Upload
ngonhan
View
221
Download
3
Embed Size (px)
Citation preview
Research Collection
Doctoral Thesis
Structural and functional characterization of proteins involved inRNA-mediated gene silencing
Author(s): Lingel, Andreas
Publication Date: 2006
Permanent Link: https://doi.org/10.3929/ethz-a-005164730
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
DISS. ETH NO. 16441
Structural and functional characterization of proteins involved in RNA-mediated gene silencing
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH
for the degree of Doctor of Sciences
presented by
ANDREAS LINGEL Dipl. Natw. ETH
born 08.02.1976 citizen of Germany
accepted on the recommendation of
Prof. Dr. Ulrike Kutay Prof. Dr. Frédéric Allain
Dr. Elisa Izaurralde
2006
Previous page: In Greek mythology, the Argonauts are a group of sailors, named after their ship, the Argo. Under the leadership of Jason, they start an expedition to find and bring back the Golden Fleece, a treasury guarded by a dragon and located in Colchis on the Black Sea. With the help of the gods, they are able to finish their mission successfully and return to Greece. The illustration shows a detail of an Athenian red-figure clay vase, dated about 425-375 BC. It shows the Argonauts on board their ship Argo, with the NMR ensemble of the Drosophila Argonaute2 PAZ domain in the foreground. The vase is located in the Museo Jatta in Ruvo, Italy.
This thesis describes work carried out in the laboratory of Dr. Elisa Izaurralde at the European Molecular Biology Laboratory (EMBL) in Heidelberg, Germany under the supervision of Prof. Dr. Ulrike Kutay (Institute of Biochemistry at the Swiss Federal Institute of Technology Zurich). The work was completed between May 2002 and December 2005 and was supported by an EMBL predoctoral fellowship. The following publications are presented in this thesis: 1. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2003) Structure and nucleic acid
binding of the Drosophila Argonaute2 PAZ domain. Nature, 426, 465-469. 2. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2004) Nucleic acid 3’-end
recognition by the Argonaute2 PAZ domain. Nat Struct Mol Biol, 11, 576-577. 3. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2004) NMR assignment of the
Drosophila Argonaute2 PAZ domain. J Biomol NMR, 29, 421-432. 4. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2005) The structure of the FHV
B2 protein, a viral suppressor of RNAi, reveals a novel mode of dsRNA recognition. Embo Rep, 6, 1149-1155.
In addition, a perspective and a review are presented: 5. Lingel, A., and Izaurralde, E. (2004) RNAi: Finding the elusive endonuclease. RNA, 10,
1675-1679. 6. Lingel, A., and Sattler, M. (2005) Novel modes of protein-RNA recognition in the
RNAi pathway. Curr Opin Struct Biol, 15, 107-115.
Acknowledgements First of all, I want to thank my supervisor Dr. Elisa Izaurralde for giving me the opportunity to work in her lab. I would like to thank her for excellent scientific guidance and all her enthusiasm, for the great help and for the constant support throughout the last years. This was all invaluable for the success of this Ph.D. thesis. Furthermore, she enabled my long-term collaboration with the laboratory of biomolecular NMR at EMBL. Especially, I want to thank Dr. Michael Sattler for the possibility to do this collaborative work with his lab. He and Dr. Bernd Simon had essential contributions to my projects and taught me NMR spectroscopy. Without them, this work would not have been possible. I am very grateful to Prof. Dr. Ulrike Kutay for the supervision of this thesis at ETH Zurich. Furthermore, I thank Prof. Dr. Frédéric Allain for being my ‘co-referent’ and Prof. Dr. Nikolaus Amrhein for his help and for chairing the examination committee at ETH Zurich. I would like to thank the members of my thesis advisory committee at EMBL for their interest in my work, for fruitful discussions and for their helpful advice. In addition to Dr. Elisa Izaurralde and Dr. Michael Sattler, these were Dr. Andreas Ladurner, and Prof. Dr. Ulrike Kutay. A special “thank you” goes to Michaela Rode and Gunter Stier. Both of them provided excellent technical support and very useful advice throughout the time of my thesis. Next, I want to thank the past and present members of the lab for being nice colleagues and friends, and for enjoyable times inside and outside of the lab. These are: Andrea Herold, David Gatfield, Leonie Unterholzner, Daniel Forler, Kerstin Gari, Jan Rehwinkel, Maria Polycarpou-Schwarz, Tamás Orban, Pavel Natalin, Isabelle Behm-Ansmant, Valerie Hilgers, Schu-Fee Yang, Ana Sofia Bregieiro Eulalio, Suheda Erener, and Foteini Christodoulou. In addition, I would like to thank the members of the NMR lab: Fabian Filipp, Christian Edlich, Cameron Mackereth, Anna Oddone, Lorenzo Corsini, Ana Messias, Malgorzata Duszczyk, Frank Gabel, and Alexander Gasch. I also want to thank other people at EMBL who contributed to my work in one way or another. These are Elena Conti, Sébastien Fribourg, Atlanta Cook, Jérome Basquin, all members of the photolab and of other service units. I am very thankful to Elisa, Sufe, Jan, Pavel, Michael and Bernd for reading this thesis and for helpful comments. Mein ganz besonderer Dank gilt meiner Familie. Meine Eltern und meine Schwester Barbara haben mich durchgehend unterstützt und haben mir immer wieder Kraft gegeben. Ohne sie wäre das alles nicht möglich gewesen. Vielen Dank!
Table of Contents
Table of Contents Zusammenfassung .................................................................................................................... 3 Summary ................................................................................................................................... 5 1. Introduction ...................................................................................................................... 7
1.1 Regulation of gene expression by RNA silencing ..................................................... 7 1.1.1 A brief history of the discovery of RNA silencing ............................................ 7 1.1.2 RNA interference ............................................................................................... 9 1.1.3 miRNA pathway............................................................................................... 13 1.1.4 RNA-mediated transcriptional gene silencing ................................................. 16
1.2 Argonaute proteins ................................................................................................... 18
1.2.1 Argonaute proteins: a conserved protein family with diverse functions.......... 18 1.2.2 The Drosophila Ago1 and Ago2 proteins ........................................................ 21 1.2.3 Drosophila and human Argonaute2 act as Slicer............................................. 22 1.2.4 Structures of archaeal Argonaute and Piwi proteins ........................................ 23
1.3 Viral suppression of RNA silencing......................................................................... 30
1.4 NMR spectroscopy: a tool to understand the function of biomolecules .................. 36
1.4.1 The basic principle of nuclear magnetic resonance ......................................... 36 1.4.2 The chemical shift ............................................................................................ 39 1.4.3 The basic NMR experiment ............................................................................. 40 1.4.4 The interaction of spins through chemical bonds: scalar coupling .................. 41 1.4.5 Resonance assignment in proteins.................................................................... 43 1.4.6 The interaction of spins through space: dipolar coupling and residual dipolar coupling .................................................................................. 45 1.4.7 The nuclear Overhauser effect ......................................................................... 47 1.4.8 Relaxation and dynamics ................................................................................. 48 1.4.9 Structure determination with NMR derived constraints .................................. 52 1.4.10 Interaction studies by NMR spectroscopy ....................................................... 54
2. Results ............................................................................................................................. 57
2.1 Structure and nucleic acid binding of the Drosophila Argonaute2 PAZ domain .... 58
2.2 Nucleic acid 3’-end recognition by the Argonaute2 PAZ domain........................... 71
2.3 NMR assignment of the Drosophila Argonaute2 PAZ domain............................... 88
2.4 The structure of the FHV B2 protein, a viral suppressor of RNAi, reveals a novel mode of dsRNA recognition ........................................................... 91
1
Table of Contents
3. Discussion ...................................................................................................................... 103
3.1 The PAZ domain represents a novel nucleic acid binding fold ............................. 103
3.2 The PAZ domain as a specificity providing module in the RNAi pathway........... 110
3.3 The FHV B2 protein binds dsRNA in a novel mode ............................................. 113
3.4 Viral suppressors of RNAi act at different levels in the RNAi pathway ............... 117 4. Additional Publications................................................................................................ 119
4.1 RNAi: Finding the elusive endonuclease ............................................................... 120
4.2 Novel modes of protein-RNA recognition in the RNAi pathway.......................... 126 5. Abbreviations................................................................................................................ 136 6. References ..................................................................................................................... 138 7. Curriculum vitae .......................................................................................................... 148
2
Zusammenfassung
Zusammenfassung Um den komplexen Prozess der Genexpression regulieren zu können, hat die Zelle ein breites
Repertoire an Kontrollmechanismen entwickelt. RNA-Interferenz (RNAi) ist ein solcher
Mechanismus, der bei Anwesenheit von doppelsträngiger RNA in den Zellen die
Genexpression herunterreguliert. Die doppelsträngige RNA wird dabei von Dicer, einer
Ribonuklease der Klasse III, in „small interfering RNAs (siRNAs)“ gespalten. Diese sind 21-
23 Nukleotide lang und bilden einen ca. 19 Basenpaare umfassenden Duplex, der auf beiden
Seiten am 3’-Ende einen Überhang von zwei Nukleotiden aufweist. In einem zweiten Schritt
werden die siRNAs in den multimeren „RNA-induced silencing complex (RISC)“ überführt,
in dem sie mit der Hilfe eines Argonaute-Proteins die Aufgabe haben, eine komplementäre
Boten-RNA zu erkennen und zu spalten. Die Mitglieder der konservierten Argonaute-
Proteinfamilie zeichnen sich durch die Anwesenheit von zwei charakteristischen Domänen,
einer zentralen PAZ-Domäne und einer C-terminalen Piwi-Domäne, aus.
Zu Beginn dieser Arbeit war bereits bekannt, dass die PAZ-Domäne nur in Dicer- und
Argonaute-Proteinen vorkommt. Darüber hinaus konnte die Interaktion zwischen Mitgliedern
beider Proteinfamilien gezeigt werden und deshalb wurde vorgeschlagen, dass die PAZ-
Domäne als Protein-Protein-Bindungsdomäne den direkten Kontakt zwischen den beiden
Proteinen vermittelt.
Der Hauptgesichtspunkt dieser Arbeit war die Charakterisierung der Funktion der PAZ-
Domäne. In einem ersten Schritt habe ich mit Hilfe der Kernspinresonanz (NMR)-
Spektroskopie die dreidimensionale Struktur der PAZ-Domäne des Argonaute2-Proteins von
Drosophila melanogaster gelöst (Kapitel 2.1). Die Domäne hat eine bisher unbekannte
Proteinfaltung, die aus einem zentralen, fünfsträngigen β-Faltblatt besteht, das auf der einen
Seite von zwei α-Helices und auf der anderen von einem konservierten β-Schleife/α-Helix-
Modul flankiert wird. Basierend auf der Struktur konnte ich durch weitere Experimente
zeigen, dass die PAZ-Domäne in vitro an Nukleinsäuren bindet; ein Ergebnis, das im Bezug
auf die vorgeschlagene Funktion erstaunlich war. Eine genauere Analyse ergab, dass die
PAZ-Domäne dabei sowohl an einzelsträngige RNA als auch an doppelsträngige RNA, die
einen zwei Nukleotide langen Überhang am 3’-Ende hat, bindet. Dies deutete bereits auf eine
Spezifität der PAZ-Domäne für einzelsträngige Nukleinsäuren hin. Konservierte
Aminosäuren in der Bindungstasche ließen darauf schließen, dass alle PAZ-Domänen
3
Zusammenfassung
Nukleinsäuren binden können, was ich tatsächlich für mehrere PAZ-Domänen von
Argonaute-Proteinen von Drosophila und des Menschen zeigen konnte.
Um die beobachtete Bindungsspezifität der PAZ-Domäne besser verstehen zu können, habe
ich im weiteren Verlauf der Untersuchungen die Strukturen der PAZ-Domäne des
Argonaute2-Proteins im Komplex mit DNA- und RNA-Oligonukleotiden gelöst (Kapitel 2.2).
Dadurch wurde deutlich, dass nur die zwei letzten Nukleotide am 3’-Ende gebunden werden
und dass die 3’-OH Gruppe des letzten Nukleotids tief innerhalb der Bindungstasche sitzt.
Das gleiche Resultat ergab die Charakterisierung der Bindung eines weiteren RNA-Liganden,
der wie siRNAs einen doppelsträngigen Teil und einen Überhang am 3’-Ende besaß. Dieses
Ergebnis ermöglichte eine Erklärung der beobachteten Bindungsspezifität der PAZ-Domäne
und die Definition der molekulare Funktion als die der 3’-Erkennung. Somit kann man
schlussfolgern, dass die PAZ-Domäne zur spezifischen Erkennung von siRNAs, die von
Dicer erzeugt wurden, beitragen kann und deren Einbau in den RNAi-Effektorkomplex
unterstützt.
RNAi ist ein Mechanismus, mit dessen Hilfe sich Zellen gegen virale Infektionen schützen
können und der bei der Anwesenheit fremder, doppelsträngiger RNA aktiviert wird. Im
zweiten Teil dieser Arbeit habe ich die Struktur und die Funktion eines viralen Proteins, von
dem gezeigt wurde, dass es den herunterregulierenden Effekt von RNAi in Insektenzellen, in
Pflanzen und in Caenorhabditis elegans außer Kraft setzen kann, charakterisiert. Eine
Vielzahl dieser viralen Faktoren ist beschrieben worden, allerdings ist bisher sehr wenig über
deren molekularen Wirkungsmechanismus bekannt. Das untersuchte Protein war das B2-
Protein des Flock House Virus (FHV), und meine Arbeit resultierte in der Aufklärung dessen
dreidimensionaler Struktur (Kapitel 2.4). Das B2-Protein bildet ein elongiertes Dimer, in dem
jedes Monomer aus drei α-Helices aufgebaut ist. Durch NMR-Experimente zur Untersuchung
der Bindung einer siRNA an das B2-Protein konnte ich zeigen, dass auf der einen Seite des
Dimers zwei α-Helices - je eine von einem Monomer - eine Interaktionsfläche mit der RNA
bilden. Die Anordnung von basischen Seitenketten legt nahe, dass die RNA überwiegend
durch Kontakte zum Phosphodiester-Ribose-Rückgrat gebunden wird, wodurch die fehlende
Sequenzspezifität erklärt wird. Da sowohl siRNAs als auch lange doppelsträngige RNAs von
B2 gebunden werden, läßt sich die Inhibition von RNAi auf zweierlei Weise erklären: zum
einen kann das B2-Protein von FHV an seine eigene, virale RNA binden und dadurch eine
Spaltung durch Dicer verhindern, und zum anderen kann B2 an siRNAs binden und so deren
Einbau in den RISC-Komplex unterbinden.
4
Summary
Summary RNA interference (RNAi) is an evolutionary conserved mechanism that regulates gene
expression in response to the presence of double-stranded RNA (dsRNA) in the cell. The
dsRNA is first cleaved by the ribonuclease (RNase) III-type enzyme Dicer into 21-23
nucleotide (nt) small interfering RNA duplexes (siRNAs) with 2 nt 3’-overhangs. In a
subsequent step, the siRNAs are incorporated into a multimeric RNA-induced silencing
complex (RISC), where they guide the selection of a complementary mRNA. Argonaute
protein family members are core components of RISCs and characterized by two conserved
domains, namely the central PAZ domain and the C-terminal Piwi domain. The Piwi domain
has endonucleolytic activity and mediates cleavage of the target mRNA in a region with
sequence complementarity to the siRNA guide. The expression of the targeted gene is thereby
strongly down-regulated.
At the time when this study was initiated, the PAZ domain had been described to be present
exclusively in two protein families, namely the Dicer and Argonaute protein family. As
members of these families had been shown to interact with each other, it was proposed that
the PAZ domain might serve as a protein-protein interaction domain, mediating physical
contact between Dicer and Argonaute proteins.
The main focus of the work described in this thesis was to reveal the function of the PAZ
domain on a molecular level. To achieve this, I first solved the three dimensional nuclear
magnetic resonance (NMR) structure of the free Drosophila melanogaster Ago2 PAZ domain
(chapter 2.1). This showed that the PAZ domain adopts a novel fold composed of a central
five-stranded β-barrel flanked by two α-helices and a conserved β-hairpin/α-helix insertion.
Based on the structure, I performed further experiments that showed that the PAZ domain
binds nucleic acids in vitro, which was an unexpected result. I could demonstrate that the
PAZ domain binds single-stranded RNAs and double-stranded RNAs with 2 nt single-
stranded 3’-overhangs, but has reduced affinity for blunt-ended double-stranded RNA
molecules. This pointed towards a specificity of the PAZ domain for single-stranded nucleic
acids. The conservation of residues located in the binding site suggested a similar function for
all PAZ domains. Indeed, I could show nucleic acid binding for multiple PAZ domains of
Drosophila and human Argonaute proteins.
5
Summary
Having established the function of the PAZ domain in nucleic acid binding, the next goal was
to understand the apparent specificity for single-stranded nucleic acids. For this, I defined
short oligonucleotides as effective ligands of the PAZ domain and determined two complex
structures of the Drosophila Ago2 PAZ domain bound to RNA and DNA oligomers,
respectively (chapter 2.2). This revealed that only the two 3’-terminal nucleotides are bound
by the PAZ domain, whereas the other nucleotides showed no stable interaction. A very
similar binding mode was observed when I studied the binding of an siRNA mimic, which
had a double-stranded stem and a 2 nt 3’-overhang. These observations provided an
explanation for the characterized specificity of the PAZ domain and demonstrated that the
PAZ domain is a 3’-end recognition domain. Together, my results indicate that the PAZ
domain contributes to the recognition and incorporation of Dicer-processed siRNAs into
RISC and thus serves as a specificity providing module in the RNAi pathway.
In the second part of my thesis, I studied the structure and function of a viral suppressor of
RNA interference. RNAi is thought to originate from an ancient endogenous defense
mechanism against viral and other heterologous dsRNAs. It had been shown before that many
viruses have evolved proteins that suppress RNA silencing and thereby counteract the cellular
response to the presence of double-stranded viral RNA. Although a large number of viral
suppressors had been identified, little is known about their molecular mechanisms.
One of these previously described suppressors of RNAi is the Flock House virus (FHV) B2
protein, which was shown to inhibit an anti-viral response in insect cells, plants and
Caenorhabditis elegans. To understand the molecular basis of its suppression activity, I
determined the solution structure of the FHV B2 protein (chapter 2.4). This study
demonstrated that B2 is an elongated dimer in solution, with each monomer composed of
three α-helices. NMR experiments demonstrated binding of the B2 dimer to a double-
stranded siRNA with high affinity. The binding site was mapped to a surface which is
comprised of one complete side of the dimer and composed of two anti-parallel helices (one
of each monomer). The distribution of positively charged side chains located in these helices
suggested that the dsRNA is contacted mainly via the sugar phosphate backbone, which
explains the lack of sequence specificity. This binding mode represents a novel way of
dsRNA recognition. As both siRNAs and long dsRNAs are bound, a dual mode of
suppression of RNAi is suggested: the B2 dimer could prevent the incorporation of viral RNA
into RISC by inhibition of Dicer-mediated cleavage of viral dsRNA and by binding directly to
siRNAs.
6
Introduction
7
1. Introduction
1.1 Regulation of gene expression by RNA silencing The basic flow of information in gene expression is from DNA to RNA to protein. This is also
known as the ‘central dogma of molecular biology’ that was established by Francis Crick in
the year 1958 (Crick, 1958). In order to ensure the right correlation between genes and the
amount of protein being produced, the expression of genes has to be tightly controlled by
complex regulatory mechanisms in both temporal and spatial manner. Recently, several post-
transcriptional mechanisms that regulate gene expression on the level of the messenger RNA
(mRNA) were discovered to be mediated by RNA. All these pathways, now commonly
referred to as RNA silencing pathways, change mRNA levels and/or influence the level of the
corresponding protein in response to the presence of dsRNA.
1.1.1 A brief history of the discovery of RNA silencing Before RNA silencing was discovered in animals, two phenomena of double-stranded RNA-
induced gene silencing had been already known for some years in plant systems, namely
virus-induced gene silencing (VIGS) and co-suppression. In the latter case it was observed
that after introducing exogenous transgenes into petunias with the goal of altering
pigmentation, the flowers did not deepen flower colour as expected. Instead, the flowers
showed variegated pigmentation, with some lacking pigment completely (Jorgensen, 1990).
This phenomenon, then called co-suppression, indicated that not only the transgenes
themselves were inactive, but also that the added DNA sequences somehow affected the
expression of the endogenous loci. Also, several laboratories found that plants responded to
RNA viruses by targeting viral RNAs for destruction (Ruiz et al., 1998 and chapter 1.3).
Nowadays, it is established that both complex transgene arrays and replicating RNA viruses
generate dsRNA, which triggers silencing as biological response.
A related phenomenon in animals was first discovered in the nematode worm Caenorhabditis
elegans as a response to double-stranded RNA, which resulted in sequence-specific gene
silencing. It was found that in an antisense RNA approach to inhibit gene expression, sense
RNA, that was used as a control, was as effective as antisense RNA for specific silencing of
the target gene (Guo & Kemphues, 1995). In a subsequent study, Fire et al. reasoned that the
Introduction
8
in vitro transcribed RNA preparations used by Guo et al. in their antisense studies were not
purely single-stranded RNA and that dsRNA in these preparations might be the key trigger of
silencing (Fire et al., 1998). Their breakthrough was to test the synergy of sense and antisense
RNAs. They could show that the dsRNA mixture was at least 10-fold more potent as a
silencing trigger than the sense or antisense RNAs alone. This phenomenon was named RNA
interference (RNAi), distinguishing it mechanistically from classical antisense-mediated
suppression. Silencing by dsRNAs had a number of remarkable properties: RNAi could be
induced by injection of dsRNA into the C. elegans gonad or by introduction of dsRNA
through either soaking of worms in dsRNA solution or feeding them with bacteria engineered
to express it (Timmons & Fire, 1998). Genetic and biochemical studies have now confirmed
that RNAi, co-suppression and virus-induced gene silencing share mechanistic similarities,
and that the biological pathways underlying dsRNA-induced gene silencing exist in many, if
not most, eukaryotic organisms (Hannon, 2002; Denli & Hannon, 2003; Meister & Tuschl,
2004; Novina & Sharp, 2004).
The initial observations in C. elegans were consistent with dsRNA-induced gene silencing
operating at the post-transcriptional level. Exposure to dsRNAs resulted in loss of the
complementary mRNAs, and promoter and intronic sequences were largely ineffective as
silencing triggers (Fire et al., 1998). A post-transcriptional mode was also consistent with data
from plant systems in which exposure to dsRNA, for example in the form of an RNA virus,
triggered depletion of mRNA sequences without an apparent effect on the rate of transcription
(Jones et al., 2001). Indeed, viral transcripts themselves were targeted, despite the fact that
they were synthesized cytoplasmically by transcription of RNA genomes (Ruiz et al., 1998).
These studies led to the notion that RNAi induces degradation of complementary mRNAs. In parallel, another mechanism was discovered that operates mainly at the level of protein
synthesis and is now known as miRNA-mediated gene silencing. One of the fist observations
was that lin-4, a gene known to control the timing of C. elegans larval development, does not
code for a protein but instead produces a pair of small RNAs (Lee et al., 1993). It was noticed
that these lin-4 RNAs had antisense complementarity to multiple sites in the 3’-UTR of the
lin-14 gene (Lee et al., 1993; Wightman et al., 1993), which were shown to be important for
regulation of lin-14 by lin-4. The fact that this regulation substantially reduced the amount of
Lin-14 protein without a noticeable change in lin-14 mRNA levels, supported a model in
which the lin-4 RNAs pair to the lin-14 3’-UTR to specify translational repression of the lin-
14 message. The shorter lin-4 RNA is now recognized as the founding member of an
abundant class of tiny regulatory RNAs called microRNAs (miRNAs). The importance of
Introduction
9
miRNA-directed gene regulation became evident as more miRNAs and their regulatory
targets were discovered (reviewed by Bartel, 2004; He & Hannon, 2004; Kim, 2005). The list
of biological processes regulated by miRNAs is already long and is expected to grow in the
future.
A third mechanism that affects gene expression in response to the presence of dsRNA acts on
the transcriptional level, i.e. on the chromatin structure (for review, see Bayne & Allshire,
2005). The initial observation of dsRNA-induced changes of chromatin was DNA
methylation of endogenous sequences that shared homology with viroids in infected plants
(Wassenegger et al., 1994). This observation was later followed by the finding that dsRNA
sharing sequence homology with promoter regions was able to induce gene silencing which
correlated with de novo methylation of promoter sequences (Mette et al., 2000). Later,
studies in fission yeast and flies showed that RNA-directed transcriptional silencing is not
confined to plants (see chapter 1.1.4).
After this brief historical overview, the different types of RNA-mediated gene silencing
mechanisms will be described in more detail in the following three subchapters.
1.1.2 RNA interference
In the course of RNA interference, a longer dsRNA precursor is cleaved into smaller RNAs
that act as guides for the silencing machinery. As a result, the expression of the corresponding
gene is down-regulated due to the degradation of the complementary mRNA (see Figure
1.1A).
The presence of these small RNAs was described for the first time in a study on transgene-
and virus-induced post-transcriptional gene silencing in plants. Hamilton et al. observed the
formation of discrete, small RNAs of around 25 nt that were complementary to the target of
silencing (Hamilton & Baulcombe, 1999). This result initiated biochemical studies to reveal
the relationship between the dsRNA trigger and the resulting small RNAs. For this, Tuschl et
al. developed a cell free Drosophila extract system that recapitulates many of the features of
RNAi (Tuschl et al., 1999). They observed sequence-specific mRNA degradation that is
promoted by dsRNA. Using the same system, it was shown that the dsRNA silencing trigger
is processed to RNA segments of 21-23 nt in length (Zamore et al., 2000). The authors termed
these RNAs small interfering RNAs (siRNAs). Simultaneously, it was found that extracts of
Drosophila cells that were transfected with dsRNA contain a nuclease activity that
Introduction
10
specifically degrades mRNAs homologous to this dsRNA (Hammond et al., 2000). The
siRNAs were shown to be associated with this nuclease, named RNA-induced silencing
complex (RISC, Figure 1.1A). Analysis of the chemical structure of siRNAs showed that they
were double-stranded and contained 5’-phosphorylated termini and 2 nt 3’-overhangs
(Elbashir et al., 2001c). This anatomy pointed towards a role of an RNase III ribonuclease in
generating siRNAs, because this family of nucleases displays specificity for dsRNAs and
generates such termini. The identity of the nuclease that cleaves the dsRNA into siRNAs in
the initiation step was revealed by a biochemical approach to be indeed a member of the
RNase III family of enzymes and was termed Dicer (Bernstein et al., 2001). The family of
Dicer enzymes is evolutionarily conserved, and proteins from many organisms besides
Drosophila, including Arabidopsis, tobacco, C. elegans, mammals and Neurospora have all
been shown to recognize and process dsRNA into siRNAs of a characteristic size for the
relevant species (Bernstein et al., 2001; Ketting et al., 2001). In general, RNase III enzymes
exhibit specificity for dsRNA and are involved in RNA metabolism in organisms ranging
from phages to animals (reviewed by Nicholson, 2003).
Figure 1.1 siRNA- and miRNA-mediated gene silencing pathways (A) Pathways of RNA silencing originating from dsRNA or endogenous hairpin pre-miRNAs. Current models of the structural and topological features of the protein–RNA complexes involved in RNAi are shown. The color coding of the different domains is the same as in B. Red arrows indicate endonucleolytic cleavage. RISC and miRNP (micro ribonucleoprotein particle) are the effector complexes of the siRNA and miRNA pathways, respectively. (B) Typical domain structure of Dicer and Argonaute proteins. Dicer comprises a DEXH helicase, a dsRBD, a PAZ domain and a domain of unknown function (DUF283), in addition to two RNase III domains. Argonaute proteins share a PAZ domain and a Piwi domain. This figure is taken from the review that is reprinted in chapter 4.2 (Lingel & Sattler, 2005).
A B
Introduction
11
Three structural classes of RNase III proteins have been described. The first class is
represented by Escherichia coli RNase III, the second by Drosha and the third by Dicer. The
first class comprises the simplest RNase III proteins, each of which contains one catalytic
endonuclease domain (RNase III domain) and a dsRNA binding domain (dsRBD). Members
of the second class of RNase III proteins comprise Drosha and homologs and contain two
RNase III domains, a dsRBD, and a long N-terminal segment, which is believed to be
involved in protein-protein interaction. Drosha was shown to have a function in miRNA
maturation (see next subchapter). The third class of RNase III enzymes, comprised of Dicer-
like proteins, are ~200 kDa multidomain proteins. Typically, Dicers contain an N-terminal
DEXH-box RNA helicase domain, a domain of unknown function (DUF283), a PAZ domain,
two RNase III domains, and a dsRBD (Figure 1.1B).
Insight into the molecular details of the RNase III activity of Dicer has been derived from the
crystal structure of the endonuclease domain of the RNase III homolog from Aquifex aeolicus
(Blaszczyk et al., 2001). This structure, together with biochemical studies, has established that
the catalytically active enzyme in prokaryotes comprises a homodimer of two RNase III
domains. As the Drosha and Dicer enzymes each contain two RNase III domains (RNase IIIa
and RNase IIIb), it was proposed that they could dimerize intra-molecularly to form an
enzymatically active unit that resembles the RNase III homodimer of A. aeolicus. For Dicer, it
was shown that the RNase IIIa and RNase IIIb domains of monomeric human Dicer do indeed
constitute the active enzyme that contains only a single center for the processing of dsRNA
(Zhang et al., 2004). This single processing center comprises two catalytic sites (one from
each RNase III domain) and cleaves both strands of the dsRNA substrate at one location,
leaving a 3’-overhang of 2 nt.
Several organisms contain more than one Dicer gene, with each Dicer preferentially
processing dsRNAs originating from a specific source. Drosophila has two paralogues: Dicer-
1 preferentially processes miRNA precursors (Lee et al., 2004b), and Dicer-2 is required for
long dsRNA processing (Pham et al., 2004). Dicer-2 is stably associated with the dsRBD-
containing protein R2D2 (Liu et al., 2003).
After the generation of siRNAs, the ribonucleoprotein particles (RNPs) are subsequently
rearranged into the RISC, which was introduced above as being the effector complex of
RNAi. For each siRNA duplex, only one strand, the so-called guide strand, is present in the
active, mature RISC, whereas the other strand, named the passenger strand, is destroyed. The
single-stranded siRNA in RISC guides the sequence-specific degradation of complementary
or near-complementary target mRNAs. The mRNA is cleaved at a position 10 nt from the 5’-
Introduction
12
end of the guiding siRNA (Elbashir et al., 2001b; Nykanen et al., 2001; Martinez et al., 2002).
Subsequently, the 5’-mRNA fragment generated by RISC cleavage is rapidly degraded from
its 3’-end by the exosome, whereas the 3’-fragment is degraded from its 5’-end by XRN1
(Orban & Izaurralde, 2005).
Like for Dicer, the identification of RISC components started with the analyses of the
associated activity in extracts of Drosophila embryos and cultured S2 cells. The first subunit
of RISC to be identified was the siRNA, which recognizes the substrate through Watson-
Crick base pairing (Hammond et al., 2000). Nykanen et al. showed that RISC is formed in
embryo extracts as a precursor complex of ~ 250 kDa which becomes activated upon addition
of ATP to form a ~100 kDa complex that can cleave substrate mRNAs (Nykanen et al., 2001),
whereas RISC from Drosophila S2 cells was purified as a ~500 kDa ribonucleoprotein
(Hammond et al., 2000; Hammond et al., 2001). The first protein component to be identified
as part of RISC was Argonaute2 (Hammond et al., 2001), a homolog of C. elegans Rde-1.
Rde-1 had previously been shown to be required for RNAi in C. elegans (Tabara et al., 1999)
and belongs to the Argonaute family of proteins (see chapter 1.2). In analogy to Argonaute2
in Drosophila, the human homologs eIF2C1 and eIF2C2 were isolated from HeLa cells in an
attempt to purify RISC activity (Martinez et al., 2002). In two subsequent studies on the
composition of RISC, three putative RNA binding proteins, the Drosophila homolog of the
fragile X mental retardation protein (FMRP), dFXR, VIG (Vasa intronic gene) and a
Drosophila homolog of the p68 RNA helicase (Dmp68), were shown to associate with a
tagged version of Ago2 (Caudy et al., 2002; Ishizuka et al., 2002). Tudor-SN, a 100 kDa
protein with 4 N-terminal Staphylococcus nuclease (SN) domains and a hybrid tudor/SN
domain, was another factor demonstrated to be bound to RISC (Caudy et al., 2003). It was
speculated that this protein might be part of the nucleolytic activity of RISC, but later studies
revealed that the actual endonuclease in RISC is a Mg2+ dependent nuclease (Martinez &
Tuschl, 2004; Schwarz et al., 2004), which ruled out Tudor-SN as it is Ca2+ dependent. A set
of structural and biochemical studies showed then that Ago2 itself is the protein responsible
for the endonucleolytic activity that cleaves the mRNA in RISC, now referred to as Slicer (see
Figure 1.1A and chapter 1.2). In addition to cleaving the target mRNA, it was recently
demonstrated that Ago2 also cleaves the passenger strand (Matranga et al., 2005; Rand et al.,
2005), thereby liberating the single-stranded guide strand. These results indicate that Ago2
receives directly the double-stranded siRNA duplex from the RISC assembling machinery
instead of binding a single-stranded siRNA after its separation by a helicase, as it was
proposed before (Tomari et al., 2004a).
Introduction
13
The possibility of specific down-regulation of target genes by RNAi provides a novel and
now frequently used tool to study gene function. In contrast to other functional inhibition
approaches like the use of antibodies or small molecule antagonists, RNAi has much higher
specificity and a less restricted applicability. For example, the transfection of mammalian
cells with siRNAs results in the potent, long-lasting post-transcriptional silencing of the
specific target genes (Caplen et al., 2001; Elbashir et al., 2001a). Beyond their value for
validation of gene function, siRNAs also hold great potential as gene-specific therapeutic
agents.
1.1.3 miRNA pathway
MicroRNAs (miRNAs) are single-stranded RNAs of 19-25 nucleotides in length and are
generated from endogenous hairpin-shaped transcripts by the same RNase III-type enzyme
Dicer (Figure 1.1A, Ambros et al., 2003; Bartel, 2004). miRNAs function as guide molecules
in post-transcriptional gene silencing by base pairing with target mRNAs, which leads to
mRNA cleavage or translational repression (Figure 1.1A). miRNAs have been shown to play
key roles in diverse regulatory pathways, including control of developmental timing,
haematopoietic cell differentiation, apoptosis, cell proliferation and organ development
(reviewed by Ambros, 2004; Bartel, 2004; He & Hannon, 2004; Kim, 2005).
miRNA genes can be clustered in the genome as it is observed for Drosophila (Lagos-
Quintana et al., 2001). In contrast to that, the majority of worm and human miRNA genes are
isolated and not clustered (Lim et al., 2003a; Lim et al., 2003b). Transcription of miRNA
genes is mediated by RNA polymerase II, and the transcripts, called primary miRNAs (pri-
miRNAs) were shown to contain both cap-structures and poly(A)-tails (Cai et al., 2004; Lee
et al., 2004a). The pri-miRNAs are usually several kilobases long and contain one or several
local hairpin structures. The stem loop structure is cleaved by the nuclear RNase III Drosha to
release the precursor miRNA (pre-miRNA) (Lee et al., 2003, Figure 1.1). The hairpin
precursors are usually ~60-80 nt in animals, but the lengths are more variable in plants. The
conserved family of Drosha ribonucleases was introduced in the previous chapter. Drosha
forms a large complex of ~500 kDa in D. melanogaster (Denli et al., 2004) or ~650 kDa in
humans (Gregory et al., 2004; Han et al., 2004). In this complex, called microprocessor,
Drosha interacts with its cofactor, the two dsRBDs containing DiGeorge syndrome critical
region gene 8 (DGCR8) protein in humans, which is also known as Pasha in D. melanogaster
Introduction
14
and C. elegans (Denli et al., 2004; Gregory et al., 2004; Han et al., 2004; Landthaler et al.,
2004). The tertiary structure of pri-miRNAs seems to be the primary determinant for substrate
specificity. Apparently, the Drosha complex can measure the length of the stem, because the
cleavage site is located approximately two helical turns (~22 nucleotides) from the terminal
loop (Zeng et al., 2005). When Drosha excises the hairpin miRNA precursor, a 5’-phosphate
and a 2 nt 3’-overhang remain at the base of the stem (Lee et al., 2003)
After being processed by Drosha, the pre-miRNA is exported out of the nucleus into the
cytoplasm by the nuclear export factor Exportin5 (Exp5) (Yi et al., 2003; Bohnsack et al.,
2004; Lund et al., 2004). Exp5 is a member of the RanGTP-binding transport receptors and
was initially identified as a factor exporting the eukaryotic elongation factor 1A together with
tRNAs (Calado et al., 2002). It was shown that the interaction between Exp5 and the RNA is
direct and involves contacts to the tRNA or miRNA double-stranded stem (Lund et al., 2004).
Following their export from the nucleus, pre-miRNAs are subsequently processed into ~22
nucleotide, usually imperfect miRNA duplexes by the cytoplasmic RNase III Dicer (Grishok
et al., 2001; Hutvagner et al., 2001; Knight & Bass, 2001). In Drosophila, the pre-miRNA
hairpins are cleaved by Dicer-1, which was shown to act in concert with the double-stranded
RNA binding domain protein Loquacious (Forstemann et al., 2005; Saito et al., 2005). As a
result of being cleaved in two subsequent steps by an RNase III enzyme, both ends of the
miRNA duplex have 2 nt 3’-end overhangs and 5’-phosphates. Like in the course of RNAi,
the double-stranded miRNA is incorporated into an Argonaute protein containing complex,
which is referred to as miRNP (Figure 1.1A). For example, the first described human miRNP
showed that miRNAs are present in a complex with the Argonaute eIF2C2, together with the
helicase Gemin3, and Gemin4 (Mourelatos et al., 2002).
MicroRNA loaded miRNPs can direct the downregulation of gene expression by either of two
post-transcriptional mechanisms: mRNA cleavage or translational repression. According to
the current model, the choice of the post-transcriptional mechanisms is not determined by
whether the small silencing RNA originated as an siRNA or a miRNA, but by the degree of
complementarity with the target. Once incorporated into a cytoplasmic RISC, the miRNA will
specify cleavage, if the mRNA has sufficient complementarity to the miRNA. However, it
will repress productive translation if the mRNA does not have sufficient complementarity to
be cleaved but does have a suitable constellation of miRNA complementary sites (Doench et
al., 2003; Zeng et al., 2003). Another determinant is the nature of the Argonaute protein that
is present in the particular miRNP, as it was demonstrated that not all Argonaute proteins are
competent for substrate cleavage (see chapter 1.2). In plants, miRNAs base pair with
Introduction
15
messenger RNA targets by precise or nearly precise complementarity and direct cleavage and
destruction of the target mRNA (Rhoades et al., 2002). When a miRNA guides cleavage, the
cut is at precisely the same site as that seen for siRNA-guided cleavage, i.e., between the
nucleotides pairing to residues 10 and 11 of the miRNA (Llave et al., 2002). In contrast, most
animal miRNAs are imprecisely complementary to their mRNA targets, with targets sites
being generally in the 3’-UTRs of the target mRNA. It was shown that not the complete
length of the miRNA is equally important for this interaction. Positions 2-7 form the critical
“seed”, and base pairing in this region is most important for target recognition. This fact is
also employed for the prediction of miRNA target sites (Enright et al., 2003; Stark et al.,
2003; John et al., 2004; Brennecke et al., 2005). Initially, it was believed that animal miRNAs
act only by inhibiting protein synthesis without affecting mRNA levels. More recently, it was
established that a miRNA can also act to guide the miRNP for cleavage of the cognate mRNA
(Hutvagner & Zamore, 2002), resulting in a change of mRNA levels (Lim et al., 2005). Thus,
it seems that miRNA-guided translational regulation and targeted mRNA degradation are
used simultaneously as natural regulatory mechanisms.
Recent observations concerning the localization of human Argonaute proteins provide a
possible and attractive explanation of the apparent repression of protein synthesis mediated by
miRNAs. It was shown that miRNA loaded Argonaute proteins and their target mRNAs
localize to mammalian processing bodies (P-bodies) (Liu et al., 2005b; Sen & Blau, 2005),
which were found previously to contain untranslated mRNAs and to serve as sites of mRNA
degradation (Ingelfinger et al., 2002; Eystathioy et al., 2003; Sheth & Parker, 2003). Such
localization could potentially explain in part or completely the translational repression and
degradation. Interestingly, one of the marker proteins of P-bodies, GW182 (Eystathioy et al.,
2002; Ingelfinger et al., 2002; Sheth & Parker, 2003), was found in a subsequent study to be
required for miRNA-mediated gene silencing (Rehwinkel et al., 2005). This emphasized the
role of P-body components in the miRNA pathway. Now, Argonaute proteins were shown to
interact physically with GW182 and it was further demonstrated that the structural P-body
integrity is crucial for miRNA-mediated silencing (Liu et al., 2005a).
Introduction
16
1.1.4 RNA-mediated transcriptional gene silencing
As outlined in chapter 1.1.1, RNA silencing mechanisms have been shown to be involved in
heterochromatin formation and transcriptional gene silencing (TGS). RNA-mediated
heterochromatin formation appears to be a natural epigenetic gene regulation mechanism.
This mechanism is believed to be active in most eukaryotes in response to environmental
changes and controls heritable changes in gene expression that are not caused by mutations
(for reviews, see Baulcombe, 2004; Lippman & Martienssen, 2004; Matzke & Birchler, 2005;
Wassenegger, 2005).
Most heterochromatin is found near centromeres and telomeres, and consists of tandem and
satellite repeats, which are sometimes interrupted by transposable elements. The expression of
heterochromatin is silenced by conserved epigenetic modifications of histones (methylation,
acetylation, phosphorylation and ubiquitination) and DNA (methylation) (Jaenisch & Bird,
2003). DNA methylation is absent, or nearly absent, in yeast, flies and nematodes, but a link
between DNA methylation and histone methylation is well established in fungi apart from
yeast, and in animals and plants.
The main fraction of endogenous siRNAs in Arabidopsis corresponds to transposons and
repeats whose histones and DNA are heavily methylated. The initiator dsRNA may be
produced by bidirectional transcription or transcription of inverted repeats. Another source of
dsRNA can be the activity of an RNA-dependent RNA polymerase (RdRP), which makes
dsRNA from a single-stranded RNA template. RNA-dependent RNA polymerases are found
in many organisms including C. elegans and plants, where they were shown to be required for
the amplification and systemic spreading of RNA-mediated silencing (Dalmay et al., 2000;
Sijen et al., 2001). However, RdRP genes have not been identified in insects and mammals.
RNA-directed DNA methylation (RdDM) was the first RNA-guided epigenetic modification
of the genome to be discovered. Originally detected in viroid-infected tobacco plants
(Wassenegger et al., 1994), RdDM was subsequently shown to require a dsRNA that is
processed into small RNAs of 21-24 nt, therefore reinforcing a link with RNAi (Mette et al.,
2000). dsRNAs which contain sequences that are homologous to promoter regions can then
trigger promoter methylation and transcriptional gene silencing. The Arabidopsis Dicer-like 3
protein and the Argonaute proteins Ago4 and Ago1 were shown to be involved in de novo
methylation or maintenance of methylation (Hamilton et al., 2002; Zilberman et al., 2003),
establishing the role of RNAi components in transcriptional gene silencing. The existence of
RdDM in other organisms is not clarified yet.
Introduction
17
In Schizosaccharomyces pombe, heterochromatin consists of simple transposon-derived
tandem arrays, which surround the central-core centromeric region of each chromosome. The
derepression of the centromeric outer-transposon repeats in mutants deficient for components
of the RNAi machinery led to the proposal that small RNAs function as guides to target the
chromatin modifications that are typical of heterochromatin (Volpe et al., 2002; , reviewed by
Grewal & Rice, 2004). Small interfering RNAs (siRNAs) are generated from centromeric
dsRNAs by the RNase III-type endonuclease Dicer. It was demonstrated later that a nuclear
effector complex known as RITS (RNA-induced initiation of transcriptional gene silencing)
complex is involved in targeting chromatin modifications. RITS contains siRNAs that
originate from heterochromatic regions, such as centromeres, and three identified proteins:
Ago1, Chp1 (a centromere-associated chromodomain protein), and Tas3 (a serine-rich protein
that is specific to fission yeast) (Verdel et al., 2004). Binding of the complex enables
recruitment of chromatin modifying proteins. Argonaute proteins were also shown to be
involved in heterochromatin formation in D. melanogaster (see next chapter).
Introduction
18
1.2 Argonaute proteins As described in the previous chapter, Argonaute proteins constitute core components of RNA-
mediated gene silencing mechanisms. They form a highly conserved family of proteins with
members found to be involved in various biological pathways ranging from RNAi to
development and stem cell determination and maintenance (reviewed by Carmell et al., 2002).
In the following subchapters, the functional diversity of Argonaute proteins will be introduced
briefly, followed by a more detailed description of the Drosophila Argonaute1 and
Argonaute2 proteins, with an emphasis on the role of Argonaute2 as Slicer. At the end, recent
structural insights into archaeal Argonaute and Piwi proteins will be provided. As the
structural and functional characterization of the PAZ domain was a main objective of this
thesis, this domain will only be introduced briefly without giving further details (for this, see
chapters 2.1, 2.2 and the discussion).
1.2.1 Argonaute proteins: a conserved protein family with diverse functions
The founding member of the Argonaute protein family is the Argonaute1 protein of
Arabidopsis thaliana. It was identified during an examination of ethyl methanesulfonate
(EMS) mutagenized A. thaliana populations for abnormal leaf morphology (Bohmert et al.,
1998). In this study, it was found that AtAgo1 mutation results in pleiotropical effects on
general plant architecture. Because of very narrow rosette leaves that looked similar to squids,
the authors called the identified genetic locus argonaute.
Argonaute family members constitute ~100 kDa highly basic proteins that contain two
signature domains, namely the PAZ and the Piwi domain (Figure 1.1B). Conservation of
amino acids in the Piwi domain of Argonaute proteins was first mentioned by Cox et al. when
analyzing the Drosophila Piwi (P-element induced wimpy testis) protein (Cox et al., 1998).
They showed that the central and C-terminal region of these proteins are well conserved and
defined a ~40 amino acid long, highly conserved C-terminal region, which they called the
Piwi box. In a subsequent bioinformatics study it was found that the Piwi box is part of a
larger ~300 amino acid domain which was termed Piwi domain (Cerutti et al., 2000). The
authors also demonstrated that the Piwi domain is not restricted to eukaryotes, but is also
Introduction
19
found in prokaryotes. In this initial study, no function could be proposed for the conserved
Piwi domain on the basis of sequence homology.
A central region of the Drosophila Piwi protein was identified to show a high level of
homology with other Argonaute proteins and with a region found in Dicer proteins (Cerutti et
al., 2000; Bernstein et al., 2001). Cerutti et al. named this novel, ~120 amino acid region PAZ
domain, after three proteins containing this domain: Piwi, Argonaute and Zwille/Pinhead. The
alignment indicated that PAZ domains can be grouped in two subfamilies, namely the
Argonaute and the Dicer family. The PAZ domains of the latter one are characterized by an
approximately 30 residue long conserved insert (see Figure S1 in chapter 2.1). Importantly,
the Dicer and Argonaute protein families are the only ones containing this domain. Based on
this fact and on co-immunoprecipitation experiments showing that both Drosophila Ago1 and
Ago2 proteins interact with Dicer (Hammond et al., 2001; Caudy et al., 2002), the function of
the PAZ domain was proposed to be protein-protein interaction, potentially mediating both
homo- and hetero-dimerization (Cerutti et al., 2000). As the PAZ domain was described to be
present only in eukaryotes, it was initially thought that also Argonaute proteins exist only in
eukaryotes. However, recent structural studies on Piwi domain containing proteins of
archaebacteria revealed that Argonaute proteins are also present in prokaryotes.
At the time of this initial bioinformatics analysis of eukaryotic Argonaute protein domains, no
structural information was available for any of the family members or of fragments of them.
Now, after intensive structural studies on the PAZ domain and on Argonaute proteins of
archaebacteria and biochemical investigations into eukaryotic Argonaute proteins, the
molecular function of both conserved domains is established and additional conserved
domains have been identified and annotated, namely the N- and Mid domain (see below).
In addition to the initial report on the developmental phenotype caused by a mutation in the
ago1 gene of A. thaliana, Argonaute family genes have been isolated from several organisms
in genetic screens for mutants that are deficient in RNAi and related phenomena, including
post-transcriptional gene silencing in plants and quelling in fungi. These include the Qde2
protein of Neurospora (Cogoni & Macino, 1997) and the Caenorhabditis elegans Rde-1
protein, of which mutants are strongly resistant to RNAi but are developmentally normal
(Tabara et al., 1999). In contrast, following the discovery of its pleiotropic effect on plant
architecture, Arabidopsis Ago1 was also shown to be necessary for post-transcriptional gene
silencing of transgenes as well as for co-suppression of transgenes and their corresponding
homologous endogenous genes (Fagard et al., 2000). Thereby, it was established that
Introduction
20
Argonaute proteins have functions in both development and RNA-mediated gene silencing.
This participation in two classes of biological functions is reflected by the fact that Argonaute
proteins can be separated according to their sequences into two subclasses: those that
resemble Arabidopsis Ago1, and those that more closely resemble Drosophila Piwi (see
Figure 1.2). Proteins of the Piwi subfamily are believed to be involved primarily in
developmental processes, whereas the Argonaute subfamily constitutes members that are
mainly implicated in RNA silencing mechanisms.
Figure 1.2 Phylogenetic tree of Argonaute proteins. The Ago subfamily is indicated in red, the Piwi subfamily in blue, orphans in black. The tree is based on an alignment done with ClustalW. Bootstrap percentages are indicated at each fork where the percentage is not 100. The tree shows that the Ago and Piwi branches are completely separated. Adapted from Carmell et al, 2002.
The diversity of functions of Argonaute proteins in development and RNA silencing can be
demonstrated by looking at the Drosophila Argonaute proteins. Drosophila contains four
characterized Argonaute proteins, namely Ago1, Ago2, Piwi and Aubergine, plus one
predicted from genomic DNA (Ago3). The first two (and the putative Ago3) belong to the
Argonaute subfamily and the latter two are included in the Piwi subfamily. DmAgo1 and
DmAgo2 function mainly in RNA silencing and will be discussed below. The piwi gene
encodes a nucleoplasmic protein known to be required for self-renewal of stem cells in the
male and female germline and the regulation of their division (Cox et al., 1998; Cox et al.,
2000). Similar to Piwi, Aubergine (also called Sting) is expressed embryonically in the
Introduction
21
presumptive gonad and is characterized by mutations that affect germline development
(Wilson et al., 1996; Schmidt et al., 1999). Aubergine is also responsible for acting post-
transcriptionally to maintain silencing of the X-linked repetitive Stellate locus that is
necessary for male fertility. The Stellate locus is silenced through a homology-dependent
mechanism mediated by an RNA transcribed from paralogous repeats called Suppressor of
Stellate, Su(Ste) (Aravin et al., 2001). Sense and antisense transcripts of Su(Ste) can be
detected and it was suggested that they form double-stranded RNA. Mutations in the
aubergine gene impair silencing by eliminating the short Su(Ste) RNA (Aravin et al., 2004).
In another study, Piwi and Aubergine were also shown to have a role in heterochromatin
silencing and HP1 localization in Drosophila (Pal-Bhadra et al., 2004). The authors
demonstrated loss of silencing as a result of mutations in piwi, aubergine, or spindle-E
(homeless), using tandem mini-white arrays and white transgenes in heterochromatin.
1.2.2 The Drosophila Ago1 and Ago2 proteins
DmArgonaute1 was the first identified Drosophila homologue of Arabidopsis Ago1, isolated
by a genetic approach to search for regulators of Wingless signal transduction. DmAgo1
maternal and zygotic mutant embryos showed developmental defects, with malformation of
the nervous system being the most prominent (Kataoka et al., 2001). Later, Ago1 was shown
to be required for efficient RNAi in Drosophila embryos, with its particular function lying
downstream of Dicer (Williams & Rubin, 2002).
Argonaute2 was identified by a biochemical approach to purify the RNAi effector nuclease
from cultured Drosophila cells (Hammond et al., 2001). After several steps of purification,
the remaining active fraction contained a ribonucleoprotein complex of around 500 kDa. The
identity of Argonaute2 was then revealed by mass spectrometry sequencing. It was also
shown that down-regulation of Ago2 by RNAi could suppress the silencing of a reporter,
indicating that Ago2 plays an important role in the silencing pathway. Soon after that, the
structure of the Ago2 PAZ domain was solved by NMR spectroscopy (see chapter 2.1),
providing the first molecular insight into the function of Argonaute proteins. At the same
time, the solution structure of the DmAgo1 PAZ domain was determined (Yan et al., 2003),
and later, the DmAgo2 PAZ domain was also solved by x-ray crystallography (Song et al.,
2003). Subsequent structural studies on the interaction of PAZ domains with nucleic acids
revealed the basis for RNA binding and specificity of the PAZ domain (chapter 2.2, and Ma et
Introduction
22
al., 2004), establishing a detailed characterization of the molecular function of this conserved
domain.
In an attempt to distinguish the functions performed by Ago1- and Ago2-associated RISCs in
Drosophila, Okamura et al. did a detailed study of RNA silencing pathways in embryos
lacking Ago1 or Ago2, complemented with experiments in cultured cells (Okamura et al.,
2004). They showed that Ago2 is an essential component for the siRNA-directed RNA
interference response and is required for the unwinding of the siRNA duplex and in
consequence for the assembly of the siRNA into RISC in Drosophila embryos. However,
Drosophila embryos lacking Ago2, which were siRNA-directed RNAi-defective, were still
capable of miRNA-directed target RNA cleavage. In contrast, Ago1, which is dispensable for
siRNA-directed target RNA cleavage, was shown to be required for mature miRNA
production that results in miRNA-directed RNA cleavage. This pointed already into the
direction that Ago2 might be very closely associated with the mRNA cleaving activity in
RISC and that Ago1 is more implicated in miRNA-mediated pathways.
1.2.3 Drosophila and human Argonaute2 act as Slicer
To figure out the relationship between DmAgo2 and the endonucleolytic activity, a stringent
biochemical purification of RISC activity from Drosophila Schneider cells was performed,
and Ago2 was found the be the only protein component present in the functional RISC that
was purified to homogeneity (Rand et al., 2004).
These results on DmAgo2 were in agreement with simultaneously published key studies
which presented compelling evidence that human Argonaute2 (HsAgo2) is the previously
unidentified endonuclease in human cells (Liu et al., 2004; Meister et al., 2004; Song et al.,
2004). Human Ago2 (also known as eIF2C2) has been shown before to be part of purified
RISC activity in HeLa cells, together with single-stranded siRNAs (Martinez et al., 2002) and
to be associated with miRNA containing ribonucleoprotein complexes (Mourelatos et al.,
2002). To get further insight into the specific functions of human Argonaute proteins in
RNAi, Liu et al. investigated their roles concerning RISC activity (Liu et al., 2004). By
transfection and immunoprecipitation experiments followed by an assay for siRNA-guided
cleavage of a complementary synthetic mRNA, they found that only HsAgo2-associated
RISC was able to catalyze the cleavage. This was true even though all Argonaute proteins
bound the co-transfected siRNA. The authors could also reconstitute RISC-mediated cleavage
Introduction
23
activity in vitro by adding an siRNA to immunopurified tagged HsAgo2, which was isolated
from cells that were not co-transfected with siRNAs. Mutational analysis, based on the
accompanying study describing the crystal structure of an Argonaute protein from the archaea
Pyrococcus furiosus (Song et al., 2004), provided strong evidence that the cleavage activity
was an intrinsic feature of Ago2, making it unlikely that the endonucleolytic activity was just
contributed by a co-purified Ago2-associated factor. Similar results providing additional
support for a role of HsAgo2 in cleavage were obtained by Tuschl and coworkers. They
demonstrated that purified Ago2 complexes, but not Ago1, Ago3 or Ago4 complexes, had
RISC activity (Meister et al., 2004).
The final biochemical evidence that Ago2 constitutes the Slicer activity was provided by a
recent reconstitution experiment (Rivas et al., 2005). The authors could show that human
Argonaute2, expressed recombinantly in E. coli and purified to homogeneity, can combine
with a small interfering RNA to form a minimal RISC that accurately cleaves substrate
RNAs.
1.2.4 Structures of archaeal Argonaute and Piwi proteins As described above, first structural insights into the molecular function of Argonaute proteins
were obtained from studies on the PAZ domain, which could be expressed recombinantly and
yielded soluble protein. Attempts of many laboratories to express full length eukaryotic
Argonaute proteins or other fragments besides the PAZ domain in amounts necessary for
structure determination were not successful up to this day. Being aware of these difficulties,
several labs turned successfully to prokaryotic homologs with unknown function (reviewed
by Hall, 2005).
The structure of the Argonaute protein of Pyrococcus furiosus and the molecular basis
of Slicer activity
To gain insight into the possible function of the larger signature domain of Argonaute
proteins, namely the Piwi domain, Song et al. performed a structural analysis of a
multidomain protein from the archaebacterium Pyrococcus furiosus (Song et al., 2004). It has
been predicted by sequence homology that this protein contains a Piwi domain. Remarkably,
the crystal structure of the protein revealed that it also comprises an unanticipated PAZ
domain (Figure 1.3). Thus, containing a Piwi and a PAZ domain, the protein represents a
bona fide Argonaute protein (now known as PfAgo), indicating that this protein family is not
Introduction
24
restricted to eukaryotes. PfAgo consists of four distinct domains held together by an
interdomain connector: an N-terminal domain, a PAZ domain, a middle (Mid) domain
structurally related to the Lac repressor (Friedman et al., 1995), and a C-terminal Piwi domain
(Figure 1.3). The N-terminal, middle and Piwi domains form a crescent-shaped structure, with
the Piwi domain sitting in the center of the crescent (Figure 1.3B). The N-terminal domain
forms a “stalk” that holds the PAZ domain on top of the concave face of the crescent, placing
it in opposition to the Piwi domain. Overall, the structure is quite compact, considering the
length of the polypeptide chain (i.e. 770 residues).
Figure 1.3 Crystal structure of the Pyrococcus furiosusArgonaute protein. (A) Schematic diagram of the domain organisation. Domain borders are indicated by the respective amino acid numbers. (B) Ribbon representation of PfAgo showing the N-terminal domain (gray), the PAZ domain (blue), the middle domain (red) and the Piwi domain (green). Putative active site residues in the Piwi domain and conserved residues in the PAZ domain are drawn in stick representation (see also Figure 1.4 and Figure 3.3). This figure and all other figures displaying molecular structures were prepared with MOLMOL (Koradi et al., 1996).
B
A
Despite very low sequence identity (<10%), the PAZ-like domain of PfAgo protein
superimposes well with the known structures of eukaryotic PAZ domains. The similarities
and differences will be presented in the discussion (see chapter 3.1).
The structure of the full length PfAgo also provided the first structural view on a Piwi
domain. The Piwi domain of PfAgo shows a striking structural similarity and conserved
secondary structure topology with the family of RNase H enzymes. This class of enzymes is
known to cleave the RNA strand of an RNA:DNA hybrid (Yang & Steitz, 1995). The RNase
H fold consists of a five-stranded mixed β-sheet surrounded by α-helices on both sides (see
Figure 1.4).
Introduction
25
Figure 1.4 The Piwi domain of PfAgo resembles the catalytic domain of RNase H endonucleases. (A) Crystal structure of RNase H1 from E. coli, showing side chains of the catalytic DDE triad residues and a bound Mg2+ ion. Conserved and unconserved secondary structure elements are colored green and yellow, respectively. (B) The Piwi domain in the crystal structure of PfAgo. Side chains of putative residues of a catalytic DDE triad and Arg627 are shown. Structural elements are colored as in (A). This figure is taken from the review reprinted in chapter 4.2 (Lingel & Sattler, 2005).
B A
The active site of RNase H enzymes is comprised of a so-called DDE motif, three acidic
amino acids with the side chain carboxylates positioned to catalyze the cleavage reaction. The
reaction is known to require the presence of divalent cations, such as Mg2+ or Mn2+ (Steitz &
Steitz, 1993; Kanaya & Ikehara, 1995; Chapados et al., 2001). The reaction mechanism
differs from most ribonucleases but resembles deoxyribonucleases, leaving 3’-OH and 5’-
phosphate termini (Wintersberger, 1990). Notably, Slicer activity results in the same products
(Martinez & Tuschl, 2004; Schwarz et al., 2004). In the Piwi domain of the PfAgo protein,
two of the three potentially catalytic carboxylates (D558 and D628) are in equivalent
secondary structure positions as those found in the RNase H fold (Figure 1.4). Mutagenesis of
the equivalent residues in HsAgo2 to alanine eliminated mRNA cleavage activity, as
described above (Liu et al., 2004). Song et al. suggested that the active site may be completed
by a third carboxylate from a conserved glutamate (E635), even though this residue is located
in a different sequence position in bacterial proteins (e.g. E48 in E. coli RNase H1). The
identity of the third carboxylate was revealed in a subsequent study on PfAgo, which made
use of PfAgo crystals that were soaked in with Mn2+ solution. The Mn2+ ion can replace the
Introduction
26
Mg2+ expected to be coordinated at the active site but is easier to detect crystallographically.
This experiment indicated that the third coordinating side chain is H745 (Figure 1.5). In
addition, the equivalent residue in an Argonaute protein of Aquifex aeolicus (AaAgo), D683,
was coordinated by a Ca2+ ion along with the two other conserved acidic residues in the
crystal structure of this protein (Yuan et al., 2005, see below). Mutation of the corresponding
residue in HsAgo2, H807, also eliminated mRNA cleavage activity, confirming its
importance for catalytic activity (Rivas et al., 2005).
Notably, not all Ago proteins are active as endonucleases, and the identification of the active
site residues of this family of proteins explains why some Ago proteins are not cleavage
competent. For example, in cleavage incompetent HsAgo1, the catalytic histidine is replaced
by arginine and mutation of the equivalent histidine in cleavage competent HsAgo2 to
arginine inactivates it (Liu et al., 2004; Rivas et al., 2005). Similarly, in cleavage incompetent
HsAgo 4, one of the catalytic aspartates is replaced by glycine.
Figure 1.5 Superimposition of the active sites of PfAgo (blue) and Bh-RNase HC (green). The RNA strand from the DNA:RNA hybrid in the Bh-RNase HC structure is shown in pink. Water molecules from the PfAgo (red) and Bh-RNase HC (dark pink) are shown as spheres. Gray dashed lines indicatecoordination of metals A and B in Bh-RNase HC. Adapted from Hall, 2005.
Additional insight into the catalytic mechanism of Ago proteins was provided by the recent
crystal structures of the RNase H of Bacillus halodurans (Bh-RNase HC) with its DNA:RNA
hybrid substrate (Nowotny et al., 2005). The authors found that the DNA:RNA hybrid binds
at the active site of Bh-RNase HC via two divalent cations. One is equivalent to the Mn2+
coordinated by two aspartates and the histidine in Ago proteins. This metal is also coordinated
by two water molecules, one that is positioned ~3 Å from the scissile phosphate group and is
predicted to be the nucleophile (metal A in Figure 1.5). Equivalent water molecules are found
coordinating the Mn2+ ion in the PfAgo structure. A third water molecule is coordinated to the
Introduction
27
Mn2+ and this corresponds to a non-bridging oxygen atom in the phosphate group preceding
the scissile phosphate.
A second divalent cation (metal B in Figure 1.5) is coordinated by an aspartate that
coordinates both metals, a glutamate, an asparagine, a water molecule, and the RNA. It is
more difficult to predict where this second metal binds in the PfAgo structure. Two possible
metal ligands are E635 and R627, although they are more distant from the first metal ion than
the ligands for the second metal in the Bh-RNase HC structure (Nowotny et al., 2005). The
two-metal ion catalytic mechanism is likely to be conserved among RNase H family enzymes,
but identification of the ligands for the second metal ion site may require a crystal structure of
an Ago protein with RNA bound at the active site.
The structures of the Piwi protein of Archaeoglobus fulgidus and the Argonaute protein
of Aquifex aeolicus: insights into 5’-end recognition
The crystal structure of a simpler archaeal, Piwi domain containing protein from
Archaeoglobus fulgidus was solved by Parker et al. They described first the structure of the
free protein together with in vitro binding data (Parker et al., 2004). Because of the lack of a
PAZ domain, the protein is not considered to be an Argonaute protein and was called AfPiwi.
In comparison to PfAgo, AfPiwi lacks also the N-terminal domain and thereby represents an
N-terminally truncated version of the PfAgo. The structure confirmed the similarity of the
Piwi domain fold with the RNase H fold and showed the presence of carboxylate side chains
in the right position in a putative active site. Most importantly, the structure revealed that the
side chains of the C-terminal residues contribute to the hydrophobic core of the protein.
Interestingly, the C-terminal four residues of Ago and Piwi proteins are highly conserved as
aliphatic and aromatic residues. Moreover, the C-terminal carboxylate group projects onto the
molecular surface, lying at the interface between the middle and the Piwi domain. There, the
carboxylate is bound to a well-ordered metal ion, which is coordinated also by other side
chains of highly conserved amino acids. Binding data revealed that AfPiwi forms a distinct
complex with an siRNA-like duplex. A mutation in the conserved region at the C-terminus
reduced the binding affinity significantly, suggesting an important role for this region in RNA
binding.
This was confirmed by two subsequent structural studies of the AfPiwi protein in complex
with a double-stranded RNA solved simultaneously (Ma et al., 2005; Parker et al., 2005).
Both structures revealed the molecular basis of RNA binding and of 5’-end recognition. For
several reasons, the 5’-ends of siRNAs and miRNAs are important for mRNA target
Introduction
28
recognition and definition of the site of RNA cleavage. The main reasons are that the
positions 2-7 of miRNAs are the most critical region for target recognition (see chapter 1.1.3)
and that a phosphate group at the 5’-end of the siRNA strand that forms the guide for mRNA
target recognition has been shown to be required for efficient RNAi and for proper cleavage
site selection (Nykanen et al., 2001; Schwarz et al., 2002). The crystal structures of AfPiwi
with siRNA-like duplexes could provide a structural perspective on the importance of the 5’-
end. In both structures, the 5’-nucleotide of the guide siRNA is unpaired and bound in a
pocket made up primarily of residues in the middle domain. The first base, U1 or A1, forms a
stacking interaction with Y123, a conserved aromatic position in Ago proteins (Figure 1.6).
The unpairing of the first base explains its relative unimportance for specifying miRNA
targets. The structures contain a highly conserved metal binding site that anchors the 5’-
nucleotide of the guide RNA. The 5’-phosphate group of the guide siRNA is bound directly
by side chains of four residues that are invariant in Ago proteins (Y123, K127, Q137, and
K163) and by the main chain nitrogen of F138 (Figure 1.6).
Figure 1.6 5’-phosphate binding pocket of AfPiwi. A divalent metal ion (M) is coordinated by the 5’-phosphate group (5’ P), Q159, the terminal carboxylate of L427, the third phosphate group of the guide RNA (yellow) and a water molecule (wat). Black dashed lines: metal coordination, red dashed lines: hydrogen bonds. Adapted from Hall, 2005.
The phosphate group is also bound by a divalent cation that is coordinated by Q159, the C-
terminal carboxylate, a water molecule, and the third phosphate group of the guide RNA,
confirming the previously observed importance of the C-terminus (see above). Mutation of
the corresponding amino acids that contact the 5’-phosphate in human Ago2 resulted in
attenuated mRNA cleaving activity, suggesting that 5’-end binding is a conserved function
Introduction
29
(Ma et al., 2005). Thus, the binding of the 5’-phosphate group observed in the AfPiwi
structures appears to represent a good model for the RNAi effector complex.
As mentioned above, a crystal structure of a full length Argonaute protein from Aquifex
aeolicus (AaAgo) in complex with a single-stranded rU8 oligonucleotide was determined
recently and reported together with binding and cleavage studies (Yuan et al., 2005).
Although the rU8 was required for successful crystallization, it is disordered and therefore not
visible in the electron density map. The structure confirms the domain organisation of the
PfAgo protein. It shows a bilobal architecture, with one lobe comprising the N-terminal and
PAZ domain and the other one comprising the middle and Piwi domain. The PAZ domain
observed in this structure superimposes with the structures of eukaryotic PAZ domains, but is
most similar to the PAZ domain of PfAgo (see chapter 3.1 for discussion). The Piwi core fold
of AaAgo is closely related to the RNase HII domain (Lai et al., 2000). A deep basic surface
pocket is lined by highly conserved residues, including an aromatic residue, a patch of basic
residues and polar residues from the Mid domain, supplemented by the C-terminal
carboxylate and conserved aromatic residue, which are contributed to the Mid-Piwi interface
by the Piwi domain.
This highly conserved basic pocket is localized to the same region that was identified in the
crystal structures of AfPiwi-RNA complexes to anchor the phosphorylated 5’-end of the guide
RNA. This implies that this pocket in AaAgo is also likely to be the site for 5’-end recognition
of the guide strand in all Argonaute proteins. In addition to describing the crystal structure,
the authors found that AaAgo could direct the cleavage of the RNA target when pre-incubated
with DNA or RNA, showing for the first time a ribonuclease activity for an archaeal
Argonaute protein. Whether this demonstration of DNA-directed cleavage of RNA is
reflecting the function of prokaryotic Argonaute proteins in vivo remains to be established. Up
to now, no regulatory mechanism similar to eukaryotic RNA silencing was reported for
prokaryotes.
Introduction
30
1.3 Viral suppression of RNA silencing RNA interference is a post-transcriptional gene regulation pathway (see chapter 1.1) that is
thought to have originated from an ancient cellular defense mechanism against viral and other
endogenous double-stranded RNAs in the cell, like transposons with dsRNA intermediates
(Plasterk, 2002). Early studies in plants and the concomitant discovery of plant viral
suppressors of RNA silencing already pointed towards a role of RNAi in antiviral defense.
Most plant viruses have positive-sense, single-stranded RNA genomes (Hull, 2002). During
their replication cycle, there are dsRNA intermediates present which are strong inducers of
RNA silencing and are believed to be the precursors of viral siRNAs (Ahlquist, 2002). Upon
incorporation of these viral-derived siRNAs into RISC, the viral RNA is specifically targeted
for degradation (Baulcombe, 1999; Waterhouse et al., 1999), allowing the plant to recover
from viral infection. Thus, it is not surprising that many plant viruses encode suppressors of
RNA silencing (reviewed by Li & Ding, 2001; Voinnet, 2001; Ding et al., 2004; Roth et al.,
2004; Voinnet, 2005), which can serve as an example of co-evolution. The expression of
these suppressors is believed to be a means for viruses to overcome the host antiviral
silencing defense.
The first viral suppressors of RNA silencing reported were the helper component proteinase
HC-Pro and the 2b protein encoded by potyviruses and cucumoviruses, respectively
(Anandalakshmi et al., 1998; Brigneti et al., 1998; Kasschau & Carrington, 1998; Li et al.,
1999), which were both identified as pathogenicity determinants of synergistic viral diseases
(Ding et al., 1996; Pruss et al., 1997). This provided a rationale for identifying novel viral-
encoded suppressors, and the re-investigation of other factors of this type revealed that several
effectively inhibit RNA-mediated silencing (Voinnet et al., 1999).
For the identification of viral suppressors, a transient expression assay using Agrobacterium
co-infiltration is used, as it provides a rapid and easy test of suppressor activity (Llave et al.,
2000; Voinnet et al., 2000). Agrobacterium tumefaciens is a commonly used bacterial
pathogen of plants. In the assay, two different strains are used: one strain is used to induce
RNA silencing of a reporter gene (usually green fluorescent protein (GFP)), and another strain
is used to express the candidate suppressor. Briefly, the strategy is to co-infiltrate mixtures of
the two bacterial strains into a plant leaf (usually tobacco leafs) and then examine the
infiltrated patch over time for silencing of the reporter. Most of the known suppressors were
identified by this simple assay (Table 1.1).
Introduction
31
Viral family Virus Suppressors Other functions
Positive-strand RNA viruses in plants Carmovirus Turnip Crinkle virus p38 Coat protein
Cucumovirus Cucumber Mosaic virus Tomato Aspermy virus 2b Host-specific movement
Closterovirus Beet Yellows virus Citrus Tristeza virus
p21 p20 p23 CP
Replication enhancer Replication enhancer Nucleic acid binding Coat protein
Comovirus Cowpea Mosaic virus S protein Small coat protein
Hordeivirus Barley Yellow Mosaic virus γb Replication enhancer, movement, seed transmission, pathogenicity determinant
Pecluvirus Peanut Clump virus p15 Movement
Polerovirus Beet Western Yellows virus Cucurbit Aphid-borne Yellows virus P0 Pathogenicity determinant
Potexvirus Potato virus X p25 Movement
Potyvirus Potato virus Y Tobacco Etch virus Turnip Yellow virus
HC-Pro
Movement, polyprotein processing, aphid transmission, pathogenicity determinant
Sobernovirus Rice Yellow Mottle virus P1 Movement, pathogenicity determinant
Tombusvirus Tomato Bushy Stunt virus Cymbidium Ringspot virus Carnation Italian Singspot virus
p19 Movement, pathogenicity determinant
Tobamovirus Tobacco Mosaic virus Tomato Mosaic virus p30 Replication
Tymovirus Turnip Yellow Mosaic virus p69 Movement, pathogenicity determinant
Negative-strand RNA viruses in plants Tospovirus Tomato Spotted Wilt virus NSs Pathogenicity determinant Tenuivirus Rice Hoja Blanca virus NS3 Unknown DNA viruses in plants
Geminivirus African Cassava Mosaic virus Tomato Yellow Leaf Curl virus Mungbean Yellow Mosaic virus
AC2 C2 C2
Transcriptional activator protein (TrAP)
Positive-strand RNA viruses in animals
Nodavirus Flock House virus Nodamura virus B2 Plaque formation
Negative-strand RNA viruses in animals
Orthomyxovirus Influenza virus A* NS1 Poly(A) binding, inhibitor of mRNA export, PKR inhibitor
DNA viruses in animals Adenovirus Adenovirus VA1 RNA PKR inhibitor Poxvirus Vaccinia virus* E3L PKR inhibitor
Table 1.1 RNA silencing suppressors encoded by plant, insect and vertebrate viruses. * These have only been demonstrated in heterologous systems (insects and plants). HC-Pro, helper component proteinase; PKR, RNA-dependent protein kinase. Adapted from Voinnet et al., 2005.
Many unrelated viral proteins have evolved silencing suppressor activities in addition to their
other functions, contributing to the remarkable diversity of these factors, which have now
Introduction
32
been identified for almost all types of plant viruses (see Table 1.1). It would not be surprising
if all plant viruses are found to encode at least one suppressor of RNA silencing, provided that
each of the viral proteins is assayed for possible suppression of both intra- and inter-cellular
(systemic) RNA silencing.
Silencing suppression has also been documented in insect cells and was discovered through
the related expression strategies and functional similarities of the Cucumber Mosaic virus
(CMV) suppressor 2b and the B2 protein of Flock House virus (FHV) (Table 1.1, Li et al.,
2002). FHV is an animal virus and belongs to the Nodaviridae family of non-enveloped
icosahedral viruses. FHV infects mainly insect cells, but replicates also in yeast, plant and
mammalian cells. All members of the Nodaviridae family have a bipartite genome comprised
of a positive-sense, single-stranded RNA (Johnson et al., 2001). The first genome segment of
FHV, RNA1, encodes the entire viral contribution to the viral RNA-dependent RNA
polymerase and replicates autonomously in the absence of RNA2, which depends on RNA1
for replication (Ball & Johnson, 1999). During replication, each RNA1 synthesizes a
subgenomic RNA3 that corresponds to the 3’-end of RNA1 and encodes two open reading
frames (ORFs, B1 and B2, see Figure 1.7).
Figure 1.7 Schematic representation of the nodavirus genome organization showing the two segments of the bipartite RNA genome (RNA 1 and 2) and the subgenomic RNA3. ORFs are shown in white and gray boxes.
Deletion of the 106 amino acids encoding B2 ORF from FHV results in a drastic loss of virus
accumulation in Drosophila S2 cells, which can be rescued by decreasing the cellular content
of Ago2 (Li et al., 2002). Therefore, B2 suppresses the effect of the Ago2-dependent silencing
response that normally restricts FHV accumulation. A subsequent study showed complete
replication of the FHV RNA genome in C. elegans (Lu et al., 2005). Similarly to the Ago2
dependence in Drosophila cells, replication of FHV in C. elegans triggered potent antiviral
silencing that required the Argonaute protein Rde-1, which is essential for RNAi mediated by
small interfering RNAs (siRNAs) but not by microRNAs. C. elegans was capable of rapid
virus clearance in the absence of FHV B2 protein but not when B2 was present. Thus, B2 acts
as a broad-spectrum RNAi inhibitor.
Introduction
33
The B2 protein of a related Nodavirus family member, Nodamura virus (NoV), which is
infectious for insect and mammalian hosts, has been shown to block RNAi in mammalian
cells (Sullivan & Ganem, 2005). Both the FHV and NoV B2 proteins bind to siRNAs and
longer dsRNAs, and therefore it was proposed that they may sequester siRNAs and prevent
the processing of long dsRNAs into siRNAs by the host RNaseIII-like enzyme Dicer (Lu et
al., 2005; Sullivan & Ganem, 2005).
Silencing suppression clearly plays a role in establishing virus infections as most of these
suppressors were previously shown to be essential for systemic virus spread in their hosts.
The molecular mechanism is unknown for most of the identified suppressors, with the
exception of p19 and p21, which will be introduced below. Structural and biochemical studies
have been initiated to reveal the molecular function of these interesting proteins, including
FHV B2 (see chapter 2.4).
Size selective recognition of siRNAs by the tombusvirus p19 protein
The linear, positive-sense ssRNA genome of tombusviruses encodes a 19 kDa protein that
was shown to be a suppressor of RNA silencing (Voinnet et al., 1999; Qiu et al., 2002; Qu &
Morris, 2002; Silhavy et al., 2002) and to bind RNA silencing-generated and synthetic 21 nt
siRNAs in vitro under stoichiometric conditions (Silhavy et al., 2002). Therefore, it was
suggested that p19 binds siRNAs and sequesters them, thereby preventing their incorporation
into RISC. This was confirmed by crystal structures of the p19 protein bound to siRNAs
(Vargason et al., 2003; Ye et al., 2003), which will be described in more detail.
The p19 homodimer bound to siRNAs revealed a C-terminal domain comprised of a four-
stranded β-sheet flanked on one side by three α-helices (Figure 1.8). Two N-terminal helices
are docked onto the C-terminal domain by salt bridges. The secondary structure topology of
the C-terminal domain corresponds to a circular permutation of the ribosomal L1 protein,
which binds dsRNA as a monomer with its β-sheet surface (Nikulin et al., 2003).
Homodimeric p19 forms an extended concave anti-parallel β-sheet surface, which mediates
multiple interactions with the sugar phosphate backbone of the double-stranded helical region
of the siRNA (Figure 1.8). Recognition of the siRNA stem by the β-sheet surface is distinct
from the mode of binding of dsRBDs, in which mainly loop regions interact with the stem of
the RNA duplex (see chapter 3.3). The p19 homodimer binds the siRNA with very high
affinity (KD ~0.17 nM) (Vargason et al., 2003). The siRNA recognition is sequence-
unspecific and employs hydrogen bonds, electrostatic contacts (involving phosphate and 2’-
Introduction
34
hydroxyl groups), and aromatic-base stacking interactions. A number of conserved serine and
threonine residues mediate key interactions with 2’-OH groups in the minor groove of the
RNA duplex, with the axis of the RNA helix being bent ~40° towards the protein.
Figure 1.8 Crystal structure of the p19 homodimer bound to a 21 nt siRNA (PDB code 1RPU). The two N-terminal helices of p19, which mediate key interactions with the terminal base pairs and contain two important tryptophan residues, are colored orange. The inset shows essential interactions that contribute to the size-selective RNA binding by the p19 homodimer; these interactions occur symmetrically at both ends of the 19 bp RNA stem. This figure is taken from the review reprinted in chapter 4.2 (Lingel & Sattler, 2005).
Size selectivity for an RNA helix of 19 bp is defined by symmetric interactions of the highly
conserved residue W42 and the more variable residue W39 with the 5’- and 3’-terminal bases,
respectively, at both ends of the RNA duplex region of a 21 nt siRNA (Figure 1.8). Consistent
with this, p19 does not interact with single-stranded nucleic acids or dsDNA, and has reduced
binding affinity for siRNAs with longer or shorter double-stranded regions (Vargason et al.,
2003; Ye et al., 2003). In one of the structures, the 2 nt single-stranded 3’-overhang makes
few interactions with p19 (Vargason et al., 2003), whereas in the other structure no electron
density is observed for the two nucleotides at the 3’-end (Ye et al., 2003). This suggests that
recognition of the 2 nt overhang is not essential, a proposal that is consistent with the
comparable binding affinities of p19 for a 19 bp RNA duplex with and without a 2 nt single-
stranded 3’-overhang (Vargason et al., 2003; Ye et al., 2003). By contrast, a hydrogen bond
between the indole amide of the conserved W42 and the 5’-phosphate seems to be important
for binding, since the affinity of p19 for an siRNA lacking a 5’-phosphate is reduced 23-fold
(Vargason et al., 2003). The major determinant of siRNA specificity is thus size selective
recognition of the 19 bp duplex - a binding mode that is unprecedented and provides a
fascinating example of structure-specific recognition of the key mediators of RNAi.
Introduction
35
General nucleic acid binding of the Beet yellow virus p21 protein
Recently, the crystal structure of a second suppressor of RNA silencing was solved by Ye et
al. (Ye & Patel, 2005), namely of the p21 protein of Beet Yellow virus (BYV). This virus is a
positive-sense, single-stranded RNA virus that comprises a large 15.5 kb genome and is a
member of the genus Closterovirus. The p21 protein, one of nine proteins encoded in the
BYV genome, was the only one that suppressed the silencing of green fluorescence protein
(GFP) induced by GFP dsRNA in the transitive agrobacterium infiltration experiment (Reed
et al., 2003). Additional biochemical studies showed that p21 binds siRNA duplexes both in
vitro and in vivo, but not single-stranded miRNAs (Chapman et al., 2004). It was also shown
that p21 has no strict binding specificity for siRNA duplexes.
Ye et al. have determined the crystal structure of p21 in the free state. The structure shows
that p21 forms an octameric ring of previously unknown topology, in which individual p21
monomers adopt an all α-helical structure (Figure 1.9).
Figure 1.9 Ribbon representation of the entire p21 octamer. Neighboring monomers are alternatively colored with green and red. The dots indicate disordered regions. Adapted from Ye & Patel, 2005.
Interestingly, the eight p21 monomer subunits associate into the closed ring structure through
two types of inter-monomer interactions, namely a head-to-head and a tail-to-tail
arrangement. Electrostatic potential analysis revealed that the ring surface displays a highly
polarized distribution of charges: negative charges dominate the outer surface of the ring,
while positive charges dominate the inner surface of the ring. Therefore, the inner surface was
proposed to bind RNA via electrostatic interactions, but mutations of conserved residues in
this putative binding site gave no conclusive results. Electromobility shift analysis revealed
that p21 is a general nucleic acid binding protein. Thus, p21 is very different from p19 in
terms of structure and RNA binding. The mechanism of RNAi suppression by p21 is therefore
not revealed by the structure, because it can interact with all kinds of nucleic acids. However,
it has been shown that transgenically expressed p21 can co-immunoprecipitate siRNAs in vivo
(Chapman et al., 2004).
Introduction
36
1.4 NMR spectroscopy: a tool to understand the function of biomolecules
a
ery important tool in the repertoire of physical methods to study biomolecules (Wüthrich,
986). It is employed mainly to determine the three dimensional structure of both proteins and
tic resonance
agnetic resonance in condensed phase was first detected by Bloch and Purcell in
asis for observation of nuclear
agnetism by NMR spectroscopy is the nuclear spin angular momentum I (representing a
Over the last 20 years, nuclear magnetic resonance (NMR) spectroscopy has developed into
v
1
nucleic acids at atomic resolution, to study their dynamics in solution and to characterize
molecular interactions (Krishna & Berliner, 2003). Unlike in x-ray crystallography, the
investigated sample is usually in liquid phase and not in solid phase, which makes it possible
to study the structural and dynamical parameters in a close to physiological environment. The
obtained results are thus essentially complementary to crystallographic investigations. The
main disadvantage of NMR spectroscopy is the size limit concerning the biomolecule of
interest. This limitation is slowly pushed towards larger molecules but further improvement
still remains one of the major tasks in method development in modern NMR spectroscopy.
In the course of this Ph.D. thesis, three main application of biomolecular NMR were
employed: solution structure determination, characterization of binding events and molecular
recognition, and analysis of dynamics. All three topics will be introduced in the following
chapters, in addition to some basic concepts.
1.4.1 The basic principle of nuclear magne Nuclear m
1946 (Bloch et al., 1946; Purcell et al., 1946). The physical b
m
vector, indicated in bold), a purely quantum mechanic property that has no analog in classical
mechanics. The magnitude of I is given by equation
( )[ ]122 +=⋅= IIhIII
in which I is the angular momentum quantum number. The value of I is specified by the
physical nature of the nucleus. Table 1.2 (adapted from et al., 1996) gives an
overview of angular momentum quantum numbers I, the gyromagnetic ratios γ (see below)
Cavanagh
and on the natural isotopic abundance of the respective nuclei that are commonly used in
NMR spectroscopy of biomolecules. Most commonly, only nuclei with I=1/2 are employed
Introduction
37
Nucleus I γ (T·s)-1 Natural abundance (%) 1H 1/2 2.6752 · 108 99.98 2H 13
1 4.107 · 107
1
--
100.00 23 100.00
0.02 C
41/2 6.728 · 107 1.11
N 5
1 1.934 · 107
7 99.64 1 N 1/2
5/2 2.712 · 10
70.36
17O 3.628 · 108
0.04 19F 1/2 2.5181 · 10
7Na 3/2 7.080 · 10831P 1/2 1.0841 · 10 100.00
by biomolecular NMR, which includes the nuclei 1H, 13C, 15N and 31P. Because of the low
natural abundance of 13C and 15N (see Table 1.2), proteins are usually uniformly labelled to b
Table Properties selected nuclei. The angular monumbe the gyrom ic ratio tural isotopic abundance for nuclei of particular importance in biomolecular NMR spectroscopy are shown.
1.2 of mentum quantum r I, agnet γ, and the na
nt of the nuclear spin angular momentum is
e
enriched in these isotopes.
The value of the z compone
mIz h=
in which m is the magnetic quantum number with m = (-I, -I+1, …, I-1, I). Thus, Iz has 2I+1
possible values. The orientation of the spin angular momentum vector in space is quantized,
is collinear with the vector representing the nuclear spin angular momentum vector
because the magnitude of the vector is constant and the z component has a set of discrete
values.
Nuclei that have a non-zero spin angular momentum also possess a nuclear magnetic moment
μ, which
and is proportionally linked to it:
Iγ=μ with a z component of mIzz hγ=γ=μ
with γ being the gyromagn istic for a give nucleus (see Table
1.2). Because the angular momentum is a quantized property, so is the nuclear magnetic
etic ratio, a constant character n
moment.
Introduction
38
In an external magnetic field B, the spin states of the nucleus have energies given by
IBB γ−=−= μE
in which B is the magnetic field vector. The magnetic pole moment μ aligns with the main
axis of the field and becomes quantized with energies proportional to their projection onto B.
is directed along the z
di
In a high field NMR spectrometer, the static external magnetic field B0
axis of the laboratory coordinate system resulting in an energy term:
00zm BmBIE γ−=γ−= h
The projection of the angular momentum of the nuclei onto of the laboratory frame
results in 2I+1 equally spaced energy levels, which are known as the Zeeman levels (Figure
At equilibrium energy states are unequally populated because lower energy
rientations of the magnetic dipole vector are more probable. The relative population of the
the z axis
1.10). For the I=1/2 nuclei usually observed in biomolecular NMR, this results in two distinct
energy levels which are commonly termed α and β, corresponding to spin states with m=+1/2
and m=-1/2, respectively.
Figure 1.10 The two precessional cones for a I=1/2 nucleus in a magnetic field B
cshowing typi al spin vectors occupying the two Zeeman states. mI=+1/2 results in the α state, whereas mI=-1/2 results in the βstate.
, the different
o
states is given by the Boltzmann distribution
TkBeE
NNα
Δ−
β
=
The bulk magnetic moment M of a macroscopic sample is given by the vector sum of the
corresponding quantities of μ for individual nuclei. The small population excess in the lower
energy levels gives rise to a bulk magnetization of the sample parallel to the static magnetic
field. This macroscopic magnetization is referred to as the longitudinal magnetization.
Introduction
39
Transitions between Zeeman levels can be stimulated by applying electromagnetic radiation,
with the selection rule for magnetic dipole transitions being Δm= ±1. The energy required to
excite a transition between the m and (m+1) state (α and β state) is proportional to the static
magnetic field B0 (see Figure 1.11).
Figure 1.11 A magnetic field B0 acting on a nucleus with I=1/2 results in two different spin states which are referred to as Zeeman levels, or α and β states. The energy difference ΔE between the states is dependent on the gyromagnetic ratio of the particular nucleus and on the field strength of B0.
The corresponding frequency is given by
00 Bω γ= or ππν 2/B2/ 000 γ=ω=
This frequency ω0 is also called t frequency and can be imagined, in analogy to
classical mechanics, as the frequency of precession of the magnetic dipole moment around the
0, a given nucleus experiences a local magnetic field
local that depends on the chemical environment. Thus, the resonance frequency of a spin is
he Larmor
main axis of the static magnetic field. The sensitivity of NMR spectroscopy depends on the
population differences between Zeeman states. Because the difference is very marginal, in the
order of 1 in 105 for 1H spins in an 11.7 Tesla magnetic field, NMR is an insensitive
spectroscopic technique compared to other methods such as visible or ultraviolet
spectroscopy. As the difference is directly correlated with the magnitude of the static
magnetic field, it is an ongoing effort to build more powerful magnets, especially for
biomolecular NMR.
1.4.2 The chemical shift
In addition to the static magnetic field B
B
governed by the effective field Beff
local0eff BBB +=
The local magnetic contribution is mainly constituted by the electronic environment of a
given nucleus, e.g. high electron densities result in large shielding of the B0 field. Thus, the
bond structure and polarity of the surrounding atoms of a spin give rise to slightly different
Introduction
40
frequencies of precession, which therefore directly correspond to different local chemical
environments. This phenomenon is called chemical shift δ. In practice, chemical shifts are
measured in parts per million (ppm) relative to a reference resonance signal of a standard
molecule (which is water in biomolecular NMR):
6
0
ref 10×ω
Ω−Ω=δ
in which Ω and Ωref are the offset frequencies of the signal of interest and the reference signal,
respectively, and ω0 is the Larmor frequency. Differences in chemical shifts of individual
nuclei are responsible for the dispersion that is observed in NMR spectra.
ple NMR experiment, a radio frequency with the associated temporary external
an axis perpendicular to the main axis of the static
agnetic field B0. This causes the equilibrium longitudinal magnetization Mz,eq to precess
1.4.3 The basic NMR experiment
In a sim
magnetic field B1 is applied along
m
around the main axis of B1. If the radio frequency irradiation is set as a pulse that makes the
magnetization to rotate by 90 degree, the magnetization is left in the xy plane, giving rise to a
so-called transverse magnetization (see Figure 1.12).
FT
Figure 1.12 In the static external magnetic field B0, the macroscopic magnetization is oriented along the z axis and has an equilibrium value of Mz,eq (green arrow). A 90 degree radio frequency pulse along the y axis causes the magnetization to rotate around this axis resulting in a transverse magnetization Mxy in the xy plane. Precession in the xy plane is recorded by a detector. Relaxation phenomena cause the signal to decay exponentially resulting in the detection of a free induction decay (FID). This information is transferred into a spectrum by Fourier transformation (FT).
Introduction
41
As a result of the 90 degree pulse, zero magnetization is left along the z axis. Under the
fluence of the static magnetic field, the magnetization is precessing in the xy plane with
ifferent radial frequencies for individual spins (because of their different chemical shift, see
s through chemical bonds: scalar coupling
ins that are close to each other can influence each other, and these interactions are
employed by NMR spectroscopy to obtain information about the chemical structure of the
ig
in
d
above). The rotation of the magnetization is recorded by a detector in the xy plane. Because of
relaxation phenomena (see chapter 1.4.8), the transverse magnetization is lost by time which
results in an exponential decay of the detected signal, which is therefore called a free
induction decay (FID). The intensity of the FID is detected over a certain time range. This
data represents the so called time domain that is transferred into a spectrum representing the
signal as a function of frequency by Fourier transformation. In case the pulse is double in
time, the magnetization is flipped by 180 degree, resulting in a magnetization with a value of
–Mz,eq. Thus, by tuning the time and power of a radio frequency pulse, the magnetization can
be changed by a desired angle, which is used extensively in pulse programs implemented in
NMR spectroscopy.
1.4.4 The interaction of spin Sp
invest ated molecule. For example, spin-spin interactions or couplings mediated by electrons
that form chemical bonds between nuclei are called scalar couplings. The strength of the
interaction is measured by the scalar coupling constant, nJIS, in which n designates the number
of covalent bonds separating the two nuclei, I and S. The magnitude is usually expressed in
Hertz (Hz) and is independent of the strength of the external field B0. Values are large (~10-
140 Hz) for nuclei separated by a single bond and become smaller (~0-10 Hz) for two- and
three-bond couplings. Typical values observed in proteins are shown in Figure 1.13.
Figure 1.13 Typical values for one- and two-bond scalar couplings (1J and 2J) in polypeptides are shown above the bond that mediates the spin interaction. Numbers are in red and given in Hertz (Hz). C’ is the carbonyl (CO) carbon.
Introduction
42
The scalar coupling modifies the energy levels of the involved spin systems and modifies
thereby the NMR spectrum. A coupled two-sp s
four energy levels corresponding to the possib
orientation of spins (up-up and down-down, β
and down-up, βα, αβ). An illustration of th
Transitions between the states occur according to the selection rule Δm = ±1 (where m is the
agnetic quantum number of each state) and result in two signals at resonance frequencies of
For J-coupling to be active, one component of the coupled spin systems needs to contain
transverse magnetization. When transfer of magnetization from one nucleus to another is
desired, scalar coupling is allowed to evolve, thus J-couplings serve as a fundamental
in system constituted of spins I and S display
le combination of spin states, these are parallel
β, αα) and anti-parallel orientations (up-down
e energy levels is shown in Figure 1.14.
m
the I and S states. Introducing a scalar coupling between I and S with the value of JIS results in
the modulation of the energy levels and each of the previously observed signals is now split
into two with a distance corresponding to the magnitude of the J-coupling (Figure 1.14).
Figure 1.14 Energy level diagram of an IS spin system. The four possible energy states (horizontal lines with numbers 1-4) of an uncoupled IS spin system are indicated on the left, whereas the same J-coupled spin system is shown on the right. Allowed transitions between them obeying the selection rule Δm=±1 are indicated by arrows and result in the corresponding spectra which are outlined below. The I and S spins and the correlated resonance lines are shown in black and red, respectively. Identical J-couplings and therefore line splittings are observed for both signals, as energy levels change by the same magnitude in opposite directions.
Introduction
43
mechanism to transfer magnetization between spins in NMR spectroscopy. In case coupling
of spins is not wanted, the spins of a coupled heteronuclear spin system can be inverted by
applying a 180 degree pulse on one of the components (decoupling).
1.4.5 Resonance assignment in proteins In order to be able to correlate the information in the NMR spectrum with a specific atomic
position in the molecule, the resonance frequency or chemical shift of each atom has to be
assigned. The sequence specific assignment of NMR resonances requires many different
experiments taking days of spectrometer time. The resonance assignment of the backbone
atoms of a protein provides already information about the secondary structure and is a
necessity to do interaction studies (see chapter 1.4.10) and to analyze the dynamics of the
protein (chapter 1.4.8). Nevertheless, backbone assignment does not reveal much of
information about the three dimensional structure of a protein and the complete assignment is
a necessary prerequisite to achieve this goal.
Assignment of proteins requires in general uniformly isotopically labelled protein samples.
This is because of the large amounts of protons in proteins that give rise to very crowded
spectra. For that reason, the signals are correlated with the resonances of heteronuclei, namely 15N and 13C, and thereby dispersed over additional frequency axes. As the natural abundance
of these nuclei is very low (see Table 1.2), protein samples have to be enriched for these
advantage of increased
solution, this type of experiments also provide better signal to noise ratio. The strategy is
usually to assign the atoms of the protein backbone first, and then the atoms of the amino acid
shifts of the Cα atoms are often not dispersed enough for unambiguous assignment, the Cβ
isotopes. This is achieved by recombinant expression of the protein in bacteria grown in
isotope enriched media which results in the uniform incorporation of 15N or 15N/13C. The
following description is based on a 15N/13C-labelled protein sample.
Resonance assignment is most commonly based on triple-resonance experiments that
correlate different nuclei by the transfer of magnetization via the bonds connecting them, i.e.
via J-couplings (Sattler et al., 1999) (see chapter 1.4.4). Besides the
re
side chains.
Backbone assignment
A basic set of experiments is outlined below to assign the resonances of the backbone atoms
and to obtain connectivities between individual amino acids. Due to the fact that the chemical
Introduction
44
resonance is normally also employed to obtain reliable assignments. A sufficient set consists
of four experiments: the HNCA-, HN(CO)CA-, CBCANH-, and CBCA(CO)NH-experiments
(Figure 1.15).
on atoms of the previous amino acid (i-1) with a given amide group. In contrast,
xperiments without this step reveal the chemical shifts of both the previous and the same
mparison of peaks can be used to get connections between residues
hain protons and transferred to the directly bound carbon nuclei. There, it is
llowed to propagate or mix along the whole of the aliphatic spin system via a transfer termed
TOCSY (total correlation spectroscopy), redirected to the amide nitrogen spin system via the
carbonyl carbon (CO) and eventually transferred from there to the HN protons for detection.
These experiments are referred to as (H)CC(CO)NH and H(CC)(CO)NH for correlation of the
aliphatic carbon resonances and the side chain protons of the previous amino acid (i-1) to the
backbone amide of residue i, respectively (see Figure 1.16).
Figure 1.15 Summary of NMR experiments for the resonance assignments of protein backbone atoms. Correlated spin systems are boxed for the respective experiments, spin systems which are employed to transfer magnetization without chemical shift evolution are outlined in gray and indicated in brackets in the respective pulse sequence names.
All of them rely on the correlation of the Cα and/or Cβ atoms with the amide nitrogen and
proton of either the same amino acid (i) and/or the previous amino acid in the polypeptide
sequence (i-1). The brackets represent thereby a step in the experiment at which the chemical
shift of the specified nucleus is used only for the transfer of magnetization but not evolved.
Experiments involving a transfer over the backbone carbonyl carbon (CO) therefore correlate
only the carb
e
residue. Therefore, a co
and to assign the resonances of the HN, N, Cα and Cβ atoms.
Side chain assignment
In a second step, triple-resonance experiments are employed to assign carbon and proton
chemical shifts of individual side chains. In general, magnetization is initially created on
aliphatic side c
a
Introduction
45
nucleus affects the magnetic
ifically, the z component of the dipolar field of
e frequency of nucleus Q by an amount that depends on
Figure 1.16 Summary of NMR experiments for the resonance assignments of protein side chain atoms. Correlated spin systems are boxed for the respective experiments, spin systems which are employed to transfer magnetization without chemical shift evolution are outlined in gray and indicated in brackets in the respective pulse sequence names.
In addition, a HCCH-TOCSY correlates all aliphatic proton resonances with all carbon spins
systems within the same amino acid residue.
1.4.6 The interaction of spins through space: dipolar coupling and residual dipolar coupling
A second type of interaction between spins is the direct dipole-dipole interaction which is
referred to as dipolar coupling. The magnetic field caused by one
field at the site of the other nucleus. More spec
nucleus P will change the resonanc
the internuclear distance and on the orientation of the internuclear vector relative to the static
field B0. This results in a splitting of resonances with a separation in frequency by
( ) 2/1cos3DD 2max −θ= PQPQ
where θ is the angle between the internuclear vector and B0 and DPQmax is the doublet splitting
that applies for the case where θ is zero. Thus, the dipolar splitting DPQ provides direct
information of the internuclear vector. Dipolar coupling splittings are very large in solid phase
NMR, but because of the fast reorientation and isotropic orientation in an isotropic solution,
dipolar couplings are usually averaged to zero and therefore not visible in liquid state NMR
pectroscopy. Nevertheless, information about the orientation of bonds can now be even
btained from samples in solution because of the availability of tunable, weak alignment,
which results in residual dipolar coupling.
s
o
Introduction
46
Residual dipolar coupling
Recently, NMR methods have been developed which allow the extraction of structural
restraints characterizing long-range order. Traditionally, structure determination was based on
distance restraints derived from quantification of NOESY spectra (see chapter 1.4.7) as well
as torsion angle restraints from measurement of 3J-couplings (chapter 1.4.4 and 1.4.9). A key
limitation inherent to this approach concerns the strictly local nature of these parameters,
since they solely define distances and angles between atoms close in space within the
tructure.
ne-bond vectors relative to the molecular
usceptibility tensor, have been introduced to derive angular restraints and are now employed
in NMR based structure calculations (Tolman et al., 1995; Tjandra & Bax, 1997; Tjandra et
lso
r as they restrain all bond orientations relative to a common
at
tunable degrees of molecular alignment can be achieved by placing the molecule under
investigation into a dilute liquid crystalline media. The alignment relies on the steric
interaction between the highly aligned medium and the biomacromolecules, which induces a
s
In particular, residual dipolar couplings (RDCs), which contain information about the
orientation of the internuclear, usually o
s
al., 1997; reviewed in Bax, 2003). Besides constraining local geometry, dipolar couplings a
have a global ordering characte
frame, independent of their individual localization in the structure.
As described before, the source of this structural information is the direct dipole-dipole
interaction between two nuclei, which is normally averaged to zero in solution. In the case of
weak, partial anisotropic alignment, however, a certain residual orientation of the dipole-
dipole vector in the magnetic field remains. Therefore, the major breakthrough with respect to
the potential use of dipolar coupling derived structural restraints was the demonstration th
weak alignment of the biomolecule itself. In this case, not all orientations of the molecule are
equally likely to occur and a residual dipolar coupling can be observed.
The most commonly measured residual dipolar coupling in structure determination of proteins
is the coupling between the 15N nucleus and a proton (Figure 1.17).
Figure 1.17 Magnetic dipole-dipole coupling, illustrated for a 15N-1H spin pair. 15N and 1H moments are aligned parallel (or anti-parallel) to the static magnetic field B0. The total magnetic field in the B0 direction at the 15N position can increase or decrease relative to B0, depending on the orientation of the 15N-1H vector and the spin state of the proton (parallel or anti-parallel to B0). Adapted from Bax, 2003.
Introduction
47
The measurement of 15N-1H RDCs provides information about the orientation of the bond
vector of amide groups in the protein backbone (Figure 1.17). Similar RDC values therefore
indicate similar orientation of the bond vectors, as observed for example for amino acids
located in helices. Tjandra et al. developed a procedure to incorporate these experimental
dipolar couplings into the structure calculation (Tjandra & Bax, 1997). This resulted in an
substantial improvement of the percentage of backbone angles in the most favoured region of
the Ramachandran plot and in considerably higher cross-validation statistics. Thus, RDC
sonance is irradiated in a
Although the dipolar interaction is averaged to zero for molecules in solution, the interaction
of spins results in relaxation, which is a second order perturbation. The transitions include W0
and W2 transitions, which involve the simultaneous change of both spins and thus give rise to
an NOE. These transitions are forbidden in the conventional sense that they cannot be directly
measurements are a valuable tool in structure validation and refinement and were employed in
the structure determinations presented in this thesis (see results).
1.4.7 The nuclear Overhauser effect
In general, the nuclear Overhauser enhancement or nuclear Overhauser effect (NOE) is the
fractional change in intensity of one NMR line when another re
double irradiation experiment. The NOE has its origin in population changes as a result of a
particular form of relaxation (see next chapter), namely dipole-dipole cross-relaxation
(Neuhaus & Williamson, 2000). This is caused by the dipolar interactions between spins,
where the magnetic moment of a particular spin induces a magnetic field at the position of the
interacting spin that is close in space (see previous chapter).
The coupling gives rise to modified energy levels shown in Figure 1.18 for a two-spin system
IS.
Figure 1.18 Energspin system (IS) w
y diagram of a two ith dipolar coupling.
n.
The NOE is a result of W0 and W2transitions which occur via spin-spin cross-relaxatio
Introduction
48
excited by a radio frequency or cause directly detectable NMR signals, but they are not
forbidden in the context of relaxation.
The NOE is so important for NMR spectroscopy of biomolecules because it is manifested by
a through space interaction and the magnitude of it directly correlates with the distance
between the interacting nuclei. The NOE is therefore used to derive proton-proton distance
restraints for the structure determination by NMR. The intensity of an NOE is measured using
NOE spectroscopy (NOESY) experiments. The cross-peak intensity is approximately
proportional to the inverse sixth power if the inter-nuclear distance and can be observed for
distances up to 6 Å.
1.4.8 Relaxation and dynamics
The macroscopic longitudinal and transverse magnetizations were introduced in chapter 1.4.1
and 1.4.3, respectively. The longitudinal magnetization (Mz) is aligned along the static field
axis of B0 with a certain equilibrium value. The transverse magnetization (Mxy) is
perpendicular to B0, around which it precesses at the Larmor frequency ω0 and it has an
equilibrium value of zero. After any disturbance caused by the action of a radio frequency
irradiation, two processes must occur for equilibrium to be restored: the longitudinal
magnetization must return to its original value and the transverse magnetization must decay to
nuclear spin relaxation is a consequence of coupling of the spin system to the
ations in the magnetic field.
urement of T1 re
Longitudinal relaxation necessarily involves transition
magnetization is a result of difference in populations of spin levels and β (chapter 1.4.1).
The longitudinal relaxation time is also called spin-lattice relaxation time or simply T1. The
lattice means basically the surrounding that includes a set of energy levels that is
given NMR transition, there will be always a possible change within the lattice that involves
zero. The first process is called longitudinal relaxation and the second transverse relaxation.
Any
surroundings and is induced by fluctu
Longitudinal relaxation and meas laxation times
s between spin states as the longitudinal
α
term
generally provided by the various translations, rotations and internal motions of the molecules
that constitute the sample. Because of the very large number of degrees of freedom that the
lattice possesses, this set is effectively a continuum. The lattice modifies the local magnetic
fields at the locations of nuclei and thereby couples the lattice and the spin system. For any
Introduction
49
the same quantity of energy. Thus, exchange of energy between the spin systems and the
lattice brings the spin system in thermal equilibrium with the lattice and the distribution of
spins returns to the equilibrium population.
The T1 relaxation time describes the longitudinal relaxation and is measured by the inversion
recovery experiment (Figure 1.19).
Following an inversion by applying an 180 degree pulse, the value of Mz equals –Mz,eq. Then,
the decay of Mz to its equilibrium value will follow the simple expression
Figure 1.19 The longitudinal relaxation time T1 is measured by the inversion recovery experiment. At equilibrium, the macroscopic magnetization (green arrow) is oriented along the z axis and has a value of Mz,eq. At the beginning of the experiment, it is inverted by a 180 degree pulse along y into the –z direction. After a time τ, the magnetization has decayed due to longitudinal relaxation. A 90 degree pulse along the y axis flips the magnetization into the xy plane, where it is detected as an FID.
⎥⎤
⎢⎡
−=ττ−
T1eqz,z 21M)(M e
⎦⎣
n in the xy plane that is After a delay τ, a 90 degree pulse is applied to generate magnetizatio
recorded as an FID. The delay is varied incrementally to obtain a set of data to fit the T1 time.
Transversal relaxation and measurement of T2 relaxation times
In contrast to T1, the transverse relaxation is a result of loss and dephasing of individual
contributions to the macroscopic transverse magnetization. This is due to transitions, but also
due to any other process that perturbs the individual Larmor frequencies like loss of phase
coherence. The transverse relaxation time is also known as spin-spin relaxation time or T2.
Introduction
50
As the line width of a resonance signal is related to T2 by equation
T21
⋅=
π 1/2νΔ
( 2/1νΔ is the full line width at half height of the line), a shorter transverse relaxation time T2
gives rise to a broader line width. Due to the fact that T2 is inversely correlated to the
molecular tumbling rate (see below and Figure 1.21), this is a major obstacle for NMR
spectroscopy of large molecules.
As explained, T2 is the relaxation rate of Mxy and T2 times are generally measured by
employing a Hahn spin-echo experiment, which is using the principle of symmetry reversal
(Figure 1.20).
magnetization in the xy
ed
s an FID. The size of this echo will only be affected by the spin-spin relaxation processes. To
obtain a whole data set, the delay τ is incrementally changed. The Carr-Purcell-Meiboom-Gill
Figure 1.20 The transversal relaxation time T2 is determined by the Hahn spin-echo experiment. At equilibrium, the macroscopic magnetization (green arrow) is oriented along the z axis and has a value of Mz,eq. A first 90 degree pulse along the y axis flips the magnetization into the xy plane. Spins are allowed to evolve for a delay time τ. The spins are inverted by a 180 degree pulse along the x axis, followed by a second delay time τ. The echo is then recorded as an FID. The T2 time can be extracted from the magnitude of the echo, because all contributions of spin-lattice relaxation (T1) are cancelled out due to the symmetrical concept of the experiment.
The Hahn echo is constituted first by a 90 degree pulse that flips the
plane. During the first τ delay, the magnetization evolves according to its chemical shift and
field inhomogeneity. Then a 180 degree pulse is applied, which inverts the magnetization.
Following this inversion pulse, another τ delay is applied. During this delay, the
magnetization refocuses. This will give rise to an echo after the 2 τ delays which is record
a
Introduction
51
(CPMG) sequence, which is derived from the Hahn spin-echo sequence, can be also used to
measure T2 relaxation times. This sequence is equipped with a "built-in" procedure to self-
correct pulse accuracy error.
modulated by molecular motion. These interactions involve
ainly dipolar coupling (see chapter 1.4.6), and chemical shift anisotropy (CSA). Other terms
used by T2. The m
us, relaxation rates are affected by intra-molecular mobility in flexible
structures and correlated with the overall rotational tumbling of the molecule in solution.
Small and medium sized molecules have a fast tumbling rate which is reflected by a small
correlation time τc. The rotational correlation time τc represents the time that it takes for a
molecule to reorient, which is much shorter for small molecules compared to large molecules,
like biomolecules. For molecules having a small τc, T2 is normally equal to T1 (Figure 1.21).
For biomolecules that are usually larger in size resulting in a slower tumbling rate and a larger
correlation time τc, T2 is shorter than T1 (Figure 1.21).
Relaxation and dynamics in biomolecules
Longitudinal and transverse relaxation rates are intimately linked to motion and the
interactions between the spins and the surroundings. This is due to the fact that relaxation is a
result of spin interactions that are
m
are responsible for additional line broadening ca ost important is chemical
exchange, which describes the interconversion of two or more states (conformations) into
each other. Th
Adobe System
s
100
10-1
10-2
101
10-3
103
102
10-12 10-11 10-10 10-9 10-8 10-7
τc [s]
T1&T2 [s]
Figure 1.21 Logarithmic plots of the T1 and T2 times versus the correlation time τcfor dipolar relaxation. The curves were computed for proton resonance frequencies νo of 400, 600 and 800 MHz.
Introduction
52
Because the slow tumbling limit applies for typical biomacromolecules with a molecular
weight above ~6 kDa, T2 is inversely proportional to the correlation time τc, which itself is
proportional to the molecular weight. Thus, by measuring T2 relaxation times (see above) it is
ossible to estimate the molecular weight of a macromolecule in solution.
otion, they are an optimal probe to characterize
be
Structure determination with NMR derived constraints
p
As the relaxation rates are directly linked to m
the dynamical properties of a molecule. For proteins, the aforementioned measurements are
usually set up as two dimensional experiments, correlating 15N nuclei with a directly bound
proton. Thus, in proteins, relaxation rates are determined for the backbone amide nitrogens.
Most commonly, the obtained data is represented in a plot of the ratio of the T1 and T2 times
of the amide nitrogens versus the position of the corresponding residue in the amino acid
sequence. If the observed value differs significantly from the average T1/T2 ratio, it indicates
regions of internal motion at time scales above or below the correlation time τc of the overall
tumbling. Thus, information about the rotational diffusion tensor of a molecule can
extracted from relaxation data. Values of T1/T2 will differ significantly within one molecule
if the shape of the molecule can not be approximated by a sphere, resulting in anisotropic
diffusion. Measurements of dynamical properties were obtained of the investigated proteins in
this thesis (chapter 2.1 and 2.4).
1.4.9
As described in the previous chapters, a variety of structural restraints can be extracted from
NMR experiments. Considering the input data for structure calculations, the most important
ones are short proton-proton distances derived from NOESY experiments (Table 1.3). With
increasing size of the biomolecule, overlap of cross-peaks becomes a problem for the
unambiguous assignment of distances. This overlap can be reduced by the introduction of
additional frequency dimensions, and therefore the NOESY spectrum is in practice usually
acquired as three dimensional 13C- and 15N-edited NOESY experiments. The distances rij can
be calculated from the NOE intensities Iij by equation
6/1
ij
refrefij I
Irr ⎟
⎟⎠
⎞⎜⎜⎝
⎛=
using an appropriate reference for distance calibration. In practice, the experimental distance
is imprecise because of peak integration errors, peak overlap, spin diffusion and internal
Introduction
53
dynamics. For this reason, the distance is defined to be between a lower and an upper bound.
Usually, the lower bound separation of a pair of protons is set to the sum of the van der Waals
radii.
In addition to NOEs, structural information can be obtained by analyzing 3J-coupling
constants. As was first described by Karplus (Karplus, 1959), the magnitude of a 3J-scalar
coupling constant is a function of the dihedral angle θ, formed by the three covalent bonds
CBA +θ+θ= coscosJ 23
The constants A, B, and C depend on the particular nuclei involved in the covalent bonds.
Most commonly, the coupling constant between the amide proton HN and the Cα bound
proton Hα is analyzed (Table 1.3). The value of the coupling constant provides a direct
correlation with the backbone dihedral angles Φ and Ψ.
NMR observable Information
NOEs (15N- and 13C-resolved NOESY) Interatomic distances (<6 Å)
J-couplings (3JHN,Hα) Dihedral angles (backbone angles Φ and Ψ)
RDCs (HN-N) Bond projection angles (backbone HN-N)
Solvent exchange (HN) Hydrogen bonds
Structure calculation
the heavy water and can therefore be
observed in a 15N,1H-correlation experiment. The information that a specific proton is located
in a regular secondary structure is then translated into distances between this proton and
another atom, based on the previously determined secondary structure (Table 1.3).
Restrained molecular dynamics Simulated annealing
3D structure ensemble
Another set of restraints is generally obtained from experimentally observed H/D exchange
rates. To detect slowly exchanging amide protons which might be part of H-bonds, the protein
is first lyophilized and then redissolved in 100% D2O. Protected amide protons do not or only
slowly exchange with the surrounding deuterium of
Table 1.3 Structure determination by NMR spectroscopy. NMR observable parameters (left) are translated into structural restraints (right), which are employed in calculations resulting in a 3D structure ensemble.
As described in chapter 1.4.6, more recent NMR observables that are implemented in
structure calculations are residual dipolar couplings. The most commonly used are the HN-N
Introduction
54
RDCs. They provide information about the directionality of the vector of the bond between
the amide nitrogen and the amide proton.
The experimental information has to be extended by chemical knowledge about bond length,
al.,
1998) and ARIA (Linge et al., 2001). The structure calculation is basically an iterative search
for an energy minimum of a given data set. The result is a fa ily of structures, also called the
ean square
eviation of the position of selected atoms in the individual structures from a mean structure.
Another set of quality checks concerns the accuracy of the structures. This includes the
analysis of backbone dihedral angle values in the Ramachandran plot and of the so-called Q-
factor for a set of RDC c
For the structural analysis of complexes betw of
NO t is employed. Usually, the d and
the ide or a nucleic a as an unlabelled compound in
saturating amounts. The edited-filtered NOESY experiment allows then to either select or
sup o-nucleus y provide
NOE intensities between protons bound to 13C nuclei an C (unlabelled)
nuc ces between protons of the protein and the ligand
can as constraints in structure ca ted complex
ee chapter 2.2).
bond angles and van der Waals radii of the molecule.
The obtained experimental restraints together with the fixed chemical restraints are then
applied in a simulated annealing protocol, e.g. by using the programs CNS (Brünger et
m
NMR ensemble, with the individual solutions fulfilling the input restraints (Table 1.3). The
final ensemble of NMR structures is then refined in a shell of water molecules (Linge et al.,
2003). The quality of the structure and of the input restraints can then be estimated by the
similarity of these structures (precision), which is normally expressed as the root m
d
onstraints.
een a polypeptide and a ligand, a special type
ESY experimen protein under investigation is fully labelle
ligand, may it be a pept cid, is added to it
press magnetization evolution of heter
lei. Thus, short inter-molecular distan
be identified and used
bou rebnd protons and can the
d protons bound to 12
lculations of the investiga
(s
1.4.10 Interaction studies by NMR spectroscopy
NMR spectroscopy is widely used to study the interaction between molecules (reviewed by
Zuiderweg, 2002). Compared to other techniques that report interactions between
biomolecules like electromobility shift assays, Biacore experiments or calorimetry and
fluorescence based assays, NMR based studies have the main advantage that not only binding
is detected but also structural details of the binding process become available. This is because
Introduction
55
the chemical shift can be used as a very sensitive probe that is reporting changes in its
physico-chemical environment.
Chemical shift perturbation mapping
Chemical shift perturbations can be analyzed in principal for all atoms of the biomolecule that
were previously assigned (see chapter 1.4.5). In practice, one is employing most frequently 15N,1H-correlation spectra (15N,1H-HSQC experiments) to detect amides in the protein
backbone that are located in or in close vicinity to the binding site. For this, the spectrum of
the free and the bound form of the biomolecule under investigation are measured and
compared by superimposition. Observation of chemical shift perturbations upon addition of a
ligand reveals atoms in which close spatial proximity the other molecule binds. An example
showing binding of ssRNA to the Drosophila Argonaute2 PAZ domain is shown in Figure
1.22 (see also chapter 2.1).
Figure 1.22 A superimposition of
15N,1H-HSQC spectra for the Ago2 PAZ domain in the absence (blue) and presence (red) of a six fold molar excess of a 21 nt ssRNA.Green arrows indicate the direction of peak movements. The insert shows additional spectra at 0.5-, 1-, 1.5- and 3-fold molar excess of RNA. Significant chemical shifts changes were observed for labelled residues. This example is taken from the publication presented in chapter 2.1 (Lingel et al., 2003).
Chemical shift perturbation measurements just yield the locations of the interfaces on the
individual binding partners. It is then still unknown how the partners interact on an atom-to-
atom basis. This can be achieved by observing inter-molecular NOEs and employing them in
calculations of complex structures (see previous chapter and chapter 2.2).
Introduction
56
Titrations with NMR
In addition to the mapping of the interface, titrations can be carried out to estimate the
ffinity, stoichiometry and the kinetics of the binding. How the chemical shifts of the labelled
rmined by the kinetics of the interaction.
ound protein (see chapter 2.4). During the titration, the ‘free set’
will disappear and will be replaced by the ‘bound set’. This regime is referred to as slow
chemical exchange and is observed usually for high affinity interactions. In slow exchange
one does not automatically know to which new location the resonance has moved, and
therefore a quantitative analysis of chemical shift changes is difficult unless the bound state is
assigned. However, the binding constant can still be quantitated by measuring the intensities
of the disappearing and/or appearing peaks as a function of the ligand concentration.
In the intermediate chemical exchange case, the frequencies of the changing resonances
become poorly defined and extensive kinetic broadening sets in, which might result in the
complete disappearance of resonance lines from the NMR spectrum.
The range of binding constants that can be determined in a quantitative way by NMR is set by
the concentration of the interaction partner that is observed by the NMR experiment. As a
rule, quantitative measures can be obtained for KD values within an order of magnitude of the
concentration of the studied species.
A practical application of interaction studies by NMR is the identification of new drugs, as the
design of drugs initially relies on the detection of binding activity of substances to target
a
protein change during the titration is dete
If the complex dissociation is very fast, there is only a single set of resonances observable
with chemical shifts being the fractionally weighted average of the free and bound chemical
shifts (Figure 1.22 and chapter 2.1). This is referred to as fast chemical exchange and is often
observed for weaker interactions. By titration with increasing amounts of ligand and by
following the peak position in the acquired spectra for each of the changing signals, a data set
is obtained which correlates the amount of ligand with the fractional change of the NMR
signal. This data can be fit to obtain the dissociation constant.
If the complex dissociation is very slow, one observes one set of resonances for the free
protein and one set for the b
proteins. Binding of low molecular weight compounds to large receptor molecules is a
therefore prominent example of the implementation of NMR in applied research. This concept
of identification and characterization of binding compounds is commonly known as structure
activity relationship (SAR) (Shuker et al., 1996).
Results
57
2. Results The major part of the results I obtained during my thesis was published in the following
papers, listed in chronological order:
Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2003) Structure and nucleic acid binding of the Drosophila Argonaute2 PAZ domain. Nature, 426, 465-469. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2004) Nby the Argonaute2 PAZ domain. Nat Struct Mol Biol, 11, 576-
ucleic acid 3’-end recognition 577.
Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2004) NMR assignment of the Drosophila Argonaute2 PAZ domain. J Biomol NMR, 29, 421-432. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2005) The structure of the FHV B2 protein, a viral suppressor of RNAi, reveals a novel mode of dsRNA recognition. Embo Rep, 6, 1149-1155. The following part is divided into four subchapters, one for each publication. Each of them
contains a summary of the results and a statement about my contribution, followed by a
reprint of the original article. The statements of contributions were confirmed by Dr. Elisa
Izaurralde and Dr. Michael Sattler.
Results
58
2.1 Structure and nucleic acid binding of the Drosophila Argonaute2 PAZ domain
he solution structure of the Drosophila Argonaute2 PAZ domain was solved. The construct
n was expressed
combinantly in E.coli und purified to homogeneity. NMR experiments showed that the
A refined NMR structure
as determined from experimentally determined short distances within the protein (NOEs),
easurement of relaxation times and heteronuclear 1H-15N NOE experiments showed that
redefines the domain borders.
n analysis of the structure revealed that the domain is composed of two structural modules, a
-stranded β-barrel, that is covered on one side by two N-terminal helices, and on the other
eic acid binding. As a consequence, the possibility of
ucleic acid binding was explored. Using NMR spectroscopy, binding of the PAZ domain to a
single-stranded 21 nucleotide RNA was detected. The binding site was mapped by chemical
shift perturbation analysis and located to the cleft mentioned above. Mutation of conserved
residues in the binding site of the PAZ domain confirmed the binding site und identified
critical residues for binding, as their mutation strongly reduced the affinity for RNA. Based
on a titration experiment, the dissociation constant for the interaction of the Ago2 PAZ
domain with a 21 nt RNA was determined to be around 20 µM. In addition to NMR titrations,
the binding of nucleic acids was also shown by UV-crosslinking assays, in which the purified
protein could be efficiently crosslinked to ssRNA. Detection of binding by electromobility
shift assays (EMSA) was not successful. To determine the relative affinities of different
Summary of results T
comprised 143 residues and was chosen based on an alignment that initially defined the
boundaries of this conserved domain (Cerutti et al., 2000). The protei
re
purified protein is properly folded. A complete set of spectra was recorded and the resonances
f the polypeptide backbone and side chains atoms were assigned. o
w
HN-N residual dipolar couplings, backbone dihedral angles and hydrogen bonds.
M
20 amino acids at the C-terminus are unstructured in solution. This information together with
the determined secondary structure elements enabled it to make a structure based sequence
lignment that a
A
5
side by a conserved module comprising a β-hairpin and a short α-helix. Inbetween these two
modules, a cleft aligned by basic and hydrophobic residues is formed. The central β-barrel is a
topological variation of the oligonucleotide/oligosaccharide (OB) fold, which is often
implicated in single-stranded nucl
n
Results
59
nucleic acid ligands, competition assays were performed with detection of binding by the
go2 PAZ domain, the DNA even with slightly higher affinity. Furthermore, a double-
tranded RNA with a 19 bp duplex region and a 2 nt 3’-overhang on each side of the duplex
s shown to bind, whereas the 19 bp duplex without the overhangs
howed a reduced affinity. This suggested that the 3’-end of nucleic acids could be critical for
imate my contribution to the manuscript to be
crosslinking experiment. It was found that not only ssRNA, but also ssDNA binds to the
A
s
(an siRNA duplex) wa
s
binding to the PAZ domain. All binding data were confirmed always by both methods, NMR
spectroscopy and the UV-crosslinking assay.
Contribution My contribution to the published data was to clone, express und purify the protein necessary
for all the NMR measurements. In addition, I did part of the NMR data acquisition and
processing. Analysis of the spectra to obtain data for the structure calculation and the
structure calculation itself was done in majority by me. I also did the NMR titration
experiments with RNA. I expressed the PAZ domain mutants as glutathione S-transferase
(GST) fusion proteins and purified them. I also contributed to the writing of the manuscript
and the preparation of figures. Overall, I est
around 50%.
Results
60
Results
61
Results
62
Results
63
Results
64
Results
65
Results
66
Results
67
Results
68
Results
69
Results
70
Results
71
2.2 Nucleic acid 3’-end recognition by the Argonaute2 PAZ domain Summary of results After solving the structure of the free Drosophila Argonaute2 PAZ domain, one important
goal was to get further insight into the interaction between the PAZ domain and nucleic acids.
Therefore, the solution structures of the domain bound to single-stranded RNA and DNA
oligomer ligands were determined. Based on the definition of the exact domain borders of the
PAZ domain in the previous study (chapter 2.1), a construct that is 20 residues shorter than
the construct investigated in the structure determination of the free protein was used. Using
the previously established competition crosslinking assay, a 5 nucleotide long oligomer could
be defined as a ligand and was used in NMR structure determination. Two complete sets of
spectra were measured to determine the complex structures with two ligands, a DNA
oligomer with the sequence CTCAC and the corresponding CUCAC RNA oligomer. As
before, the structures were calculated based on experimentally determined short distances
within the protein (NOEs), HN-N residual dipolar couplings, backbone dihedral angles and on
hydrogen bonds. Inter-molecular short distances between the ligand and the protein revealed
that only the two 3’-terminal nucleotides are bound by the PAZ domain. The interaction
surface of the protein mapped essentially to the region that was identified before as the
binding site for nucleic acids. In both complexes, the interaction is based mainly on
hydrophobic contacts of aromatic and nonpolar residues of the protein with the bases and
sugar backbone of the two 3’-terminal nucleotides. The 3’-OH group of the last nucleotide is
buried deeply inside of the protein, making it impossible to extend the polynucleotide chain
into the 3’-direction. These results enabled the definition of the PAZ domain as a 3’-end
recognition domain.
Contribution I prepared all samples of the Drosophila Argonaute2 PAZ domain needed for the
determination of the two complex structures. Data acquisition and processing was done in part
by me. I did all assignments of the DNA and RNA oligomers and of the protein in its bound
state and analyzed the spectra to get short distances within the protein and between the PAZ
domain and the nucleic acid ligands. I was involved in writing the manuscript and prepared all
figures. My contribution to the publication is around 70%.
Results
72
Results
73
Results
74
Results
75
Results
76
Results
77
Results
78
Results
79
Results
80
Results
81
Results
82
Results
83
Results
84
Results
85
Results
86
Results
87
Results
88
2.3 NMR assignment of the Drosophila Argonaute2 PAZ domain Summary of results In the process of the structure determination of the Drosophila Argonaute2 PAZ domain,
chemical shifts of the protein backbone and side chain atoms were assigned. Triple-resonance
NMR experiments were performed to get essentially complete sequence-specific assignments.
To allow public access to the obtained results, the chemical shifts were deposited in the
BioMagResBank (BMRB) and an assignment note was published.
Contribution My contribution to this publication was to do the major part of chemical shift assignment of
the Drosophila Argonaute2 PAZ domain. Furthermore, I wrote the manuscript and prepared
the figure. The overall contribution is estimated to be 80%.
Results
89
Results
90
Results
91
2.4 The structure of the FHV B2 protein, a viral suppressor of RNAi, reveals a novel mode of dsRNA recognition
Summary of results The structure of the Flock House virus B2 protein was solved by NMR spectroscopy. The full
length protein comprises 106 amino acids and was expressed recombinantly in E.coli. NMR
experiments indicated that a portion of the full length protein is not well defined. Based on
secondary structure predictions and NMR spectra, C-terminally deleted versions of the protein
were made. The shortest folded construct comprised 72 amino acids, lacked flexible residues
and was therefore chosen for further studies. Gel filtration of the purified protein and the line
width of the NMR signals suggested that both the full length and the B2 (1-72) construct are
dimers in solution.
Based on triple-resonance NMR experiments, chemical shift assignment was performed for
the polypeptide backbone and side chains atoms to almost completeness. The structure of the
B2 (1-72) dimer was determined by structure calculations based on experimentally
determined short distances in the protein (NOEs), residual dipolar couplings (RDCs), dihedral
angles and hydrogen bonds.
The NMR structure shows that B2 (1-72) is a symmetric, anti-parallel dimer in solution. The
dimeric conformation is confirmed by NMR relaxation experiments. One monomer is
comprised of three α-helices, which are arranged in a triangular manner and thereby the N-
and C-termini are placed in close proximity. Helix α1 and α2 of one monomer are anti-
parallel and the interaction between them is stabilized by hydrophobic residues.
The B2 dimer is formed by a head-to-tail interaction of the two monomers, with the interface
being stabilized mainly by hydrophobic side chains of residues in helix α1 and α2. In
addition, the dimer is held together by three electrostatic interactions.
To characterize the interaction of B2 with RNA and to map the binding surface, a double-
stranded siRNA was added to the B2 dimer and 15N,1H-correlation spectra were recorded.
Slow exchange in the NMR timescale between the free and the RNA-bound form of the
protein was observed, which indicates that B2 binds the siRNA with a high affinity in the
nanomolar range. Chemical shift perturbations were located mainly at helix α2, with some
additional changes also in the small helix α3. Helix α2 contains a number of positively
charged residues with one of them being arginine 54, a residue previously identified to be
Results
92
critical for dsRNA binding. The spacing of positively charged side chains indicates that B2
. This
in-line with the reported sequence-independent binding specificity of B2. Based on a
tration experiment, the stoichiometry of the dsRNA-B2 dimer complex could be determined
the observed length of the binding site, this suggests that the siRNA
ulations themselves. Furthermore, I performed the characterization of the
contacts the dsRNA primarily by polar interactions with the sugar phosphate backbone
is
ti
to be 1:1. Together with
binds with its helix axis oriented in parallel to the α2 helix axis.
Contribution I cloned, expressed and purified all constructs of the B2 protein and prepared all NMR
samples. I also did part of the data acquisition and processing. I did the complete chemical
shift assignment, the analysis of the spectra to get constraints for the structure calculations
and the structure calc
dsRNA binding to the B2 protein by NMR. I was involved in writing the manuscript and
prepared all the figures. I estimate my overall contribution to this publication to be 80%.
Results
93
Results
94
Results
95
Results
96
Results
97
Results
98
Results
99
Results
100
Results
101
Results
102
Discussion
3. Discussion
.1 The PAZ domain represents a novel nucleic acid binding fold
he PAZ domain is one of the two signature domains of Argonaute proteins and was initially
entified by a bioinformatics approach which searched for conserved sequence blocks in
roteins involved in RNA interference and cell differentiation (Cerutti et al., 2000). It was
und that the PAZ domain is exclusively present in only two protein families: the Argonaute
rotein family, both of them being known to constitute core components of the
initiation and effector step of RNA silencing, respectively (Figure 1.1). In the course of RNAi
or miRNA-mediated gene silencing, the small RNAs generated by Dicer are transferred into
RISC to perform their function in the effector complex. A direct transfer seems likely, as
Dicer was shown to interact with Argonaute proteins in several studies (Hammond et al.,
2001; Caudy et al., 2002). This was the reason for the proposal that the PAZ domain might act
as a protein-protein interaction domain, mediating physical contact between Dicer and
Argonaute proteins.
Structure of the PAZ domain
The structure of Drosophila Argonaute2 PAZ domain, presented in chapter 2.1 of this thesis,
provided the first molecular insight into the function of this conserved protein domain. The
structure revealed that the domain has a bimodular composition, with a cleft formed
inbetween the two modules. The core of the PAZ domain is formed by a central five-stranded
β-barrel that resembles structures of known β-barrels, such as the Sm fold and the
oligosaccharide/oligonucleotide binding (OB) fold (Figure 3.1, dark blue), which were both
shown to have a function in nucleic acid binding. The Sm fold is shared by members of the
LSm protein family and is known to form hexameric or heptameric rings that bind to single-
stranded RNA (reviewed by Khusial et al., 2005). It constitutes a closed barrel comprising 5
anti-parallel β-strands with an α-helix on one side (Figure 3.1B). RNA binding is found not
for the monomer alone, but only when the protein is part of a multimeric ring. The binding
site is located in the lumen of the ring with one nucleotide binding to each subunit (Thore et
al., 2003) and is formed of the loops between strands β2 and β3 and between β4 and β5 (see
Figure 3.1B, black arrows). Members of the closely related OB fold superfamily consist of
two 3-stranded anti-parallel β-sheets, where strand β1 is shared by both sheets (Figure 3.1C,
3
T
id
p
fo
and the Dicer p
103
Discussion
104
Theobald et al., 2003). The observed variability in length among OB fold domains is due to
ngth of the loops between the well-conserved elements of secondary
tructure. The nucleic acid binding subfamily of OB fold proteins is the largest within the OB
nding. OB
differences in the le
s
superfamily, and members of this subfamily are involved in ssRNA or ssDNA bi
folds tend to use a common ligand binding interface centered on β-strands 2 and 3 (Figure
3.1C, arrow).
The OB-like barrel of the DmAgo2 PAZ domain is covered on one side by two N-terminal
helices and on the other side by an inserted β-hairpin/α-helix module (Figure 3.1A, light
blue), which is around 35 amino acids long and contains several highly conserved residues.
One of them is a phenylalanine (F72) in strand β4, which is invariant in PAZ domains.
Inbetween the central β-barrel and the conserved module, a hydrophobic cleft can be
identified. This domain architecture of a β-barrel covered by a β-hairpin/α-helix module was
not observed in previously solved structures and thus represents a novel domain fold.
The structure of the DmAgo2 PAZ domain reported in chapter 2.1 was confirmed by a crystal
structure of an MBP-DmAgo2 PAZ fusion protein that was solved by Song et al. shortly
afterwards (Song et al., 2003). The two structures determined by different methods are
Figure 3.1 Comparison of β-barrel protein folds. The strands of the central β-barrel are colored in dark blue in all panels. (A) The DmAgo2 PAZ domain, with the conserved insertion shown in light blue. (B) Sm fold of the human Sm D2 protein (PDB code 1B34, chain B) (Kambach et al., 1999). The black arrows indicate the common Sm fold binding site. (C) OB fold of the E. coli Rho transcription termination factor (PDB code 1A8V) (Bogden et al., 1999). The black arrow indicates the common OB fold nucleic acid binding site on β2 and β3.
A B C
Discussion
105
essentially identical. The same is true for the PAZ domain of DmAgo1, that was also solved
as a solution structure and published simultaneously (Yan et al., 2003). Both structures show
the same secondary structure and domain architecture.
Function of the PAZ domain in nucleic acid 3’-end recognition
Unlike previously suggested, the PAZ domain was found to bind nucleic acids in vitro.
inding of ssRNA, ssDNA, and dsRNA with 3’-end overhangs to the DmAgo2 PAZ domain
bserved low affinity for blunt-ended dsRNA and double-stranded RNA with 5’-overhangs.
human Ago1 PAZ domain and for the PAZ domains of all four Argonaute proteins of
B
could be detected by chemical shift perturbation analysis and crosslink experiments (chapter
2.1). Using NMR, the binding site for a single-stranded siRNA could be mapped to be in the
hydrophobic cleft between the β-barrel and the β-hairpin/α-helix module. Furthermore, two
critical residues for nucleic acid binding, namely F50 and F72, which are located on opposite
faces of the clamp in the PAZ domain, could be identified by mutational analysis. Notably,
the two mutant proteins were still folded as confirmed by NMR spectroscopy, but showed no
chemical shift changes upon addition of ssRNA and were not able to form a photo-induced
product in the UV-crosslinking assay. Competition experiments revealed that ssRNA and
ssDNA are bound with similar affinities by the DmAgo2 PAZ domain. Different siRNAs
competed equally well in the assay, indicating that the RNA is bound with little sequence
preference. The observation that blunt-ended dsRNA molecules were bound with a reduced
binding affinity to the PAZ domain pointed towards a specificity for ssRNA.
These results were in-line with UV-crosslinking experiments that were reported together with
the crystal structure of the DmAgo2 PAZ domain (Song et al., 2003). The authors showed
binding of a single-stranded siRNA and of double-stranded siRNAs with 3’-overhangs, but
o
This suggested a role of a single-stranded 3’-end in nucleic acid binding.
As the identified region of interaction includes highly conserved residues, it was suggested
that also other PAZ domains function in nucleic acid binding (chapter 2.1). Indeed, in the
simultaneously published study of the Drosophila Argonaute1 PAZ domain, Yan et al.
showed binding of a small, single-stranded RNA oligonucleotide (Yan et al., 2003). The
authors also reported a reduced affinity for DNA, showing that the DmAgo1 PAZ domain
discriminates between RNA and DNA, whereas the Ago2 PAZ domain does not (chapter 2.1
and 2.2). Furthermore, RNA binding was shown by NMR for the PAZ domains of human
Ago1, Ago2 and Piwi proteins (Yan et al., 2003).
In agreement with that, binding of nucleic acids was detected by the crosslinking assay for the
Discussion
106
Drosophila (A. Lingel and E. Izaurralde, unpublished). In these experiments, PAZ domains
showed a clear preference for RNA over DNA, except for the DmAgo2 PAZ domain (see
above). If this difference in specificity is employed in vivo in different RISCs is presently not
nown.
o2-RNA (Y44)
ting that the 3’-end
nd makes only a few contacts with the PAZ
k
The molecular basis of nucleic acid recognition has been revealed by NMR solution structures
of the DmAgo2 PAZ domain bound to single-stranded RNA and DNA oligonucleotides and
by the characterization of the binding of an siRNA mimic to the PAZ domain, as described in
chapter 2.2 of this Ph.D. thesis. In both complex structures, only the two 3’-terminal
nucleotides are bound at similar binding sites, which are located in the previously identified
hydrophobic cleft of the PAZ domain (Figure 3.2A). This established the molecular function
of the PAZ domain as 3’-end recognition. Simultaneously, the crystal structure of the HsAgo1
PAZ domain in complex with an siRNA mimic was published (Ma et al., 2004) and
confirmed these results. Like in the structures of the DmAgo2 PAZ complexes, the 2 nt 3’-
overhang of the siRNA-like ligand is bound in the same cleft (Figure 3.2B). Recognition of
the 3’-terminal nucleotide involves a stacking interaction with a conserved aromatic residue
(F292 and F72 in HsAgo1 and DmAgo2, respectively) of the β-hairpin/α-helix module. The
phosphate groups of the two 3’-terminal nucleotides interact with tyrosine and histidine
residues in the HsAgo1-RNA (Y309, Y314, H269 and Y277) and DmAg
complexes. It is conceivable that some of these residues (such as Y314 and H269, which are
not conserved in DmAgo2) contribute to the selectivity of most PAZ domains for RNA over
DNA. This may also be linked to the different conformations of the sugar phosphate
backbone in the complexes of DmAgo2 with RNA and DNA (chapter 2.2).
Very few contacts involve the sugar of the 3’-terminal nucleotide, indica
of the nucleic acid is specified mainly by steric exclusion. In the complexes of the DmAgo2
PAZ domain with single-stranded nucleic acids, the penultimate nucleotide interacts with
conserved residues Y57 and K107. Notably, this binding pocket for the penultimate
nucleotide corresponds to the location of a single uridine nucleotide bound to the DmAgo1
PAZ domain (Yan et al., 2003), suggesting that it is a high-affinity site. By contrast, in the
complex of HsAgo1 with an siRNA mimic, the base of the penultimate nucleotide is in a
different more exposed location (Figure 3.2B, Ma et al., 2004). It acts as a spacer connecting
the 3’-terminal nucleotide with the dsRNA stem a
domain. This observation indicates that there is conformational plasticity with respect to the
location of the penultimate nucleotide. The mode of recognition of this nucleotide may
Discussion
107
depend on whether the upstream nucleic acid adopts a single-stranded or double-stranded
conformation. In both studies, the double-stranded stem of the different siRNA mimics
extends along the β3 strand of the central β barrel of the PAZ domain (chapter 2.2 and Ma et
al., 2004).
The phosphodiester backbone of the RNA strand which is bound via its 3’-end mediates
numerous contacts to positively charged arginine and lysine side chains (Ma et al., 2004).
These additional contacts depend on the presence of the dsRNA stem and contribute
substantially to the binding affinity, as the extension of the 3’-end single-stranded overhang
from two to ten uridines reduces the interaction 50-fold (corresponding to a change in the
dissociation constant, KD, from ~2 nM to 100 nM). Notably, the complementary RNA strand,
including its 5’-end, does not seem to be recognized by the PAZ domain (Figure 3.2B).
Figure 3.2 Recognition of the 3’-end of nucleic acids by the PAZ domain. The central β-barrel and the insertion module are shown in dark and light blue, respectively. Amino acid side chains that are involved in nucleic acid recognition (see text) are indicated. (A)DmAgo2 PAZ bound to a single-stranded 5 nt RNA (PDB code 1T2R). Only the two 3’-terminal nucleotides that interact with the protein are shown. (B) HsAgo1 PAZ bound to
an siRNA mimic (PDB code 1SI3). Only one half of the protein-RNA dimer iunit of the crystal structure is shown. (C) The OB fold domain of S. pombe Pot
n the asymmetric 1 (SpPot1) bound
to telomeric DNA (PDB code 1QZG) (Lei et al., 2003).
A B
C
Discussion
108
Taken together, the structural studies on the Drosophila Argonaute2 PAZ domain show that
the PAZ domain binds the two 3’-terminal nucleotides of single-stranded nucleic acids by
burying them in a hydrophobic cleft between the characteristic β-hairpin/α-helix module and
a central β-barrel.
ucture
lements suggests that the overall fold of Dicer PAZ domains is similar to that of Argonaute
The mode of recognition of the two terminal nucleotides is unique and distinct from that used
by other single-stranded nucleic acid binding domains, like the Sm or the OB fold (Theobald
et al., 2003; Messias & Sattler, 2004). In strong contrast to OB and Sm fold proteins, where
the solvent-exposed surface of the β-barrel is used for nucleic acid recognition (Figure 3.1),
the major part of the interaction surface of the PAZ domain involves the inner part of the
central β-barrel near β7. An even more significant difference is the presence of the highly
conserved β-hairpin/α-helix module in the PAZ domain, which is absent in the OB fold
(Figure 3.1). The burial of the 3’-end by the monomeric PAZ domain also differs from 3’-end
recognition by OB fold domains, such as the recognition of telomeric DNA by the OB fold of
Schizosaccharomyces pombe Pot1 (Figure 3.2C, Lei et al., 2003). SpPot1 engages in several
base- and DNA-specific contacts, including intra-DNA interactions, consistent with sequence-
specific recognition and its preference for DNA over RNA. The 3’-end of the single-stranded
telomeric DNA is exposed and recognized by specific contacts between its 3’-terminal
hydroxyl group and the side chains of arginine and threonine residues (Figure 3.2C).
Thus, the PAZ domain represents not only a novel nucleic acid binding fold, also the mode of
selectively binding the 3’-end of nucleic acids was not precedenced.
Diversity of PAZ domains
In eukaryotes, only Dicer and Argonaute proteins contain the PAZ domain. Unlike for the
PAZ domain of Argonaute proteins, little is known about the Dicer family PAZ domains.
Sequences of Dicer PAZ domains align well with the sequences of Argonaute proteins, except
for a stretch of approximately 30 amino acids in Dicer PAZ domains that are inserted in the
loop between the β6 and β7 strand of Argonaute PAZ domains (for sequence alignments, see
Supplementary Figure 1 in chapter 2.1 and Supplementary Figure 2 in chapter 2.2). This
insertion is in close proximity to the nucleic acid binding site and is rich in arginine and lysine
residues. This suggests that it might be involved in nucleic acid binding, but it could modulate
the function of the PAZ domain. Conservation of hydrophobic residues in secondary str
e
Discussion
109
proteins. A proposed model for the function of Dicer PAZ domains will be discussed in the
next chapter.
Initially, the PAZ domain was believed to occur only in eukaryotes, as there were no
conserved sequences identified in prokaryotes (Cerutti et al., 2000). Later, crystal structures
interactions with conserved aromatic residues. Residues Y190 and Y216 of PfAgo are
of archaeal proteins with identified Piwi domains revealed that the PAZ domain is not
restricted to eukaryotes. First, Song et al. solved the structure of an Argonaute protein of
Pyrococcus furiosus (Song et al., 2004) and found a PAZ domain which can be superimposed
with eukaryotic PAZ domains, despite very low sequence identity. However, the structural
similarity is not complete, and there are structural elements unique to the PfAgo PAZ domain.
The main difference is that in the PfAgo protein, the β-hairpin/α-helix module is replaced by
two α-helices (Figure 3.3). In addition, some loop conformations are different. Despite these
variations, amino acids reported to be important for binding to the 3’-end of nucleic acids (see
chapter 2.2), including those in the β-hairpin/α-helix module, are conserved. Their side chains
are located in similar spatial positions in the Pyrococcus protein, although some of these
residues are at different positions in the protein sequence and protrude from different
secondary structural elements (Figure 3.3).
As described above, key features of nucleic acid recognition by the PAZ domain involve base
B
Figure 3.3 Comparison of the Drosophila Argonaute2 PAZ domain (DmAgo2, A) and the PAZ domain of the Argonaute protein from Pyrococcus furiosus (PfAgo, B). Structures are shown as ribbon diagrams. The β-hairpin/α-helix module is shown in light blue. This module contains two α-helices in the PAZ domain of PfAgo. Side chains of residues shown to contact nucleic acids inthe DmAgo2 PAZ domain are shown and labeled. Residues in equivalent positions in the PfAgo protein are indicated.
A
Discussion
110
equivalent to Y57 and Y84 of the DmAgo2 PAZ domain (Figure 3.3). The position of residue
F72 of the DmAgo2 PAZ domain, which was identified to be critical for nucleic acid
providing module in the RNAi
recognition, is occupied by another aromatic residue, W213, in PfAgo. On the basis of these
similarities, Song et al. suggested that the PAZ domain of PfAgo could also bind to single-
stranded 3’-ends of RNAs.
A PAZ domain is also found in the crystal structure of the Argonaute protein of another
archaea, Aquifex aeolicus (Yuan et al., 2005). Similarly to the PAZ domain of the Argonaute
protein of Pyrococcus furiosus, the β-hairpin/α-helix module found in eukaryotic PAZ
domains is replaced by two α-helices. Also like in the PfAgo PAZ domain, conserved
residues occupy similar spatial positions with their counterparts in other PAZ domains (Figure
3.3), suggesting a conserved function in 3’-end binding. However, nucleic acid binding to
archaeal PAZ domains has not been demonstrated yet and thus the function of this class of
PAZ domains remains to be established.
3.2 The PAZ domain as a specificitypathway
The structural studies on eukaryotic Argonaute PAZ domains revealed that the PAZ domain
provides a unique platform for the recognition of the two 3’-terminal nucleotides in single-
stranded nucleic acids. A key feature of the binding is the burial of the terminal 3’-OH group.
This binding mode provides new insights into the biological role of this domain within Dicer
and Argonaute proteins, which constitute the cores of RNA silencing complexes.
In the initiation step of RNA-mediated gene silencing, the RNase III enzyme Dicer generates
small double-stranded RNAs. Based on the established function of the Argonaute PAZ
domain, a model of PAZ function in Dicer was proposed by Filipowicz and co-workers in a
recent study addressing the function of human Dicer (Zhang et al., 2004). The authors suggest
that the characteristic size (~20 bp) of the siRNAs resulting from Dicer cleavage may be
linked to 3’-end recognition of the dsRNA substrate by the Dicer PAZ domain. In this model,
the PAZ domain recognizes the 2 nt 3’-overhang generated by Drosha in the miRNA
pathway, or the last cut made by Dicer in the RNAi pathway. PAZ and RNase III domains of
Dicer are spaced such that when the 3’-overhang is bound, they act as a ruler to position the
cleavage site exactly 22 nt away from the 3’-end. This model is consistent with the known
preference of Dicer for free ends of dsRNAs (Zhang et al., 2002). This results in Dicer
Discussion
111
cleavage at the ends of dsRNA and pre-miRNAs, shortening the substrate consecutively by
siRNA-sized segments, and not cutting internally, which would lead to the generation of
multiple larger fragments (Zhang et al., 2002; Zhang et al., 2004; Vermeulen et al., 2005). In-
line with this model, both the PAZ and dsRBD domains are required for full Dicer activity.
Intact Dicer can slowly process blunt-ended dsRNA, however deletion of the dsRBD makes
cleavage of long dsRNA, the Dicer-2-R2D2-siRNA
rnary complex is an assembly intermediate in the formation of RISC (Pham et al., 2004;
Tomari et al., 2004b) that is proposed to recruit Ago2 directly to the siRNA. This interaction
uplex is
hereb d from the initiator complex into the RISC effector complex, as it was
Dicer more dependent on substrates with 3’-overhangs (Zhang et al., 2004). This suggests that
the dsRBD contributes to non-specific binding of incoming substrates and potentially allows
Dicer to cleave long dsRNA substrates processively. As no experimental structural data is
available for the PAZ domain of Dicer proteins, the exact details of the catalytic mechanism
and the role of the PAZ domain must await the structure of a Dicer-dsRNA complex.
In Drosophila, Dicer-2 is stably associated with the dsRBD containing protein R2D2 (Liu et
al., 2003), and they are together required for cleavage of dsRNA and for RISC assembly (Lee
et al., 2004b; Pham et al., 2004). After
te
might be mediated directly by Ago2 and Dicer-2 (Tahbaz et al., 2004). The siRNA d
y transferret
demonstrated recently (Matranga et al., 2005). The results on the PAZ domain of Argonaute
proteins presented in this thesis show that it specifically recognizes the 2 nt 3’-overhang of
siRNAs. This suggests that the PAZ domain is involved in receiving the siRNA from Dicer
and to facilitate its incorporation into RISC.
The role of the PAZ domain in the subsequent catalytic cycle of RISC can be envisioned as
follows: The PAZ domain of Ago2 binds the 3’-end of the strand that is considered the guide
strand of the siRNA duplex. At the same time, the 5’-end of the guide strand is docked in the
phosphate binding pocket of the Ago2 Piwi domain (see chapter 1.2.4). The decision which
strand of the siRNA duplex functions as guide strand is determined by the orientation of the
siRNA in the RISC loading complex. The passenger strand would then be cleaved by Ago2
and the resulting fragments released (Matranga et al., 2005; Rand et al., 2005). This release
might be facilitated by an ATP-dependent cofactor, like the release of the products of target
mRNA cleavage which was shown to be stimulated by ATP (Haley & Zamore, 2004). The
resulting RISC contains Ago2 with the single-stranded guide RNA bound. This ‘mature’
RISC provides the sequence-specific binding site for the complementary mRNA. The mRNA
can then base pair with the guide strand and is cleaved by the endonucleolytic activity
Discussion
112
constituted by the Piwi domain of the Argonaute protein (see chapter 1.2.3). The reaction
cycle is completed with the release of the cleavage products.
It is not established yet that the PAZ domain interacts with the 3’-end of double-stranded or
single-stranded siRNAs in vivo. Experiments testing this are possible and will be done
certainly in the future. A strong indication that the PAZ domain is important for RNAi
function in vivo is provided by results of a genetic screen in C. elegans (Tabara et al., 1999).
Worms deficient in RNAi were isolated in which only an absolutely conserved glutamate in
the PAZ domain of Rde-1 is mutated to lysine. The structure of the Drosophila Ago2 PAZ
domain suggests that this mutation destabilizes the fold, which most probably also disrupts
the structural integrity of the PAZ domain in vivo.
Despite the diversity of functions attributed to the different Argonaute-associated complexes
(see chapter 1.2.1), a common feature of these complexes is the use of non-coding RNAs to
select their targets in a sequence-specific manner. These non-coding RNAs (siRNAs or
microRNAs) have the specific features of RNAs processed by the RNase III enzyme Dicer in
common, i.e. a 19-22 nucleotide double-stranded RNA body with two nucleotide 3’-
overhangs and 5’-monophosphate groups. This raises the question of how siRNAs or
miRNAs are discriminated from other cellular RNAs and specifically incorporated into
RISCs. With the binding specificity of the PAZ domain of Argonaute proteins, it is likely that
it provides a contribution to the specific recognition of the 2 nt 3’-overhang feature of
siRNAs. It would then act as a selectivity filter in RISCs, allowing only small RNAs
processed by Dicer to enter the effector complex and excluding all other RNAs that are
present in the cell. Without such a filter, all double-stranded RNA that is present in a cell
could enter the RNAi pathway. This selectivity is very important, as the incorporation of
unwanted dsRNA would lead to the down-regulation of the expression of complementary
genes.
Discussion
113
3.3 The FHV B2 protein binds dsRNA in a novel
mode
as previously shown to act as a potent
ppressor of RNAi in insect cells, C. elegans and plants (Li et al., 2002; Lu et al., 2005). The
e distance over the minor groove of A-form dsRNA. These observations, together
ith the determined 1:1 binding stoichiometry and the fact that the complete α2/α2’ surface
dsRNA suggests that the dsRNA binds parallel to the α2 helix
axis. The length of the RNA binding surface is about 45 Å, which corresponds to an
approximately 17 base pair dsRNA. This proposal was confirmed by a crystal structure of a
B2 dimer bound to a palindromic 18 bp RNA duplex that was published simultaneously
(Chao et al., 2005). In this protein-RNA complex, the α2/α2’ helices of the B2 dimer are
indeed oriented almost parallel to the helical axis of the A-form RNA (the angle between the
To counteract the cellular response to the presence of dsRNA, viruses have evolved proteins
that can act as suppressors of RNA interference. The detailed mechanism of suppression is
unknown for most of the identified proteins, with the exception of the p19 protein of
tombusviruses (see chapter 1.3). p19 was shown to bind specifically to siRNAs and thereby
blocks their incorporation into RISC (Vargason et al., 2003; Ye et al., 2003).
As part of this thesis, the solution structure of the B2 protein of Flock House virus was
determined and is described in chapter 2.4. B2 w
su
NMR structure shows that B2 is an anti-parallel, symmetric dimer in solution with each
monomer composed of three α-helices. Searches in databases of known structures did not
result in significant hits, indicating that this dimer fold is novel. Binding of dsRNA was
detected by chemical shift perturbation analysis and the binding site could be mapped to an
elongated surface formed by the two α2 helices of the dimer (α2/α2’). Notably, the helices
are highly positively charged, as many lysine and arginine residues are located on this surface.
Arginine 54, which has been shown to be crucial for dsRNA binding (Lu et al., 2005), is
located at the centre of this binding interface, which confirmed the importance of residues in
the identified binding surface. The exposed arginine and lysine residues suggest that
recognition of dsRNA is mediated by non-sequence-specific electrostatic contacts with the
sugar phosphate backbone of the dsRNA stem. Interestingly, the spacing of positively charged
side chains corresponds to known distances in regular A-form dsRNA. R54 and K47 are
separated with a spacing of approximately 10 Å (Figure 3.4A), which is approximately the
distance of phosphates across the major groove. Furthermore, R36 and K62 are positioned to
bridge th
w
provides the binding site for
Discussion
114
B2 dimer axis and the axis of the dsRNA is ~10°) and B2-RNA interaction is made up almost
gure 3.4B). entirely of contacts with the phosphate backbone and ribose sugars (Fi
As suggested by the solution structure of B2, K47 and R54 of both α2 and α2’ helices contact
the phosphate backbone in the major groove, whereas R36 and K62 contact the phosphates
over the minor groove. Both the monomer and the dimer folds of the B2 dimer in complex
with dsRNA are very similar to the solution structure of the free B2 protein (backbone
coordinate r.m.s.d. 1.2/1.3 Å for monomer/dimer). A slightly different orientation of the α3
helices is potentially linked to electrostatic contacts of residues in the preceding linker with
the RNA. In addition, the helix dipole moment of the small helix α3 is oriented towards the
RNA and the partial positive charge could be also used to interact with the phosphate
backbone.
Figure 3.4 Comparison of the solution structure of the free B2 dimer (A) with the crystal structure of the B2 dimer in complex with a palindromic 18 base pair dsRNA (B). The two monomers of the B2 dimer are shown in dark und light blue, respectively. Lysine and arginine residues in the RNA binding surface (helices α2 and α2’) mediating contacts to the phosphate backbone are shown in both structures and labelled in A. Distances between these residues and the total length of the RNA binding site are indicated (see also text). The major groove (MG) and minor grooves (mg) in the dsRNA are indicated.
A B
Discussion
115
The structure of the B2 dimer reveals a novel mode of sequence-independent dsRNA
recognition, a feature that is shared with the double-stranded RNA binding domain (dsRBD,
ia coli. The Rop protein forms a four-helical bundle and binds an RNA-kissing
op structure. Several models of the Rop-RNA complex have been built on the basis of
also called double-stranded RNA binding motif, dsRBM). The dsRBD comprises about 70-75
amino acids and adopts an αβββα fold in which the two α-helices pack against one face of
the 3-stranded β-sheet. It uses residues in both the α-helical and β-strand loop regions for
RNA binding (Figure 3.5, Ryter & Schultz, 1998; Stefl et al., 2005). The dsRBD it found in
many RNA binding proteins with various functions, including pre-mRNA editing, RNA
transport, and RNA-mediating silencing. In the dsRBD, the α-helix 1, the N-terminal part of
α-helix 2, and the loop between β-strands β1 and β2 recognize the shape of the dsRNA stem.
Like the B2 dimer, the dsRBD discriminates between dsRNA and dsDNA but has no
sequence-specificity. Both the dsRBD and the B2 dimer recognize two minor grooves and the
intervening major groove along one face of the dsRNA, but the recognition is performed by
completely different protein architectures and backbone contacts (Figure 3.4 and Figure 3.5).
Thus, B2 and the canonical dsRBD represent distinct solutions to the problem of sequence-
independent dsRNA recognition.
Figure 3.5 Recognition of dsRNA by the dsRBD. The dsRBD of the Xenopus laevis RNA-binding protein A (Xlrbpa2) is shown in complex with dsRNA (Ryter & Schultz, 1998). The protein interacts with two successive minor grooves and across the intervening major groove on one face of the dsRNA. Recognition of the shape of the dsRNA stem is achieved by resdues in helices α1 and α2 and by residues in the loop connecting β1 and β2. Side chains that interact with the RNA are shown.
Notably, the topology of the four-helical bundle formed by α1, α2, α1’and α2’ in the B2
dimer resembles the small dimeric Repressor of primer (Rop) protein (Figure 3.6B, Banner et
al., 1987), which was shown to be important for the regulation of ColE1 plasmid copy number
in Escherich
lo
Discussion
116
mutagenesis and NMR chemical shift mapping (Predki et al., 1995). The predicted binding
interface with kissing RNA hairpins comprises an anti-parallel pair of the first helix (α1) of
each monomer, somewhat resembling the B2 dimeric RNA binding interface. But in contrast
to B2, in which all four helices are almost parallel to each other, Rop forms a canonical left-
handed four-helical bundle, as reflected in the rather large r.m.s.d. of the atomic backbone
coordinates (3.7 Å) between the B2 and Rop dimers (Figure 3.6A,B).
BA C
Similarly, the N-terminal RNA binding domain of the Influenza A virus NS1 protein is a
symmetric homodimer formed by two three-helix monomers (Figure 3.6C, Chien et al., 1997;
Liu et al., 1997). It corresponds to the first 73 amino acids of the NS1 protein and possesses
all of the dsRNA binding activity of the full length protein (Qian et al., 1995). Site-directed
mutagenesis experiments indicated that the side chains of positively charged amino acids in
Figure 3.6 Small, dimeric and α-helical proteins that bind dsRNA. The two monomers are shown in each panel in dark and light blue, respectively. Side chains of amino acids that were shown to be important for dsRNA binding are shown, placing the binding site for dsRNA on the right side.(A) Solution structure of the B2 dimer as described in chapter 2.4. (B) Crystal structure of the Rop protein of E. coli (Banner et al., 1987). (C) Crystal structure of the dsRNA binding domain of the Influenza A virus non structural protein NS1 (Liu et al., 1997).
e second α-helix (α2) are the only amino acid side chains that are required for the dsRNA th
binding activity of the intact dimeric protein (Figure 3.6C).
Despite these similarities, the structure of B2 reveals clearly a novel fold and an
unprecedenced way of interaction with dsRNA.
Discussion
117
3.4 Viral suppressors of RNAi act at different levels in the RNAi pathway The B2 protein of FHV was the first identified suppressor of RNA silencing of an animal
virus (Li et al., 2002), after the presence of suppressors of RNAi was already established for
plant viruses. The fact that FHV suppresses RNA silencing during replication in plant, insect
tep of RNAi (Lu et al., 2005). B2
affinity. Interestingly, inhibition of Dicer
leavage of long dsRNA after addition of B2 was indeed demonstrated very recently in vitro
(Chao et al., 2005; Lu et al., 2005). This is supported by recent studies on the B2 protein of
and worm hosts suggested that B2 has evolved a host-independent evasion mechanism (Li et
al., 2002; Lu et al., 2005). The solution structure of the FHV B2 protein dimer and the novel
dsRNA binding mode of contacting the sugar phosphate backbone to enable sequence-
independent recognition of dsRNA provide an explanation of the suppressor activity on a
molecular level. These results are consistent with a suppression of RNAi by B2 at two
different stages: first, it can bind the viral dsRNA that is generated during viral replication,
thereby protecting it from being cleaved by Dicer. Secondly, in the case viral siRNAs are
generated, they can be bound by B2 blocking their incorporation into RISC (Figure 3.7).
Figure 3.7 Suppression of RNAi by the FHV B2 protein. B2 can act at two stages of the RNAi pathway: by binding viral dsRNA it can block the initiation step(Dicer cleavage), preventing the generation of siRNAs. Like p19, it can also bind to siRNAs and thereby taking them out of the pathway.
This model is in-line with genetic analysis in C. elegans showing that B2 is active in both
wild-type worms and worms that are deficient of the Argonaute protein Rde-1, which
indicated that B2 might act upstream of the RISC-mediated s
has been demonstrated to bind siRNAs in vitro, but unlike p19 of tombusviruses, B2 can also
bind to long dsRNA, even with much higher
c
Discussion
118
Nodamura virus (NoV, Sullivan & Ganem, 2005). The two B2 proteins share only little
,
lmost all of the residues that contact the RNA are conserved between these proteins. Sullivan
es a
econd mechanism to protect the viral RNA from being cleared from the cell. The structure of
sequence similarity, but probably adopt similar structures (Johnson et al., 2001). Interestingly
a
et al. showed that NoV B2 is able to inhibit processing of long dsRNA into siRNAs in virus
infected mammalian cells (Sullivan & Ganem, 2005). Furthermore, cellular NoV B2 was
found to be associated with small ~20 nt RNAs which derived from a longer precursor. In
addition to the inhibition of Dicer in vivo, the authors showed the binding of B2 to dsRNA
and blocking of Dicer in vitro.
In conclusion, the B2 protein represents the first example of a dual mode of suppression of
RNAi: it is not only targeting siRNAs as it was demonstrated for p19, but also the long
precursor RNA that becomes thereby protected from Dicer cleavage. The protection of the
viral dsRNA might be not complete, and a certain fraction of the replication intermediates
might be processed anyway. Then, the binding of the generated siRNAs by B2 provid
s
the B2 dimer and the characteristics of dsRNA binding described in chapter 2.4 of this thesis
provide an explanation for that on a molecular level.
The list of identified suppressors is long (Table 1.1) and future studies on the molecular
function of these proteins might reveal novel modes of RNAi inhibition that are unknown to
date. Suppressors of RNA silencing may provide very useful tools to study RNA silencing
mechanisms, e.g. to identify endogenous targets of siRNA or miRNAs.
Additional Publications
119
4. Additional Publications
In addition to the publications containing primary data, I was involved in writing one
perspective and one review during the time of my Ph.D. thesis:
Perspective Lingel, A., and Izaurralde, E. (2004) RNAi: Finding the elusive endonuclease. RNA, 10, 1675-1679.
Review
Lingel, A., and Sattler, M. (2005) Novel modes of protein-RNA recognition in the RNAi pathway. Curr Opin Struct Biol, 15, 107-115.
For each of the manuscripts, a summary of the discussed topic is given, followed by the
reprint of the publication.
Additional Publications
120
4.1 RNAi: Finding the elusive endonuclease
ummary
e Dicer containing complex
to smaller RNAs, called siRNAs. These are transferred and incorporated into a second
omplex, which contains an Argonaute protein as key player and name giving component.
called the effector complex, as it targets an mRNA that is
plex was termed Slicer, according to the activity that cleaves RNA in the first
tep, Dicer. The molecular identity of Slicer remained unknown despite intensive studies until
Then, Song et al. solved the crystal structure of an Argonaute protein of the archaebakterium
adopts an RNase H fold. This class of enzymes is known to cleave the RNA strand of an
cture, Liu et al. did mutations in the putative active site
f human Argonaute proteins and showed that the Argonaute2-associated complex is not
rgonaute2 protein itself is Slicer came from results of Tuschl and
coworkers (Meister et al., 2004). They showed that of different Argonaute-associated
plexes purified from human cells, only the Ago2 complex was cleavage competent.
This perspective summarizes the results of these three papers which led to the identification of
Ago2 as the molecular identity of Slicer.
S As described in the introduction, the process of RNAi can be divided into two steps. Briefly,
in the first step, dsRNA is cleaved by the RNase III-type enzym
in
c
This complex is also
complementary to the siRNA for degradation. The endonucleolytic activity of this Argonaute-
associated com
s
the middle of the year 2004.
Pyrococcus furiosus (Song et al., 2004). The key result of their study is that the Piwi domain
RNA:DNA hybrid. Based on this stru
o
longer capable of cleaving a target mRNA (Liu et al., 2004). Additonal evidence for the
hypothesis that the human A
com
Additional Publications
121
Additional Publications
122
Additional Publications
123
Additional Publications
124
Additional Publications
125
Additional Publications
126
4.2 Novel modes of protein-RNA recognition in the RNAi pathway
Summary Over the last couple of years, an increasing number of structures of proteins or protein
domains involved in RNAi were determined. These structures include PAZ domains, either in
their free form or in complex with nucleic acid ligands, the structure of the Argonaute protein
of the archaea Pyrococcus furiosus, structures of RNase III-type enzymes and crystal
structures of the p19 protein, a viral suppressor of RNA silencing, in complex with an siRNA.
In this review, the novel structures are described und the modes of interaction of the proteins
with RNA are discussed.
Additional Publications
127
Additional Publications
128
Additional Publications
129
Additional Publications
130
Additional Publications
131
Additional Publications
132
Additional Publications
133
Additional Publications
134
Additional Publications
135
Abbreviations
136
5. Abbreviations Aa Aquifex aeolicus Af Archaeoglobus fulgidus At Arabidosis thaliana ATP Adenosin triphosphate BMRB BioMagResBank bp Base pair BYV Beet Yellow virus cDNA Complementary DNA CMV Cucumber Mosaic virus CSA Chemical shift anisotropy DGCR8 DiGeorge syndrome critical region gene 8 Dm Drosophila melanogaster DNA Deoxyribonucleic acid ds Double-stranded DUF Domain of unknown function Ec Escherichia coli EDTA Ethylenediaminetetraacetic acid EMS Ethyl methanesulfonate EMSA Electro mobility shift assay FHV Flock House virus FID Free induction decay FMRP Fragile X mental retardation protein FT Fourier transformation GST Glutathione S-transferase GTP Guanosin triphosphate Hs Homo sapiens HSQC Heteronuclear single quantum correlation IPTG Isopropyl-β-D-thiogalactopyranoside kDa Kilo Dalton miRNA MicroRNA mRNA Messenger RNA NMR Nuclear magnetic resonance NOE Nuclear overhauser effect / enhancement NOESY NOE spectroscopy NoV Nodamura virus NPC Nuclear pore complex nt Nucleotide
Abbreviations
137
OB Oligosaccharide/oligonucleotide en reading frame
AGE Polyacryl-amid gel electrophoresis
R
oupling methylation
polymerase omplex transcriptional silencing
ncing
d gene silencing
ORF OpPPAZPf
Piwi, Argonaute, Zwille Pyrococcus furiosus RNA-dependent protein kinase PK
RBD RBM
RNA binding domain RNA binding motif
RDC Residual dipolar cRdDMRdRP
RNA-directed DNA RNA-dependent RNA RNA-induced silencingRISC cRNA-induced initiation oRITS fRoot mean square deviationRMSD
RNA
Ribonucleic acid RNAiRNase
RNA interference Ribonulease
RNP Rop
Ribonucleoprotein Repressor of primer
S2 Drosophila Schneider cell line 2 Structure activity relationship SAR
SDS Sodium dodecyl sulfate siRNA Small interfering RNA Sm Stephanie Smith SN Staphylococcus nuclease
Schizosaccharomyces pombe Sp ss Single-stranded TEV Tobacco Etch virus TGS Transcriptional gene silTOCSYUV
Total correlation spectroscopy Ultra-violet
VIG Vasa intronic gene VIGS Virus-induce
References
138
6. References
ist P. 2002. RNA-d s, viruses, and RNA silencing. Science 296:1270-1273.
s V. 2004. The fun re 431:350-355. s V, Bartel B, Bar gton JC, Chen X, Dreyfuss G, Eddy SR, Griffiths-Jones S, Marshall M, Matzke M, Ruvkun G, Tuschl T. 2003. A uniform system for microRNA annotation. Rna 9
kshmi R, Pruss , Smith TH, Vance VB. 1998. A viral suppressor of gen ci U S A 95:13079-13084. AA, Klenov MS , Gvozdev VA. 2004. Dissection of a natural RNA sil Mol Cell Biol 24:6742-6750.
A, Naumova N zovsky YM, Gvozdev VA. 2001. Double-stranded RNA-m genomic tandem repeats and transposable elements in the D. melanogaster :1017-1027. Johnson KL. 19 etics of nodaviruses. Adv Virus Res 53:229-244.
r DW, Kokkinidis 1987. Structure of the ColE1 rop protein at 1.7 A resolution. J Mol
DP. 2004. MicroR on. Cell 116:281-297. mbe D. 2004. RNA 63. mbe DC. 1999. Fa Curr Opin Plant Biol 2:109-113. 003. Weak alignment offers new NMR opportunities to study protein structure and dynamics.
Protein Sci 12:1-e EH, Allshire RC. 2 tional gene silencing in mammals. Trends Genet
21:370-373. nstein E, Caudy AA, annon GJ. 2001. Role for a bidentate ribonuclease in the
initiation step of re 409:363-366. zyk J, Tropea JE M, Waugh DS, Court DL, Ji X. 2001.
Crystallographic II suggest a mechanism for double-stranded A cleavage. S
F, Hansen WW, Pa . Phys Rev 69:127. n CE, Fass D, Ber , Berger JM. 1999. The structural basis for terminator
recognition by th ctor. Mol Cell 3:487-493. Bohmert K, Camus I, Bellini C, Bouchez D, Caboche M, Benning C. 1998. AGO1 defines a novel
locus of Arabidopsis controlling leaf development. EMBO J 17:170-180. Bohnsack MT, Czaplinski K, Gorlich D. 2004. Exportin 5 is a RanGTP-dependent dsRNA-binding
protein that mediates nuclear export of pre-miRNAs. Rna 10:185-191. Brennecke J, Stark A, Russell RB, Cohen SM. 2005. Principles of microRNA-target recognition. PLoS
Biol 3:e85. Brigneti G, Voinnet O, Li WX, Ji LH, Ding SW, Baulcombe DC. 1998. Viral pathogenicity
determinants are suppressors of transgene silencing in Nicotiana benthamiana. Embo J 17:6739-6746.
quAhl ependent RNA polymerase
Ambro ctions of animal microRNAs. NatuAmbro tel DP, Burge CB, Carrin
:277-279. Anandala GJ, Ge X, Marathe R, Mallory AC
e silencing in plants. Proc Natl Acad SAravin , Vagin VV, Bantignies F, Cavalli G
encing process in the Drosophila melanogaster germ line.
Aravin A M, Tulin AV, Vagin VV, Roediated silencing of germline. Curr Biol 11
Ball LA, 99. Reverse genBanne M, Tsernoglou D.
Biol 196:657-675. Bartel NAs: genomics, biogenesis, mechanism, and functi
Nature 431 56-3Baulco silencing in plants. :3Baulco st forward genetics based on virus-induced gene silencing.
Bax A. 216.
Bayn 005. RNA-directed transcrip
Ber Hammond SM, HRNA interference. Natu
Blaszc , Bubunenko M, Routzahn Kand modeling studies of RNase I
RN tructure (Camb) 9:1225-1236. Bloch ckard M. 1946Bogde gman N, Nichols MD
e Rho transcription termination fa
References
139
Brünger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Nilges M, Pannu NS, Read RJ, Rice LM, Simonson T, Warren GL. 1998. y & NMR system: A new software suite for macromolecular structure
determination. Acta Crystallogr D Biol Crystallogr 54:905-921.
diated nuclear export of
v 16:2733-
Caudy asterk RH. 2003. A micrococcal nuclease homologue in RNAi effector
Caudy :2491-2496.
y: Academic
Cerutti ins in gene silencing and cell differentiation proteins: the
iamson JR. 2005. Dual modes of
/nsmb1005.
rimer recognition and removal during DNA replication. J Mol Biol
Chapma evsky AI, Gopinath K, Dolja VV, Carrington JC. 2004. Viral RNA silencing
Cogoniing in Neurospora crassa. Proc Natl Acad Sci
Dalmay RNA-dependent RNA
Denli Acessor complex. Nature 432:231-235.
Kuszewski J, Crystallograph
Cai X, Hagedorn CH, Cullen BR. 2004. Human microRNAs are processed from capped, polyadenylated transcripts that can also function as mRNAs. Rna 10:1957-1966.
Calado A, Treichel N, Muller EC, Otto A, Kutay U. 2002. Exportin-5-meeukaryotic elongation factor 1A and tRNA. Embo J 21:6216-6224.
Caplen NJ, Parrish S, Imani F, Fire A, Morgan RA. 2001. Specific inhibition of gene expression by small double-stranded RNAs in invertebrate and vertebrate systems. Proc Natl Acad Sci U S A 98:9742-9747.
Carmell MA, Xuan Z, Zhang MQ, Hannon GJ. 2002. The Argonaute family: tentacles that reach into RNAi, developmental control, stem cell maintenance, and tumorigenesis. Genes De2742.
AA, Ketting RF, Hammond SM, Denli AM, Bathoorn AM, Tops BB, Silva JM, Myers MM, Hannon GJ, Plcomplexes. Nature 425:411-414.
AA, Myers M, Hannon GJ, Hammond SM. 2002. Fragile X-related protein and VIG associate with the RNA interference machinery. Genes Dev 16
Cavanagh J, Fairbrother WJ, Palmer III AG, Skelton NJ. 1996. Protein NMR SpectroscopPress. L, Mian N, Bateman A. 2000. Domanovel PAZ domain and redefinition of the Piwi domain. Trends Biochem Sci 25:481-482.
Chao JA, Lee JH, Chapados BR, Debler EW, Schneemann A, WillRNA-silencing suppression by Flock House virus protein B2. Nat Struct Mol Biol:doi:10.1038
Chapados BR, Chai Q, Hosfield DJ, Qiu J, Shen B, Tainer JA. 2001. Structural biochemistry of a type 2 RNase H: RNA p307:541-556. n EJ, Prokhnsuppressors inhibit the microRNA pathway at an intermediate step. Genes Dev 18:1179-1186.
Chien CY, Tejero R, Huang Y, Zimmerman DE, Rios CB, Krug RM, Montelione GT. 1997. A novel RNA-binding motif in influenza A virus non-structural protein 1. Nat Struct Biol 4:891-895.
C, Macino G. 1997. Isolation of quelling-defective (qde) mutants impaired in posttranscriptional transgene-induced gene silencU S A 94:10233-10238.
Cox DN, Chao A, Baker J, Chang L, Qiao D, Lin H. 1998. A novel class of evolutionarily conserved genes defined by piwi are essential for stem cell self-renewal. Genes Dev 12:3715-3727.
Cox DN, Chao A, Lin H. 2000. piwi encodes a nucleoplasmic factor whose activity modulates the number and division rate of germline stem cells. Development 127:503-514.
Crick FH. 1958. On protein synthesis. Symp Soc Exp Biol 12:138-163. T, Hamilton A, Rudd S, Angell S, Baulcombe DC. 2000. An polymerase gene in Arabidopsis is required for posttranscriptional gene silencing mediated by a transgene but not by a virus. Cell 101:543-553.
Denli AM, Hannon GJ. 2003. RNAi: an ever-growing puzzle. Trends Biochem Sci 28:196-201. M, Tops BB, Plasterk RH, Ketting RF, Hannon GJ. 2004. Processing of primary microRNAs by the Micropro
References
140
Ding SW, Li H, Lu R, Li F, Li WX. 2004. RNA silencing: a conserved antiviral immunity of plants and animals. Virus Res 102:109-115.
, Shi BJ, Li WX, Symons RH. 1996. An interspecies hybrid RNA virus is significantly more virulent than either parental virus. Proc Natl Acad Sci U S A 93:7
Ding SW470-7474.
Eystathi
dies. Rna 9:1171-1173.
NA
Fire A, d RNA in Caenorhabditis elegans. Nature 391:806-811.
ble-stranded RNA-binding domain protein. PLoS Biol 3:e236.
11-620. complex. Nat Struct Mol Biol
Hall TMNA in
Doench JG, Petersen CP, Sharp PA. 2003. siRNAs can function as miRNAs. Genes Dev 17:438-442. Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K, Tuschl T. 2001a. Duplexes of 21-
nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411:494-498.
Elbashir SM, Lendeckel W, Tuschl T. 2001b. RNA interference is mediated by 21- and 22-nucleotide RNAs. Genes Dev 15:188-200.
Elbashir SM, Martinez J, Patkaniowska A, Lendeckel W, Tuschl T. 2001c. Functional anatomy of siRNAs for mediating efficient RNAi in Drosophila melanogaster embryo lysate. EMBO J 20:6877-6888.
Enright AJ, John B, Gaul U, Tuschl T, Sander C, Marks DS. 2003. MicroRNA targets in Drosophila. Genome Biol 5:R1. oy T, Chan EK, Tenenbaum SA, Keene JD, Griffith K, Fritzler MJ. 2002. A phosphorylated cytoplasmic autoantigen, GW182, associates with a unique population of human mRNAs within novel cytoplasmic speckles. Mol Biol Cell 13:1338-1351.
Eystathioy T, Jakymiw A, Chan EK, Seraphin B, Cougot N, Fritzler MJ. 2003. The GW182 protein colocalizes with mRNA degradation associated proteins hDcp1 and hLSm4 in cytoplasmic GW bo
Fagard M, Boutet S, Morel JB, Bellini C, Vaucheret H. 2000. AGO1, QDE-2, and RDE-1 are related proteins required for post-transcriptional gene silencing in plants, quelling in fungi, and Rinterference in animals. Proc Natl Acad Sci U S A 97:11650-11654. Xu S, Montgomery MK, Kostas SA, Driver SE, Mello CC. 1998. Potent and specific genetic interference by double-strande
Forstemann K, Tomari Y, Du T, Vagin VV, Denli AM, Bratu DP, Klattenhoff C, Theurkauf WE, Zamore PD. 2005. Normal microRNA maturation and germ-line stem cell maintenance requires Loquacious, a dou
Friedman AM, Fischmann TO, Steitz TA. 1995. Crystal structure of lac repressor core tetramer and its implications for DNA looping. Science 268:1721-1727.
Gregory RI, Yan KP, Amuthan G, Chendrimada T, Doratotaj B, Cooch N, Shiekhattar R. 2004. The Microprocessor complex mediates the genesis of microRNAs. Nature 432:235-240.
Grewal SI, Rice JC. 2004. Regulation of heterochromatin by histone methylation and small RNAs. Curr Opin Cell Biol 16:230-238.
Grishok A, Pasquinelli AE, Conte D, Li N, Parrish S, Ha I, Baillie DL, Fire A, Ruvkun G, Mello CC. 2001. Genes and mechanisms related to RNA interference regulate expression of the small temporal RNAs that control C. elegans developmental timing. Cell 106:23-34.
Guo S, Kemphues KJ. 1995. par-1, a gene required for establishing polarity in C. elegans embryos, encodes a putative Ser/Thr kinase that is asymmetrically distributed. Cell 81:6
Haley B, Zamore PD. 2004. Kinetic analysis of the RNAi enzyme 11:599-606. . 2005. Structure and function of argonaute proteins. Structure (Camb) 13:1403-1408.
Hamilton A, Voinnet O, Chappell L, Baulcombe D. 2002. Two classes of short interfering RRNA silencing. Embo J 21:4671-4679.
Hamilton AJ, Baulcombe DC. 1999. A species of small antisense RNA in posttranscriptional gene silencing in plants. Science 286:950-952.
References
141
Hammond SM, Bernstein E, Beach D, Hannon GJ. 2000. An RNA-directed nuclease mediates post-transcriptional gene silencing in Drosophila cells. Nature 404:293-296.
HannonHe L, Hannon GJ. 2004. MicroRNAs: small RNAs with a big role in gene regulation. Nat Rev Genet
Hull R. San Diego, CA: Academic Press.
-838.
John B, MicroRNA targets. PLoS
Johnson larger
Jones L can
96:375-387.
Kasscha sion of
Kataoka Y, Takeichi M, Uemura T. 2001. Developmental roles and molecular characterization of a .
NA involved in developmental timing in C. elegans.
Hammond SM, Boettcher S, Caudy AA, Kobayashi R, Hannon GJ. 2001. Argonaute2, a link between genetic and biochemical analyses of RNAi. Science 293:1146-1150.
Han J, Lee Y, Yeom KH, Kim YK, Jin H, Kim VN. 2004. The Drosha-DGCR8 complex in primarymicroRNA processing. Genes Dev 18:3016-3027. GJ. 2002. RNA interference. Nature 418:244-251.
5:522-531. 2002. Matthews' Plant Virology.
Hutvagner G, McLachlan J, Pasquinelli AE, Balint E, Tuschl T, Zamore PD. 2001. A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7 small temporal RNA. Science 293:834
Hutvagner G, Zamore PD. 2002. A microRNA in a multiple-turnover RNAi enzyme complex. Science 297:2056-2060.
Ingelfinger D, Arndt-Jovin DJ, Luhrmann R, Achsel T. 2002. The human LSm1-7 proteins colocalize with the mRNA-degrading enzymes Dcp1/2 and Xrnl in distinct cytoplasmic foci. Rna 8:1489-1501.
Ishizuka A, Siomi MC, Siomi H. 2002. A Drosophila fragile X protein interacts with components of RNAi and ribosomal proteins. Genes Dev 16:2497-2508.
Jaenisch R, Bird A. 2003. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat Genet 33 Suppl:245-254. Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS. 2004. HumanBiol 2:e363. KN, Johnson KL, Dasgupta R, Gratsch T, Ball LA. 2001. Comparisons among thegenome segments of six nodaviruses and their encoded RNA replicases. J Gen Virol 82:1855-1866. , Ratcliff F, Baulcombe DC. 2001. RNA-directed transcriptional gene silencing in plantsbe inherited independently of the RNA trigger and requires Met1 for maintenance. Curr Biol 11:747-757.
Jorgensen R. 1990. Altered gene expression in plants due to trans interactions between homologous genes. Trends Biotechnol 8:340-344.
Kambach C, Walke S, Young R, Avis JM, de la Fortelle E, Raker VA, Luhrmann R, Li J, Nagai K. 1999. Crystal structures of two Sm protein complexes and their implications for the assembly of the spliceosomal snRNPs. Cell
Kanaya S, Ikehara M. 1995. Functions and structures of ribonuclease H enzymes. Subcell Biochem 24:377-422.
Karplus M. 1959. J Phys Chem 30:11-15. u KD, Carrington JC. 1998. A counterdefensive strategy of plant viruses: suppresposttranscriptional gene silencing. Cell 95:461-470.
Drosophila homologue of Arabidopsis Argonaute1, the founder of a novel gene superfamilyGenes Cells 6:313-325.
Ketting RF, Fischer SE, Bernstein E, Sijen T, Hannon GJ, Plasterk RH. 2001. Dicer functions in RNA interference and in synthesis of small RGenes Dev 15:2654-2659.
References
142
Khusial P, Plaag R, Zieve GW. 2005. LSm proteins form heptameric rings that bind to RNA via repeating motifs. Trends Biochem Sci 30:522-528.
Kim VN. 2005. MicroRNA biogenesis: coordinated cropping and dicing. Nat Rev Mol Cell Biol 6:376-385.
Knight SW, Bass BL. 2001. A role for the RNase III enzyme DCR-1 in RNA interference and germ line development in Caenorhabditis elegans. Science 293:2269-2271.
a program for display and analysis of
Krishna NR, Berliner LJ. 2003. Protein NMR for the Millenium. New York: Kluwer
Lai L, Kim R, Kim SH. 2000. Crystal structure of archaeal RNase HII: a
Landthaler M, Yalcin A, Tuschl T. 2004. The human DiGeorge syndrome critical region gene 8 and
Lee RCisense complementarity to lin-14. Cell 75:843-854.
ing. Nature 425:415-419.
icer-1 and Dicer-2 in the siRNA/miRNA silencing pathways. Cell 117:69-81.
Li H, L ession of RNA silencing by an animal virus.
Li HW,l suppressor of the plant gene silencing defence mechanism. Embo J 18:2683-
Li WX, NA silencing. Curr Opin Biotechnol 12:150-154.
s. Nature 433:769-773. hakim A, Yekta S, Rhoades MW, Burge CB, Bartel DP.
Linge J nment of ambiguous nuclear overhauser
Linge J 50:496-506.
Koradi R, Billeter M, Wüthrich K. 1996. MOLMOL: macromolecular structures. J Mol Graph 14:51-55.
Academic/Plenum Publishers. Lagos-Quintana M, Rauhut R, Lendeckel W, Tuschl T. 2001. Identification of novel genes coding for
small expressed RNAs. Science 294:853-858. Yokota H, Hung LW, homologue of human major RNase H. Structure Fold Des 8:897-904.
Its D. melanogaster homolog are required for miRNA biogenesis. Curr Biol 14:2162-2167. , Feinbaum RL, Ambros V. 1993. The C. elegans heterochronic gene lin-4 encodes small RNAs with ant
Lee Y, Ahn C, Han J, Choi H, Kim J, Yim J, Lee J, Provost P, Radmark O, Kim S, Kim VN. 2003. The nuclear RNase III Drosha initiates microRNA process
Lee Y, Kim M, Han J, Yeom KH, Lee S, Baek SH, Kim VN. 2004a. MicroRNA genes are transcribed by RNA polymerase II. Embo J 23:4051-4060.
Lee YS, Nakahara K, Pham JW, Kim K, He Z, Sontheimer EJ, Carthew RW. 2004b. Distinct roles for Drosophila D
Lei M, Podell ER, Baumann P, Cech TR. 2003. DNA self-recognition in the structure of Pot1 bound to telomeric single-stranded DNA. Nature 426:198-203. i WX, Ding SW. 2002. Induction and supprScience 296:1319-1321. Lucy AP, Guo HS, Li WX, Ji LH, Wong SM, Ding SW. 1999. Strong host resistance targeted against a vira2691. Ding SW. 2001. Viral suppressors of R
Lim LP, Glasner ME, Yekta S, Burge CB, Bartel DP. 2003a. Vertebrate microRNA genes. Science 299:1540.
Lim LP, Lau NC, Garrett-Engele P, Grimson A, Schelter JM, Castle J, Bartel DP, Linsley PS, Johnson JM. 2005. Microarray analysis shows that some microRNAs downregulate large numbers of target mRNA
Lim LP, Lau NC, Weinstein EG, Abdel2003b. The microRNAs of Caenorhabditis elegans. Genes Dev 17:991-1008. P, O'Donoghue SI, Nilges M. 2001. Automated assigeffects with ARIA. Methods Enzymol 339:71-90.
P, Williams MA, Spronk CA, Bonvin AM, Nilges M. 2003. Refinement of protein structures in explicit solvent. Proteins
Lingel A, Sattler M. 2005. Novel modes of protein-RNA recognition in the RNAi pathway. Curr Opin Struct Biol 15:107-115.
References
143
Lingel A, Simon B, Izaurralde E, Sattler M. 2003. Structure and nucleic-acid binding of the Drosophila Argonaute 2 PAZ domain. Nature 426:465-469.
.
NAi. Science 305:1437-
Liu J, L erman HM. 1997. Crystal structure of the
Liu J, Rivas FV, Wohlschlegel J, Yates JR, Parker R, Hannon GJ. 2005a. A role for the P-body
Liu J, V . 2005b. MicroRNA-dependent localization of
Liu Q, 2, a bridge between the
Llave C ne
by the A. fulgidus Piwi protein. Nature 434:666-670.
Dev
Meister orsett Y, Teng G, Tuschl T. 2004. Human Argonaute2
Meister Nature 431:343-
Messias -stranded RNA recognition. Acc Chem Res
Mette M tzke MA, Matzke AJ. 2000. Transcriptional silencing
Mourelatos Z, Dostie J, Paushkin S, Sharma A, Charroux B, Abel L, Rappsilber J, Mann M, Dreyfuss G. 2002. miRNPs: a novel class of ribonucleoproteins containing numerous microRNAs. Genes Dev 16:720-728.
Lippman Z, Martienssen R. 2004. The role of RNA interference in heterochromatic silencing. Nature 431:364-370
Liu J, Carmell MA, Rivas FV, Marsden CG, Thomson JM, Song JJ, Hammond SM, Joshua-Tor L, Hannon GJ. 2004. Argonaute2 is the catalytic engine of mammalian R1441. ynch PA, Chien CY, Montelione GT, Krug RM, Bunique RNA-binding domain of the influenza virus NS1 protein. Nat Struct Biol 4:896-899.
component GW182 in microRNA function. Nat Cell Biol. alencia-Sanchez MA, Hannon GJ, Parker R
targeted mRNAs to mammalian P-bodies. Nat Cell Biol 7:719-723. Rand TA, Kalidas S, Du F, Kim HE, Smith DP, Wang X. 2003. R2Dinitiation and effector steps of the Drosophila RNAi pathway. Science 301:1921-1925. , Kasschau KD, Carrington JC. 2000. Virus-encoded suppressor of posttranscriptional gesilencing targets a maintenance step in the silencing pathway. Proc Natl Acad Sci U S A 97:13401-13406.
Llave C, Xie Z, Kasschau KD, Carrington JC. 2002. Cleavage of Scarecrow-like mRNA targets directed by a class of Arabidopsis miRNA. Science 297:2053-2056.
Lu R, Maduro M, Li F, Li HW, Broitman-Maduro G, Li WX, Ding SW. 2005. Animal virus replication and RNAi-mediated antiviral silencing in Caenorhabditis elegans. Nature 436:1040-1043.
Lund E, Guttinger S, Calado A, Dahlberg JE, Kutay U. 2004. Nuclear export of microRNA precursors. Science 303:95-98.
Ma JB, Ye K, Patel DJ. 2004. Structural basis for overhang-specific small interfering RNA recognition by the PAZ domain. Nature 429:318-322.
Ma JB, Yuan YR, Meister G, Pei Y, Tuschl T, Patel DJ. 2005. Structural basis for 5'-end-specific recognition of guide RNA
Martinez J, Patkaniowska A, Urlaub H, Luhrmann R, Tuschl T. 2002. Single-stranded antisense siRNAs guide target RNA cleavage in RNAi. Cell 110:563-574.
Martinez J, Tuschl T. 2004. RISC is a 5' phosphomonoester-producing RNA endonuclease. Genes18:975-980.
Matranga C, Tomari Y, Shin C, Bartel DP, Zamore PD. 2005. Passenger-Strand Cleavage Facilitates Assembly of siRNA into Ago2-Containing RNAi Enzyme Complexes. Cell 123:607-620.
Matzke MA, Birchler JA. 2005. RNAi-mediated pathways in the nucleus. Nat Rev Genet 6:24-35. G, Landthaler M, Patkaniowska A, DMediates RNA Cleavage Targeted by miRNAs and siRNAs. Mol Cell 15:185-197. G, Tuschl T. 2004. Mechanisms of gene silencing by double-stranded RNA. 349. AC, Sattler M. 2004. Structural basis of single37:279-287. F, Aufsatz W, van der Winden J, Ma
and promoter methylation triggered by double-stranded RNA. Embo J 19:5194-5201.
References
144
Neuhaus D, Williamson MP. 2000. The Nuclear Overhauser Effect in Structural and Conformational Analysis: Wiley-VCH.
Nicholson AW. 2003. The ribonuclease III family: forms and functions in RNA maturation, decay, and gene silencing. Cold Spring Harbor, New York: Cold Spring Harbor Press.
rance in the ribosome. Nat Struct Biol 10:104-108.
Nowotn n
Nykane all interfering RNA structure in the
Okamu gonaute proteins in small
Orban T mplex,
Pal-Bhailencing and HP1 localization in Drosophila are dependent on the RNAi
Parker suggests mechanisms for
Parker Jplex. Nature 434:663-666.
NAs during RNAi in Drosophila. Cell 117:83-94.
ene
ated by the coat protein and maintained by p19 through suppression of gene silencing.
Rand TA, Ginalski K, Grishin NV, Wang X. 2004. Biochemical identification of Argonaute 2 as the
Nikulin A, Eliseikina I, Tishchenko S, Nevskaya N, Davydova N, Platonova O, Piendl W, Selmer M, Liljas A, Drygin D, Zimmermann R, Garber M, Nikonov S. 2003. Structure of the L1 protube
Novina CD, Sharp PA. 2004. The RNAi revolution. Nature 430:161-164. y M, Gaidamakov SA, Crouch RJ, Yang W. 2005. Crystal structures of RNase H bound to aRNA/DNA hybrid: substrate specificity and metal-dependent catalysis. Cell 121:1005-1016. n A, Haley B, Zamore PD. 2001. ATP requirements and smRNA interference pathway. Cell 107:309-321.
ra K, Ishizuka A, Siomi H, Siomi MC. 2004. Distinct roles for ArRNA-directed RNA cleavage pathways. Genes Dev 18:1655-1666. I, Izaurralde E. 2005. Decay of mRNAs targeted by RISC requires XRN1, the Ski coand the exosome. Rna 11:459-469. dra M, Leibovitch BA, Gandhi SG, Rao M, Bhadra U, Birchler JA, Elgin SC. 2004. Heterochromatic smachinery. Science 303:669-672.
JS, Roe SM, Barford D. 2004. Crystal structure of a PIWI protein siRNA recognition and slicer activity. Embo J 23:4727-4737. S, Roe SM, Barford D. 2005. Structural insights into mRNA recognition from a PIWI domain-siRNA guide com
Pham JW, Pellino JL, Lee YS, Carthew RW, Sontheimer EJ. 2004. A Dicer-2-dependent 80s complex cleaves targeted mR
Plasterk RH. 2002. RNA silencing: the genome's immune system. Science 296:1263-1265. Predki PF, Nayak LM, Gottlieb MB, Regan L. 1995. Dissecting RNA-protein interactions: RNA-RNA
recognition by Rop. Cell 80:41-50. Pruss G, Ge X, Shi XM, Carrington JC, Bowman Vance V. 1997. Plant viral synergism: the potyviral
genome encodes a broad-range pathogenicity enhancer that transactivates replication of heterologous viruses. Plant Cell 9:859-868.
Purcell EM, Torrey HC, Pound RV. 1946. Phys Rev 69:37-38. Qian XY, Chien CY, Lu Y, Montelione GT, Krug RM. 1995. An amino-terminal polypeptide
fragment of the influenza virus NS1 protein possesses specific RNA-binding activity and largely helical backbone structure. Rna 1:948-956.
Qiu W, Park JW, Scholthof HB. 2002. Tombusvirus P19-mediated suppression of virus-induced gsilencing is controlled by genetic and dosage features that influence pathogenicity. Mol Plant Microbe Interact 15:269-280.
Qu F, Morris TJ. 2002. Efficient infection of Nicotiana benthamiana by Tomato bushy stunt virus is facilitMol Plant Microbe Interact 15:193-202.
sole protein required for RNA-induced silencing complex activity. Proc Natl Acad Sci U S A 101:14385-14389.
Rand TA, Petersen S, Du F, Wang X. 2005. Argonaute2 Cleaves the Anti-Guide Strand of siRNA during RISC Activation. Cell 123:621-629.
References
145
Reed JC, Kasschau KD, Prokhnevsky AI, Gopinath K, Pogue GP, Carrington JC, Dolja VV. 2003. Suppressor of RNA silencing encoded by Beet yellows virus. Virology 306:203-209.
1:1640-1647.
Rivas F -Tor L. 2005. Purified Argonaute2 l 12:340-349.
05-7513.
ells. PLoS Biol 3:e235.
and meiotic drive in the male
Schwar 2. Evidence that siRNAs function as guides, not
Schwar complex is a Mg2+-
Shuker 6. Discovering high-affinity ligands for proteins:
Silhavy vazza M, Burgyan J. 2002. A viral protein
Song JJ an J, Smith SK, Martienssen RA, Hannon GJ, Joshua-Tor L.
Song J L. 2004. Crystal Structure of Argonaute and Its
Stark A
6:33-38.
Rehwinkel J, Behm-Ansmant I, Gatfield D, Izaurralde E. 2005. A crucial role for GW182 and the DCP1:DCP2 decapping complex in miRNA-mediated gene silencing. Rna 1
Rhoades MW, Reinhart BJ, Lim LP, Burge CB, Bartel B, Bartel DP. 2002. Prediction of plant microRNA targets. Cell 110:513-520. V, Tolia NH, Song JJ, Aragon JP, Liu J, Hannon GJ, Joshuaand an siRNA form recombinant human RISC. Nat Struct Mol Bio
Roth BM, Pruss GJ, Vance VB. 2004. Plant viral suppressors of RNA silencing. Virus Res 102:97-108.
Ruiz MT, Voinnet O, Baulcombe DC. 1998. Initiation and maintenance of virus-induced gene silencing. Plant Cell 10:937-946.
Ryter JM, Schultz SC. 1998. Molecular basis of double-stranded RNA-protein interactions: structure of a dsRNA-binding domain complexed with dsRNA. Embo J 17:75
Saito K, Ishizuka A, Siomi H, Siomi MC. 2005. Processing of pre-microRNAs by the Dicer-1-Loquacious complex in Drosophila c
Sattler M, Schleucher J, Griesinger C. 1999. Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Prog NMR Spectrosc 34:93-158.
Schmidt A, Palumbo G, Bozzetti MP, Tritto P, Pimpinelli S, Schafer U. 1999. Genetic and molecular characterization of sting, a gene involved in crystal formationgerm line of Drosophila melanogaster. Genetics 151:749-760.
z DS, Hutvagner G, Haley B, Zamore PD. 200primers, in the Drosophila and human RNAi pathways. Mol Cell 10:537-548.
z DS, Tomari Y, Zamore PD. 2004. The RNA-induced silencing dependent endonuclease. Curr Biol 14:787-791.
Sen GL, Blau HM. 2005. Argonaute 2/RISC resides in sites of mammalian mRNA decay known as cytoplasmic bodies. Nat Cell Biol 7:633-636.
Sheth U, Parker R. 2003. Decapping and decay of messenger RNA occur in cytoplasmic processing bodies. Science 300:805-808. SB, Hajduk PJ, Meadows RP, Fesik SW. 199SAR by NMR. Science 274:1531-1534.
Sijen T, Fleenor J, Simmer F, Thijssen KL, Parrish S, Timmons L, Plasterk RH, Fire A. 2001. On the role of RNA amplification in dsRNA-triggered gene silencing. Cell 107:465-476. D, Molnar A, Lucioli A, Szittya G, Hornyik C, Tasuppresses RNA silencing and binds silencing-generated, 21- to 25-nucleotide double-stranded RNAs. Embo J 21:3070-3080. , Liu J, Tolia NH, Schneiderm2003. The crystal structure of the Argonaute2 PAZ domain reveals an RNA binding motif in RNAi effector complexes. Nat Struct Biol 10:1026-1032.
J, Smith SK, Hannon GJ, Joshua-Tor Implications for RISC Slicer Activity. Science. , Brennecke J, Russell RB, Cohen SM. 2003. Identification of Drosophila MicroRNA targets. PLoS Biol 1:E60.
Stefl R, Skrisovska L, Allain FH. 2005. RNA sequence- and shape-dependent recognition by proteins in the ribonucleoprotein particle. EMBO Rep
References
146
Steitz TA, Steitz JA. 1993. A general two-metal-ion mechanism for catalytic RNA. Proc Natl Acad Sci U S A 90:6498-6502.
Sullivan CS, Ganem D. 2005. A virus-encoded inhibitor that blocks RNA interference in mammalian cells. J Virol 79:7371-7379.
Tabara H, Sarkissian M, Kelly WG, Fleenor J, Grishok A, Timmons L, Fire A, Mello CC. 1999. The rde-1 gene, RNA interference, and transposon silencing in C. elegans. Cell 99:123-132.
Rep 5:189-
Theobald DL, Mitton-Fry RM, Wuttke DS. 2003. Nucleic Acid Recognition by OB-Fold Proteins.
Thore S ck D. 2003. Crystal structures of the Pyrococcus abyssi Sm
Tjandra es in biomolecules by NMR in a
Tjandra the dependence of heteronuclear relaxation times on rotational
Tolman in solution. Proc Natl Acad Sci
Tomari uf WE, Zamore
Voinnet O. 2001. RNA silencing as a plant immune system against viruses. Trends Genet 17:449-459.
Voinne protein prevents spread of the gene
Voinne ion of gene silencing: a general strategy used
Volpe all IM, Teng G, Grewal SI, Martienssen RA. 2002. Regulation of
Tahbaz N, Kolb FA, Zhang H, Jaronczyk K, Filipowicz W, Hobman TC. 2004. Characterization of the interactions between mammalian PAZ PIWI domain proteins and Dicer. EMBO194.
Annu Rev Biophys Biomol Struct 32:115-133. , Mayer C, Sauter C, Weeks S, Sucore and its complex with RNA. Common features of RNA binding in archaea and eukarya. J Biol Chem 278:1239-1247.
Timmons L, Fire A. 1998. Specific interference by ingested dsRNA. Nature 395:854. N, Bax A. 1997. Direct measurement of distances and angldilute liquid crystalline medium. Science 278:1111-1114. N, Garrett DS, Gronenborn AM, Bax A, Clore GM. 1997. Defining long range order in NMR structure determination fromdiffusion anisotropy. Nature Struct Biol 4:443-449. JR, Flanagan JM, Kennedy MA, Prestegard JH. 1995. Nuclear magnetic dipole interactions in field-oriented proteins: information for structure determinationU S A 92:9279-9283.
Y, Du T, Haley B, Schwarz DS, Bennett R, Cook HA, Koppetsch BS, TheurkaPD. 2004a. RISC assembly defects in the Drosophila RNAi mutant armitage. Cell 116:831-841.
Tomari Y, Matranga C, Haley B, Martinez N, Zamore PD. 2004b. A protein sensor for siRNA asymmetry. Science 306:1377-1380.
Tuschl T, Zamore PD, Lehmann R, Bartel DP, Sharp PA. 1999. Targeted mRNA degradation by double-stranded RNA in vitro. Genes Dev 13:3191-3197.
Vargason JM, Szittya G, Burgyan J, Tanaka Hall TM. 2003. Size selective recognition of siRNA by an RNA silencing suppressor. Cell 115:799-811.
Verdel A, Jia S, Gerber S, Sugiyama T, Gygi S, Grewal SI, Moazed D. 2004. RNAi-mediated targeting of heterochromatin by the RITS complex. Science 303:672-676.
Vermeulen A, Behlen L, Reynolds A, Wolfson A, Marshall WS, Karpilow J, Khvorova A. 2005. The contributions of dsRNA structure to Dicer specificity and efficiency. Rna 11:674-682.
Voinnet O. 2005. Induction and suppression of RNA silencing: insights from viral infections. Nat Rev Genet 6:206-220.
t O, Lederer C, Baulcombe DC. 2000. A viral movement silencing signal in Nicotiana benthamiana. Cell 103:157-167.
t O, Pinto YM, Baulcombe DC. 1999. Suppressby diverse DNA and RNA viruses of plants. Proc Natl Acad Sci U S A 96:14147-14152. TA, Kidner C, Hheterochromatic silencing and histone H3 lysine-9 methylation by RNAi. Science 297:1833-1837.
References
147
Wassenegger M. 2005. The role of the RNAi machinery in heterochromatin formation. Cell 122:13-16.
Wassenegger M, Heimes S, Riedel L, Sanger HL. 1994. RNA-directed de novo methylation of genomic sequences in plants. Cell 76:567-576.
William , Rubin GM. 2002. ARGONAUTE1 is required for efficient RNA interference in
Wilson enhances oskar translation in the Drosophila
Winters
Ye K, Patel DJ. 2005. RNA Silenc
ort hairpin RNAs. Genes Dev 17:3011-3016.
d bacterial RNase III. Cell 118:57-68.
16-719.
Waterhouse PM, Smith NA, Wang MB. 1999. Virus resistance and gene silencing: killing the messenger. Trends Plant Sci 4:452-457.
Wightman B, Ha I, Ruvkun G. 1993. Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75:855-862. s RWDrosophila embryos. Proc Natl Acad Sci U S A 99:6889-6894. JE, Connell JE, Macdonald PM. 1996. aubergine ovary. Development 122:1631-1639. berger U. 1990. Ribonucleases H of retroviral and cellular origin. Pharmacol Ther 48:259-280.
Wüthrich K. 1986. NMR of proteins and nucleic acids: Wiley-Interscience. Yan KS, Yan S, Farooq A, Han A, Zeng L, Zhou MM. 2003. Structure and conserved RNA binding of
the PAZ domain. Nature 426:468-474. Yang W, Steitz TA. 1995. Recombining the structures of HIV integrase, RuvC and RNase H.
Structure 3:131-134. Ye K, Malinina L, Patel DJ. 2003. Recognition of small interfering RNA by a viral suppressor of RNA
silencing. Nature 426:874-878. ing Suppressor p21 of Beet Yellows Virus Forms an RNA Binding
Octameric Ring Structure. Structure (Camb) 13:1375-1384. Yi R, Qin Y, Macara IG, Cullen BR. 2003. Exportin-5 mediates the nuclear export of pre-microRNAs
and shYuan YR, Pei Y, Ma JB, Kuryavyi V, Zhadina M, Meister G, Chen HY, Dauter Z, Tuschl T, Patel DJ.
2005. Crystal structure of A. aeolicus argonaute, a site-specific DNA-guided endoribonuclease, provides insights into RISC-mediated mRNA cleavage. Mol Cell 19:405-419.
Zamore PD, Tuschl T, Sharp PA, Bartel DP. 2000. RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101:25-33.
Zeng Y, Yi R, Cullen BR. 2003. MicroRNAs and small interfering RNAs can inhibit mRNA expression by similar mechanisms. Proc Natl Acad Sci U S A 100:9779-9784.
Zeng Y, Yi R, Cullen BR. 2005. Recognition and cleavage of primary microRNA precursors by the nuclear processing enzyme Drosha. Embo J 24:138-148.
Zhang H, Kolb FA, Brondani V, Billy E, Filipowicz W. 2002. Human Dicer preferentially cleavesdsRNAs at their termini without a requirement for ATP. EMBO J 21:5875-5885.
Zhang H, Kolb FA, Jaskiewicz L, Westhof E, Filipowicz W. 2004. Single processing center models for human Dicer an
Zilberman D, Cao X, Jacobsen SE. 2003. ARGONAUTE4 control of locus-specific siRNA accumulation and DNA and histone methylation. Science 299:7
Zuiderweg ER. 2002. Mapping protein-protein interactions in solution by NMR spectroscopy. Biochemistry 41:1-7.
Curriculum vitae
148
7. Curriculum vitae
Name
Place of
Educa
Since 0 European Molecular Biology Laboratory (EMBL) Heidelberg, Germany and , Switzerland
r at ETH Zurich: Prof. Dr. Ulrike Kutay
10/1999
10/2000
rgency service
Personal details
Andreas Lingel Date of birth February 8th, 1976
birth Marbach am Neckar, Germany Marital status Single Nationality German
tion
5/2002 Swiss Federal Institute of Technology (ETH) Zurich International Ph.D. Programme at EMBL Heidelberg Ph.D. research project under the supervision of Dr. Elisa Izaurralde: “Structural
and functional characterization of proteins involved in RNA-mediated gene silencing” Superviso
11/2001-02/2002 Rockefeller University, New York City, New York, USA Laboratory of Cell Biology, Prof. Dr. Günter Blobel
Research internship, topic: “The Role of POM121 and gp210 on Nuclear Pore Complex (NPC) Structure and Assembly”
-11/2001 ETH Zurich, Switzerland Diploma in biology (biochemistry), passed with distinction, grade: 5.93
-04/2001 University of Zurich, Switzerland Department of Biochemistry, laboratory of Prof. Dr. Andreas Plückthun Diploma thesis, entitled “Stability Enhancement of Proteins by Intein mediated Cyclization”, grade: 6, with honors
04/2000-07/2000 ETH Zurich, Switzerland Institute of Biochemistry, laboratory of Prof. Dr. Ari Helenius Semester work, topic: “Quality Control of Newly Synthesized Proteins in the Endoplasmatic Reticulum”
04/1997-09/1999 Eberhard-Karls-University Tübingen, Germany Vordiplom in biochemistry, passed with distinction, grade: “sehr gut”
01/1996-01/1997 German Red Cross, District Ludwigsburg, Germany Civil service as an ambulance man in the eme
06/1995 Friedrich-Schiller-Gymnasium in Marbach am Neckar, Germany Abitur, grade: 1.4, passed with distinction
Curriculum vitae
149
ublications P
1. Lingel, A., Simon, B., Izaurralde, E., and Sattler, M. (2005) The structure of the FHV B2 protein, a viral suppressor of RNAi, reveals a novel mode of dsRNA recognition.
Rep
l, A
Embo , 6, 1149-1155.
2. Ling .e , and Sattler, M. (2005) Novel modes of protein-RNA recognition in the urr Opin Struct Biol, 15, 107-115.
3. Lingel, A.
RNAi pathway. C
, and Izaurralde, E. (2004) RNAi: Finding the elusive endonuclease. RNA, 675-1679.
l, A.
10, 1
4. Linge , Simon, B., Izaurralde, E., and Sattler, M. (2004) NMR assignment of the Drosophil
. Lingel, A.
a Argonaute2 PAZ domain. J Biomol NMR, 29, 421-432.
5 , Simon, B., Izaurralde, E., and Sattler, M. (2004) Nucleic acid 3’-end n by the Argonaute2 PAZ domain. Nat Struct Mol Biol, 11, 576-577. recognitio
. Lingel, A.6 , Simon, B., Izaurralde, E., and Sattler, M. (2Drosophila Argonaute2 PAZ domain.
003) Structure and nucleic acid binding of the Nature, 426, 465-469.
. Iwai, H., L7 ingel, A., Plückthun, A. (2001) Cyclic green fluorescent protein produced in vivo usChem, 276
Conferences
He EMBO Co
Poster pr recognition by the Argonaute2 PAZ
. Keystone Keystone
Poster presentation: “Structural Basis for 3’-end recognition of nucleic acids by the
3. Max Planctein complexes”, 17.-20.
SeptembPoster pre nucleic acid binding fo
4. Cold Spring Harbor Laboratory, New York, USA Cold Spring Harbor Conference on “Eukaryotic mRNA Processing”, 20.-24. August 2003 Oral presentation: “Structure and nucleic acid binding of the Drosophila Argonaute 2 PAZ domain”
ing an artificially split PI-PfuI intein from Pyrococcus furiosus. J Biol , 16548-16554.
1 idelberg, Germany . EMBL nference “Structures in Biology”, 10.-13. November 2004
esentation: “Nucleic acid 3’-enddomain”
2 Resort, Colorado, USA Symposium on “siRNAs and miRNAs”, 14.-19. April 2004
Drosophila Argonaute2 PAZ domain”
k Institut for Biophysical Chemistry, Göttingen, Germany Symposium “Structure, function and dynamics of RNA-pro
er 2003 sentation: “The Argonaute 2 PAZ domain adopts a novel ld”