Information

How to find miRNA binding sites on a specific gene?

How to find miRNA binding sites on a specific gene?


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

I am trying to find miRNAs that bind to the 3'UTR of a specific gene. What is the best way of doing that (that is, with a good scoring analysis that is most commonly used by researchers in this area)? I would also like to know the other possible methods if there are multiple ways of doing this.


There are some tools for predicting the binding:

  1. TargetScan (based on seed match [primary], extra pairing, sequence context 1 - nucleotide composition around the site etc [secondary])
  2. miRanda (based on hybridization stability and seed match[primary] and sequence context [secondary])
  3. PicTar (adds a layer of evolutionary conservation criteria)

1 Context means the position of the target site in the 3'UTR and the surrounding nucleotide composition (which is also considered an indirect metric of secondary structure)

miRanda and TargetScan also have classes called conserved and non-conserved targets. miRanda reports miRSVR scores and TargetScan reports context scores but these measures are based on few experimental results and may not always be meaningful. miRSVR is based on change in target expression upon miRNA overexpression/knockdown. It also uses metrics for target site context. Context+ score of TargetScan includes many metrics which include target abundance and conservation along with the regular context scoring. Some of these metrics may be more useful than the others. But scores based just on miRNA OE/KD experiments can be misleading- especially if target expression is quantified just at the RNA level. Target abundance also varies between different cell types. For more details on these metrics refer to the papers corresponding to these tools.

There are some experimental procedures for target determination which mainly involve Protein-RNA crosslinking followed by immunoprecipitation of Ago and subsequent quantification by RNA sequencing. See HITS-CLIP and PAR-CLIP. These techniques do not really find the targets. All they do is to correlate the levels of miRNA and their predicted targets attached to Ago (you won't exactly know if the ternary mRNA-miRNA-Ago complex really formed or not).

CLASH is a more recent technique which tries to address this issue by ligating mRNA with miRNA. This way you can capture the miRNA-mRNA interaction. However, I am a little skeptical of CLASH; I myself was working on this principle some time back and faced this one challenge. CLASH roughly involves the following steps:

  • Crosslink protein-RNA
  • Immunoprecipitate Ago
  • Use RNAse to digest unbound regions of mRNA (and make it small enough for a short read sequencing experiment)
  • Ligate miRNA and mRNA bound to Ago using RNA ligase
  • Degrade Ago using protease
  • Sequence the miRNA-mRNA ligated pair

My doubt was how would miRNA and mRNA be ligated when the footprint of Ago is bigger than an miRNA (reported ~50-60 nt in HITS-CLIP and PAR-CLIP). When the RNA is nuclease protected then how can it be accessible to ligase. When I was thinking about it then I thought that a partial protein degradation was necessary (after RNAse step and before ligase step) to give some space for ligase to act. Eventually I did not work on it further. Three years later CLASH paper was published and I was happy that someone made it work. But the ligase issue was not addressed (It seems to have worked but I don't know how!!).

To test the predictions you can use a reporter assay (such as GFP or Luciferase) with the 3'UTR cloned downstream of the reporter. These can have artifacts too. Overexpressing the miRNA should be avoided and seed mutations should be done to ascertain the targetability.

Another technique to determine a target mRNA is to mask its predicted miRNA binding site and see the effect on the expression. This has been done in zebrafish using morpholinos complementary to target sites but not on other models AFAIK. This seems to me as an elegant assay- no overexpression and you can precisely determine the target site.


I would thoroughly check the literature on the UTR you are interested in, a lot of this has already been done for many genomic regions since nextgen seq began.

You will want to first off use computational prediction algorithms to help guide you in getting a good Candidate list of miRNAs

The interaction between a miRNA and its target mRNAs is usually studied by co-transfection of a reporter expression vector containing the 3'-UTR region of the mRNA and an inhibitory or precursor molecule for the miRNA.

But, this does not measure the direct and physical interaction between a miRNA and a specific mRNA. To more specifically measure binding you will need to do a ElectroMobility Shift Assay.

Here's a commercial tool which will work well if your just trying to identify it.

http://www.activemotif.com/catalog/905/lightswitch-synthetic-mirna-target-reporter-collection

I don't think there are any other very effective/easy methods.

I guess you could do something like hybridize it, run the hyb on a gel, pop out the band and seq it


How to test if a given microRNA regulates a gene of interest - microrna target (Sep/30/2009 )

Hi,
I am a newbie to this field so please forgive this simple question.
I have a gene of interest, X, which is predicted to have many
conserved target sites for microRNAs.
How do I go about finding out
1) which of the many microRNAs that target this gene are actually real and functional?
One way I can think of is to look at gene expression data and see
if the microRNA and the target show anticorrelation across many cell types.
Turns out that there are many microRNAs that are anticorrelated with this gene
(could be because I don't have that many expression data to rule out
spurious or chance anticorrelation from bonafide ones).
Another way is for me to clone the 3'UTR of the gene and also the
mutant 3'UTR with the 6bp pair deleted in a luciferase vector and
see if in a transfection system they are downregulated upon addision of
the microRNA.
But if many microRNAs are targetting this gene, I would have to do this
one by one. Is there an alternative?

What is the standard way of validating the action of a given microRNA on
a target or finding a list of microRNAs that act on this gene and their
relative effect?
I would be grateful if someone can also point me to relevant literature on
how to do this experimentally.

lesande on Sep 30 2009, 02ᛞ PM said:

Hi,
I am a newbie to this field so please forgive this simple question.
I have a gene of interest, X, which is predicted to have many
conserved target sites for microRNAs.
How do I go about finding out
1) which of the many microRNAs that target this gene are actually real and functional?
One way I can think of is to look at gene expression data and see
if the microRNA and the target show anticorrelation across many cell types.
Turns out that there are many microRNAs that are anticorrelated with this gene
(could be because I don't have that many expression data to rule out
spurious or chance anticorrelation from bonafide ones).
Another way is for me to clone the 3'UTR of the gene and also the
mutant 3'UTR with the 6bp pair deleted in a luciferase vector and
see if in a transfection system they are downregulated upon addision of
the microRNA.
But if many microRNAs are targetting this gene, I would have to do this
one by one. Is there an alternative?

What is the standard way of validating the action of a given microRNA on
a target or finding a list of microRNAs that act on this gene and their
relative effect?
I would be grateful if someone can also point me to relevant literature on
how to do this experimentally.

Choi WY, Giraldez AJ, Schier AF. Target Protectors Reveal Dampening and Balancing of Nodal Agonist and Antagonist by miR-430. Science. 2007 Oct 12318(5848):271-4. Epub 2007 Aug 30.

My opinion is that the luciferase assay is the best way to go initially, although later in vivo morpholino experiments would be nice after you've validated repression by luciferase. I think that it will be difficult to show that the morpholino is really specific for your mRNA. How could you show that any changes in protein "A" you see is not due to an indirect effect of repression of genuine target "B" being repressed, resulting in indirect repression of your gene of interest "A" (if that makes any sense - i.e. "B" is directly (or indirectly) regulating "A" and there is no direct miRNA-regulation of "A"). The morpholino might be specific for the miRNA binding site, but how many binding sites are there in the genome? Jon - please correct me if I'm wrong!

You are right, it takes careful controls to show specificity of a Morpholino-RNA interaction. The Morpholino oligos are more specific than most other antisense types (due to low per-base affinity and consequent long complementarity requirement), but off-target binding sometimes happens enough to cause misleading results. The problem of proving specificity for an oligo protecting an miRNA target site hasn't been thoroughly worked out as far as I know. The last paragraph of this discussion specifically addresses a new technique for controlling for the specificity of target protectors. Many of these techniques require careful control of oligo dose, so the solution concentrations of the Morpholinos should be confirmed by UV spectrometry prior to dosing.

For translation blocking or splice modifying Morpholinos, a common method of controlling for specificity is to use two distinct Morpholinos with non-overlapping target sites. If the oligos are used in separate experiments and produce the same change, this supports the hypothesis that the observed change is due to interactions with the targeted RNA and not due to unintended Morpholino-RNA interactions (since a random RNA binding to one oligo would be unlikely to have sufficient complementarity with the second oligo to cause the same change with each). Note that the two oligos will usually have somewhat different efficacies, and the minimum dose that causes the change will need to be independently determined. A second specificity test using the same pair of oligos involves coinjecting the pair of oligos and looking for dose synergy -- If a phenotype elicited by a single Morpholino can be reproduced by a coinjection where the sum of the concentrations of the two coinjected oligos is considerably less than the concentration of the original single oligo, this supports the hypothesis that the phenotype was specific.

Back to blocking miRNA activity with Morpholinos. First, when targeting the miRNA itself, one Morpholino is used which targets the guide strand (and usually overlaps the guide-Dicer site) and another oligo is used targeting the star strand (usually star-Dicer targeted). While the star-targeted oligo cannot bind to the miRNA on RISC (assuming the star strand doesn't load into RISC), the star-Dicer oligo can inhibit maturation of the miRNA (by protecting the Dicer site from cleavage). This means that the two oligos can be used as a specificity control pair while the star oligo might produce a slightly weaker phenotype, the ability to mostly phenocopy the guide oligo lends support to the specificity hypothesis. However, this is complicated by the close sequence similarity within miRNA families Morpholinos can generally still bind well enough to have some biological effect even if there is a mispair or two between the oligo and its target. To target one member of a family, I try to design oligos overlapping more loop sequence, since within a miRNA family the loop region is generally far less conserved than the guide or star sequences. The star-strand control was first proposed by Wigard Kloosterman: Kloosterman WP, Lagendijk AK, Ketting RF, Moulton JD, Plasterk RH. Targeted Inhibition of miRNA Maturation with Morpholinos Reveals a Role for miR-375 in Pancreatic Islet Development. PLoS Biol. 2007 Jul 245(8):e203

Next, targeting miRNA target sites. There are not very many reports of target-protecting with Morpholinos in the literature, so a consensus has not been reached about the techniques for specificity control. One possibility is to use two target-protecting oligos, each covering half of the miRNA target site and extending in different directions across the flanking sequence. First one or the other of these is used and the concentrations determined that produce a weak phenotype. When they are both used together, the phenotype should strengthen considerably, showing a dose-synergy effect IF the single-oligo phenotypes are caused by blocking the intended miRNA target site. A good confirmation is to drop the concentration of the coinjected oligos and determine whether they are still producing the phenotype of interest at a summed concentration below that required for a single oligo. This is not a well-vetted technique and I've not seen it published yet, but it is an approach that parallels the standard techniques used for other classes of Morpholino targets.

On rereading our original post I see that I did not directly address the situation you posed. While the two-oligo strategy I outlined helps to show that a particular effect is due to modulation of the expression of a particular mRNA, it does not prove the pathway for that modulation. It is still possible that the oligo pair is modulating expression of a different mRNA which encodes a protein (r.g. a transcription factor) itself modulating the expression of the mRNA of interest. However, the probability is very low that the different mRNA bears sufficient homology with the miRNA response element and its surroundings (50 bases for two oligos) that the two oligos would pass a specificity test by modulating this different RNA.

You present an interesting problem. The luciferase assay is useful in that it shows that the oligos are interacting with the putative miRNA response element (MRE), but is that sufficient proof that the expression of the gene of interest is primarily modulated through binding its MRE and not the more circuitous pathway as you proposed? What experiment would rigorously demonstrate that there is no other pathway involved in vivo?

Your double morpholino strategy is a very good one and I agree that it would be unlikely for another 3' UTR to have similar 50 bp of sequence. What I gathered from the paper you cited was that they used a single morpholino to block the MRE in the 3' UTR. In this case it was convincing because they came at the problem from several other directions (for example the in vivo GFP reporters and Dicer knockout) and came to the same conclusion. Without the other corroborating data their conclusion wouldn't be as strong. Of course the same principle applies to the luciferase approach since it usually has drawbacks, such as being an overexpression system in many cases.

The most rigorous experiment that I can think of would be to use HITS-CLIP or other similar precipitation/sequencing technologies that are being developed. Hopefully these techniques will become the norm in the near future, given time and reduced costs.

I agree, HITS-CLIP looks like a great strategy to directly show interactions. I think it will be hard to find a more convincing technique.


Introduction

MicroRNAs (miRNAs) are small, non-coding RNA molecules of 21–25 nucleotides base length. They are involved in gene expression regulation by alignment with target gene, resulting in cleavage or repression of the target genes at post-transcriptional level [1]. They play important regulatory roles in many biological processes, including differentiation, metabolism, development and cellular signaling. Thus, identification of gene targets is important for functional characterization of miRNAs and gives new insights into biological processes that could leads to biomarkers and predictors of drug response for disease. Processes for identification and validations of microRNA targets in laboratory are mostly time consuming and expensive. These limitations have led to the development of sophisticated computational approaches of microRNA target predictions allows for narrowing down the potential targets for experimental validation.

Several computational methods to identify target genes have already been developed. Some methods rely on the conservation of binding sites (e.g. TargetScan) [2], other relies on site accessibility and thermodynamic properties to filter the seed binding sites (e.g. miRanda) [2]. Prediction algorithms use a combination of different features to increase their accuracy and compensate for the limitations of the individual features. However, there is still need of accurate with high sensitivity computational approach needed to overcome the problem generated by traditional based algorithm. Machine learning based algorithms rely on parameterization of biological data and other predicted features and are growing new era in genomics. This technique used by many prediction algorithm that generate more accurately validated miRNA-target interaction (for e.g., TarpmiR, miRGen++, MBSTAR) [3–5].

Based on prediction accuracy algorithm and the fact that most of the prediction databases were not updated for some years, we have decided to launch state of the art learning based technique with new features and transfer to miRWalk repository to an another server on a new framework to increase the accuracy and sensitvity, which allows exhaustive use of other application in this study.


Experimental Methods

Isolation and quantification of miRNAs

For profiling expression of miRNAs four pools of liver total RNA, each consisting of 10 different liver donor tissues were prepared. The origin of the human liver samples was described before (Klein et al., 2010). In brief, liver tissues were previously collected from patients undergoing liver surgery at the Department of General, Visceral, and Transplantation Surgery (A. K. Nuessler, P. Neuhaus, Campus Virchow, University Medical Center Charité, Humboldt University Berlin, Germany). The study was approved by the ethics committees of the medical faculties of the Charité, Humboldt University, and of the University of Tuebingen and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from each patient. All tissue samples had been examined by a pathologist and only histologically normal liver tissue was collected and stored at �ଌ. Total RNA was extracted using the mirVANA kit from Ambion (Austin, TX, USA). Reverse transcription of total RNA (1 μg) was performed by using stem-loop RT primer pool A v 2.0 from Applied Biosystems (Foster City, CA, USA) following the protocol provided by the manufacturer. qPCR for miRNA expression was performed with micro fluidic cards TaqMan ® Low Density Array A v 2.0 for humans (Applied Biosystems) and detected with the ABI 7900HT Fast real time RT-PCR system. Relative quantification (RQ) was calculated by normalization to the endogenous control MammU6 RQ =�xp(−㥌t).


Discussion

In this study, we examined both single and multiple targeting of miRNAs and their effects on repression. Because of the far-ranging effects of miRNA repression, it is likely that miRNAs are involved in many diseases as well. In the case of multiple targeting, we show that cancer genes tend to be targeted by more miRNAs, supporting the notion that miRNAs play a role in cancer. In the case of single targeting, we describe below a possible relationship between miRNAs and DM1, using observations about the repression of genes containing multiple pairs of overlapping binding sites. These links to diseases underline the importance of studying the mechanisms behind miRNA targeting, which we discuss in the following.

Understanding multiple targeting

The fundamental motivation for having multiple miRNAs target a gene is so that these presumably important genes can be regulated in a variety of conditions such as different tissue types or transcriptional programs. While it is known that some genes have recognition sites for multiple different miRNAs, it is uncertain whether multiple miRNAs simply supplement each other in different conditions or they act in concert to provide enhanced gene repression. In a simple model, each miRNA would be independently responsible for regulating genes that need to be repressed for a given condition (for example, a specific tissue type). For genes that need to be active under a number of specific conditions, a different miRNA could be expressed under each condition that that gene needed to be regulated. MiRNAs would, therefore, act independently of each other, so that in the case that multiple miRNAs happened to be expressed simultaneously, there would not necessarily be any enhanced repression.

A more intriguing model involves multiple miRNAs working in concert to repress a gene. In this case, two different miRNAs expressed independently could each repress a given gene. If both miRNAs are expressed simultaneously, however, then that gene is much more strongly repressed than the repression exerted by each miRNA on its own. This coordinated regulation is achieved in transcriptional regulation when two transcriptional factors interact in a transcriptional complex while binding to the promoter of a gene. Since miRNAs are much smaller than transcription factors and, therefore, have little of a binding interface, it is unlikely that miRNAs could directly interact. It is possible that miRNA complexes (as part of a RISC complex) could instead interact, but the binding interfaces to these complexes would need to exhibit some unique characteristic of that miRNA to differentiate one complex from another. Another possibility is that two binding sites responsive to different miRNAs may have different repressive potential depending on the distances between the two sites, similar to the results shown above for a single miRNA on multiple sites. Independent of the mechanism, the fine degree of regulation allowed by coordinate miRNA repression makes this an appealing model that deserves further attention. Our observation that genes targeted by more miRNAs tend to be repressed more than genes targeted by fewer miRNAs is consistent with the combinatorial model where both the degree and specificity of miRNA regulation can be modulated by various combinations of relevant miRNAs.

CTG repeat-binding miRNAs and link to myotonic dystrophy

Our observation that CTG repeat length correlates with miRNA repression led us to surmise a possible role for this phenomenon in disease, in particular DM1. If a 3' UTR were to gain CTG repeats, it would be possible to abnormally repress that transcript, affect the stoichiometry of free to bound CTG-repeat binding miRNAs, or otherwise disrupt CTG-repeat binding miRNA function. We focused on DM1 because CTG repeat expansion in DMPK has been shown to be the cause of the disease and because the detailed mechanism for DM1 pathogenesis remains unresolved.

We therefore propose that miRNA repression of CTG repeats plays a role in the mechanism of DM1. In this model, CTG repeat-binding miRNAs, such as mir-107 and mir-103, preferentially bind to the mutated DMPK transcript. This could have two miRNA-leaching effects: first, the amount of unbound miRNA that would normally be regulating other genes is reduced and could no longer repress other target genes or second, the strength of the repression due to long CTG repeats could result in the sequestration of large amounts of miRNA machinery and prevent normal miRNA repression in general. MiRNA involvement could have significant consequences on known proteins in DM1 disease progression. In the current view, the CTG repeat expansion triggers sequestration of the DMPK transcript into nuclear foci [48] along with MBNL [49], implicating MBNL as a key player in DM pathogenesis. Instead of the prevalent view that MBNL binds directly to DMPK mRNA, MBNL might instead be responsive to the complex with miRNAs binding to the DMPK 3' UTR.

This proposed relationship between MBNL and miRNAs might explain why colocalization of MBNL with RNA foci does not necessarily trigger DM1 downstream events [50] and why MBNL1 apparently binds to other types of repeats even better than CTG repeats [51] in both cases, miRNA-binding to CTG repeats might mediate this interaction. One potential complication to this theory is that while, traditionally, miRNA biogenesis assumes that mature miRNA is active only within the cytoplasm, miRNA binding to DMPK transcripts would require that miRNAs and their machinery exist within the nucleus. In fact, recent evidence has shown that miRNAs are also active within the nucleus [52], facilitated by a nuclear import mechanism [53].

Several lines of evidence support our theory that miRNAs might be involved in DM1. First, as we showed earlier, miRNA repression increased as the length of CTG repeats increased, suggesting a relationship between CTG repeat length and miRNA repression. Second, we also showed that wild-type DMPK is responsive to repression by CTG repeat-binding miRNAs. Additionally, disrupting miRNA biogenesis through the knockdown of Dicer has been shown to increase DMPK expression [54], suggesting that miRNAs regulate DMPK. Finally, the model implies that CTG repeat-binding miRNAs should be expressed in the tissues that exhibit DM1 symptoms. Using published mouse miRNA expression data [55], we found that our strongest candidates, mir-107 and mir-103, are indeed strongly expressed in brain, heart, and muscle. Together, these results support a role for miRNAs in DM1 pathogenesis, and, in particular, highlight mir-107 and mir-103 as attractive candidates for binding to DMPK.

Observations about the relative expression metric

The RE metric as used in this paper is unique compared to previous efforts in understanding miRNA targeting in that both miRNA and mRNA expression data have been used together to measure miRNA repression. Previous efforts utilizing only expression microarray or in situ hybridization data indirectly measured differential expression of mRNA targets by taking advantage of knowledge about tissue- or stage-specific expression of miRNAs. For example, since mir-1 tends to be expressed in skeletal muscle, it is expected that targets of mir-1 should be downregulated in muscle samples. However, miRNA expression characteristics can be inferred for only a limited number of miRNAs for a given sample type. In this approach, the experimentally derived expression of many different miRNAs is known for multiple samples, making it possible to calculate the differential expression of target genes using measurements of actual miRNA expression levels. Using this approach, we performed an in silico genome-wide assessment of binding site-specific characteristics of miRNA repression, including the number of binding sites, distances between binding sites, pairs of extensively overlapping sites, and length of the 3' UTR. While some of these morphological features have been previously discussed as factors linked to or contributing to miRNA repression, they had generally been studied under specific and limited experimental conditions or in the context of miRNA target predictions without regard to expression data.

Because the Lu et al. dataset [21] used here comprises a heterogeneous set of samples drawn from different tissue types and cancer status, tissue- and cancer-specific gene expression complicates analysis of miRNA repression. For this reason, housekeeping genes, which are universally expressed, served to reduce variation in gene expression across different tissues due to non-miRNA specific effects. It is also for this reason that we chose to employ the RE metric and not a correlation metric. Because of the complexity of the data, the expected anti-correlations that correspond to repression tend to be very slight and, therefore, difficult to interpret. Additionally, the RE metric corresponds more closely to the concept of degree of repression, where smaller RE values correspond to down-regulation and thus greater repression.

While larger changes have been observed in translational inhibition by miRNAs compared with transcriptional repression, the relatively small changes in RE values that we observed nevertheless emphasize the importance of miRNA-mediated transcriptional repression. As we showed, the 5-10% lower RE values for a set of gene interactions is in line with the average repression of target genes in cells transfected with miRNA in Lim et al. [15]. Importantly, both calculations are based on a large number of gene targets, and, therefore, subject to various sources of noise and uncertainty. These include the possibility that some genes might be more strongly repressed by a miRNA than others, that some gene targets might have been mis-predicted by PicTar, or that some gene targets might only be expressed or responsive to a miRNA in certain tissues. Despite these potential sources of noise, our ability to detect the observed trends shows that the results are applicable genome-wide and emphasize the role of miRNA repression at the transcriptional level. Since it appears that the same sequence features (that is, the distance between binding sites) can influence repression both at the translational level [29] and transcriptional level (shown here), this suggests that the mechanisms driving miRNA-mediated transcriptional and translational repression may be linked.

One unexpected observation was the presence of RE values greater than 1.0 in various analyses. This effect is possible if considered within the context of total gene regulation, where multiple factors compete to up- and downregulate a gene. In this scenario, transcription factors and miRNAs that are simultaneously expressed exert opposing effects on the regulation of genes. If, on balance, a gene experiences greater transcriptional activation than miRNA repression, then this gene could exhibit RE values greater than 1.0 despite the miRNA repression. This apparent upregulation in the presence of miRNA repression should not be considered surprising given the belief that miRNAs serve to fine-tune gene regulation in feedback loops, increasing the precision and robustness of gene expression [31, 56, 57].

We anticipate that the RE metric will be able to reveal additional features of miRNA repression when applied to larger datasets containing more uniform data, such as those containing the same tissues or cancer state. Some potential experiments include testing for cooperative effects of multiple miRNAs working together to repress a gene, interactions between miRNAs and transcription factors when targeting a gene, and binding site specific effects, such as the importance of the seed region or the tolerability of G/U mismatches.


Gandomani HS, Aghajani M, Mohammadian-Hafshejani A, Tarazoj AA, Pouyesh V, Salehiniya H. Colorectal cancer in the world: incidence, mortality and risk factors. Biomed Res Ther. 20174:1656–75.

Colace L, Boccia S, De Maria R, Zeuner A. Colorectal cancer: towards new challenges and concepts of preventive healthcare. Ecancermedicalscience. 201711:ed74–98.

Simonian M, Khosravi S, Mortazavi D, Bagheri H, Salehi R, Hassanzadeh A, et al. Environmental risk factors associated with sporadic colorectal cancer in Isfahan, Iran. Middle East J Cancer. 20189:318–22.

Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 201868:394–424.

Terzić J, Grivennikov S, Karin E, Karin M. Inflammation and colon cancer. Gastroenterology. 2010138:2101–14. e5

Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation. Nature. 2008454:436.

Greten FR, Grivennikov SI. Inflammation and cancer: triggers, mechanisms, and consequences. Immunity. 201951:27–41.

Wang S, Liu Z, Wang L, Zhang X. NF-κB signaling pathway, inflammation and colorectal cancer. Cell Mol Immunol. 20096:327.

Schottenfeld D, Beebe‐Dimmer J. Chronic inflammation: a common and important factor in the pathogenesis of neoplasia. CA Cancer J Clin. 200656:69–83.

Rajput S, Wilber A. Roles of inflammation in cancer initiation, progression, and metastasis. Front Biosci (Sch Ed). 20102:176–83.

Sethi G, Shanmugam MK, Ramachandran L, Kumar AP, Tergaonkar V. Multifaceted link between cancer and inflammation. Biosci Rep. 201232:1–15.

Kundu JK, Surh Y-J. Emerging avenues linking inflammation and cancer. Free Radic Biol Med. 201252:2013–37.

Chams S, Badran R, Sayegh SE, Chams N, Shams A, Hajj Hussein I. Inflammatory bowel disease: Looking beyond the tract. Int J Immunopathol Pharmacol. 201933:2058738419866567.

Hnatyszyn A, Hryhorowicz S, Kaczmarek-Ryś M, Lis E, Słomski R, Scott RJ, et al. Colorectal carcinoma in the course of inflammatory bowel diseases. Hereditary Cancer Clin Pract. 201917:18.

Voronov E, Apte RN. IL-1 in colon inflammation, colon carcinogenesis and invasiveness of colon cancer. Cancer Microenviron. 20158:187–200.

Feagins LA, Souza RF, Spechler SJ. Carcinogenesis in IBD: potential targets for the prevention of colorectal cancer. Nat Rev Gastroenterol Hepatol. 20096:297.

Grivennikov SI, Greten FR, Karin M. Immunity, inflammation, and cancer. Cell. 2010140:883–99.

Zhang K, Civan J, Mukherjee S, Patel F, Yang H. Genetic variations in colorectal cancer risk and clinical outcome. World J Gastroenterology. 201420:4167.

Rustgi AK. The genetics of hereditary colon cancer. Genes Dev. 200721:2525–38.

Liu Q, Tan Y-Q. Advances in identification of susceptibility gene defects of hereditary colorectal cancer. J Cancer. 201910:643.

Jasperson KW, Tuohy TM, Neklason DW, Burt RW. Hereditary and familial colon cancer. Gastroenterology. 2010138:2044–58.

Dominic OG, McGarrity T, Dignan M, Lengerich EJ. American College of Gastroenterology guidelines for colorectal cancer screening 2008. Am J Gastroenterol. 2009104:2626.

Wen J, Xu Q, Yuan Y. Single nucleotide polymorphisms and sporadic colorectal cancer susceptibility: a field synopsis and meta-analysis. Cancer Cell Int. 201818:155.

Xin J, Du M, Gu D, Ge Y, Li S, Chu H, et al. Combinations of single nucleotide polymorphisms identified in genome‐wide association studies determine risk for colorectal cancer. Int J Cancer. 2019 145:2661–2669.

Tomlinson IP, Webb E, Carvajal-Carmona L, Broderick P, Howarth K, Pittman AM, et al. A genome-wide association study identifies colorectal cancer susceptibility loci on chromosomes 10p14 and 8q23. 3. Nat Genet. 200840:623.

Collins FS, Brooks LD, Chakravarti AA. DNA polymorphism discovery resource for research on human genetic variation. Genome Res. 19988:1229–31.

Jin Y, Wang J, Bachtiar M, Chong SS, Lee CG. Architecture of polymorphisms in the human genome reveals functionally important and positively selected variants in immune response and drug transporter genes. Hum Genomics. 201812:43.

Sana J, Faltejskova P, Svoboda M, Slaby O. Novel classes of non-coding RNAs and cancer. J Transl Med. 201210:103.

MacFarlane L-A, R Murphy P. MicroRNA: biogenesis, function and role in cancer. Curr Genomics. 201011:537–61.

O’Brien J, Hayder H, Zayed Y, Peng C. Overview of microRNA biogenesis, mechanisms of actions, and circulation. Front Endocrinol. 20189:402.

Landi D, Gemignani F, Naccarati A, Pardini B, Vodicka P, Vodickova L, et al. Polymorphisms within micro-RNA-binding sites and risk of sporadic colorectal cancer. Carcinogenesis. 200829:579–84.

Mosallayi M, Simonian M, Khosravi S, Salehi AR, Khodadoostan M, Sebghatollahi V, et al. Polymorphism (rs16917496) at the miR-502 binding site of the lysine methyltransferase 5A (SET8) and Its correlation with colorectal cancer in iranians. Adv Biomed Res. 20176:77.

Saunders MA, Liang H, Li W-H. Human polymorphism at microRNAs and microRNA target sites. Proc Natl Acad Sci USA. 2007104:3300–5.

Chen L, Deng H, Cui H, Fang J, Zuo Z, Deng J, et al. Inflammatory responses and inflammation-associated diseases in organs. Oncotarget. 20189:7204.

Freire MO, Van Dyke TE. Natural resolution of inflammation. Periodontology 2000. 201363:149–64.

Muñoz-Carrillo JL, Cordero JFC, Gutiérrez-Coronado O, Villalobos-Gutiérrez PT, Ramos-Gracia LG, Hernández-Reyes VE. Cytokine profiling plays a crucial role in activating immune system to clear infectious pathogens. In: Immune Response Activation: IntechOpen 201825:6287.

Kany S, Vollrath JT, Relja B. Cytokines in inflammatory disease. Int J Mol Sci. 201920:6008.

Mager LF, Wasmer M-H, Rau TT, Krebs P. Cytokine-induced modulation of colorectal cancer. Front Oncol. 20166:96.

Ferreira VL, Borba HHL, Bonetti ADF, Leonart L, Pontarolo R. Cytokines and interferons: types and functions. In: Autoantibodies and Cytokines: IntechOpen 201874:550.

Shrihari T. Dual role of inflammatory mediators in cancer. Ecancermedicalscience. 201711:721.

Zamarron BF, Chen W. Dual roles of immune cells and their factors in cancer development and progression. Int J Biol Sci. 20117:651.

Setrerrahmane S, Xu H. Tumor-related interleukins: old validated targets for new anti-cancer drug development. Mol Cancer. 201716:153.

Azijli K, Weyhenmeyer B, Peters G, De Jong S, Kruyt F. Non-canonical kinase signaling by the death ligand TRAIL in cancer cells: discord in the death receptor family. Cell Death Differ. 201320:858.

Lin W-W, Karin M. A cytokine-mediated link between innate immunity, inflammation, and cancer. J Clin Investig. 2007117:1175–83.

Meira LB, Bugni JM, Green SL, Lee C-W, Pang B, Borenshtein D, et al. DNA damage induced by chronic inflammation contributes to colon carcinogenesis in mice. J Clin Investig. 2008118:2516–25.

Colotta F, Allavena P, Sica A, Garlanda C, Mantovani A. Cancer-related inflammation, the seventh hallmark of cancer: links to genetic instability. Carcinogenesis. 200930:1073–81.

Hussain SP, Hofseth LJ, Harris CC. Radical causes of cancer. Nat Rev Cancer. 20033:276.

Wang Z, Li S, Cao Y, Tian X, Zeng R, Liao D-F, et al. Oxidative stress and carbonyl lesions in ulcerative colitis and associated colorectal cancer. Oxid Med Cell Longev. 20162016:1–15.

Jeong H-J, Chung H-S, Lee B-R, Kim S-J, Yoo S-J, Hong S-H, et al. Expression of proinflammatory cytokines via HIF-1α and NF-κB activation on desferrioxamine-stimulated HMC-1 cells. Biochem Biophys Res Commun. 2003306:805–11.

Kaur P, Asea A. Radiation-induced effects and the immune system in cancer. Front Oncol. 20122:191.

Chang CL, Marra G, Chauhan DP, Ha HT, Chang DK, Ricciardiello L, et al. Oxidative stress inactivates the human DNA mismatch repair system. Am J Physiol-Cell Physiol. 2002283:C148–C54.

Fleisher AS, Esteller M, Harpaz N, Leytin A, Rashid A, Xu Y, et al. Microsatellite instability in inflammatory bowel disease-associated neoplastic lesions is associated with hypermethylation and diminished expression of the DNA mismatch repair gene, hMLH1. Cancer Res. 200060:4864–8.

Mirzaee V, Molaei M, Shalmani HM, Zali MR. Helicobacter pylori infection and expression of DNA mismatch repair proteins. World J Gastroenterol. 200814:6717.

Hussain SP, Amstad P, Raja K, Ambs S, Nagashima M, Bennett WP, et al. Increased p53 mutation load in noncancerous colon tissue from ulcerative colitis: a cancer-prone chronic inflammatory disease. Cancer Res. 200060:3333–7.

Hofseth LJ, Khan MA, Ambrose M, Nikolayeva O, Xu-Welliver M, Kartalou M, et al. The adaptive imbalance in base excision–repair enzymes generates microsatellite instability in chronic inflammation. J Clin Investig. 2003112:1887–94.

Jedinak A, Dudhgaonkar S, Kelley MR, Sliva D. Apurinic/Apyrimidinic endonuclease 1 regulates inflammatory response in macrophages. Anticancer Res. 201131:379–85.

Ramana CV, Boldogh I, Izumi T, Mitra S. Activation of apurinic/apyrimidinic endonuclease in human cells by reactive oxygen species and its correlation with their adaptive response to genotoxicity of free radicals. Proc Natl Acad Sci USA. 199895:5061–6.

Balkwill F, Mantovani A. Inflammation and cancer: back to Virchow? Lancet. 2001357:539–45.

Munkholm P. The incidence and prevalence of colorectal cancer in inflammatory bowel disease. Aliment Pharmacol Ther. 200318:1–5.

Hamoya T, Fujii G, Miyamoto S, Takahashi M, Totsuka Y, Wakabayashi K, et al. Effects of NSAIDs on the risk factors of colorectal cancer: a mini review. Genes Environ. 201638:6.

Hsu H-H, Lin Y-M, Shen C-Y, Shibu M, Li S-Y, Chang S-H, et al. Prostaglandin E2-induced COX-2 expressions via EP2 and EP4 signaling pathways in human LoVo colon cancer cells. Int J Mol Sci. 201718:1132.

Gupta RA, DuBois RN. Colorectal cancer prevention and treatment by inhibition of cyclooxygenase-2. Nat Rev Cancer. 20011:11.

Tahamtan A, Teymoori-Rad M, Nakstad B, Salimi V. Anti-inflammatory MicroRNAs and their potential for inflammatory diseases treatment. Front Immunol. 20189:1377.

Fernández‐Messina L, Gutiérrez‐Vázquez C, Rivas‐García E, Sánchez‐Madrid F, de la Fuente H. Immunomodulatory role of microRNAs transferred by extracellular vesicles. Biol Cell. 2015107:61–77.

Qu Z, Li W, Fu B. MicroRNAs in autoimmune diseases. BioMed Res Int. 20142014:1–11.

Sonkoly E, Ståhle M, Pivarcsi A. MicroRNAs and immunity: novel players in the regulation of normal immune function and inflammation. Semin Cancer Biol. 200818:131–40.

Tomankova T, Petrek M, Gallo J, Kriegova E. MicroRNAs: emerging regulators of immune‐mediated diseases. Scand J Immunol. 201275:129–41.

Asirvatham AJ, Magner WJ, Tomasi TB. miRNA regulation of cytokine genes. Cytokine. 200945:58–69.

Nakasa T, Miyaki S, Okubo A, Hashimoto M, Nishida K, Ochi M, et al. Expression of microRNA‐146 in rheumatoid arthritis synovial tissue. Arthritis Rheum. 200858:1284–92.

Pfeiffer D, Roßmanith E, Lang I, Falkenhagen D. miR-146a, miR-146b, and miR-155 increase expression of IL-6 and IL-8 and support HSP10 in an In vitro sepsis model. PLoS ONE. 201712:e0179850.

Cargill M, Altshuler D, Ireland J, Sklar P, Ardlie K, Patil N, et al. Characterization of single-nucleotide polymorphisms in coding regions of human genes. Nat Genet. 199922:231.

Sukhumsirichart W. Polymorphisms. In Genetic Diversity and disease susceptibility. IntechOpen. 201876:728.

Altshuler D, Donnelly P, Consortium IH. A haplotype map of the human genome. Nature. 2005437:1299.

Guo L, Du Y, Chang S, Zhang K, Wang J. rSNPBase: a database for curated regulatory SNPs. Nucleic Acids Res. 201342(D1):D1033–D9.

Steri M, Idda ML, Whalen MB, Orrù V. Genetic variants in mRNA untranslated regions. Wiley Interdiscip Rev: RNA. 20189:e1474.

Schwerk J, Savan R. Translating the untranslated region. J Immunol. 2015195:2963–71.

Skeeles LE, Fleming JL, Mahler KL, Toland AE. The impact of 3′ UTR variants on differential expression of candidate cancer susceptibility genes. PLoS ONE. 20138:e58609.

Iorio MV, Croce CM. MicroRNA dysregulation in cancer: diagnostics, monitoring and therapeutics. A comprehensive review. EMBO Mol Med. 20124:143–59.

Calin GA, Dumitru CD, Shimizu M, Bichi R, Zupo S, Noch E, et al. Frequent deletions and down-regulation of micro-RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci USA. 200299:15524–9.

Ding H-X, Lv Z, Yuan Y, Xu Q. MiRNA polymorphisms and cancer prognosis: a systematic review and meta-analysis. Front Oncol. 20188:596.

Ryan BM, Robles AI, Harris CC. Genetic variation in microRNA networks: the implications for cancer research. Nat Rev Cancer. 201010:389.

Sivanesan D, Beauchamp C, Quinou C, Lee J, Lesage S, Chemtob S, et al. IL23R (interleukin 23 receptor) variants protective against inflammatory bowel diseases (IBD) display loss of function due to impaired protein stability and intracellular trafficking. J Biol Chem. 2016291:8673–85.

Yang B, Kang H, Fung A, Zhao H, Wang T, Ma D. The role of interleukin 17 in tumour proliferation, angiogenesis, and metastasis. Mediators Inflamm. 20142014:752-12.

Zwiers A, Kraal L, van de Pouw Kraan TC, Wurdinger T, Bouma G, Kraal G. Cutting edge: a variant of the IL-23R gene associated with inflammatory bowel disease induces loss of microRNA regulation and enhanced protein production. J Immunol. 2012188:1573–7.

Zheng J, Jiang L, Zhang L, Yang L, Deng J, You Y, et al. Functional genetic variations in the IL-23 receptor gene are associated with risk of breast, lung and nasopharyngeal cancer in Chinese populations. Carcinogenesis. 201233:2409–16.

Chen J, Lu Y, Zhang H, Ding Y, Ren C, Hua Z, et al. A nonsynonymous polymorphism in IL23R gene is associated with risk of gastric cancer in a Chinese population. Mol Carcinogenesis. 201049:862–8.

Mosallaei M, Simonian M, Esmaeilzadeh E, Bagheri H, Miraghajani M, Salehi AR, et al. Single nucleotide polymorphism rs10889677 in miRNAs Let-7e and Let-7f binding site of IL23R gene is a strong colorectal cancer determinant: report and meta-analysis. Cancer Genet. 2019239:46–53.

Zhou S, Ruan Y, Yu H, Chen Y, Yao Y, Ma Y, et al. Functional IL-23R rs10889677 genetic polymorphism and risk of multiple solid tumors: a meta-analysis. PLoS ONE. 20138:e80627.

Omrane I, Baroudi O, Bougatef K, Mezlini A, Abidi A, Medimegh I, et al. Significant association between IL23R and IL17F polymorphisms and clinical features of colorectal cancer. Immunol Lett. 2014158:189–94.

Okazaki T, Wang M-H, Rawsthorne P, Sargent M, Datta LW, Shugart YY, et al. Contributions of IBD5, IL23R, ATG16L1, and NOD2 to Crohn’s disease risk in a population-based case-control study: evidence of gene–gene interactions. Inflamm Bowel Dis. 200814:1528–41.

Karimkhani S, Chaleshi V, Balaii H, Tarban P, Nourian M, Irani S, et al. Lack of association between interleukin 23R (IL-23R) rs10889677 polymorphism and inflammatory bowel disease susceptibility in an Iranian population. Rep Biochem Mol Biol. 20187:16.

Mattner F, Fischer S, Guckes S, Jin S, Kaulen H, Schmitt E, et al. The interleukin‐12 subunit p40 specifically inhibits effects of the interleukin‐12 heterodimer. Eur J Immunol. 199323:2202–8.

Monteleone G, Biancone L, Marasco R, Morrone G, Marasco O, Luzza F, et al. Interleukin 12 is expressed and actively released by Crohn’s disease intestinal lamina propria mononuclear cells. Gastroenterology. 1997112:1169–78.

Seegers D, Zwiers A, Strober W, Pena A, Bouma G. A TaqI polymorphism in the 3′ UTR of the IL-12 p40 gene correlates with increased IL-12 secretion. Genes Immun. 20023:419.

Ma X, Yan W, Zheng H, Du Q, Zhang L, Ban Y, et al. Regulation of IL-10 and IL-12 production and function in macrophages and dendritic cells. F1000Research. 20154:F1000.

Brunda MJ, Luistro L, Rumennik L, Wright RB, Dvorozniak M, Aglione A, et al. Antitumor activity of interleukin 12 in preclinical models. Cancer Chemother Pharmacol. 199638:S16–S21.

Agarwal V, Bell GW, Nam J-W, Bartel DP. Predicting effective microRNA target sites in mammalian mRNAs. Elife. 20154:e05005.

Huang H-Y, Lin Y-C-D, Li J, Huang K-Y, Shrestha S, Hong H-C, et al. miRTarBase 2020: updates to the experimentally validated microRNA–target interaction database. Nucleic Acids Res. 2019:48:D148–D154.

Sun R, Jia F, Liang Y, Li L, Bai P, Yuan F, et al. Interaction analysis of IL-12A and IL-12B polymorphisms with the risk of colorectal cancer. Tumor Biol. 201536:9295–301.

Zhou L, Yao F, Luan H, Wang Y, Dong X, Zhou W, et al. Functional polymorphisms in the interleukin-12 gene contribute to cancer risk: Evidence from a meta-analysis of 18 case–control studies. Gene. 2012510:71–7.

Chen H, Cheng S, Wang J, Cao C, Bunjhoo H, Xiong W, et al. Interleukin-12B rs3212227 polymorphism and cancer risk: a meta-analysis. Mol Biol Rep. 201239:10235–42.

Mathy N, Scheuer W, Lanzendörfer M, Honold K, Ambrosius D, Norley S, et al. Interleukin‐16 stimulates the expression and production of pro‐inflammatory cytokines by human monocytes. Immunology. 2000100:63–9.

Silva EM, Mariano VS, Pastrez PRA, Pinto MC, Castro AG, Syrjanen KJ, et al. High systemic IL-6 is associated with worse prognosis in patients with non-small cell lung cancer. PLoS ONE. 201712:e0181125.

Kovacs E. The serum levels of IL-12 and IL-16 in cancer patients. Relation to the tumour stage and previous therapy. Biomedicine Pharmacother. 200155:111–6.

Yellapa A, Bitterman P, Sharma S, Guirguis AS, Bahr JM, Basu S, et al. Interleukin 16 expression changes in association with ovarian malignant transformation. Am J Obstet Gynecol. 2014210:272. e1-. e10

Yellapa A, Bahr JM, Bitterman P, Abramowicz JS, Edassery SL, Penumatsa K, et al. Association of interleukin 16 with the development of ovarian tumor and tumor-associated neoangiogenesis in laying hen model of spontaneous ovarian cancer. Int J Gynecologic Cancer. 201222:199–207.

Azimzadeh P, Romani S, Mohebbi SR, Mahmoudi T, Vahedi M, Fatemi SR, et al. Association of polymorphisms in microRNA-binding sites and colorectal cancer in an Iranian population. Cancer Genet. 2012205:501–7.

Chen K, Song F, Calin GA, Wei Q, Hao X, Zhang W. Polymorphisms in microRNA targets: a gold mine for molecular epidemiology. Carcinogenesis. 200829:1306–11.

Stanifer ML, Pervolaraki K, Boulant S. Differential regulation of type I and type III interferon signaling. Int J Mol Sci. 201920:1445.

Lu C, Klement JD, Ibrahim ML, Xiao W, Redd PS, Nayak-Kapoor A, et al. Type I interferon suppresses tumor growth through activating the STAT3-granzyme B pathway in tumor-infiltrating cytotoxic T lymphocytes. J Immunother Cancer. 20197:157.

Zitvogel L, Galluzzi L, Kepp O, Smyth MJ, Kroemer G. Type I interferons in anticancer immunity. Nat Rev Immunol. 201515:405.

Muñoz-Fontela C, Macip S, Martínez-Sobrido L, Brown L, Ashour J, García-Sastre A, et al. Transcriptional role of p53 in interferon-mediated antiviral immunity. J Exp Med. 2008205:1929–38.

Chang L-C, Fan C-W, Tseng W-K, Chein H-P, Hsieh T-Y, Chen J-R, et al. IFNAR1 is a predictor for overall survival in colorectal cancer and its mRNA expression correlated with IRF7 but not TLR9. Medicine. 201493:349.

Araya RE, Goldszmid RS. IFNAR1 degradation: a new mechanism for tumor immune evasion? Cancer Cell. 201731:161–3.

Lu S, Pardini B, Cheng B, Naccarati A, Huhn S, Vymetalkova V, et al. Single nucleotide polymorphisms within interferon signaling pathway genes are associated with colorectal cancer susceptibility and survival. PLoS ONE. 20149:e111061.

Sobolewski C, Cerella C, Dicato M, Ghibelli L, Diederich M. The role of cyclooxygenase-2 in cell proliferation and cell death in human malignancies. Int J Cell Biol. 20102010:1920–87.

Ricciotti E, FitzGerald GA. Prostaglandins and inflammation. Arterioscler Thromb Vasc Biol. 201131:986–1000.

Greenhough A, Smartt HJ, Moore AE, Roberts HR, Williams AC, Paraskeva C, et al. The COX-2/PGE 2 pathway: key roles in the hallmarks of cancer and adaptation to the tumour microenvironment. Carcinogenesis. 200930:377–86.

Roelofs HM, Te Morsche RH, van Heumen BW, Nagengast FM, Peters WH. Over-expression of COX-2 mRNA in colorectal cancer. BMC Gastroenterol. 201414:9005.

Williams CS, Mann M, DuBois RN. The role of cyclooxygenases in inflammation, cancer, and development. Oncogene 199918:7908.

Elzagheid A, Emaetig F, Alkikhia L, Buhmeida A, SYRJÄNEN K, El-Faitori O, et al. High cyclooxygenase-2 expression is associated with advanced stages in colorectal cancer. Anticancer Res. 201333:3137–43.

Tsujii M, Kawano S, DuBois RN. Cyclooxygenase-2 expression in human colon cancer cells increases metastatic potential. Proc Natl Acad Sci USA. 199794:3336–40.

Li P, Wu H, Zhang H, Shi Y, Xu J, Ye Y, et al. Aspirin use after diagnosis but not prediagnosis improves established colorectal cancer survival: a meta-analysis. Gut. 201564:1419–25.

Andersen V, Holst R, Kopp TI, Tjønneland A, Vogel U. Interactions between diet, lifestyle and IL10, IL1B, and PTGS2/COX-2 gene polymorphisms in relation to risk of colorectal cancer in a prospective Danish case-cohort study. PLoS ONE. 20138:e78366.

Ali I, Luke B, Dean M, Greenwald P. Allellic variants in regulatory regions of cyclooxygenase-2: association with advanced colorectal adenoma. Br J Cancer. 200593:953.

Gong Z, Bostick RM, Xie D, Hurley TG, Deng Z, Dixon DA, et al. Genetic polymorphisms in the cyclooxygenase-1 and cyclooxygenase-2 genes and risk of colorectal adenoma. Int J Colorectal Dis. 200924:647–54.

Cox D, Pontes C, Guinó E, Navarro M, Osorio A, Canzian F, et al. Polymorphisms in prostaglandin synthase 2/cyclooxygenase 2 (PTGS2/COX2) and risk of colorectal cancer. Br J Cancer. 200491:339.

Iglesias D, Nejda N, Azcoita MM, Schwartz S, González-Aguilera JJ, Fernández-Peralta AM. Effect of COX2-765G> C and c. 3618A> G polymorphisms on the risk and survival of sporadic colorectal cancer. Cancer Causes Control. 200920:1421–9.

Mosallaei M, Simonian M, Ahangari F, Miraghajani M, Mortazavi D, Salehi AR, et al. Single nucleotide polymorphism rs4648298 in miRNAs hsa-miR21 and hsa-miR590 binding site of COX gene is a strong colorectal cancer determinant. J Gastrointest Oncol. 20189:448.

Kobayashi M, Nakata M, Shioi G, Miyachi H, Honjo T, Nagaoka H. In vivo analysis of Aicda gene regulation: a critical balance between upstream enhancers and intronic silencers governs appropriate expression. PLoS ONE. 20138:e61433.

Hoesel B, Schmid JA. The complexity of NF-κB signaling in inflammation and cancer. Mol Cancer. 201312:86.

He G, Karin M. NF-κB and STAT3–key players in liver inflammation and cancer. Cell Res. 201121:159.

Arlıer S, Kayışlı ÜA, Arıcı A. Tumor necrosis factor alfa and interleukin 1 alfa induced phosphorylation and degradation of inhibitory kappa B alpha are regulated by estradiol in endometrial cells. Turkish J Obstet Gynecol. 201815:50.

Liu T, Zhang L, Joo D, Sun S-C. NF-κB signaling in inflammation. Signal Transduct Target Ther. 20172:17023.

Simonian M, Mosallayi M, Miraghajani M, Feizi A, Khosravi S, Salehi AR, et al. Single nucleotide polymorphism rs696 in miR449a binding site of NFKBIA gene is correlated with risk of colorectal cancer. Gastroenterol Hepatol Bed Bench. 201811:48.

Zhang G-L, Zou Y-F, Feng X-L, Shi H-J, Du X-F, Shao M-H, et al. Association of the NFKBIA gene polymorphisms with susceptibility to autoimmune and inflammatory diseases: a meta-analysis. Inflamm Res. 201160:11–8.

Song S, Chen D, Lu J, Liao J, Luo Y, Yang Z, et al. NFκB1 and NFκBIA polymorphisms are associated with increased risk for sporadic colorectal cancer in a southern Chinese population. PLoS ONE. 20116:e21726.

Gao J, Pfeifer D, He L-J, Qiao F, Zhang Z, Arbman G, et al. Association of NFKBIA polymorphism with colorectal cancer risk and prognosis in Swedish and Chinese populations. Scand J Gastroenterol. 200742:345–50.

Klein W, Tromm A, Folwaczny C, Hagedorn M, Duerig N, Epplen JT, et al. A polymorphism of the NFKBIA gene is associated with Crohn’s disease patients lacking a predisposing allele of the CARD15 gene. Int J Colorectal Dis. 200419:153–6.

Szamosi T, Lakatos PL, Szilvasi A, Lakatos L, Kovacs A, Molnar T, et al. The 3′ UTR NFKBIA variant is associated with extensive colitis in Hungarian IBD patients. Digestive Dis Sci. 200954:351–9.

Allen IC. Non-inflammasome forming NLRs in inflammation and tumorigenesis. Front Immunol. 20145:169.

Saxena M, Yeretssian G. NOD-like receptors: master regulators of inflammation and cancer. Front Immunol. 20145:327.

Sidiq T, Yoshihama S, Downs I, Kobayashi KS. Nod2: a critical regulator of ileal microbiota and Crohn’s disease. Front Immunol. 20167:367.

Bonen DK, Cho JH. The genetics of inflammatory bowel disease. Gastroenterology. 2003124:521–36.

Negroni A, Pierdomenico M, Cucchiara S, Stronati L. NOD2 and inflammation: current insights. J Inflamm Res. 201811:49.

Mukherjee T, Hovingh ES, Foerster EG, Abdel-Nour M, Philpott DJ, Girardin SE. NOD1 and NOD2 in inflammation, immunity and disease. Arch Biochem Biophys. 2019670:69–81.

Kozomara A, Birgaoanu M, Griffiths-Jones S. miRBase: from microRNA sequences to function. Nucleic Acids Res. 201847(D1):D155–D62.

Ahangari F, Salehi R, Salehi M, Khanahmad H. A miRNA-binding site single nucleotide polymorphism in the 3′-UTR region of the NOD2 gene is associated with colorectal cancer. Med Oncol. 201431:173.

Ognjanovic S, Yamamoto J, Saltzman B, Franke A, Ognjanovic M, Yokochi L, et al. Serum CRP and IL-6, genetic variants and risk of colorectal adenoma in a multiethnic population. Cancer Causes Control. 201021:1131–8.

Slattery ML, Curtin K, Poole EM, Duggan DJ, Samowitz WS, Peters U, et al. Genetic variation in C‐reactive protein in relation to colon and rectal cancer risk and survival. Int J Cancer. 2011128:2726–34.

Omrane I, Medimegh I, Baroudi O, Ayari H, Bedhiafi W, Stambouli N, et al. Involvement of IL17A, IL17F and IL23R polymorphisms in colorectal cancer therapy. PLoS ONE. 201510:e0128911.

Poole EM, Curtin K, Hsu L, Duggan DJ, Makar KW, Xiao L, et al. Genetic variability in IL23R and risk of colorectal adenoma and colorectal cancer. Cancer Epidemiol. 201236:e104–e10.

Gong J, Shen N, Zhang H-M, Zhong R, Chen W, Miao X, et al. A genetic variant in microRNA target site of TGF-β signaling pathway increases the risk of colorectal cancer in a Chinese population. Tumor Biol. 201435:4301–6.

Dimberg J, Skarstedt M, Löfgren S, Zar N, Matussek A. Protein expression and gene polymorphism of CXCL10 in patients with colorectal cancer. Biomed Rep. 20142:340–3.

Shi M-D, Chen J-H, Sung H-T, Lee J-S, Tsai L-Y, Lin H-H. CXCL12-G801A polymorphism modulates risk of colorectal cancer in Taiwan. Arch Med Sci. 20139:999.

Dimberg J, Hugander A, Löfgren S, Wågsäter D. Polymorphism and circulating levels of the chemokine CXCL12 in colorectal cancer patients. Int J Mol Med. 200719:11–5.

Zanetti KA, Haznadar M, Welsh JA, Robles AI, Ryan BM, McClary AC, et al. 3′-UTR and functional secretor haplotypes in mannose-binding lectin 2 are associated with increased colon cancer risk in African Americans. Cancer Res. 201272:1467–77.


Discussion

For most of the trait-associated loci identified by GWAS, pinpointing the causative variant has been a major challenge. Often, the associated SNPs are located outside of the coding regions, indicating that they are involved in modulating spatiotemporal gene expression, rather than protein structure. Previous work has shown a significant overrepresentation of trait-associated SNPs in 3' UTRs, highlighting the potential significance of post-transcriptional regulation [14]. Here, we provide a systematic survey of 3' UTR variants and examine their potential influence on mRNA expression levels via miRNA binding. We used the largest and most robust cis-eQTL dataset available, identified from the peripheral blood of 5,311 individuals from seven independent cohorts [21]. To the best of our knowledge, the present study is the first to systematically investigate the impact of MRE polymorphisms on a genome-wide scale by taking into account the direction of expression changes and by incorporating the public eQTL, miRNA expression profiling, and AGO-CLIP data into the analysis.

Although it is estimated that >10% of the total length of all 3' UTRs is covered by in silico predicted miRNA target sites [18], the extent to which gene expression variation is mediated by miRNA binding alterations has not been determined. SNPs located in MREs may influence mRNA expression post-transcriptionally by disrupting or creating functional miRNA binding sites, by changing the effectiveness of MREs, or by replacing the binding site of one miRNA with that of another.

Previous studies have identified a number of SNPs with the potential to influence miRNA binding. Some of the results of these studies are summarized in online databases [18,31,33,53,54], however the sensitivity and specificity of in silico target prediction algorithms are not perfect [55], and additional support is needed to validate an miRNA target site such that it is useful for identifying functionally relevant MRE-SNPs.

Limiting putative MRE-SNPs to the subset overlapping with Argonaute binding sites [14] may add confidence to in silico predictions. However, it is noteworthy that there are significant differences between individual AGO-CLIP datasets, possibly due to heterogeneity between cell types studied. Several studies have also utilized the MRE-SNPs from experimentally supported target sites [56,57] or the co-expression of the target and corresponding miRNA [56,58,59].

So far, only a handful of studies have systematically investigated the link between eQTLs and MRE-SNPs. Landmark study by Richardson et al. utilized miRanda algorithm for MRE-SNP identification, applying miRNA-target co-expression and SNP presence in GWAS Catalog as additional filters [60]. Out of 39 filtered associations they identified 4 which showed marginally significant eQTL trend, supporting the logic of miRNA regulation.

Gamazon et al. used lymphoblastoid cell lines (LCL) from 60 European (CEU) and 60 Yoruban (YRI) individuals to analyze genetic variants, mRNA, and miRNA expression data [61]. They identified a number of putative MRE-SNPs by using in silico target prediction algorithms, assuming a negative miRNA-target correlation and eQTL effect on the target expression. They reported that 18–25% of SNPs having trans-eQTL effects on miRNA expression also have associations with mRNA levels, indicating the importance of genetic variants on miRNA regulatory networks.

In another study, the authors integrated available genomic and gene expression information from 149 HapMap 3 LCL samples to identify 2,262 putative MRE-SNPs, of which 5.74% also had cis-effects on their target genes [62].

In a more focused study, Wei et al. [63] intersected cis-eQTLs previously identified from LCL, liver, and brain with in silico predicted MREs from a limited set of 409 xenobiotic-associated genes and identified 27 putative MRE-SNPs.

In the present study, we identified 5,992 putative MRE-SNPs, which we narrowed down to a prioritized set of 217 MRE-SNPs using more stringent criteria. In either case, we did not detect any significant overrepresentation of MRE-SNPs for which the cis-eQTL direction and effect on miRNA binding would be concordant. The effect sizes did not differ between exclusively concordant and exclusively unconcordant cis-eQTLs, and, likewise, we did not observe any significant association between the average effect of a given MRE-SNP and the cis-eQTL effect size.

Ambiguity in our results can be explained by several factors. First, most of the 3' UTRs contain binding sites for several different miRNAs, and quite often, there is more than one site for a specific miRNA. This means that the effect of disrupting or creating a single binding site may be reduced by the action of other sites. Second, the effect of a 3' UTR SNP can be manifested through different mechanisms, since both miRNA binding and mRNA stability in general are affected by several different factors. These mechanisms may include alternative polyadenylation or splicing, mRNA structural alterations, or the accessibility to RNA-induced silencing complex (RISC). Some of those mechanisms were addressed in a recent study [14], suggesting that the majority of 3' UTR SNPs influence MREs rather than splicing sites or 3' UTR folding. Third, miRNAs and their targets form complex regulatory modules that contain feedback loops to minimize the effects of genetic variation and stochastic noise [64,65]. Fourth, miRNA-mediated effects on target expression are rapid in time but modest in magnitude and may therefore escape detection. In addition, the major biological role of miRNAs seems to be to dampen the stochasticity of gene expression [66]. Therefore, the loss of an miRNA site may not result in a change in expression level, but rather an increase in expression variability.

Although we were not able to detect a statistically significant enrichment of MRE-SNPs with concordant-type regulation in our analyses, we did uncover several interesting candidates where miRNA-mediated regulation may take place. The SNP with one of the most striking effects on its predicted miRNA-binding site is rs10187, which disrupts the MRE of miR-210-3p in the ISCU gene. As the site is covered by AGO-CLIP reads and targeting is experimentally well validated [44–46,67], we have reason to believe that this miRNA activity may be the underlying cause of the observed cis-eQTL effect. Interestingly, a significant cis-eQTL effect is present for only one out of two Illumina probes detecting ISCU, demonstrating possible technical variability in expression datasets. ISCU codes a protein with a functional role in the assembly of mitochondrial iron-sulfur clusters. Its decreased mRNA level has been associated with myopathy with exercise intolerance [68] and decreased cancer survival [69]. However, as the minor allele of rs10187 correlates with increased expression of ISCU, a direct link between the MRE-SNP and these conditions is unclear. miR-210-3p is a hypoxia-inducible miRNA (hypoxamir) controlled by hypoxia inducible factors (HIF, for a review see [70]) and, among other functions, down-regulates mitochondrial metabolism under hypoxic conditions by directly targeting ISCU [69].

Two potential MRE-SNPs—rs4245739 in MDM4 (creating an MRE for miR-191-5p) and rs2239680 in BIRC5 (disrupting the MRE for miR-335-5p)—have been experimentally shown to affect miRNA binding in cell lines [47,50]. Interestingly, only rs4245739 has a C-type cis-eQTL effect in peripheral blood. The lack of concordance between the allelic trend of rs2239680 and the expected effect on miRNA-mediated regulation may be explained by several factors: miR-191-5p is one of the most highly expressed miRNAs in blood (detectable in nine out of 11 blood samples) and represents

1% of the detected miRNome, whereas miR-335-5p is detectable in only six out of 11 blood samples and represents roughly 0.01% of the detected miRNome (S2 Fig). At the same time, miR-335-5p has a significantly larger in silico predicted targetome than miR-191-5p (3,046 target genes versus 568 target genes, respectively TargetScan v6.2), suggesting that the effect of miR-335-5p on any single MRE may be diluted in this tissue.

Our study also has several limitations. The primary analysis was restricted to a single tissue type: peripheral blood. Since miRNAs are known to play a role in development and often play a very specific role in a particular tissue, it is likely that similar studies focused on other tissue types will reveal additional information about the role of genetic variation in MREs. We investigated the heterogeneity of cis-eQTLs by using the independent data from GTEx project. Although the sample sizes and resulting replication power are currently much smaller than in blood meta-analysis, we demonstrated that several exclusively C-type MRE-SNPs have also nominally significant cis-eQTL effect in other tissues than blood. Unfortunately, rs10187 was not available in GTEx datasets but concordant cis-eQTL effect between rs4245739 and MDM4 was also present in several other tissues than blood, suggesting the widespread action of this miRNA-mediated mechanism in multiple tissue types (S9 Fig).

The levels of miRNA expression also vary between individuals, and it is likely that further analyses comparing SNP, gene expression, and miRNA expression data collected from the same individuals would be helpful. In addition, our analysis relies to a large extent on computationally predicted miRNA target motifs.

To overcome some of the limitations, we applied several filters to our dataset. To minimize the effect of tissue specificity of miRNA expression, we defined a blood-expressed miRNome using publicly available data. Due to the limitations of in silico target prediction algorithms, we also used the intersection of miRNA target predictions from three databases and the set of miRBase “high confidence” mature miRNAs. Although we integrated the information from experimentally validated targets into our analyses, it should be noted that an experimentally confirmed MRE does not necessarily provide solid proof of the functional consequence of an MRE-SNP on miRNA binding and its subsequent effect on mRNA and protein levels. Regardless of these limitations, our analysis identified four trait-associated concordant-type MRE-SNPs as a proof of concept, for which three variants were related to cancer.

One of the most interesting SNPs from the results of our filtered analysis is the aforementioned rs4245739 (A/C), which creates a functionally verified MRE for miR-191-5p in MDM4 (Fig 3D and 3F, Fig 4A). The MDM4-encoded protein inhibits p53 post-translationally and is upregulated in tumors [71,72], while the minor allele of rs4245739, carried by approximately 20% of the general population, is associated with a protective effect for several cancers [47,73–75] and may serve as a potential biomarker. Most importantly, the effect of rs4245739 on miR-191-5p binding and subsequent down-regulation of MDM4 mRNA and protein expression has been experimentally verified in ovarian cancer cell lines [47], serving as an example of a functional MRE-SNP identified independent of our systematic genome-wide approach.

The second gene containing both a GWAS hit and an MRE-SNP in this study is N4BP1. rs6500395, which is located in the first intron of N4BP1, has been associated with the response of rheumatoid arthritis patients to tocilizumab treatment [76], but this gene also contains an AGO-CLIP-supported C-type MRE-SNP proxy (rs1224) for miR-330-3p in its 3' UTR (Fig 4C).

In two cases, the absolute proxies of cis-eQTLs were located in the 3' UTR MRE of the nearby gene. rs3771570, which is associated with aggressive prostate cancer [74], is located in the intronic region of FARP2 gene. However, it has a perfect proxy (rs1056801) within the 3' UTR of a gene next to it, SEPT2. The minor allele of rs1056801 disrupts the binding of cancer-associated miR-17-92 family members (Table 2, [46]), and aberrant expression of SEPT2 has been reported in different tumor types [77]. As SEPT2 is also the only gene influenced by a significant cis-eQTL effect in a corresponding LD block (FDR = 0.04, Z = 3.9 [21]), we propose that MRE-SNP-mediated alterations in the binding of miR-17-92 family miRNAs may be related to abnormal expression of SEPT2 and could be a causative SNP in the GWAS locus identified by rs3771570.

In the esophageal squamous-cell carcinoma susceptibility region tagged by rs2239815 [78], we identified an MRE-SNP within the 3' UTR of CCDC117. This LD block contains two apparent candidate genes for cancer susceptibility (XBP1 and CHEK2) (Fig 4D). Although all three of these genes are affected by cis-eQTLs, the largest effect of this LD block is associated with the XBP1 gene (FDR < 0.01, Z = 23 [21]), casting doubt on the miRNA-mediated cis-eQTL mechanism.

In summary, we conducted a systematic and comprehensive survey to identify miRNA-mediated cis-eQTLs effects. Integrating data from different sources, we identified a number of potentially functional MRE-SNPs, serving as putative causative variants for allele-specific gene expression and for the development of complex traits.


Results

MiRNA knockdown by CRISPR/cas9

We aim to determine if CRISPR/cas9 targeting miRNA genomic DNA loci can robustly repress miRNA expression. As a result, we constructed CRISPR/cas9 vectors containing the individual sgRNAs with complementary sequences to miR-17, miR-200c, and miR-141 genes, respectively (Fig. 1a). Two sgRNAs were designed for each miRNA by using CRISPR DESIGN (http://crispr.mit.edu/), an online program that was developed and is maintained by Dr. Feng Zhang’s Lab at MIT. This tool can recommend sgRNA sequences for each input DNA fragment and analyze the potential off-target sites of individual sgRNA by bioinformatics blast with the whole genome DNA sequences 9 . In our study, two sgRNAs targeting the same miRNA were designed accordingly (Fig. 1b–d). Given the importance of Drosha and Dicer processing sites in process of miRNA biogenesis 14 , we triggered the PAMs (NGG) within/adjacent to these sites. The individual CRISPR/cas9 constructs were transiently transfected to HCT116 cells and the cells with CRISPR/cas9 were selected by puromycin treatment and expanded accordingly. As shown in Fig. 1e, the expression levels of mature miR-17, miR-200c and miR-141 declined up to 96% in the cells transfected with the designated CRISPR/cas9 constructs when compared to the control vectors, supporting the high efficiency of CRISPR/cas9 system in downregulation of mature miRNA expression.

(a) Schematic diagram of CRISPR/cas9 system. (b–d) Design of sgRNAs for miR-17, miR-200c, and miR-141. Two sgRNAs are designed for each miRNA by using CRISPR DESIGN (http://crispr.mit.edu/). (e) CRISPR/cas9 with designated sgRNAs can significantly inhibit the expression of miR-17, miR-200c, and miR-141, respectively (n = 3). *P < 0.05. RQ: relative quantification. (f) sgRNA -miR-17 can upregulate the expression of the validated target of miR-17, E2F1. (g,h) sgRNA -miR-200c and sgRNA -miR-141 can upregulate the expression of the validated target of miR-200c and miR-141, ZEB1, which in turn represses the expression of E-cadherin. Ctrl is the CRISPR/cas9 vectors with sgRNA sequences targeting GFP. All the blots were run under the same experimental conditions and presented by using cropped images. Please refer to the supplementary for the full blots.

The key signature of miRNA is the master role in repressing the expression of multiple genes at the post-transcription and/or translation levels. Thereby, we extended our observation to the expression of selected targets of these miRNAs. As shown in Fig. 1f, E2F1, one of the validated targets of miR-17 3 , appears to be induced in HCT116 cells transfected with CRISPR/cas9 against miR-17 knocking down miR-200c and miR-141 by CRISPR/cas9 can upregulate the expression of their co-target, ZEB1, which in turn enhances the repression of E-cadherin (Fig. 1gh). The interaction between ZEB1 and E-cadherin has been exclusively studied in EMT of human cancer cells 15 . As expected, miR-200c or miR-141 knockdown by CRISPR/cas9 in HCT116 cells can also facilitate the cell invasion (Supplementary 2 Figure S2). Therefore, these results support that CRISPR/cas9 system can be used for miRNA loss-of-function studies.

Indels generated by CRISPR/cas9 on miRNAs processing sites impede their biogenesis

As mentioned above, CRISPR/cas9 system can edit genome DNA sequences resulting in indels, which can be recognized by T7EN1 assay in an efficient manner 16 . After transfected CRISPR/cas9 vectors with individual sgRNA sequences targeting miR-17, miR-200c, and miR-141 to HCT116 cells, we detected the cleavages at target loci of these miRNAs by T7EN1 assay. As shown in Fig. 2a, the multiple site-specific bands were detected, which demonstrated the existence of indels. These results were further confirmed by DNA sequencing in which the random sizes of indels adjacent to PAM sequence were identified (Fig. 2b).

(a) DNA cleavage by CRISPR/cas9 is detected by T7EN1 assay. (b) DNA sequencing confirms the deletions (Red) and insertions (Blue) generated by CRISPR/cas9 in selected miRNA genes. PAM sequences are highlighted by green. All the gels were run under the same experimental conditions and presented by using cropped images. Please refer to the supplementary for the full-length gel images.

Pri-miRNA is the direct transcript of miRNA gene and then processed by Drosha and Dicer to form short mature miRNA. Studies demonstrated that both flanking and internal structure of pri-miRNA can decide the efficiency of Drosha processing and, in turn, the biogenesis of miRNA 17 . As shown above, our results have shown that CRISPR/cas9 cleaves miRNA Drosha and Dicer processing sites leading to random sizes of indels on genomic DNA sequences. Then, we examined the primary miRNA expression. MiR-17 is one of six members in miR-17-92 cluster, and their primary miRNA is the gene transcript containing all 6 miRNA sequences. Likewise, miR-200c and miR-141 also share a same primary miRNA transcribed from miR-200c/141 cluster. As shown in Fig. 3a, pri-miR-17-92 and pri-miR-200c/141 clusters were upregulated, which imply that cleavage of miRNA Drosha and/or Dicer processing sites by CRISPR/cas9 could interrupt biogenesis of mature miRNA resulting in accumulation of pri-miRNAs. In order to further determine the effects of these indels on mature miRNA biogenesis, we cloned the wild-type miR-17 sequences as well as mutated miR-17 sequences with selected indels into pWPXL vectors, and then transfected them into HCT116 cells for exogenous expression of miR-17 in different single clones. As shown in Fig. 3bc, mature miR-17 was significantly elevated in the cells transfected with wild-type miR-17 sequences, but not in the cells transfected with the single clones containing mutated miR-17 sequences. Moreover, not only single clones with large pieces of deletion (6–18 bp) can lead to the failure of mature miR-17 biogenesis, but also the clone with only 2 deletions and 1 insertion can impede the exogenous expression of miR-17 as well. These data strongly support that mutations generated by CRISPR/cas9 on the stem-loop structure of primary miRNA can result in the downregulation of mature miRNA by interrupting the process of biogenesis.

(a) CRISPR/cas9 induces the accumulation of primary miR-17-92 and miR-200c/141 clusters (n = 3). (b) Exogenous expression of miR-17 in HCT116 cells after transfected with wild-type miR-17 sequences and mutated sequences, respectively (n = 3). (c) DNA sequencing confirms the deletions (Red) and insertions (Blue) in the single clones with mutated miR-17 sequences. *P < 0.05. RQ: relative quantification.

Off-target effect of CRISPR/cas9 on miRNA editing

Off-target effect is the hallmark of all gene silence methodologies, including CRISPR/cas9 system. By using CRISPR DESIGN (http://crispr.mit.edu/), we predicted the off-targets of sgRNA-miR-17-1 and sgRNA-miR-200c-1 (Table 1), and examined their off-target effects accordingly. As shown in Fig. 4a, if the target sgRNA has more than 3 mismatches to the non-specific DNA sequences, there are no cleavages that can be detected by T7EN1 assay however, if the mismatches are equal to or less than two, CRISPR/cas9 system can unbiasedly cleave the off-target sequences. The T7EN1 results were also confirmed by DNA sequencing (Fig. 4b). These results not only support the high specificity of CRISPR/cas9 system, but also provide a reference to guide the design of sgRNA sequences to reduce the off-target effects.

(a) CRISPR/cas9 cannot cleave the off-target sequences with > = 3 mismatches with sgRNA-miR-17-1 and sgRNA-miR-200c-1 by T7EN1 assay. “On” means on-target sequences while “off” means off-target sequences. (b) DNA sequencing confirms the deletions (Red) and insertions (Blue) generated by CRISPR/cas9 in the off-target sequences containing < = 2 mismatches with sgRNA-miR-17-1 and sgRNA-miR-200c-1. (c) Crossing off-target effects between sgRNA-200c and sgRNA-141 as determined by qRT-PCR (n = 3). (d) T7EN1 assay confirms the DNA integrity of miR-200c subjected to sgRNA-miR-141 and that of miR-141 subjected to sgRNA-miR-200c. (e) Co-transfection of sgRNA-ZEB1 with sgRNA-miR-200c or sgRNA-miR-141 can significantly reduce crossing off-target effects on miR-141 or miR-200c, respectively (n = 3). *P < 0.05. RQ: relative quantification. All the gels were run under the same experimental conditions and presented by using cropped images. Please refer to the supplementary for the full-length gel images.

It was known that the traditional miRNA silence strategies, such as antisense inhibitors and sponge, are less robust when being used to inhibit the expression of miRNAs from the same family, given their highly conserved sequences with a few mismatches. For example, miR-200c and miR-141 are transcribed from the same genomic locus, and only have 4 mismatches in their mature sequences. Based on our results shown above, the CRISPR/cas9 targeting miR-200c should not influence the expression of miR-141 because the mismatches are more than 3. However, as shown in Fig. 4c, both sgRNAs-miR-200c generated cross expression inhibition on miR-141, vice versa. To address this unexpected result, we first examined DNA integrity of miR-200c and miR-141 genes in cells transfected with sgRNA-miR-141 and sgRNA-miR-200c by using T7EN1 assay, but only single bands were identified (Fig. 4d), which suggests that no cleavage was induced by CRISPR/cas9 and downregulation of miR-200c and miR-141 might not be caused by off-target effects. Given the regulatory loop between miR-200c/141 and ZEB1, we hypothesize that ZEB1 transcriptional regulation may be involved in these unexpected “off-target” phenotypes. It means when miR-200c is downregulated by CRISPR/cas9, ZEB1 will be inversely upregulated, which can in turn repress the expression of miR-141 at transcriptional level. Likewise, sgRNA-miR-141 can downregulate miR-200c via alternation of ZEB1 as well. To prove our hypothesis, we co-transfected CRISPR/cas9 constructs targeting ZEB1 and miR-200c or miR-141 into HCT116 cells, respectively. As shown in Fig. 4e, when ZEB1 was silent by CRSPR/cas9, the cross inhibitory phenotypes between miR-200c and miR-141 were nearly dismissed.

We also extend our observation to mature miR-200b that only has 2 different nucleotides from mature miR-200c. First, we blasted the sgRNA-miR-200c sequences to mature miR-200b sequence, and found that sgRNA-miR-200c-1 had 2 mismatches with miR-200b but sgRNA-miR-200c-2 did not align with miR-200b at all. Then, we performed the T7EN1 assay and found sgRNA-miR-200c-1 can edit the genomic DNA locus of miR-200b but not sgRNA-miR-200c-2 (Fig. 5a). These results suggest that the crossing off-target effects for the miRNAs at same family or with highly conserved sequences can be minimized by proper design of sgRNAs, which is very flexible and feasible because the design of sgRNAs can be anchored to dispersive PAMs that are easily identified in DNA sequences.

(a) MiR-200b and miR-200c have only 2 different nucleotides in their mature sequences, and T7EN1 assay shows sgRNA-miR-200c-1 (with 2 mismatches) can cleave miR-200b DNA sequences, but not sgRNA-miR-200c-2 (without mismatches). We randomly made point mutations (1–4 bp) within the sequence of sgRNA-miR-200c-1, and examined their targeting effects on miR-200c (b) and potential off-target effects on miR-200b (c) by using T7EN1 assay. There are no cleavages if mutations (MU) are more than 3 bp. The digits in the figure can be interpreted by an example: 3[1,9,20] means there are 3 mutations locating at the 1 st , 9 th , and 20 th locus of sgRNA-miR-200c-1. All the gels were run under the same experimental conditions and presented by using cropped images. Please refer to the supplementary for the full-length gel images.

In order to further support our assumption, we randomly made point mutations (1–4 bp) within the sequence of sgRNA-miR-200c-1 (Table 2), and examined their targeting effects on miR-200c and potential off-target effect on miR-200b by T7EN1 assay (Fig. 5bc). The results suggest that, if the numbers of mismatches are more than two, the mutated forms of sgRNA-miR-200c-1 show neither targeting effects on miR-200c, nor off-target effects on miR-200b, which suggest that high specificity and low off-target effect of CRISPR/cas9 in knockdown of miRNA. The T7EN1 results were also confirmed by DNA sequencing (data not shown).

Stability of miRNA knockdown phenotypes by CRISPR/cas9

It was reported that CRISPR/cas9 could result in permanent cleavage on DNA sequence. However, to our knowledge, there is no published data showing if CRISPR/cas9 can establish stable phenotypes of gene knockdown for a long term in both in vitro and in vivo models after transfection of CRISPR/cas9. Therefore, we transiently transfected CRISPR/cas9 against miR-17 and the control vectors to HT-29 and HCT116 cells, respectively, and collected cells at Day 10, 20, and 30 after transfection. As shown in Fig. 6a, miR-17 is continuously downregulated in the cells with CRISPR/cas9 targeting miR-17, even though the amount of vehicle vectors is gradually dismissed over the course up to 30 days. To confirm the findings from our in vitro studies, we injected HT-29 cells with the miR-17 knockdown phenotype by CRISPR/cas9 to nude mice subcutaneously. We collected xenograft tumors at Day 14 and Day 28 for analysis, respectively. As shown in Fig. 6b, stable downregulation of miR-17 was validated in these xenograft tissues with gradual loss of the vehicle vectors. In addition, we checked the DNA cleavage of miR-17 in the xenograft tissues and found stable phenotypes in the samples collected in Day 14 and Day 28 (Fig. 6c). These results support that CRISPR/cas9 can generate stable miRNA knockdown phenotypes in both in vitro and in vivo models.

(a) MiR-17 knockdown phenotype can be maintained as long as 30 days in both HT-29 and HCT116 cells with transit transfection of CRISPR/cas9 constructs, although the vehicle vectors are being gradually lost (n = 3). (b) Xenograft tissues generated by subcutaneously injecting HT-29 cells with CRISPR/cas9 constructs targeting miR-17 in mice show stable knockdown phenotype versus gradually lost vehicle vectors up to 28 days (n = 3). (c) T7EN1 assay confirms the permeant cleavage of CRISPR/cas9 on miR-17 DNA sequences in xenograft tissues collected at Day14 and Day 28. RQ: relative quantification. All the gels were run under the same experimental conditions and presented by using cropped images. Please refer to the supplementary for the full-length gel images.


PubMed miRWalk2.0 is an improved version of the previous database (i.e. miRWalk). miRWalk2.0 is so far the only freely accessible, comprehensive archive, supplying the biggest available collection of predicted and experimentally verified miRNA-target interactions with various novel and unique features (missing in a previous version i.e. miRWalk) to greatly assist the miRNA research community.

miRWalk2.0 not only documents miRNA binding sites within the complete sequence of a gene, but also combines this information with a comparison of binding sites resulting from 12 existing miRNA-target prediction programs (DIANA-microTv4.0, DIANA-microT-CDS, miRanda-rel2010, mirBridge, miRDB4.0, miRmap, miRNAMap, doRiNA i.e.,PicTar2, PITA, RNA22v2, RNAhybrid2.1 and Targetscan6.2) to build novel comparative platforms of binding sites for the promoter (4 prediction datasets), cds (5 prediction datasets), 5’- (5 prediction datasets) and 3’-UTR (13 prediction datasets) regions. It also documents experimentally verified miRNA-target interaction information collected via an automated text-mining search and data from existing resources (miRTarBase, PhenomiR, miR2Disease and HMDD) offer such information.

miRWalk2.0 novelties are as follows:

The web-interface of miRWalk2.0 is broadly classified into the Predicted Target (PTM) and the Validated Target (VTM) modules. These two modules are further categorized into different search pages, allowing users to fetch miRNA associated information using different identifiers.

The Predicted Target module hosts miRNA-target interactions information within the complete sequence of all known genes of human, mouse and rat including all transcripts and mitochondrial genomes. It also provides novel comparative platforms of miRNA binding sites resulting from 13 data-sets for promoter, cds, 5'-, and 3'-UTR, mitochondrial genes and miRNA-miRNA interactions. miRBase release 20 is utilized to generated miRNA-target interaction information for this module.

The Validated Target module hosts experimentally verified miRNA interaction information associated with genes, pathways, organs, diseases, cell lines, OMIM disorders and literature on miRNAs. This module is last updated on 29th September 2014. In addition, it provides the information on proteins known to be involved in miRNA processing.


TRFs and the Argonautes: gene silencing from antiquity

In Greek mythology, the Argonauts are a band of heroes who accompany Jason on his quest to find the Golden Fleece, a garment whose origins likely lie in the use of sheep fleeces as sieves to collect gold flakes from running water.

In a new paper published in BMC Biology, Anindya Dutta and colleagues mine Argonaute (sic) datasets for biology's very own hidden gold: previously neglected fragments of tRNA molecules, known as tRFs.

Here are seven awesome things you need to know about tRFs:

1) tRNA molecules are routinely degraded by the cell into tRNA halves and smaller fragments (tRFs), which can be created from both the 5' and 3' ends of each tRNA. Some studies have argued that these degradation products are not merely waste, but instead have their own biological functions.

2) Argonaute proteins bind microRNA (miRNA) molecules as part of a process of gene silencing. These interactions are studied using techniques such as CLIP and CLASH.

Anindya Dutta and colleagues have noticed that hidden in CLIP and CLASH datasets is plenty of evidence that tRFs also bind to Argonautes – a fact that has often been overlooked owing to a focus on miRNAs.

3) miRNAs and tRFs share many similarities. Both are small RNAs. As with miRNAs, tRFs (when bound to Argonautes) seem to downregulate, or 'silence', mRNAs containing complementary sequences.

They form similar complexes with Argonautes and target mRNAs, even using the same 'seed' rules for selecting target binding sites. And they are present in the cell at a similar abundance to one another.

Given that the similarities between miRNAs and tRFs are so striking, it is perhaps unsurprising that in at least one case a miRNA reported in the literature turned out to be a tRF.

4) But tRFs are not the same as miRNAs. As an obvious difference, tRFs are created from pol III-transcribed tRNAs, whereas miRNAs are the products of pol II-transcribed pre-miRNA molecules. miRNA production requires the DROSHA and DICER1 enzymes, but tRF generation is DROSHA- and DICER1-independent.

5) Another difference between tRFs and miRNAs is the identity of the Argonaute proteins to which they bind. For the most part, miRNAs form complexes with Argonaute 2 but not Argonautes 1, 3 or 4. For tRFs, Argonaute preference is reversed, with Argonautes 1, 3 and 4 being favored, and Argonaute 2 shunned.

6) Perhaps the most curious difference of all is that, unlike metazoan-and-plant-specific miRNAs, tRFs are ubiquitous.

You can find them everywhere from bacteria, to people, to plankton (a BMC Genomics paper recently spotted tRFs in the diatom Phaeodactylum tricornutum), to plants (as shown in this Biology Direct paper).

tRFs offer one explanation as to why bacteria have Argonaute proteins despite not possessing a miRNA pathway. Jason and the Argonauts is a legend of great antiquity, and we can infer from their ubiquity that tRFs are also very ancient – most likely an ancestral gene-silencing mechanism present in the last universal common ancestor.

7) Dutta and colleagues have constructed a database (maybe they used JSON?) of Argonaute-interacting tRFs, which they hope will launch other curious biologists on their own odysseys of discovery.

But, wait, there's more!

The tendency of tRNAs to break down doesn't stop with tRFs. As stated above, they can degrade into larger pieces, known as tRNA halves, which have also been shown to play a regulatory role in gene expression.

Whereas small tRFs target mRNAs for silencing through interactions with Argonautes, tRNA halves use an altogether different mechanism. Taking advantage of their ability to mimic full-length tRNAs, the halves seek out tRNA-binding sites in ribosomes. Occupation of these sites can stall mRNA translation and so downregulate protein expression.

Even before they are transcribed into RNA, tRNAs can fall apart. That is, tRNA genes are sometimes split. A number of organisms, mostly but not exclusively archaeal species, have split some of the tRNA genes in their genomes into two. In other cases, they are not split, but permuted.

Between tRFs, tRNA halves and fragmented tRNA genes, it seems that tRNAs just don't like to be whole.

Topics:

Share this post

< PREVIOUS POST

NEXT POST >

3 Comments

[…] In Greek mythology, the Argonauts are a band of heroes who accompany Jason on his quest to find the Golden Fleece, a garment whose origins likely lie in the use of sheep fleeces as sieves to collect gold flakes from running water. […]

[…] In the years since the first draft of the human genome was released, our understanding of the incredible complexity that lies beyond protein coding regions has grown considerably. Of particular note is the growing family of non-coding RNAs. In a recent study in BMC Biology, Anindya Dutta from the University of Virginia School of Medicine, USA, and colleagues probe one of the newest members of this family, tRNA derived RNA fragments (tRFs). There is still much debate over their biogenesis and function, leading Dutta and colleagues to carry out a meta-analysis of publically available data on tRFs. Through their analysis it was revealed that these fragments are more evolutionarily conserved than their better known contemporaries, the microRNAs, and moreover uncovered that they, like microRNAs, are able to bind Argonaute proteins to form complexes that are involved in gene silencing. Here Dutta explains some of their more surprising results, speculates upon the function of tRFs, and discusses what comes next in the tRF story. More on what makes tRFs such an intriguing class of non-coding RNAs can be found in this BioMed Central blog. […]

[…] tRFs and the Argonautes: gene silencing from antiquity – BioMed Central blog. […]