A functional SNP rs895819 on pre-miR-27a is associated with bipolar disorder by targeting NCAM1

The aberrant expression or genomic mutations of microRNA are associated with several human diseases. This study analyzes the relationship between genetic variations of miRNA and schizophrenia or bipolar disorder. We performed case-control studies for ten SNPs in a total sample of 1584 subjects. All these ten SNPs were on or near mature microRNAs. We identified the association between bipolar disorder and the T/C polymorphism at rs895819. To illustrate the function of miR-27a, we constructed several miR-27a knockout (KO) cell lines, determined candidates of miR-27a, and then verified NCAM1 as a target gene of miR-27a. Further studies revealed that the T/C polymorphism on miR-27a led to the differential expression of mature and precursor miR-27a without affecting the expression of primary miR-27a. Furthermore, the C mutation on pre-miR-27a suppresses cell migration and dopamine expression levels. Our study highlights the importance of miR-27a and its polymorphism at rs895819 in bipolar disorder.

This is a straightforward and largely well conducted study. The experimental set up is sound and the data are presented clearly. miRNAs have gained increasing interest in many diseases, including neuropsychiatric disorders but much of their functions are still not known. In the literature, there has been an increasing number of studies on miRNA/target relationships both on the descriptive and on the functional level. The strength of this investigation lies in the association of genetic variations identified in a larger cohort of subjects with functional analysis of a potential BDassociated mechanism, at least on the cellular level. Thus, this study adds to a growing literature on genetic predisposition in BD by identifying a novel T/C polymorphism of miR-27a and suggesting a hypothetical mechanism in disease pathology. The findings are somewhat consistent with previous data suggesting an association of increased ICAM1 levels with BD and the here described mechanism adds to further understanding the complexity of BD pathology.

Major:
To make a tighter case for miRNA function on ICAM1 regulation, in Fig. 5C, data on miR-27a WT and mutant should be shown in addition to mimics and inhibitors. Also, protein levels in Fig. 5C should be added and all experiments in Fig. 5 should be done in both U-251MG cells and NPCs.
The Discussion could be a bit sharper with regard to a role of ICAM1 in BD. For example, elevated ICAM1 levels are not specific to BD and have been found in other neuropsychiatric disorders and in aging as well. Also, there could be a bit more detailed discussion on the mechanisms of ICAM1 in disease pathogenesis. This can be at the expense of redundant text with the Result section.
Minor: Lines 50-74: Please cite more original article(s) on miRNA biogenesis. Line 109: To follow the flow of the text, Supplementary Table 2 should be Supplementary Table 1 and vice versa in the Methods section. Fig. 1B: It should read "Circulatory system development" Fig. 1D: It should read "Protein homodimerization activity" Lines 331-351: There is quite a bit of redundant text with the Result section which should be avoided.
In my view, Fig. 6 is a mood point as it doesn't add much additional and clarifying information to this manuscript.
The English language is largely OK but needs corrections of some minor mistakes.
Reviewer #2 (Remarks to the Author): The manuscript by Yang and colleagues has analyzed SNPs in microRNA genes in individuals with bipolar disorder, schizophrenia, and unrelated healthy people (~500 subjects each). The topic of the study is exciting and innovative as most studies so far have assessed microRNA expression levels in brain disorders, whereas much less have analyzed mutations in microRNA genes. They make the interesting discovery that a SNP in the gene coding for miR-27 is associated with bipolar disorder but not schizophrenia, and that this SNP leads to reduced expression of the mature microRNA. Using RNAseq and qPCR in miR-27 KO astrocytoma cell lines and mimics and inhibitors in neuronal precursor cells they show differentially expressed mRNAs and identify ICAM-1 as a putative target that is regulated by miR-27. This is an interesting and novel finding, but the study could be strengthened in its rigor and significance. The major weakness is that it does not provide clear evidence for how the SNP effects neuronal function, which would be crucial to confirm its relevance for bipolar disorders. Below are my specific comments.
Major comments: 1) The authors do not provide evidence for a causal relationship of the identified SNP and bipolar disorder, as neuronal phenotypes were not assessed. The paper would be strengthened considerably if the authors would provide some evidence for a neuronal dysfunction caused by the SNP.
2) I suggest combining table 2 and supplemental table 2 into one table. I think the association with bipolar disorder but not schizophrenia is particularly interesting.
3) The authors state that they identified two astrocytoma KO clones. The rigor of the study would be increased considerably if both were analyzed, at least using qRT-PCR to confirm RNASeq results. 4) The authors base their predicted target genes on mRNA expression. This approach does not consider the function of microRNAs as translational repressors. It is thus likely that many of the changes in gene expression that the scientists identified are secondary effects -in line of this assumption, more mRNAs were downregulated than upregulated in the KO cell lines. The authors should discuss this carefully. 5) Related to point 4, I think it would be helpful to analyze how many of the DEG are predicted targets of miR-27. Also, it would be helpful to show more direct evidence that ICAM is a target of miR-27, e.g. via luciferase assays using ICAM reporters along with ICAM reporter constructs with mutations in the seed sequence. If this data cannot be provided, I suggest toning down the statement that they identified ICAM as a target. 6) The authors find that the SNP, when expressed from plasmids in NPCs, significantly reduces microRNA expression. So, in my opinion the main finding is that the SNP reduces the levels of mature microRNA (which is interesting per se), and the reduced effects on target genes is then just secondary. It is thus not surprising that the SNP has a reduced effect on potential targets of miR-27. I thus suggest that this data (Fig. 4) should be shown before Fig. 3. 7) The quality of the western blots shown in 5A is low. Can the authors provide a different western blot and/or additional blots?
Other comments: 1) The authors state that they have not corrected for multiple testing in their studies but in figure legends and the methods they state that they did FDR. This should be clarified. 2) Statistics: Figure 3 should be analyzed as a 2-way ANOVA followed by posthoc and Fig. 4A should be a one-sample t-test to account for lack of variance in the WT. 3) In general, n=3 is low for these types of experiments. 4) The abstract seems to jump back and forth between bipolar disorders and cell lines, therefore suggest editing. 5) There are a few typos, e.g.: line 71: "interacts", line325: "to" validate Response to reviewers 1) While we agree with Referee #2 that performing RNA-seq on the second miR27a-null astrocytoma clone would improve the impact of the study, any revision should, at a minimum, validate similar changes in gene expression between clonal lines using qRT-PCR. On a related note, we believe that any investigation of how rs895819 affects neuronal physiology would be interesting, but out of scope for the current manuscript, and should be discussed as a limitation.
Response: Thanks for your suggestion. We have reconstructed a miR27a-null neuroblastoma SH-SY5Y cell line and found the similar changes in gene expression, which were mostly significantly up-regulated in both miR-27a KO cell line by qPCR (see Fig 2).
We really agree to the importance of how rs895819 affects neuronal physiology. So we designed two functional experiments to find the possible link. One is to detect neurotransmitter expression of two stable NPCs containing mir-27a-C (mutant) or mir-27a-T(ctrl) , the other is focus on the cell migration because we identified NCAM1 as the target of miR-27a and NCAM1 is reported to have an effect on cell adhesion and cell migration. Indeed, we found that mir-27a-C(mutant) affects both neurotransmitter expression and cell migration in neurons differentiated from NPCs.
In the manuscript we added one paragraph to describe this (Page 15, the first paragraph). We also added the description of these two methods on methods section (page 26).
(2) As suggested by Referee #2, examine a direct vs. indirect relationship between miR27a and ICAM, potentially using a 3' UTR luciferase assay. On a related note, we also agree with Referee #2 that the study would benefit from an analysis of which DEGs are predicted to harbor miR27a seed sequences or are potential targets.
Response: Good suggestion. According to the advice of Referee #2, we reanalyzed the DEGs which are predicted to harbor miR-27a seed sequence and acquired the top 20 gene list . (Table 2) and then did Luciferase Reporter Assay.
ICAM1 and NCAM1 were predicted as candidate targets of miR-27a by the verification of qPCR and western blotting.
To further examine the direct or indirect relationship between mir-27a and ICAM1 & NCAM1, we utilize 3' UTR luciferase reporter assay.
We found relative luciferase activity of 3'UTR-NCAM1 was decreased dramatically in cells transfected with miR-27 mimic compared with miR-27 NC group, while up-regulation of miR-27 showed almost no effect on the relative luciferase activity of NCAM1-Mutant ( Figure 5). This indicated that NCAM1 is a direct target as mir-27a.
(3) As noted by Referee #1, include WT and mutant miR27 as controls for the U-251MG experiments in Figure 5c, and quantify protein levels in all experimental conditions. If feasible, we also encourage you to repeat these experiments in NPCs, as a potentially relevant cell type for BD.

Response：
As suggested, we repeated all these experiments in NPCs .
However, the results in NPC was not as effective in U251.This might be caused by low transfection efficiency of primary cells.
Additionally, we tried many times and found that knocking out miR-27a will make NPC cells lose the ability to proliferate. On the other hand we successfully constructed the stable overexpression of miR-27a in NPCs. (page 23 ) In addition, we have quantified protein levels in all experimental conditions. The findings are somewhat consistent with previous data suggesting an association of increased ICAM1 levels with BD and the here described mechanism adds to further understanding the complexity of BD pathology.
Response: We appreciate the positive comments on our paper.

Major:
To make a tighter case for miRNA function on ICAM1 regulation, in Fig. 5C, data on miR-27a WT and mutant should be shown in addition to mimics and inhibitors. Also, protein levels in Fig. 5C should be added and all experiments in Fig. 5 should be done in both U-251MG cells and NPCs. Response: Thanks for your advice. Besides ICAM1, we also found another Response： OK，we have combined Table 2 and Supplementary   Table 2 into one table，which was listed in advance along with the associated method.  Thank you for pointing out this, and we have deleted this .
In my view, Fig. 6 is a mood point as it doesn ' t add much additional and clarifying information to this manuscript.
Response: We have replaced this figure with the functional data and outlined the possible mechanism of a functional SNP rs895819 on pre-miR-27a associated with bipolar disorder by targeting NCAM-1 in Figure 7.
The English language is largely OK but needs corrections of some minor mistakes.
Response: Thanks. We have asked for paper editing service to improve our manuscript.
Reviewer #2 (Remarks to the Author): The manuscript by Yang and colleagues has analyzed SNPs in microRNA genes in individuals with bipolar disorder, schizophrenia, and unrelated healthy people (~500 subjects each Response: Thank you for the suggestions. We have fused 3'UTR sequences (WT vs. seed region mutants) of candidate genes with luciferase as reports to verify whether these gene are direct targeted by mir-27a (see Fig 5). Interestingly, NCAM1 but not ICAM1 is a direct target of mir-27a.
6) The authors find that the SNP, when expressed from plasmids in NPCs, significantly reduces microRNA expression.
So, in my opinion the main finding is that the SNP reduces the levels of mature microRNA (which is interesting per se), and the reduced effects on target genes is then just secondary. It is thus not surprising that the SNP has a reduced effect on potential targets of miR-27. I thus suggest that this data (Fig. 4) should be shown before Fig. 3.
Response: We agree to this. We have shown the "SNP reduces the levels of mature microRNA " data in advance. What is more, to investigate the possible SNP function in the pathology of bipolar disorder, we designed the function analysis to detect neurotransmitter level instead of the reduced effect of potential targets of miR27. We think it is redundant information because we have identified NCAM1 as direct target as miR-27a , and NCAM1 was reported to have an important role in psychiatry diseases including Bipolar disorder.
7) The quality of the western blots shown in 5A is low. Can the authors provide a different western blot and/or additional blots?
Response: To improve our data quality, we utilize JESS (Protein Simple Company) to do most of our protein quantification.
Other comments: 1) The authors state that they have not corrected for multiple testing in their studies but in figure legends and the methods they state that they did FDR. This should be clarified.
Response: Here "Multiple testing" specially denotes Bonferroni test in the population research. To avoid misunderstanding, we added "perform in the population " 2) Statistics: Figure 3 should be analyzed as a 2-way ANOVA followed by posthoc and Fig. 4A should be a one-sample t-test to account for lack of variance in the WT.
Response: Thanks. We have reanalyzed the data according to your suggestions.
3) In general, n=3 is low for these types of experiments.
Response: As you suggested, it is low for n=3. Fortunately, we repeated it several times and also used different methods to verify.
4) The abstract seems to jump back and forth between bipolar disorders and cell lines, therefore suggest editing.
Response: Thanks. We have revised the abstract.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors have sufficiently addressed the reviewers' points and added additional data to improve the quality and message of the manuscript. There are a few minor revisions with regards to my original comments: -Please review the legend to revised Figure 4. It should be clear which data are mRNA and which are protein.
-The observation that k.o. of miR-27a in NPCs negatively affects their proliferation is interesting and may warrant a bit more attention. With regards to cell function, there seems to be a difference between knocking out miR-27a and reducing its maturation via the C polymorphism. Do the miR-27a mutant NPCs also have reduced cell proliferation? In my view, discussing the mechanistic difference between gene k.o. versus polymorphism-mediated regulation of miRNA maturation is an important point, particularly in context of cell type-specific functions, e.g., cell proliferation in NPC, as an important factor in brain development. I suggest to adding a few sentences to address this issue in more detail.
Reviewer #2 (Remarks to the Author): In response to the previous comments, the authors added new data, which, in theory, addresses the concerns. However, many details are missing from results and legend sections, the description of the results is scarce and even misleading in some cases, and for some, the number of repeats and statistical tests are not indicated. While western blots are quantified now, no bar diagrams with error bars to illustrate variability, nor statistics are provided. Some of the data are inconsistent (effects of the three sgRNAs in 4C,D) but not discussed, and there seems to be a mix up both in legend and quantification.
1) The new data in Figure 4 are interesting but somewhat confusing and need clarification and a more accurate description. a. It seems as if the mimic and/or inhibitor controls have an effect themselves, in particular in the ICAM1 experiments. In both cell lines, ICAM levels in cells treated with the negative control inhibitor are strongly reduced compared to in those cells treated with the mimic negative control. This strongly confounds the experiment and should be explained. b. The left two blots in B and the blots in D look as if they were generated using a Jess system whereas the two right blots in B look like traditional western blots. This should be specified. More importantly, quantifications should be shown. c. The legend is incorrect (B shows western blots but is described as qRT data) and lacks a lot of detail. d. The results section states that three different KO cells generated with three different sgRNAs increased NCAM1 expression; however, only 1 of 3 increased the mRNA, and ICAM1 was unchanged, that needs to be better described and discussed. D only showed one example and the quantification is confusing (see point f). e. The authors list values below their western blots -it is unclear if these are the averages of the three replicates (if there were three replicates, that is also unclear from the legend) or just the quantification of the single experiment. The authors should show bar diagrams with individuals dots for each replication throughout the figure.
f. The quantification in D does not match what is seen in the blot (e.g., value 0 for ICAM1 although there is clearly a band). They might have mixed up NCAM1 with ICAM1.
2) For figure 5, the authors should better describe which nucleotides were mutated in their "mutant control".
3) In the description of figure 6, the authors should better explain that they analyzed the cell culture media. In addition, the transwell assay should be better described so that the reader can assess what it shows. The legend does not indicate repeats, number of cultures or statistics. 4) Figure 7 (model): it should make clear that the last sentences about the role of NCAM1 in dopamine release and cell migration is not shown in this study and rather speculative.
5) The statistics section should be more specific about what type of tests were used for which data, and all legends should indicate the specific test as well as type of posthoc. When 2-way ANOVA was done, the statistics should be reported. 6) Sequences of antagomirs, mimics and sg RNAs (in addition to the mutation in NCAM1 as indicated above) should be listed.

Referee expertise:
Referee #1: miRNAs in neuropsychiatric disease and neuroinflammation Referee #2: miRNAs in neurological disease Reviewers' comments: Reviewer #1 (Remarks to the Author): The authors have sufficiently addressed the reviewers' points and added additional data to improve the quality and message of the manuscript.
Response：Thanks for your positive comments for our revised manuscript.
There are a few minor revisions with regards to my original comments: Please review the legend to revised Figure 4. It should be clear which data are mRNA and which are protein.
Response ： Thanks for your correction. We have corrected this in Figure 4 legend.
In our former revision, we used two quantification methods of protein expression including traditional western blotting (semi-quantitative) and Jess (automatic protein expression quantitative analysis system, see SUPPLEMENTARY method. Taking into account good repeat-ability, more accurate quantitative data, and data consistency, we have repeated all the experiments of protein quantification at least 3 times using Jess system.
The observation that k.o. of miR-27a in NPCs negatively affects their proliferation is interesting and may warrant a bit more attention. With regards to cell function, there seems to be a difference between knocking out miR-27a and reducing its maturation via the C polymorphism.
Response: Yes. It's really interesting that KO of miR-27a in NPCs affects their proliferation. In our study, the rs895919 mutation on pre-miR-27a affects the maturation of miR-27a and inhibits the miR-27 expression in NPC. To discriminate the difference between CRISPR/Cas9-mediated knocking out miR-27a and reducing its maturation via the C polymorphism, we investigated the WT, knockout, C allele, T allele of miR-27a, and found that miR-27 KO inhibits the cell growth in SH-SY5Y at 72 hours using CCK8 method. No Proliferation of several cell lines (WT, C allele, T allele and miR-27a KO in SH-SY5Y) was examined using the CCK8 assay at 24, 48 and 72 hours, respectively.
Do the miR-27a mutant NPCs also have reduced cell proliferation? Response: In order to answer this question, we used CCK8 method to detect the OD450 values of NPC-C allele / NPC-T allele at 48 hours and we found no significance between them. In summary, the rs895919 mutation on pre-miR-27a affects the maturation of miR-27a but has no effect on the cell growth rate in NPCs. (in supplementary Fig 4) Supplementary Figure 4. The C mutant has no effect on NPC proliferation.
Proliferation of NPC Ctrl (T allele) and NPC Mutant (C allele) was examined using the CCK8 assay at 48 hours. Unpaired t-test was performed between two groups with five replicates(n=5) for each group. Results were expressed as the mean ± SEM. P<0.05 as a sign of significance; n.s., not significant.
In my view, discussing the mechanistic difference between gene k.o. versus polymorphism-mediated regulation of miRNA maturation is an important point, particularly in context of cell type-specific functions, e.g., cell proliferation in NPC, as an important factor in brain development. I suggest to adding a few sentences to address this issue in more detail.
Response: Thanks for your good suggestion. "miR-27a knockout cell line and the miR-27a point mutation cell line are different in the cell physiological character and it functional mechanism. In our study, we found that nucleotide polymorphisms (SNPs) or mutations occurring in the miRNA gene region may affect the property of miRNAs through altering miRNA expression and/or maturation (see Fig. 3). On the other hand, miR-27a KO in NPCs affects their proliferation (see FigS4) . Meanwhile, the cell growth is inhibited in miR-27a knockout of SH-SY5Y. CRISPR/Cas9-mediated knockout of all miR-27/24 in ESCs leads to serious deficiency in ESC differentiation in vitro and in vivo [1].
Therefore, we think that miR-27a also plays an important role in the proliferation and differentiation of NPC, and its function can be further studied in future experiments." We added this above paragraph in the discussion section (Line 285-296).
Reviewer #2 (Remarks to the Author): In response to the previous comments, the authors added new data, which, in theory, addresses the concerns. However, many details are missing from results and legend sections, the description of the results is scarce and even misleading in some cases, and for some, the number of repeats and statistical tests are not indicated. While western blots are quantified now, no bar diagrams with error bars to illustrate variability, nor statistics are provided.
Some of the data are inconsistent (effects of the three sgRNAs in 4C,D) but not discussed, and there seems to be a mix up both in legend and quantification.
Response: Thanks a lot for your efforts to improve the quality of our data. We have provided more details for the data in the figure  Firstly, we replaced Lipofectamine 3000 with RNAimax (Invitrogen, which is more effective for RNA transfection and less toxic than lipofectamine 3000).
We replaced another Negative control for RNA sequence and reduced its dosage. Especially in NPC, we used 1 pmol or 10 pmol mimic or inhibitor RNA in total. Then we evaluated the effect of miR-27a mimics or inhibitor by qPCR method. The relative expression of mature miR-27a was dramatically increased by the miR-27a mimic and decreased by the miR-27a inhibitor (p<0.0001) in U251 and NPC ( Figure 4A)，which showed the effectiveness of miR-27a mimics and inhibitors . In addition, the negative control changes little compared to wild types after optimization.
b. The left two blots in B and the blots in D look as if they were generated using a Jess system whereas the two right blots in B look like traditional western blots. This should be specified. More importantly, quantifications should be shown.
Response：We agree with you. Taking into account Jess's good repeat-ability, more accurate quantitative data, and data consistency, we have repeated this part of the experiments using Jess system with at least three replicates.
c. The legend is incorrect (B shows western blots but is described as qRT data) and lacks a lot of detail.
Response: We are very sorry for our incorrect description of legend to figure   4B. In addition, we have added experimental information in detail according to reviewer's suggestion.
d. The results section states that three different KO cells generated with three different sgRNAs increased NCAM1 expression; however, only 1 of 3 increased the mRNA, and ICAM1 was unchanged, that needs to be better described and discussed. D only showed one example and the quantification is confusing (see point f).
Response：To avoid the off-target of CRISPR, three KO cells were generated with three different sgRNAs. These three sgRNAs effectively inhibited the mature of miR-27a (Fig 4D). In addition, we found that in those KO cell lines the protein levels of both NCAM1 and ICAM1 are increased (Fig. 4F), but are not necessarily consistent with their mRNA levels ( Figure 4E). Response: According to reviewer's suggestion, we have showed bar diagrams for the replications throughout the figures.
f. The quantification in D does not match what is seen in the blot (e.g., value 0 for ICAM1 although there is clearly a band). They might have mixed up NCAM1 with ICAM1.
Response: We are very sorry for negligence of this. We have corrected it and provided more details in it.
2) For figure 5, the authors should better describe which nucleotides were mutated in their "mutant control".
Response: Thanks for your valuable suggestion. We have provided the mutated information in Fig 5A. 3) In the description of figure 6, the authors should better explain that they analyzed the cell culture media. In addition, the transwell assay should be better described so that the reader can assess what it shows. The legend does not indicate repeats, number of cultures or statistics.
Response: To give a better understanding of Figure 6, we described the following paragraph.  ). This information should be added.
Response : As you suggested, we have reported that p value is reported to one of the main effects for two-way ANOVA statistics .
(Line 554-555, Page 22) Figure 4: The new quantification in G should indicate the statistical test used. It would also be helpful to indicate that G is the quantification of NCAM1 and ICAM1 protein levels in U251 cells. D should indicate in the y axis label that miR-27 was quantified.
Response: 1.We have added the the statistical test used in Figure 4 G. (Line 570-572,Page 23) 2.We also indicated that G is the quantification of NCAM1 and ICAM1 protein levels in U251 cells.(Line 570,Page 23).
3. Figure 4D indicated in the y axis label that miR-27 was quantified.
Line 506: This should read "the cell culture supernatant" or "…cell culture media" Response: Thanks for your correction. We have corrected it as "the cell culture media" in the new revision. (Line 441, Page 16)