD163 antibody (1).
Surface Plasmon Resonance (SPR) Analyses-The interaction between CD163 and complexes of Hp (Hp1-1 purified from human plasma (Sigma) or purified recombinant Hp variants) and Hb A 0 (Sigma) was studied by SPR analysis on a Biacore 3000 instrument, essentially as described (1, 18 -20). Biacore sensor chips type CM5 were activated with a 1:1 mixture of 0.2 M N-ethyl-NЈ-(3-dimethylaminopropyl)carbodiimide and 0.05 M N-hydroxysuccimide in water according to instructions by the manufacturer. CD163 purified from human spleen (1) or purified truncated recombinant CD163 SRCR1-5 variants were immobilized in 10 mM sodium acetate, pH 4.0, and the remaining binding sites were blocked with 1 M ethanolamine, pH 8.5. A control flow cell was made by performing the activation and blocking procedure only. Samples were dissolved in 10 mM Hepes, 150 mM NaCl, 2.0 mM CaCl 2 , 1.0 mM EGTA, and 0.005% Tween 20, pH 7.4. Sample and running buffers were identical. Regeneration of the sensor chip after each analysis cycle was performed with 10 mM glycine, pH 4.0, containing 20 mM EDTA and 500 mM NaCl. Fig. 1 shows an alignment of the MARCO SRCR domain and the nine SRCR domains of CD163 demonstrating that the metal binding residues of MARCO are conserved in CD163 SRCR domains 2, 3, 7, and 9.

Acidic Triads in CD163 SRCR Domains 2 and 3 Are Essential for Hp-Hb Binding-
To investigate whether the acidic triads in CD163 SRCR domains 2 and 3 play a role in Hp-Hb binding, we expressed recombinant CD163 SRCR1-5 mutated in the candidate metalbinding residues of SRCR domains 2 and 3 ( Fig. 2A). Expression of a variant carrying mutations in the acidic triad of SRCR domain 2 (D185A,D186A,E252A) revealed a stable protein but absent Hp-Hb binding as measured by Hp-Hb-mediated pulldown analysis (data not shown). Mutation of the entire acid triad of SRCR domain 3 rendered the protein unstable, but sin-gle (E359A) and double mutations (D292A,D293A) in this acidic triad completely abrogated Hp-Hb-mediated pulldown ( Fig. 2B and data not shown). In line with these results, simultaneous mutation of one or two residues in each of the acidic triads of SRCR domains 2 and 3 (SRCR1-5 E252A,E359A and SRCR1-5 D185A,D186A,D292A,D293A) led to stable expression products but inactivity in terms of Hp-Hb binding (Fig.  2B).
To study the Hp-Hb binding properties in a more sensitive assay, three of the mutants (SRCR1-5 D185A,D186A,E252A; SRCR1-5 D292A,D293A; and SRCR1-5 E359A) were purified by antibody affinity chromatography (Fig. 2C) for use in SPR experiments. The resulting data showed that the SRCR domain 2 mutant protein (D185, D186A,E252A) has a significant reduction in affinity compared with the WT counterpart, and that effects of the single (E359A) and double mutations (D292A, D293A) in SRCR domain 3 are equally dramatic (Fig. 2D). Collectively, these binding analyses disclosed an essential role for the acidic triads of CD163 SRCR domains 2 and 3 in Hp-Hb recognition.
Two Positively Charged Residues in Hp Loop 3 Are Crucial for the CD163-(Hp-Hb) Interaction-In this part of the study we tested the hypothesis that basic residues in the ligand pair with Ca 2ϩ -coordinated acidic triads in CD163, an electrostatic pairing mechanism, that has previously been proposed to be a common theme of Ca 2ϩ -dependent ligand-receptor interactions (24). In this model, Ca 2ϩ plays an indirect role in ligand binding by positioning two or three negatively charged residues from the acidic triad for direct interaction with one positively charged residue from the ligand. In compliance with Lys/Arg playing an essential role in the CD163-(Hp-Hb) interaction, the presence of free Lys or free Arg efficiently inhibited Hp-Hb binding to CD163 in SPR binding experiments (data not shown). In contrast, free His was severalfold less efficient in inhibiting Hp-Hb binding to CD163.
Basic residues in the CD163 binding region in Hp loop 3 were subsequently subjected to mutagenesis to identify potential Arg/Lys residues involved in receptor binding. A strong candidate was Lys-262 because a previous comparison of Hp and Hpr has shown that simultaneous substitution with Hpr-specific residues in positions 261, 262, and 264 in Hp loop 3 completely abrogates binding of Hp-Hb to CD163 ((20) and Fig. 3, A and  C). Analysis of single mutants (Hp E261A; Hp K262A; and Hp T264) (Fig. 3B) in SPR binding experiments revealed that the K262A substitution alone accounted for the eliminated binding of the Hp-Hb complex to CD163 (Fig. 3C). For comparison, the triple mutant with mutations corresponding to Hpr-specific residues (Hp E261K,K262W,T264A (Fig. 3A)) and a triple mutant with Ala substitutions (Hp E261A,K262A,T264A) were also included in this analysis (Fig. 3, B and C).
To hunt down a second positively charged Hp residue with a crucial role in receptor interaction, we again turned to the previous mutational analysis of Hp/Hpr hybrids which had furthermore revealed that the triple substitution of Hp residues Asp-248, Gln-249, and Arg-252 into the corresponding residues of Hpr (Fig. 3A) resulted in inhibition of receptor binding (20). These residues are positioned in an ␣-helix located in the loop 3 region. Mutation of Arg-252 on its own gave rise to a significant reduction in receptor affinity (Fig. 3,  B and C).
We also investigated the effect of substitution of surfaceexposed Hb residues that are positioned closely to the Hp loop 3 in the recently solved crystal structure of the Hp-Hb complex (3). Neither of the following Hb mutations affected the interaction between Hp-Hb and CD163: K61Q(␣Hb), K96Q(␤Hb), K121H(␤Hb), R41A(␤Hb) (data not shown). In summary, these mutational studies identified Arg-252 and Lys-262 in Hp loop 3 as essential residues in the CD163-(Hp-Hb) interaction. The three-dimensional model in Fig. 3D illustrates the position of these two positively charged residues within the Hp molecule.
Alignment of available Hp sequences revealed that Lys-262 is highly conserved whereas Arg-252 only is present in primate Hp (Fig. 4). Some primates such as Macaca fascicularis have a Lys corresponding to Arg-252 and an Arg corresponding to Lys-262. Hp from this primate species showed similar induction of high affinity binding of human Hb to human CD163 in the SPR binding assay (data not shown).

DISCUSSION
The present mutagenesis study now reveals the essential receptor and ligand residues involved in high affinity binding of the Hp-Hb complex to CD163 in humans. In CD163, the identified ligand contact region comprises two triplets of acidic residues in SRCR domains 2 and 3, respectively, both of which conform to the calcium-and ligand-binding motif described in MARCO (23). In the CD163 ligand, the two basic residues Arg-252 and Lys-262 residing in loop 3 of Hp are crucial for binding to CD163. As further described below, we suggest that electrostatic interactions between the Ca 2ϩ -coordinated acidic residues in CD163 and the basic residues in Hp loop 3 are crucial for the high affinity contact between ligand and receptor. This also corresponds with the well described calcium dependence Cell culture medium harvested from cell lines expressing various CD163 variants was subjected to pulldown analysis using Sepharose beads coupled with Hp-Hb or BSA. Bound CD163 SRCR1-5 variants were visualized by immunoblotting using a polyclonal antibody against CD163. In parallel reactions harvested cell culture medium was incubated in the absence of ligand-bound beads. The amount loaded from the latter reactions represents 3% of the amount that was incubated with Hp-Hb beads or BSA beads in the binding reactions. C, silver-stained SDS-polyacrylamide gel of the purified recombinant CD163 SRCR1-5 variants. Positions of molecular size standard markers are indicated. D, SPR analysis of Hp-Hb binding to immobilized CD163 SRCR1-5 recombinant variants. Purified Hp derived from human plasma, purified human Hb, or the complex of Hp with Hb was analyzed for binding to purified recombinant CD163 variants immobilized on a Biacore sensor chip. The CD163 variant analyzed is indicated above each panel.
of the ligand-receptor interaction (19) and probably explains why Hp-Hb segregates from CD163 after internalization into endosomes where the calcium concentration has been measured to be less than 10 M at pH 6.5 (25). In addition, the acidic environment of the endosomes may also contribute to a weakening of the proposed receptor-ligand electrostatic pairings (19).
The data presented here do not disclose exactly how the two basic residues in Hp combine with the two acidic clusters in CD163 SRCR domains 2 and 3, but it seems likely that the basic residues each combines specifically with one of the two receptor domains. Such mechanism of dual-point interaction by two basic residues with two Ca 2ϩ -coordinated acidic clusters has previously been disclosed in crystal structures of two other receptor-ligand complexes, that between receptor-associated protein and the LDL receptor (26) and that between the intrinsic factor-vitamin B 12 complex and the receptor cubilin (24). The complement-type/LDL receptor-type A repeats and CUB domains constitute the ligand binding domains in these two different receptors. Fig. 5 is a schematic representation of this dual-point interaction that now seems to be a common mechanism of three rather different receptor-ligand structures. Whereas the electrostatic nature of the binding may be essential for the coupling at physiological Ca 2ϩ concentrations and neutral pH as well as the uncoupling at low Ca 2ϩ concentrations and acidic pH, the presence of two contact sites may increase the functional affinity (avidity) of the ligand. In the case of the CD163-(Hp-Hb) system, the binding of Hp-Hb to CD163 embedded in the cell membrane may be further stabilized by ligand-mediated CD163 cross-linking because of the dimeric or multimeric nature of Hp (1,27). The ability of Hp-Hb to interact with two (dimeric Hp-Hb) or more (multimeric Hp-Hb) CD163 receptors simultaneously has recently been confirmed by small angle x-ray scattering measurements of human Hp-Hb in complex with recombinant CD163 SRCR domains 1-5 (3) and by SPR binding experiments, 4 respectively.
In addition to the electrostatic ligand-receptor interactions proposed in the present study, other structural features of Hp-Hb including the loop 3 conformation and parts of Hb may be essential for the ligand-receptor fit. So far it is unknown whether the Hp-complexed Hb contributes directly to CD163 binding. Free Hb has a low affinity for CD163 (20,21) but is unable to compete out binding of Hp-Hb to CD163 in our hands, 4 suggesting that the CD163 region involved in binding free Hb may be different from the one binding Hp-Hb. Neither is it known how Hb complex formation with Hp promotes CD163 binding. Hb may induce a conformational change in Hp allowing its high affinity interaction with CD163. Another possibility is that Hb participates in binding through a direct interaction with CD163. Solving the crystal structure of free Hp and Hp-Hb in complex with the ligand binding part of CD163 may provide the final insights into CD163 recognition of Hp-Hb.   (20).
Interestingly, Hp Arg-252 seems specific for primates, and this coincides with the observation that human and monkey Hp, when complexed with Hb, elicits a pronounced increase in CD163 affinity. Other striking Hp and CD163 features specific for primates or for humans are: (i) a high Hp level in plasma, (ii) human-specific Hp multimerization induced by the Hp2 gene product, (iii) evolution of the Hpr protein (for review, see Refs. 8,28), and (iv) duplication of the CD163 gene giving rise to the gene encoding the macrophage surface protein CD163-L1 (29). Hpr is known to play a major role in the innate defense against the Trypanosoma brucei brucei parasites causing sleeping sickness. It is tempting to speculate that the other genetic expan-sions also rely on the evolvement of novel defense mechanisms against infectious diseases that have been a major threat to survival during evolution of the primates.
As presented in the alignment in Fig. 1, the Ca 2ϩ -coordinated ligand-binding residues of MARCO are conserved not only in CD163 SRCR domains 2 and 3, but also in SRCR domains 7 and 9. Based on previous binding studies with truncated CD163, SRCR domains 7 and 9 are not part of the Hp-Hb binding regions (19). However, it is possible that these domains bind free Hb, as well as other ligands. Multiligand preference is a general feature of macrophage scavenger receptors, and other ligands such as the cytokine TWEAK, erythroblasts, as well as some bacteria and viruses have been reported to bind to CD163 (reviewed in detail in Ref. 30). Compared with the binding of Hp-Hb to CD163, these other reported ligand-CD163 interac-

A B C
Hp-Hb tions are less characterized, but in the case of both erythroblasts and bacteria it has been suggested that ligand binding to CD163 involves an 11-residue motif positioned just amino-terminally to the two Asp residues of the acidic triad in CD163 SRCR domain 2 (31,32). This SRCR domain may thus recognize several distinct ligands, but the mechanism of interaction between CD163 SRCR domain 2 and Hp-Hb is fundamentally different from the one between CD163 SRCR domain 2 and erythroblasts (and probably also bacteria) as the latter interaction is independent of Ca 2ϩ (31). In view of the present report suggesting that the high affinity binding of Hp-Hb complexes to CD163 is a relatively late event in evolution, it is tempting to speculate that the receptor may have other not yet identified ligands that are removed by macrophages in the process of inflammation. Furthermore, the consensus sequence for the Ca 2ϩ -coordinated acidic cluster is conserved in more than 50% of all human SRCR domains (Table  1), including SR-AI (MSR1) and the close CD163-relative CD163-L1 (29,33). Ligands that bind to these putative Ca 2ϩ sites await identification.