Mitochondrial dynamics controls anti-tumour innate immunity by regulating CHIP-IRF1 axis stability

Macrophages, dendritic cells and other innate immune cells are involved in inflammation and host defense against infection. Metabolic shifts in mitochondrial dynamics may be involved in Toll-like receptor agonist-mediated inflammatory responses and immune cell polarization. However, whether the mitochondrial morphology in myeloid immune cells affects anti-tumor immunity is unclear. Here we show that FAM73b, a mitochondrial outer membrane protein, has a pivotal function in Toll-like receptor-regulated mitochondrial morphology switching from fusion to fission. Switching to mitochondrial fission via ablation of Fam73b (also known as Miga2) promotes IL-12 production. In tumor-associated macrophages, this switch results in T-cell activation and enhances anti-tumor immunity. We also show that the mitochondrial morphology affects Parkin expression and its recruitment to mitochondria. Parkin controls the stability of the downstream CHIP–IRF1 axis through proteolysis. Our findings identify mechanisms associated with mitochondrial dynamics that control anti-tumor immune responses and that are potential targets for cancer immunotherapy.

It is not clear to me why the authors chose to go down this laborious route in the last third of the manuscript. They made a KO of Fam73b and have a very striking phenotype in vitro as well as in vivo. They implicate IRF1 and mito fission/fusion in the phenotype. By jumping to CHIP, Ube2w, and Parkin, the manuscript rapidly loses its focus on Fam73b (for which the data is very nice), in an effort to establish connections between some proteins that are not well characterized previously or in this manuscript. For this reviewer, most of the data in the last two figures do not add to and instead detract from the story.
There is some interesting material here. Unfortunately, the story becomes mired in an increasingly complicated pathway that remains poorly characterized and for which some key pieces of data are not very strong or well-explained. I suggest that the reviewers trim some of the data in the last third of the manuscript to keep the story focused on the role of Fam73b and mitochondrial fission/fusion in regulation of IRF1 levels and consequent effects on IL-12 production. In addition, it is recommended that the authors seek an editing service to correct typos and improve the English/grammar and readability of the manuscript. 2) The authors need to better introduce Fam73b in the Introduction section, since this is a major focus of the story and most readers would not be familiar with the protein. Likewise, if the author decides to keep CHIP and Ube2w in the story, they need to be better introduced so that the reader understands the biological relevance of their control of IL12a gene induction. In this manuscript, the authors report that mitochondrial fission, in an ubiquitin-dependent manner, stimulates IL-12 expression while inhibits that of IL-10 and IL-23 without affecting other proinflammatory cytokines. They observed that stimulation of BMDM with LPS or poly I:C promotes a rapid mitochondrial fission by downregulating the mitochondrial fusion effectors Mfn1/2 and the downstream effector Fam73b (initially named mitogardin or Miga). Genetic ablation of Fam73b confirmed that this mitochondrial protein is required in TLRs-regulated fragmentation/fission of the mitochondrial network and this phenotype specifically promotes IL-12 production. At the molecular level, the authors have observed that mitochondrial fission triggers the accumulation and recruitment of the E3 ubiquitine ligase Parkin at the mitochondria. Parkin then induces the degradation of monubiquitinated CHIP and stabilizes the transcription factor IRF1 that is involved in IL-12 production. In vivo, this increased Il-12 production by macrophages enhances the antitumor T cells responses in a murine melanoma model as well as in a MCAinduced fibrosarcoma model. These findings suggest therefore an unexpected role of mitochondrial dynamics in cytokine production and ensuing anti-tumoral immunity. This study is really extensive and complete while sometime, quite complicated. Moreover I have few concerns.
To decipher the molecular mechanisms promoting IL-12 production downstream of mitochondrial fission, the authors have knocked down IRF1, Parkin and CHIP using shRNA in WT or Fam73 KO BMDM. As KO mice are available for IRF1 and Parkin, I guess that their conclusions should be confirmed by using BMDM from these KO mice. I mean for instance, IL12 production in IRF1 KO BMDM should be less important than in WT BMDM after TLR4 stimulation. The use of Parkin KO BMDM would be also interesting.

Specific comments:
In Fig 1a. It appears that 50 % of the BMDM have already short/fragmented mitochondria. TLR4 stimulation for 2 hrs with 500 ng/m of LPS leads to a 20% increase in mitochondrial fragmentation. It does not seem to be the extensive mitochondrial fragmentation described in Fig  1c with time-lapse microscopy. 500 ng/ml of LPS is also quite a high concentration to stimulate BMDM. Do lowers concentrations of LPS also trigger mitochondrial fragmentation? In fig 1g,  We thank the reviewer for the constructive comments and suggestions. Following the reviewer's suggestion, we delete some parts of complicated data in last two figures such as all Ube2w data and polyubiquitinated sites by Parkin, which are not associated with the final conclusion tightly. Furthermore, we apologize for not clearly describing the data linked first 4 Figures to last 3. We observe similar phenotypes among Fam73b KO ( Figure. Figure. 6e) cells, which indicates that IL-12a increment is due to the mitochondrial fission, but not special function mediated by Fam73b. In last two figures, we focus on the mechanism how mitochondrial fission regulates IRF1 accumulation. So we still remain some parts data of CHIP monoubiquitination and Parkin function, which well character IRF1 accumulation.
2. There is some interesting material here. Unfortunately, the story becomes mired in an increasingly complicated pathway that remains poorly characterized and for which some key pieces of data are not very strong or well-explained. I suggest that the reviewers trim some of the data in the last third of the manuscript to keep the story focused on the role of Fam73b and mitochondrial fission/fusion in regulation of IRF1 levels and consequent effects on IL-12 production. In addition, it is recommended that the authors seek an editing service to correct typos and improve the English/grammar and readability of the manuscript.

Response:
Following this excellent suggestion, we have removed some data which is not associated with our crucial conclusion as mentioned above. Mitochondrial fission mediated by all of Fam73a, Fam73b, Mfn1/2 Opa1 deficiency performs the similar phenotypes, indicating mitochondrial morphology is required for IL-12 regulation, but not special function of one molecule. Therefore, we still keep the data of CHIP monoubiquitination and Parkin to clarify the underlying mechanism.
To improve the readability of our manuscript, we reorganize the data and correct typos with a language editing service.

Response:
We apologize for not clearly describing our data. In ubiquitination assay, we always pretreated with MG132 to block proteasome activity. MG132 treatment promotes IRF1 stability and normalizes the difference between WT and KO BMDMs. This data is consisting with IB shown in Figure. 5f.

Response:
The reviewer's point is well taken. We first perform the IRF1 level with longer explorer in Fig.5d. The data indicate steady state level of IRF1 in KO cells remains higher than WT one. Furthermore, we agree with the reviewer's concern for the difference of IRF1 in steady state in Fig. 6a. Actually, we have already repeated these results for other twice. We replace the IRF1 IB data with our repeating data. The new data together indicate both of basal and inducible IRF1 are upregulated in Fam73b KO BMDMs.
• Most of the ChIP-Ub blots are not entirely convincing.

Response:
We apologize for not presenting CHIP-Ub data with best quality. However, we are still confidence for our conclusion, because the specific band of monoubiquitinated and regular CHIP disappeared when silencing its expression with shRNA in supplementary Fig. 8c. Furthermore, to respond to the reviewer's concern, we repeated some crucial ChIP-Ub blots as shown in Fig. 6a, 6f. In Parkin KO BMDMs, we also evaluated ChIP-Ub level (Fig. 7f). All of these results indicate that monoubiquitinated CHIP degradation is regulated by Parkin and tightly control IRF1 stability.

Response:
This is a quite excellent question. Based on previous data in Fig. 6a, monoUbi-ChIP performs modern degradation in total cell lyses. However, due to the colocalization with mitochondria, we assume that monoUbi-ChIP may mainly be degraded in cytoplasm. Therefore, we carefully evaluated the level of monoUbi-ChIP in cytoplasm with LPS stimulation. As shown in supplementary Fig.8d, the results indicated that monoUbi-ChIP was degraded more significantly in cytoplasm than total cell lyses.

• The authors need to better introduce Fam73b in the Introduction section, since this is a major focus of the story and most readers would not be familiar with the protein.
Likewise, if the author decides to keep CHIP and Ube2w in the story, they need to be better introduced so that the reader understands the biological relevance of their control of IL12a gene induction.

Response:
The reviewer's point is well taken. We add some introductions of Fam73b and CHIP biological function on IRF1 stability in first section on Page 3 and Page 4-5. In additionally, we follow the reviewer suggestions to delete all Ube2w data to make the story more clearly.
• Line 346: "IRF1 silencing specifically promoted Il12a induction" is the opposite of what the authors want to say.

Response:
We apologize for the error, and we have made the correction.

Response:
We apologize for not explaining the results clearly. Actually, we find both of Fam73a and Fam73b are potential suppressors for IRF1 stability. We replace the description as "Interestingly, IB analyses revealed that both of Fam73b (Fig. 5d) and Fam73a KO (Supplementary Fig. 8a) macrophages displayed a higher IRF1 protein level".
• Fig S6f: What is the difference between the left and right panels?

Response:
We apologize for not labeling the results clearly in supplementary Figure 7e. The left panel is basal level of OCR. The right one is OCR level with LPS stimulation for 6 hours. We add the legends back for these two panels.

Main concerns:
To decipher the molecular mechanisms promoting IL-12 production downstream of mitochondrial fission, the authors have knocked down IRF1, Parkin and CHIP using shRNA in WT or Fam73 KO BMDM. As KO mice are available for IRF1 and Parkin, I guess that their conclusions should be confirmed by using BMDM from these KO mice. I mean for instance, IL12 production in IRF1 KO BMDM should be less important than in WT BMDM after TLR4 stimulation. The use of Parkin KO BMDM would be also interesting.

Response:
This is an excellent question. Following the reviewer's suggestion, we carefully evaluate cytokine induction either in IRF1 KO or Parkin KO BMDMs. Consistent with our conclusion, loss of IL-12 mRNA expression is observed in activated macrophages from IRF-1−/−mice (Fig. 5g). Furthermore, Parkin deficiency suppresses IL-12 expression, but promotes IL-10 and IL-23 induction (Fig. 7e). Depletion of Parkin also stabilizes monoubiquitinated CHIP and inhibits IRF1 level (Fig. 7f). These data together suggest that Parkin-IRF1 signal pathway contributes to Fam73b KO-mediated sorts of phenotypes.

Specific comments:
• In Fig 1a. It appears that 50 % of the BMDM have already short/fragmented mitochondria. TLR4 stimulation for 2 hrs with 500ng/m of LPS leads to a 20% increase in mitochondrial fragmentation. It does not seem to be the extensive mitochondrial fragmentation described in Fig 1c with time-lapse microscopy.

Response:
We apologize for not clearly describing the data. We do not evaluate the extension of mitochondrial fission with time-lapse microscopy. We have already modified the text in this section.
• 500 ng/ml of LPS is also quite a high concentration to stimulate BMDM. Do lowers concentrations of LPS also trigger mitochondrial fragmentation?

Response:
This is an excellent question. Following the reviewer's suggestion, we carefully detect the effect of LPS with lower doses (10ng/ml and 100ng/ml). As shown in supplementary Fig. 1a, the intensity of mitochondrial fragment is dose dependent. The high dose LPS do not trigger mitochondrial fragmentation unlimited, because the dose with 100ng/ml have already performed similar phenotype of mitochondrial fission.

Response:
To respond to the reviewer's concern, we performed new experiments. The new data showed that IRF1 K48-linked ubiquitination was significantly impaired with Fam73b deficiency in Fig. 5i.
• In Fig 7k, it is not that obvious that Parkin preferentially promotes K48-linked ubiquitination of CHIP.

Response:
To respond to the reviewer's question, we perform new experiments and show that Parkin significantly promoted K48-linked ubiquitination of CHIP in Fig. 7k. However, we still observe a moderate enhancement of CHIP polyUbiquitination in K48R group. This data indicate Parkin also enhance other subtypes of CHIP polyUbiquitination. Therefore, we modified our conclusion that CHIP was preferred but not only to select K48-linked polyubiquitin chains by Parkin (Fig. 7k).

Response:
To address the reviewer's question, we perform new experiments and show that Mdivi1 treatment significantly suppresses Parkin expression and stabilize CHIP monoubiquitination in Fam73b KO BMDMs (Fig.6g). In additionally, CHIP recruitment is also restored due to mitochondrial fusion (Fig.6h).