Molecular Mechanisms for Synchronized Transcription of Three Complement C1q Subunit Genes in Dendritic Cells and Macrophages*

Background: C1q is assembled in a 1:1:1 ratio from three subunits demanding synchronized expression of three genes in a cluster; C1q deficiency causes lupus. Results: A PU.1/IRF8-stimulated “core promoter” appears to exist in this gene cluster. Conclusion: A core promoter helps synchronize transcription of three C1q genes. Significance: This helps explain the highly conserved clustering of three C1q genes in all animal species, including zebrafish. Hereditary homozygous C1q deficiency is rare, but it almost certainly causes systemic lupus erythematosus. On the other hand, C1q levels can decline in systemic lupus erythematosus patients without apparent C1q gene defects and the versatility in C1q production is a likely cause. As an 18-subunit protein, C1q is assembled in a 1:1:1 ratio from three different subunits. The three human C1q genes are closely bundled on chromosome 1 (C1qA-C1qC-C1qB) and their basal and IFNγ-stimulated expression, largely restricted to macrophages and dendritic cells, is apparently synchronized. We cloned the three gene promoters and observed that although the C1qB promoter exhibited basal and IFNγ-stimulated activities consistent with the endogenous C1qB gene, the activities of the cloned C1qA and C1qC promoters were suppressed by IFNγ. To certain extents, these were corrected when the C1qB promoter was cloned at the 3′ end across the luciferase reporter gene. A 53-bp element is essential to the activities of the C1qB promoter and the transcription factors PU.1 and IRF8 bound to this region. By chromatin immunoprecipitation, the C1qB promoter was co-precipitated with PU.1 and IRF8. shRNA knockdown of PU.1 and IRF8 diminished C1qB promoter response to IFNγ. STAT1 instead regulated C1qB promoter through IRF8 induction. Collectively, our results reveal a novel transcriptional mechanism by which the expression of the three C1q genes is synchronized.

The complement system is an enzymatic cascade, consisting of Ͼ20 abundant plasma proteins, and its activation on microbial pathogens leads to killing and clearance through the formation of membrane lytic complexes, opsonins, and anaphylatoxins (1)(2)(3). These complement proteins are mostly of hepatic origin (4), but C1q is extrahepatically produced by tissue macrophages and dendritic cells (DCs) 2 (5)(6)(7)(8). These are potent phagocytes and major antigen-presenting cells instrumental to the induction of adaptive immunity and tolerance (9 -11). The evolutionary advantages and transcriptional mechanisms that sustain this distinct mode of C1q production by antigen-presenting cells are not understood.
C1q recognizes antibody-bound pathogens to trigger the complement classical pathway (1)(2)(3). In addition, C1q also recognizes apoptotic cells and thus augments phagocytosis (12)(13)(14). It binds to membrane blebs on apoptotic cells and, on the molecular level, C1q can bind to surface-exposed calreticulin, which is otherwise a resident protein of the endoplasmic reticulum in live cells (12,13). Apoptotic cells absorb polyclonal IgM, which then recruits C1q (14). Apoptotic cells are sources of self-antigens, and their impaired clearance by phagocytes, which can occur in C1q deficiency, can elicit autoantibody production (15). The dominant manifestation in genetic C1q deficiency is the development of systemic lupus erythematosus (SLE) (16 -18).
In addition to these ligand-dependent C1q functions, which can involve complement activation and phagocytosis, ligandindependent C1q functions have also been reported (19). For example, both soluble and solid phase C1q were shown to affect monocyte differentiation into DCs so that these cells were less immunostimulatory or more tolerogenic (20,21). Second, C1q Ϫ/Ϫ mouse DCs produced IL-12 abnormally, which affected IFN␥ induction from T cells (22,23). These can impact on the induction of immunity or autoimmunity.
Homozygous hereditary C1q deficiency is rare but it represents the strongest known genetic risk for SLE pathogenesis (24). Without known gene defect, acquired C1q deficiency can occur (25,26). C1q consumption by immune complexes and anti-C1q autoantibodies are likely causes (27,28 affect C1q production (19), which can contribute to the decline of C1q levels in SLE patients. C1q is a delicate 18-subunit assembly in 1:1:1 ratio from three types of subunits (29). The C1q genes are highly clustered (C1qA-C1qC-C1qB) within a genomic region of ϳ25 kb (30). How the transcription of these three genes is synchronized in macrophages and DCs is relevant to SLE pathogenesis but remains to be investigated. Here, we report a 53-bp region in the C1q gene cluster that responds to IFN␥ through PU.1 and IRF8, two DC/macrophage-associated transcription factors, and potentially regulates all three genes from the center of this gene cluster.

EXPERIMENTAL PROCEDURES
RNA Ligase-mediated Rapid Amplification of 5Ј cDNA Ends (RLM-RACE)-The transcription start site (TSS) for each C1q gene was determined using the RLM-RACE kit (Invitrogen) with RNA isolated from cultured human macrophages (1-5 g) (21). After dephosphorylation with calf intestinal phosphatase, RNA was decapped at the 5Ј end with tobacco acid pyrophosphatase and a 44-bp GeneRacer RNA oligo was then ligated to the 5Ј ends. cDNA was synthesized from the RNA using C1q gene-specific primers (supplemental data). 5Ј end sequences were PCR-amplified using common forward primers embedded in the GeneRacer oligonucleotides and respective reverse gene-specific primers (supplemental data). The PCR products were cloned using the TOPO cloning reaction kit (Invitrogen) and sequenced using the BigDye Terminator v3.1 cycle sequencing reagents (Applied Biosystems).
Transfection-These cloned C1q gene promoters were transfected into the mouse RAW264.7 macrophage cells (ATCC) following a published protocol (31). Briefly, RAW264.7 cells were cultured in DMEM supplemented with 10% (v/v) bovine calf serum (HyClone, Logan, UT), 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine at 37°C, and 5% CO 2 . For transfection, cells were harvested and resuspended in serum-free RPMI 1640 medium (2.5 ϫ 10 7 cells/ml). The C1q promoter plasmid (10 g) was mixed in a 40-l volume with the pRL-␤-actin promoter plasmid (1 g). In a 4-mm cuvette, the plasmid mix was further mixed with RAW264.7 cells (0.4 ml), which was then pulsed in the Gene Pulser Xcell system (Bio-Rad) at 300 V with 975 F capacitance and infinite resistance. The cells were left stand for 5 min and then resuspended in RPMI 1640 medium (8.5 ml) containing 10% (v/v) bovine calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine, and 10 M chloroquine. Cells were cultured in 48-well plates (0.5 ml/well) overnight at 37°C and 5% CO 2 and then changed to fresh RPMI 1640 medium (1 ml/well) where chloroquine was replaced by 1.2% (v/v) dimethyl sulfoxide. After 24 h, luciferase activities were measured. Where cells were stimulated with mouse IFN␥ (10 ng/ml), it was added 4 h after the change of medium.
Luciferase Assay-The Dual-Luciferase reporter assay kit (Promega) was used to determine firefly luciferase expression resulting from the C1q gene promoters and constitutive Renilla luciferase expression under a ␤-actin promoter. The transfected RAW264.7 cells, with or without IFN␥ treatments, were washed in PBS and lysed in the passive lysis buffer (65 l/well) for 45 min with shaking. The cell lysate (2 l) was first mixed with the firefly luciferase substrate (luciferase assay reagent II, 25 l) and then measured in the TD-20/20 luminometer (Turner Designs). The Renilla luciferase substrate was then added (Stop & Glo, 25 l) to measure the control ␤-actin promoter activity. Relative luciferase activity was calculated for each experiment (firefly luciferase reading/Renilla luciferase reading ϫ 100), and data were presented as mean Ϯ S.D. of triple experiments.
Mutagenesis-Potential cis-acting regions in the cloned C1q gene promoters were initially mapped by 5Ј and 3Ј deletions of the promoters. Specified regions were PCR-amplified using primers listed in supplemental data. 5Ј deletion of the pGL-C1qB-Luc vector to Ϫ273 bp yielded a basic C1qB gene promoter (B273). Point mutations were introduced to the B273 promoter using the QuikChange® site-directed mutagenesis kit (Stratagene) and primers listed in supplemental Table S4. In some experiments, the B273 promoter was cloned at the 3Ј end of the firefly luciferase gene with C1qA or C1qC promoters at the 5Ј end.
shRNA Knockdown-Oligonucleotides representing PU.1, IRF1, IRF8, and STAT1 target sequences (supplemental data) were synthesized, and double-stranded DNA was generated by annealing complementary oligonucleotides. These were cloned into the pSUPER.basic vector (OligoEngine, Seattle, WA). To transfect RAW264.7 cells, each shRNA plasmid (10 g) was mixed with the B273 plasmid (1 g) and the control ␤-actin promoter plasmid (1 g) in a 40-l volume for electroporation. Cells were cultured to measure luciferase activities. To evaluate shRNA knockdown of the transcription factors, cell lysates were subjected to Western blotting.
DNA Binding Assay-A double-stranded, 53-bp DNA fragment (Ϫ133 to Ϫ81 bp) was generated by annealing a sense oligonucleotide (5Ј-ctcatttacagtaaatccagtgggttgcagaaataggacctgaaactgcctga-3Ј) and a biotinylated antisense oligonucleotide (5Ј-biotin-tcaggcagtttcaggtcctatttctgcaacccactggatttactgtaaatgag-3Ј). The two oligonucleotides were mixed at 1 pmol/l each in an annealing buffer (10 mM Tris, pH 8.0, 1 mM EDTA, 50 mM NaCl) and after boiling for 5 min were cooled to room temperature. On the other hand, nuclear extracts were prepared from RAW264.7 cells, with or without IFN␥ treatment, using the NE-PER nuclear extraction reagents (Thermo Scientific, Rockfold, IL).
Streptavidin-agarose resins (50 l of packed volume) were washed and resuspended in 100 l of TEN buffer (10 mM Tris, pH 8.0, 1 mM EDTA, and 100 mM NaCl). Half of the resins were mixed with the 53-bp probe (30 l) overnight at 4°C and, as a control, the other half of the resins were mixed with the annealing buffer. After washing once with TEN buffer and twice with TGEDN buffer (20 mM Tris, pH 8.0, 10% (v/v) glycerol, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, and 0.1% (v/v) Triton X-100), the resins were resuspended in 0.5 ml of TGEDN buffer and incubated for 4 h at 4°C with the nuclear extract (200 g proteins) and sonicated salmon sperm DNA (25 g). The resins were washed five times each with 0.5 ml of 0.5ϫ TGEDN buffer and resuspended in 50 l of elution buffer (0.5% (w/v) SDS and 1 M NaCl). Proteins were eluted by heating at 50°C for 20 min and then separated on 10% (w/v) SDS-PAGE gels. As controls, total nuclear extracts were loaded on the gels. Precipitated PU.1, IRF1, IRF8, and STAT1 were detected by Western blotting.
ChIP-Macrophages (ϳ1 ϫ 10 7 ) were cultured in a 100-mm plate with macrophage colony-stimulating factor (20 ng/ml) (21). On day 6, cells were fixed by adding formaldehyde to 1% (w/v). After gently mixing for 10 min, reactions were stopped by adding glycine to 150 mM. After 5 min, cells were washed three times with PBS, harvested, and resuspended in 1 ml of L1 buffer (50 mM Tris, pH 8.0, 2 mM EDTA, 1% (v/v) Triton X-100, 10% (v/v) glycerol, and protease inhibitors). After incubation on ice for 5 min, the insoluble nuclear fraction was pelleted and resuspended in 1 ml of L2 buffer (50 mM Tris, pH 8.0, 5 mM EDTA, 1% (w/v) SDS, and protease inhibitors), and 0.2-ml aliquots were made in 1.5-ml Eppendorf tubes. Sonication was performed in the Bioruptor at its high setting (Diagenode, Inc., Liege, Belgium) for 15 cycles (30 s on/30 s off) in an ice bath. Soluble nuclear lysate was obtained by centrifugation for 30 s at 13,000 rpm and 4°C.
The nuclear lysate (50 g) was diluted 1:10 with a dilution buffer (50 mM Tris, pH 8.0, 5 mM EDTA, 0.2 M NaCl, and 1% (v/v) Triton X-100). A 1:10 dilution of the L2 buffer was used in further incubation and washing steps. Protein A-Sepharose resins (20 l) were washed and mixed for 3 h at 4°C with the PU.1, IRF1, IRF8, or STAT1 antibody (10 g/ml). The resins were similarly incubated with non-immune rabbit IgG (Bio-Rad) and used to absorb the nuclear lysate for 3 h at 4°C. After addition of sonicated salmon sperm DNA (25 g), the preabsorbed lysate was then incubated overnight at 4°C with resins loaded with the PU.1, IRF8, STAT1, or IRF1 antibodies or, as controls, resins loaded with the anti-His antibody or resins without antibody.

Coordinated Transcription of Three Endogenous C1q
Genes-The fact that C1q is assembled in a 1:1:1 ratio from its C1qA, C1qB, and C1qC subunits (29) demands synchronized transcription of the three C1q genes. Furthermore, these mechanisms apparently also ensure that C1q gene transcription is largely restricted to DCs and macrophages. As shown in Fig. 1A, DCs constitutively express all three mRNA species that were, after IFN␥ treatment, all up-regulated (Fig. 1A). Similar basal and IFN␥-stimulated expression of the three C1q mRNA species was observed in macrophages (Fig. 1B). The magnitudes of up-regulation were the highest with the C1qB mRNA and the lowest for C1qA mRNA. Overall, the three C1q genes showed largely coordinated basal and IFN␥-stimulated transcription. Mechanistically, this could be achieved by having similar relevant cis-acting elements in all three gene promoters. The fact that the three genes are in a rare but highly conserved chromosomal cluster (Fig. 2) implies that synchronization may also be achieved through a core element inside this gene cluster (30,32).
Cloned C1qA and C1qC Gene Promoters Lack Basal and IFN␥-stimulated Activities of Endogenous Genes-To examine the first hypothesis, i.e. that the three C1q gene promoters function independently and synchronization is achieved by having the relevant cis-acting elements duplicated in all three promoters, these promoters were each cloned into a firefly luciferase reporter plasmid (pGL3-basic) and transfected into the mouse RAW264.7 macrophage cells. Cloning was performed after validation of the TSS for each C1q gene by RLM-RACE (supplemental data). In the transfected RAW264.7 cells, the three promoters displayed varying degrees of basal activity with the C1qB promoter being many times stronger than the C1qC promoter (Fig. 2). Similar to its endogenous counterpart (Fig. 1), the cloned C1qB promoter was highly inducible by IFN␥ (Fig.  2C). In contrast, the cloned C1qC promoter dramatically departed from the activity of its endogenous counterpart by exhibiting weak basal activity, which was further suppressed by IFN␥ (Fig. 2). The C1qA promoter was also suppressed by

PU.1 and IRF8 Regulation of C1q Genes
IFN␥. These results are not in favor of the first hypothesis that synchronized transcription is achieved by having the relevant cis-acting element duplicated in the three independent C1q gene promoters.
Search for Discrete Regions for IFN␥ Inhibition in Cloned C1qA and C1qC Promoters-A series of 5Ј deletions were introduced in the C1qA and C1qC gene promoters in an attempt to determine whether specific regions exist that mediated IFN␥ inhibition of these promoters. For the C1qA promoter (Ϫ2497 bp), deletions were introduced at Ϫ1800, Ϫ1067, Ϫ627, and Ϫ231 bp, whereas the C1qC promoter (Ϫ2396 bp) was truncated at Ϫ1612, Ϫ997, Ϫ615, and Ϫ258 bp. A progressive 5Ј deletion of the C1qA promoter gradually increased its basal activity, e.g. the Ϫ627-bp promoter was 3-fold stronger than the Ϫ2497-bp promoter (Fig. 2B). Further deletion to Ϫ231 bp had no additional effect. However, all of these truncated C1qA pro- DCs and macrophages were cultured from blood monocytes and were, at day 6, cultured for a further 24 h either untreated (PBS) or treated with human IFN␥ (100 ng/ml). RNA was isolated for real-time PCR using primers specific for C1qA, C1qB, and C1qC. Results were normalized to the levels of ␤-actin mRNA in each experiment, and data were presented as folds of C1q mRNA induction by IFN␥ taking that detected in untreated cells as 1. Conventional PCR was also performed, and the PCR products were detected on 1% agarose gels. were specified and cloned into the pGL3-basic vector (A). The cloned C1qA (B), C1qB (C), and C1qC (D) gene promoters were transfected into RAW264.7 cells to measure the basal activities, and cells were treated with mouse IFN␥ (10 ng/ml) to measure IFN␥-stimulated activities. After 5Ј deletions, the shortened promoters were examined similarly. In each experiment, a Renilla luciferase reporter plasmid under the ␤-actin promoter was co-transfected to normalize the firefly luciferase activities derived from C1q promoters. Data were presented as means Ϯ S.D. of triplicate experiments. moters remained inhibited by IFN␥, showing that the remaining promoter region of the C1qA gene, despite the increased basal activity, still lacks adequate cis-acting elements to confer IFN␥ induction as observed in the endogenous C1qA gene.
Deletion of the C1qC promoter from Ϫ2396 to Ϫ1612 bp dramatically raised the otherwise trivial basal activity, but further deletion to Ϫ997, Ϫ615, and Ϫ258 bp had no additional effects (Fig. 2D). All of these truncated promoters also remained inhibited by IFN␥. It should be noted that even the raised basal activity of the truncated C1qC promoter remained much lower than that of the C1qB promoter. Furthermore, the truncated C1qA and C1qC promoters remained inhibited by IFN␥. It shows that the C1qA and C1qC genes are not effectively regulated by their 5Ј regions upstream of the TSS, and additional endogenous mechanisms may exist, in the chromosomal context, for synchronized transcription of the three C1q genes.
Identification of Novel IFN␥-stimulated Region in C1qB Promoter-Similar 5Ј deletions of the C1qB promoter (Ϫ2303 bp) at Ϫ1724, Ϫ1068, or Ϫ684 bp did not affect its strong basal and IFN␥-stimulated activities. Deletion to Ϫ273 bp only slightly reduced its basal activity, but the IFN␥-stimulated activity remained intact. Deletion further to Ϫ37 bp halved the basal activity, but it completely eliminated the IFN␥-stimulated activity (Fig. 2C). It suggests that the 236-bp region between Ϫ273 and Ϫ37 bp is responsible for its IFN␥-stimulated activity and may also partially contributes to its basal activity (Fig. 2C).
Inspection of this region using the MatInspector program (Genomatix), revealed a TSS-proximal chimeric GAS-ISRE sequence and a c-Rel site further upstream (Fig. 3). 5Ј deletion to Ϫ141 bp removed the c-Rel site but still retained the GAS-ISRE sequences in the promoter. This slightly reduced the IFN␥-stimulated activity (Fig. 4A). However, deleting a further 8 bp to Ϫ133 bp began to impair the IFN␥-stimulated activity: a 50% reduction compared with the Ϫ273-bp promoter (Fig. 4A), suggesting a significant contribution of the 11-bp segment between Ϫ144 and Ϫ133 bp, which is 5Ј outside of the GAS-ISRE region, to the ability of the promoter to respond to IFN␥. Deletion of another 8 to Ϫ125 bp still maintained the integrity of the GAS-ISRE region (Fig. 4B), but it completely abolished the IFN␥-stimulated activity (Fig. 4A). These results suggest that the putative GAS-ISRE region is inadequate for the IFN␥ response of the C1qB gene if it plays any roles.
To further assess this GAS-ISRE region, 3Ј deletions were made using the Ϫ133-bp promoter as a template. 3Ј deletion to Ϫ28 bp, which removed the TSS, dramatically reduced its basal activity, but the promoter still retained some IFN␥-inducible activity (Fig. 4B). Progressive deletion to Ϫ45, Ϫ72, and Ϫ81 bp showed no impairment of this IFN␥-stimulated activity (Fig.  4B). However, deletion to Ϫ90 bp, which began to disrupt the ISRE region, completely abolished IFN␥-stimulated activity. These 5Ј and 3Ј deletion studies identified a basic 53-bp region in the C1qB promoter (Ϫ133 to Ϫ81 bp) that was sufficient and essential to confer most of IFN␥-stimulated activity to the promoter.
The involvement of different regions along this 53-bp fragment to IFN␥-stimulated activity was further evaluated by sitedirected mutagenesis (Fig. 4C). Most of these mutants showed varying degrees of impairment in the IFN␥-stimulated activity. The activity appeared to be completely abolished by some single nucleotide mutations, i.e. A96G, A102G, A120G, A121G, T122C, T127C, and T128C (Fig. 4D). It shows that the integrity of this 53-bp element is essential for the C1qB gene promoter to respond to IFN␥. Interestingly, although the A96G mutation completely inactivated this element, mutations at the flanking nucleotides, i.e. C95T and G97A, substantially raised the basal activity of the promoter without compromising IFN␥-stimulated activity. This implies a potent regulatory role of this CAG trinucleotide segment in C1q gene transcription.
3Ј B273 Promoter Can Correct 5Ј C1qA and C1qC Promoters across Luciferase Reporter Gene-At this point, we evaluated the second hypothesis, i.e. that the expression of the three C1q genes is synchronized through a core cis-acting element within this gene cluster. We reckoned that the B273 promoter was such a candidate element. B273 was cloned at the 3Ј end of the luciferase reporter gene in the pGL-C1qA-Luc and pGL-C1qC-Luc vectors, which had 5Ј C1qA and C1qC promoters, respectively (Fig. 5A). This placed the luciferase gene under the influence of a 5Ј promoter (C1qA or C1qC) and a 3Ј B273 promoter (Fig. 5A). As a control, B273 was cloned at the 3Ј end of the luciferase gene, which lacked a 5Ј promoter (Fig. 5A). The B273 promoter clearly exhibited both basal and IFN␥-stimulated activities from the 3Ј end of the luciferase gene in the absence of a 5Ј promoter (Fig. 5B). When the C1qA promoter was present at the 5Ј end, the 3Ј B273 promoter had no obvious effect on its basal activity across the luciferase gene, but it reverted the 5Ј C1qA promoter from a negative to a slightly positive response to IFN␥ (Fig. 5B). Likewise, the 3Ј B273 promoter also reverted the 5Ј C1qC promoter from inhibition to induction by IFN␥ (Fig. 5B). To a lesser extent, a shorter (Ϫ258 bp) C1qC promoter was similarly superseded by the 3Ј B273 promoter. This experimental design mimicked the natural chromosomal ori-  Predicted cis-acting elements in the B273 promoter. The sequence of the Ϫ273-bp C1qB promoter (B273) was analyzed using MatInspector software, and the predicted cis-acting elements were highlighted. Two noticeable elements identified are the GAS-ISRE chimeric site proximal to the TSS (ϩ1) and an upstream c-Rel site. OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40  Ϫ125, and Ϫ119 bp were made to identify the 5Ј boundary of the IFN␥-stimulated region in the C1qB promoter. B, 3Ј deletions were made at Ϫ28, Ϫ45, Ϫ72, Ϫ81, and Ϫ90 bp to identify the 3Ј boundary of the IFN␥-stimulated site. C, illustration of site-directed mutagenesis of the B273 promoter to determine the involvement of specific nucleotides for B273 response to IFN␥. D, basal and IFN␥-stimulated activities of site-directed B273 mutants. Data were presented at mean Ϯ S.D. of triplicate experiments. entation of the B273 promoter relative to the C1qA and C1qC genes, and the results suggest that the B273 promoter, which may fall in the usual C1qB promoter region, may actually act as a core promoter element for all three C1q genes in vivo.
Knockdown of IRF1 in RAW264.7 cells showed no inhibition to the IFN␥ -stimulated activity of the B273 promoter (Fig. 6A). On the contrary, it slightly but consistently increased B273 response to IFN␥. Knockdown of STAT1, IRF8, or PU.1 all markedly impaired B273 response to IFN␥ (Fig. 6). Therefore, these three transcription factors are required for the C1qB promoter to respond to IFN␥.
PU.1 and IRF8 Bind to 53-bp IFN␥-stimulated Element-To determine whether PU.1, IRF8, and STAT1 directly interact with the C1qB promoter, we synthesized the 53-bp element as a double-stranded fragment to examine whether it was able to pull down these transcription factors from the nuclear extracts. It was synthesized with a 3Ј-biotin tag and then immobilized on streptavidin-Sepharose resins. From the nuclear extracts of untreated macrophages, none was co-precipitated (Fig. 7). However, when the macrophages were treated with IFN␥ for 24 h, IRF8 and PU.1 were both co-precipitated with the 53-bp element (Fig. 7). IRF1 and STAT1 remained absent from these precipitates. As a control, the resins without the 53-bp element immobilized failed to pull down any of these transcription factors. These results show direct IRF8 and PU.1 interaction with the 53-bp element and also suggest that the role for STAT1 in C1qB promoter regulation is indirect.
STAT1 Regulates IRF8 Expression-A likely mechanism by which STAT1 may regulate the C1qB promoter, without direct association with the 53-bp element, is its ability to regulate transcription factors (e.g. PU.1 or IRF8) that directly interact with this element. PU.1 was found constitutively expressed in RAW264.7 cells (Fig. 6), but IRF8 showed little expression unless the cells were stimulated with IFN␥ (Fig. 8). In fact, it is known that IRF8 expression is regulated by STAT1 (38).
Whether STAT1 indeed regulates IRF8 expression in RAW264.7 cells was assessed using shRNA. IRF8 mRNA was effectively induced in IFN␥-treated RAW264.7 cells, and this was effectively knocked down when IRF8 shRNA was co-transfected (Fig. 8). As a control, it was not affected by scramble shRNA. This IFN␥-induced IRF8 mRNA expression was also suppressed by STAT1 shRNA, showing that STAT1 is required for IRF8 induction by IFN␥ in RAW264.7 cells. Therefore, PU.1 and IRF8 directly interact with the C1qB promoter upon IFN␥ stimulation, but IRF8 expression is dependent on IFN␥-induced STAT1 signaling.
Association of Endogenous PU.1 and IRF8 with C1q Promoters-Whether PU.1 and IRF8 bind to the C1qB promoter in vivo was examined by ChIP using macrophages. After formaldehyde fixation, nuclei were isolated from macrophages, sheared, and immunoprecipitated using PU.1, IRF8, STAT1, or IRF1 antibodies. C1qB promoter sequences were sought in the precipitated chromatin fragments by PCR using primers spanning Ϫ167 and Ϫ31 bp, which contained the 53-bp element. C1qA and C1qC gene promoter sequences were also PCR-amplified from the precipitated chromatin fragments, i.e. C1qA (Ϫ122 to ϩ 79 bp) and C1qC (Ϫ110 to ϩ81 bp). Immunoprecipitation of IRF1 and STAT1 failed to co-precipitate any of the three C1q gene promoters regardless of IFN␥ treatment (Fig. 9).  OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40

PU.1 and IRF8 Regulation of C1q Genes
With regard to IRF8, it failed to precipitate the C1qA and C1qC promoters. However, it co-precipitated with the C1qB promoter, but this only occurred after macrophages were treated with IFN␥ (Fig. 9). It shows that, in live macrophages, IFN␥ induces IRF8 expression enabling its binding to the C1qB promoter. PU.1 also co-precipitated with the C1qB promoter, and this was also dependent on IFN␥ stimulation (Fig. 9). Regardless of IFN␥ treatment, co-precipitation between PU.1 and the C1qA promoter was not detected. However, PU.1 coprecipitated with the C1qC promoter irrespective of IFN␥ treatment.
Overall, the ChIP experiments validated the DNA pulldown results regarding IFN␥-dependent PU.1 and IRF8 binding to the C1qB promoter and were in line with PU.1 and IRF8 being shown to mediate IFN␥-stimulated C1qB promoter activity. The question remains how the C1qA promoter, which showed no interaction with PU.1 or IRF8, may respond to IFN␥. The implications of constitutive C1qC promoter association with PU.1 are not understood. However, our finding that the C1qB promoter potentially regulates all three C1q genes suggests contributions from multiple regions in this C1q gene cluster for the synchronization of basal and inducible expression of all three genes.

DISCUSSION
A number of reasons warrant investigation of the C1q gene promoters. First, genetic C1q deficiency, which impairs the production of functional C1q, has profound impact on the immune system as it almost certainly causes SLE. Although genetic C1q deficiency is rare, acquired C1q deficiency is more frequently found in SLE patients, which could contribute to SLE pathogenesis. For example, increased IFN␣ production by plasmacytoid DCs could result from reduced C1q levels (39). The reduced C1q levels could attribute partly to inflammatory consumption and anti-C1q autoantibodies (27,28). However, C1q expression is also versatile, which is affected by diverse microbial stimuli, cytokines, hormones, and drugs (19). Second, C1q production is largely restricted to DCs and macrophages, but the transcription mechanisms are unknown. Thirdly, C1q is a precise assembly from 18 subunits, equally from three different subunits. Deficiency in a single subunit gene abolishes the production of the entire C1q molecules. This interdependence of the three subunits demands transcriptional synchronization.
We proposed two hypotheses regarding the transcriptional regulation of the three C1q genes: 1) the three C1q gene promoters contain replicated cis-acting elements so as to respond to common regulatory signals that are unique to DCs and macrophages, and 2) the three C1q genes lack strong independent promoters and synchronization is achieved through a superseding core promoter inside the C1q gene cluster. Our results support the second hypothesis and suggest that the C1qB promoter functions similar to such a core promoter for the cluster.
The IFN␥-stimulated C1qB promoter activity, which can regulate genes from the 3Ј as well as 5Ј end of coding genes, has been defined from both the cis-and trans-acting facets. First, a 53-bp region in the C1qB promoter was essential to its response to IFN␥ induction. Second, this 53-bp element was found directly recognized by PU.1 and IRF8, which are two macrophage/DC-associated transcription factors and known to mediate IFN␥ stimulation of target genes (33)(34)(35). The finding that STAT1, a key intracellular messenger in IFN␥ receptor signaling (40), regulates the C1qB promoter through IRF8 induction appears to complete a signaling pathway that helps explain the IFN␥-induced C1qB gene expression (Fig. 10). Our data suggest that, through the C1qB promoter, the C1qA and C1qC genes are similarly and coordinately regulated. This "core promoter" concept reconciles with the highly conserved clustering of the three C1q genes (32).
Although these data help explain the synchronized induction of the three C1q genes by IFN␥, these data are insufficient to understand how the basal expression of the three C1q genes is coordinated. The ability of a 3Ј C1qB promoter to regulate the activity of 5Ј C1qA and C1qC promoters across the coding gene implies that C1qB promoter somehow also synchronizes the basal expression of the three C1q genes.  ؊81 bp). This 53-bp DNA fragment was synthesized with 3Ј-biotin tag, which was then immobilized on streptavidin-Sepharose resins. Nuclear extract was prepared from untreated (PBS) and IFN␥-stimulated (IFN␥) RAW264.7 cells and then incubated with the immobilized 53-bp DNA fragment. After washing, the precipitated proteins were analyzed by Western blotting to detect IRF1, STAT1, IRF8, and PU.1 with specific antibodies. As a positive control, nuclear extract (10% of the input amount) was also included in the blots. Proteins precipitated with streptavidin-Sepharose without the 53-bp DNA fragment were used as negative controls. FIGURE 8. STAT1 is required for IFN␥-stimulated IRF8 expression. IRF8 mRNA was detected in RAW264.7 cells after transfection with IRF8 shRNA (shIRF8 -1 and shIRF8 -2) or STAT1 shRNA plasmids (shSTAT1). As a control, the cells were transfected with scramble shRNA plasmid. Cells were either untreated (PBS) or IFN␥-stimulated (IFN␥) for 24 h and IRF8 mRNA was measured by real-time PCR. Data were presented as mean Ϯ S.D. of triplicate experiments.
C1q is preferentially, if not exclusively (41), produced by DCs and macrophages. The finding that PU.1 and IRF8 play key roles in the regulation of C1q gene expression help explain DC and macrophage production of C1q. First, PU.1 is a key transcription factor in DCs and macrophages, which is required for lineage commitments and is not significantly associated with other cell lineages (42,43). Second, IRF8 is also strongly associated with macrophage and DC development and functions (44). In zebrafish, in which the three C1q genes are in even denser cluster (32), IRF8 is a definitive developmental switch in favor of macrophages over neutrophils (45).
In this context, it must be noted that whether PU.1 and IRF8 interact with wider regions of the three promoters has not been assessed. For example, PU.1 appears to associate with the C1qC promoter regardless of IFN␥ stimulation (Fig. 9). More extensive ChIP studies may result in a comprehensive map of PU.1/ IRF8-binding regions in the three promoters and help understand DC/macrophage-associated expression of C1q.
C1q is a prototype for non-collagen proteins that contain Gly-Xaa-Yaa collagen-like regions in their primary amino acid sequences. These collagen-like regions determine and facilitate the polymerization of multiple subunits into large molecular structures. However, C1q remains distinct from the rest of these other molecules for its assembly from three different subunits. Cbln1 and adiponectin are similar to C1q in primary sequences over both the collagen and non-collagen regions, but they are each assembled from a single type of subunit encoded by a single gene (46,47). The sequence similarity between C1q and collectins and ficolins is limited to the collagen-like region and these C1q-like molecules are also each assembled from a single type of subunit (3,48). Therefore, synchronized transcription remains a C1q-specific affair.
We hypothesize that, by having an independent promoter for each C1q subunit gene, the transcription of the three genes is difficult to synchronize. The three independent C1q genes are likely to drift apart during evolution resulting in three distinct, C1q-like molecular identities. The highly conserved 1:1:1 assembly of C1q from its three subunits must have offered key physiological advantages that are not compensated by C1q-like molecules assembled from a single type of C1q subunit. The fact that deficiency in a single gene is able to cause SLE (49) testifies the essentiality of this 1:1:1 assembly of C1q to the host immune system. C1q therefore offers an excellent model for dissecting SLE pathogenesis and a genomic template for understanding synchronized gene transcription. FIGURE 9. Endogenous PU.1 and IRF8 association with the C1q gene promoters. Macrophages were cultured from human blood monocytes and fixed with formaldehyde. Nuclear fraction was harvested and sonicated to generate nuclear lysate. The chromatin fragments were incubated with protein G-Sepharose immobilized with antibodies specific for IRF1, IRF8, STAT1, and PU.1. As negative controls, immunoprecipitation was performed with anti-His antibody or without antibody. Precipitated chromatin fragments were eluted, and DNA was extracted for PCR detection of specific C1qA, C1qB, and C1qC promoter regions. DNA isolated from sonicated nuclei before immunoprecipitation was used as a positive control. PCR products were examined on 1% agarose gels. The B273 promoter falls in the classic promoter region of its downstream C1qB gene. However, its upstream C1qA and C1qC genes lack the activities expected from results obtained with the endogenous genes. The fact that the B273 promoter is able to supersede these expected C1qA and C1qC gene promoters from the 3Ј ends, across the luciferase reporter gene, suggests B273 regulation of C1qA and C1qC gene expression from the 3Ј ends of these genes. This model explains the highly conserved chromosomal clustering and synchronized transcription of the three C1q genes.