Update Trends in Plant Science Vol.13 No.7

describes the emerging link between small-RNA-mediatedheterochromatin formation and the enigmatic epigeneticphenomenon of paramutation. Future research willaddress the environmental sensitivity of transposon con-trol by small RNAs, and the extent to which this influencesgene expression in a way that can reprogram plant de-velopment and physiological state in an adaptive manner.

Although the field of ncRNAs has come a long way in ashort time, there remainmany unanswered questions evenin the comparatively well-studied area of small RNAs, asdiscussed byGaryRuvkun in a Perspective Essay followingthis editorial. One such question is the general validity ofthe ARGONAUTE and PIWI protein structures, whichform the basis of our current mechanistic models. Thesestructures are derived from information gained fromorganisms such as archae and bacteria, which are notknown to haveRNA silencing. Ruvkun presents argumentsthat these proteins might instead function in DNA clea-vage in these species, emphasizing the need to obtain morerelevant structures. Looking towards the future, Ruvkun

Corresponding author: Ruvkun, G. (ruvkun@molbio.mgh.harvard.edu).

suggests that it is still uncertain that current technologiesare really sampling the entire spectrum of small RNAmolecules in the cell, as emphasized by the discovery ofpiRNAs. Technologies that are insensitive to specific bio-chemical modifications and approaches based on associ-ations with putative RNA-processing proteins are likely toreveal new undiscovered classes of RNAs for which func-tions will need to be ascribed. Ruvkun points out thatresearch in plants and Caenorhabditis worms has beenat the frontier of the ncRNA field, and there is every reasonto believe that this will continue into the future as the fieldkeeps up its rapid expansion.

We hope that you enjoy reading these articles as muchas we have.

The Guest Editors

Reference1 Ruvkun, G. (2001) Glimpses of a tiny RNA world. Science 294, 797–799

1360-1385/$ – see front matter � 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.tplants.2008.05.003 Available online 16 June 2008

Perspective essay – Special Issue: Noncoding

and small RNAs

Tiny RNA: Where do we come from?What are we? Where are we going?Gary Ruvkun

Department of Molecular Biology, Massachusetts General Hospital, and Department of Genetics, Harvard Medical School Boston,

MA 02114, USA

The tiny RNA field has been a tornado for the past decade,with miRNAs and other small RNAs emerging from theirobscurity and RNAi exploding onto the scene. Due to thepower of deep sequencing, we have the most completedescription of the small RNAs that are produced inanimals, fungi, plants, and even protozoa. But in mostcases this does not immediately reveal much about theirfunctions, except in plants, where the near-perfect com-plementarity of mRNAs to tiny RNAs has allowed a muchmore precise mapping of their functions [1]. From the veryfirst identification of miRNAs in plants, they mapped ontothe canon of the field: their targets corresponded to some ofthe previously identified regulatory mutations in tran-scription factors that regulate flower and leaf patterning[1]. The plant miRNA surveys have been jaw-droppinglybeautiful to geneticists because they instantly reveal agenetic pathway from the miRNA to its easily identifiabletarget mRNA. No other type of gene identification soaccurately produces an outline of a pathway in one step.In this way, the plant miRNA field has fulfilled the falsepromise that wasmade years ago for transcription factors –that a code would be discovered connecting all transcrip-tion factors with their target genes, revealing the genetic

cascades of biology. The small number of base pairs actu-ally recognized by each transcription factor, the degener-acy in transcription factor binding site sequences, and thecombinatorial binding of transcription factors have madethis decoding anything but simple. On the other hand, thelong term employment of the transcription community isassured.

The mapping of miRNAs to clearly important targetmRNA functions is only now beginning in animals. It is atestament to how a few bits of extra information, thenear-perfect duplexes of plant miRNA::mRNA comple-mentarity versus the bulged and loopy animal miR-NA::mRNA complementarity, can make the differencebetween pathway assignment and suggestive evidencefor function.

And the most recent findings from plants loop back tothe heterochronic mutations that revealed small RNAs inthe first place in animals. Plant miRNAs have recentlybeen identified that mediate the switch from juvenile toadult leaf and branch patterns ([2], Scott Poethig, pers.commun.). There is even evidence that miRNA changesmay have been important in the domestication of crops. Forexample the Corngrass1mutation of maize is an activationof an miRNA by a transposable element, and a validatedtarget of thismiRNA is one of the fewgene variations known

313

Update Trends in Plant Science Vol.13 No.7

to mediate the domestication of teosinte to maize. Thisreveals the power of miRNAs in the variation under dom-estication, andmay be the sort ofmacromutation picked outby impatient plant breeders in antiquity [3,4]. One wonderstoo about the domestication of Homo sapiens.

A taxonomy of small RNAs has begunAs impressive as the genomic exploration of small RNAshas been, we are still only viewing the tip of the iceberg.The powerful cloning methods using in vitro RNA selectionand ligation techniques, and the amazing production capa-bilities of deep sequencing metholologies reveal a rathercomplete survey of this one aspect of the small RNA world.But they are like one kind of telescope observing at onewavelength: in this case, mainly studying those smallRNAs with 50 monophosphates for example, the likelyproducts of DICER. But one of the major findings of thepast few years, the piRNAs associated with the PIWIsubtype of PIWI proteins, are generated by a process thatdoes not depend onDICER [5]. So a universe of small RNAsthat might be generated by other nucleases was suddenlysuggested. Not all of those other pathways might generate50 phosphates. Thus the RNA ligation strategy, howeverclever and useful, might reveal only a subset of smallRNAs. The current deep sequencing approaches now allowless selective approaches for other classes of small RNAs tobe discovered.

But other modifications of small RNAs may haunt thesebiochemical surveys: methylation of 30 hydroxyls are onetype of modification that than plague ligation reactions.Other modifications of RNA are well known from 50 yearsof studying ribosomal and tRNAs, including thioylation,glycosylation, acetylation, and other more obscure decora-tions. If any of these modifications preclude RNA ligase orreverse transcriptase from recognizing these modifiedRNAs, they will be systematically missed by such bio-chemical studies. On the other hand, the ability of suchmodified RNAs to base pair to DNA or RNA is likely to beintact, given the informational functions of small RNAs, sobiochemical purifications that depend on base pairing foranalysis may detect such outliers. And genome sequencesignatures of small RNAs, their conservation in relatedspecies and their secondary structures, will continue toidentify candidates. In this regard, the plant genome withits more perfect complementarity to small RNAs may beinformatically powerful in the identification of small RNAsby their complementarity to conserved sequences inmRNAs.

A taxonomy of small RNA protein cofactors is at aneven earlier stageThe biochemical analysis of the protein cofactors thatmediate the processing of miRNAs and siRNAs proceededvery quickly after their existence was inferred from theRNAi and miRNA genetic phenomonology. The bio-chemical study of the translational control aspects ofanimal miRNAs has not proceeded as dramatically, mostlikely because it intersects the ribosome, which has acomplex set of associated proteins, and is therefore difficultto disentangle from other translational controls. But satur-ation genetic analysis of miRNA [6] and other small RNA

314

pathways [7] has been done and some and perhaps evenmany of the factors revealed by these studieswill turn out tomediate other steps in the process. As in any complexprocess, there may be many steps downstream of DICERand ARGONAUTE that interpret these small RNAs, andthere may be paralogous pathways of other small RNAsmediated by paralogs of the miRNA and siRNA cofactors,just as an ARGONAUTEparalogmediates piRNA function.Most of the field has been a bit myopic towards DICER orPIWI proteins because they are at the center of the pro-duction and first activities of some small RNAs. But there isincreasing evidence of a cell biology of RNA regulation,specifically in P bodies and P granules, so a more complexchoreography is likely. It may be significant in this regardthat cytoskeletal elements have emerged as strong candi-dates for activity in the miRNA pathway [6]. Interestingly,RNA has been shown to associate with the cytoskeleton[8].

As much as the field has tended to focus its biochemicallens on DICER and PIWI proteins, this has been a veryproductive focus: for example the piRNAs emerged mostdirectly from simple immunoprecipitations of the PIWIsubtype of PIWI proteins [7[. Similarly miRNAs interactwith one Argonaute subclass from animals. A good case canbe made for all of the dozens of possible small RNA path-way proteins to be similarly analysed for small RNAbinding. In this way, particular sizes and sequence classesof small RNAs can be mapped to particular protein cofac-tors. This would both endorse the small RNA and thecandidate protein cofactor by virtue of its interaction.

Side trips to the structural biology of very distantcousins of Argonaute proteinsThe interpretation of the structures of the Argonauteproteins has been overly enthusiastic, given that no eukar-yotic protein structures have been solved. One would haveexpected that the crystallographers would have targetedfor structural work the PIWI domain proteins with wellannotated functions, for example the C. elegans RNAifactor RDE-1 or the miRNA or siRNA cofactors of theARGONAUTE subtype, or the germline piRNA cofactorsof the PIWI subtype. But because of technical difficulties inobtaining good crystals from these proteins, andpresumably because the tiny RNA field has been so inten-sely competitive, three groups came up with strange dis-tant relatives of the eukaryotic PIWI proteins for theirstructure determinations: one protein from the archaeaPyrococcus furiosus [9], another from the archaea Archae-oglobus fulgidus [10] and another from the deeply diver-gent bacterium, Aquifex aeolicus [11], which is thought toderive about 10% of its genes from the Archaea. Theseproteins chosen for crystallographic studies are very dis-tant relatives of the eukaryotic PIWI PAZ proteins as wellas distantly related to each other: when I use either PIWIsubtype or ARGONAUTE subtype eukaryotic PIWIproteins as a query in Blast analysis of all archaeal orbacterial genome sequences, the proteins whose structureswere solved score just barely above background, and atleast 50 logs lower e-value than for example the PIWI /PAZorthologues or paralogues in animals or plants or fungi orprotozoa. And when the Pyrococcus, Archaeoglobus, or

Update Trends in Plant Science Vol.13 No.7

Aquifex so-called PIWI proteins are Blast compared to theentire database of proteins, they are only distantly relatedto each other and detect only a handful of very distantbacterial or archaeal proteins in the entire database. Andthese solved proteins detect the eukaryotic PIWI proteinswith an e value of e-5 or less; this level of homology wouldnot be interpreted as significant bymost homologymavens.The lack of PIWI proteins in the Archaea and Bacteria issupported by their lack of DICER as well.

So it has been surprising that the structure determi-nation of these Pyrococcus, Archaeoglobus, and Aquifexdistant relatives of PIWI proteins has been successfullyused to predict structural features of ARGONAUTE andPIWI proteins. In fact, the key catalytic residues from theArchaeal structures were used to successfully predictthe same for the eukaryotic PIWI proteins, showing thatthe structures are likely to be predictive, even with the lowamino acid sequence homology. However, the finding thatthe archaeal PIWI proteins bind DNA better than RNAsuggest that these distant relatives may have a functionrelated to DNA modification rather than RNA cleavage ortranslational control [11].

The view that these genes mediate DNA functions isendorsed by the operon structure of these archaeal genes.Archaea and Bacteria have the wonderful annotation toolof genes in operons, so that one can infer genetic pathwayfunction if other genes in the same operon have moreconvincing annotation. In the case of the Pyrococcus fur-iosus gene distantly related to PIWI, it is not in an operon.But when I searched with this gene as a query, the top hit,scoring at e–58, or 50 logs better homology than eukaryoticPIWI proteins, is open reading frame MJ1321 in thearchaean Methanococcus jannaschii. This is clearly anorthologue, but the only one that can be detected in thedatabase so far. MJ1321 is in an operon with four othergenes, annotated to be 50MJ1323, a homologue of the DNArepair protein RAD32, MJ1322, a homologue of the DNArepair exonuclease SbcC, MJ1321, the distant relative ofthe eukaryotic PIWI proteins, andMJ1320, no homology 30.Similarly, the Aquifex distant PIWI protein is located in anoperon with DNA repair annotion: 50 aq 1449, with an 8-oxo-dGTPase domain orMutT domain, aq 1447, the distantPIWI relative, aq 1446, with no protein homology, aq1445,a copper transporting ATPase, and aq1444, no homology. Inthe case of the Archaeoglobus fulgidus PIWI relativeAF1318, it is in an operon with AF1317, but this gene hasno homology.

One interpretation of these operon structures isthat these very distant PIWI proteins mediate DNA modi-fication functions ancestrally related to the RNA cleavageand modification functions of eukaryotic PIWI proteins.However the very sparse number of these PIWI relativesin the archaeal and bacterial genome databases suggeststhat the functions mediated by the solved archaeal andbacterial PIWIproteins are not a general feature of archaealor bacterial life. Possibly they represent the very rarelineages that have received a horizontal gene transfer fromeukaryotes. Alternatively, they may represent the last ves-tiges in the archaeal and bacterial clades that have not yetdeleted a DNA modification gene function still required inthe eukaryotes.

To the nucleusThe suggested DNAmodification function for these distantPIWI proteins has implications for the function of theARGONAUTE proteins from eukaryotes. The DNA modi-fication function may also be maintained in the eukaryoticPIWI proteins. They have been observed in both thenucleus and the cytoplasm [12], suggesting that theseproteins could do more than just catch RNA transcriptsas they emerge from DNA duplexes; they may sling themback at the DNA tomediate modifications such as the DNAmethylations, noted in silenced plant genes [13].

Of course it would be grand to see a structure of theactual eukaryotic PIWI proteins. While it is difficult tointerpret why the eukaryotic PIWI proteins have been sorecalcitrant to crystallography, one can speculate thatthere has been a systematic missing component in thecocktails used for crystallization. It could be a missingprotein cofactor, but it also could be amissing RNA cofactorthe absence of which allows many possible conformationsof PIWI proteins, disabling or disordering crystallization.

The hints of DNA repair function for Argonaute proteinsfrom the PIWI operon analysis in the Archaea andBacteriaare strongly endorsed by the function of PIWI proteins andrelated RNAi factors in DNA elimination in the protists[11]. Similarly, the ARGONAUTE and DICER proteinshave been implicated in chromatin silencing in the fungi,plants and animals [14]. But it is surprising how littleintersection of small RNAs with chromatin factors hasbeen documented, now more than 5 years after the S.pombe work showed that small RNAs regulate heterochro-matinization. I would have expected dozens of chromatinfactors, especially the known epigenetic remodeling factorslike the polycomb group genes to have been shown todirectly bind small RNAs. One explanation might be thatthat a simple interaction between small RNAs and chro-matin factors is not taking place. The S. pombe work doessuggest elaborate protein machines that directly interactwith PIWI proteins mediate the heterochromatin changesby small RNAs [15].

Why have worms and plants driven the small RNArevolution?Of course, we would like to believe it is our intellectualsuperiority. Good taste in science is perhaps a more defen-sible position. But perhaps the truth lies in a biologicalunity in these disparate taxons. The hallmark distinctionof worms in the animal kingdom is that it is just a dozen orso cell divisions from a totipotent fertilized egg to a fullyformed adult. So the heterochronic mutations that firstrevealed the tiny RNA world in worms mediate multi-potent vs committed cell fate determinations at these veryfirst steps in the reprogramming of a cell from totipotency.The cell lineage analysis that was the great advance inearly C. elegans genetic analysis was the lens throughwhich these cell fate commitments were viewed, and theless than 1000 cells of a worm allowed such major trans-formations in cell fate to be viable enough to survive forthis analysis. Similarly, the major organs of a plant derivefrom totipotent shoot apical meristem tissues and arejust a few cell divisions away from this totipotency, somutant plants defective in small RNA pathways of cell

315

Update Trends in Plant Science Vol.13 No.7

commitment could be interpreted. It is not a coincidencethat the ARGONAUTE and DICER pathways emergedfirst from plant floral and leaf patterning genetics [16,17].

It is not surprising that RNAi was also discovered inplants andworms: gene silencing bydsRNA is somuchmoreintense in plants and worm, due to the amplification ofsiRNAs by RNA dependent RNA polymerases andthe systemic spread of these small RNAs. But why do theseclades use RNAi so much more intensely than others? Oneview is that the onslaught of viruses is somehow a largerproblem for these soil dwelling taxa. But, even thoughRNAipathways do intersect viral resistance, viruses do not seemtobemore of a plant orwormproblem thanother organisms.Another view is that small RNAs are spread systemically inplants and worms as a form of hormonal signaling, eitheraboutviral statusorother informational onslaughts thataremore endemic in plants and worms. The systemic spread ofsiRNAs in plants and worms does suggest that other smallRNAs constitute signaling molecules. In fact, the Teopodmutations in plants, which are so similar to the maizemiRNA mutation, have been shown to be non-autonomous[4], althoughmostmiRNAs in bothplants andanimals seemtoact in the cellswhere theyareexpressed [18].Even thoughit may not be miRNAs that constitute an intercellularsignaling superhighway in plants and worms, other as yetunidentified tiny RNA signaling packets may reveal them-selves as our genetic, informatic, and biochemical analysesbecome more sophisticated.

AcknowledgementsI thank Scott Poethig and Venkatesan Sundaresan for discussing theseideas with me and pointing me in a number of directions that I would nothave found on my own.

References1 Baulcombe, D. (2004) RNA silencing in plants. Nature 431, 356–3632 Wu, G. and Poethig, R.S. (2006) Temporal regulation of shoot

development in Arabidopsis thaliana by miR156 and its targetSPL3. Development 133, 3539–3547

316

3 Chuck, G. et al. (2007) The heterochronic maize mutant Corngrass1results from overexpression of a tandem microRNA. Nat. Genet. 39,544–549

4 Dudley, M. and Poethig, R.S. (1993) The heterochronic Teopod1 andTeopod2 mutations of maize are expressed non-cell-autonomously.Genetics 133, 389–399

5 Klattenhoff, C. and Theurkauf, W. (2008) Biogenesis and germlinefunctions of piRNAs. Development 135, 3–9

6 Parry, D.H. et al. (2007) A whole-genome RNAi Screen for C. elegansmiRNA pathway genes. Curr. Biol. 17, 2013–2022

7 Kim, J.K. et al. (2005) Functional genomic analysis of RNA interferencein C. elegans. Science 308, 1164–1167

8 Blower, M.D. et al. (2007) Genome-wide analysis demonstratesconserved localization of messenger RNAs to mitotic microtubules.J. Cell Biol. 179, 1365–1373

9 Song, J.J. et al. (2004) Crystal structure of Argonaute andits implications for RISC slicer activity. Science 305, 1434–1437

10 Parker, J.S. et al. (2004) Crystal structure of a PIWI protein suggestsmechanisms for siRNA recognition and slicer activity. EMBO J. 23,4727–4737

11 Yuan, Y.R. et al. (2005) Crystal structure of A. aeolicusargonaute, a site-specific DNA-guided endoribonuclease, providesinsights into RISC-mediated mRNA cleavage. Mol. Cell 19, 405–419

12 Mochizuki, K. et al. (2002) Analysis of a piwi-related gene implicatessmall RNAs in genome rearrangement in tetrahymena. Cell 110,689–699

13 Qi, Y. et al. (2006) Distinct catalytic and non-catalytic roles ofARGONAUTE4 in RNA-directed DNA methylation. Nature 443,1008–1012

14 Pal-Bhadra, M. et al. (2002) RNAi related mechanisms affect bothtranscriptional and posttranscriptional transgene silencing inDrosophila. Mol. Cell 9, 315–327

15 Motamedi, M.R. et al. (2004) Two RNAi complexes, RITS and RDRC,physically interact and localize to noncoding centromeric RNAs. Cell119, 789–802

16 Jacobsen, S.E. et al. (1999) Disruption of an RNA helicase/RNAse IIIgene in Arabidopsis causes unregulated cell division in floralmeristems. Development 126, 5231–5243

17 Bohmert, K. et al. (1998) AGO1 defines a novel locus of Arabidopsiscontrolling leaf development. EMBO J. 17, 170–180

18 Parizotto, E.A. et al. (2004) In vivo investigation of the transcription,processing, endonucleolytic activity and functional relevance of thespatial distribution of a plant miRNA. Genes Dev. 18, 2237–2242

1360-1385/$ – see front matter � 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.tplants.2008.05.005 Available online 16 June 2008