Endosomal TLR3 co-receptor CLEC18A enhances host immune response to viral infection

Human C-type lectin member 18A (CLEC18A) is ubiquitously expressed in human, and highest expression levels are found in human myeloid cells and liver. In contrast, mouse CLEC18A (mCLEC18A) is only expressed in brain, kidney and heart. However, the biological functions of CLEC18A are still unclear. We have shown that a single amino acid change (S339 →R339) in CTLD domain has profound effect in their binding to polysaccharides and house dust mite allergens. In this study, we further demonstrate that CLEC18A and its mutant CLEC18A(S339R) associate with TLR3 in endosome and bind poly (I:C) specifically. Compared to TLR3 alone, binding affinity to poly (I:C) is further increased in TLR3-CLEC18A and TLR3-CLEC18A(S339R) complexes. Moreover, CLEC18A and CLEC18A(S339R) enhance the production of type I and type III interferons (IFNs), but not proinflammatory cytokines, in response to poly (I:C) or H5N1 influenza A virus (IAV) infection. Compared to wild type (WT) mice, ROSA-CLEC18A and ROSA-CLEC18A(S339R) mice generate higher amounts of interferons and are more resistant to H5N1 IAV infection. Thus, CLEC18A is a TLR3 co-receptor, and may contribute to the differential immune responses to poly (I:C) and IAV infection between human and mouse.

1. The labelling of human Macrophages as M-M in figure legends is very confusing for the reader. Prior to reading the legend the reader may think this data represents values from Mouse Macrophages (M-M). Can authors please explain or change this to H-M or other? 2. Figure 1, Expression in mouse DC is not as convincing as in BMDM however, I appreciate the antibodies may not be great and the majority of the data in the paper is in BMDM.
Reviewer #2 (Remarks to the Author): The manuscript byYa-Lang Huang et al. roles of human CLEC18A function in TLR3 dependent signaling event. They generated transgenic mice expressing human CLEC18A and mutant CLEC18A S339R for ex vivo and in vivo experiment. Author firstly demonstrated Ifn and other cytokines expression in BMDM and alveolar epithelial cells and found that hCLEC18A and mutant CLEC18A expression enhances expression of Ifnb and after poly IC stimulation and IAV and DV infection. Then author demonstrated TLR3 and CLEC18a associated each other. Author performed in vivo inflammatory model using poly I:C and found that CLEC18A and mutant CLEC18A mice showed sevir inflammation. In contrary, IAV infection to CLEC18A and mutant CLEC18A mice showed reduced inflammatory response and prolonged survival comparing to wild mice. Author tried to explained the discrepancy of TLR3 innate immune response between mice and human, and CLEC18A may causes changes of TLR3 response in human and mice. However, logics of manuscript and these results are not fully support the author`s point. In addition, control of experiments in each experiments and references are unsuitable.
The concept is certainly intriguing and is of interest. However, there are several limitations in the data as following. 4. Author mentioned that Clec18A is low expression in BMDM, however expression level of endogenous clec18a is unclear in Fig1. 5. In Fig2a and b, author included results of human macrophages. These are completely different cell line from mouse cells, therefore it is not suitable to put in the same panel and cannot discuss the production level of Ifns in between mouse and human by these results. 6. Poly I:C treatment in vivo showed increase of inflammation in CLEC18A and mutant CLEC18A mice compared with wild type mice, however IAV treatment in these mice showed reduces of inflammation. Why these difference induced by these stimulation? 7. It is unclear why author use CLEC18A mutant mice to show difference of TLR3 responses between human and mice. 8. Author tied to explain the discrepancy of TLR3 innate immune response between mice and human using transgenic mice which expressed human CLCE18A. However, ectopic expression in mice to human gene does not help to explain the difference of human and mice. It is hardly compared with wild type and human gene transgenic mice. Author should compare mice gene and human gene transgenic mice. If author say the expression of Clec18a in mice is low, author show the evidence of expression level in human and mice and use knockdown or knockout technique in human cells line.
Overall, the data shown do not support the main conclusions drawn by the authors.
Reviewer #3 (Remarks to the Author): In this study, Huang et al. described essential roles of human CLEC18A and CLEC18A-1 in TLR3 signaling pathway by utilizing the ROSA-CLEC18A and ROSA-CLEC18A-1 mice. Their results suggested that human CLEC18A and CLE18A-1 directly interacted with TLR3 and poly(I:C), enhancing IFN production without the enhancement of pro-inflammatory cytokine production. ROSA-CLEC18A and ROSA-CLEC18A-1 mice were more resistant to H5N1 IAV infection compared to WT mice, suggesting that these proteins are important in antiviral immune responses in human. Overall, their findings that human CLEC18A and CLEC18A-1 act as a co-receptor with TLR3 is particularly interesting in view of differential roles of them between human and mouse, in TLR3-mediated antiviral responses. However, data presented in this manuscript is not sufficient to support their conclusion. Furthermore, there seems to be over-interpretation throughout their manuscript. To strengthen their conclusion, many concerns listed below should be addressed.
Specific comments 1. Figure S1A, S2B and S2C should be explained in their manuscript. 2. In figure 2A, data showing IP-10 production in response to poly(I:C) (1 microG/mL) is missing. 3. In figure 2A and S1A, there seems to be some discrepancy, especially data on pro-inflammatory cytokines, they should kindly explain about that. Furthermore, they should place ELISA data but not qPCR data onto the main figure.
4. Figure 2a did not correspond to supplementary fig. 1a. While mRNA level of TNF showed no significant difference between WT and CLEC18 expressing cells, the protein level was increased in CLEC18 expressing cells. 5. They should describe about how they prepared human macrophages in [materials and methods] section. 6. In figure 4D, to support their conclusion, they should perform statistical analysis. 7. To strengthen their conclusion, they should perform knockdown experiment using human cells. 8. In figure 6 and 7, they should show data using control (untreated) mice. 9. In figure 1A, because the expression level of mRNA does not always reflect that of protein, they should show CLEC18 protein expression in each tissues and cells (especially immune cells). Same for figure 1B. 10. Because their previous paper (ref. 1) also suggests that CLEC18 is secreted by cells, they should clearly refer it in their manuscript. 11. In figure 1C, they should clearly describe about the antibody they used in figure legend. Furthermore, in case that they used antibody against endogenous human CLEC18 protein but not against tags, they should clearly describe about the information of the antibody used. 12. There are many typos and grammatical errors throughout the manuscript that should be corrected (e.g. 14. Although they claim that CLEC18A-1 is more potent to induce interferon response, this reviewer could not find any statistical analysis between CLEC18A and CLEC18A-1. 15. In figure 2A and 2B, they claim that CLEC18A and CLEC18A-1 increased type I interferon expression when even 1.1-1.2 fold. On the other hand, they claim that, in figure S1B, S1C and S1D, CLEC18 did not affect to ifna or ifnb mRNA expression even 2-4 fold. This is quite strange. They should perform statistical analysis in figure S1 and precisely interpret their data. 16. In figure 4C, statistical analysis seems to be strange. The difference in ifnb between WT (pIC 0) and WT (pIC 30) was significant, but that between WT (pIC 0) and WT (pIC 10) was not significant. If the figure is correct, latter also seems to be significant. They should confirm the method of statistical analysis throughout their manuscript. 17. In figure 5D, the means of [+] and [-] is not clear. 18. In figure 5D, it is interesting to investigate which domain of CLEC18 is involved in its binding to poly(I:C).

Reviewer #1 (Remarks to the Author):
Major 1. Line 233-235: Authors state that unlike IFNa and IFNb, ROSA-CLEC18A and ROSA-CLEC18A-1 BMDMs do not upregulate TNF-α and IL-6 in Fig. 2a. However, TNF and IL-6 are both clearly upregulated on the RNA (Fig. 2a) and protein level (Supp Fig. 1a) the difference is simply that expression is not significantly increased compared to WT BMDM levels. This is important to note and suggests that the mouse model in which human CLEC18A is overexpressed may not match what is known to be true for TLR3 responses in human macrophage. Importantly however, this does not negate the significant amount of data in the remainder of the paper demonstrating the interaction of CLEC18A/A-1 with poly I:C and TLR3.  Fig. 3). We found that CLEC18 knockdown in human cells only affects poly (I:C) -, but not LPS-, induced cytokine production ( Supplementary Fig. 3). This observation further supports the argument that CLEC18 is the co-receptor of TLR3.
2. It is interesting that in this system CLEC18A-1 seems to be inhibitory on the TRAF6/NF-kB side of the TLR3 pathway following poly (I:C) stimulation ( Supp Fig 1a), yet enhances the TRIF side. On the other hands in CLEC18A BMDM, 'endogenous' CLEC18A appears important for both. Have the authors identified the residue responsible for the differences we observe between human and murine signaling in response to poly I:C? What's the conservation in sequence between human and mouse Answer: 1) As we show in Figure 1

Answer:
We repeat this experiment with two different doses of poly (I:C) to show dose-dependent effect. We also compared the effect of poly (I:C) with other TLR ligand in cytokine expression in mRNA and protein levels ( Supplementary Fig. 2) in this revised version.
4. The TLR3/dsRNA inhibitor suppresses IFN at the same concentration in WT and CLEC18 and CLEC18A-1 cells (100uM) (Fig. 4e). This is also the case for IFN. This result weakens the claim that C18A is in-fact a co-receptor for TLR3. Could the TLR3/dsRNA inhibitor be added to FRET experiments to definitively prove C18A/C18A-1 as co-receptor of TLR3? If not perhaps remove. The bio-layer interferometry data is much more convincing.

Answer:
1. We performed FRET experiments in Figure 4f. The high concentration of TLR3/dsRNA inhibitor also disrupts the association between poly(I:C)-CLEC18A as well as poly(I:C)-CLEC18A-1. The FRET assay is in accord with what we observed in cytokine production (Figure 4e).

Lower concentration of inhibitor (10 ,tiM and 30 ,tiM) can only suppress cytokine release from WT, but not CLEC18A/18A-1 KI mice. However, when the inhibitor is up to 100 ,tiM, it also disrupts the interactions between TLR3 and poly (I:C). This observation is in
accord with our claim that the CLEC18 act as a TLR3 co-receptor and can increase binding affinity between poly (I:C)/TLR3-CLEC18A complex.

The confocal data in supplemental showing EEA1, TLR3 and CLEC18 co-localization is really nice and could perhaps be moved to the main figures.
Answer: Yes, we move these supplementary figures to figure 4c &d in this revised version.
6. Figure 5b -it is unfortunate that the authors present the input blots for the immunoprecipitation studies on separate gels/western blots. It is generally expected that the Input/expression of plasmids in both the Isotype control lane and the Flag lanes be shown to confirm adequate expression of plasmid in these samples and in order to confirm a legitimate interaction. Blots in Figure 5d with deletion clones shows this appropriately. Figure 5b.

3) Thank you for your comment on Figure 5d that our presentation is OK. But I would like to
say that the total input and IP are also developed separately in western blot analysis.  Figure 1, Expression in mouse DC is not as convincing as in BMDM however, I appreciate the antibodies may not be great and the majority of the data in the paper is in BMDM.

Answer:
We try a couple of times and find that human CLEC18 expression in mouse DC are much weaker than that BMDM. It may due to the differential promoter activity in mouse DC and BMDM.
Author tried to explained the discrepancy of TLR3 innate immune response between mice and human, and CLEC18A may causes changes of TLR3 response in human and mice. However, logics of manuscript and these results are not fully support the author`s point. In addition, control of experiments in each experiments and references are unsuitable.
The concept is certainly intriguing and is of interest. However, there are several limitations in the data as following.  Fig2a and b, author included results of human macrophages. These are completely different cell line from mouse cells, therefore it is not suitable to put in the same panel and cannot discuss the production level of Ifns in between mouse and human by these results.

Answer:
We present the cytokine production profile of human (Supplementary Fig. 3) and mice (Fig. 2a) macrophages in separate figures.
6. Poly I:C treatment in vivo showed increase of inflammation in CLEC18A and mutant CLEC18A mice compared with wild type mice, however IAV treatment in these mice showed reduces of inflammation. Why the difference induced by these stimulations? 7. It is unclear why author use CLEC18A mutant mice to show difference of TLR3 responses between human and mice.

Answer:
We noticed the presence of CLEC18A(S339R) in the human genome database when we initiated the projects, so we are curious to understand whether the S339R339 has any effect in host response to TLR3 ligand and IAV infection. Because mice do not express endogenous CLEC18A in immune cells (Figure 1), so we take this advantage to compare the differential response of CLEC18A and CLEC18A(S339R) in knock-in mice.
8. Author tried to explain the discrepancy of TLR3 innate immune response between mice and human using transgenic mice which expressed human CLCE18A. However, ectopic expression in mice to human gene does not help to explain the difference of human and mice. It is hardly compared with wild type and human gene transgenic mice. Author should compare mice gene and human gene transgenic mice. If author say the expression of Clec18a in mice is low, author should show the evidence of expression level in human and mice and use knockdown or knockout technique in human cells line.

Answer:
1) As we show in Figure 1 a~c & Supplementary Fig. 1 Fig. 3). We found that CLEC18 knockdown in human cells only affects poly (I:C)-, but not LPS-, induced cytokine production ( Supplementary Fig. 3). This observation further supports the argument that CLEC18 is the co-receptor of TLR3.
Overall, their findings that human CLEC18A and CLEC18A-1 act as a co-receptor with TLR3 is particularly interesting in view of differential roles of them between human and mouse, in TLR3mediated antiviral responses. However, data presented in this manuscript is not sufficient to support their conclusion. Furthermore, there seems to be over-interpretation throughout their manuscript. To strengthen their conclusion, many concerns listed below should be addressed.
Specific comments 1. Figure S1A, S2B and S2C should be explained in their manuscript.
Answer: We follow the suggestion, and add explanation for what we observed in this revised manuscript. The S1A, S2B and S2C in previous version are moved to Fig. 2A, S2A, S2B, respectively, in this revised version. Description of these findings is shown in section 2 (line 256266).
Answer: We present the IP-10 production in Figure 2A 3. In figure 2A and S1A, there seems to be some discrepancy, especially data on proinflammatory cytokines, they should kindly explain about that. Furthermore, they should place ELISA data but not qPCR data onto the main figure. 4. Figure 2a did not correspond to supplementary fig. 1a. While mRNA level of TNF-alpha showed no significant difference between WT and CLEC18 expressing cells, the protein level was increased in CLEC18 expressing cells.

Answer:
We present the new ELISA results in Figure 2a and

Answer:
We change those labeling accordingly.
14. Although they claim that CLEC18A-1 is more potent to induce interferon response, this reviewer could not find any statistical analysis between CLEC18A and CLEC18A-1.

Answer:
1) We add the symbols to represent statistical significance in the figures in this revised version. 2) We rename CLEC18A-1 as CLEC18A(S339R) in this revised version.
15. In figure 2A and 2B, they claim that CLEC18A and CLEC18A-1 increased type I interferon expression when even 1.1-1.2 fold. On the other hand, they claim that, in figure S1B, S1C and S1D, CLEC18 did not affect to ifna or ifnb mRNA expression even 2-4 fold. This is quite strange. They should perform statistical analysis in figure S1 and precisely interpret their data.

Answer:
We repeat the experiment and present the data with statistical analysis in Figure 2a and Supplementary Figure 2. Compared to WT BMDMs, higher expression of type I interferons was in CLEC18A and CLEC18A(S339R) cells after poly (I:C) stimulation. In contrast, other TLR ligand has no obvious effect to upregulate cytokine production.
16. In figure 4C, statistical analysis seems to be strange. The difference in ifnb between WT (pIC 0) and WT (pIC 30) was significant, but that between WT (pIC 0) and WT (pIC 10) was not significant. If the figure is correct, latter also seems to be significant. They should confirm the method of statistical analysis throughout their manuscript. Figure 4c (changed to Figure 4e in this revised version). In this experiment, all groups were incubated with the same concentration of the poly (I:C), but with the different concentration of TLR3/dsRNA complex inhibitor as indicated. We re-analyze data by one-way ANOVA method.