The protective effect of the anti-Toll-like receptor 9 antibody against acute cytokine storm caused by immunostimulatory DNA

Toll-like Receptor 9 (TLR9) is an innate immune receptor recognizing microbial DNA. TLR9 is also activated by self-derived DNA, such as mitochondrial DNA, in a variety of inflammatory diseases. We show here that TLR9 activation in vivo is controlled by an anti-TLR9 monoclonal Ab (mAb). A newly established mAb, named NaR9, clearly detects endogenous TLR9 expressed in primary immune cells. The mAb inhibited TLR9-dependent cytokine production in vitro by bone marrow-derived macrophages and conventional dendritic cells. Furthermore, NaR9 treatment rescued mice from fulminant hepatitis caused by administering the TLR9 ligand CpGB and D-(+)-galactosamine. The production of proinflammatory cytokines induced by CpGB and D-(+)-galactosamine was significantly impaired by the mAb. These results suggest that a mAb is a promising tool for therapeutic intervention in TLR9-dependent inflammatory diseases.

Since the administration of the anti-TLR7 mAb can ameliorate TLR7-dependent systemic inflammation in Unc93b1 D34A/D34A mice 22 , the antibody bound to NA-sensing TLRs is expected to act as a tool for therapeutic approaches. Although TLR9 is another promising target of therapeutic intervention in autoimmune diseases, previously established anti-TLR9 mAbs fail to inhibit TLR9 responses. It is, therefore, currently unknown whether anti-TLR9 mAb can control TLR9-dependent inflammatory diseases.
In this report, we established a novel anti-mouse TLR9 mAb that functions in confocal microscopy and flow cytometry, and inhibits TLR9 responses in vitro and in vivo. Administration of the anti-TLR9 mAb protected mice from acute hepatitis caused by the TLR9 ligand and D-(+ )-galactosamine, and these results demonstrate that the anti-TLR9 mAb is a promising tool to control TLR9-dependent inflammatory diseases.

Results
New anti-mouse TLR9 mAb recognizing the N-terminal fragment of TLR9. We have generated three anti-mouse TLR9 mAbs, B33A4, C33A1 and J15A7 14 , but they do not clearly inhibit the TLR9 response. We tried to establish a new anti-TLR9 with an inhibitory effect on the TLR9 response and generated a new anti-TLR9 mAb named NaR9, classed as IgG2a/κ . To locate the epitope of NaR9, Ba/F3 cells expressing full length, N-terminal or C-terminal fragments of TLR9 were stained by NaR9 and the other anti-TLR9 mAbs. NaR9 specifically reacted with the N-terminal fragment of TLR9 (Fig. 1a), and a crossblocking study showed NaR9 bound to the region close to the epitope of J15A7 but not to the epitope of B33A4 (Fig. 1b).
For further analysis, we stained various TLR9s with mutations at the CpGB binding sites 23 . Among the mutants, the H642A mutation abolished NaR9 binding to TLR9 (Fig. 1c). Because the other two mAbs, J15A7 and B33A4, also impaired binding to the H642A mutant, TLR9 conformation may be altered by the H642A mutation. The other mutants did not alter NaR9 binding to TLR9, so the epitope of NaR9 might not be close to the CpGB binding sites. Endogenous TLR9 is detected by NaR9. The expression of endogenous TLR9 in primary immune cells has not been extensively studied. To detect endogenous TLR9 in primary immune cells, we stained spleen cells from WT or Tlr9 −/− mice with 3 anti-TLR9 mAbs (Fig. 2a). NaR9 detected endogenous TLR9 in WT spleen cells but not in Tlr9 −/− spleen cells. B33A4 showed a much lower intensity of staining than NaR9. J15A7 showed the highest intensity of staining among the three mAbs tested. According to these results, NaR9 and J15A7 were used for further experiments.
In a previous report, we showed that TLR9 is expressed on the surface of splenic conventional dendritic cells (cDCs) and plasmacytoid DCs (pDCs) as well as in intracellular vesicles of bone marrow-derived cDCs (BM-cDCs) and BM-pDCs by using J15A7 14 . In this report, we have improved the method of staining and analysed the TLR9 expression patterns again in various immune cells using NaR9 and J15A7. Both NaR9 and J15A7 detected TLR9 in splenic cDCs (CD11c hi , Siglec-H − ) and pDCs (CD11c + , Siglec-H + ) under the membrane permeabilizing condition (Fig. 2b-d). Cell surface TLR9 was also detected by these anti-TLR9 mAbs, although the staining intensity with NaR9 was lower than that with J15A7 ( Fig. 2c,d). TLR9 is also detectable in B cells by membrane-permeabilized staining. However, cell surface TLR9 was hardly detected (Fib. 2b, e). Next, we studied TLR9 expression in splenic monocytes. CD11b + Ly-6G + neutrophils did not express TLR9, whereas CD11b + CD49b − Ly6G − monocytes weakly but significantly expressed TLR9 (Fig. 3a-c). TLR9 was weakly detected on the surface of Ly-6G − monocytes by J15A7 but was undetectable by NaR9 (Fig. 3c).
The expression of TLR9 is increased in cells by Type I IFN stimulation. The molecular mechanism controlling TLR9 expression was studied next. A macrophage cell line RAW264.7, bone marrow-derived macrophages (BM-Macs) and bone marrow-derived conventional DCs (BM-cDCs) were treated with lipid A, CpGB, IFN-β or TNF-α for 24 h, and cells were stained by NaR9 with or without membrane permeabilization. IFN-β strongly induced TLR9 expression, whereas lipid A, CpGB and TNF-α weakly increased TLR9 expression, (Fig. 4a,c,e). Interestingly, induction of TLR9 expression by IFN-β in BM-Macs was weaker than in RAW264.7 cells or BM-cDCs.
To investigate the correlation between TLR9 expression and TLR9 response, these cells were pre-treated with IFN-β or TNF-α for 24 h and then stimulated with the TLR9 ligands CpGB or CpGA. Pre-treatment with IFN-β enhanced the production of RANTES by TLR9 ligands in RAW264.7 cells (Fig. 4b) and enhanced the production of IL-6 in BM-Macs and BM-cDCs (Fig. 4d,f).
Endogenous TLR9 localizes in late endosome and lysosome. TLR9 is synthesized in the endoplasmic reticulum (ER) and transported to the endosome or lysosome by a chaperone molecule Unc93B1 14 . Previous studies have shown the distribution of transfected and epitope-tagged TLR9 10,12 , but the distribution of endogenous TLR9 has not been reported. NaR9 was employed to detect endogenous TLR9 in BM-Macs, and endogenous TLR9 was specifically detected in WT but not detected in Tlr9 −/− BM-Macs. These data verify the specificity of NaR9 in confocal microscopy analysis (Fig. 5a).
Next, ligand-dependent trafficking of endogenous TLR9 was observed in RAW264.7 cells. Without CpGB stimulation, TLR9 colocalized with LAMP1, LAMP2 and Rab7a, markers for endosomes and lysosomes, but showed much less colocalization with Calnexin, an ER marker (Fig. 5b,c). After CpGB stimulation for 24 h, TLR9 colocalization with LAMP1 was significantly increased (Fig. 5b,d). TLR9 colocalization with LAMP2 was weakly increased, whereas TLR9 colocalization with a late endosome marker Rab7a was weakly decreased. These results suggest that ligand stimulation transports endogenous TLR9 into the LAMP1 + compartment.

NaR9 inhibits TLR9 responses of BM-derived cells.
We previously reported that the anti-mouse TLR7 mAb inhibited the TLR7 response and ameliorated inflammatory disorders in Unc93b1 D34A/D34A mice 22 . Here, the effect of NaR9 on the TLR9 response was studied in bone marrow-derived Macs (BM-Macs), conventional Scientific RepoRts | 7:44042 | DOI: 10.1038/srep44042 DCs (BM-cDCs) and BM-pDCs. NaR9 drastically inhibited the production of TNF-α and IL-6 induced by CpGB but did not inhibit the response to loxoribine, a TLR7 ligand (Fig. 6a,b). In contrast to BM-Macs and BM-cDCs, TLR9-dependent IFN-α production by BM-pDCs was not impaired by NaR9 (Fig. 6c).
Next, we tested the inhibitory effect of NaR9 under severe conditions. As shown above, IFN-β enhanced the expression of TLR9 and the response to the TLR9 ligand (Fig. 4). Even with pre-treatment of IFN-β , NaR9 inhibited the production of IL-6 by BM-Macs and BM-cDCs, induced by a drastically enhanced response to CpGB (Fig. 6d,e). Next, we investigated why NaR9 failed to inhibit the TLR9 response in BM-pDCs. We stained cell surface TLR9 with NaR9 and J15A7, based on the hypothesis that TLR9 on the cell surface is required for inhibition and BM-pDCs do not express TLR9 on cell surfaces. In contrast to DCs and macrophages in the spleen (Figs 2 and 3), cell surface TLR9 was hardly detected on BM-Macs, BM-cDCs and BM-pDCs (Fig. 7a-c).
Another hypothesis for the different sensitivity levels of the different cells is that sensitivity is dependent on their antibody uptake activity. BM-Macs, BM-cDCs and BM-pDCs were incubated with Alexa 488-labelled antibodies, and antibody internalization was examined by flow cytometry analysis with or without quenching of the antibody on the cell surface. The antibody was internalized by BM-Macs and BM-cDCs but not by BM-pDCs (Fig. 7d).

NaR9 protects mice from lethal inflammation induced by CpGB and D-gal. Because NaR9 inhib-
ited TLR9-dependent cytokine production in BM-Macs and BM-cDCs, the inhibitory effect of NaR9 in vivo was studied next. Administration of CpGB and D-(+ )-galactosamine (D-gal) induces lethal fulminant hepatitis in a TLR9-dependent manner 1,24 , and TNF-α production in the liver is thought to cause hepatocyte cell death 24 . Mice were treated with NaR9 or isotype-matched IgG2a for 15 h and administered with CpGB and D-gal. More than 80% of mice pre-treated with PBS or isotype-matched IgG2a died within 24 h, whereas 80% of mice treated with NaR9 survived the treatment with CpGB and D-gal (Fig. 8a). To examine the effect of NaR9 on TLR9-dependent cytokine production in vivo, the serum concentration of TNF-α and IL-12p40 in mice treated with CpGB and D-gal was determined by ELISA. Serum TNF-α and IL-12p40 reached a peak at 1 or 3 h after CpGB and D-gal injection, respectively (Fig. 8b). NaR9 treatment significantly impaired the production of these cytokines in vivo. These results indicated that NaR9 rescued mice from fulminant hepatitis by inhibiting the TLR9 response in vivo.

Discussion
The present study focused on endogenous TLR9 by establishing a novel mAb to mouse TLR9. Although the distribution of over-expressed, epitope-tagged TLR9 has been extensively studied 10,12 , it is still important to study endogenous TLR9. The present study showed that TLR9 is expressed in CD11b + Ly-6G − macrophages, CD11c hi Siglec-H − cDCs, CD11c + Siglec-H + pDCs and CD19 + B cells in the spleen. Although human granulocytes are reported to express cell surface TLR9 25 , CD11b + Ly-6G + mouse granulocytes in the spleen did not express TLR9. Interestingly, TLR9 was also detected on the surface of cDCs and pDCs but to a much lesser extent on B cells and monocytes. The cell surface expression of TLR9 seems to vary not only with cell types but also with the status of cell differentiation and activation. In contrast to splenic DCs, TLR9 expression was very low on the surface of BM-derived cells. Furthermore, IFN-β , but not TNF-α , enhanced TLR9 expression in a macrophage cell line CpGB-induced TLR9 trafficking into the LAMP1 + compartment and TLR9 was detected not only in endolysosomes but also on the cell surface. It is possible that TLR9 shuttles between the cell surface and endolysosomal compartments 26 , and our results suggest that cell surface TLR9 comes from the LAMP1 + compartment.

cells (b), BM-Macs (d) and
BM-cDCs (f) were pre-treated with medium, IFN-β (1000 U/ml) or TNF-α (25 ng/ml). Twenty-four hours later, the cells were stimulated with the indicated concentration of CpGB or CpGA. Twenty-four hours after stimulation with CpGB or CpGA, culture supernatants were collected, and the concentration of RANTES (b) or IL-6 (d,f) was determined by ELISA. The results are represented by the mean value and the s.d. from triplicate wells. These experiments were repeated at least twice, and representative data are shown.  Newly established anti-TLR9 mAb NaR9 inhibited TLR9-dependent cytokine production in BM-Macs and BM-cDCs but not in pDCs. The difference in the inhibitory effect of NaR9 among cells is related to the internalization of antibodies, but the mechanism by which NaR9 inhibits the TLR9 response following the internalization is unclear. One possibility is the inhibition of TLR9 binding to CpG-ODN; however, NaR9 bound to TLR9 mutants at the CpGB binding site (W47A, R74A and F108A) as much as to WT TLR9. These results did not support the possibility that NaR9 inhibits TLR9 binding to CpG-ODN. Another possibility is the inhibition of TLR9 dimerization. The extracellular domain of TLR9 consists of 26 tandem repeats of the leucine-rich repeat (LRR) protein motif 23 . Proteolytic cleavage of the loop between LRR14 and LRR15 generates N-terminal and C-terminal fragments of TLR9, and the cleavage is required for ligand-dependent TLR9 dimerization. If NaR9 reacts with the loop or LRR14, it is possible that NaR9 inhibits TLR9 dimerization. These possibilities should be addressed by performing an assay to detect TLR9 dimerization.
NaR9 not only inhibited TLR9 responses in an in vitro assay but also rescued mice from lethal liver failure induced by the TLR9 ligand CpGB and D-gal. Because serum TNF-α and IL-12p40 induced by CpGB and D-gal was significantly decreased by the treatment with NaR9, the cells producing proinflammatory cytokines might be sensitive to NaR9. In addition to the acute inflammation studied here, it is reported that chronic inflammation in liver such as non-alcoholic steatohepatitis (NASH) is driven by TLR9 27 . Mitochondrial DNA (mtDNA) from hepatocytes activates immune cells, leading to chronic hepatitis and cirrhosis. Given that anti-TLR7 mAb rescues Unc93b1 D34A/D34A mice from chronic hepatitis 22 , anti-TLR9 mAb is also expected to cure TLR9-dependent chronic diseases.
Compared with the other reagents inhibiting the response of TLR9, the use of an antibody as a therapeutic intervention has advantages and disadvantages. First, the inhibitory effect of anti-TLR9 is dependent on the uptake activity of the cells. This might be a disadvantage if the corresponding cells do not have antibody uptake activity, but has the advantage of specificity if the corresponding cells do have uptake activity. Next, antibodies have much higher specificity than small chemicals or short ODNs. Some of these reagents not only inhibit the response of TLR9 but also inhibit the response of TLR7 [28][29][30] . It is advantageous to use these reagents if the targeting diseases are driven by both TLR7 and TLR9; however, specificity is important for medicine to avoid unexpected results when being used as a therapeutic intervention. In conclusion, the present study suggests that the anti-TLR9 mAb is a promising tool for therapeutic intervention in TLR9-dependent inflammatory diseases.

Methods
Generation of anti-mouse TLR9 monoclonal antibody. To establish an anti-mouse TLR9 monoclonal antibody (mAb), Tlr9 −/− mice on a BALB/c background were immunized by intraperitoneally administering Ba/ F3 cells expressing mouse TLR9 (Ba/F3-mTLR9) with complete Freund's adjuvant/incomplete Freund's adjuvant as adjuvants. Five days after the final immunization, splenic cells were fused with SP2/O myeloma cells. A hybridoma producing anti-TLR9 mAb was selected by flow cytometry staining of Ba/F3-mTLR9. A subclass of this mAb was determined as IgG2a/κ and named NaR9.
Mice. C57BL/6 mice were purchased from Japan SLC, Inc. (Shizuoka, Japan). Tlr9 −/− mice on a C57BL/6 background were generated in our laboratory. Tlr9 −/− mice on a BALB/c background were backcrossed seven times with BALB/c wild-type mice purchased from Japan SLC, Inc. All animal experiments were approved by the Animal Research Committee of the Institute of Medical Science, The University of Tokyo, and performed in accordance with the guidelines.
Reagents and antibodies. Anti-TLR9 mAbs, NaR9, J15A7, B33A4 and C34A1, were purified from ascitic fluid as shown previously 12 . Streptavidin-PE, anti-mouse IgG1-PE, anti-mouse IgG2a-PE, isotype using LSM710 ZEN software according to the manufacturer's manual. Pearson's correlation coefficients were used for the statistical analysis of co-localization. A Pearson's coefficient of < 0.2 represents poor co-localization.
Ligand stimulation and ELISA. RAW264.7 cells, BM-Macs or cDCs (2 × 10 5 cells/well) were seeded in 24-well plates and stimulated with Lipid A, CpGB, IFN-β or TNF-α for 24 h. The cells were subjected to TLR9 staining and analysed using LSRFortessa X-20. RAW264.7 cells (2 × 10 4 cells/well) were seeded in 96-well plates and treated with medium, IFN-β or TNF-α . Twenty-four hours later, the cells were stimulated with CpGB or CpGA for 24 h. RANTES in the supernatant was measured using a DuoSet ELISA Development System (R&D Systems, Minneapolis, MN, USA). BM-Macs, cDCs or pDCs (5 × 10 4 cells/well) were plated in 96-well plates, and treated with medium, NaR9 or IgG2a control Ab. Four hours later, the treated cells were stimulated by CpGB, CpGA, loxoribine or PolyU (with DOTAP) for 24 h. TNF-α and IL-6 in the supernatant were measured using Ready-SET-Go! ELISA Sets (eBiosciences), and IFN-α was measured using a VeriKine ELISA Kit (PBL Assay Science, Piscataway, NJ, USA).
Antibody-uptake assay. BM-Macs, BM-cDCs or BM-pDCs (1 × 10 6 cells/well) were seeded in 6-well plates and incubated with 10 μ g/ml of NaR9 or IgG2a control Ab conjugated to Alexa 488 (Alexa Fluor ® 488 Antibody Labeling Kit, Thermo Fisher) at 37 °C for 24 h. The cells were collected and treated with/without anti-Alexa488 pAb (Thermo Fisher) for quenching cell surface fluorescence. The prepared cells were subjected to flow cytometry analysis using LSRFortessa X-20.
In vivo administration of CpGB and D-(+)-galactosamine. C57BL/6 WT mice were treated intraperitoneally with the indicated mAbs or PBS. Fifteen hours later, mice were administered intraperitoneally with 10 nanomoles of CpGB and 20 mg of D-(+ )-galactosamine per head. Blood samples were collected from mice at 0, 1, 3 and 6 h after stimulation, and separated serum was subjected to ELISA to measure the production of TNF-α and IL12-p40. Statistical analysis. Statistical significance was calculated by performing a two-tailed unpaired Student's t-test. A p value of < 0.05 was considered statistically significant.