Collectin CL-P1 Utilizes C-Reactive Protein for Complement Activation

Background: C-reactive protein (CRP) is a plasma pentraxin family protein that is massively induced as part of the innate immune response to infection and tissue injury. CRP and other pentraxin proteins can activate a complement pathway through C1q, collectins, or on microbe surfaces. It has been found that a lectin-like oxidized LDL receptor 1 (LOX-1), which is an endothelial scavenger receptor (SR) having a C-type lectin-like domain, interacts with CRP to activate the complement pathway using C1q. However it remains elusive whether other lectins or SRs are involved in CRP-mediated complement activation and the downstream effect of the complement activation is also unknown. Methods: We prepared CHO/ldlA7 cells expressing collectin placenta-1 (CL-P1) and studied the interaction of CRP with cells. We further used ELISA for testing binding between proteins. We tested for C3 fragment deposition and terminal complement complex (TCC) formation on HEK293 cells expressing CL-P1. Results: Here, we demonstrated that CL-P1 bound CRP in a charge-dependent manner and the interaction of CRP with CL-P1 mediated a classical complement activation pathway through C1q and additionally drove an amplification pathway using properdin. However, CRP also recruits complement factor H (CFH) on CL-P1 expressing cell surfaces, to inhibit the formation of a terminal complement complex in normal complement serum conditions.


Introduction
C-reactive protein (CRP) is an acute-phase plasma protein produced by hepatocytes in response to inflammation, tissue damage and trauma (1). Like other acute phase proteins, CRP is normally present in trace levels (<10 mg/L) in serum but increases rapidly and dramatically in response to a variety of infectious or inflammatory conditions (2). Mild inflammation and viral infections cause elevation of CRP to 10-40 mg/L while bacterial infections produce levels of 40-200 mg/L. Levels higher than 200 mg/L are found in several bacterial infections and burns (2). CRP is capable of interacting with a variety of ligands such as phosphocholine residues, modified low density lipoprotein (mLDL), and damaged cells as well as activating the complement and opsonizing biological particles (3,4). The site directed mutagenesis model shows how one globular head group of C1q interacts through the central pore of CRP on the A face of the pentamer (5). The classical pathway is mostly activated in an antibody dependent manner, but it is also initiated by C1q directly through recognition of distinct structural moieties on microbial and apoptotic cells or through various soluble pattern recognition molecules, such as CRP and Pentraxin 3 (PTX3) (6,7).
Lectin-like oxidized LDL receptor 1 (LOX-1) was discovered as an endothelial scavenger receptor (SR), having a calcium dependent lectin-like protein (8). Recent papers demonstrated that CRP bound to Chinese hamster ovary cells (hLOX-1-CHO) express human LOX-1, and that CRP bind to recombinant 6 human LOX-1 in a cell-free system (9,10). These studies also showed that LOX-1 promotes CRP-induced xenogeneic complement activation by interacting with CRP to develop an inflammatory pathogenic response. Moreover, subsequent research has demonstrated that scavenger receptor type I (SR-AI) binds CRP whereas other SRs, CD36 or SR-B1, do not (11).
The ligand specificity of the SR family overlaps considerably (12,13). We previously identified collectin placenta 1 (CL-P1) (14), a type II membrane protein which contains carbohydrate recognition domain (CRD), a long coiled-coil domain, and a transmembrane domain and showed it to be a SR, in addition to its role as a collectin. It is also referred to as a scavenger receptor C-type lectin (SRCL) and collectin 12 (15). Recent genomic analysis verified CL-P1 as an SR-AI gene of SCARA4 as well as a collectin gene of COLEC12 (16, 17). CL-P1 was originally found and identified as an endothelial receptor that can endocytose and phagocytose Gram-negative and -positive bacteria and yeast as well as oxidized low density lipoprotein (OxLDL) in vascular endothelial cells, and it interacts with OxLDL and microbes through the collagen-like domain whereas it utilizes the CRD to bind sugar ligands (18,19). Furthermore, very recently Ma et al. have hypothesized the existence of a fluid phase molecule of CL-P1 which may initiate complement activation on Aspergillus fumigatus (20).
The aim of this study was to investigate the involvement of CL-P1 in CRP-mediated complement activation, as well as to investigate the downstream effects of this complement activation. In the present

Fluorescent labelling and biotinylation of CRP
CRP was fluorescently labeled with an Alexa Fluor 555 antibody-labeling kit (Invitrogen) and dialyzed 3 times (12 h, 4°C) against a 3000-fold volume of PBS. Biotinylation of CRP was performed using EZ-Link Sulfo-NHS-LC-LC-Biotin (Invitrogen) according to the manufacturer's instructions and dialyzed 3 times (12 h, 4°C) against a 1000 fold volume of PBS.

Binding and inhibition assay using the CHO/ldlA7 cell line
The binding of Alexa 555-CRP with full length CL-P1, CL-P1 deletion mutants and LOX-1 was performed as previously described (10). In brief, 24 h after transfection cells were washed twice with ice-cold HAM's F12 medium with 5% FBS. Then, the medium was replaced with ice-cold Ham's F-12/10 mM HEPES containing 10 µg/ml of Alexa 555-CRP and cells were incubated at 4 °C for 1 h.
After being washed with ice cold PBS, the cells were fixed with 4% phosphate-buffered paraformaldehyde (Wako Pure Chemical Industries). For the inhibition assay, 10 µg/ml poly(I) or poly(C) were pre-incubated with cells before the addition of Alexa 555-CRP. For the phosphocholine inhibition assay, 1 mM phosphocholine was mixed with Alexa 555-CRP and then incubated with cells for 1 h at 4 °C. The expression of CL-P1 and LOX-1 was visualized using anti-myc antibody followed by Alexa 488 anti-mouse IgG. Nuclear counterstaining was performed with Hoechst 33342 obtained from Invitrogen. The images were taken using a fluorescent microscope (BZ-9000, Keyence). Signal intensity was calculated by using the BZ-HIC program (Keyence). Lab.). For the analyses of the binding of heat-denatured biotin-CRP, CRP solution in PBS was heated in boiling water for 5 min before use. In some experiments a concentration (0-0.1 mM) of phosphocholine or 10 mM EDTA was added to the biotin-CRP in TBSTC and detected as described above.
Finally we determined peroxidase activity with SureBlue TMB microwell peroxidase substrate as described above.

Terminal complement complex (TCC) deposition ELISA
For the TCC deposition assay we used the same procedure as described in the complement activation ELISA except that we used 10% CFH depleted serum. For inhibition of TCC formation, CFH depleted serum was supplemented with purified CFH (400 µg/ml full serum). We detected TCC using rabbit polyclonal anti-human C5b-9 antibody.

Complement activation on cultured cells
We performed the complement activation assay on cultured cells as previously described (13). In brief, we transfected HEK293 cells seeded in collagen-coated dishes with pcDNA3.1 vector and CL-P1. After washing the cells with PBS, we added 10% human complement serum, C1q depleted serum, and properdin depleted serum in the presence or absence of 20 µg/ml CRP (full serum) in DMEM-high glucose medium. In some assays, C1q depleted serum or properdin depleted serum was replenished with native C1q (200 µg/ml full serum) or native properdin (6 µg/ml full serum). After incubation at 37 °C for 1 h, the cells were washed twice with PBS and fixed with 4% phosphate-buffered paraformaldehyde. We detected the deposition of C3 fragments on the cells by immunostaining with rabbit anti-human C3d antibody (3.5 µg/ml) in combination with Alexa Fluor 594 anti-rabbit IgG antibody. The nuclei of the cells were counterstained with Hoechst 33342. We performed quantitative analysis with the BZ-HIC program.

TCC formation assay
We performed the formation of the TCC on cell surfaces on cultured cells as previously described with few modifications (22). Briefly, HEK293 cells were plated on collagen-coated dishes, in DMEM supplemented with 10% FBS. The next day, cells were transfected with pcDNA3.1 vector and CL-P1. 24 h after transfection cells were washed three times with PBS followed by incubation with a final concentration of 10% human complement serum or CFH depleted serum in the presence or absence of 20 µg/ml of CRP (full serum) at 37 °C for 1 h. In the case of recovery experiments, CFH depleted serum was supplemented with (400 µg/ml full serum) purified human CFH. Cells were washed three times with PBS and fixed with 4% paraformaldehyde for 30 minutes. After three washes in PBS, cells were incubated for 30 minutes at room temperature with rabbit polyclonal anti-human C5b-9 and anti-myc monoclonal antibody and then for 30 minutes with goat Alexa Fluor 594 anti-rabbit IgG antibody and goat Alexa Fluor 488 anti-mouse IgG antibody. The nuclei were counterstained with Hoechst 33342. Cells were imaged using a fluorescence microscope.

CFH and properdin recruitment assay
HEK293 cells were plated on collagen-coated dishes, in growth medium. Cells were transfected with pcDNA3.1 vector and CL-P1. 24 h after transfection cells were incubated with a final concentration of 10% human complement serum in the presence or absence of 20 µg/ml of CRP (full serum) at 37 °C for 1 h. Cells were washed three times with PBS and fixed with 4% paraformaldehyde for 30 minutes. Cells were then incubated for 30 minutes at room temperature with goat polyclonal anti-human CFH or goat anti-human properdin with anti-myc monoclonal antibody and then for 30 minutes with goat Alexa Fluor 488 anti-mouse IgG antibody or donkey Alexa Fluor 594 anti-goat IgG antibody. We imaged the cells using a fluorescence microscope.
Quantitative measurement of SC5b-9 HEK293 cells transfected as described above were incubated with CFH depleted serum or CFH depleted serum supplemented with purified CFH (400 µg/ml full serum) in the presence or absence of 20 µg/ml of CRP (full serum). 1 h after incubation the medium was collected and the levels of SC5b-9 in the serum-containing medium were determined using a commercially available ELISA kit (Quidel Corporation).

Statistical Analysis
Statistical analysis was conducted using the unpaired two-tailed Student's t test included in the JMP statistics software package (version 7, SAS). Data are mean ±S.E. p<0.0001 is considered statistically significant.

CRP interacts with CL-P1 and LOX-1
It has been reported that LOX-1 binds CRP in cell experiments and ELISA (10). To investigate whether CL-P1 interacts with CRP, we inquired into the binding of Alexa Fluor 555-labeled CRP with CL-P1 and LOX-1 using transiently transfected CHO cells. As shown in Fig. 1a, Alexa 555-CRP binds CL-P1 and LOX-1 whereas no CRP binding was observed on pcDNA3.1 control vector transfected cells.
The CL-P1 and LOX-1 were co-localized with CRP on the cell surface.
We further characterized the CRP and CL-P1 interaction in the ELISA system. The presence of a soluble form of CL-P1 is still under debate. The soluble CL-P1 used in this study was prepared by forcing the extracellular domain to be a secreted protein and attaching an insulin leader peptide. We observed that biotin-CRP bound CL-P1 in a dose dependent manner (0-300 µg/ml) (Fig. 1b) and the heat denatured biotin-CRP lost its ability to bind CL-P1.

Charge dependent interaction of CRP with CL-P1
It is known that CRP interacts with phosphocholine and activates the classical complement pathway (23). The complement activation site in CRP is located in the A-face, whereas the phosphocholine binding site is in the B-face. We found that biotin-CRP and CL-P1 binding in ELISA was inhibited by phosphocholine in a dose dependent manner (0-0.1 mmol/L) (Fig. 2a). We demonstrated that incubation of a phosphocholine and Alexa 555-CRP mixture with CL-P1 transfected cells reduced the CRP binding to CL-P1 (Fig. 2b). Quantitative analysis indicates that 1 mM phosphocholine can completely inhibit the CRP binding (Fig. 2c). These results suggest that CL-P1 might interact with the B-face in CRP.
To check whether this binding of biotin-CRP and CL-P1 was calcium dependent, we performed an EDTA inhibition assay in ELISA which showed no significant decrease in biotin-CRP and CL-P1 binding (Fig. 2d). These results suggest that the interaction of CL-P1 with CRP might be calcium independent.
We further checked whether the interaction of CL-P1 with CRP is charge dependent. The pre-incubation of Poly(I) with the CL-P1 transfected cells before the addition of Alexa 555-CRP reduced CRP binding to CL-P1, whereas poly(C) did not reduce the binding (Fig. 2e, f). Therefore we concluded that the interaction of CL-P1 with CRP might be mainly charge dependent.

CL-P1 interacts with CRP mainly through the collagen-like and coiled-coil domain
We prepared several CL-P1 deletion mutants to find out the binding domain with different ligands as previously described (19) (Fig. 3a). We confirmed that MIF and western blotting analysis for the expression of full length and deletion mutants of CL-P1 showed almost similar expression levels and patterns on the cell surface except Δcc-col-CRD (19). We found that the collagen-like domain and coiled-coil domain were involved in CL-P1 and CRP interaction ( Fig. 3b and supplemental fig. 1). The deletion of CRD could not inhibit the interaction. There are three positively charged clusters in the collagen-like domain of human CL-P1 (19). To verify which positive cluster was involved in the CL-P1 and CRP interaction, we prepared the positively charged cluster mutants (Fig. 3c). We then performed the binding experiments of Alexa 555-CRP with CL-P1 positively charged cluster mutants. Our results showed cluster I and cluster III in the collagen-like domain of CL-P1 were involved in the interaction with CRP ( Fig. 3d and supplemental fig. 2). These results also show that the interaction between CRP and CL-P1 takes place in a charge dependent manner but does not show Ca 2+ dependent lectin activity, although the detailed mechanism is still unknown.

CRP and CL-P1 interaction activates the classical complement pathway
It is reported that the interaction of LOX-1 with CRP mediates complement activation (9). We tried to detect the interaction of CL-P1 with the CRP activating complement pathway. We examined the complement activation using CRP and recombinant human CL-P1 in an ELISA system. We found that there was complement activation only in the recombinant CL-P1-coated wells but boiled CRP failed to activate the complement system (Fig. 4a). Supplementation of polymyxin B did not affect complement activation (Fig. 4a). These data suggest that the complement activation observed therein is an effect of C-reactive protein in a CRP solution, but not caused by lipopolysaccharide.
Next, we determined the deposition of C3 fragments using CL-P1 expressing HEK293 cells from a human embryonic kidney cell line. The C3 fragments were deposited on CL-P1 in a CRP positive condition (Fig. 4b), although there were some non-specific deposition of C3 fragments. We observed that the degree of specific C3 fragment deposition was 78% on CL-P1 expressing cells and unspecific deposition was 22% in these images. Phase contrast images showed the edges of the cells and that the deposition of C3 fragments occurs all over the surface of CL-P1 expressing cells (Fig. 4b). The signal intensity of the deposition of C3 fragments was increased in CL-P1 expressed cells compared with those of control pcDNA3.1 when human complement serum was used (Fig. 4c). It is known that CRP can interact with the C1q (19). Next, we tested the possible involvement of C1q in the complement activation in our system. Our ELISA analysis demonstrated that the deposition of C3 fragments occurred in the presence of CRP and C1q (Fig. 5a).
We then analyzed the co-staining of the C3 fragments and C1q. Our results clearly demonstrate the co-staining of C3 fragments and C1q on the cell surface (Fig. 5b). Next, we focused in the deposition of C3 fragments using C1q depleted serum in our HEK293 cell system. In absence of C1q, complement activation was not detected regardless of the presence or absence of CRP (Fig. 5c, d). Supplementation of C1q recovered the deposition of the C3 fragments and it indicates the interaction of C1q and CRP in this system (Fig. 5c, d).

CRP induces the additional activation in the amplification pathway via CL-P1
Activation of the classical pathway inevitably initiates the alternative pathway. To determine whether CRP plays a role in classical pathway-triggered alternative pathway complement amplification, we used ELISA and found a significant decrease in the deposition of C3 fragments which was recovered when purified properdin was added back to the properdin depleted serum (Fig. 6a). We then confirmed the recruitment of properdin on the cell surface in the presence of CRP depending on CL-P1 (Fig. 6b). We next determined, to what degree classical pathway-triggered alternative pathway complement activation depends on properdin in our assay system. The deposition of C3 fragments was observed even using properdin depleted serum and the supplementation of purified properdin increased the C3 fragments deposition intensity (Fig. 6c, d). Purified properdin apparently has a propensity to form aggregates. So we took extreme care while using purified properdin. Commercially available properdin forms higher oligomers or aggregates upon repeated freezing and thawing were originally called "activated" properdin due to their ability to promote complement activation and consumption when added to serum. So, we avoided repeated freezing and thawing of purified properdin and used in experiments upon receipt to minimize aggregation of the properdin that can occur with prolonged storage. Thus, the interaction of CRP with CL-P1 basically activates the C1q dependent complement pathway and furthermore performs the additive amplification using the alternative pathway.

CRP negatively regulates TCC assembly on CL-P1 expressing cells by recruiting CFH
To determine the effect of CRP on downstream complement components, we examined the assembly of the TCC on CL-P1 expressing cells. As shown in Fig. 7a, we found no TCC formation using human complement serum. It is reported that CRP drives the classical pathway of the complement on nucleated cells without TCC formation or causing cytolysis (24). Other studies have shown that CRP bound to CFH (25,26), a complement regulatory protein that accelerates the decay of the C3 and C5 convertases and inhibits the assembly of the terminal complement components (27). Our results demonstrated the binding of CFH with CRP using human complement serum depended on CL-P1 (Fig. 7b). To determine whether CFH was required for the prevention of the TCC formation, we incubated CL-P1 expressing cells with CFH depleted serum with or without CRP. CFH depleted serum failed to prevent the TCC deposition (Fig.   7c). This inhibitory activity could be restored by the addition of CFH (Fig. 7c). Phase contrast images clearly demonstrated the pattern of TCC deposition on the surface of CL-P1 expressing cells (Fig. 7c).
We found that a similar phenomenon occurs in the ELISA system (Fig. 7e). Next, we analyzed the soluble TCC formation in the serum-containing culture medium 1 h after TCC formation reaction. The level of SC5b-9 significantly increased in the serum-containing medium after the reaction CL-P1 transfected cells including CRP compared with no CRP addition (Fig. 7d). Therefore, we concluded that the recruitment of CFH by CRP is required for the prevention of the TCC assembly on CL-P1 expressing cells (Fig. 8).

CRP is an acute phase protein involved in complement activation through the classical pathway (3).
Recently CRP has shown to bind LOX-1 and induce complement activation (9). However the mechanism by which CRP induces complement activation is poorly understood. Our study aimed to identify whether collectin CL-P1 can interact with CRP. Here, we demonstrated that CRP might be a novel ligand for CL-P1 and mediate some biological effects through complement activation.

Fujita et al. reported that LOX-1 interacted with CRP in a calcium dependent manner as EDTA
completely ended the interaction (9). In the case of CL-P1, EDTA was unable to inhibit the interaction of recombinant CL-P1 with CRP and this suggests that the interaction of CL-P1 and CRP might be calcium independent. It was proposed that LOX-1 interacted with the B-face of CRP since recombinant LOX-1 and CRP interaction was inhibited by phosphocholine (9). Two inhibition studies using phosphocholine in ELISA and cell experiments indicate that CL-P1 might also interact with the B-face of CRP.
Next, we focused on the inhibition effects by polycations or polyanions on the interaction of CRP with CL-P1. It was shown that polycations bound CRP on the B-face (28,29). Our microscope data showed the interaction was completely inhibited by the pre-incubation of polyanions but not by polycataions before the addition of CRP in CL-P1 expressed cells. This could be explained by the neutralization of a positive charge on CL-P1 by polyanions that inhibit the interaction with CRP. Lee et al.
has found phosphocholine can apparently bind cationic sites on CRP and the binding site does not overlap with the polycationic binding site (30). Therefore we hypothesize that CL-P1 might interact with the B-face of CRP in a charge dependent manner.
The site directed mutagenesis study has shown that the OxLDL and CRP binding site on LOX-1 is different but shares some common features because a carrageenan and anti-LOX-1 antibody were able to act as competitors to both (31). Recently, we have shown that the collagen-like domain of CL-P1 is responsible for the interaction with OxLDL and microbes using CL-P1 deletion mutants (19). Our results also suggest that the collagen-like domain and coiled-coil domain of CL-P1 is involved in CRP binding.
The study using mutants of the positively charged cluster in the collagen-like domain of CL-P1 revealed that the first and third positively charged cluster is involved in interaction with CRP. However, previous mutant experiments demonstrated that the second and third charged cluster was important for the binding of OxLDL and microbes (19). These results suggest that the third positively charged cluster of CL-P1 is an important domain for CRP and OxLDL binding.
CRP activates the complement system in human and mouse serum because CRP is known to bind C1q and activate pathways through its A-face (32). Here, we prepared a complement activation system using human complement serum and human cell line HEK-293 cells. Our Figs. 4 and 5 also demonstrate that the interaction of CRP with CL-P1 mediates complement activation depending on C1q as a classical pathway without antibodies. The deposition of C3 fragments using soluble CL-P1 is low in our ELISA analysis. This is could be due to the randomness of the orientation of the soluble CL-P1 in the ELISA plate.
Activation of the alternative pathway can occur secondary to classical pathway activation or be initiated independently. CRP has been shown to enhance the alternative pathway activation via the C3 convertase generation and function with it as a pre-antibody host defense mechanism (33). Properdin plays an important role in the activation of the alternative pathway not only by maintaining the stability of C3bBb, but also by providing a platform for the de novo C3 convertase assembly (34). Our results suggest that the interaction of CRP with CL-P1 on the cells first activates the classical pathway through C1q and next drives the alternative pathway as an amplification loop using properdin (Fig. 6). We believe activation. It has been found that CRP activates the classical pathway on nucleated cells without activating the TCC or causing cytolysis (24). CRP has been shown to bind apoptotic cells and protects the cells from assembly of the TCC by recruiting CFH (35). Our results also demonstrated that the complement activation by CRP and CL-P1 interaction could not form the TCC. Our data show that depletion of CFH allows TCC formation which was prevented when CFH was added back to the CFH depleted serum (Figs. 7), although the interaction of CL-P1 and CRP initiates complement activation. We suggest that, the complete CFH level in the serum might be able to suppress terminal complement activation even if acute inflammation or injury occurs with up-regulation of CRP. Thus, the interaction of elevated levels of CRP with CL-P1 could be disastrous in a CFH compromised condition. Furthermore, another study showed no enhanced CFH binding or any protective effect of CRP regarding C9 deposition (36). Why these observations differ from some other reports may relate to the quality of proteins and the semantics regarding the stages of cells.
The role of CRP in cardiovascular disease is controversial (37,38). For example, reports of atherosclerosis resulting in apolipoprotein E deficient mice overexpressing CRP were conflicting, with one study finding a positive relationship between CRP and cardiovascular disease (39), whereas other studies did not (40,41). Our study suggests that in CFH dysfunctional individuals, CRP might augment endothelial injury by activating the complement via CL-P1 to initiate cardiovascular disease.
Du Clos et al. have found that local production and activation of the complement has an important role in I/R injury and allograft rejection (42). By localizing at sites of tissue damage CRP has the potential to contribute to the complement activation at these sites, but CRP may also regulate this activation by its interaction with CFH depending on CL-P1.
In this paper, we found an important role of collectin CL-P1 in the limited activation of