IL-10 Controls Cystatin C Synthesis and Blood Concentration in Response to Inflammation through Regulation of IFN Regulatory Factor 8 Expression Free

Cysteine proteases play fundamental roles in multiple biological processes such as protein turnover, proprotein processing, bone remodeling, Ag presentation, and apoptosis (1, 2). However, they are also involved in numerous pathological processes such as cardiovascular disease and inflammation (3, 4). The activity of these enzymes thus needs to be tightly controlled by inhibitors. Cystatin C (CstC) belongs to the cystatin type II superfamily of protease inhibitors and strongly inhibits papain-like cysteine proteases and legumain (5, 6). It is a small (13-kDa) secreted protein and is abundant in most body fluids.

The importance of CstC in health and disease is clearly established. Reduced CstC in circulation and atherosclerostic plaques correlates with elevated activity of the cysteine proteases cathepsins K and S, which are involved in the breakdown of elastic laminase in the blood vessel wall (7). Altered CstC expression and extracellular proteolysis reportedly play a role in several other clinical conditions including tumor metastasis (8), autoimmune disease (9), and liver dysfunction (10). Indeed, the association between CstC and disease is so strong that the level of extracellular CstC has diagnostic value and is a marker of disease prognosis in cancer (11), cardiovascular disease (12), and inflammatory lung disorders (13).

In addition to its protease inhibitory function, CstC also plays regulatory roles unrelated to proteolysis. It conditions the sensitivity of macrophages to IFN-γ (14), induces NO production (15), and blocks TGF-β signaling in normal and cancer cells (16). An additional property of CstC that makes this protein clinically relevant is that it can form aggregates by a mechanism known as domain swapping, similar to that involved in formation of amyloid β (Aβ) in Alzheimer’s disease (17). This phenomenon is exacerbated in individuals suffering from hereditary CstC amyloid angiopathy (HCCAA) (18), a fatal disease caused by a point mutation that provokes a single Leu to Gln amino acid substitution at position 68, which favors CstC aggregation and formation of amyloid fibrils in brain vessels. In contrast, it has been suggested that CstC plays a protective role in Alzheimer’s disease by binding to and preventing formation of Aβ (19, 20).

The diverse functions and clinical relevance of CstC raise the question of whether its expression is regulated. CstC has been considered a ubiquitously expressed protein with no regulatory elements in its gene promoter (21). However, we previously reported that the expression of CstC varies widely among hematopoietic cell types, with dendritic cells (DC) expressing much higher levels than B or T cells (22). Furthermore, three subtypes of DC have been identified in mouse spleen, and CstC is expressed primarily by the population characterized by expression of CD8, with little or no expression in the closely related CD8⁻ DC or in plasmacytoid DC (pDC) (22, 23). These findings foretold the existence of regulatory mechanisms of expression of CstC. Consistent with this prediction, an IFN regulatory factor (IRF)-external transcribed spacer composite sequence (IECS) in the CstC promoter region was found to be essential and sufficient for CstC expression in a differentiating myeloid progenitor cell line (24).

To characterize the mechanisms controlling the expression of CstC in primary cells in vivo, we have identified the transcription factors that determine the pattern of CstC expression among DC subtypes. We found that not only IRF-8, but also the transcription factor PU.1 are required to induce high CstC expression. We also describe a role for IL-10 in regulation of CstC expression. We found that systemic inflammation caused downregulation of IRF-8 and, consequently, CstC expression in DC, and this effect was mediated by IL-10. Furthermore, inflammation caused a drop in CstC concentration in blood, and this effect was also mediated by IL-10. Our results define the mechanisms that regulate CstC transcription in vivo and indicate that manipulation of IL-10 levels might be a therapeutic strategy for the treatment of diseases caused by insufficient expression of CstC (e.g., atherosclerosis) or deposition of CstC amyloid (e.g., HCCAA). Our results also urge caution regarding the reliance on CstC serum levels as a measurement of renal function (25) in situations in which inflammation may alter CstC production rather than its removal from circulation.

C57BL/6, CstC-deficient (8), and IRF-8–deficient (26) mice were bred in the animal facilities of the Walter and Eliza Hall Institute. IL-10–deficient mice (27) were bred in the animal facilities of the Baker IDI Heart and Diabetes Institute. Where indicated, C57BL/6 mice were injected s.c. with 5 × 10⁶ B16 melanoma cells secreting murine Flt3 ligand (Flt3L) (28) to expand DC numbers in vivo and killed after 9 or 10 d. All animal breeding and experimentation was carried out according to institutional guidelines.

Splenic DCs were isolated as previously described (29). The CD8⁺ and CD8⁻ DC were further purified as described (30). Briefly, the splenic DC preparation was first incubated with anti-CD8 mAb YTS 169.4 and magnetic beads (Miltenyi Biotec) to purify the CD8⁺ DC by positive selection. The remaining cells were depleted of CD8⁺ DC precursor cells (30) by incubation with anti-CD205 (NLDC-145) and anti-CD24 (M1/69) mAbs and MACS beads. Finally, the CD8⁻ DC were isolated by positive selection using MACS beads after incubation with anti-CD11b (M1/70) mAb. To obtain pDC, splenic DCs were isolated from mice injected with Flt3L-producing B16 cells as described (31). CD8⁺ DC were positively selected from this preparation using mAb YTS 169.4 and magnetic beads, leaving a preparation containing both pDC and CD8⁻ DC. The CD8⁺ DC line 1940 was derived from spleens of transgenic mice expressing the oncogenic SV40 T Ag under the control of the CD11c promoter (32). Bone marrow-derived DCs (BMDC) were generated in RPMI 1640 culture medium containing GM-CSF and IL-4 (GM/IL4-DC) or Flt3L (FL-DC) as described (33). Where indicated, CD11c⁺ cells were enriched with MACS beads and incubated overnight in medium alone or containing TNF-α (50 ng/ml), IFN-α (75 ng/ml), TGF-β (20 ng/ml), or IL-10 (50 ng/ml).

Chimeric mice were generated by irradiation of recipient B6.Ly5.1 mice with two doses of 550 rad 3 h apart. A total of 0.5 × 10⁶ B6.Ly5.1 together with 0.5 × 10⁶ CstC-deficient (Ly5.1/Ly5.2) bone marrow cells were injected i.v. Antibiotic (neomycin 1.1 g/l) was provided in the drinking water for the duration of the experiment. The mice were allowed to reconstitute for 6 wk before use. DCs were sorted by preparative FACS using expression of the Ly5 marker to distinguish their origin.

Chromatin immunoprecipitation (ChIP) was performed using the EZ ChIP kit (Upstate Biotechnology) according to the manufacturer’s instructions. The Abs used for IRF-8 and PU.1 immunoprecipitation were goat anti–IRF-8 (C-19; Santa Cruz Biotechnology) or rabbit anti-PU.1 (T-21; Santa Cruz Biotechnology). Amplification of the cst3 IECS was carried out by semiquantitative or real-time PCR using the oligonucleotides listed in the section below.

Oligonucleotides specific for irf8 inhibition (irf8 short hairpin RNAs [shRNAs]) were amplified from oligonucleotide templates by PCR (shRNA1: 5′-TGCTGTTGACAGTGAGCGATCCAACCTAGTTTGTAAGTAATAGTGAAGCCACAGATGTATTACTTACAAACTAGGTTGGACTGCCTACTGCCTCGGA-3′; and shRNA2: 5′-TGCTGTTGACAGTGAGCGCCAGAGTGTGTGATACATGCAATAGTGAAGCCACAGATGTATTGCATGTATCACACACTCTGTTGCCTACTGCCTCGGA-3′) and cloned into the retroviral vector LMP (34). Production of the retrovirus and infections were conducted as described (35) with some modifications. Briefly, packaging Phoenix cells were transfected with 12 μg LMP by CaPO₄ precipitation, and after 48 h, the retrovirus-containing supernatant was harvested and filtered. The 1940 cells were infected with the retrovirus in the presence of 4 μg/ml polybrene, and infected cells were selected by incubation in culture medium containing puromycin for 1 wk.

Cells were metabolically labeled with [³⁵S]Met/Cys as described (22). Where indicated, cells were cultured overnight in medium with or without rIL-10 (PeproTech) at 50 ng/ml before labeling. The efficiency of [³⁵S]Met/Cys incorporation in each sample was determined by counting the amount of radioactivity precipitated with 1% TCA from a small volume of cell lysate. Immunoprecipitation, SDS-PAGE analysis, and Western blot were performed as described (22). To precipitate CstC from blood samples, we used carboxymethylated papain conjugated to agarose beads (Calbiochem).

The concentration of CstC in serum was determined using the mouse DuoSet ELISA kit (R&D Systems) according to the manufacturer’s instructions.

RNA extraction and cDNA synthesis were performed using the RNeasy kit (Qiagen) and the Thermoscript RT-PCR System (Invitrogen), respectively. The sequences of the primers used for cDNA amplification were as follows: cst3 IECS forward: 5′-GCAATGACCAACTTCTCTGGTG-3′; cst3 IECS reverse: 5′-CTTACCAGTTCCTCTTCTGTGC-3′; irf8 forward: 5′-TGACACACACCATTCAGCTTTCT-3′; irf8 reverse: 5′-CATCCGGCCCATACAACTTAG-3′; pu.1 forward: 5′-GGAGAAAGCCATAGCGATCACT-3′; pu.1 reverse: 5′-TCTCAAACTCGTTGTTGTGGACAT-3′; cst3 forward: 5′-CAACAAGGGCAGCAACGAT-3′; cst3 reverse: 5′-CCAGCCACGAGCTGCTTAC-3′; hmbs forward: 5′-GACCTGGTTGTTCACTCCCTGAAG-3′; hmbs reverse: 5′-GACAACAGCATCACAAGGGTTTTC-3′; gapdh forward: 5′-CATTTGCAGTGGCAAAGTGGAG-3′; and gapdh reverse: 5′-GTCTCGCTCCTGGAAGATGGTG-3′.

Real-time PCR reactions were carried out using Power SYBR Green PCR Master Mix (Applied Biosystems) and ABI Prism 7900HT sequence detection system (Applied Biosystems) according to the manufacturer’s instructions. Control reactions without cDNA template were performed alongside test samples to ensure the absence of contaminating genomic DNA. Data were analyzed using the ABI Prism software (Applied Biosystems). For the experiment in Fig. 1D, reactions were performed using the QuantiTect SYBR Green PCR Kit (Qiagen) and a Light cycler (Roche), as per the manufacturer’s instructions. The specific primer pairs for cst3 are as follows: 5′-CGCTCCTTGCTGTTCCTGCT-3′ and 5′-TGCCCTTGTTGTACTCGCTCAC-3′. The expression level for each gene was determined using a standard curve prepared from 10⁻²–10⁻⁶ pg specific DNA fragment.

Mean and SD values and two-tailed Student t tests (unpaired) were calculated using Microsoft Excel software (Microsoft). A p value ≤0.05 was considered significant.

Mouse spleens contain two major types of DC: pDC (CD11c^intCD45RA⁺) and conventional (CD11c^highCD45RA⁻) DCs. The latter group can be subdivided into two subtypes that differ in expression of CD8 (CD8⁺ DC and CD8⁻ DC hereafter). These three populations can also be obtained in bone marrow cultures supplemented with the growth factor Flt3L (23). We have previously shown that among these three DC populations, only CD8⁺ DC contain significant levels of CstC as measured by Western blot (22, 23). However, this analysis did not formally prove that the CD8⁺ DC themselves expressed CstC, as it was possible that the protein was captured from the extracellular medium only by this DC subset.

To establish the origin of CstC contained in CD8⁺ DC, we generated mixed bone marrow chimeras by injecting irradiated C57BL/6 mice with bone marrow from wild-type (C57BL/6) mice expressing the marker Ly5.1 and from CstC-null animals expressing Ly5.1 and Ly5.2. Following reconstitution of the bone marrow compartment in the recipient animals, DC derived from the control and CstC-null bone marrow were purified based on Ly5 expression. Western blot analysis showed that only the DC derived from control bone marrow contained CstC (Fig. 1A). A similar result was obtained in analyses of DC generated in vitro in Flt3L cocultures of control and CstC-null bone marrow (Fig. 1B). These results showed that the CstC contained within the CD8⁺ DC was of endogenous origin rather than endocytosed from the extracellular medium. This was confirmed by metabolic labeling of the DC subtypes followed by immunoprecipitation of CstC, showing that only CD8⁺ DC synthesized the protein (Fig. 1C). Furthermore, real-time PCR analysis of CstC gene (cst3) transcription showed that CstC expression was transcriptionally regulated among DC types (Fig. 1D).

The only regulatory element described in the promoter region of cst3 is an IECS recognized by the transcription factor IRF-8, which was essential and sufficient to drive CstC expression in a transduced macrophage cell line (24). IRF-8 is highly expressed in CD8⁺ DC but not in CD8⁻ DC, consistent with the differential expression of CstC in these two subsets, but it is also highly expressed in pDC (Table I). As pDC do not express CstC, this suggests that some additional factor is responsible for CstC expression only in CD8⁺ DC. One candidate might be the transcription factor PU.1, which was found to synergize with IRF-8 to drive CstC expression in transfected macrophages (24) and is expressed at high levels in CD8⁺ and CD8^- DC but only at low levels in pDC (Table I) (23, 36).

To investigate whether IRF-8 or PU.1 are involved in CstC expression in primary DCs, we purified CD8⁺ and CD8⁻ DC from mouse spleens and examined their physical association with the IECS of the cst3 promoter by ChIP. Assessment of the amount of IECS that was coprecipitated with Abs for IRF-8 or PU.1 was carried out by semiquantitative PCR (Fig. 2A) or real-time PCR (Fig. 2B). The IECS region of CD8⁺ DC was coprecipitated with either IRF-8 or PU.1, indicating that both transcription factors were bound to the CstC promoter in CD8⁺ DC (Fig. 2). Immunoprecipitation of IRF-8 from CD8⁻ DC did not pull down the IECS (Fig. 2) as expected because these cells do not express IRF-8 (Table I). PU.1 immunoprecipitation pulled down a much smaller amount of the IECS from CD8⁻ DC than from CD8⁺ DC (Fig. 2), even though this transcription factor is similarly expressed in both DC subsets (Table I).

Analysis of pDC by ChIP was hampered by the paucity of this population (15% of splenic DC), as it was not possible to purify them in high enough numbers from normal mice for this type of assay. To overcome this limitation, we induced expansion of DC in vivo by injecting mice with a B16 tumor cell line that secretes the DC growth factor Flt3L (28). We then analyzed by ChIP a preparation of CD8⁺ DC and a preparation containing both pDC and CD8⁻ DC in which the number of pDC was adjusted to be equivalent to that of CD8⁺ DC. These two preparations contained the same number of IRF-8⁺ cells (CD8⁺ DC in one, pDC in the other) expressed at equivalent levels (Table I). However, the results of the ChIP assay using the preparation containing both pDC and CD8⁻ DC were similar to those using CD8⁻ DC alone (Fig. 2). This indicated that IRF-8 was not bound to the cst3 IECS in pDC.

These results support the notion that cst3 gene transcription requires binding of both IRF-8 and PU.1 to the IECS, and such binding requires high expression of both transcription factors, a situation found only in CD8⁺ DC (Table I).

The observed binding of IRF-8 to the CstC promoter did not prove per se that IRF-8 was required for transcription, so we sought to assess the effect of IRF-8 elimination. For this, we first used shRNA to knockdown IRF-8 expression in the CD8⁺ DC line 1940. This DC line was derived from CD8⁺ splenic DCs in mice that express the oncogenic SV40 T Ag under the control of the DC promoter CD11c (32) and expresses characteristic markers of CD8⁺ DC (CD11c⁺Clec9A⁺CD24⁺Sirpα⁻CD45RA⁻MHC class II^int (Supplemental Fig. 1). Importantly, the physical interaction of IRF-8 and PU.1 with the cst3 IECS region was recapitulated in this cell line (Fig. 3A), providing a suitable model to evaluate the impact of IRF-8 downregulation on cst3 transcription. We infected 1940 cells with the retrovirus vector LMP encoding GFP alone (control) or with either of two shRNAs specific for irf8. Following the selection of the infected (puromycin-resistant) cells, these were radiolabeled for 30 min with [³⁵S]Met/Cys, and IRF-8 and CstC were immunoprecipitated from normalyzed amounts of radiolysate, along with MHC class I (MHC I) molecules as an internal control. The immunoprecipitates were analyzed by SDS-PAGE and the amount of newly synthesized protein in each sample visualized by autoradiography. Synthesis of IRF-8 was inhibited in cells expressing the shRNAs, but not in those expressing the vector alone (Fig. 3B). Synthesis of CstC, but not that of MHC I, was also inhibited, confirming the requirement of IRF-8 in CstC expression. Cells transfected with shRNA constructs that failed to inhibit IRF-8 did not downregulate CstC synthesis (not shown), proving that the effect of the two inhibitory shRNAs that worked was specific.

Although IRF-8 knockout mice have been described, it is not possible to assess the impact of this deficiency on CD8⁺ DC because these cells do not develop in the absence of IRF-8 (37). However, it is possible to generate in vitro monocyte-derived DCs from IRF-8–deficient bone marrow incubated with GM-CSF and IL-4 (GM/IL-4–BMDC) (38). We found that this type of DC express IRF-8 and CstC (Fig. 3C). Consistent with the experiments employing shRNA, the synthesis of CstC was reduced in GM/IL-4–BMDC lacking IRF-8 expression (Fig. 3C).

Certain inflammatory conditions such as atherosclerosis are accompanied with reduction in CstC levels in atherosclerotic lesions and in circulation, and this in turn associates with disease exacerbation (7). Therefore, we examined the effect of inflammation on CstC production by DC. We injected mice i.v. with the synthetic analog of bacterial DNA, CpG, a treatment that causes systemic activation of DC (39). After 16 h, we measured the synthesis of MHC I molecules in CD8⁺ DC, which was upregulated upon activation as previously described (Fig. 4A) (40). In contrast, synthesis of CstC and, to a lesser extent, IRF-8, was reduced (Fig. 4A). This correlated with reduced transcription of irf8 and cst3 genes in activated DC (Fig. 4B). Interestingly, pu.1 transcription was not altered (Fig. 4B), providing an opportunity to further assess binding of IRF-8 and PU-1 to the IECS motif of the cst3 promoter and its role on gene expression. ChIP assays revealed that binding of IRF-8 was reduced in activated CD8⁺ DC, whereas binding of PU.1 did not change (Fig. 4C). This supports the notion that PU.1 is not sufficient to drive CstC expression; efficient binding of IRF-8 is required, so that downregulation of IRF-8 expression caused by DC activation was sufficient to cause a drastic reduction in CstC synthesis.

To assess the effect of systemic inflammation on the level of circulating CstC, we measured the amounts of CstC in the serum of animals treated with CpG 16 or 48 h earlier. We precipitated CstC from serum prepared from each mouse using carboxymethylated papain coupled to agarose, a reagent that can only recognize active CstC. The precipitates were run in SDS-PAGE, and CstC was detected by Western blot (Fig. 4D). The protein band identified as CstC was not observed if the serum sample was from a CstC-deficient mouse, confirming that this band indeed corresponded to CstC. No additional CstC could be recovered by immunoprecipitation using anti-CstC Ab from the serum samples after removal of the papain-reactive species, indicating that all of the CstC in serum was active (Fig. 4D). The amount of CstC relative to total protein was greatly reduced in the sera of CpG-treated mice (Fig. 4D). This reduction in serum CstC concentration was confirmed by ELISA (Fig. 4E). Therefore, downregulation of CstC synthesis by CD8⁺ DC in response to inflammation correlated with reduced levels of circulating CstC in serum. This suggested that either most of the circulating CstC is produced by CD8⁺ DC or that the mechanisms controlling CstC synthesis in CD8⁺ DC are recapitulated in other cell types that contribute to the overall levels of serum CstC.

The effect of systemic CpG administration on IRF-8 and CstC expression in DC could be caused by direct encounter of CpG or by a factor produced in response to CpG. Strikingly, direct exposure of splenic CD8⁺ DCs to CpG in vitro increased irf8 and cst3 expression (Fig. 4F), suggesting that the downregulation of their expression in vivo was indeed caused by an intermediate cytokine.

Injection of CpG causes the release of both proinflammatory (e.g., IFN-α and TNF-α) and anti-inflammatory (e.g., IL-10, TGF-β) cytokines (41). GM/IL-4-BMDC incubated with IFN-α or TNF-α upregulated cst3 expression, whereas IL-10 induced its downregulation, and TGF-β caused no major change (Fig. 5A). To investigate which of these effects was mediated by IRF-8, we repeated the same experiment using IRF-8–deficient GM/IL-4-BMDC. Consistent with the results of the analysis of protein expression (Fig. 3C), these cells expressed lower basal levels of cst3 than their wild-type counterparts (Fig. 5A). Furthermore, IFN-α and TNF-α also caused upregulation of cst3 expression in IRF-8–deficient GM/IL-4-BMDC, suggesting an IRF-8–independent process (Fig. 5A). Importantly, IL-10 did not cause downregulation of cst3 expression in IRF-8–deficient GM/IL-4-BMDC, indicating that the suppressing effect of this cytokine was mediated by IRF-8. Indeed, incubation with IL-10 in vitro caused reduced irf8 and cst3 transcription in 1940 cells (Fig. 5B) and downregulation of IRF-8 and CstC synthesis in splenic CD8⁺ DC (Fig. 5C).

Because IL-10 is readily detectable in the serum of mice treated with CpG (Fig. 5D), and our results indicated that this cytokine downregulates expression of cst3, we hypothesized that the downregulation of serum CstC concentration in mice treated with CpG might be mediated by IL-10. This was tested using IL-10 knockout mice. These mice are hyperreactive to systemic inflammatory stimuli and do not survive after i.v. injection of the dose of CpG used in the previous experiments, so we used a lower dose and analyzed the mice after only 7 h. Even after this milder treatment, the CD8⁺ DC of control mice synthesized lower levels of IRF-8 and CstC, but no effect was observed in the CD8⁺ DC of IL-10–deficient mice (Fig. 5E). MHC class II synthesis was downregulated in the CD8⁺ DC of both sets of CpG-injected mice (Fig. 5E), indicating that DC activation had occurred (42). We also determined the role of IL-10 in downregulation of serum CstC in mice treated with CpG. In contrast to control mice, IL-10–deficient mice had a higher concentration of CstC in their serum after CpG treatment, although this did not reach statistical significance. This result suggests that CpG injection causes the release of cytokines that induce CstC secretion to the extracellular medium (as IFN-α and TNF-α did in vitro; Fig. 5A), but the simultaneous or subsequent production of IL-10 overrides this effect, causing downregulation of CstC synthesis and therefore a net reduction of its concentration in serum.

The mechanisms regulating CstC expression are of major clinical importance. Firstly, low levels of CstC are associated with excessive extracellular proteolysis in atherosclerosis or aortic aneurysm (7). Secondly, CstC plays other, still poorly defined regulatory roles as an inducer of NO release (15) or as a TGF-β antagonist (16). Thirdly, CstC can have causal (HCCAA) (18) or protective (Alzheimer’s disease) (19, 20) roles in brain amyloidosis. Finally, the serum concentration of CstC has become a widely used parameter of renal function (43). Despite this, little is known about the mechanisms that control CstC expression. In this study, we have described the critical function of IRF-8 in cst3 transcription and the role of IL-10 in the downregulation of CstC production during inflammation.

The cst3 promoter was previously considered to lack regulatory elements (21), but our previous observation that macrophages (data not shown) and DC (22) express much more CstC than lymphocytes suggested otherwise. Furthermore, CstC was abundant within CD8⁺ DC but not in the closely related CD8⁻ DC (22, 23). In this study, we have extended our analyses to pDC, showing that this DC subset also expresses little CstC. We confirmed that most of CstC contained in CD8⁺ DC in steady state is of endogenous origin. Indeed, only this DC subset contained high levels of cst3 transcripts. This pattern of expression has allowed us to assess the role of two transcription factors reportedly involved in CstC expression, IRF-8 and PU.1 (24).

We confirmed using shRNA that cst3 transcription requires expression of IRF-8. This explains why CD8⁻ DC, which lack IRF-8 (23), do not express CstC. However, expression of IRF-8 was not sufficient for cst3 transcription because pDC also express IRF-8 at levels comparable to those in CD8⁺ DC, yet they do not express CstC. Previous reports (36) and our own results indicate that this is because simultaneous high expression of PU.1 and binding of this transcription factor to the IECS are also required to promote IRF-8–mediated cst3 transcription. Thus, an IRF-8 mutant that could not associate with PU.1 bound poorly to the IECS (44). Because the level of PU.1 in pDC is one fifth to one tenth of that observed in CD8⁺ DC (36), and the role of PU.1 is dose dependent (45), the most likely explanation for the lack of cst3 transcription in pDC is that these cells express too little PU.1 to promote efficient IRF-8 binding to the cst3 IECS region.

Given the critical role of IRF-8 in regulation of the cst3 gene, it is not surprising that factors that induce changes in IRF-8 expression in turn affect cst3 transcription. Indeed, CD8⁺ DC incubated in vitro with the inflammatory TLR9 ligand CpG upregulated transcription of irf8 and, consequently, cst3. We do not know which signaling pathway was responsible for irf8 upregulation in vitro. TLR9 engagement triggers multiple changes in DC, including the secretion of cytokines that might have secondary autocrine effects on the DC themselves (46). Upregulation of irf8 transcription might be mediated by the first wave of TLR9-initiated signals or by subsequent signals induced by secondary cytokines. Surprisingly, though, treatment with CpG in vivo caused downregulation of irf8 transcription and CstC production in CD8⁺ DC. In this case, we could demonstrate that CstC downregulation was not directly caused by CpG encounter, but required IL-10. Two proinflammatory cytokines released in response to CpG injection, IFN-α and TNF-α, stimulated cst3 expression via an IRF-8–independent pathway, but in vivo their effect was overridden by IL-10, which caused downregulation of cst3 expression. The downregulatory effect of IL-10 was mediated by IRF-8, at least in vitro. Other cytokines may also be capable of inducing downregulation of cst3 expression, for instance IL-6 (47), but if this is the case, they cannot compensate for the absence of IL-10 in vivo because in mice deficient in the latter, CpG injection caused no downregulation of CstC synthesis or serum concentration. We conclude that IL-10 is the major mediator of CstC downregulation in response to inflammation in vivo.

Our analysis of the mechanisms controlling IRF-8 and CstC expression has been based on CD8⁺ DC, but the conclusions of our studies appear applicable to the whole organism. Thus, induction of systemic inflammation with CpG led to reduced concentration of CstC in blood serum, and this effect was mediated by IL-10. This implies that either CD8⁺ DC are the main contributors to maintenance of CstC concentration in blood or that expression of CstC by other cells is regulated by similar mechanisms to those operating in CD8⁺ DC. The second explanation is more likely and consistent with the observation that macrophages also express abundant CstC in an IRF-8–dependent manner (24).

In conclusion, our results demonstrate that expression of CstC is more tightly regulated than generally believed. In addition to varying among closely related cell types, it is also controlled by external factors. This raises the possibility of designing strategies to reduce CstC expression as a treatment for conditions associated with amyloid formation (e.g., HCCAA). Conversely, treatments that promote CstC expression might help restore the extracellular protease balance and reduce the damage caused by excessive proteolysis (e.g., atherosclerosis). Our results also raise caution about the use of serum CstC concentration as a measurement of the glomerular filtration rate. Serum CstC is removed from the bloodstream by glomerular filtration and catabolized in the renal tubules (25). The concentration of serum CstC has thus become a useful indirect parameter of kidney function, in the assumption that such concentration is directly dependent on the rate of CstC elimination (25). Our results show that CstC production is also regulated, so a change in the level of this protein in serum may not necessarily reflect an alteration in kidney function.

We thank Dannielle Cooper and Melissa Smith for assistance in animal injection and organ extraction.

The authors have no financial conflicts of interest.