Structural Basis for Inflammation-driven Shedding of CD163 Ectodomain and Tumor Necrosis Factor-α in Macrophages^*

Introduction

ADAM17,² alias TNF-α-converting enzyme, belongs to the “a disintegrin and metalloproteinase (ADAM)” family of transmembrane proteases. It has a multi-substrate specificity, and it is involved in several cellular processes such as cytokine, hormone, and growth factor release (1). ADAM17-dependent shedding is also involved in regulated intramembrane proteolysis releasing cytoplasmic mediators and in inhibition of extracellular signaling and ligand uptake by a rapid down-regulation of surface receptors. Because of the many different substrates of ADAM17, the protein has a key role in both homeostatic and pathologic processes (2). ADAM17 is widely known for its activation of macrophage TNF-α by cleavage of membrane-bound proTNF-α (3, 4), and as a consequence, it is designated TNF-α-converting cleavage enzyme. TNF-α is one of the most potent proinflammatory cytokines known, which indicates an indirect proinflammatory role of ADAM17 in inflammatory macrophages (2, 5). proTNF-α, which is a homotrimeric type II transmembrane protein (N terminus in the cytoplasmic tail), is cleaved by ADAM17 between Ala⁷⁶ and Val⁷⁷ (⁷⁶A↓V) in the juxtamembrane region (3, 4, 6).

In accordance with promotion of the proinflammatory state in macrophages, ADAM17 was also recently shown to cleave and thereby down-regulate the surface expression of the hemoglobin scavenger receptor CD163 (7). This receptor is suggested to serve an anti-inflammatory function by clearance of the extracellular haptoglobin-hemoglobin complexes and the heme oxygenase-mediated conversion of proinflammatory heme to anti-inflammatory heme metabolites (8, 9). The ADAM17-mediated cleavage leads to generation of soluble CD163 (sCD163), a byproduct that is highly up-regulated in plasma during infectious and inflammatory conditions (7, 10). CD163 is a type I transmembrane protein with an opposite orientation as compared with proTNF-α. The ADAM17-dependent cleavage of the CD163 ectodomain occurs in the juxtamembrane region in an as yet unidentified cleavage site and results in shedding of the sCD163 ectodomain, consisting of nine SRCR domains (11, 12).

In contrast to TNF-α, which is almost absent in normal plasma, sCD163 is present in a rather high concentration (1–4 mg/liter) (12). Upon proinflammatory stimulation the plasma level of both proteins is rapidly elevated. This was recently demonstrated in a human study of experimental endotoxemia where a bolus injection of LPS caused a fast and simultaneous increase of TNF-α and sCD163 (7). Both proteins peaked after 1.5 h, but although TNF-α was no longer detectable in circulation after 3 h, the LPS-induced increase in sCD163 persisted for more than 24 h. In line with this an elevated level of sCD163 is reported in patients suffering from various infectious and inflammatory diseases such as sepsis, tuberculosis, diabetes, human immunodeficiency virus, rheumatoid arthritis, and hemophagocytosis (13).

The physiological role, if any, of sCD163 is so far unknown, but it has been speculated that release of sCD163 and subsequent formation of an sCD163-haptoglobin-hemoglobin complex may suppress the heme iron supply to hemolytic bacteria and trypanosomes (14, 15). Moreover sCD163 has been suggested to inhibit activated T lymphocyte proliferation and recently also to promote recognition and phagocytosis of Staphylococcus aureus (16, 17).

CD163 is expressed in all mammalian species, and the human protein displays more than 71% amino acid similarity to its reported orthologs including chimpanzee, green monkey, swine, dog, cow, rat, and mouse CD163. Despite the very high amino acid similarity, the mechanism of haptoglobin-mediated hemoglobin scavenging differs substantially between mice (18) and humans (19). In humans, efficient uptake of hemoglobin via the CD163 system requires preformation of the haptoglobin-hemoglobin complex, whereas mouse CD163 efficiently mediates uptake of hemoglobin independent of haptoglobin-hemoglobin complex formation.

To define the molecular basis for ADAM17-mediated cleavage of CD163, we have now made a comparative mutagenesis analysis of proTNF-α and CD163 cleavage in mice and humans. This led to identification of a common motif for ADAM17-mediated cleavage of the two protein humans and the surprising finding that inflammation-driven down-regulation of CD163 by ADAM17-mediated cleavage seems specific for primate species.

EXPERIMENTAL PROCEDURES

RESULTS

A progressive protein alignment algorithm was used to compare the juxtamembrane sequences of human CD163 and human proTNF-α (Fig. 1A); this identified a palindromic sequence consisting of ⁷⁸Arg-Ser-Ser-Ser-Arg situated next to the ADAM17 cleavage site in proTNF-α (⁷⁶A↓V) and a similar palindrome, ¹⁰⁴⁴Arg-Ser-Ser-Arg, in CD163. A cross-species protein alignment revealed that the Arg-Ser-Ser-Ser-Arg sequence is rather well conserved in the juxtamembrane region of proTNF-α (Fig. 1B), whereas the Arg-Ser-Ser-Arg sequence is only present in human and monkey CD163 (Fig. 1C). The Arg-Ser-(Ser)-Ser-Arg sequences both in proTNF-α (all species) and in human and monkey CD163 are preceded by a small hydrophobic residue, which in TNF-α represents the final N-terminal residue. A distance-based phylogenetic tree based on the cross-species alignment of the juxtamembrane region of CD163 (Fig. 1C) showed a high sequence identity within groups of primates, domestic animals, and rodents.

The importance of the Arg-Ser-Ser-Arg sequence in PMA-induced cleavage of hCD163 was analyzed by comparing the PMA-induced shedding of CD163 in HEK293 cells expressing either wild type hCD163 or the hCD163 ¹⁰⁴⁴RS/AA and ¹⁰⁴⁶SR/AA mutants (Fig. 2, A and C). The mutants were expressed at the cell surface at comparable levels to wild type hCD163. When we incubated HEK293 cells expressing hCD163 with PMA, the level in the medium increased severalfold to ∼9 μg/ml (Fig. 2C). In contrast, both the constitutive soluble CD163 release and the PMA-induced release from HEK293 cells expressing hCD163 ¹⁰⁴⁴RS/AA were greatly reduced. Release of sCD163 was also inhibited in HEK293 cells expressing the hCD163 ¹⁰⁴⁶SR/AA mutant, although to a lower extent. This indicates that the entire Arg-Ser-Ser-Arg motif is essential for efficient cleavage, albeit the initial Arg-Ser dipeptide seems particularly important. Inhibition of PMA-induced shedding is also observed when introducing the RS/AA mutation in proTNF-α (proTNF-α ⁷⁸RS/AA) (Fig. 2, B and D). Incubation of HEK293 cells expressing proTNF-α with PMA resulted in highly increased TNF-α levels. However, despite similar surface expression of proTNF-α ⁷⁸RS/AA and wild type proTNF-α, no increase of TNF-α is measured in the medium when incubating HEK293 cells expressing proTNF-α ⁷⁸RS/AA with PMA (Fig. 2D). Interestingly, introduction of the RS/AA mutation inhibits both the constitutive and the PMA-induced shedding in CD163, whereas only the PMA-induced shedding is affected when the RS/AA mutation is introduced in proTNF-α (Fig. 2, B and D).

To establish whether CD163 is shed in mice, the sCD163 level in mouse serum was estimated by an ELISA specific for mouse CD163 (Fig. 3). The concentration of sCD163 in naive mice was ∼300 ng/ml, whereas no sCD163 was detected in Cd163^−/− mice (Fig. 3A). To analyze the effect of LPS on sCD163 serum concentration in mice, an experimental endotoxemia study was performed (Fig. 3B). Mice received an LPS bolus (2.5 mg/kg) intravenously in the tail vein, and serum concentrations of both sCD163 and TNF-α were monitored over a 240-min time period. As expected, a considerable increase in the TNF-α level was observed after 60 min followed by a fast clearance with no detectable TNF-α after 240 min. In contrast, no change in serum concentration of sCD163 was observed at any of the analyzed time points. To investigate the involvement of ADAM17 in shedding of mouse CD163, the experimental endotoxemia study was repeated in hypomorphic ADAM17^ex/ex mice, which exhibit over 95% reduction in ADAM17 expression (26). Challenging ADAM17^wt/wt and ADAM17^ex/ex mice with LPS resulted in a strong release of TNF-α to the circulation in ADAM17^wt/wt mice, whereas no detectable levels of TNF-α were measurable in ADAM^ex/ex mice (Fig. 3C), highlighting that proinflammatory shedding of TNF-α solely depends on ADAM17 activation. Opposed to this, comparable levels of sCD163 were detected in plasma of both ADAM17^wt/wt and ADAM17^ex/ex mice after LPS challenge (Fig. 3D).

Analysis of LPS-induced shedding of CD163 and TNF-α was repeated in vitro using MBDMs. MBDMs were maturated by M-CSF, which led to a substantial CD163 expression as verified by image cytometric analysis (Fig. 4A). When we incubated MBDMs with LPS, a large increase in TNF-α was detected in the culture medium, whereas no increase was observed in the level of sCD163 (Fig. 4B).

To further validate the involvement of the juxtamembrane region in shedding of human CD163, juxtamembrane chimeras were prepared and expressed in HEK293 cells (Fig. 5A). The chimeras consisted of either mouse CD163 with the human juxtamembrane region (mCD163hjux) or human CD163 with the mouse juxtamembrane region (hCD163mjux). The chimeras were expressed at the cell surface at either comparable levels (hCD163mjux) or lower levels (mCD163hjux) as compared with the wild type protein (Fig. 5B , panels a and b). Fig. 5, C and D, show an immunoblot analysis of CD163 shedding after induction with various PMA concentrations. When we exposed HEK293 cells expressing hCD163 to increasing concentrations of PMA, a dose-dependent release of sCD163 was observed (Fig. 5C). In contrast, only a very low and dose-independent release of sCD163 observed when incubating HEK293 cells expressing hCD163mjux with increasing PMA concentrations. However, in HEK293 cells expressing either mCD163 or mCD163hjux, only mCD163hjux-expressing cells displayed a PMA dose-dependent release of sCD163 (Fig. 5D). Interestingly, preincubation of the transfected cells with TIMP3, an ADAM17 inhibitor, inhibited the PMA-induced release of hCD163 and mCD163hjux (Fig. 5, E and F) The effect of hCD163 juxtamembrane sequence on shedding of mouse CD163 was further investigated using chimeras (mCD163chim1–4) consisting of mouse CD163 substituted with varying length of human CD163 (Fig. 6A). PMA-induced and TIMP3-inhibited release of sCD163 was only observed in chimeras containing the human CD163 juxtamembrane region (Fig. 6, C and D). This also included mCD163chim2, which showed a lower CD163 surface expression as compared with wild type mCD163 and mCD163 chimeras (Fig. 6, B and C).

Having established a positive effect of the human CD163 juxtamembrane region on shedding of mouse CD163, the importance of the Arg-Ser-Ser-Arg sequence was analyzed by introducing this specific motif next to the conserved Thr-Gly dipeptide in the juxtamembrane region of mouse CD163 (Fig. 7A). Introduction of the Arg-Ser-Ser-Arg sequence in mouse CD163 did not change the surface expression of CD163 (Fig. 7B). Instead, the 4-amino acid insert induced a PMA-stimulated and TIMP3-inhibited release of mouse sCD163 (Fig. 7, C and D). When we transfected mCD163 RSSR-expressing HEK293 cells with siRNA against ADAM17, the mRNA level of ADAM17 was reduced to ∼60% of controls (Fig. 7E), which resulted in an inhibition of the PMA-induced release of sCD163 (Fig. 7F).

DISCUSSION

In the present study, we have demonstrated that a palindromic sequence (Arg-Ser-(Ser)-Ser-Arg) present in both human CD163 and TNF-α is necessary for an efficient ADAM17-mediated cleavage. Mutation of the palindromic sequence in human CD163 affected both the induced and the constitutive release of sCD163, which corresponds with earlier findings suggesting that ADAM17 functions as both a constitutive and an inducible sheddase of CD163 (7). Cross-species sequence alignment of the juxtamembrane region in CD163 revealed that species other than primates completely lacked the Arg-Ser-(Ser)-Ser-Arg sequence. In line with this, we show that despite the existence of sCD163 in mouse serum, the protein is not released to the circulation via ADAM17-mediated cleavage. Importantly, we were able to verify the importance of the Arg-Ser-Ser-Arg sequence by enabling ADAM17-mediated shedding in mouse CD163 using chimeras of mouse CD163 including the human juxtamembrane sequence or the Arg-Ser-Ser-Arg sequence.

So far, more than 75 different proteins (2) have been identified as substrates for ADAM17. Despite the vast number of proteins, no consensus sequence for ADAM17-mediated proteolysis has been identified, and therefore it seems obvious that ADAM17 may cleave in rather variant sequences. Screening of peptide libraries has shown a prevalence of small hydrophobic residues at the substrate P1′ position, basic residues at P2′, and small aliphatic residues at P3′, although with weaker selectivity (27). Recently, it was also reported that the ADAM17 activity relied on the secondary structure of the substrate and other noncatalytic domains in the enzyme (28). The present findings comply with the P1′-P3′ consensus rule in the sense that the Arg-Ser-(Ser)-Ser-Arg sequence in both proTNF-α and CD163 is preceded by a small hydrophobic residue (P1′ position). Arg and Ser in the tetrapeptide then represent the basic and small aliphatic residue at P2′ and P3′, respectively. As such, It is possible that the a small hydrophobic residue followed by Arg-Ser-(Ser)-Ser-Arg represents an optimized sequence for ADAM17-mediated cleavage. However, it should be noted that exact Arg-Ser-(Ser)-Ser-Arg motifs are absent from most proteins known to be cleaved by ADAM17, and the motif is therefore not a prerequisite for ADAM17-mediated cleavage.

N-terminal sequencing has identified Val-Arg-Ser-Ser-Ser-Arg as the N terminus of TNF-α, whereas the precise cleavage site of CD163 is unknown. However, the P1′-P3′ consensus rule and the present data including the knock-in data in mouse CD163 indicate that cleavage may occur between Thr¹⁰⁴² and Gly¹⁰⁴³. (Fig. 8). However, because CD163 is a type I transmembrane protein, which is cleaved in the C terminus, the ¹⁰⁴³Gly-Arg-Ser-Ser-Arg motif will not, as in the case of proTNF-α, represent the N terminus of the released soluble product. Instead, this sequence will constitute the N-terminal end of the CD163 receptor stub in the membrane.

ProTNF-α is a type II transmembrane protein, whereas CD163 is a type I transmembrane protein, and as such, they have opposite orientation. The identification of a similar palindromic sequence important for ADAM17-mediated cleavage in opposite oriented substrates could suggest a parallel to palindromic sequences known from DNA restriction enzymes. However, palindromic peptide sequences are not identical structures in both directions due to the unidirectional orientation of the carboxyl and amino groups involved in the peptide bonds. Consequently, both proTNF-α and CD163 have to present the Arg-Ser-(Ser)-Ser-Arg sequence to the membrane-associated ADAM17 in the same orientation. Flexibility of the juxtamembrane regions in CD163 or proTNF-α may allow the right orientation for recognition by ADAM17. Alternatively, high flexibility of the extracellular domain of ADAM17 could also allow recognition of the motif in either orientation.

In the present study, we have shown that mouse CD163 lacks the ability to undergo ADAM17-dependent shedding, although sCD163 is easily detectable in mouse serum. So far, it is not known whether mouse sCD163 is released via regulated proteolysis or merely as a byproduct of macrophage turnover. Recently, it was shown that another human ADAM17 substrate, the interleukin-6 receptor, was not shed by this enzyme in mice (29). Instead, the substrate was shed by the closely related ADAM10, which only had minimal activity toward the substrate in humans. One could also speculate that CD163 in mouse serum is an uncleaved receptor present in macrophage-released exosomes, as has been shown for the structurally related protein CD163-L1 in humans (30). Given the relatively high level of sCD163 in mouse and human plasma, it is possible that a common basal level of exosome-associated CD163 exists in both species and that humans under inflammatory conditions are able to increase the concentration of the receptor in plasma by ADAM17-dependent release of sCD163.

Interestingly, the striking difference in human and mouse CD163 in terms of ADAM17-mediated cleavage parallels a number of other differences in the function of primate haptoglobins versus the haptoglobins in non-primates. One example is the primate-specific defense mechanism against trypanosome parasites. Here, the primate-specific haptoglobin gene duplication product, haptoglobin-related protein, binds together with another primate-specific protein, apolipoprotein-L1, to exert a trypanolytic function (15). Another example is our recent finding that human haptoglobin upon complex formation with hemoglobin has a much higher affinity for CD163 than mouse haptoglobin in complex with hemoglobin (18). In return, mouse free hemoglobin binds with higher affinity to CD163 as compared with human hemoglobin. Apparently, the difference in affinity between haptoglobin-hemoglobin complexes in the two species relies on a primate-specific presence of two basic residues in a haptoglobin loop that interacts with CD163 when haptoglobin is complexed to hemoglobin (31). A third example, the plasma concentration of haptoglobin, is generally much higher in primates versus in non-primates (32, 33).

The serious threat of trypanosome parasites causing sleeping sickness most likely may have driven the evolution of resistance toward certain types of trypanosome parasites by means of primate-specific haptoglobin-related protein and apolipoprotein-L. It is tempting to speculate that other infectious diseases such as malaria or other hemolytic infections may also have caused an evolutionary pressure leading to development of additional protection by haptoglobin. However, the increased protection obtained by higher concentrations of haptoglobin and a faster CD163-mediated clearance of haptoglobin-hemoglobin complexes may lead to an inexpedient overdamping of the inflammatory response induced by anti-inflammatory heme metabolites. Hence, an ADAM17-dependent mechanism to down-regulate CD163 expression in primates may have evolved to tune the system and thereby achieve the right balance between protection from hemolysis and overdamping of the proinflammatory response.