You'd Prefer An Argonaute

Secreted Monocytic miR-150 Enhances Targeted Endothelial Cell Migration

Yujing Zhang, Danqing Liu, Xi Chen, Jing Li, Limin Li, Zhen Bian, Fei Sun, Jiuwei Lu, Yuan Yin, Xing Cai, Qi Sun, Kehui Wang, Yi Ba, Qiang Wang, Dongjin Wang, Junwei Yang, Pingsheng Liu, Tao Xu, Qiao Yan, Junfeng Zhang, Ke Zen, and Chen-Yu Zhang

Molecular Cell 39, 133–144, 9 July 2010.
DOI: 10.1016/j.molcel.2010.06.010

This week’s summary and gloves-off analysis by Anonymous:

This group had previously examined microRNA (miRNA) profiles in the serum samples of patients with certain cancers and diabetes, and found them to be able to serve as biomarkers for these diseases (Chen et al, 2008). In that study, they also found that serum miRNAs were resistant to RNase A digest and this study follows up on that. Exosomes/microvesicles (MVs) are small vesicles shed from many cell types of endocytic origin. These are delimited by a lipid bilayer and have been found to contain proteins, mRNAs and miRNAs. MVs can deliver their contents to recipient cells and while it has been shown previously that delivered proteins can alter cellular functions in recipient cells (Skog et al, 2008; Valadi et al, 2007), there has been no direct evidence of miRNAs being delivered to alter target gene expression in recipient cells. This study thus set out to fill that gap.

Briefly, the group first shows that MVs generated by THP-1 cells (a human macrophage/monocytic cell line) contained miRNAs that were resistant to RNase A digest by virtue of the protection afforded by the MV membrane. Next, the authors attempted to show that upon treatment by various stimuli, cellular miRNAs are selectively packaged into MVs such that the miRNA profile in MVs differs from that in the origin cells. However, the evidence was not convincing. The entire study uses quantitative real-time PCR (qRT-PCR) to measure miRNA expression levels. Aside from concerns that qRT-PCR measurements of miRNAs can be wildly noisy, this study is also handicapped by the fact that a reliable internal control that can be found in both cells and MVs is hard to find (it is unclear which control was used in this study, if any). Although the authors attempted to get around this issue by measuring absolute levels of miRNAs normalized to the total protein content in MVs, the miRNA levels in the “no-treatment control” for three different sets of stimuli are not very comparable (even though they should be if absolute levels were measured), underscoring the noise inherent in the miRNA qRT-PCR and/or normalization method. As such, it cannot be said conclusively that miRNAs are selectively packaged into MVs upon different stimulation. It would have been better if the authors had used deep sequencing to quantify miRNA expression instead.

It is, however, fair to say that MVs from THP-1 cells contain high levels of miR-150, which can be delivered to recipient HMEC-1 cells (an endothelial cell line). Upon incubation with THP-1 MVs, miR-150 levels (originally low in HMEC-1 cells) were increased in the recipient cells. The authors also checked that this was not because interactions with the MVs caused the HMEC-1 cells themselves to upregulate expression of miR-150 by checking the levels of pre-miR-150 (which were unaltered) in the HMEC-1 cells. The delivered miR-150 was shown to repress the protein levels of c-Myb, a known miR-150 target, in HMEC-1 cells, and this downregulation enhanced the migration capability of the HMEC-1 cells. Numerous controls were done here to demonstrate that this effect could only be seen when the donor MVs came from cells with high levels of miR-150, which is perhaps the redeeming factor in this paper. Although the authors showed that miR-150 repressed c-Myb protein expression via the 3′ untranslated region (3′UTR), they did not mutate the miR-150 target sites in the 3′UTR to show direct targeting definitively. The paper ends by showing that MVs that were intravenously injected into mouse tail veins can be taken up by the endothelium of mouse blood vessels. Interestingly, the authors also found that MVs from the plasma of patients with atherosclerosis have high levels of miR-150 and that incubation of recipient HMEC-1 cells with these MVs replicated the effects seen (repressed c-Myb protein levels, increased cellular migration) when HMEC-1 cells were incubated with THP-1 MVs.

Several questions remain. As the evidence for selective packaging of miRNAs into MVs is tenuous, it remains to be determined if this is indeed true. If this is true, the mechanism of miRNA packaging would be a natural question to address and miRNAs that are processed differently might behave differently in this respect. In the immunology field, MVs are thought to be “zipcoded” by having different combinations of markers/receptors on their surface (Théry et al, 2002). This paper only tested HMEC-1 cells as the recipient cells and it would be interesting to see if monocytic MVs can be targeted to different cell types and thus modulate the cellular environment differently. In the paper, the delivered miR-150 appeared to repress c-Myb protein levels by ~4-fold, which seems rather high, even after taking into account that the c-Myb 3′UTR has two conserved 8mer seed matches to miR-150. It would have been nice if the authors had determined the concentration reached by miR-150 in the recipient cells, relative to endogenous miRNA concentrations, to see if this could explain the strong repression. Alternatively, as monocytic MVs (of a different cell line) were previously found to be enriched in GW182 (Gibbings et al, 2009), it would be interesting to see if this enrichment also occurs in THP-1 MVs and had somehow contributed to the strong repression observed. At the end of the paper, the authors suggest that finding high levels of miR-150 in the plasma MVs of atheroschlerotic patients may indicate that a contributing factor to atherosclerosis might be the secretion of MVs with high levels of miR-150 by stimulated macrophages, which then cause target endothelial cell migration. However, the cellular origin(s) of these plasma MVs was not determined. This hypothesis thus remains to be tested.

References:

Chen et al (2008) Characterization of microRNAs in serum: a novel class of biomarkers for diagnosis of cancer and other diseases. Cell Res 18: 997-1006

Gibbings et al (2009) Multivesicular bodies associate with components of miRNA effector complexes and modulate miRNA activity. Nat Cell Biol 11:1143-1149

Skog et al (2008) Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide diagnostic biomarkers. Nat Cell Biol 10: 1470-1476

Théry et al (2002) Exosomes: composition, biogenesis and function. Nat Rev Immunol 2:569-579

Valadi et al (2008) Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells. Nat Cell Biol 9:654-659

Citation for researchblogging.org:

Zhang Y, Liu D, Chen X, Li J, Li L, Bian Z, Sun F, Lu J, Yin Y, Cai X, Sun Q, Wang K, Ba Y, Wang Q, Wang D, Yang J, Liu P, Xu T, Yan Q, Zhang J, Zen K, & Zhang CY (2010). Secreted monocytic miR-150 enhances targeted endothelial cell migration. Molecular cell, 39 (1), 133-44 PMID: 20603081

Target RNA–Directed Trimming and Tailing of Small Silencing RNAs

Stefan L. Ameres, Michael D. Horwich, Jui-Hung Hung, Jia Xu, Megha Ghildiyal, Zhiping Weng, Phillip D. Zamore

Science 328: 1534–1539, 18 June 2010.
DOI: 10.1126/science.1187058

This week’s thorough summary and analysis by Anonymous:

Like you, Drosophila siRNAs’d prefer an Argonaute; in fact their Argonaute of choice is Ago2. And this preference comes with a bonus: Hen1 adds 2′-O–methyl groups to all Ago2-bound small RNAs. In this paper, Ameres et al. attempted to uncover the mystery of the methyl group present in the 3’ end of Drosophila siRNAs, which is absent for most microRNAs. Adding an evolutionary twist to the story, in plants, all small RNAs (including microRNAs) are methylated. The obvious difference between the plant microRNAs and those of animals is the degree of complementarity to their targets. In plants, microRNAs are almost completely complementary to their target, hinting that the addition of the methyl group might be related to how tightly the small RNA is bound to its target.

Ameres et al. transfected several Drosophila cell lines with microRNA sensors, which had one or more fully complementary sites for an intended microRNA.  Normally these microRNAs are loaded to the Ago1; hence, they lack a methyl group in their 3’ end. They showed that levels of endogenously expressed (miR-34 and bantam) and induced (let-7 and miR-125) microRNAs decreased whence the corresponding microRNA sensor is present. Moreover, for another microRNA (miR-277), which can be loaded into both Ago1 and Ago2, they showed that the Ago1-loaded microRNA population is destabilized upon transfection of a microRNA sensor for miR-277. Therefore, Ago1-loaded small RNAs are prone to destabilization when a fully complimentary target is present.

In order to follow the fate of Ago1-loaded and destabilized microRNAs, the authors radiolabeled the 5’ end of let-7 with ³²P in the presence of let-7 sensor in vitro and observed new “tailed” and “trimmed” forms of let-7. They also detected tailed and trimmed forms of endogenously expressed bantam in the presence of a bantam-sensor using Drosophila embryo lysates. An addition of 3’ methyl group protected let-7 from being tailed and trimmed, and absence of Hen1 lead to tailing and trimming of let-7 siRNA. Putting all this evidence into context, the authors conclude that the methyl group protects small RNA from destabilization.

To determine the extent of tolerable complementarity between a microRNA and its target that does not lead to microRNA destabilization, the authors changed the sequence of complementarity site of the microRNA sensor. They found that targets resembling classical microRNA target sites did not result in tailing and trimming, whereas the targets that had less than 8 mismatches to the 3’ end of microRNA triggered destabilization. Furthermore, they also discovered that a small central bulge of 3nt leads to trimming and tailing, but not larger bulges. Hence, target RNA-triggered tailing and trimming require extensive but not necessarily perfect complementarity to the microRNAs.

Since Hen1 plays such a major role in protecting Ago2-bound small RNAs, the authors sequenced the small RNAs from hen1 mutant flies. The abundance and the length of microRNAs were not affected by lack of functional Hen1, but endo-siRNAs showed drastic changes in their lengths and abundance. The authors analyzed the content of the tails that arose due to absence of Hen1 and they found that the tail was either a single adenine or a single or multiple uridines, which is a mark of RNA turnover. Lastly, the authors also looked for target RNA-directed trimming and tailing in mammalian cells and confirmed the presence of a similar mechanism with slight modifications.

The Zamore Lab previously characterized RNA sorting in Drosophila, showing that the degree of complementarity between the small RNA and its star strand was responsible for being selectively loaded into Ago1 or Ago2. MicroRNAs are generally loaded into Ago1 to repress target translation and decrease target stability, often mediated by seed-paring between the microRNA and its target. The presence of highly complementary targets for Ago1 loaded small RNAs lead to remodeling of the small RNA by tailing or trimming. Therefore, Ago1 bound small RNAs are specialized in regulating partially complementary sites, which explain why microRNA targets lack extensive complementarity. The authors also speculate that extensive 3’ pairing might result in release of the small RNA from the PAZ domain of Argonaute, which in turn exposes the small RNA to the nucleotidyl transferases and 3’-to-5’ exonuclease enzymes.

Unfortunately we do not know which factors lead to target RNA-directed tailing and trimming and this will be an interesting venue of research in upcoming years. Moreover, the current model fails to explain the existence of highly complementary microRNA target sites present in both mammals and flies. But the paper formulates the question it tries to answer clearly from the beginning and delivers a satisfactory answer.

Citation for researchblogging.org:

Ameres SL, Horwich MD, Hung JH, Xu J, Ghildiyal M, Weng Z, & Zamore PD (2010). Target RNA-directed trimming and tailing of small silencing RNAs. Science, 328 (5985), 1534-9 PMID: 20558712

Down-Regulation of a Host MicroRNA by a Herpesvirus saimiri Noncoding RNA

Demián Cazalla, Therese Yario, Joan Steitz

Science Vol. 328: 1563 – 1566, 18 June 2010.
DOI: 10.1126/science.1187197

This week’s attentive summary and analysis by David Koppstein, his first on the blog:

In this study, Cazalla and colleagues found complementarity of three endogenous microRNAs–miR-27, miR-16, and miR-142-3p–to two noncoding RNAs encoded by Herpesvirus saimiri, HSUR 1 and HSUR 2, which have conserved motifs reminiscent of cellular U snRNAs. They pursued this observation by pulling down Ago2 and looking for these HSURs, and conversely by pulling down Sm proteins and looking for the miRNAs. Interestingly, these CoIPs specifically pulled down the HSURs and miRNAs, respectively, that were bioinformatically predicted to interact. During these experiments, they noticed that a mutant strain of the virus that lacked HSURs 1 and 2 expressed miR-27 at significantly higher levels.

Cazalla and colleagues then performed experiments to determine what was causing the increased levels of miR-27 in the mutant strain. They designed a pulse-chase nucleofection protocol with a radiolabeled synthetic miRNA, and noted a significantly shorter half-life in the presence of HSURs 1 and 2. Since levels of the pre-miRNA and the passenger strand were unchanged, they concluded that the mature miRNA itself must be destabilized.

It was also noted that the destabilization of miR-27 had consequences on the transcriptional landscape of the cell. A validated target of miR-27, FOXO1, which is a transcription factor that is dysregulated in breast, prostate, and endometrial cancers, was observed to be significantly downregulated in the absence of HSUR 1. Cazalla et al. also recapitulated the downregulation of miR-27 in Jurkat T-cells by
combinations of stable lines expressing HSURs and knockdowns using chimeric oligoribonucleotides. Strikingly, they observed that the specificity of HSUR 1 could be artificially switched to target miR-20.

There are several questions raised by this paper. First, there is the tantalizing prospect that the mechanism of destabilization of miR-27 is the same as that described by Ameres et al. in the same edition of Science. Further work, especially deep sequencing analysis of small RNAs, will likely reveal whether this is the case. There is also the mystery of where the interactions between HSURs and miRNAs are taking place; HSURs are thought to be nuclear, but may shuttle to the cytoplasm during their maturation. It is also still unclear what genes are perturbed as a consequence of the decreased miR-27 levels, and how this may affect viral fitness. Finally, one wonders about the functional significance of the interaction of HSURs 1 and 2 with miR-142-3p and miR-16. In summary, this paper presents a novel mechanism by which a virus perturbs host gene expression using noncoding RNA.

Citation for researchblogging.org:

Cazalla D, Yario T, & Steitz J (2010). Down-regulation of a host microRNA by a Herpesvirus saimiri noncoding RNA. Science, 328 (5985), 1563-6 PMID: 20558719

A coding-independent function of gene and pseudogene mRNAs regulates tumour biology

Laura Poliseno, Leonardo Salmena, Jiangwen Zhang, Brett Carver, William J. Haveman & Pier Paolo Pandolfi

Nature 465: 1033–1038, 24 June 2010.
doi:10.1038/nature09144

No formal write-up for this week, rather just some points to consider, raised during our journal club discussion:

The authors’ probing of expression levels in prostate cancer cells showed that PTENP1 was expressed at much lower levels, perhaps up to 100-fold less, than PTEN. If this is true, how can CMV driven overexpression of PTENP1 RNA come close to recapitulating the relative levels and biological interactions of these two transcripts in the cell? The abundance of each transcript will strongly influence their abilities to act as miRNA sponges for one another. This concern wasn’t really alleviated even when they looked at expression in normal human and prostate tissue samples.

Since their central argument was that PTENP1 acts as a sponge for miRNAs that can also bind PTEN, since many miRNA binding sites are present in both transcripts, they were remiss in not showing the effect of expressing the PTENP1 3’UTR with mutated miRNA binding sites. This is a crucial control. While they showed that there is a difference in expression when the sites are mutated, using luciferase reporter constructs, the difference was small for the number of sites they predict are targeted by various miRNAs.

They also observed PTENP1 overexpression reduced colony formation significantly more than repressing PTEN protein alone, indicating expression of PTENP1 is important for reasons other than regulating how miRNAs bind PTEN (perhaps also not surprising given the near perfect maintenance of a large block of nucleotide sequence between the gene and pseudogene, not just the miRNA binding sites). Together, the data don’t provide adequate support to the authors’ claim that PTENP1 is an endogenous miRNA sponge for PTEN.

Update 7/8/10: A quick search for this paper on the web revealed it is quite popular with the ID folk–they’re eating it up good! And then I found the press release from Beth Israel Deaconess Medical Center, where Dr. Pandolfi says:

… not only have we discovered a new language for mRNA, but we have also translated the previously unknown language of up to 17,000 pseudogenes and at least 10,000 long non-coding (lnc) RNAs. Consequently, we now know the function of an estimated 30,000 new entities, offering a novel dimension by which cellular and tumor biology can be regulated, and effectively doubling the size of the functional genome.

Oh, well if you put it that way… now I see why it was in Nature.

Citation for researchblogging.org:

Poliseno L, Salmena L, Zhang J, Carver B, Haveman WJ, & Pandolfi PP (2010). A coding-independent function of gene and pseudogene mRNAs regulates tumour biology. Nature, 465 (7301), 1033-8 PMID: 20577206

Update 9/8/15:

The Reproducibility Project: Cancer Biology is attempting to reproduce some results central to this study. Godspeed to you!

http://elifesciences.org/content/4/e08245

Mitotic cell-cycle progression is regulated by CPEB1 and CPEB4-dependent translational control

Isabel Novoa, Javier Gallego, Pedro G. Ferreira  &  Raul Mendez

Nature Cell Biology 12: 447 – 456, May 2010.
doi:10.1038/ncb2046

This week’s lucid summary and analysis by Noah Spies. This is Noah’s second contribution to the blog:

A normal eukaryotic messenger RNA is capped, spliced and polyadenylated within the nucleus in preparation for export into the wild world of the cytoplasm. Once in the cytoplasm, the translational machinery will recognize the cap and poly(A) tail and recruit ribosomes to translate the message until its eventual decapping/deadenylation and decay. In the late 1980s, it was discovered that many messages in Xenopus oocytes lay dormant in the cytoplasm until oocyte maturation or fertilization. These messages are recognized by a Cytoplasmic Polyadenylation Element (CPE) in their 3′ UTR by the CPE binding protein CPEB, and are deadenylated in the cytoplasm and stored in a translationally inert form until activation and re-adenylation at some later time point. Previous in vitro experiments suggested this system may activate some messages during the embryonic cell cycle, but there was little follow-up on these old results prior to this paper.

Novoa, et al (2010) used a differential elution system to purify mRNAs with short poly(A) tails (<30bp) and compared this pool to total mRNA (poly(A) tail length > 9bp) using microarrays. The authors isolated HeLa cells in S phase and compared these to cells in G2/M phases. The microarray experiment, largely validated by PCR (75% concordant results), showed several hundred genes whose poly(A) tails were either longer or shorter in G2/M than in S phase, and polysome profiling confirmed translational changes for a handful of these.

To connect these results to cytoplasmic polyadenylation, Novoa and colleagues showed the CPEB1 and CPEB4 were expressed in HeLa cells and CPEB1/2/4 were functional in an in vitro binding assay. Upon shRNA knock-down of CPEB1 and CPEB4, a number of genes showed differential polyadenylation profiles by microarray, hinting at a possible global role for CPEBs in regulating genes during the cell cycle. Experimental follow-up on one of these hits showed shRNA knock-down of CPEB lengthened the poly(A) tail of the Mnt mRNA, and this lengthening increased the protein levels. Other genes (e.g. CDKN3) showed similar effects.

If there are so many genes regulated by CPEB1/4 during the cell cycle, it should follow that there would be a significant phenotype following knock-down of these factors. Indeed, the authors found cell proliferation defects in both knock-downs, but not in a control knock-down. Much FACS analysis later, the authors conclude that CPEB knock-down results in a mitotic entry defect.

This work suggests that there is yet another layer of gene regulation operating during the cell cycle, and like any good research, raises more questions than it answers.

First, poly(A) tail lengths clearly change after CPEB knock-downs, but it remains an open question how many of these effects are direct. The authors attempted a cursory motif enrichment analysis, but by their own admission, this was limited in its scope by the loose definition of these elements. Differentially polyadenylated genes were enriched for a number of regulatory motifs (CPE, ARE, microRNA), suggesting that some of the changes might have been due to effects other than cytoplasmic polyadenyation. A simple follow-up would be to express a reporter containing a putative regulated 3′ UTR and follow poly(A) tail length after CPEB knock-down or after mutation of the CPE element. Further work might include immunoprecipitation of CPEB-bound RNAs at different cell cycle stages to confirm a direct interaction. And the motif work could be expanded upon with more specific control sets (e.g. only expressed genes) and with only microRNAs expressed in HeLa cells (it’s unclear how this was done).

Secondly, it is important to identify which CPEB interactions are important for the cell proliferation defect, as it is plausible that such an effect could result from shRNA off-target effects.

Finally, the CPEB1 knock-out mouse is viable, though sterile. If CPEB1 plays an important role during the cell cycle, this suggests that CPEB4 may be partially redundant with CPEB1. Seemingly inconsistent with this result, CPEB1 and CPEB4 knock-downs show only a small proportion of the same regulated genes.

Citation for Research Blogging:

Novoa I, Gallego J, Ferreira PG, & Mendez R (2010). Mitotic cell-cycle progression is regulated by CPEB1 and CPEB4-dependent translational control. Nature cell biology, 12 (5), 447-56 PMID: 20364142

Most “Dark Matter” Transcripts Are Associated With Known Genes

Harm van Bakel, Corey Nislow, Benjamin J. Blencowe, Timothy R. Hughes

PLoS Biology, 8 (5): e1000371, 18 May 2010.
doi:10.1371/journal.pbio.1000371

This week’s illuminating summary and analysis by Igor Ulitsky. It’s Igor’s second contribution to the blog:

The paper is one of at least five interesting RNA-Seq papers that came out in the past month or so (see also Guttman et al. and Trapnell et al. in the May issue of Nature Biotechnology, Kim et al. from Nature 465 (7295): 182-7 and De Santa et al. from the same issue of PLoS Biology). All these studies harness the awesome power of Illumina RNA-Seq to look at (mainly) the murine transcriptome and to try to figure out what it consists of, and how what RNA-Seq tells us differs from what we knew previously. Unfortunately, since the reads in those studies are still <75 long, RNA-Seq still can’t tell us exactly what the transcripts in the cell are, but rather what regions of the genome seem to give rise to RNA (or more precisely – which regions of the genome can we uniquely align reads to). This problem is partially alleviated by paired-end RNA-Seq used in the Nature Biotech papers, but it still has limited power for deciphering transcripts expressed at low levels. The five studies mentioned above tell three different stories about the transcriptome – the two Nature Biotech papers talk about how it is possible to identify thousands of novel exons in known genes, and also to give significantly more accurate exonic structures to some of the previously proposed long non-coding RNAs (lincRNAs). Kim et al. and De Santa et al. talk about a surprising amount of RNA coming from enhancer regions in the mouse genome, RNA whose exact function remains a mystery. The study we’re focusing on – van Bakel et al., tackles a more global question – how much of the polyA+ RNA comes from “known genes”, and how much from everything else – the “dark matter”. This question is naturally of high interest, but addressing it involves a wealth of caveats:

• What are “known genes” (i.e., “non-dark matter”) – protein-coding ones? miRNAs? Coding and non-coding ones with known functions?
• What expression levels can be considered functional? Are all transcripts with relatively low expression levels just noise?
• Are short RNA-Seq reads really informative in terms of the number of different RNA species?
• Are we confident enough in the annotation of the genome with pseudogenes and repeats, both of which can contribute to spurious mappings in intergenic regions?

Despite these caveats, van Bakel et al. do a thorough job of at least trying to answer this question, and do their best to convince the readers that, in fact, very little transcription from the mammalian genome is “dark matter”. When analyzing RNA-Seq data, the majority of the genome does not seem to give rise to detectable polyA+ RNA segments. Then how did ENCODE and related studies report as much as 80% of the genome as being transcribed? The authors begin by showing that tiling arrays are prone to give rise to many false positive calls. They do so by comparing their own tiling array data from human and mouse tissues to published and novel RNA-Seq datasets. Unfortunately, the sets are not completely matched (different labs/starting materials), but the data very convincingly shows that for transcripts with low expression data, the signal from tiling arrays is practically the same as the background – a fertile ground for false positive calls. It is interesting to note that the first part of the paper shows in fact that the data that the authors generated themselves (the tiling arrays) is worse than previously published data (RNA-Seq).

From that point on, the authors focus on RNA-Seq data. They find relatively few completely intergenic stand-alone transcripts that are not captured in some way in the “known genes” databases or at least in existing EST/mRNA collections. This is not very surprising given the effort involved in sequencing ESTs in mouse/human – it could hardly be expected that a lot of polyadenylated transcripts would be abundant in RNA-Seq, but missing from those datasets. It should be kept in mind though, that many of what the authors call “known genes” are in fact non-coding transcripts (based on lack of a long/conserved ORF) with completely unclear function. What about the sequence fragments (seqfrags) that do fall outside of the “known genes” boundaries? About 80% of those reads fall within 10kb of known genes and are likely to represent either unannotated parts of those genes, or transcripts whose biogenesis function is related to the gene adjacent to them, as their expression is generally highly correlated with their neighbors. What about the rest? Are there any interesting RNAs out there in the intergenic space? Well, there are some – the authors identify about 11,000-16,000 seqfrags that are located >10kb away from any known gene and that are significantly different from expected. The novel intergenic transcripts tend to overlap regions of open chromatin –identified using DNAse I hypersensitivity – which suggests that at least some of them could be the enhancer-associated transcripts reported in the parallel studies.

The authors then go on to show that by looking at splice junctions derived from the reads (using the popular TopHat tool) they can reach roughly the same conclusion – most of the spliced polyA+ RNA is already “known” to us. They can still identify about 5,000 novel exons in “known genes”, and those share the general characteristics of the known exons, albeit with lower expression and conservation levels. The imminent problem from this section, and from all the other recent RNA-Seq studies is this: Is anyone keeping track of all those new exons? Updating RefSeq/UCSC/Ensembl? How to update these databases is also an excellent question, as short-read-based studies cannot give us a complete (or close to complete) snapshot of the actual transcript. Anyhow, at this pace, we expect to see many additional papers re-discovering the same set of 5,000 novel exons.

The bottom line?

1. The outback of the genome rarely gives rise to highly expressed and polyA+ transcripts. This does not mean that there is shortage of putative lincRNAs – hundreds of them are already in the “known genes” set, and others may be functional despite low expression levels/proximity to known genes. The jury is still out on the polyA-transcriptome.
2. Annotation of the “canonical genes” in the mouse/human genomes is still not complete and both can be complemented with several thousand additional exons. Let’s hope somebody is keeping track.
3. Many intergenic RNAs are likely to be enhancer-associated (but we still don’t understand why).

This paper (as well as the other recent RNA-Seq studies) was definitely interesting to read, and we can only look forward to what we will learn once long-read RNA-Seq (e.g., Pacific Biosciences) kicks in.

Citation for researchblogging.org:

van Bakel H, Nislow C, Blencowe BJ, & Hughes TR (2010). Most “dark matter” transcripts are associated with known genes. PLoS biology, 8 (5) PMID: 20502517

A dicer-independent miRNA biogenesis pathway that requires Ago catalysis

Sihem Cheloufi, Camila O. Dos Santos, Mark M. W. Chong  & Gregory J. Hannon

Nature, 465: 584–589, 3 June 2010.
Nature AOP, 27 April 2010.
doi:10.1038/nature09092

This week’s summary and ruminative analysis by Vikram Agarwal. It’s Vikram’s second contribution to the blog:

In this article, Cheloufi and colleagues demonstrate an alternative biogenesis pathway for the maturation of a microRNA. A key question that this study seeks to address is why a member of the Argonaute family of proteins has retained its catalytic activity throughout millions of years of mammalian evolution. Though a handful of examples are known of miRNA-mediated cleavage events in animals, none have been shown to be crucial for target gene regulation and cell viability. The evidence thus suggests the possibility that the catalytically active residues of Argonaute have been conserved for purposes that are distinct from target cleavage.

The authors initially explore the consequences of losing catalytic activity of Argonaute during early mouse embryogenesis. They find that Ago2 is expressed ubiquitously in the early mouse embryo and in nearby placental tissues. Mice with a mutated catalytic residue of Ago2 develop normally through embryogenesis, but exhibit an anemic phenotype and die shortly after birth. Further cell sorting experiments confirm a problem in the maturation of erythrocytes, leading to an accumulation of pro-erythroblasts prior to birth.

During a check of global miRNA expression patterns in wild type and mutant mice, the authors discover that a single miRNA, miR-451, is aberrantly expressed in catalytically deficient mice. This miRNA is already known to have important roles in erythrocyte differentiation. More interestingly, small RNAs mapping to its precursor do not correspond to a canonical pattern one would expect if the miRNA were processed by Dicer. The precursor lacks a detectable miR* fragment, and the 3′ end of the mature miRNA is derived from the loop of the hairpin rather than the stem. The processing of the precursor is biochemically shown to be Drosha, but not Dicer, dependent, when compared to canonical miRNAs such as miR-294 and miR-16. The concluding experiments of the work investigate the role of Argonaute in the processing of pre-miR-451. In vitro purified Ago2 is demonstrated to be sufficient to produce a mature miR-451 cut, although the cleavage product is nearly 8-10 bases longer than expected. Detection of shorter products in in vivo samples, along with untemplated U addition in the 3′ end, implicate the role of an unknown exonuclease in the ultimate maturation of miR-451 into its shorter, functional form.

Overall, the biochemical evidence presented in this study is compelling, and strongly supports a model in which the Piwi domain of Ago2 catalytically cleaves miR-451. A conclusion that is less clear, though presented as an underlying motivation for the study, is that this non-canonical processing pathway explains the retention of Ago2 catalytic activity. An experiment that would provide support to this hypothesis would be to see whether a miR-451 knock-out recapitulates a similar phenotype as a catalytically-deficient Ago2 mutant. At present, it is possible that the target cleavage ability of Argonaute, or a yet unknown mechanism requiring its catalytic potential, could explain the evolutionary pressure for its conservation. Intuitively, it is difficult to comprehend why this non-canonical pathway has been conserved for only a single miRNA out of the hundreds encoded in the genome. The time spans involved during the evolution of the precursor are so vast that there must have been at least some opportunity for small insertions to have converted the miRNA into one that is processed canonically, or produced a novel miRNA convergently with the same seed sequence. Only further work will help elucidate why such strong evolutionary pressure exists to preserve this non-canonical processing mechanism.

Citation for researchblogging.org:

Cheloufi S, Dos Santos CO, Chong MM, & Hannon GJ (2010). A dicer-independent miRNA biogenesis pathway that requires Ago catalysis. Nature PMID: 20424607

Cooperation Between Translating Ribosomes and RNA Polymerase in Transcription Elongation

Sergey Proshkin, A. Rachid Rahmouni, Alexander Mironov, Evgeny Nudler

Science, Vol. 328. no. 5977: pp. 504 – 508, 23 April 2010.
DOI: 10.1126/science.1184939

This week’s smart summary and analysis by Pavan Vaidyanathan:

In this paper, the authors propose a mechanism for active physical cooperation between an elongating RNA polymerase (RNAP) and a translating ribosome following the polymerase on the transcribed message.

It has been known for several decades that in prokaryotes the processes of transcription and translation are coupled and occur simultaneously. Ribosomes are loaded on to a message as it is getting transcribed by RNAP. This coupled process serves to maximize efficiency of protein synthesis and allows for rapid changes in gene expression. It is also well established that the coupling of translation to transcription can serve as a means of regulation of RNAP. For instance, if the rate of translation is not as high as the rate of transcription, the increasing distance between the ribosome and RNAP allows for the transcription terminator Rho to bind the message and cause RNAP dissociation from the message. Additionally, numerous amino acid biosynthesis operons are regulated by the mechanism of transcription attenuation in which the rate of translation determines the formation of secondary structure behind RNAP that either allows the continuation of transcription or terminates transcription. In this paper however, the authors propose a novel means of regulation of transcription that depends on a physical interaction between RNAP and the first trailing ribosome behind the polymerase.

The authors first showed that in vivo, the ratio of the rate of transcription (nt/sec) to the rate of translation (aa/sec) is consistently ~3.0 in a variety of environmental conditions and stages of growth. Based on this observation, the authors hypothesize that the rate of transcription in vivo is determined by the rate of translation. To test this, they determined the rate of transcription of a plasmid-derived lacZ message in cells grown in media containing low concentrations of ribosome-binding antibiotic (chloramphenicol). They observed that slowing down the ribosome also slowed down the rate of transcription. They confirmed this by testing the rate of transcription of various messages harboring increasing amounts of rare codons, which are expected to slow down translation. As expected, the rate of transcription was inversely proportional to the percentage of rare codons in the message.

RNAP, like other polymerases, is known to backtrack on a message quite frequently. However, the presence of multiple polymerases on the same message significantly decreased backtracking. The authors hypothesized that the ribosome could improve the efficiency of transcription by serving as a physical block to backtracking and ‘forcing’ the polymerase to go forward. In order to test their model, the authors developed an assay to monitor RNAP backtracking in vivo. The assay employed the single-stranded DNA binding probe, chloroacetaldehyde, to monitor the migration of the transcription bubble. They observed that when a single RNAP molecule was forced to encounter a roadblock (lac repressor bound to DNA), it paused and backtracked. However, when there were two RNAP elongation complexes transcribing the same message, this backtracking was reduced substantially. Similarly, the presence of a ribosome behind RNAP also significantly reduced the incidence of backtracking suggesting that the ribosome could improve transcription by preventing backtracking of the leading elongation complex (EC). Using a similar roadblock system, the authors additionally showed by Northern blot analysis of the transcribed mRNA that the leading EC could had much higher rates of readthrough of the roadblock when it was followed by a second EC or by a ribosome.

The authors conclude that the ribosome directly controls the rate of transcription by preventing RNAP backtracking. Because of this cooperation, the rate of transcription is determined by codon usage and nutrient availability as sensed by the ribosome thus allowing precise regulatory adjustment of transcription to translational needs.

Citation for researchblogging.org:

Proshkin S, Rahmouni AR, Mironov A, & Nudler E (2010). Cooperation between translating ribosomes and RNA polymerase in transcription elongation. Science, 328 (5977), 504-8 PMID: 20413502

Caspase-Dependent Conversion of Dicer Ribonuclease into a Death-Promoting Deoxyribonuclease

Akihisa Nakagawa, Yong Shi, Eriko Kage-Nakadai, Shohei Mitani, Ding Xue

Science Express, 11 March 2010.
doi: 10.1126/science.1182374

This week’s methodische summary and analysis–impressively her third contribution to this blog–by Anna Drinnenberg:

In this paper Nakagawa and colleagues describe a new role for Dicer (DCR-1) in the apoptotic pathway of C. elegans. Briefly, the apoptotic pathway can be triggered by many different cellular stimuli, resulting in caspase activation and subsequent fragmentation of nuclear DNA. DNA fragmentation can be separated into two steps: (i) During the first phase a caspase-activated DNA endonuclease catalyzes formation of DNA nicks and breaks (ii) In mammals Caspase 3 activates the Caspase-activated deoxyribonuclease CAD generating 3’ hydroxylated (3’ OH) DNA breaks that can be detected by the TUNEL (TdT-mediated dUTP nick end Labeling) assay. Other endonucleases including EndoG, which translocates from the mitochondria to the nucleus, function at later stages to complete degradation of genomic DNA. Mutants of one or combinations of those endonucleases result in the accumulation of TUNEL-stained nuclei because the resolution of 3’ OH DNA breaks is impaired.

C. elegans has homologues of Caspase3 (CED-3) and EndoG (CPS-6), as well as other endonucleases involved in the second stage of DNA fragmentation, however, a homologue of CAD had not been found in the C. elegans genome. This group therefore aimed to identify the endonuclease that catalyzes the initial formation of 3’ hydroxylated DNA breaks. They performed an RNAi screen in the cps-6 deletion background and selected mutants that showed a decrease in TUNEL signal. DCR-1, a protein so far only known to be involved in RNAi pathways of C. elegans and other species was one hit from the screen.

They further confirmed the TUNEL results using DCR-1 deletion alleles, indicating that DCR-1 acts upstream of CPS-6 and other endonucleases. Moreover, they showed that DCR-1 has pro-apoptotic activity by counting the number of cell corpses during C. elegans embryogenesis in wild-type and dcr-1 deletion strains. Deletion strains of other factors involved in the C. elegans RNAi pathway did not show the same phenotypes indicating that the pro-apoptotic function of DCR-1 is independent of its role in the RNAi pathway. Furthermore, they found that DCR-1 is processed by CED-3, which cleaves its first RNAseIII domain. They named the c-terminal part of truncated form of DCR-1, “tDCR-1”, which lacks most of the full-length protein domains including the Helicase, the PAZ, and even one of the RNaseIII domains.

Studies in vitro incubating tDCR-1 with dsRNA or plasmid DNA showed loss of RNase activity, but a gain of DNase activity. The DNase activity of tDCR-1 appears to be weak, resulting in just a single cut of plasmid DNA instead of complete fragmentation. This would be consistent with a role for tDCR-1 in generating 3’OH DNA nicks, whereas completion of DNA fragmentation is carried out by other endonucleases like CPS-6. It would be therefore interesting to compare the enzymatic activity of tDCR-1 to CAD in the same in vitro assay. They also did some very interesting experiments that convincingly tied the function of the different forms of DCR-1 to either being involved in the apoptotic pathway or the RNAi pathway. For example, they rescued the mutant RNAi phenotype, but not the mutant apoptosis phenotype, of dcr-1 deletion animals by expression of an allele of DCR-1 resistant to CED-3 cleavage. In a complementary experiment, expression of tDCR-1 in the dcr-1 deletion background resulted in rescue of the apoptotic pathway but not the RNAi pathway. Finally, they showed the same acidic amino acids are important for both RNase and DNase activity, demonstrating the similarity of both enzymatic activities.

Nakagawa and collegues performed a very comprehensive study characterizing the role of DCR-1 acting as a functional analog of CAD, indicating that a conserved, caspase-mediated mechanism activates the apoptotic DNA degradation process in both C. elegans and mammals. Further studies could include visualizing the relative abundance of tDCR-1 versus full-length DCR-1 in the nucleus and the cytoplasm. Moreover, studies in other organisms, including those that have a homolog of CAD, could determine if the pro-apoptotic role of DCR-1 is specific to C. elegans or is conserved in other species as well. In conclusion, this study shows the participation of the same protein in two different unrelated pathways, an economical use of the genetic repertoire.

Citation for researchblogging.org:

Nakagawa A, Shi Y, Kage-Nakadai E, Mitani S, & Xue D (2010). Caspase-Dependent Conversion of Dicer Ribonuclease into a Death-Promoting Deoxyribonuclease. Science PMID: 20223951

Mammalian cell penetration, siRNA transfection, and DNA transfection by supercharged proteins

Brian R. McNaughton, James J. Cronican, David B. Thompson and David R. Liu

PNAS 106 (15): 6111-6116, 14 April 2009.
doi: 10.1073/pnas.0807883106

This week’s shrewd summary/analysis–impressively his third contribution to this blog–by David Weinberg:

In their 2009 PNAS paper, David Liu and colleagues demonstrate that a green fluorescent protein (GFP) variant that has been engineered to have a positively charged surface can penetrate mammalian cells and also chaperone nucleic acids into those same cells. The story began in 2007 when Liu’s lab published their initial characterization of so-called “supercharged” proteins. The motivation was to determine how changing the net charge of a protein can affect its stability. To do this, they began with an extra-stable GFP and changed as many surface-exposed residues as possible to positively charged residues (i.e., lysine and arginine). This resulted in a GFP variant with a net charge of +36 (herein referred to as +GFP) that folded and fluoresced similarly to the original GFP. Amazingly, however, +GFP was highly resistant to aggregation: boiling of the protein eliminated activity, but cooling the boiled protein restored most of the activity. This stabilizing effect of supercharging was not unique to GFP, as similarly supercharged variants of GST (a dimer) and streptavidin (a tetramer) showed similar properties.

So what does all of this have to do with RNA? In 2007, Liu’s lab also noted that +GFP could use its positively charge surface as “molecular Velcro” that can reversibly bind to RNA (tRNA) or DNA (plasmid dsDNA). Of course, +GFP-bound RNA is useless (to a first approximation) in a test tube. Where the PNAS paper begins is with the hypothesis that +GFP might be able to enter cells and thereby escort its bound nucleic acid cargo into the cell as well. The authors begin by conclusively showing that +GFP penetrate a variety of mammalian cells with an efficiency that varies with its charge. To satisfy the cell biologists in the audience, they provide an initial dissection of the mechanism of +GFP uptake using a variety of (mostly chemical) perturbations. From this, they conclude that the mechanism involves energy-dependent endocytosis that is dependent on actin polymerization and sulfated (positively charged) cell surface peptidoglycans but does not require caveolin or clathrin. Now focusing back on RNA, the authors demonstrate that +GFP can bind to siRNAs in vitro – not surprising given their previously published data showing that it can bind to tRNA. Moreover, +GFP-bound Cy3-labeled siRNAs can enter HeLa cells – their FACS data suggest that virtually all cells in the population take up the siRNAs to a similar degree, yielding a quite homogenous population of “transfected” cells. Although Lipofectamine (a standard transfection reagent) can also deliver Cy3-siRNA to HeLa cells, +GFP delivers ~100-fold more siRNA based on fluorescence. More impressively, in 4 other cell lines that are virtually resistant to Lipofectamine-mediated transfection, +GFP delivers huge amounts of Cy3-siRNA without any significant cytotoxicity. Not only do these siRNAs enter the cell, but they can interact with the RNA interference machinery and mediate gene silencing. Further characterization of +GFP-siRNA complexes reveals that +GFP enhances the stability of siRNA in serum and the protein is itself relatively stable in serum. Although less impressive in its efficiency, +GFP (with an HA tag) can also be used to transfect plasmid DNA that gets expressed in the nucleus.

The authors conclude that +GFP may provide an attractive alternative to nucleic acid delivery. Because it uses a general pathway (endocytosis) for delivery, it should in theory work in all cell types. In addition, it is relatively easy to use since it only requires mixing of recombinant +GFP (which can be readily made in E. coli) and the nucleic acid of interest, and adding this to cells. But beyond this immediate application as a transfection reagent, it seems that +GFP and supercharged proteins more generally may become a useful tool for all sorts of biology (which I have no doubt the Liu lab is already exploring).

Citation for researchblogging.org:

McNaughton BR, Cronican JJ, Thompson DB, & Liu DR (2009). Mammalian cell penetration, siRNA transfection, and DNA transfection by supercharged proteins. Proceedings of the National Academy of Sciences of the United States of America, 106 (15), 6111-6 PMID: 19307578

Regulation of Alternative Splicing by Histone Modifications

Reini F. Luco, Qun Pan, Kaoru Tominaga, Benjamin J. Blencowe, Olivia M. Pereira-Smith, Tom Misteli

Science 327: 996 – 1000, 19 February 2010.
DOI: 10.1126/science.1184208

This week’s substantial summary and analysis by Mohini Jangi:

This study is the latest in a string of papers within the last year that have sought to take a closer look at the link between chromatin modifications and splicing regulation.  It has now been analyzed and accepted in the field that splicing occurs co-transcriptionally, and that transcription can boost splicing efficiency and vice versa.  Focusing on transcription regulating splicing, there are thought to be two main mechanisms (although inter-related and not mutually exclusive) by which this occurs.  The first is that pre-mRNA processing machinery, including the spliceosome, can directly associate with the C-terminal domain of RNA Pol II as it is transcribing the gene.  This basically increases the chance that a splice site will get recognized because of the proximity of the splicing machinery to the nascent transcript.  The other is that nucleosome position can affect the speed of the elongating polymerase, which in turn will influence splice site recognition.  Interest in this second mechanism spurred this set of papers, which generally showed enrichment of specific chromatin marks, including H3K36me3, within exons compared to introns.  Here, the authors set out to address whether any specific chromatin marks were associated with alternative exon usage, specifically those events regulated by the splicing factor polypyrimidine tract binding protein (PTB).

They began by looking at a well-studied model in splicing regulation, fibroblast growth factor receptor 2 (FGFR2), which has a set of mutually exclusive exons IIIb and IIIc.  Exon IIIb is repressed by PTB in mesenchymal cell types (human mesenchymal stem cells in this study) and included in epithelial cell types (PNT2 cells).  ChIP assays on a number of histone marks showed H3K36me3 and H3K4me1 enrichment to correlate with PTB-dependent splicing of FGFR2.  When they extended this beyond FGFR2 to other PTB-dependent transcripts, this correlation also held, whereas PTB-independent transcripts did not show this enrichment.  Next they wanted to determine if these methylation marks are causal in the splicing regulation.  To this end, they overexpressed or depleted the H3K36 methyltransferases SET2/SETD2 or the H3K4 methyltransferase ASH2 and looked for repression of exon IIIb.  As expected, increased H3K36me3 and decreased H3K4me3 after modulation of methylation led to increased repression of IIIb.  To get at the molecular mechanism for this, they hypothesized that a component of the H3K4 demethylase complex that binds H3K36me3, MRG15, may also be able to drive this splicing switch.  Indeed, ChIP experiments showed MRG15 enrichment correlating with IIIb repression, and knockdown and overexpression experiments showed enhancement and repression of IIIb, respectively.  Co-IP’s also showed a small pool of PTB and MRG15 associating, and RNA-IP similarly showed both proteins associating with the nascent exon IIIb.  To bring this into a larger context, the authors performed high-throughput cDNA sequencing in hMSCs in the presence or absence of PTB, MRG15, or SETD2.  They saw that of 447 PTB-dependent and 186 MRG15-dependent splicing events, 65 were common, of which 61 changed in the same direction upon knockdown.  Furthermore, transcripts weakly regulated by PTB made up the largest fraction of co-dependent transcripts.  From their data, they suggested a model that MRG15 is serving as an adaptor between the splicing machinery, or more specifically PTB, and the chromatin, mediated by H3K36me3.

Not being a computational biologist myself, I glossed over the fact that their sequencing analysis was not very stringent nor did it go into much depth regarding the nature of the PTB and MRG15 co-dependent transcripts.  The discussion that came up during the journal club brought this to the forefront.  For example, they did not include basic controls, such as addressing what fraction of PTB-regulated transcripts are only weakly PTB-dependent, regardless of whether they are also MRG15-dependent. Another point of contention was how this might actually work in the cell – when you imagine PTB bound to chromatin at a specific locus, it is difficult to picture how this would be noticeably more efficient than PTB associated with the CTD of Pol II. The strength of the paper lay basically in the idea that methylation marks can actually drive changes in alternative splicing, and that there is a direct association between these marks and splicing factors involved in this regulation.  It further raises the questions of what other splicing factors and other histone marks might be working in this manner.  Overall, it was an interesting study that raised some good discussion, and there is most likely much more interesting work to come from it in the future.

Allosteric regulation of Argonaute proteins by miRNAs

Sergej Djuranovic, Michelle Kim Zinchenko, Junho K Hur, Ali Nahvi, Julie L Brunelle, Elizabeth J Rogers & Rachel Green

Nature Structural & Molecular Biology 17, 144 – 150, February 2010.
doi:10.1038/nsmb.1736

This week’s summary and analysis by David Garcia:

It’s a good thing that one of the proteins central to RNAi, Argonaute, has a beefy, determined sounding name, because it’s everywhere these days. (Fortunately the wussy sounding “P-element induced wimpy testes,” or “PIWI” proteins are only a subclass of Argonautes, and not the other way around.) Argonautes owe their celebrity status to, of course, their relationship with widespread small RNAs. Numerous groups have studied Argonaute structure and function in vitro and in vivo. This week’s paper from Rachel Green’s group I think mostly convincingly shows Drosophila Ago1 operating allosterically, through a pleasing mix of structural bioinformatics, biochemistry, and cell culture experiments.

Focusing on the MID domain of eukaryotic Argonautes (Agos), the paper begins with bioinformatic analyses demonstrating sequence and structural similarities to the ligand binding domains of bacterial proteins that exhibit allosteric behavior, namely the coupling of metabolite binding in one site to active function in another distant site. Using the CLANS program they showed that MID domains themselves clustered into groups of sequence similarity that reflect their function: inhibition of translation (via miRNAs) or mRNA cleavage (via siRNAs). Agos that they classified as involved in translational repression, including DmAgo1, HsAgo1-4, and CeAlg1-2, formed a tight cluster separate from DmAgo2 and CeRde1, both which are involved in siRNA directed repression. They saw looser clustering for other domains in Ago, like PAZ or PIWI. Thus they argue that MID domain function, reflected in sequence, largely distinguishes various Ago family members. These conclusions might be tempered by the fact that some Agos, including HsAgo2 and DmAgo1, have dual-functionality: they can mediate mRNA degradation/translational repression and have slicer activity. (A growing body of evidence is showing that mRNA degradation is a substantial part of miRNA directed repression, although this doesn’t preclude their final model.)

The authors next searched for biochemical signatures that distinguish different Agos. Running tagged, purified MID domains over m7-GTP-Sepharose resins, they observed the miRNA Agos like DmAgo1 and CeAlg-1 bound more tightly than the siRNA Agos DmAgo2 and CeRde-1. Free nucleotides could compete with binding for DmAgo2 MID, suggesting a single site in this domain that binds the 5’ nucleotide of the sRNA was responsible for binding to the resin. In contrast, for the MID of DmAgo1, free nucleotides actually stimulated binding to the cap-like structure, suggesting a second allosteric site.

They next tested full-length Agos (DmAgo2 was slightly truncated) in the presence of miRNA, and saw strongly increased affinity for the cap-like structure for DmAgo1, but not for DmAgo2, indicating some crosstalk between two distinct sites that bind the sRNA and cap-like structure. The binding affinity of DmAgo1 for the cap-like structure increased with increasing [miRNA].

Filter binding assays were used to do the inverse experiment—whether binding of a miRNA can be stimulated by addition of free nucleotides/analogs with varying resemblance to the cap-like structure. They only saw stimulated miRNA binding with tri-phosphorylated nucleotides, which most closely resembled the cap. No such effect was seen for DmAgo2. Consistent with this result, addition of miRNA duplex stimulated binding of a labeled, capped mRNA ten-fold for DmAgo1, but no-fold for DmAgo2. Free nucleotides that resembled caps could compete away the labeled mRNA.

With these pleasant in vitro results in hand, demonstrating allostery for DmAgo1 but not DmAgo2, they moved in vivo into S2 cells. But here, while supporting their hypothesis, I find the results less satisfying due to their choice of reporter system. They employed a tethered Ago reporter system (reference) that examines Ago mediated inhibition of a luciferase reporter without need for a miRNA. The utility of the system comes from the ability to directly assess the effect of mutations in the tagged and tethered Ago on the reporter, without competition from endogenous Ago. But in my opinion the compromise is too great, ignoring the effect of a targeting miRNA on Ago structure and activity.

Anyways, they saw that DmAgo1 repressed the reporter strongly while DmAgo2 did not. Then they tested mutations in conserved residues in DmAgo1 predicted to be responsible for interacting with the 5’ nucleotide of the sRNA. These mutations actually caused substantial de-repression, which as they note, was very unexpected because the Ago protein is tethered to the mRNA, so it shouldn’t matter whether or not it can bind a 5’ nucleotide. They surmise that Ago needs to bind a miRNA for full activity, even if it’s tethered. (Another possibility is that even binding free nucleotides would be sufficient, and this was still perturbed in the mutants.)

Next they pursued an exposed region in the MID domain that could bind a cap. Mutation of a specific residue caused reduced binding to m7-GTP-Sepharose resin, and complete de-repression in the reporter assay. They declared this site the second allosteric site, which binds mRNA caps. In cell lysates, they observed that only the Ago variants that repressed in the reporter assays could effectively bind miRNA, cap structures, and GW182, a protein that has been implicated in miRNA directed repression.

So overall I found the bioinformatics and in vitro data very convincing, but for the in vivo data, despite the fact that it supports the allosteric model, I’m less enthusiastic because of the experimental system.

A burning question arising from their allosteric model is how the cap binds Ago. I wouldn’t naturally assume that the cap would be free and close to Agos moving along 3’ UTRs. Messenger RNAs are most efficiently translated when they’re circularized by protein-protein-RNA interactions between the PABPs and proteins that bind the cap. Could Ago or some other co-factor then destabilize these interactions to compete for the cap? Would this change be a prerequisite for Ago mediated repression? How reversible would the exchange be? What about mRNAs with multiple target sites–each message has only a single cap but may host multiple Agos? As the authors imply in the discussion section, Ago binding the cap may only be favorable when RISC is bound to an authentic target (potentially avoiding indiscriminate activation of Ago by the caps of other mRNAs floating nearby that aren’t necessarily themselves actively repressed), so there may be a kinetic argument to be made. Finally, has the allostery of DmAgo1, and potentially other Agos, co-evolved with its preference for miRNAs that usually target the 3’ UTR, a region perhaps physically closer to the cap of a circularized mRNA? Hopefully Rachel Green’s group and others are asking these questions. We crave answers.

Targeted 3′ Processing of Antisense Transcripts Triggers Arabidopsis FLC Chromatin Silencing

Fuquan Liu, Sebastian Marquardt, Clare Lister, Szymon Swiezewski, Caroline Dean

Science Express: 3 December 2009.
doi: 10.1126/science.1180278

This week’s punctilious summary and analysis by Igor Ulitsky:

The group of Caroline Dean in Norwich, as well as several other groups, has been extensively studying the regulation of the FLC gene in Arabidopsis. FLC is a key transcription factor in the flowering process, which is very tightly controlled in plants. Their studies have uncovered an extensive regulatory network that converges at the FLC locus, the mechanisms of which include transcriptional regulation, chromatin modification, small RNAs, and now also RNA processing. Two papers on FLC from Dean’s group were published last week – one in Science (Liu et al., the one I’ll focus on) and another in Nature (Swiezewski et al. Nature Vol 462). The Science paper focuses on the so-called “autonomous pathway” of FLC regulation, which promotes flowering by repressing FLC. Previous genetic analyses from the same group have uncovered a number of genes involved in this process, including two RNA-binding proteins, FCA and FPA, the cleavage and polyadenylation specificity factor FY, and an H3K4me2 demethylase, FLD. They have also previously shown (mostly in Liu et al., Molecular Cell 2007) that FCA is physically associated with the chromatin in an intron of FLC, between exons 6 and 7, but did not find clear evidence that FCA binding leads to processing of the FLC gene itself. Instead, it seemed that FCA played a role in regulating the transcription from the FLC locus, through chromatin modification by FLD. In addition, they characterized two anti-sense transcripts in the FLC locus, one short, ending around the physical location of FCA on the chromatin, and one long, ending at the FLC promoter. Interestingly, fca mutants had a higher long form/short form ratio than WT plants.

In this paper, Liu et al. first conducted a genetic screen for suppressors of FCA over-expression (which strongly repressed FLC). They used a fusion of FLC to LUC, and identified several genes that could activate FLC in presence of a strongly activated FCA. These included the known players FY and FLD, but also two additional genes, subunits of the CstF 3’ processing complex, which is highly conserved and essential in many species, including Arabidopsis. Through epistasis analysis, they could show that the CstF subunits indeed function in the FCA pathway. In addition, they show that these mutants have increased transcription of FLC, as evidenced by nascent RNA levels, Pol II binding, and H3K4me3 signatures. Their previous findings now pointed to a role of these subunits in regulating the long form/short form ratio of the antisense transcript. The paper doesn’t convincingly show that the sense FLC transcript is not affected, but it seems that the authors are convinced somehow that only the antisense is affected. Indeed, they find that the CstF mutants fail to process the 3’ of the antisense transcript, which leads in general to higher transcription of the anti-sense, which is similar to the elevation observed in fca mutants, and coincides with an increase in the FLC sense transcript. Overall, in different mutants in the FCA pathway, there is an increase in the abundance of the long form of the anti-sense transcript over the short form. The authors’ model for what happens in the WT strain, in which FCA represses FLC, is as follows: FCA/FY/CstF, through 3’ processing of the antisense transcript, causes shift in long-form/short-form ratio, which leads somehow to recruitment of FLD, which removes the H3K4me2 marks from the body of the FLC gene, leading eventually to down-regulation of both the sense and the anti-sense transcripts. They speculate on what may be the missing link between FCA/FY/CstF and FLD, but there is no clear evidence supporting it.

While the paper focuses on a single gene in Arabidopsis, there are several lines of evidence that this kind of regulation through 3’ processing of an antisense transcript occurs in other regulatory programs. In general, although the paper does not mention it, this kind of mechanism could explain why some regulatory proteins have conserved biding sites in the introns of their targets, as at least some of them may play roles in RNA processing of both the sense and the anti-sense transcipts.

To summarize, this paper, as well as another report on the FLC locus, show how amazingly complicated a relatively straight forward regulatory program (shutting down a target gene) can be. Another global implication that I found in this paper is how useful yet misleading genetic interactions can be. The authors found a genetic link between FCA and FLD, the chromatin modifier at the end of the pathway, in one of their previous studies. However, they could not find a direct physical link between them. As it appears now, FLD appears several pathway steps downstream of FCA, with complex machinery for 3’ processing of the antisense transcript separating them.

The paper was very interesting to read, and it gave a decent introduction to FLC pathway regulation. However, it wasn’t an easy feat to understand it thoroughly. The details of the unfolding of the pathway were not really clear in the first scan, and it was very difficult to understand which results were actually novel, and which have been reported previously. Part of the problem could be the extreme length limitations of Science publications, and lack of simple “textbook” figures in the paper.