Extracellular mRNA transported to the nucleus exerts translation-independent function

RNA in extracellular vesicles (EVs) are uptaken by cells, where they regulate fundamental cellular functions. EV-derived mRNA in recipient cells can be translated. However, it is still elusive whether “naked nonvesicular extracellular mRNA” (nex-mRNA) that are not packed in EVs can be uptaken by cells and, if so, whether they have any functions in recipient cells. Here, we show the entrance of nex-mRNA in the nucleus, where they exert a translation-independent function. Human nex-interleukin-1β (IL1β)-mRNA outside cells proved to be captured by RNA-binding zinc finger CCCH domain containing protein 12D (ZC3H12D)-expressing human natural killer (NK) cells. ZC3H12D recruited to the cell membrane binds to the 3′-untranslated region of nex-IL1β-mRNA and transports it to the nucleus. The nex-IL1β-mRNA in the NK cell nucleus upregulates antiapoptotic gene expression, migration activity, and interferon-γ production, leading to the killing of cancer cells and antimetastasis in mice. These results implicate the diverse actions of mRNA.

1. Page 4-5. I guess it would be good to highlight that the experiment treating cells from spleen with TCM has as a goal to show that cells that are not within the tumour environment can respond to TCM treatment. Why authors did not compared cells from wt vs ZC3H12D KO mice? 2. I don't understand the reasoning behind using spleen gene expression profile to identify the potential regulatory nex-mRNAs. Wouldn't make sense to analyse the supernatant of cultured tumours (TCM) after purification of extra vesicle RNA? Is really the analysis used adequate to identify nex-mRNA by transcriptomic profiling? Maybe I have missed something here. Moreover, I don't think I fully understand the reasoning behind the selection of IL1B mRNA as from my point of view is not fully justified here. Weren't other RNAs as upregulated as IL1B mRNA (based on Fig S4)? Was this a lucky guess or there was a reason to think that IL1B mRNA was going to be a) extracellular and extra vesicle and b) would produce a phenotype. As I said, I may have missed something here and if so, apologies for this. 3. Page 6. I am not familiar with the extra-exosomal sample preparation used here for the RNA analysis by RT-qPCR. After reading the mat and met I am still confused as I am not familiar with the kits. Maybe some sentences to clarify the purpose of each step would help. Do authors control for extra-exosomal RNA using RNase treatment (that should digest unprotected RNA)? 4. Page 6 / Figure 2a. While authors discuss an interaction between ZC3H12D and IL1B mRNA, the microscopy shows a clear exclusion of the fluorescence derived from each molecule. Since the microscopy used is expected to have low resolution (200-300 µm), if both molecules do interact, they would be expected to be present at (at least partially) overlapping areas instead of juxtaposed. Moreover, I don't understand why other FITC-labelled RNAs weren't used to show specificity and why the NoCM control is not shown in the microscopy above. 5. Fig. 2b. How is copy number estimated? Why other (control) RNAs aren't tested? Ideally this immunoprecipitation should be combined with RNAseq of the eluates to provide an overview of the RNAs bound by ZC3H12D and show that IL1B mRNA is a dominant target. Otherwise, it is difficult to justify why focusing on IL1B mRNA and not others. It is also difficult to identify positive and negative RNA controls to use in downstream experiments. Figure legends of panels d-e are not sufficiently elaborated to understand what these plots show (at least for people not familiar with them). At a glance, I can see some coherency regarding the Y axis, but a large data distribution in the X axis. Is this what authors expected? 6. Fig. 3b. A control with unlabelled IL1B mRNA is missing. The signal observed is very close to noise levels, so this control is critical to differentiate between signal and noise. Moreover, the experiments in Figure 3 are generated with large amounts of exogenous IL1B mRNA. Authors should prove that in physiological TCM IL1B mRNA concentrations (i.e. the one found in TCM), the RNA would be detected in the nucleus. 7. Figure 3c. Not sure what is the Wt panel as authors only show Zc3h12d +/-and -/-. Without the +/+ control this experiment is inconclusive and difficult to evaluate. All the experiments in figure 3 should include an unlabelled RNA control to define the autofluorescence level. Without this, it is very difficult to assess whether the observed fluorescence is signal or autofluorescence. 8. Figure 4. Can authors show what is the affinity of Zc3h12d for the target RNA in the EMSA assay? In Figure 4e, without a splitting of the channels, the inclusion of a negative RNA control, and a paralleled Zc3h12d KO line analysis, it is not possible to fully interpret and draw solid conclusions from this microscopy experiment. 9. Figure 5. If the effect is independent of translation, I would expect to see the same effect if one or more stop codons are inserted into the IL1B mRNA to avoid the production of the protein. This would be the only way to show that the effect is translation-independent although it cannot be excluded the possibility that IL1B mRNA interaction with membrane proteins tigers a signalling cascade through, for example, RNA-binding toll-like receptors. Controls showing that toll-like receptor pathway is not activated should be shown. Unfortunately, the Cap/poly(A) -is a good control but not sufficient to rule translation and toll-like receptor roles in the observed effects. 1) An mRNA lacking cap and poly(A) can be translated if added in large quantities. 2) An mRNA lacking cap and poly(A) will exhibit lower stability as it is susceptible to exonucleases. These two factors can explain why authors see the effects when adding large amounts of cap and poly(A) less RNA. Controls using RNAs with near identical sequence but that cannot be translated are thus essential here. Also showing that other RNAs do not produce this effect when used at the very same (or even higher) concentration, would be critical to rule out RNA sensing and triggering of the antiviral programme. This criticism is applicable to Fig 6. 10. Figure S4b. Where are the controls to determine the basal levels of H2AX phosphorylation and the specificity of the effect (e.g. does it happens when adding another exogenous RNA at similar concentrations? 11. Overall figure legends must be expanded to allow readers to understand the experiments (see above). For example, where is derived the fluorescence measured in Fig 5c derived from? How is this related to necrosis?
Minor points Page 5. Why is the data from DNase treatment not shown? I think this is very important data and should be presented to back up the conclusions. Figure 1. Not sure LTCM, BTCM and ETCM are defined in the text or figure legend. Figure 2a. What is the label of the Y axis in the barplot? Figure S2c. The co-localisation experiment is not very clear and to me it seems that Zc3h12d may colocalise partially with almost every marker showed. A more refined analysis using monitoring the fluorescence distribution profile of each fluorophore should be done to draw any conclusion. Page 7. Do authors refer to FITC-labelled RNA with or without cap and poly(A) tail? Otherwise I do not understand what authors meant by 'its cap and poly(A) adducts'. I suggest to be very clear here. In their manuscript "Extracellular mRNA transported to the nucleus exerts translation-independent function", Tomita et al describe a novel function for the protein ZC3H12D as a mediator RNA uptake from the extracellular space. They also provide evidence that ZC3H12D-mediated uptake of nonvesicular, extracellular IL1B mRNA leads to cellular stress, and, in the case of a specific NK cell population, their activation and increased anti-tumoral activity.
Although ZC3H12D is known to be an RNA binding protein, up to now its function has been assumed to be similar to that of ZC3H12A or "Regnase", which binds and degrades the mRNA of specific proinflammatory cytokines. Thus, the RNA uptake function described by Tomita and colleagues is truly novel and surprising. Moreover, the described non-transcriptional function of IL1B mRNA in NK cells is also novel and unexpected.
Altogether, this study represents a significant advance for the field and will be of great interest to immunologists and cell biologists. However, precisely due to the really surprising nature of its findings, this reviewer also thinks that several control experiments are necessary before the manuscript is suitable for publication. In general, these controls include (i) investigating whether other mRNAs are also bound and internalized by ZC3H12D, in particular those found in the array in Figure 1 / Table S2, (ii) using β-actin or another non-ZC3H12D-binding RNA as a control for experiments, (iii) including human and other murine cell lines for RNA uptake, cellular stress, etc. The specific experiments are detailed below.

Figure 1
Major points: - Figure 1a: According to array data in biogps (http://biogps.org/#goto=genereport&id=340152), ZC3H12D should be broadly expressed in PBMCs. However, it is unclear if all of these cells really express ZC3H12D protein and internalize ZC3H12D upon exposure to tumor supernatant. Could the authors please co-stain for T-cells, B-cells, NK-cells and Monocytes in this experiment? This would provide valuable information on the role of ZC3H12D in these cell types.
-The authors then switch to the use of RAW macrophages. Since murine macrophages also express ZC3H12D, these data are valuable, but do not necessarily reflect the situation in PBMC. Could the authors also include another cell line but perhaps more representative for blood immune cells, e.g. a murine T-cell or B-cell line, if necessary with overexpression of ZC3H12D? -Since data with human tumor cells and supernatants are also used in the figure, human PBMC and/or using a human leukocytic cell line, e.g. THP-1, Daudi, NK-92 cells, should be included to visualize ZC3H12D internalization. If no antibody can be found, this could be done using overexpressed tagged protein in a one of the cell lines. -Please include the other genes from the gene expression array/ -What cells are shown in 1c? "Some cells from the spleen" is rather vague.
-Could you include a graphical schema of the experiment in 1g? This would greatly help the reader.  Major points: -As above, the experiment in Fig. 2a should also be performed with 1-2 further cell lines (murine blood cells, human cells) -The FCS experiment in Fig. 2c should also be performed with a non-binding RNA such as β-actin as a control to make the data easier to interpret. Doesn't the 3'UTR of IL1B also interact with ZCH12D?
Minor points: -Please calculate the colocalization in Fig. 2a using Pearson, Manders or a similar approach.   It is undoubtedly interesting that IL1B mRNA induces H2AX phosphorylation in a ZC3H12Ddependent manner, but it does not necessarily mean that cell-intrinsic cell stress leads to NK-cell activation and IFNγ induction --in particular, because H2AX phosphorylation is indicative of DNA damage and often followed by apoptosis (rather than increased survival) in other cell types.
Please address the following questions: -Does the observed response to IL1B mRNA/H2AX phosphorylation specific to NK cells? oCan other DNA-damaging agents (e.g. camptothecin) provoke a similar response (activation, IFNγ, increased cytotoxicity) in NK cells? oDo other cell types (RAW macrophages, etc) react differently to ZCH12D-mediated IL1B mRNA uptake? Is there H2AX phosphorylation? Do cells become apoptotic? -As NK cells can be activated by interaction with "stressed cells", is the response really cell intrinsic? This can be tested by incubating other cells, such as IL1bm RNA-stimulated WT and Zc3h12d-/-BMDM, with NK cells in an in vitro killing assay. Alternatively, one could overexpress ZC3H12D in a target cell line.
-Are the genes upregulated in ZC+ RAW in Fig 5e and 5f also upregulated in B220+CD11c+NK1.1+ cells? oHow were the genes in Table S3 picked? Are they part of the same microarray as in figure 1? -In the assays in Fig 6b and 6c, it would be important to include a control RNA (e.g. β-actin). Although the authors have thought to cap and 2'O-methylate the IVT mRNA, unmodified RNA could activate RIG-I in NK cells, which also leads to their activation. (In Fig. 6a this control was performed.) Minor points: -The necrosis assay in 5c should be explained a little. It is somewhat unclear what exactly was measured. -It is interesting that IL1b mRNA induces IFNγ in the CD56dim, but not the CD56bright, NK cell population. Please compare IL1B mRNA uptake in these cell populations.
-If the uptake is still observed for CD56bright NK cells, does IL1B mRNA uptake lead to the upregulation of other NK cell activation markers on CD56bright NK cells (e.g. NKG2D, NKp46)? What about killing activity? Discussion: Several important points are currently missing in the discussion.
-Please place your data into the context of what has been previously published about ZC3H12D, i.e. you do not observe RNase activity, you do not observe "Regnase-like" activity.
-Which human NK cell population is closer to B220+CD11c+NK1.1+? According to Blasius et al, 2007. These cells should be more similar to the CD56bright human population, yet this doesn't fit with your observations. This should be discussed.
-Do you think that ZC3H12D is also involved in the uptake of other RNAs? ZC3H12D is evolutionarily highly conserved, with putative orthologs found even in protozoa(https://www.uniprot.org/uniprot/?query=taxonomy:5811%20zc3h12d). In contrast, IL1B is only found in vertebrates. Thus, it seems highly unlikely that ZC3H12D only acts to transport IL1B mRNA.
-How do these extravesicular RNAs survive in the serum? #Q1: Authors report here a very interesting story that if is true will be paradigmatic for extracellular RNA function. However, with the data presented and the controls missed, I am not sure that all the central claims of this work are well supported. It is often the case that figures lack critical controls or seem incomplete (see below). I have also missed the use of a marker to delimit the cytoplasm in nucleus/cytoplasm localisation studies (e.g. Phalloidin) as the cytoplasm of the cells used appears to be small. Anyway, I think the 'killing' experiment is still missing. It would be to perform the experiments with an IL1B mRNA at physiological concentration including one or more stop codons to avoid the production of IL1B protein.
This, with additional controls to show that the toll-like receptor pathway is not the cause of the observed effects (and I think this possibility is likely), would give a lot of value to the paper. #A1: We appreciate the reviewer's comments; they were very thoughtful and constructive. We have responded to them and believe we have provided more concrete evidence, especially concerning the translation-independent role of nex-mRNA, in the revised manuscript. Strikingly, pretreatment of TCM with RNase (TCM+RNase) did not affect the localization pattern of ZC3H12D (wt panel in Fig. 1d). The addition of RNA isolated from TCM resulted in the same phenomenon (wt panel, RNA-TCM column in Fig. 1d).
These data suggest that RNA drives the translocation of ZC3H12D. On the contrary, this was not observed when we used splenic leukocytes from Zc3h12d-/-mice (Fig 1d). #A3: We apologize for the confusion. We did not show whether the lung-conditioned media from lungs cultured with TCM (Lung-TCM) contained IL1β-mRNA and βactin-mRNA. We added the new microarray data (GSE161219) in a revised Table S2.
High amounts of non-vesicular IL1β-mRNA and βactin-mRNA were detected in the Lung-TCM, and both genes were ranked in the top 20 genes in the microarray dataset.
We obtained candidates including IL1β-mRNA and βactin-mRNA from the Lung-TCM data. These genes should be further screened to compare wild-type, and Zc3h12d knockout mice since the expression level of ZC3H12D-related genes were affected by Zc3h12d knockout. Preparation of nex-mRNA from Lung-TCM requires many mice at once. Due to our animal facility's limitations, we were unable to prepare samples from Zc3h12d knockout mice. In addition, ZC3H12D expression was prominent in the spleen.
We considered that the spleen is more sensitive than other tissues in terms of the ZC3H12D knockout effect. Thus, we conducted a second screening using the microarray data shown in the revised Table S3 (Table S2 in the first submitted paper) generated by wild-type and Zc3h12d knockout mice. Based on the data in Table S3, we decided to add an immunoglobulin kappa variable (Igkv), resistin like gamma (Retnlg), matrix metallopeptidase 8 (Mmp8), βactin, and gapdh data analyses to Fig. 1. This criticism was also mentioned by reviewer 2. We used βactin as a negative biological control in this paper. We added sentences in the revised text shown below. (page 5, line

16-28)
Based on our data that TCM induced RNA-depending response of ZC3H12D location ( Fig. 1d), we hypothesized that some specific exRNA 24 bind to ZC3H12D. Because exRNA should be exposed to the ZC3H12d protein on leukocytes, we tried to examine which non-vesicular exRNA (nex-RNA) was increased in lung microenvironment that was stimulated by a tumor. We carried out a microarray analysis on the lungEC-TCM.
The top 20 ranked genes are listed in the table (Table S2). We then searched for mRNA differences in expression between wild-type and Zc3h12d-/-mouse spleens. Spleens were chosen because they had the highest ZC3H12D expression compared to other organs tested in this study. The rationale was that ZC3H12D depletion effects might be seen by comparing wild-type and Zc3h12d-/-spleen data. This comparison revealed that immunoglobulin kappa variable (Igkv), resistin like gamma (Retnlg), interleukin 1 beta (Il1β), and matrix metallopeptidase 8 (Mmp8) were upregulated in the Zc3h12d-/sample (Table S3). Among the upregulated genes, we focused on IL1β-mRNA because this gene exhibited relatively high expression of the nex-RNA derived from TCM-stimulated lung cells (Table S2). In addition, we chose βactin as a negative control in our study because it was abundantly found in the TCM-stimulated lungs (Table S2), and its expression levels did not change between wild-type and Zc3h12d-/mice (Table S3). Exosomes can be fractionated by size exclusion chromatography, ultracentrifugation, magnetic bead pull-down, and polymer precipitation (Methods Mol Biol, 728, 235-246, 2011). Among them, ExoQuick-TC (System Biosciences), a polymer used to precipitate exosomes, is the best method to eliminate exosomes from conditioned media.

#A5: We included a NoCM control-and TCM-induced ZC3H12D co-localization with
FITC-labeled IL1β-mRNA in Fig. 2a and Fig. S2a. In these images, Phalloidin was used as a marker. To quantify the co-localization signals of FITC-labeled IL1β-mRNA and ZC3H12D, we calculated Pearson's R values and described them in the revised text.
This criticism was also mentioned by reviewer 2. In addition, we showed clear co-localization data without Phalloidin staining (For reviewer data 1). We found that the signal became obscure after the membrane perforation process, required for Phalloidin staining. We added the sentence below to the text. (page6, line 27-29) Thirty minutes after TCM stimulation, the co-localization signals in ZC + RAW cells were more prominent (Pearson's R value is 0.47 ± 0.04, n=3) than the stimulated control, NoCM (arrow in Fig. 2a  #A6: The reason for selecting IL1β-mRNA as the target gene and other control genes such as βactin, gapdh, retnlg, mmp8, and Igkv is described above. Estimations of copy number and physiological concentrations of RNA are discussed below.
We quantified IL1β-mRNA in various TCMs or Lung-TCMs using a real-time PCR.
Data for the standard curves were prepared using diluted IL1β-cDNA. Copy number of the standard sample was calculated based on the absorbance at 260 nm. Real-time PCR gave us cycle threshold (Ct) values for IL1β-mRNA of the various TCMs or Lung-TCMs in Fig. 1. These Ct values were then converted to copy numbers by using the standard curve. To further convert copy number into concentration, we used the molecular weight of full-length IL1β-mRNA as 300 x 1500 (nts) = 4.5 x 10 5 (g). Thus, 1 ng of full-length IL1β-mRNA corresponds to 1.3 x 10 9 copies of the molecule. While this is not accurate, we considered it an acceptable approximation. According to this estimate, conditioned media contained about 1 pg/ml of extra-exosome RNA. Based on Fig. 2b, the amount of IL1β-mRNA bound to ZC3H12D + cells was almost 200~300 copies after the application of Lung-TCM originally containing 1.3 x 10 6 copies of IL1β-mRNA. Given the cell numbers, sample degradation in the media, and sample loss during the immunoprecipitation process, roughly 0.1%-1% of IL1β-mRNA was detected in ZC3H12D + cells within TCMs or Lung-TCMs. In the experiments, we observed clear biological responses such as migration and IFNγ induction of ZC3H12D + cells by 1~10 ng/ml of IL1β-mRNA. In addition, confocal microscopy systems required at least 10 ng/ml of FITC-labeled IL1β-mRNA to detect signals in the nucleus. Moreover, it was reported that 1,000-5,000 ng/ml of B-DNA and poly (I:C) were used as stimulators on nucleic acid-mediated innate immune responses in vitro (PNAS 108: 11542-11547, 2011). We also showed that passing TCM through a ZC3H12D protein column did not induce the migration activity in Fig S7a. Taken together, although it would be difficult to estimate the physiological concentration of IL1β-mRNA, somewhere between 1 pg/ml ~ 1 ng/ml of IL1β-mRNA might be expressed when tumors influence the tissues. In this revision, we carried out most experiments using minimal concentrations (10 ng/ml = 22 pM) of RNA, as the reviewer suggested. #A8: This is an important point. We repeated all the data acquisitions using non-labeled IL1β-mRNA, and these data were used as basal signals. As described above, we used 10 ng/ml (22 pM) of IL1β-mRNA instead of 100 ng/ml. As shown in Fig. 3, the signal intensities remained at a low level when 10 ng/ml of IL1β-mRNA was used, but at the same time, the difference between the labeled and non-labeled IL1β-mRNA became significant. We also added IL1β-stop-mRNA data as suggested by the reviewer. We  EMSA was conducted by one concentration only; it is not possible to draw any conclusion. In Fig. 4b, probe 5 displays a shifted band slightly more intense than the unshifted band. This implies that protein concentration in this assay was close to its Kd value, allowing us to make a very rough estimation of 10 nM-100 nM. To make a solid scientific statement about the affinity of the ZC3H12D-RNA target, we need to have more data. We want to let you know that our biochemical project to analyze the ZC3H12D-RNA target interaction in detail is ongoing. Figure 4e, without a splitting of the channels, the inclusion of a negative RNA control, and a paralleled Zc3h12d KO line analysis, it is not possible to fully interpret and draw solid conclusions from this microscopy experiment.

#Q11: In
#A11: We compared the uptake of FITC-labeled-IL1β-mRNA in B220 + CD11c + NK1.1 + cells derived from TCM-stimulated wild-type, Zc3h12d-/-, and Regnase-1-/-mice. In this experiment, the same littermate pairs were used. We used βactin-mRNA as a negative control. The data is shown in Fig. 5a and Fig. S5b.  #Q18: Figure S2c. The co-localisation experiment is not very clear and to me it seems that Zc3h12d may co-localise partially with almost every marker showed. A more refined analysis using monitoring the fluorescence distribution profile of each fluorophore should be done to draw any conclusion.
#A18: We calculated the co-localization signal of ZC3H12D and nuclear body markers to find out that the ZC3H12D signal was relatively high in nuclear speckles. (page 7, line 31-page 8, line 2) The quantitative data is shown in (Fig. S2d). function for the protein ZC3H12D as a mediator RNA uptake from the extracellular space. They also provide evidence that ZC3H12D-mediated uptake of non-vesicular, extracellular IL1B mRNA leads to cellular stress, and, in the case of a specific NK cell population, their activation and increased anti-tumoral activity.
Although ZC3H12D is known to be an RNA binding protein, up to now its function has been assumed to be similar to that of ZC3H12A or "Regnase", which binds and degrades the mRNA of specific proinflammatory cytokines.
Thus, the RNA uptake function described by Tomita and colleagues is truly novel and surprising. Moreover, the described non-transcriptional function of IL1B mRNA in NK cells is also novel and unexpected.
Altogether, this study represents a significant advance for the field and will be of great interest to immunologists and cell biologists. However, precisely due to the really surprising nature of its findings, this reviewer also thinks that several control experiments are necessary before the manuscript is suitable for publication. In general, these controls include (i) investigating whether other mRNAs are also bound and internalized by ZC3H12D, in particular those found in the array in Figure 1  #A21: We appreciate your valuable comments. We repeated the experiments using βactin and genes found in Table S2 (Table S3 in the revised manuscript) as suggested by the reviewer. Our FCS revealed that ZC3H12D has a non-specific binding site for RNA, but EMSA and other biological assays clarified that ZC3H12D recognizes the RNA sequence. Thus, we used the above mentioned RNAs, including βactin, as a negative control. In addition, we included a new human cell line, THP-1, as recommended. In the revised manuscript, modified or newly added parts are highlighted in red. The responses are shown one by one below. #A25: We showed splenic leukocytes after eliminating the red blood cells in Fig 1c, then we narrowed it down to NK cells because we had reported that those cells worked with anti-metastatic ability in the lungs. In the figure and text, we specified them as splenic leukocytes. We also renewed the data to show the relocation of ZC3H12D on splenic leukocytes derived from wild-type and Zc3h12d-/-mice in tumor-conditions in Fig. 1d, as reviewer 1 requested.
We added graphical schema to Fig. 1f (Fig. 1g in the first submission paper).
In the viewpoint of physical interactions between ZC3H12D and IL1β-mRNA, it would be displayed in Fig. 2. In this case, because we would like to emphasize the existence of nex-mRNA in biological samples, we decided Fig 1j data will stay in Fig. 1.

Figure 2
Major points: #Q26: -As above, the experiment in Fig. 2a should also be performed with

1-2 further cell lines (murine blood cells, human cells)
#A26: We added the ZC3H12D + THP-1 cell data. The cells were purified using a cell sorter and anti-ZC3H12D antibody. While we tried to establish ZC3H12D-overexpressing cells in murine and human cells, only 2 cell lines, murine RAW and human 786-O, have been established and used in this study. We observed the relocation of ZC3H12D protein from the cell surface to inside the cell by stimulation with IL1β-mRNA but not βactin in THP-1 cells (Fig. S2b). Incorporated IL1β-mRNA was transferred into the nucleus (Fig S3a). We also calculated the co-localization signals between the ZC3H12D protein and FITC-labeled  Fig. 2a and Fig. S2a). In ZC3H12D + THP-1 cells, isolated using a cell sorter and an anti-ZC3H12D antibody, similar co-localization signals were observed after applying IL1β-mRNA-FITC (Fig. S2b, Pearson's R value is 0.62 ± 0.08, n=3). Fig. 2c should also be performed with a non-binding RNA such as β-actin as a control to make the data easier to interpret. Doesn't the 3'UTR of IL1B also interact with ZCH12D?

#Q27: -The FCS experiment in
#A27: FCS has been used to demonstrate direct protein-RNA interactions in vitro. One reference was added in the text to support this statement. We performed FCS measurements using βactin-mRNA and IL1β-mRNA to find out both RNA bound ZC3H12D equally. The 3'UTR IL1β-mRNA also bound ZC3H12D. The results revealed that ZC3H12D has a non-sequence-specific binding site. On the other hand, EMSA clearly showed that ZC3H12D has a sequence-specific binding site, other bioassay data supported this conclusion. In EMSA, RNA with the ARE sequence remained on the ZC3H12D protein despite the presence of 100-fold amounts of competitors. FCS shows a direct interaction between two purified objects in solution but does not distinguish specific binding sites from non-specific binding sites and we could not conduct competitor experiments in FCS. The sequence-specific binding was emphasized in the biochemical assays because FITC-labeled-IL1β-mRNA but not βactin-mRNA seemed to bind to ZC3H12D on the cell membrane (Fig2a), implying that physiological binding structures play important roles. Our new project is deciphering detailed interactions between ZC3H12D and various RNAs, including βactin-mRNA and IL1β-mRNA and their fragments, using FCS and other spectroscopic techniques. It is not surprising to see non-specific binding and sequence-specific sites in the same protein (for example, p53).
A non-specific binding site may have a different role from a sequence-specific binding site. We modified sentences in the text as shown below. (page 7, line 4-9) (page 7, line

13-17)
The data showed that the ZC3H12D protein enriched IL1β-mRNA in the ZC+RAW cells (Fig. 2b). On the contrary, the gapdh-βactin-mRNAs control showed little enrichment in the condition (Fig 2b). We investigated direct interactions between the ZC3H12D protein and IL1β-mRNA using fluorescence correlation spectroscopy (FCS). This technique has been used to demonstrate a direct protein-RNA interaction in vitro (PNAS 104, 12306-12311, 2007).

The FCS data indicates that the dissociation constant for ZC3H12D-βactin and
IL1β-mRNA were in the order of 1 nM. We also confirmed that the 3′ untranslated region (3′UTR) of IL1β-mRNA had direct interaction with the ZC3H12D protein ( Fig.   2e and 2f). Because the 3'UTR of IL1β-mRNA had a different sequence from βactin-mRNA, the binding might include non-specific and sequence-specific interactions.
Minor points: #Q28: -Please calculate the colocalization in Fig. 2a using Pearson, Manders or a similar approach.

Figure 3:
Major points: #Q29: -As above, please include a human cell line  in the experiments for Fig. 3c and 3d.
#A29: We carried out FITC-labeled mRNA uptake in ZC3H12D + THP-1 cells and found that IL1β-mRNA but not βactin-mRNA was incorporated in these cells, although the signals in the nucleus (Fig S3a) were 30% lower than RAW and NK cells (Fig. 3a, 3c).
#Q30: -Please also use a control RNA such as β-actin for the experiments in  #Q33: - Fig. 4f would fit better in Figure 5.

Figure 5/6:
Major points: It is undoubtedly interesting that IL1B mRNA induces H2AX phosphorylation in a ZC3H12D-dependent manner, but it does not necessarily mean that cell-intrinsic cell stress leads to NK-cell activation and IFNγ induction --in particular, because H2AX phosphorylation is indicative of DNA damage and often followed by apoptosis (rather than increased survival) in other cell types.
Please address the following questions: #Q34: -Does the observed response to IL1B mRNA/H2AX phosphorylation specific to NK cells?
In the new Fig. 5b, IL1β-mRNA but not βactin-mRNA and gapdh-mRNA induced phospho-H2AX signals in the nucleus of wild-type NK cells. ZC3H12D + THP-1 also increased phospho-H2AX signals after the application of IL1β-mRNA (For reviewer Data 3). Then, we tested RAW cells. However, it was difficult to evaluate whether immortalized cell lines became apoptotic under this condition; we only presented the data for the reviewer.
#A35: We examined whether 1 µM of Camptothecin (CPT) induced IFNγ in B220 + CD11c + NK1.1 + NK cells during the 3 hr incubation time used for the mRNA, then checked IFNγ expression 24 hr after application. In this assay, CPT did not induce IFNγ in NK cells. We this data is shown as 'for reviewer data 4'.
#A36: We described this above. Zc3h12d-/-mice and incubated them in an L-cell conditioned medium for 7 days. Next, BMDMs were stimulated with IL1β-mRNA for 24 hr. After the IL1β-mRNA stimulation, cells were washed to remove IL1β-mRNA, and B220 + CD11c + NK1.1 + NK cells from wild-type mice were co-incubated with the primed BMDMs for 24 hr. Finally, the co-incubated NK cells were mixed with labeled tumor cells for 24 hr to evaluate their viabilities of labeled tumor cells. Dead cells were counted using the Zombie Green Fixable Viability Kit (BioLegend). We did not observe significant differences between the two groups (wt-BMDM and Zc3h12d-/--BMDM). We showed graphic schema and data in Fig. 6d and added sentences in the text as shown below. Fig 6d's legend was modified accordingly. (page 12, line 1-10) In this assay system, to account for the effects of RNA priming on the tumor cells, the NK cells were washed before being applied to tumor cells. Our data revealed that IL1β-mRNA priming increased the tumoricidal activity of NK cells originating from wild-type mice but not from the Zc3h12d-/-mice (Fig. 6c). Again, B220 + CD11c + NK1.1 + NK cells from Zc3h12d+/-mice had similar tumoricidal activity (Fig. S7b). To check if IL1β-mRNA-primed macrophages enhance this activity, we co-incubated NK cells with bone marrow-derived macrophages (BMDMs) stimulated by IL1β-mRNA (Fig 6d, upper). Both IL1β-mRNA-primed BMDMs derived from wild-type and Zc3h12d-/-mice did not have additive effects on the tumoricidal activity in vitro (Fig. 6d, lower). #A38: We added B220 + CD11c + NK1.1 + NK data in Fig. 5f and sentences in the text as shown below. (page11, line 9-10) Furthermore, expressions of Dusp1 and IL1rn were upregulated in B220 + CD11c + NK1.1 + NK cells after the application of IL1β-mRNA (Fig. 5f).
#Q39: oHow were the genes in Table S3 picked? Are they part of the same microarray as in figure 1?
#A39: We picked up the genes from the same array data shown in Table S3 and Table   S4 (Table S2 and Table S3 in the first submitted paper) (GSE104002). Genes listed in the new Table S4 were picked up based on gene ontology (GO). First, we picked up genes classified as nucleus component by GO, because our focus was on exotic RNA functions transported into the nucleus. Then, genes were sorted by fold-change value to identify genes with a fold-change value larger than 1.5. They are part of the same microarray shown in Fig. 1.   NK cell activation marker, NKG2D, in CD56 bright NK cells was slightly increased after the application of hIL1β-mRNA (Fig. S8e).
Discussion: Several important points are currently missing in the discussion.
We described the several issue about reviewer's comments in the discussion.  -2, IL-6, IL-10, and TNFα (J. Immunol 192, 1512-1524, 2014 1700051,2017). It is suggested that a unique substrate recognition ability was adopted in this protein to regulate RNA based on their structures (BioEssays 39, 1700051, 2017) during branching between ZC3H12D and other Regnase family genes.
According to Blasius et al, 2007. These cells should be more similar to the CD56bright human population, yet this doesn't fit with your observations. This should be discussed. - Figure 1d. There is some ZC3H12D in ZC3H12D KO cells. Can authors comment on this? It would be nice to have error bars and t test analysis to see if the differences are significant.
-RT-qPCR experiments. Can authors add error bars and statistics to all RT-qPCR experiments throughout the text?
- Figure 2 panels c and e. It is difficult to distinguish the different conditions given the choice of colours.
-Because the 3'UTR of IL1β-mRNA had a different sequence from βactin-mRNA, the binding might include non-specific and sequence-specific interactions. Maybe good to add 'in vitro'.
-I might be mistaken but I have the impression that the number of cells analysed by microscopy included in the bar plots in Figure 2 and 3 is very small (4-10 cells per condition). Since in a given experiment the number of cells is expected to be high, I don't understand why authors do not include more cells to exclude real signal from artefacts. The fact that data is consistent throughout the figures indicate the this is not the case, but it would be important to comment on this.
-Interestingly the IL1β-stop-mRNA clearly transported to the nucleus (Fig 3a). Why authors didn't include an illustrative image of this construct, which I think is critical for the interpretation of the results.
-To determine the effect of the ZC3H12D protein, TCM was divided into two parts, and these samples were passed through anti-FLAG-beads, with or without the ZC3H12D-FLAG tag protein. The nex-RNA was isolated from each sample independently. The first nex-RNA was expected to have less IL1β-mRNA than the second because it was captured in the ZC3H12D-FLAG-beads.
Thank you for the reviewer's further comments. We have responded to them and appreciate the reviewer's idea to use stop-codon-inserted IL1β-RNA to demonstrate that our observations were translation-independent.
For reviewer 1 Comment 1: Figure 1d. There is some ZC3H12D in ZC3H12D KO cells. Can authors comment on this? It would be nice to have error bars and t test analysis to see if the differences are significant.
Answer) In the flowcytometric analyses, we used anti-ZC3H12D antibody to stain spleen cells derived from tumor-conditioned medium (TCM)-stimulated mice. To set a gate to isolate ZC3H12D + cells, an isotype control antibody-staining data was used. Our results showed that ZC3H12D + cells were clearly separated from ZC3H12D-cells.
Nevertheless, small number of non-specific signals appeared in the gated region. Fc block treatment reduced the non-specific signal, but trace amount of signals remained in the gated region. We added sentence in the legend shown below. Answer) In the previous manuscript, Figure 2b showed simple bar graph depicting mean values for two biologically independent experiments. In the revised manuscript, we added the plot data. In addition, data used in the first submission were displayed in Figure S9. Although they lack βactin and gapdh data, we consider that the date help to show enrichment of IL1β-mRNA by ZC3H12D pull-down. In Figure S9, we showed Figure 1d, 1h, 1i, 1j, and 5f data obtained from biologically independent preparations.
In the Figure 1i dataset, the most important point is that extra-exosomal IL-1β-mRNA was increase in Lu-TCM compared with Lu-CM. This point was reproduced in the repeated experiment although the absolute values were different from the initial results.
Because an increase of βactin was observed in the repeated experiment, we would like to remove one sentence describing that βactin scarcely increased in the TCM condition.
Similarly, in the Figure 1j dataset, the most important point is that IL-1β-mRNA was capture by ZC3H12D protein beads. This point was reproduced in the repeated experiment, but the absolute values were different from the initial results. We also added note in the figure legend for Fig. 1, 2, 3, and 5, and added sentences in supplementary figure 9 shown below.
Repeated experiment data for Figs. 1d and h to j, 2b, 3a, and 5f. These data show the key points (IL-1β in Fig. 1i and j and 2b) were reproduced. Answer) To increase the visibility of the data, FCS data for ZC alone were depicted by dotted lines (2c and e), ZC+IL1β(3'UTR) 1 nM data was removed (2e), and line colors were changed (2e).
Comment 4-Because the 3'UTR of IL1β-mRNA had a different sequence from βactin-mRNA, the binding might include non-specific and sequence-specific interactions.
Maybe good to add 'in vitro'.
Answer) We added 'in vitro' in the text shown below.
Because the 3'UTR of IL1β-mRNA had a different sequence from βactin-mRNA, the binding might include nonspecific and sequence-specific interactions in vitro.
(page 7 in the revised text) Comment 5-I might be mistaken but I have the impression that the number of cells analysed by microscopy included in the bar plots in Figure 2 and 3 is very small (4-10 cells per condition). Since in a given experiment the number of cells is expected to be high, I don't understand why authors do not include more cells to exclude real signal from artefacts. The fact that data is consistent throughout the figures indicate the this is not the case, but it would be important to comment on this.
Answer) Thank you for this comment. We consider that we should explain more about the method, regarding with Figure 3a, for the image data acquisition. After application of FITC-labeled RNAs, we stopped the uptake using PFA fixation, and immediately started to obtain the fluorescent signals in nucleus using confocal microscopy. As shown in Methods section, we first detected a cell and decided top and bottom positions of the nucleus in the cell. Then, we obtained 15 continuously sliced images using scan mode.
These sliced images were combined for the 3D image re-construction. Thus, to make 6 cell images we accumulated 6 x 15 = 90 images. In the graph, 6 data points were plotted per group. This number may look small as cell research data, but the data acquisition process is more laborious than ordinary cell research. The reason why we took this complicated method to obtain the RNA uptake data that labeled RNA emits only weak signals. In this experiment, to minimize the effect of FITC-labeling to the structure of RNA, we reduced the amount of labeled nucleotides in the RNA syntheses as low as possible. Therefore, fluorescence signals were so small that slow scan mode was absolutely required to reduce the noise level, it took about 10 min to obtain one cell data.
Thus, it took a total of 5 hrs to obtain 5-6 cells data per sample from 5 samples (non-labeled IL1β, βactin RNA, gapdh, RNA, IL1β RNA, and IL1β-stop RNA). To make accurate comparisons among the samples, we had to complete the date acquisition at least within two days. Therefore, cell numbers in each condition in Figure 3a were limited. To confirm reproducibility of the data, we added repeated data for figure 3a, in supplementary Figure 9. We would like to note that IL-1β RNA uptake was observed in Figure 3a data and repeated experiment (Fig S9). This conclusion is further supported by Figure 5a data, in which IL1β RNA uptake are also demonstrated. We also added the sentences in Methods section and figure 5 legend shown below.

Uptake of RNA
… For accurate comparisons among samples, image data were acquired within 2 days.
Image data was analyzed by Leica Application Suite X (LAS X v3, Leica). All cells were checked with a single image of the central portion of the nucleus to confirm the nuclear RNA uptake.
(page 23 in the revised text) Repeated experiment data are shown in Fig. S9. (page 41 in the legend in revised text) Comment 6-Interestingly the IL1β-stop-mRNA clearly transported to the nucleus (Fig   3a).
Why authors didn't include an illustrative image of this construct, which I think is critical for the interpretation of the results.
Answer) Thank you for this comment. We added the illustrative image in the top of figure 3a.
Comment 7-To determine the effect of the ZC3H12D protein, TCM was divided into two parts, and these samples were passed through anti-FLAG-beads, with or without the ZC3H12D-FLAG tag protein. The nex-RNA was isolated from each sample independently. The first nex-RNA was expected to have less IL1β-mRNA than the second because it was captured in the ZC3H12D-FLAG-beads.
This isn't clear.
Answer) We changed the sentences as follows: To confirm the effect of ZC3H12D protein, migration assays were repeated using TCM passed through ZC3H12D-FLAG tag bound on anti-FLAG-beads. In this experiment, TCM passed through ZC3H12D protein-bound beads [designated as ZC3H12D(+) column] and passed through anti-FLAG beads [designated as ZC3H12D(-) column, used as a control] were prepared. nex-RNA were isolated from both TCMs and used for the migration assay.