C1q/TNF-Related Protein 6 Is a Pattern Recognition Molecule That Recruits Collectin-11 from the Complement System to Ligands

Adipose tissue is no longer considered as a passive energy storage but has emerged as an active endocrine organ involved in metabolic regulation and secretion of innate immune factors (1–3). Furthermore, overlapping pathways that regulate both metabolism and immunity have recently been identified (4), with adiponectin as the prototypic example of such cross-talk. Adiponectin increases insulin sensitivity through receptor-dependent mechanisms in metabolic tissues and thereby improves whole-body energy homeostasis (5–7). At the same time, it provides protection against several pathological events in various tissues by suppressing cell death and reducing inflammation (8–12), highlighted by low levels of adiponectin in both obesity and asthma, two conditions associated with chronic inflammation (12–14). Adiponectin belongs to the C1q/TNF-related protein (CTRP) family, named after the homology with the pattern recognition molecule (PRM) C1q of the classical complement system and the inflammatory cytokine TNF-α. The family consists of 15 members apart from adiponectin (15). One of these paralogues, CTRP6, has been shown to have the inverse relationship to the development of type 2 diabetes compared with adiponectin. Serum levels of adiponectin are markedly reduced in obese patients and are correlated with insulin insensitivity (16), whereas the opposite is observed for CTRP6 (17–19). This is supported by mouse studies in which adiponectin-deficient mice also show increased expression of CTRP6 (15). The inverse relationship between CTRP6 and adiponectin, and the observation that CTRP6 interacts with one of the adiponectin receptors (20, 21), suggests that CTRP6 may have a counterbalancing effect on adiponectin. This raises the question of whether the increased levels of CTRP6 in serum directly contributes to the state of low-grade inflammation seen in obesity and type 2 diabetes.

The structure of the CTRP family is strikingly similar to the PRMs of the complement system, the humoral arm of the innate immune response. Like the PRMs (C1q, mannose-binding lectin, collectin-10 and -11, as well as the three ficolins) of the classical and lectin pathways of complement, the CTRP family has a cysteine rich N-terminal V region, a collagen-like domain with Gly-X-Y triplets, and a C-terminal globular domain. The CTRP family has been shown to form trimers and higher-order oligomers by cross-linking cysteine residues in the V region forming a characteristic bouquet-like structure (15). Adiponectin has been shown to bind factor H, regulating the alternative complement pathway (9), and to recruit C1q from the classical pathway (10). Also, CTRP6 has been shown to interact with the complement system by regulating the formation of the alternative pathway C3 convertase (22).

We hypothesized that CTRP6 might indeed function as a PRM based on the similarities in structure and the interaction with the complement system. In this study, we present evidence that CTRP6 recognize different sugar molecules and acetylated ligands, as well as live enteroaggregative Escherichia coli (EAEC) and Pseudomonas aeruginosa bacteria. Moreover, we can show that CTRP6 interacts specifically with collectin-11 (CL-11) and activates the complement cascade through recruitment of CL-11/MASP-2 complexes and downstream cleavage and deposition of C4.

The following buffers were used: PBS (0.2 M Na₂HPO₄, 35 mM K₂HPO₄, 137 mM NaCl, 15 mM KCl), PBS-T (PBS, 0.5% v/v Tween 20), TBS-T (10 mM Tris, 150 mM NaCl, 0.05% Tween 20, 2.5 mM CaCl₂, ± 10 mM EDTA), wash buffer (PBS, 300 mM NaCl), and elution buffer (PBS, 300 mM NaCl, 300 mM Imidazole).

Recombinant CTRP6 was produced in Flp-In Chinese hamster ovary (CHO) cells (Invitrogen) using the CTRP6 coding sequence (NM_031910) cloned into the pcDNA5/FRT vector as described elsewhere (23). A C-terminal hexa-histidine tag was added for purification purposes. When collecting supernatant for purification, the cells were kept in Ham’s F-12 Nutrient Mixture media free from FCS. CTRP6 was purified by incubating the cell supernatant with Ni-NTA beads (Qiagen) rotating end over end overnight at 4°C. The beads were subsequently spun down at 800 × g for 5 min and washed with 20 ml wash buffer. Bound protein was eluted with elution buffer and dialyzed against TBS + 2.5 mM CaCl₂ and stored in aliquots at −80°C. The concentration of CTRP6 in the supernatant and elution fractions was calculated with an in-house sandwich ELISA using commercially available CTRP6 (Origene) as calibrator, whereas the purity was assessed by InstantBlue staining (Expedeon) as described below.

The purified CTRP6 was denatured for 5 min at 90°C, and a total of 10 μg was loaded on a NuPage 4–12% Bis-Tris gel (Invitrogen) and run for 1 h in NuPage MOPS running buffer (Invitrogen) at 150 V and 300 mA. The gel was subsequently stained in InstantBlue solution on a shaker for 1 h. The gel was washed in distilled water before scanning.

A total of 1 μg of the purified CTRP6 and commercial CTRP6 (Origene) was added to the SDS-PAGE as described above and subsequently blotted onto a 0.45-μm nitrocellulose membrane (Amersham) in NuPage transfer buffer (Invitrogen) for 1 h at 30 V and 300 mA. Equal volumes of His-tag purified CTRP1 and undiluted cell supernatant from CTRP1-transfected CHO Flp-in cells were loaded as negative control. After transfer the membrane was blocked with 5% nonfat dry milk (Merck Millipore) in PBS-T for 15 min at room temperature (RT), followed by wash for 30 min. The primary Ab E-256 (AdipoGen Life Sciences), a mouse mAb raised against human CTRP6 was added to the membrane in a dilution of 1:10,000 in PBS-T and incubated for 1 h. After the wash the secondary Ab, rabbit anti-mouse-HRP (Dako) was added in a dilution of 1:10,000. Images were taken using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and MicroChemi System (DNR Bio Imaging Systems). Protein Standard Precision Plus All Blue (Bio-Rad) was used as the protein ladder.

Maxisorp microtiter plates (Nunc) were coated overnight with 2 μg/ml anti-CTRP6 Ab E-256 (AdipoGen Life Sciences). Serial dilutions of commercial CTRP6 (Origene) was used as calibrator. Purified CTRP6 and the commercial calibrator were diluted in TBS-T buffer and applied to the plate for 2 h, followed by detection with biotinylated anti-CTRP6 Ab E-256 in a concentration of 0.5 μg/ml in TBS-T. Using the same Ab for coat and detection is possible because of the repetitive structure of CTRP6. Finally, HRP-conjugated streptavidin (Sigma-Aldrich) was added to the wells diluted in TBS-T at a 1:2000 dilution for 1 h. TMB One (KemEnTec Diagnostics) was used as HRP substrate. The reaction was stopped with 0.2 M H₂SO₄, and the OD was measured at 450 nm with the ELx80 absorbance reader (BioTek). Between steps, the plate was washed with TBS-T, and, except for the coating step, all incubations took place at RT on an orbital shaker. The optimal combination of coat and detection concentration was determined using a two-dimensional titration of the capture and detection Abs against serial dilutions of purified CTRP6 and rated according to their signal intensity and signal-to-noise ratio.

Acetylated BSA (AcBSA), BSA, mannan, and zymosan were coated overnight in PBS in a 2-fold serial titration starting at 10 μg/ml in a Maxisorp microtiter plate. LPS from E. coli and Salmonella typhi (Sigma-Aldrich) were coated likewise in PBS + 1 M NaCl. CTRP6 was incubated with the ligands in a concentration of 2 μg/ml in TBS-T ± 10 mM EDTA. CTRP6 was detected with the anti-CTRP6 Ab E-256 in a concentration of 1 μg/ml in TBS-T. HRP-conjugated rabbit anti-mouse IgG (Dako) was used as a secondary Ab in a dilution of 1:2000 in TBS-T. Signal development and detection were performed as described previously.

AcBSA was coated overnight in a concentration of 10 μg/ml in a Maxisorp microtiter plate. A serial dilution of mannan, BSA, AcBSA, zymosan, LPS (starting at 20 μg/ml), l-fucose, d-mannose, GlcNAc, GalNAc, and galactose (starting at 62.5 mM) in TBS-T were preincubated with 2 μg/ml of CTRP6 for 1 h at RT before transfer to the AcBSA-coated plate. CTRP6 and the ligands were incubated for 2 h at RT before wash and addition of the anti-CTRP6 Ab E-256 in a concentration of 1 μg/ml in TBS-T. Signal development and detection were performed as described previously. The IC₅₀ was calculated with Quest Graph IC50 Calculator (AAT Bioquest).

A Maxisorp microtiter plate was coated overnight with 2 μg/ml mAbs against ficolin 3 [FCN334 (24)], ficolin 2 [FCN219 (24)], MBL (HYB131-11; BioPorto), or CL-11 [Hyb-15 (23)]. Purified recombinant-produced in-house ficolin-2 (25), ficolin-3 (26), MBL (27), or CL-11 (23) diluted in TBS-T was applied to the plate in a starting concentration of 2 μg/ml and diluted 2-fold for 2 h at RT before wash and addition of CTRP6 in a concentration of 2 μg/ml. The amount of CTRP6 bound was measured by a biotinylated anti-CTRP6 E-256 Ab and HRP-conjugated streptavidin in a dilution of 1:2000. Signal development and detection were performed as described previously. Additionally, we coated a Maxisorp microtiter plate with the anti-CTRP6 Ab E-256 in a concentration of 2 μg/ml or anti-MBL (HYB131-11; BioPorto). Purified CTRP6 or MBL was added in a concentration of 2 μg/ml diluted in TBS-T for 2 h at RT before wash and addition of recombinant MASP-1, MASP-2, MASP-3 (28), and MAP-1 (29) or buffer as control. MASP-1, MASP-3, and MAP-1 were all detected with a biotinylated mAb recognizing the common H chain of the three molecules [8B3 (29)]. MASP-2 binding was measured with a rabbit polyclonal Ab (Abnova). HRP-conjugated streptavidin or swine anti-rabbit Ig (Dako) was used as a secondary Ab in a dilution of 1:2000 in TBS-T. Signal development and detection were performed as described previously.

A Maxisorp microtiter plate was coated with 2 μg/ml mouse mAb anti–CL-11 clone Hyb-15 overnight. A serial dilution of CL-11 starting in a concentration of 2 μg/ml in TBS-T was added to the plate. It was subsequently washed in TBS-T and then incubated with 2 μg/ml CTRP6. Bound CTRP6 was determined with a biotinylated anti-CTRP6 Ab E-256 Ab and developed as previously described.

A suspension of 2 μg/ml CTRP6 and CL-11 in TBS-T was incubated for 1 h before addition of Dynabeads (Invitrogen) coupled to anti-CTRP6 Ab E-256, an anti–CL-11 Ab Hyb-15, or an IgG1κ isotype control. The mix of proteins and Ab-coupled Dynabeads was incubated for 2 h rotating end over end at RT. The beads were then removed and washed three times in TBS-T. The beads were resuspended in 30 μl H₂O and applied to SDS-PAGE with purified protein as positive control. In the Western blot of the immunoprecipitated CTRP6, coprecipitated CL-11 was detected by a rabbit polyclonal Ab in a concentration of 0.05 μg/ml (Abnova). In the CL-11 immunoprecipitation, coprecipitated CTRP6 was detected with biotinylated anti-CTRP6 E-256, after which the Western blot was developed as previously described.

AcBSA was coated overnight in a Maxisorp microtiter plate in a concentration of 10 μg/ml. CTRP6 was added in a concentration of 5 μg/ml for 2 h at RT in TBS-T. CL-11 was added after wash in a concentration of 2 μg/ml in TBS-T. The amount of recruited CL-11 was measured using anti–CL-11 Hyb-15 in a concentration of 2 μg/ml, followed by HRP-conjugated rabbit anti-mouse (Dako) in a 1:2000 dilution in TBS-T. The ELISA was developed as previously described.

AcBSA was coated overnight in a Maxisorp microtiter plate in a concentration of 10 μg/ml. CTRP6 was added in a concentration of 5 μg/ml for 2 h at RT in TBS-T. Recombinant CL-11 was added in a concentration of 2 μg/ml and incubated for 2 h at RT in TBS-T. Recombinant MASP-2 was incubated for 4 h at RT in TBS-T. Purified human C4 (Comptech) was incubated in the ELISA plate for 1 h at 37°C in a concentration of 0.5 μg/ml, and the amount of C4 deposited was measured with a polyclonal anti-C4c Ab (Dako) followed by HRP-conjugated swine anti-rabbit (Dako), both in a 1:2000 dilution in TBS-T. The ELISA was developed as previously described.

Clinical isolates of EAEC and laboratory type strains were provided by Statens Serum Institut (Copenhagen, Denmark). Refer to Table I for bacterial strain details. Stock cultures were frozen at −80°C in Luria–Bertani broth (LB; Sigma-Aldrich) containing 10% (v/v) glycerol. The bacteria cultures were cultivated in LB medium (Table I, [1-9]) or DMEM containing 4.5 g/l d-glucose (DMEM-HG; Life Technologies) (Table I, [10-15]). All strains were cultivated overnight, with shaking at 37°C. The OD (OD600 nm) was measured for each bacterial strain, and the corresponding CFU/ml were determined.

The overnight bacterial cultures were centrifuged at 5000 × g for 5 min and washed three times in PBS. The cell pellet was resuspended in TBS-T, and 2.5 × 10⁸ CFU were incubated with 0.2 μg/ml CTRP6 rotating end over end for 2 h. After centrifugation (10,000 × g for 5 min), the supernatant was applied to a Maxisorp microtiter plate (Nunc), and the concentration of CTRP6 in the supernatant was measured using the CTRP6-specific sandwich ELISA described earlier. The level of binding was evaluated by comparing the amount of remaining protein with a control sample containing no bacteria.

CTRP6 binding to bacteria was confirmed by microscopy. The strains E. coli [1], P. aeruginosa [8], and EAEC [10] were incubated with 2 μg/ml CTRP6. A mouse mAb E-256 or a mouse isotype control Ab was added to the bacteria for 30 min at 4°C, followed by another round of wash and centrifugation. The secondary Ab Alexa Fluor 488–coupled goat anti-mouse (Thermo Fisher Scientific) was added to the samples for 30 min at 4°C. The bacteria were spun at 500 × g for 5 min onto slides with Cytospin (Thermo Fisher Scientific) and mounted with Diamond Antifade Mountant (Life Technologies). The images were obtained using a Zeiss Axio Observer through a ×63/1.40 oil DIC Plan-Apochromat objective and Zen pro software (Zeiss, Germany). Image processing was performed with the Zen Blue software (Zeiss). Imaging conditions were kept constant during sample acquisition.

All statistical analyses were performed in GraphPad Prism software v8.0 (GraphPad). Significance of binding was calculated by paired parametric t test, and p values <0.05 were considered significant. All data are represented as mean of three individual experiments ± SD.

The purification of CTRP6 from cell supernatant was performed using Ni-NTA beads incubated with the supernatant overnight. The purity of the purification was assessed by total protein stain using InstantBlue (Fig. 1B), and the bands were identified by Western blot using an anti-CTRP6 specific mAb (Fig. 1C). CTRP6 presents a ladder-like pattern with bands larger than 250 kDa when under nonreducing conditions in SDS-PAGE. The pattern is disrupted under reducing conditions, indicating that CTRP6 forms higher-order oligomers stabilized by cysteine bridges. The monomeric band of CTRP6 migrates as a band of ∼30 kDa corresponding to the theoretical size of CTRP6 (aa 47–278) (30). We observed no cross-reaction of the Ab when using purified CTRP1 or cell supernatant from cells transfected with CTRP1 as negative control under reduced or nonreduced conditions (Fig. 1C).

We addressed the hypothesis of CTRP6 acting as a PRM by screening the binding to ligands routinely used in the measurement of complement activity. The ligands were probed with purified CTRP6 in ELISA, and the amount of CTRP6 bound was measured with an anti-CTRP6 mAb. We tested AcBSA, mannan, zymosan, and LPS derived from E. coli and S. typhi in the presence of calcium or EDTA (as cationic chelator). BSA was used as a negative control. The results of the ELISA are displayed in Fig. 2. We observed a dose-dependent binding to AcBSA and zymosan in the presence of calcium. When calcium was chelated with EDTA, we observed an increased binding of CTRP6 to AcBSA and zymosan, as well as binding to the two strains of LPS employed. No binding to BSA or mannan was detected regardless of the buffer used.

To determine if the different ligands shared the same binding site(s) with AcBSA, we coated AcBSA and measured if addition of ligands to the liquid phase interfered with the binding of CTRP6 to the coat. We observed a dose-dependent inhibition of the binding to AcBSA by zymosan, AcBSA, LPS from E. coli, (Fig. 3) and l-fucose (Fig. 4) when the buffer contained calcium. In addition, we observed an inhibition in the binding to AcBSA with LPS from S. typhi in the presence of EDTA, like the direct binding to the immobilized ligand. We observed no interference of binding with any of the other sugars employed. We calculated the IC₅₀ of l-fucose to 4.46 mM. As previously observed the total binding was increased with EDTA.

A total of 14 Gram-negative bacterial strains were screened in a consumption assay in which we measured the residual CTRP6 in the supernatant after being incubated with bacteria (Fig. 5). We observed a depletion from the supernatant from seven of the strains. Two strains of P. aeruginosa and five prototypic strains of EAEC. The consumption was evaluated by comparing the depletion of CTRP6 to a sample that contained no bacteria. The depletion of CTRP6 was unaltered when calcium was chelated by EDTA (data not shown). The binding to the EAEC [10] and P. aeruginosa strain [8] was confirmed by microscopy using E. coli strain [1] as a negative control (Fig. 6).

We screened for the binding of CTRP6 to ficolin-2, ficolin-3, MBL, and CL-11; the lectin pathway activating enzymes MASP-1, MASP-2, and MASP-3; and the splice variant from the MASP1 gene MAP-1 in a sandwich ELISA. We did not observe any notable interaction with the MBL, the ficolins (Supplemental Fig. 1), or the MASPs (Supplemental Fig. 2). However, we observed a strong interaction with CL-11.

Further defining this interaction, we coated an ELISA plate with an in-house produced Ab raised against CL-11 and titrated CL-11 (Fig. 7). We measured the amount of CTRP6 bound to CL-11 and observed a CL-11 dose-dependent binding of CTRP6 (Fig. 7). The interaction between CTRP6 and CL-11 was not dependent on calcium (data not shown). We observed no signal when either CTRP6 or CL-11 was not present, or when the plate was coated with an unspecific isotype control Ab. We confirmed the interaction of CTRP6 and CL-11 by immunoprecipitation and Western blotting in which we detected coprecipitation of CL-11 when immunoprecipitating CTRP6 and vice versa (Fig. 8).

To address the functionality of the CTRP6–CL-11 interaction, we investigated if CTRP6 could recruit CL-11 to AcBSA and thereby function as an adaptor molecule for CL-11. We observed that not only could CL-11 bind to AcBSA in the presence of CTRP6 (Fig. 9) but that the recruited CL-11 was able to form complex with MASP-2 and cleave and deposit C4, resulting in a downstream activation of the complement cascade (Fig. 10).

Recombinant human full-length his-tagged CTRP6 was produced in CHO Flp-In cells and purified to >90% purity. We found that the recombinant protein displayed a characteristic oligomerization pattern when separated on SDS-PAGE under nonreducing conditions with oligomeric complexes higher than 250 kDa. The pattern is very similar to the pattern seen when applying the ficolins to Western blot (31), suggesting that CTRP6 forms oligomers in a similar way and that the oligomerization is dependent on the presence of cysteine bridges, as the pattern disappears under reducing conditions (Fig. 1B, 1C). This is further established when comparing similarity of domain organization of the PRMs of complement and the CTRPs (Fig. 1A). We have found evidence that CTRP6, in addition to its role as hormone (15, 21, 32), act as a PRM. AcBSA has previously been used to describe the activity of the ficolins, which are well-established PRMs of the complement system (24). In this study, we showed that CTRP6 bound to AcBSA but also, to some extent, to zymosan and LPS, but no binding was observed to mannan or BSA. The calcium-independent binding to AcBSA contrasts with the ficolins that require calcium to bind to many of their ligands (24, 25). CL-11, however, display both calcium-dependent binding to mannan and a calcium-independent binding to zymosan under conditions in which CL-11 is able to initiate the complement cascade via the MASPs (23, 33, 34). We suggest that the increased binding of CTRP6 is through a conformational change to either CTRP6 or the surface when calcium is removed, which makes it more accessible for CTRP6 to bind a ligand on a solid surface.

We found a high degree of inhibition when CTRP6 was incubated with l-fucose (IC₅₀ = 4.46 mM), but not to the other investigated sugars (d-mannose, GlcNAc, GalNAc, and galactose), thereby establishing CTRP6 as a novel lectin. This inhibition indicates that the binding site is shared between l-fucose and AcBSA. Interestingly, Bayarri-Olmos et al. (23) have showed a similar inhibition by l-fucose in the binding of CL-11 to mannan (IC₅₀ = 36.48 mM), meaning that CTRP6 and CL-11 partially shares ligand specificity.

In addition to the capacity of CTRP6 to bind artificial ligands coated on a solid surface, we tested the capacity of CTRP6 to bind 14 randomly selected Gram-negative strains of bacteria. We found that two strains of P. aeruginosa and five prototypic strains of EAEC depleted CTRP6 from the supernatant and confirmed the binding to two of these strains by fluorescence microscopy.

Although we have described the capacity to bind selectively to both artificial and natural surfaces the collagen region of CTRP6 does not contain the putative protease binding motif: Hyp-Gly-Lys-Xaa-Gly-(Pro/Tyr) (35). This motif has been associated with the binding of MASPs and hence complement activation via C4/C2 (35). Although we found no interaction between CTRP6 and the MASPs we observed a strong interaction with CL-11 but none with the other investigated PRMs of the lectin pathway (Supplemental Fig. 1). The interaction with CL-11 persisted from capture in a sandwich ELISA (Fig. 7), to coprecipitation and detection in Western blot (Fig. 8), to a more native scenario where CTRP6 recruited CL-11 to AcBSA (Fig. 9). In addition, we were able to show that the recruitment of CL-11 to AcBSA was able to bind MASP-2 and initiate the complement cascade through C4 deposition (Fig. 10). Thus, our results suggest that CTRP6 is not initiating complement activation per se but rather functions as an adaptor molecule facilitating the binding of CL-11 and directs complement activation. A similar mechanism has been described for adiponectin, which binds the globular region of C1q and activates the classical complement pathway (10). Furthermore, it is well established that C1q and some of the ficolins use the pentraxins (CRP, PTX3, and SAP) apart from Igs as guiding molecules for ligand binding (36). However, the guiding and activating capabilities of CTRP6 are in conflict with the work of Murayama et al. (22), in which CTRP6 is described as a regulator of the alternative pathway through direct interaction with C3(H₂O), inhibiting the formation of the alternative pathway C3 convertase. These experiments were, to our knowledge, performed using recombinant protein produced in E. coli, and thus the difference in activating capabilities between the two studies might be due to the differences in CTRP6 oligomerization.

In conclusion, we have shown that CTRP6 may act as a PRM that interacts with sugars, microbial ligands, and live bacteria and that CL-11 uses CTRP6 to extend its recognition repertoire. At this stage, we can only conjecture if the rest of the CTRP family also possess the ability to bind ligands and recruit/activate the complement system; however, this study provides evidence that CTRP6, but possibly also other members of CTRP family, may have roles both in the endocrine and the innate immune system.

We appreciate the excellent technical assistance performed by Jytte Bryde Clausen.

The authors have no financial conflicts of interest.