Tiny RNA Biology
Small RNA pathways regulate gene expression, chromatin structure and defense against invading elements such as transposons and viruses. Small RNA deep sequencing experiments have shown that about half of all genes produce endogenous siRNAs suggesting that this regulatory axis controls a wide range of gene activities. siRNAs are incorporated into effector complexes, comprised of an Argonaute protein and accessory factors, which direct silencing of complementary RNAs and in certain species, such as C. elegans, recruit RNA-dependent RNA polymerases (RdRPs) to the target, resulting in siRNA amplification. All small RNAs act sequence-specifically through base pairing with their target mRNA, but the outcome of the small RNA:target interaction can vary from suppression of transcription to mRNA degradation or translational repression and this is likely governed in part by the specific Argonaute cofactor and in part by other coupled proteins that mediate the sorting of these small RNAs within the cell and between cells. Much remains to be learned about miRNAs and siRNAs and how they silence gene expression. There are hints that they sequester target mRNAs in P bodies and other subcellular structures, and constitute elements in an RNA trafficking system. There are hints of extracellular movement of small RNAs, perhaps constituting a shadow endocrine system. The genes identified are likely to be key factors in the function of small RNAs and their identification may enable more potent RNAi based drug development.
Our work follows up on our past full genome screens for candidate cofactor genes and our development of molecular signatures for small RNA defects using deep sequencing. The gene lists identify the key elements in the choreography of small RNAs in C. elegans. Because the genes are nearly all conserved across phylogeny, it is likely that the rules we work out in C. elegans will apply generally. Our goals are to use our deep sequencing analyses and new reporter genes for endogenous siRNAs and miRNAs to transform these lists of about 500 genes into coherent pathways of RNA processing, trafficking and presentation to the genome. We expect that the implications of the work will be to discover to new elements in how miRNAs and siRNAs are generated and engage cellular proteins to in turn engage their targets, as well as the discovery of other small RNA pathways from our analysis of the genes we have identified.
There have been 2 general areas of small RNA biology that we have developed: the mechanisms of exogenous and endogenous RNAi and the mechanism of miRNA regulation of target mRNAs.
The exonuclease ERI-1 plays a conserved dual role in RNA interference and ribosomal 5.8S RNA processing. Gabel, H. W. and G. Ruvkun. 2008. Nat Struct Mol Biol. 15: 531-3.
In C. elegans and fungi and plants, RNA dependent RNA polymerases amplify primary siRNAs to produce a more robust response. This two stage amplifier with feedforward amplification, like PCR, can amplify noise, so as in good electronic design, is likely to be attenuated. To identify negative regulators of C. elegans RNAi, we carried out a genetic screen for mutants that show an enhanced RNAi (Eri) phenotype. Mutants in eri-1 display an enhanced RNA interference (Eri) phenotype in which dsRNAs that are ineffective in silencing target mRNAs in wild-type trigger robust silencing in an eri-1 null mutant. eri-1 has a conserved endogenous function that was previously unsuspected: the 5.8s RNA inappropriately retains 1 to 4 nucleotides of its precursor in the eri-1 mutant compared to wild type. The homolog of ERI-1 in the fission yeast S. pombe also negatively regulates RNAi and we also observed that the 5.8S rRNA in the S. pombe eri1∆ mutant has 2 to 8 additional 3’ nucleotides. The Rao group at Harvard also discovered that mammalian ERI-1 targets the 5.8S RNA analogously (1). Thus ERI-1 plays a conserved dual role in the RNAi and rRNA processing pathways.
Trans-splicing in C. elegans generates the negative RNAi regulator ERI-6/7. Fischer, S.E.J., M. D. Butler, Q. Pan and G. Ruvkun 2008, Nature 455: 491-6.
Two other mutants from the enhanced RNAi genetic screen, eri-6 and eri-7 map to two adjacent ORFs present on opposite strands but failed to complement. The syntenic region in C. briggsae revealed that eri-6 and eri-7 constitute one gene on one strand in this species. The eri-6/7 protein is a member of the superfamily I DNA and RNA helicases. The divergently transcribed RNAs for eri-6 and eri-7 are assembled in C. elegans into one mRNA by trans-splicing. A 930 bp repeat 3’ to eri-6 and 5’ to eri-7 mediates this trans-splicing. We proved this by the detection adenosine to inosine editing of the proposed RNA duplex mediated by an adenosine deaminase (ADAR). A strain with mutations in both adenosine deaminase genes enhances transgene silencing, suggesting that editing of dsRNAs normally inhibits RNAi. One model for the increase in transgene silencing the absence of these ADARs is that the dsRNA element in eri-6/7 is a better target for the RNAi pathway when it is not edited, thus decreasing eri-6/7 gene activity.
The ERI-6/7 Helicase Acts at the First Stage of an siRNA Amplification Pathway that Targets Recent Gene Duplications (Fischer, SEJ., TA Montgomery, C Zhang, N Fahlgren, A Hwang, CM Sullivan, JC Carrington and G Ruvkun 2011, PloS Genetics 7(11): e1002369.
To characterize the role of eri-6/7 in endogenous siRNA pathways, we compared the small RNA profiles of eri-6/7 mutants as well as ergo-1 and eri-1 mutants to wild type. siRNAs corresponding to about 80 target genes were depleted in eri-6/7 mutants and the corresponding mRNA levels were up-regulated. The genes targeted by eri-6/7 show extensive nucleotide homology to each other but are not conserved between C. elegans and C. briggsae. These genes also have few introns, supporting the model that these genes have recently been acquired by C. elegans, perhaps from viruses. ERGO-1/ERI-6/7 pathway may constitute elements of such a viral surveillance pathway that extends beyond the initial infection.
An endogenous siRNA sensor reporter gene to easily assay the production or response to endogenous C. elegans siRNAs Montgomery TA, E. Rim, N Fahlgren, C Sullivan, JC Carrington, and G. Ruvkun.).
We developed a GFP based sensor for endogenous siRNA activity in C. elegans. 22G siR-1 is derived from a cluster of 22G siRNAs on the X chromosome that are dependent on embryonically expressed upstream ERGO-1 generated 26G siRNAs. The siRNA sensor, ubl-1::gfp-siR-1-sensor, contains a single engineered perfect duplex target site for an abundant endogenous siRNA, 22G siR-1, embedded in the 3’ UTR of ubiquitin-like1 (ubl-1) and expressed under the control of the ubl-1 promoter, which is active in all tissues throughout development (Figure 1). A control construct ubl-1::gfp, lacks the siRNA target site. Each construct was introduced into C. elegans using Mos1-mediated single copy insertion. GFP expression was ubiquitous in C. elegans containing the control, which lacks the 22G siR-1 target site, but was nearly absent in C. elegans containing the reporter with the 22G siR-1 sensor element in the 3’ UTR. The silencing of the siRNA sensor is suppressed by ergo-1 or eri-6/7 RNAi.
To determine if siRNAs trigger production of siRNAs in the genomic vicinity of the initial target site, we deep sequenced small RNAs from C. elegans containing either the ubl-1::gfp or ubl-1::gfp-siR-1-sensor transgene. 22G siRNAs were uniformly distributed across both transgenes and were derived exclusively from coding and vector sequence and not from the ubl-1 5’ and 3’ untranslated regions (Figure 1CD). A large peak was observed at the siRNA target site of the sensor, it is 22G siR-1 that may originate from the endogenous X-cluster siRNA locus, as the levels of 22G siR-1 were identical between control- and siRNA sensor-transgenic C. elegans. But this abundant siRNA does not attract more siRNAs to the region. These results suggest that, unlike primary exogenous siRNAs and endogenous 26G siRNAs, 22G siRNAs do not trigger siRNA amplification or spreading outside of the siRNA target site.
A genome-wide RNAi screen identifies RNAi and miRNA pathway components.
Sylvia E. J. Fischer, Qi Pan, Peter C. Breen, Taiowa A. Montgomery, Chi Zhang, Ellen Rim and Gary Ruvkun, data generation and analysis finished, writing paper.
We developed a fusion gene that enabled a full genome RNAi screen for enhanced siRNA or enhanced miRNA silencing. This fusion gene bears a dominant mutation in the collagen gene rol-6 that causes animals to move in an easily scorable rolling motion fused to the lin-14 3’ UTR which mediates the translational silencing of the lin-14 open reading frame or a lacZ reporter gene by the miRNAs lin-4 and let-7. The fusion gene was designed to study how mutations or gene inactivations that disable the miRNA pathway might fail to silence the fusion gene bearing the lin-14 3’ UTR. However, even though a similar lacZ fusion gene bearing the lin-14 3‘ UTR is silenced by lin-4 in wild type, the rol-6/lin-14 3’ UTR fusion gene is not repressed by lin-4 in wild type. The rol-6/lin-14 fusion gene is silenced by mutants that enhance RNAi and transgene silencing (for example the eri-1 mutant, or eri-6/7 RNAi, or synMuvB mutant RNAi) and non rolling animals are easily observed. This transgene was used in a full genome screen to discover gene inactivations that either enhance transgene silencing or enhance the ability of the miRNAs to repress via the lin-14 3’ UTR. Transgenic animals were exposed to dsRNA corresponding to 16,557 genes and scored for a rolling phenotype. After ruling out suppressors of the rol-6 phenotype, 69 gene inactivations cause silencing of the rol-6::lin-14 3’ UTR transgene. By testing all the hits from the primary screen with a panel of simpler transgenes without the lin-14 UTR, we classified the hits as enhancing RNAi or miRNAs. We detected dozens of gene inactivations that generally increase transgene silencing and RNAi, and a handful of gene inactivations that specifically enhance miRNA silencing. Available viable mutants corresponding to these genes were tested for enhanced RNAi and RNAi-defective phenotypes and many showed defects. Using additional molecular assays for siRNA silencing, identified new genes that act in ERGO-1 pathway: puf-8 RNA binding protein, F29A7.6/phosphoprotein, T26A8.4/CCCH Zn Finger protein, tcer-1/CA150 transcriptional elongation factor, ent-1/nucleoside transporter, mel-32/serine hydroxymethyltransferase which modifies uracil to thymidylate in the nucleus, C25F9.11/novel, B0001.6/novel, F57C9.7/novel, rde-4/RNA helicase and F58G11.2/RNA helicase. It also identified several factors in clathrin-dependent endocytosis that affect silencing. Non-lethal mutant alleles annotated to act in endocytosis showed an enhanced response to dsRNA: dpy-23/AP2 adaptin complex, chc-1/clathrin heavy chain, and rme-4/DENN domain. Our model for endocytosis involvement is that it may resolve a cycle of siRNA activity, so that in the absence of endocytosis, siRNAs remain active longer. These siRNAs could act at the primary step of production from dsRNA templates or at the amplification step by RdRps.
The gene inactivations that only enhanced silencing of a lin-14 3’UTR fusion gene identify negative regulators of microRNA activity. The vig-1/RNA binding protein, ccr-4/ CCR4-NOT transcription complex subunit 6, the P granule component exos-1/ exosomal 3’-5’ exoribonuclease ski4, htz-1/ H2A.Z histone variant, sams-3/4/ S-adenosylmethionine synthetase and F58G11.2/RNA helicase gene inactivations may enhance the silencing potency of miRNAs. Quantitative RT-PCR assays measuring lin-14 mRNA and mir-35 microRNA showed increased lin-14 mRNA levels in vig-1, ccr-4 and npp-12 mutant embryos, and reduced microRNA levels in vig-1, npp-12, and prmt-1 mutants and arl-8(RNAi).
mut-16 and other mutator-class genes modulate 22G and 26G siRNA pathways in Caenorhabditis elegans. (published Zhang. C., TA Montgomery, HW Gabel, SE Fischer, CM Phillips, N Fahlgren, C Sullivan, JC Carrington, and G. Ruvkun. 2011. mut-16 and other mutator-class genes modulate 22G and 26G siRNA pathways in Caenorhabditis elegans. Proc. Natl. Acad. Sci., 2011 108:1201-8.
Mutator-class (mut) genes mediate siRNA-guided repression of transposons as well as exogenous RNAi, but their roles in endogenous RNA silencing pathways are not well understood. To characterize the endogenous small RNAs dependent on mutator-class genes, small RNA populations from a null allele of mut-16 was subjected to deep sequencing. Additionally, each of the mutator-class genes was tested for a requirement in 26G siRNA pathways. The results indicate that mut-16 is an essential factor in multiple endogenous germline and somatic siRNA pathways involving several distinct Argonautes and RNA-dependent RNA polymerases. We also found that the soma-specific mut-16(mg461) mutant allele is present in multiple C. elegans laboratory strains, suggesting a possible advantage to this allele in the laboratory setting.
RDE-10/RDE-11 complex regulates RNAi and endogenous siRNA pathways in Caenorhabditis elegans. Current Biology, in press Zhang. C., TA Montgomery, SE Fischer, CM Phillips, N Fahlgren, C Sullivan, JC Carrington, and G. Ruvkun.
From a forward genetic screen for factors required for RNAi, we identified rde-10 and through proteomic analysis of RDE-10-interacting proteins, we identified a protein complex containing the new RNAi factor RDE-11, the RNAi factor RSD-2 as well as other candidate RNAi factors. The newly identified genes rde-10 and rde-11 encode a novel proteins. Mutations in rde-10 and rde-11 are resistant to low dosage, but sensitive to high dosage of double-stranded RNAs. Exogenous siRNA levels are greatly reduced in rde-10 and rde-11 mutants, indicating that they are required for siRNA formation or stability during RNAi. We assessed the roles of rde-10, rde-11, rsd-2 and rsd-6 in endogenous small RNA pathways using high-throughput RNA sequencing. These genes are required for a subset of secondary siRNAs derived from repeated C. elegans genes. Some of the RDE-10 interacting proteins have been identified as potential regulators of small RNA pathways in our previous RNAi-based genome-wide screens or proteomics; MATH-33 protein is a highly conserved ubiquitin carboxyl-terminal hydrolase and is required for cosuppression, an RNAi-related pathway, CAR-1 is an LSM RNA binding protein and PAB-1 is a multiple RRM poly A binding protein—both are components of the germline P granule, a site of small RNA activity.
Cytological analysis of the RNAi factors mut-2, mut-7, mut-16 and their roles in transposon silencing and RNAi. Carolyn Phillips, Genes and Development in review
In C. elegans, there are six families of transposons (Tc1-Tc5, Tc7). RNAi silences Tc1 in the germline: mutator (mut) class genes that activate transposition in the germline, also cause defects in RNAi. These genes includes a nucleotidyl transferase (mut-2), a 3’-5’ exonuclease (mut-7), a DEAD box RNA helicase (mut-14), a glutamine/asparagine rich protein (mut-16), two proteins of unknown function (rde-2 and mut-15). Functional GFP fusions to mut-7, rde-2, and mut-16 that rescue the corresponding mutants for temperature-sensitive sterility, production of endogenous siRNAs, and response to exogenous RNAi, are each visible throughout the germline and display distinct perinuclear localization. The pattern of localization is reminiscent of germline P-granules, which are nuclear pore-associated RNA-rich cytoplasmic germ granules. P-granules have previously been shown to contain numerous RNA-binding proteins including known components of small RNA pathways (2). MUT-16:GFP localizes to these foci in the absence of each of the other Mutator-class proteins, but the other GFP fusions fail to localize in a mut-16 mutant.
Identification of new small RNA pathway genes from correlated phylogenetic profiles and other genome analyses. Yuval Tabach, Nature submitted.
Hundreds of candidate protein cofactors for small RNA pathways have emerged from our functional genomic screens for defects in small RNA pathways. The Argonaute class proteins, which are central players in RNAi and miRNA, are entirely missing in 9 out of 31 species of Ascomycota, in 1 of 3 species of Basidomycota, in 6 of 14 protist species, and in one of two green algae species, but have not been lost in any of the 33 animal or 6 land plant species compared. The loss of the Argonaute correlates with reported competences to inactivate genes by RNA interference in those species that have been tested. To Identify other genes that share an evolutionary history with validated small RNA pathway proteins, we determined the phylogenetic profiles of all 20,000 proteins encoded by C. elegans genes in 85 disparate animal, fungal, plant and protist genomes. Genes that have coevolved but bear no homology to Argonautes were identified, for example the asparaginase/threonine aspartase/taspase encoded by K01G5.9, implicated in cleavage of the Mixed-Lineage Leukemia/trithorax histone methyltransferase MLL (3), the CAND-1 elongation factor and another elongation factor, THOC-1. THO complex members have emerged from genetic screens for defective transgene and RNAi silencing in Arabidopsis thaliana as well (4).
We used a full genome RNAi screen for gene inactivations that suppress the transgene silencing up-regulation of the eri-1 mutant as an experimental test of these phylogenetic profile predictions. C. elegans tandem array transgenes are silenced in the germline. Such transgenes are also silenced in many somatic cells in an eri-1 mutant. Screening animals grown on E coli strains expressing dsRNA targeting 18,578 genes, we identified 855 genes required for transgene silencing in the eri-1(mg366) background, including a number of previously identified genes such as rde-1 and dcr-1. The potency of transgene desilencing gene inactivations was ranked in four classes. The genes predicted to be RNAi cofactors from the phylogenetic profile analysis are highly enriched on these lists of candidate RNAi factors, including the strongest class of transgene desilencing gene inactivations.
*The Caenorhabditis elegans SOMI-1 zinc finger protein and SWI/SNF promote regulation of development by the mir-84 microRNA. Hayes, G. D. C.G Riedel, and G. Ruvkun. 2011. Genes Dev. 25:2079-92. *
The conserved miRNA let-7 and three paralogs of let-7, mir-84, mir-48 and mir-241, control the timing of stage-specific developmental events. These miRNAs are identical at the 5’ ends, but diverge at the 3’ ends. To identify factors required for the activity of mir-84 and possibly other miRNAs, we screened for mutations that suppress the developmental defects caused by overexpression of mir-84. Mutations in the somi-1 gene prevent these defects without affecting the expression level of mir-84. Loss of somi-1 also causes phenotypes similar to deletion of mir-84, showing that somi-1 is necessary for the normal function of this miRNA. somi-1 encodes a zinc-finger protein that localizes to nuclear foci and binds the promoters of let-60/RAS, lin-14 and lin-28, genes that may be targeted by mir-84 and similar miRNAs. Proteomic and genetic screens identified conserved chromatin-remodeling and homeodomain transcription factor complexes that work with somi-1 to regulate differentiation. Enrichment of SOMI-1 at the promoters of lin-14, lin-28 and let-60/RAS, transcripts of which are targets of let-7 and mir-84, and the requirement for somi-1 for the defects caused by misexpression of mir-84 and let-7, suggest that somi-1 mediates a nuclear response to the activity of these miRNAs.
The mevalonate pathway modulates microRNA activity via protein N-glycosylation. Zhen Shi, PNAS in press
A strong hit from our RNAi screen for miRNA defective gene inactivations encodes C. elegans HMG-CoA synthase, which synthesizes mevalonate, a key intermediate in the isoprenoid pathway, Inactivation of F25B4.6/HMG-CoA synthase by RNAi causes let-7 mutant-like phenotypes that are suppressed by the loss-of-function of validated let-7 target genes. The silencing of lin-14 mRNA translation by the lin-4 miRNA is also dependent on F25B4.6/HMG-CoA synthase: F25B4.6 RNAi causes a two fold derepression of LIN-14 protein levels. This de-silencing of LIN-14 by F25B4.6/HMG-CoA synthase inactivation is due to reduced lin-4 miRNA repression of lin-14 via its 3’ untranslated region. In the lin-14(n355gf) strain which lacks all the miRNA (lin-4 and let-7 and others) complementary sites in the lin-14 3’UTR, LIN-14 protein levels are not further increased when F25B4.6/HMG-CoA synthase is inhibited. The miRNA abundance is unaffected in F25B4.6/HMG-CoA synthase RNAi. These results show that F25B4.6/HMG-CoA synthase is required for the miRISC activity, downstream of miRNA biogenesis/accumulation.
Isoprenoids, the end products of the mevalonate pathway, feed into a wide range of downstream biosynthetic pathways in additional to the better known sterol synthesis: dolichol serves as the lipid carrier of the oligosaccharide moiety for protein N-glycosylation; ubiquinone and heme A function in the electron transport chain; prenylated proteins; and isopentenyl adenine, present in position 37 of tRNAs that read codons starting with U. Addition of mevalonate to plates rescues the let-7-deficient-like phenotypes of F25B4.6/HMG-CoA synthase inactivation, but not inactivation of the miRNA Argonaute alg-1/2 by RNAi. Cholesterol does not rescue the RNAi depletion of F25B4.6/HMG-CoA synthase. This indicates that instead of cholesterol, other biosynthetic product(s) of the mevalonate pathway modulate miRNA activity. HMG-CoA reductase (encoded by F08F8.2) also acts in the miRNA pathway. Homozygous F08F8.2(tm4368) animals arrest at the L1 stage. High concentration of mevalonate rescues the homozygous F08F8.2(tm4368) animals. F08F8.2(tm4368) homozygous mutant animals growing with low mevalonate fail to express the col-19::GFP in the let-7(mg279) mutant background. Statins inhibit HMG-CoA reductase activity. When fluvastatin is added to the growth medium, let-7(mg279 animals fail to express col-19::GFP, similar to the RNAi depletion of F25B4.6/HMG-CoA synthase or the F08F8.2(tm4368) mutation. Thus in C. elegans, the non-sterol output of the mevalonate pathway modulates miRNA activity.
Even though the RNA modification output was the most attractive model for how the mevalonate pathway intersects with miRNA regulation of target mRNAs, mass spectroscopy on nuclease P1 digested C. elegans 18-28nt small RNAs (~30% are miRNAs) did not detect N6-(∆2-isopentenyl)adenosine, whereas we could easily detect it in nuclease P1 treated C. elegans tRNA. We therefore broadened the search to include cherry-picked RNAi clones targeting the genes that function in protein prenylation, tRNA isopentenyltransferase, ubiquinone and heme A biosynthesis, dolichol synthesis and protein N-glycosylation. We performed an RNAi screen for gene inactivations that cause a col-19::GFP expression defect in the let-7(mg279) mutant background. The hits that cause failure of col-19::GFP expression all function in the pathway for protein N-glycosylation. The C. elegans oligosaccharyltransferase (OST) complex that carries out protein N-glycosylation has five subunits: the catalytic subunit T12A2.2/STT-3; and four accessory subunits: T22D1.4/ribophorin I, ostd-1, ostb-1 and dad-1. Knocking down any of these genes by RNAi causes col-19::GFP expression defect in the let-7(mg279) mutant, but not in wild type, suggesting a strong genetic interaction with let-7. Inactivation of the OST complex causes de-repression of miRNA targets, with no effect on the mature miRNA levels. This indicates that the activity, rather than the biogenesis or accumulation of miRNAs, requires the N-glycosylation.
Statins block cholesterogenesis by inhibiting HMG-CoA reductase, the rate-limiting enzyme in the mevalonate pathway and are prescribed to hundreds of millions of patients to lower the risk of heart diseases. While the efficacy of statins to lower cholesterol is the general view of their molecular mechanism, the central role that mevalonate regulation of N-glycosylation to in turn mediate the function of miRNA repression of target mRNAs suggest a miRNA axis of statin treatment of heart disease (5). Significantly, a deletion of the cardiac miRNA mir-208 strongly suppresses the high blood pressure induction of fibrosis caused by the adult stage misexpression of fetal myosin heavy chain gene (6). The mevalonate has also emerged from another miRNA analysis: mir-122 is required for the replication of hepatitus C virus and statins act in the same manner as loss of mir-122 to inhibit hepatitis C replication (7-9). mir-122 mediates the recruitment of HCV replication and associated small RNA proteins such as Argonautes to lipid vesicle replication zones within cells (8). The finding that in miRNA repression of target mRNAs in animal cells engages a vesicle sorting pathway is consistent with a sorting to such organelles in miRNA regulation of target mRNAs (10-12).
RNAs linked to the proteome. The small RNA field has now collected millions of small RNAs, via library construction and high throughput DNA sequencing approaches. We have been in the descriptive phase of small RNA biology for almost a decade: for the endogenous siRNAs, there are thousands of genes represented by antisense small RNAs and millions of distinct siRNAs that tile along mRNAs represented, but no clear function has emerged. Protein cofactors for small RNAs, identified either biochemically in assays for production of siRNAs or miRNAs or piRNAs, or genetically via screens for defective small RNA function have also emerged: For example, Dicer and the proteins of the Argonaute class that mediate the maturation and nuclease guidance activities and more recently, via RNAi screens for protein cofactors of small RNA pathways in our lab and other labs that revealed many other possible components that may sort and guide small RNAs to their targets (1,2).
We believe that the clever small RNA cloning approaches used by the field have systematically missed key small RNAs that are modified at their 5’ and 3’ ends to dramatically change the ability of these RNAs to interact with the proteome and other molecules in the cell. For example, the now established 5’ ligation of dephospho-coenzyme A to RNAs allows such small RNAs to form disulphides covalently with cysteine residues on proteins (or to glutamate or aspartate residues via thioester bonds), to essentially bar code those proteins with small RNA moieties that can base pair with RNA, DNA, or other RNA-tagged proteins. Such modified proteins could then be sorted via these tags, or asssemble onto other nucleic acid scaffolds, such as the kinetochore, or scaffolds that interact with nucleic acids, such as the centrosome or cytoskeleton.
Dephospho coenzyme A has a ligatable 3’ OH. Acetyl coenzymeA as well as the synthetase that generates it is localized to the nucleus for acetylation of histones and other transcription factors. S adenosyl methionine also bearing a 3’ hydroxyl that could ligate to small RNAs. SAM is the methyl donor in a large array of methylation reactions, including methylation of histones, a connection that is not uninteresting with regard to small RNA pathways. The linkage of S adenosyl methionine or S adenosyl homocysteine to the 5’ end of small RNAs could also allow these RNAs to accompany histone methylases to the nucleus and perhaps ligate small RNAs as they transfer a methyl group. NAD. Notice that the 3’ OH of the ATP moiety could be ligated to RNA and has been shown to be ligated (4). The ligation of NAD to the 5’ end of small RNAs would allow them to participate in adenylation reactions of glutamate, aspartate and arginine, leaving the small RNA tagging the target protein. Similarly, poly ADP ribosylation could multiply tag proteins with small RNAs. Interestingly, the poly ADP polymerases pme-1 and pme-2 are the closest of all C. elegans proteins in our phylogenetic profile to the argonautes suggesting that poly ADP modification is correlated with small RNA pathways in animals, plants, fungi and protists.
Support for the idea of ligation of coenzyme A to the 5’ end of small RNAs has emerged from multiple experiments as well. Based on the concept that dephosphocoenzyme A could ligate to RNAs and to then allow them to interact with the cysteine proteome, we searched for incorporation of 35S into RNA preparations from C. elegans and S. pombe. We saw incorporation of 35S in a molecule of about 20 to 25 nt only in C. elegans, that co purifies with RNA in phenol extractions in the trireagent or SDS proteinase K digestion isolation, ethanol precipitation, and binding to RNA purification resins. We could see the expected charged tRNAs for Met and Cys in both C. elegans and S. pombe. However the 35S RNA, which we have named suRNA is not sensitive to various RNAses, suggesting that either it is highly modified or not an RNA. One concern was that sulfated glycosaminoglycans might be 35S-labeled in these preparations, though we did not expect such a narrow size range for such polysaccharide polymers. Treatment with a bacterially derived heparinase degrades the 20 to 25 nt 35S fraction band. However, the annotation of commercially bought heparinase II, isolated from Flavobacteria, is not definitive at all. It is possible that these enzymes are actually involved in disruption of small RNA pathways, especially suRNAs and that they are promiscuous for other sulfated polymers.
Even more strongly supporting the model that modified nucleotides are ligated to small RNAs is mass spectroscopy of E. coli RNA by the Liu group, who reported that succinyl coenzyme A is ligated to RNAs isolated from E. coli and other bacteria (4). Their mass spec analysis revealed a number of mystery peaks that might correspond to other ligations to RNAs, and proved that NAD is added in an analogous manner (5). The focus of their work is to reveal other enzymatic capabilities of RNAs, with major implications for the capacities in the RNA world.
We expect to find that protein elements of the centrosome for example, are ligated to siRNAs corresponding to the centromere elements of C. elegans and other organisms to mediate the binding of particular proteins to these DNA regions. We expect to find that proteins that form multisubunit complexes may use complementary small RNAs to form and disassemble complexes. Ligation of small RNAs to proteins allows the addition of a handle that can be a bar code for assembly of protein complexes and for interaction of proteins with other nucleic acids. In the same way that PCR primers allow us to bar code and preferentially amplify segments in the lab, a small RNA tag on a protein allows that protein to be recognized by other proteins bearing the complementary RNA tag.