Defective mitochondrial COX1 translation due to loss of COX14 function triggers ROS-induced inflammation in mouse liver

Mitochondrial oxidative phosphorylation (OXPHOS) fuels cellular ATP demands. OXPHOS defects lead to severe human disorders with unexplained tissue specific pathologies. Mitochondrial gene expression is essential for OXPHOS biogenesis since core subunits of the complexes are mitochondrial-encoded. COX14 is required for translation of COX1, the central mitochondrial-encoded subunit of complex IV. Here we describe a COX14 mutant mouse corresponding to a patient with complex IV deficiency. COX14M19I mice display broad tissue-specific pathologies. A hallmark phenotype is severe liver inflammation linked to release of mitochondrial RNA into the cytosol sensed by RIG-1 pathway. We find that mitochondrial RNA release is triggered by increased reactive oxygen species production in the deficiency of complex IV. Additionally, we describe a COA3Y72C mouse, affected in an assembly factor that cooperates with COX14 in early COX1 biogenesis, which displays a similar yet milder inflammatory phenotype. Our study provides insight into a link between defective mitochondrial gene expression and tissue-specific inflammation.

The assembly of cytochrome c oxidase (COX), the last enzyme of the mitochondrial respiratory chain, requires the function of numerous assembly factors, two of which, COX14 and COA3, are required for the expression and assembly of COX subunit 1.Mutations in both factors have been reported in patients with mitochondrial disorders, although with different severity.How mutations comparably affecting COX lead to different pathological manifestations remains to be fully elucidated.To gain insights into the pathophysiology of these disorders, in the manuscript by Aich et al, the authors generated and characterized two murine models recapitulating the COX14 and COA3 patient mutations.The mutation in COX14 is associated with multisystemic alterations in several tissue, with the liver being the most affected organ.Notably, mutant hepatocytes present an increase in mitochondrial ROS levels triggering mitochondrial nucleic acid release in the cytosol and type I interferon-mediate inflammation.In my opinion, the experiments are well designed and convincingly established the molecular mechanisms leading to the observed severe liver inflammation, which represent a central aspect of the pathological phenotype.However, although emphasis is placed on tissue specificity, it seems that the severity of the pathology correlates for most part with CIV levels/activity.Moreover, compared to the COX14 model, CIV is less affected in liver from the COA3 mutant mice, which present a milder phenotype.The reason why CIV accumulates at different levels in different organ/mice remains to be elucidated.Lastly, the authors observed the formation of extensive and numerous contact sites between mitochondria and lipid droplets in COX14 mutant hepatocytes.While it remains descriptive, this is an interesting observation that grants further investigation.Minor points: -It is very clear that lack of COX14 dramatically affects COX1 synthesis.However, the defect in newly synthesized COX1 stability is less convincing.In lane 4 of figure 1E, COX1 is undetectable, despite the lane appears overloaded, and it is difficult to assess its degradation.Please also indicate whether the mitochondria analyzed in figure 1D and E were isolated from liver.Additionally, the authors could comment on the stability of the other mitochondrion-encoded COX subunits, which stability has been reported to be compromised by defects in COX1 biogenesis.
Reviewer #2 (Remarks to the Author): The authors discovered that the release of mitochondrial RNA is stimulated by elevated production of reactive oxygen species.This conclusion was reached through an analysis of two genetically modified mice, each carrying a mutation in either COX14 or COA3.Both of these mutations are related to the biogenesis of COX1, which is a component of complex IV.By studying these mice, the authors elucidated a new pathway in which the increased production of reactive oxygen species, resulting from the loss of complex IV, triggers the release of mitochondrial RNA.This process also leads to the simultaneous onset of inflammation in the liver.These findings are highly intriguing as they indicate a strong connection between metabolic burdens like type 2 diabetes and steatohepatitis.The experiments were executed with great proficiency, and the interpretation is deemed satisfactory.In an effort to enhance the comprehensibility of this paper for the readers, there exist various remarks that necessitate further clarification.
1) Please provide an in-depth analysis of the results depicted in figures 1b, 1c, 5a, 5e, 5f, 7c, and 7d, highlighting the significance of these findings, as well as the individual data.
2) Please quantify all the results obtained from the Western blot analysis.
3) I believe it would be more advantageous to present more precise data in figure 2c rather than focusing on general changes of increase or decrease.4) Fig. 5G lacks data on MAVS, despite the author's mention of it in the results section.It is worth noting the heightened presence of ZBP1 protein in the COX1 mutant.
5) An explanation is lacking for Figure 6. 6) Could you provide further insight regarding the liver phenotype's preeminence and whether COX14 expression exhibits variations across different tissues?7) What is the underlying cause for the abbreviated lifespan observed in female mice with the COX14 mutation?Do disparities in liver phenotype exist between the sexes?Reviewer #3 (Remarks to the Author): This study addresses tissue-specific manifestations of mitochondrial diseases in two novel transgenic mouse models carrying patient-specific variants in COX14 and COA3, which are involved in the translation of cytochrome C oxidase (COX).Such studies are important because mitochondrial diseases are considered the largest group of inherited metabolic disorders, with more than >400 disease genes reported to date.Moreover, there are no effective therapies for these diseases and improved understanding of the molecular mechanisms may help in development of much needed treatments.
A long-standing question in the field has been why do mitochondrial diseases exhibit tissue-specific pathologies, when one might expect that the synthesis of ATP would be of importance to all tissues.Such tissue-specific manifestation does not seem to be explained simply by tissue-specific expression of disease-related genes as most are ubiquitously expressed and are considered typically essential for development and function in most organ systems.
During the last 5-10 years there has been a substantial amount of evidence published describing release of mitochondrial nucleic acids into the cytosol and activating the cellular antiviral signalling response leading to inflammation including in monogenic mitochondrial disease (Dhir, A. et al. Mitochondrial double-stranded RNA triggers antiviral signalling in humans. Nature 560, 238-242 (2018).Here, Aich et al demonstrate that this pathway is activated in COX14 mouse tissues, in particular the liver, and to a lesser extent the COA3 liver.Treatment of COX14 hepatocytes with the antioxidant N-acetycysteine for 24 hours attenuated accumulation of both mitochondrial RNAs in the cytosol and expression of inflammatory proteins.Therefore the main finding of this study and mechanistic insight concluded is that liver-specific inflammation is caused by a ROS-dependent release of mitochondrial nucleic acids into the cytosol.
In general experiments and data presented are to a high standard, the transgenic mice are wellphenotyped and manuscript is written well.The evidence supporting release of mitochondrial nucleic acids into the cytosol seems convincing but evidence for ROS involvement less so.

My specific comments:
-There is not a great deal of evidence to support the claim that the effects seen are ROS-dependent.Increased superoxide is shown in COX14 primary hepatocytes however I did not see anywhere the superoxide measurements following NAC treatment.Did NAC treatment cause a reduction in superoxide levels?-Can NAC treatment work through other redox mechanisms than ROS that might also explain the effects seen?-NAC only has a moderate effect on inflammatory mRNA gene expression (Fig 7C) but near complete reduction at protein level (Fig 7E )…..and a complete reduction in mitochondrial mRNA abundance in cytosol (Fig 7D).An explanation to reconcile these observations is lacking.
-Apparently, another antioxidant, Mito-tempo, was used (ln 352) with similar effects to NAC but data is not shown.
-The liver-specific claim of inflammation seems to be a little overplayed, especially when inflammation is observed in other tissues at mRNA level and in kidney at protein level (and in my opinion also the spleen.Spleen has increased OAS1A.And in other tissues IFIT1 is not measured, so it doesn't seem to be entirely fair comparison).
-It isn't really clear from the discussion to what extent we learn about tissue-specific manifestation other than in the two models presented there is a bit more inflammation in the liver than in other tissues.This could be important but there is isn't much in the discussion that puts this in the context of COX14 patients, and their multisystem disease, and mitochondrial diseases in general.In this regard, the final two concluding sentences of the discussion are somewhat at odds with the main message of the paper: "The concomitant loss of complex IV leads to increased ROS production, mitochondrial damage, and mtRNA release, which induces type I IFN inflammation in mice that affects not only liver but other major organ systems as well.Our findings provide a mechanistic explanation on how defective mitochondrial translation due to loss of COX14 function triggers ROS-induced type I IFN inflammation in liver leading to worsening pathology with time.-Minor comments -It is not specified always which sex or age mice are used e.g. in biochemical analyses Figure 1  -Ln P48 -"absence of complex IV" would reduction or deficiency be more appropriate term?-Ln 49 -"Additionally, we generated a COA3Y72C mouse, affected in COX1 biogenesis" -this is rather vague sentence, it would be good to state what COA3 is or does.-Ln "M19I exchange" -substitution or missense variant -Ln139 "Among the tested tissues, liver appeared to be the most affected tissue.In agreement with this observation, COX14M19I 140 mitochondria displayed a significant reduction in respiration (Fig. 1H)."How can such agreement be postulated when only liver in OCR shown?Please rephrase.
-Ln 202 "Since the liver fulfills key metabolic functions, we addressed changes in the gene expression pattern by RNA sequencing (RNA-seq) of wild type and COX14M19I mice liver samples" -I don't see the logic in this statement, what about the other tissues, they are also important right?I get that not everything can be done.You chose the liver because it seems worse affected?-Ln228 "Interestingly, many mitochondrial ribosomal proteins in this network map were upregulated" -Perhaps downregulated rather than downregulated.There is a mistake in the figure legend as both red and blue are said to be upregulated.

The assembly of cytochrome c oxidase (COX), the last enzyme of the mitochondrial respiratory chain, requires the function of numerous assembly factors, two of which, COX14 and COA3, are required for the expression and assembly of COX subunit 1. Mutations in both factors have been reported in patients with mitochondrial disorders, although with different severity. How mutations comparably affecting COX lead to different pathological manifestations remains to be fully elucidated.
To gain insights into the pathophysiology of these disorders, in the manuscript by Aich et al, the authors generated and characterized two murine models recapitulating the COX14 and COA3 patient mutations.The mutation in COX14 is associated with multisystemic alterations in several tissue, with the liver being the most affected organ.Notably, mutant hepatocytes present an increase in mitochondrial ROS levels triggering mitochondrial nucleic acid release in the cytosol and type I interferon-mediate inflammation.In my opinion, the experiments are well designed and convincingly established the molecular mechanisms leading to the observed severe liver inflammation, which represent a central aspect of the pathological phenotype.
We thank the reviewer for the insights and comments to improve the manuscript.We have strived to address all concerns raised experimentally.
However, although emphasis is placed on tissue specificity, it seems that the severity of the pathology correlates for most part with CIV levels/activity.Moreover, compared to the COX14 model, CIV is less affected in liver from the COA3 mutant mice, which present a milder phenotype.The reason why CIV accumulates at different levels in different organ/mice remains to be elucidated.
Although a corelation with complex IV levels/activity exists, this appears to be not the only factor that contributes to the mitochondrial damage response.As we demonstrate below, e.g.ROS levels seem to be also tissue specifically altered.The topic of tissue specificity of mitochondrial dysfunction is an unresolved and very much investigated question in the field.Current concepts consider the mitochondrial turnover rate, what substrate are used by a given tissue to drive metabolism, and how easily each tissue regenerates as important aspects of the pathology.Accordingly, a plethora of factors add to the complexity of the tissue specificity and thus pathology.We have tried to make this clearer in the revised text.
As requested, we examined complex IV activity in different organs of the COA3 mice (new Fig 5d).Interestingly we observe a similar trend of reduced CIV activity in brain, muscle and liver.Yet, the COA3 mice did not display a significant CIV reduction in the heart.
The observed tissue phenotypes in liver and heart of COX14 mice correlate with complex IV activity.Brain and muscle display similar complex IV reduction, yet a brain phenotype was not apparent in our analyses.Moreover, while we notice the most drastic reduction of the mutant COX14 protein in liver, one may argue that the reduced levels of COX14 in liver correlate with the complex IV defect.Yet, this correlation is not as apparent in the other tissues.Accordingly, as described above, there seem to be several factors that contribute to tissue specificity of mitochondrial disorders and the exact correlation between molecular defect and pathology remains an open and very much investigated issue in the field.
Lastly, the authors observed the formation of extensive and numerous contact sites between mitochondria and lipid droplets in COX14 mutant hepatocytes.While it remains descriptive, this is an interesting observation that grants further investigation.
To further investigate this, we isolated lipid body-associated and non-associated (cytosolic) mitochondria from WT and COX14 M19I mice livers (Supplementary Figure 5a).However, we found no protein candidates that differed significantly in these fractions and would explain the formation of the extensive contact sites (Supplementary Figure 5b).Additionally, we did not find functional difference in the respiratory capacities of the mitochondria (Supplementary Figure 5c).We included this data into the manuscript to provide a full report of the phenotypic analyses.Yet, the molecular detail of the altered cell biology remains elusive even though we extend the analysis significantly in the revised version.
Minor points: -It is very clear that lack of COX14 dramatically affects COX1 synthesis.However, the defect in newly synthesized COX1 stability is less convincing.In lane 4 of figure 1E, COX1 is undetectable, despite the lane appears overloaded, and it is difficult to assess its degradation.Please also indicate whether the mitochondria analyzed in figure 1D and E were isolated from liver.Additionally, the authors could comment on the stability of the other mitochondrion-encoded COX subunits, which stability has been reported to be compromised by defects in COX1 biogenesis.
As requested, we have replaced Figure 1E with a higher exposed version from the same experiment.Here the COX1 is clearly visible and correlates with the quantification.In addition, a similar quantification was done for the COX2/COX3 band and it is represented bellow.There is no difference between the WT and Mutant.We would be happy to do this experiment.Yet, the experiment would require a specific permission.Due to political pressure to minimize animal experiments in the state of Lower Saxony and elsewhere in Germany, the permissions will take about nine months to get.Hence, we are unable to perform these analyses within a reasonable timeframe.
We corrected the citation.
Reviewer #2 (Remarks to the Author): The authors discovered that the release of mitochondrial RNA is stimulated by elevated production of reactive oxygen species.This conclusion was reached through an analysis of two genetically modified mice, each carrying a mutation in either COX14 or COA3.Both of these mutations are related to the biogenesis of COX1, which is a component of complex IV.By studying these mice, the authors elucidated a new pathway in which the increased production of reactive oxygen species, resulting from the loss of complex IV, triggers the release of mitochondrial RNA.This process also leads to the simultaneous onset of inflammation in the liver.These findings are highly intriguing as they indicate a strong connection between metabolic burdens like type 2 diabetes and steatohepatitis.The experiments were executed with great proficiency, and the interpretation is deemed satisfactory.In an effort to enhance the comprehensibility of this paper for the readers, there exist various remarks that necessitate further clarification.
1) Please provide an in-depth analysis of the results depicted in figures 1b, 1c, 5a, 5e, 5f, 7c, and 7d, highlighting the significance of these findings, as well as the individual data.
As requested, the data in Figure 1 has been updated with p-values for individual changes and indicated by a *.For the panels of qPCR data in Figures 5 and 7, due to limited figure spacing, we generated tables with the statistical analysis.They are added at the bottom of this document (See pages 9 -18).
2) Please quantify all the results obtained from the Western blot analysis.
As requested, data obtained from Western blot analyses has been quantified and represented as bar graphs in the figures.
3) I believe it would be more advantageous to present more precise data in figure 2c rather than focusing on general changes of increase or decrease.
As requested, the actual change relative to WT is represented in revised Figure 2C and the raw data provided as Supplementary Table 1.We apologise for the error.The experiment tested for MDA5 and not MAVS; the mention of MAVS in the text was an error and has been corrected to be MDA5.As requested, we have mentioned the heightened presence of ZBP1 in the text.
5) An explanation is lacking for Figure 6.
As requested, we provide an explanation for Figure 6.

6) Could you provide further insight regarding the liver phenotype's preeminence and whether COX14 expression exhibits variations across different tissues?
We have extended the discussion on the tissue phenotypes as suggested by reviewer 1.As requested, we have determined COX14 protein levels (western blot) and gene expression (qPCR) across tissues (see below).Differences observed did not clearly correlate with the observed tissue phenotypes.The liver phenotype is similar in both the sexes.What could be different is how inflammation is responded to systemically.Additionally, the physiological response is different amongst sexes, see changes in body weight (Fig. 2a).It could be that additional factors such as hormones, pregnancy, and ROS production during aging could lead to the effect on lifespan.Reviewer #3 (Remarks to the Author):

Heart Brain Liver
This study addresses tissue-specific manifestations of mitochondrial diseases in two novel transgenic mouse models carrying patient-specific variants in COX14 and COA3, which are involved in the translation of cytochrome C oxidase (COX).Such studies are important because mitochondrial diseases are considered the largest group of inherited metabolic disorders, with more than >400 disease genes reported to date.Moreover, there are no effective therapies for these diseases and improved understanding of the molecular mechanisms may help in development of much needed treatments.
A long-standing question in the field has been why do mitochondrial diseases exhibit tissue-specific pathologies, when one might expect that the synthesis of ATP would be of importance to all tissues.Such tissue-specific manifestation does not seem to be explained simply by tissue-specific expression of disease-related genes as most are ubiquitously expressed and are considered typically essential for development and function in most organ systems.
During the last 5-10 years there has been a substantial amount of evidence published describing release of mitochondrial nucleic acids into the cytosol and activating the cellular antiviral signalling response leading to inflammation including in monogenic mitochondrial disease (Dhir, A. et  In general experiments and data presented are to a high standard, the transgenic mice are wellphenotyped and manuscript is written well.The evidence supporting release of mitochondrial nucleic acids into the cytosol seems convincing but evidence for ROS involvement less so.

My specific comments:
-There is not a great deal of evidence to support the claim that the effects seen are ROS-dependent.Increased superoxide is shown in COX14 primary hepatocytes however I did not see anywhere the superoxide measurements following NAC treatment.Did NAC treatment cause a reduction in superoxide levels?
As requested, we measured the superoxide levels after NAC treatments and add this experimental data as Fig. 7f.As expected, NAC treatment caused a reduction in superoxide levels.
-Can NAC treatment work through other redox mechanisms than ROS that might also explain the effects seen?
We clarify in the revised text that it is possible that NAC acts via other redox mechanisms through glutathion.Yet, since we obtain similar results with MitoTempo (Fig. 7g), we provide additional support indicating that ROS are the central mediator of mitochondrial damage and mtRNA release. -

NAC only has a moderate effect on inflammatory mRNA gene expression (Fig 7C) but near complete reduction at protein level (Fig 7E)…..and a complete reduction in mitochondrial mRNA abundance in cytosol (Fig 7D
).An explanation to reconcile these observations is lacking.
As requested, an explanation to reconcile these observations has been added into the main text.
-Apparently, another antioxidant, Mito-tempo, was used (ln 352) with similar effects to NAC but data is not shown.
As requested, the data is shown as Figure 7G.
-The liver-specific claim of inflammation seems to be a little overplayed, especially when inflammation is observed in other tissues at mRNA level and in kidney at protein level (and in my opinion also the spleen.Spleen has increased OAS1A.And in other tissues IFIT1 is not measured, so it doesn't seem to be entirely fair comparison).
As requested, the data for IFIT1 was added as Supplementary Figure 6.In addition we address the inflammatory phenotype across tissues in the revised discussion.
-It isn't really clear from the discussion to what extent we learn about tissue-specific manifestation other than in the two models presented there is a bit more inflammation in the liver than in other tissues.This could be important but there is isn't much in the discussion that puts this in the context of COX14 patients, and their multisystem disease, and mitochondrial diseases in general.In this regard, the final two concluding sentences of the discussion are somewhat at odds with the main message of the paper: "The concomitant loss of complex IV leads to increased ROS production, mitochondrial damage, and mtRNA release, which induces type I IFN inflammation in mice that affects not only liver but other major organ systems as well.Our findings provide a mechanistic explanation on how defective mitochondrial translation due to loss of COX14 function triggers ROSinduced type I IFN inflammation in liver leading to worsening pathology with time.
As requested, we have tried to discussed the phenotype of the mice in the context of the human patient.Yet, keeping in mind that the patient died shortly after birth and the phenotype has not been fully assessed.Based on the reviewer's recommendation, we have modified these two sentences in the main text.We have added details to the figure legends.
-Fig 1C and Supp Fig 1B. COX14 protein amount is measured relative to WT. COX14 have been normalised to VDAC for total protein for each sample but also ideally normalisation would be done to another mitochondrial protein to control for mitochondrial mass.
COX14 has been normalised to RIESKE and not VDAC in the original analysis.We are grateful to the reviewer to address this aspect.We have corrected this in the revised legend.Additionally, as requested, we performed an additional normalisation to ATP5B and see the pattern to be the same (see graph bellow).We have made modifications as suggested.

B r a i n L i v
-Ln P48 -"absence of complex IV" would reduction or deficiency be more appropriate term?
Changed accordingly.
-Ln 49 -"Additionally, we generated a COA3Y72C mouse, affected in COX1 biogenesis" -this is rather vague sentence, it would be good to state what COA3 is or does.
Changed accordingly.
-Ln139 "Among the tested tissues, liver appeared to be the most affected tissue.In agreement with this observation, COX14M19I 140 mitochondria displayed a significant reduction in respiration (Fig. 1H)."How can such agreement be postulated when only liver in OCR shown?Please rephrase.
Changed accordingly.
-Ln 202 "Since the liver fulfills key metabolic functions, we addressed changes in the gene expression pattern by RNA sequencing (RNA-seq) of wild type and COX14M19I mice liver samples" I don't see the logic in this statement, what about the other tissues, they are also important right?I get that not everything can be done.You chose the liver because it seems worse affected?
Changed accordingly.
-Ln228 "Interestingly, many mitochondrial ribosomal proteins in this network map were upregulated" -Perhaps downregulated rather than downregulated.There is a mistake in the figure legend as both red and blue are said to be upregulated.
Changed accordingly.

Figure 5A
Number and Supp fig 1, Fig 2C.-Fig 1C and Supp Fig 1B.COX14 protein amount is measured relative to WT. COX14 have been normalised to VDAC for total protein for each sample but also ideally normalisation would be done to another mitochondrial protein to control for mitochondrial mass.Fig 3a/b -some of the most significant changes are not annotated apparently because they do not belong the relevant pathways highlighted.But it is not really clear until you come to Fig 3c which gives a summary of pathways.Perhaps order could be rearranged so that 3C comes first?

-
Treatment with NAC reduces mtRNA-dependent ISG up-regulation.Does the treatment ameliorate the liver pathology?

4
) Fig.5Glacks data on MAVS, despite the author's mention of it in the results section.It is worth noting the heightened presence of ZBP1 protein in the COX1 mutant.

Figure a :
Figure a: Quantification of steady state protein levels of COX14 in wild type mouse heart, brain, liver, muscle, spleen, kidney and eye in relation to RIESKE.

-
It is not specified always which sex or age mice are used e.g. in biochemical analyses Figure 1 and Supp fig 1, Fig 2C.
Fig 3a/b -some of the most significant changes are not annotated apparently because they do not belong the relevant pathways highlighted.But it is not really clear until you come to Fig 3c whichgives a summary of pathways.Perhaps order could be rearranged so that 3C comes first?