Y. Takeda, Department of Environmental and Preventive Medicine, Hyogo College of Medicine, Mukogawa-cho 1-1, Nishinomiya, Hyogo 663-8501, Japan. E-mail: firstname.lastname@example.org
Cellulose acetate (CA) beads are often used for leucocyte apheresis therapy against inflammatory bowel disease. In order to clarify the mechanism of the anti-inflammatory effects of CA, global analysis of the molecules generated in blood by the interaction with CA beads was performed in this study. An activated medium was collected from whole blood that had been preincubated with CA beads, and the effects of the CA-activated medium on leucocyte function were investigated. Fresh blood was stimulated with lipopolysaccharide (LPS) or interferon (IFN)-β in the presence of the activated medium, and levels of chemokines and cytokines, including CXCL10 (IFN-inducible protein-10), and phosphorylated STAT1 (signal transducer and activator of transcription 1), which is known to be essential for CXCL10 production in leucocytes, were measured. IFN-β- or LPS-induced CXCL10 production, expression of CXCL10 mRNA and phosphorylation of STAT1 were significantly reduced in the presence of the medium pretreated with CA beads compared with the control without the CA bead treatment. The factors inhibiting CXCL10 production were identified as the C3 and C4 fragments by mass spectrometry. The monomeric C3bi and C4b proteins were abundant in the medium pretreated with CA beads. Furthermore, purified C3bi and C4b were found to inhibit IFN-β-induced CXCL10 production and STAT1 phosphorylation. Thus, STAT1-mediated CXCL10 production induced by stimulation with LPS or IFN was potently inhibited by monomeric C3bi and C4b generated by the interaction of blood with CA beads. These mechanisms mediated by monomeric C3bi and C4b may be involved in the anti-inflammatory effects of CA.
The discrimination between self and non-self is a basic immune function. In general, the interaction of blood with foreign bodies initiates inflammation and an immune reaction. Although cellulose acetate (CA) is an artificial material, CA beads induce anti-inflammatory responses and immune suppression . CA beads are used for granulocyte/monocyte apheresis, which has been established as an effective therapy for ulcerative colitis and Crohn disease . Granulocytes and monocytes are activated on the beads, but anti-inflammatory responses are induced . Identification of the molecules generated by the CA bead interaction may help not only to elucidate the paradoxical reaction induced by the beads, but may also lead to discovery of a new immune mechanism.
Leucocytes have been shown to adhere readily to CA beads and thereby release large amounts of tumour necrosis factor (TNF)-α receptors [4,5] and interleukin (IL)-1 receptor antagonist , both of which are strongly anti-inflammatory. The therapeutic efficacy of leucocyte apheresis using CA beads is known to persist for more than 6 months after therapy [6–8], so we speculated that the interaction between the CA beads and the blood prevented the development of acute inflammation in patients with inflammatory disease, and also protected those patients from self-reaction induced by lymphocyte functions.
In this study, we performed a global analysis of molecules generated by this interaction and found specific inhibitory effects of CA beads on release of interferon (IFN)-inducible protein-10 (IP-10/CXCL10), which is known to be an important molecule in the generation of effector T cells and the induction of various autoimmune diseases [9,10].
Materials and methods
We used RPMI-1640 medium, lipopolysaccharide (LPS, Escherichia coli 055:B5) and ImmunoProbe Biotinylation kit from Sigma (St Louis, MO, USA); Sypro Ruby and SilverQuest silver staining kit from Invitrogen (Carlsbad, CA, USA); Isogen-LS from Wako (Tokyo, Japan); RNeasy Micro Kit from Qiagen (Hilden, Germany); ReverTra Ace -α- from Toyobo (Osaka, Japan); LightCycler-Primer set and LightCycler FastStart DNA master SYBR Green I from Roche Diagnostics (Mannheim, Germany); C3bi and C4b from Calbiochem-Merck, EMD Biosciences (San Diego, CA, USA); IFN-β1a from PBL Biomedical Laboratories (Piscataway, NJ, USA); two-dimensional clean-up kit, DeStreak Rehydration Solution, immobilized pH gradient (IPG) buffer and Immobiline DryStrip gel from GE Healthcare Biosciences (Buckinghamshire, UK); and BlockAce from Dainippon (Osaka, Japan). All reagents were of the highest purity available commercially. Cytokines and chemokines were measured with BD™ Cytometric Bead Array System (BD Biosciences, San Jose, CA, USA) or with a Quantikine® human CXCL10/IP-10 immunoassay kit from R&D Systems (Minneapolis, MN, USA).
Anti-α-actinin was from Chemicon International (Temecula, CA, USA); mouse anti-C3 monoclonal antibody [mAb, clone H11, immunoglobulin (Ig)G1] from Progen Biotechnik (Heidelberg, Germany); normal goat IgG and goat IgG to human complement C4 from MP Biomedicals, LLC (Solon, OH, USA); EnVision+ kit and horseradish peroxidase (HRP)-conjugated streptavidin from Dako (Carpinteria, CA, USA); and R-phycoerythrin-conjugated mouse anti-human CD3 mAb (clone UCHT1, IgG1) and AlexaFluor 488-conjugated mouse anti-human signal transducer and activator of transcription 1 (STAT1)(pY701) mAb (clone 4a, IgG2a) from BD Biosciences. Normal goat IgG and goat IgG to human C4 were labelled by ImmunoProbe biotinylation kit according to the instruction.
Generation of opsonized CA beads and activated medium
Although minor molecules are concealed in large amounts of plasma proteins, plasma is required for granulocyte/monocyte adhesion on CA beads . Thus, we prepared plasma-poor blood (washed with RPMI-1640) and opsonized CA beads (preincubated with plasma). After institutional review board and informed consent was obtained, peripheral blood was collected from healthy volunteers. Blood was mixed with 5 U/ml of low-molecular-weight heparin and then centrifuged at 450 g for 10 min at room temperature. After centrifugation, the supernatant plasma was harvested and the packed cells were kept for preparing plasma-poor blood to facilitate purification of soluble factors. CA beads from JIMRO Co., Ltd (Takasaki, Japan) were autoclaved in saline and washed with saline prior to use. The beads were mixed with plasma in a syringe (the ratio of plasma to CA beads was 1 ml : 2 g) and incubated with one time-inverting rotation per min (1 i-rpm) for 1 h at 37°C. The beads were washed twice in saline before further exposure to blood cells.
Plasma-poor blood cell suspensions were prepared as follows. After removing the plasma, packed cells were washed once with 10 volumes of RPMI-1640 and then were resuspended in RPMI-1640 to obtain the same volume as the initial blood (approximately 10% of the original plasma was retained). The plasma-poor blood cell suspension was drawn into syringes containing the opsonized beads (1 ml–2 g). The syringes were rotated gently at 1 i-rpm at 37°C for 1 h. Cell suspensions incubated without CA beads were used as a control. After incubation, the suspension was removed from the syringe and the supernatant was collected following centrifugation at 450 g for 10 min.
Stimulation of whole blood
Fresh peripheral blood from healthy volunteers was collected into low-molecular-weight heparin and then mixed with the activated medium (or control) and RPMI-1640 containing fetal calf serum (FCS). The mixture consisted of blood (10%), activated medium (50%), FCS (10%) and RPMI-1640 (30%). The mixture was incubated in the presence or absence of a stimulant (LPS or IFN-β1a) in a CO2 incubator. After incubation, samples were centrifuged, and the conditioned medium was then collected and stored at −80°C until the cytokines were assayed.
Total RNA was purified by using Isogen-LS and cleaned up by RNeasy Micro Kit according to the manufacturer's instructions. cDNA was synthesized using oligo(dT) primers by ReverTra Ace -α-. Quantitative real-time RT–PCR was performed in an ABI Prism 7900HT system (Life Technologies, Carlsbad, CA, USA) using LightCycler FastStart DNA master SYBR Green I. The specific primer pairs and standards were prepared form a LightCycler-Primer set of human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and human IP10 (CXCL10). The thermal cycling conditions were as follows. For amplification of both GAPDH and CXCL10, we used 37 cycles at 95°C for 10 s, 64°C for 10 s and 72°C for 45 s. CXCL10 and GAPDH mRNA expressions in the samples were quantified using the corresponding standard curve. CXCL10 mRNA levels were corrected by reference to the GAPDH mRNA levels.
Two-dimensional gel electrophoresis to identify substances responsible for the function of the activated medium
Albumin and immunoglobulin in the samples were removed by using an Aurum Serum Protein Mini Kit (Bio-Rad, Hercules, CA, USA) and concentrated by ultra-filtration. The first dimension of electrophoresis (isoelectric focusing) was performed with an Ettan IPGphor II System (GE Healthcare Biosciences). Briefly, the samples were treated with a two-dimensional clean-up kit and dissolved in DeStreak rehydration solution containing 0·5% IPG buffer and were then applied to an Immobiline DryStrip gel (3–11 NL). After the isoelectric focusing, the strip was soaked in Laemmli sample buffer (Bio-Rad) containing 6 m urea and 0·25% dithiothreitol for 15 min. The strip was set on a 10–20% two-dimensional polyacrylamide gel electrophoresis (PAGE) mini gel (Daiichi, Tokyo, Japan) with 0·5% agarose for sodium dodecyl sulphate (SDS)-PAGE. Following electrophoresis, the gel was silver-stained.
Ion exchange column chromatography to separate components of activated medium
Activated medium (15–20 ml) was desalted in 2·5 volumes of 10 mm Tris-HCl (pH 8·0) by ultra-filtration, using an Amicon Ultra-15 Centrifugal Filter Unit with Ultracel-50 membrane (Nihon Millipore, Tokyo, Japan). The desalted sample was applied to diethylaminoethyl cellulose (DEAE)-sepharose (GE Healthcare Biosciences) and washed with 100 ml of 10 mm Tris-HCl (pH 8·0) containing 0·1 m NaCl. After washes, the samples were eluted by a gradient of 0·1–0·3 m NaCl (flow rate 0·25 ml/min) and collected as 1 ml fractions by a fast protein liquid chromatography (FPLC) system (LCC-500; GE Healthcare Biosciences). The fractions were concentrated to <0·2 ml by ultrafiltration with phosphate-buffered saline (PBS) and used for a culture assay and SDS-PAGE analysis.
Mass spectrometry for identification of molecules in the gel
The gels were stained with Sypro Ruby. The candidate bands or spots were picked up from the gels for mass spectrometric (MS) analysis, which was performed by ProPhoenix (Hiroshima, Japan). Pieces of the gels were digested enzymatically with purified protease and analysed using Ultraflex (Bruker Daltonics, Billerica, MA, USA). The results of the peptide mass fingerprint analysis were confirmed by an MS/MS ion search analysis and were verified further by using the predictive value from databases with mascot search (Matrix Science, London, UK).
Measurement of STAT phosphorylation
For the staining method that involved the use of an anti-STAT1 mAb, whole blood was treated by BD Phosflow Lyse/Fix buffer and BD Phosflow Perm buffer III (BD Biosciences) according to the BD Phosflow Protocols (Protocol III). The flow cytometry analysis was performed using fluorescence activated cell sorter (FACS)Calibur and CellQuest Pro software version 4·02 (BD Biosciences).
The study groups were compared by analysis of variance (anova) with Bonferroni's post-hoc test or Student's t-test. The calculations were performed with Prism Software version 5·03 (GraphPad Software, San Diego, CA, USA). P < 0·05 was considered significant.
Suppression of leucocyte-induced CXCL10 production by activated medium
CXCL10 production from whole blood by stimulation with LPS or IFN-β1a was detected after 3 h of incubation and reached the plateau level after 12 h of incubation (Fig. 1a,b). Addition of the activated medium to whole blood caused a marked decrease in LPS-induced CXCL10 production after 12 and 18 h of incubation compared with the control in the absence of the activated medium (Fig. 1a). IFN-β1a-induced CXCL10 production after 12 and 18 h of incubation was also inhibited in the presence of the activated medium (Fig. 1b). Only a trace level of CXCL10 mRNA expression was detected without stimulation. The level of CXCL10 mRNA expression was increased markedly at 3 h after stimulation with LPS or IFN-β1a, and this increased expression was inhibited strongly in the presence of the activated medium. In contrast, the activated medium did not suppress LPS-induced production of chemokines (CCL2, CCL5, CXCL8 and CXCL9) and cytokines (TNF-α, IL-1β, IL-6, IL-10 and IL-12p70) significantly (data not shown).
Presence of C3bi in activated medium
By using two-dimensional gel electrophoresis, we attempted to identify the molecule that was mediating CXCL10 inhibition. Despite the difference between activated medium and the control in inducing CXCL10, the major spots in the activated medium were also observed in the control (Fig. 2a). The predominant spots in two-dimensional gels were identified as the plasma proteins transferrin, albumin and immunoglobulins, all listed in the SWISS-2DPAGE database (http://www.expasy.org/swiss-2dpage). This suggested that the major molecules were proteins carried over from plasma. However, the spot designated as (*) on the basic side of the transferrin spot was observed only in the activated medium gel and not in the control (Fig. 2a, lower panel). The sequence-tagged mass from the marked spot (*) was identified by MS/MS analysis as a complement C3 β-chain fragment (Table 1).
Table 1. Identification of spot and bands in the activation medium by MS/MS analysis.
Spot or bands
Identified molecule (accession#)
Mascot research results
Spot and bands (*, **−1, **−2, and ***) shown in Fig. 2.
Inclusion of the C4 fragment and α-actinin-4 in the fraction of activated medium that inhibits CXCL10 release
The activated medium was fractionated by DEAE ion-exchange column to separate and identify the fraction involved with the inhibition of CXCL10 production induced by LPS. Prominent inhibitory action was observed in fractions 33, 35 and 37, and inhibition was highest with fraction 35 (Fig. 2b, upper panel). By using SDS-PAGE, we identified the specific bands unique to fractions 33, 35 and 37 (Fig. 2b, lower panel): 80 kD (**−1), 28 kD (**−2) and 110 kD (***).Using MS, we also identified the 80 kD band (**−1) and 28 kD band (**−2) as complement C4 α-chain fragments and the 110 kD band (***) as α-actinin-4 (Table 1).
Formation of C3 and C4 fragments in activated medium
To verify the results of the MS identification, we analysed activated medium by immunoblotting. C3 and C4 fragments were present in the activated medium, but not in the control medium (Fig. 3a,b). The molecular mass based on Western blotting under non-reducing conditions suggested that the C3 fragment was in the form of C3/C3b(i) (Fig. 3a) and the C4 fragment was in the form of C4/C4b and C4c (Fig. 3b). To elucidate the fragmentation of C4, activated medium was immunoblotted with anti-C4 under reducing conditions. C4 α-chain in activated medium disappeared and fragmented to α′ (27 kD) and α′′ (14 kD); however, the amount of C4 β-chain did not change in activated medium (Fig. 3c). These results suggest that the C4 α-chain in activated medium was digested, while almost all the C4 β- and γ-chain remained.
Inhibition of CXCL10 production by exogenous C3bi and C4b
To verify the function of the complement opsonins, we tested whether C3bi and C4b directly suppress CXCL10 production induced by authentic IFN-β1a. Exogenous C3bi and C4b suppressed the release of CXCL10 dose-dependently by IFN-β-stimulated leucocytes (Fig. 4a). A combination of C3bi and C4b at each 10 and 50 µg/ml did not inhibit CXCL10 production, while the high-dose combination at each 250 µg/ml inhibited CXCL10 production significantly. No additive inhibitory effect of the combination was observed on CXCL10 production (Fig. 4a).
Suppression of STAT phosphorylation in lymphocytes by authentic C4b and C3bi
To understand more clearly the inhibitory function of C4b and C3bi, we determined which leucocyte subpopulation was inhibited during phosphorylation of STAT1 induced by IFN-β1a (10 U/ml). The C3bi and C4b mixture inhibited IFN-β-induced STAT1 (Y701) phosphorylation significantly in CD3+ T cells and CD3- B cells (Fig. 4b).
The CA beads induced production of C3bi and C4b in the fluid phase. Activated medium containing C3bi and C4b specifically inhibited CXCL10 release. Further, exogenous C3bi and C4b also inhibited CXCL10 release and phosphorylation of STAT1. This is the first report of inhibitory actions of C3bi and C4b on type I IFN stimulation.
Naive C3 and C4 circulate in the plasma but do not bind to any receptors or pathogens. When complement is activated by foreign bodies, C3 is converted into C3a and C3b (C3b is converted immediately to C3bi), and C4 is converted into C4a and C4b. C3b and C4b are the major cleavage products that bind to targets. However, the binding of these products to their targets is not efficient. Typically, only about 10% of the C3bi and C4b binds to targets; the other 90% remains in the fluid phase . Indeed, the amount of C4 β-chain (β-chain is not digested after the C4 conversion to C4a and C4b) in activated medium was almost same as that in the control (Fig. 3c). These results indicate that C3 and C4 are fragmented to C3bi and C4b by interaction with CA beads, while almost all the C4b is retained in the fluid phase. As shown in Fig. 3b, the major bands in the activated medium were C4 complex and C4/C4b, and only a trace level of C4c was detected in the medium by the Western blot analysis. Thus, most C4b molecules are thought to remain intact and not to be digested to C4c. Furthermore, in non-reducing conditions, C3 and C4 in the control easily formed complexes, while C3 and C4 in the activated medium did not (Fig 3a,b). This result suggests that the internal thioester in C3b and C4b may be deactivated immediately by interaction of blood with CA beads.
Although we detected C3 fragments as specific spots in the activated medium on two-dimensional electrophoresis gels, C4 fragments were not detected in the activated medium on two-dimensional electrophoresis gels. It is difficult to detect small amounts of proteins in serum using two-dimensional electrophoresis, because a large amount of albumin or immunoglobulin conceals small amounts of proteins. The concentration of C4 in blood is about one-fourth that of C3 [normal range: 123–167 mg/dl (C3) versus 22–40 mg/dl (C4)]. Thus, C4 fragments may not be detected stably on the gels. Another possible reason for no detection of C4 fragments is that the removal of albumin and immunoglobulin in the samples by an affinity mini-column, as described in Materials and methods, might consume C4 through activating the classical pathway induced by immunoglobulin aggregation.
Conversely, C3 fragments were not detected in the fractions eluted from the DEAE-sepharose column. This may be explained by the fact that the molecular weight of C3 β-chain is similar to that of major proteins, such as albumin and transferrin, in serum. Indeed, the fractions containing albumin (fractions 15–27) slightly inhibited CXCL10 production, suggesting contamination of the C3 fragments.
The mechanism by which monomeric C3bi and C4b inhibit type I IFN stimulation remains unknown. However, this study shows that the effect of a combination of purified C3bi and C4b on CXCL10 production was not additive. Furthermore, the inhibition disappeared when lower doses of C3bi and C4b (each 10 and 50 µg/ml) were combined (Fig. 4a). Therefore, we speculate that by mixing purified C3bi and C4b, a C3b–C4b complex may be generated via the remaining thioester, and then the complex might not be able to inhibit the CXCL10 release.
We speculate that CD46 is a candidate molecule for the regulation of the type I IFN signalling by complements as a complement receptor, CD46 (membrane co-factor protein, MCP), expressed on T and B cells binds to both C3bi and C4b . In RAW264·7, a macrophage cell line, ligation of CD46 with the C3b multimer has been shown to enhance IFN-γ (type II IFN)-dependent nitric oxide production, which is known to be mediated through STAT1 activation, while this enhancement was not observed in the presence of monomeric C3b . Although the receptor for type I IFN is different from that for type II IFN, cellular responses to both types of IFN are known to be mediated by STAT1 . Thus, C3b multimer-induced ligation of CD46 on cell membrane enhances STAT1 activation stimulated with IFN. In the present study, monomeric forms of C3bi and C4b were detected in the activated medium. We thus postulate that CD46 ligation is inhibited by monomeric C3bi or C4b originating from the activated medium, resulting in the suppression of STAT1 activation. Another possible mechanism for the actions of C3bi and C4b is inhibition of STAT1 by recruitment of SHP1 protein tyrosine phosphatase, which has been shown to be mediated by CD46 . Further studies are needed to test whether monomeric C3bi or C4b affects the recruitment of SHP1 protein tyrosine phosphatase.
The level of CXCL10 mRNA expression was markedly increased at 3 h after stimulation with LPS or IFN-β, and this increased expression was inhibited strongly in the presence of the activated medium. Thus, C3bi and C4b are thought to inhibit CXCL10 transcription induced by LPS or IFN-β. The degree of the inhibitory effect on STAT1 phosphorylation was not as prominent as the inhibitory effects on CXCL10 production and its mRNA induction (Fig. 4b). Thus, there is a possibility that besides the inhibition of the STAT1 pathway, another mechanism(s) is involved in the inhibitory actions of C3bi and C4b on IFN-stimulated CXCL10 production, and further studies are needed to test this possibility.
The stimulation of CD46, which binds to C3b/C4b , temporarily alters acquired immunity via induction of regulatory T cells [16–19]. It has been reported that lymphocytes of haemodialysis patients show an increased response on CD46 co-stimulation, because complement activation occurs during haemodialysis . Conversely, the treatment with CA beads also induces complement activation and regulatory T cells in vivo[21–23]. These observations allow us to speculate that there is a relation between complement activation by CA beads and the CD46 stimulation by C3b/C4b.
CXCL9 production in monocytes is known to be mediated by STAT1 . However, in our preliminary experiments, CXCL9 production was not increased by stimulation with LPS in whole blood. Further studies are needed to clarify whether the production of other chemokines and cytokines is influenced by complements.
STAT1 activation is associated closely with some diseases due to immune disorder, such as coeliac disease and chronic graft-versus-host disease [25,26]. Recently, STAT1 has been identified as a novel therapeutic target for atherosclerosis . Adacolumn containing cellulose acetate (CA) is a useful tool for therapy against inflammatory bowel diseases. Therefore, a mechanism underlying this therapy is proposed as follows: monomeric C3bi and/or C4b produced by the interaction of blood with CA beads inhibit STAT1-mediated production of chemokines such as CXCL10, resulting in the attenuation of inflammatory responses.
Induction of type I IFN (IFN-α/β) by viruses and other pathogens has been recognized as an important step in the innate immune response , while overproduction of type I IFN by an autoimmune response has been recognized as a risk factor for peripheral immune tolerance breakdown and the development of systemic lupus erythematosus (SLE) . Recently, it has been reported that C1q, the first protein of the classical complement pathway, is required for inhibition of IFN-α and CXCL10 production . In this study, monomeric C3bi and C4b suppress type I IFN signalling and CXCL10 production. C1q and C4 deficiency are associated strongly with increased risk for the development of systemic lupus erythematosus (SLE) . Thus, we propose that these complement factors act as a rheostat controlling type I IFN response.
The C4 gene is located between the human leucocyte antigen (HLA) class I gene and the HLA class II gene; thus the C4 gene is linked with homologous recombination of HLAs and is highly polymorphic, with various allotypes [32,33]. It would be interesting to learn whether there is a relationship between the polymorphic C4 and the ability to bind CD46 or to control type I IFN response.
In conclusion, we show that C3bi and C4b inhibit CXCL10 release. The inhibitory action of the monomeric opsonins may lead to a new understanding of the complement feedback system.
We are grateful to Mr Hiromu Shibusawa for his valuable technical support.
No competing interests to declare. No external funding was received for this study.