The regulation of microRNA activity

Abstract

MicroRNAs are short, single-stranded, non-protein coding RNAs that negatively regulate gene expression. MicroRNAs have been shown to play a critical role in many biological processes and diseases. Therefore, it is critical to gain an understanding of the factors that can regulate microRNA activity. It has been established that, microRNA activity is exerted by a class of RNA-binding proteins called Argonautes. MicroRNAs arise from a transient duplex (the microRNA duplex) from which only one strand (the microRNA strand) is specifically sorted into an Argonaute protein, whereas the other strand (the microRNA* strand) is not. This observation provokes the question; how is the microRNA specifically sorted into an Argonaute protein? To this effect we have found that the differing 5' terminal nucleotides of the microRNA and microRNA* strand are a critical determinant for strand specific sorting of the microRNA into Argonaute. Knowing that microRNAs proceed through a duplex intermediate we set out to identify a microRNA duplex unwindase. Biochemical purification of a microRNA duplex specific unwinding activity from cells yielded P68 RNA Helicase. We showed that P68 RNA Helicase is sufficient to unwind a microRNA duplex, and also that P68 is necessary for microRNA function in cells. Next we asked if silent mutations could create novel, non- canonical microRNA target sites. We identified a silent variant (G4304A) in the BRCA1 mRNA, creates a novel microRNA target site for microRNA-501.

Full text

The Regulation of MicroRNA Activity
David W. Salzman, Ph.D.
University of Connecticut, 2009
MicroRNAs are short, single-stranded, non-protein coding RNAs that negatively
regulate gene expression. MicroRNAs have been shown to play a critical role in many
biological processes and diseases. Therefore, it is critical to gain an understanding of the
factors that can regulate microRNA activity. It has been established that, microRNA
activity is exerted by a class of RNA-binding proteins called Argonautes. MicroRNAs
arise from a transient duplex (the microRNA duplex) from which only one strand (the
microRNA strand) is specifically sorted into an Argonaute protein, whereas the other
strand (the microRNA* strand) is not. This observation provokes the question; how is
the microRNA specifically sorted into an Argonaute protein? To this effect we have
found that the differing 5' terminal nucleotides of the microRNA and microRNA* strand
are a critical determinant for strand specific sorting of the microRNA into Argonaute.
Knowing that microRNAs proceed through a duplex intermediate we set out to identify a
microRNA duplex unwindase. Biochemical purification of
a
microRNA duplex specific
unwinding activity from cells yielded P68 RNA Helicase. We showed that P68 RNA
Helicase is sufficient to unwind a microRNA duplex, and also that P68 is necessary for
microRNA function in cells. Next we asked if silent mutations could create novel, non-
canonical microRNA target sites. We identified a silent variant (G4304A) in the BRCAl
mRNA, creates a novel microRNA target site for microRNA-501.

The Regulation of MicroRNA Activity
David W. Salzman
B.S.,
University of Hartford, 2004
A Dissertation
Submitted in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy
at the
University of Connecticut
2009

UMI Number: 3383933
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted.
Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3383933
Copyright 2009 by ProQuest LLC
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346

APPROVAL PAGE
Doctor of Philosophy Dissertation
The Regulation of MicroRNA Activity
Presented by
David W. Salzman, B.S.
Major Advisor X
lenry M. Furneaux
Associate Advisor
Associate Advisor
^(^(TpO
Asis Das
Chris Heinen
University of Connecticut
2009
11

Table of Contents
1.
Synopsis (1)
2.
Introduction (2)
3.
Premise (22)
4.
The Differing Termini of Let-7 and Let-7* Play a Critical Role in Strand Specific
Sorting into Argonaute2 (28)
4.1 Introduction (28)
4.2 Results (30)
4.2.1 The let-7* Strand Does Not Silence Gene Expression In Vivo or In
Vitro (30)
4.2.2 The let-7* Strand Does Not Readily Complex With Argonaute2
(38)
4.2.3 The let-7* Strand Does Not Recruit mRNA (41)
4.2.4 Analysis of the let-7 and let-7* Strand Binary Complexes Reveals
That the let-7* Strand Does Not Adopt an Appropriate Structure
for mRNA Recruitment (45)
4.3 Discussion (51)
5.
P68 RNA Helicase Unwinds the Human let-7 MicroRNA Duplex and Is Required
for let-7-Directed Silencing of Gene Expression (54)
5.1 Introduction (54)
5.2 Results (56)
in

5.2.1 Identification of an ATP-dependent Activity in Hela Cells That
Can Unwind the let-7 MicroRNA Duplex (57)
5.2.2 P68 RNA Helicase Co-Purifies With the let-7 MicroRNA Duplex
Unwinding Activity (62)
5.2.3 Recombinant P68 RNA Helicase Is Sufficient to Unwind the let-7
MicroRNA Duplex (66)
5.2.4 P68 RNA Helicase Is Required For let-7 MicroRNA Function in
Hela Cells (71)
5.3 Discussion (75)
6. The G4304A BRCAl Cancer Susceptibility Allele Creates a Novel Target Site
For the MicroRNA, MiR-501 (77)
6.1 Introduction (77)
6.2 Results (78)
6.2.1 Silent Mutations in the BRCAl mRNA Can Strengthen
Potential MicroRNA Target Sites (78)
6.2.2 The G4304A Mutation Suppresses mRNA Expression (82)
6.2.3 Suppression of the G4304A Mutant mRNA is MicroRNA-
Mediated (87)
6.2.4 MiR-501 Targets the G4304A Mutant mRNA (90)
6.3 Discussion (95)
7.
Methods and Materials (97)
8. Summary (104)
9. References (105)
iv

List of Figures
Figure 1: Crystal structure of
T.
thermophilus Argonaute bound to a 5'
phosphorylated 21-nucleotide DNA-guide strand (5)
Figure 2: Drosha processes the pre-microRNA from the pri-microRNA (9)
Figure 3: Dicer processes a microRNA duplex from the pre-microRNA (12)
Figure 4: Crystal structure of the
G.
intestinalis Dicer (16)
Figure 5: Model of microRNA and siRNA guide-strand specific-sorting
into Argonaute2 (20)
Figure 6: How does Argonaute know which strand of the microRNA duplex to
bind? (23)
Figure 7: Is there an activity in cells that can unwind the microRNA duplex? (25)
Figure 8: Can silent mutations create novel microRNA target sites? (26)
Figure 9: The let-7* strand does not silencing gene expression (32)
Figure 10: The let-7* strand does not readily complex with Argonaute2 (38)
Figure 11: The Argonaute2-let-7* strand binary complex does not recruit mRNA (41)
Figure 12: The let-7* strand does not adopt an appropriable structure for mRNA
recruitment (46)
Figure 13: Identification of a novel activity from human cells that unwinds the let-
7 microRNA duplex (58)
Figure 14: Purification of the let-7 microRNA duplex unwinding activity yields
P68 RNA Helicase as a likely candidate (62)
v

Figure 15: Purified recombinant His-P68 RNA Helicase is sufficient to unwind the
let-7 microRNA duplex (67)
Figure 16: P68 RNA Helicase is required for let-7 microRNA function in Hela
cells (71)
Figure 17: Silent mutations in the BRCAl mRNA can create microRNA target
sites (79)
Figure 18: The BRCAl G4304A mutant mRNA is suppressed in breast and ovarian
tumor cell lines. (83)
Figure 19: Suppression of
the
G4304A mutant mRNA is microRNA-mediated (87)
Figure 20: MiR-501 targets the G4304A mutant mRNA (90)
VI

Chapter 1: Synopsis
MicroRNAs are short, single-stranded, non-protein coding RNAs that negatively
regulate gene expression. MicroRNAs have been shown to play a critical role in many
biological processes and diseases. Therefore, it is critical to gain an understanding of the
factors that can regulate microRNA activity. It has been established that, microRNA
activity is exerted by a class of RNA-binding proteins called Argonautes. MicroRNAs
arise from a transient duplex (the microRNA duplex) from which only one strand (the
microRNA strand) is specifically sorted into an Argonaute protein, whereas the other
strand (the microRNA* strand) is not. This observation provokes the question; how is
the microRNA specifically sorted into an Argonaute protein? To this effect we have
found that the differing 5' terminal nucleotides of the microRNA and microRNA* strand
are a critical determinant for strand specific sorting of
the
microRNA into Argonaute.
Knowing that microRNAs proceed through a duplex intermediate we set out to identify a
microRNA duplex unwindase. Biochemical purification of a microRNA duplex specific
unwinding activity from cells yielded P68 RNA Helicase. We showed that P68 RNA
Helicase is sufficient to unwind a microRNA duplex, and also that P68 is necessary for
microRNA function in cells. Next we asked if silent mutations could create novel, non-
canonical microRNA target sites. We identified a silent variant (G4304A) in the BRCAl
mRNA, creates a novel microRNA target site for microRNA-501.
1

Chapter
2:
Introduction
MicroRNAs are short (19-24 nucleotide), single-strand, non-protein coding RNAs
that negatively regulate gene expression by annealing to complementary elements in the
3'
UTR of a target mRNA (Battel, 2004; He and Hannon, 2004; Nelson et al.,
2003;
Zamore and Haley, 2005). MicroRNAs have been shown to regulate many cellular
processes including differentiation, development, cell growth, and cell death (Bussing et
al.,
2008; Giraldez et al., 2005; Mishima et al., 2007; Nimmo and Slack, 2009). Mutation
or deletion of certain microRNAs has also been shown to cause disease (Calin et al.,
2002;
Esquela-Kerscher et al., 2008; Johnson et al, 2005; Kumar et al., 2008).
MicroRNAs were discovered in C. elegans where their mutation radically
affected developmental transitions required for the worm to proceed though the larval
stage to the adult life stage (Lee et al., 1993; Pasquinelli et al., 2000; Reinhart et al.,
2000;
Slack et al., 2000; Wightman et al., 1993). Since the discovery of the first
microRNAs in
C.
elegans microRNAs have been identified in a wide range of organisms
including plants, flies, mice, humans and even viruses (Aravin et al.,
2003;
Lagos-
Quintana et al., 2001; Lagos-Quintana et al., 2002; Lau et al., 2001; Lee and Ambros,
2001;
Pfeffer et al., 2005; Pfeffer et al., 2004; Reinhart et al., 2002).
The silencing effect of microRNAs is exerted through a novel class of RNA-
binding proteins called the Argonaute family (Carmell et al., 2002; Farazi et al., 2008;
Hock and Meister, 2008; Hutvagner and Simard, 2008; Peters and Meister, 2007; Tolia
and Joshua-Tor, 2007). Argonaute proteins were first identified in plants and worms
where their mutation abolished posttranscriptional gene silencing (Bohmert et al., 1998;
2

Fagard et al., 2000; Lynn et al., 1999; Tabara et al., 1999). The name Argonaute is
derived from the squid-like phenotype of the mutant plants (Argonaute in Greek, means
squid) (Bohmert et al., 1998). Members of the Argonaute protein family have been
shown to bind a variety of
small,
single stranded RNA including microRNAs, piRNAs
and esiRNAs (these RNAs are generically referred to as guideRNAs) (Carmell et al.,
2002;
Farazi et al., 2008; Hock and Meister, 2008; Hutvagner and Simard, 2008; Peters
and Meister, 2007; Tabara et al., 1999). Argonaute/guideRNA complexes have been
shown to silence gene expression through a variety of different mechanisms, including
translational inhibition, deadenylation and mRNA cleavage (Giraldez et al., 2006;
Hutvagner and Zamore, 2002; Llave et al., 2002; Wakiyama et al., 2007; Wightman et
al.,
1993).
Since the discovery of Argonaute proteins, 4 Argonaute homologues have been
identified in humans (Carmell et al., 2002; Farazi et al., 2008; Hock and Meister, 2008;
Hutvagner and Simard, 2008; Peters and Meister, 2007; Tolia and Joshua-Tor, 2007). All
Argonaute proteins have 3 conserved domains: the PIWI domain, the MID domain and
the PAZ domain (Cerutti et al., 2000). The PIWI domain of Argonaute proteins harbors
characteristics of the RNase H endonuclease domain (Patel et al., 2006; Tolia and Joshua-
Tor, 2007). The PIWI domain of human Argonaute2 contains an RNase catalytic triad
(DDH) and is the only human Argonaute protein believed to have endonuclease activity
(Felice et al., 2009; Liu et al., 2004; Rivas et al., 2005). Indeed, human Argonaute2 has
been shown to be an RNA-directed endonuclease, capable of cleaving a target mRNA
(Felice et al., 2009; Hammond et al., 2001; Hutvagner and Zamore, 2002; Liu et al.,
2004;
Martinez et al., 2002; Meister et al., 2004; Rivas et al., 2005). Moreover, site
3

directed mutagenesis has confirmed that the DDH catalytic triad is in fact required for
RNA-directed endonuclease activity (Felice et al., 2009; Liu et al., 2004; Rivas et al.,
2005).
Biochemical and crystallographic data has proven that the MID domain contains a
5'
phosphate-binding pocket for the guideRNA (Rivas et al., 2005; Wang et al., 2008).
Indeed, mutation or deletion of the MID domain abolishes guideRNA-directed mRNA
silencing (Okamura et al., 2004; Wang et al., 2008). Interestingly, the MID domain also
displays a high level of specificity for the 5' terminal nucleotide of an associated
microRNA (Mi et al., 2008). The PAZ domain (standing for PIWI, Argonaute and
Zwille/Pinhead) is a highly conserved domain among genes involved in
posttranscriptional gene silencing (Cerutti et al., 2000). Biochemical and
crystallographic data of the PAZ domain from various Argonaute proteins has implicated
a role in the recognition of the 3' end of the guideRNA (Lingel et al.,
2003;
Song et al.,
2003;
Yan et al., 2003). However, mutation and/or deletion of the PAZ domain in
Argonaute proteins showed that it is dispensable for guideRNA-directed mRNA-cleavage
(Okamura et al, 2004; Wang et al., 2008).
4

Figure 1:
3'
terminal
nucleotide
v
'X- v>
S' terminal
nucleotide
4
/" .• <-*!
-'*«,
m
W
V,/
V
' i » fc.
ti
N LI PAZ L2 MID 1 D D H
FIWI
3'
end binding 5' end binding RNA-directed endonuclease
5

Figure 1: Crystal structure of
T.
thermophilus Argonaute bound to a 5'
phosphorylated 21-nucleotide DNA-guide strand. Ribbon (Argonaute), and stick
(DNA) diagram of the Argonaute/DNA-guide strand co-crystal (Wang et al., 2008). The
individual domains are color-coded. The bound 21 nucleotide DNA-guide strand is
colored red (with phosphorous atoms in yellow). The 5' and 3' terminal nucleotides are
indicated. Below is a linear diagram of the
T.
thermophilus Argonaute protein. The
approximate positions of the amino acids comprising the catalytic triad are indicated
(DDH).
6

Importantly, analysis of the microRNA component of Argonaute-containing
silencing complexes from cells has shown an association of single stranded RNA (Dostie
et al.,
2003;
Mi et al., 2008; Mourelatos et al, 2002). Indeed, biochemical analysis of
recombinant and affinity-purified Argonaute proteins has shown a strict requirement for
single stranded RNAs (Liu et al., 2004; Maniataki and Mourelatos, 2005; Martinez et al.,
2002).
However, microRNAs are not transcribed from the genome as short, single
stranded RNAs, rather they are generated through a series of processing events.
MicroRNAs are transcribed from the genome as a long primary transcript (or pri-
microRNA), often several kilobases long by RNA polymerase-II (Lee et al., 2002; Lee et
al.,
2004). The pri-microRNA adopts a hairpin structure, which is recognized and
processed by the microprocessor complex into an approximately 70 nucleotide hairpin
RNA called the pre-microRNA in the nucleus (Denli et al., 2004; Gregory et al., 2006;
Gregory et al., 2004; Lee et al., 2002; Seitz and Zamore, 2006). The microprocessor
complex is comprised of the RNaselll enzyme Drosha and the double-stranded RNA-
binding protein DGCR8 (Denli et al., 2004; Gregory et al., 2006; Gregory et al, 2004;
Seitz and Zamore, 2006). Drosha is a highly conserved, 160kDa protein containing 2
tandem RNaselll domains and one double-stranded RNA-binding domain, both of which
have been shown to be critical for pri-microRNA processing (Han et al., 2004; Han et al.,
2006).
DGCR8 (DiGeorge syndrome critical region gene 8, also known as Pasha in
worms and flies) is a highly conserved, 120kDa protein containing 2 tandem double-
stranded RNA-binding domains (Yeom et al., 2006).
Biochemical analysis of the pro-microRNA processing event has shown that
DGCR8 recognizes the terminal loop of the pri-microRNA hairpin and directs Drosha
7

cleavage approximately 7 helical turns away from the terminal loop (Han et al., 2006;
Lee et al., 2003). Drosha cleaves both strands of the stem, resulting in an approximately
70 nucleotide hairpin RNA with a
2
nucleotide overhang on the 3' end and a recessed 5'
phosphate (Basyuk et al.,
2003;
Han et
al.,
2006; Lee et al., 2003). This product is
termed the pre-microRNA (Figure 2).
8

Figure
2:
pri-microRNA
5' 1 3'
Drosha/DGCR8
pre-microRNA
9

Figure 2: Drosha processes the pre-microRNA from the pri-microRNA.
MicroRNAs (red) are transcribed as a long primary transcript (pri-microRNA). The pri-
microRNA adopts a hairpin structure, which is recognized by the Drosha/DGCR8
microprocessor complex. Drosha cleaves both strands of the pri-microRNA resulting in
an approximately 70 nucleotide hairpin RNA with 2 nucleotide over hangs on the 3' end
and a recessed 5' phosphate.
10

Following nuclear processing by the microprocessor complex, it is believed that
the pre-microRNA is exported to the cytoplasm by Exportin-5, a member of the Ran-GTP
nuclear export family (Lund et al., 2004; Yi et al.,
2003;
Zeng and Cullen, 2004). It has
been shown that Exportin-5 is required for pre-microRNA and microRNA accumulation
in the cytoplasm (Lund et al., 2004; Yi et al.,
2003;
Zeng and Cullen, 2004). However,
the pre-microRNA does not accumulate in the nucleus following Exportin-5 knockdown,
thus,
it has been speculated the Exportin-5 is also required for pre-microRNA stability
(Yi et al., 2003).
Following export to the cytoplasm, the pre-microRNA is recognized and
processed by the RNaselll enzyme, Dicer and a double-stranded RNA-binding protein
TRBP (HIV trans-activating response element binding protein) (Bernstein et al., 2001;
Chendrimada et al., 2005; Gregory et al., 2005; Haase et al., 2005). Dicer was originally
discovered for its ability to cleave long double-stranded RNA into approximately 21
nucleotide RNAs called siRNAs (Bernstein et al., 2001; Hutvagner et al., 2001; Ketting
et al., 2001). Therefore, it was speculated that Dicer would also be involved in
microRNA biogenesis. Indeed, it was shown that recombinant and affinity-purified Dicer
could generate approximately 22 nucleotide RNAs from the pre-microRNA (Provost et
al.,
2002[Zhang, 2004 #88; Zhang et al., 2002) (Figure 3). Dicer has been shown to be
critical for microRNA biogenesis in a wide variety of organisms, as depletion of Dicer
resulted in a decrease of mature microRNA and a concomitant accumulation of
pre-
microRNA (Bernstein et al.,
2003;
Giraldez et al, 2005; Hutvagner et al.,
2001;
Ketting
etal.,2001).
11

Figure
3:
pre-microRNA
Dicer
microRNA
duplex
12

Figure
3:
Dicer processes a microRNA duplex from the pre-microRNA. The pre-
microRNA is recognized and processed by Dicer. Dicer cleaves both strands of the pre-
microRNA producing an approximately 20 nucleotide base paired duplex, with 2
nucleotide over hangs on the 3' ends and recessed 5' phosphates. The strands of the
microRNA duplex are called
the
microRNA (red) and microRNA* (black).
13

Dicer is a highly conserved protein found in almost all eukaryotic organisms
including plants and fungi (Bernstein et al., 2001 [Bernstein, 2003
#90)].
Some
organisms such as plants and flies have multiple Dicer homologues each with differing
specificity. Flies for example, express 2 Dicer homologues, Dicer-1 andDicer-2. Dicer-
1 is required for pre-microRNA generation, whereas Dicer-2 is required for siRNA
generation, the substrate specificity of these homologues is not fully understood
(Forstemann et al., 2007; Vagin et al., 2006). Mammalian cells on the other hand,
express only one Dicer. Mammalian Dicer is a 220kDa protein containing an N-terminal
DExD box helicase domain, a PAZ domain, 2 tandem RNaselll domains and a double-
stranded RNA-binding domain (Ma et al., 2004; MacRae et al., 2007; Macrae et al.,
2006).
While the exact function of the DExD box helicase domain remains elusive,
biochemical data has shown that the addition of ATP can stimulate siRNA processing by
Dicer, however the addition of ATP had no effect on pre-microRNA processing (Provost
et al., 2002; Zamore et al., 2000). The PAZ domain of Dicer is required for siRNA and
pre-microRNA processing. Structural analysis of the PAZ domain has implicated a role
in binding to the 3' end structure of
the
pre-microRNA (Ji, 2008; Ma et al., 2004;
MacRae et al., 2007; Macrae et al., 2006). Crystallographic studies on G. intestinalis
Dicer have invoked a 'ruler' model, in which the, 3' end structure of the pre-microRNA
hairpin is recognized by the PAZ domain of Dicer. The PAZ domain acts as a platform,
which is connected to a 'ruler', which orients the 2 RNaselll domains approximately 2
helical, turns (22 nucleotides) on the stem of the pre-microRNA (MacRae et al., 2007;
Macrae et al., 2006) (Figure 4). Mutational analysis of recombinant Dicer has shown that
the fidelity of both RNaselll domains is required for pre-microRNA and siRNA
14

processing (Zhang et al., 2004). The RNaselll domains of Dicer cleave both strands of
the pre-microRNA stem generating an approximately 22 nucleotide, symmetrical RNA
duplex harboring 2 nucleotide overhangs on the 3' ends and recessed 5' phosphates
(Zhang et
al.,
2004). This 22 nucleotide RNA duplex is termed the microRNA duplex (in
the case of microRNAs) or siRNA duplex (in the case of
siRNAs).
The strands of
microRNA duplex are termed the microRNA and microRNA* strand, whereas the strands
of the siRNA duplex are termed the guide and passenger strand.
15

Figure 4:
semester 7^ ©^ <-, ' '•••'•••'-
^_ %,. ''SFV
M sapiens
Hlxlt
G. intestinalis
PAZ Nascllli
-Vase UlfaH
>asc lllbj
16

Figure
4:
Crystal structure of the
G.
intestinalis
Dicer.
Ribbon diagram with color-
coded domains (Macrae et al., 2006). Magnesium ions are indicated (purple spheres) and
linear box diagram of H.
sapiens
and
G.
intestinalis
Dicer
proteins.
Colored domains are
oriented to their relative position compared to human Dicer.
17

TRBP (loquacious in Drosophila) is a highly conserved 55kDa protein containing
3 double-stranded RNA-binding domains (Chendrimada et al., 2005; Haase et al., 2005).
TRBP has been shown to associate with Dicer and is believed to be involved in strand-
specific sorting of the microRNA into Argonaute2 (Chendrimada et al., 2005; Haase et
al.,
2005). It is thought that Dicer and TRBP form a heterodimeric complex that
asymmetrically binds to the microRNA duplex, with TRBP bound to the
thermodynamically stable end, and Dicer bound to the thermodynamically weaker end of
the microRNA duplex (Chendrimada et al., 2005; Haase et al., 2005). TRBP is believed
to recruit Argonaute2 to the ribonucleoprotein complex (Chendrimada et al., 2005).
However, the exact mechanism by which this occurs is not clear and is the focus of much
debate in the field.
Nevertheless, despite the apparent symmetrical structure of the duplex, there is a
remarkable level of specificity in the subsequent expression of the two strands. Deep
sequencing analysis of cellular cDNA has shown that only one strand (the microRNA) is
present in cells, whereas the other strand (the microRNA*) is not readily detectable
(Aravin et al.,
2003;
Lagos-Quintana et al., 2002; Lau et al., 2001; Lim et al., 2003).
Insight into this selectivity was originally provided by the reconstitution of pre-
microRNA-directed mRNA silencing. In these experiments, incubation of the let-7 pre-
microRNA with affinity-purified silencing complexes led to the robust and specific
cleavage of a let-7 target mRNA, with little effect upon a let-7* strand target mRNA
(Gregory et al., 2005). This result invoked a model in which, the differing
thermodynamic stability of
the
duplex ends as an asymmetric feature, and a critical
determinant of strand selectivity.
18

Indeed, subsequent experiments suggested that the asymmetrical binding of
a
Dicer/TRBP complex was the key determinant for the selective sorting of
let-7
into
Argonaute2. Whereas the let-7* strand is believed to be catabolized by nonspecific
nucleases. This model for loading distinctly echoes the model for the specific sorting of
the siRNA guide strand into Argonaute2. In which Dicer-2 forms a heterodimeric
complex with the double-stranded RNA-binding protein R2D2 (TRBP homolog in
Drosophila) (Tomari et al., 2004). The Dicer-2/R2D2 heterodimer asymmetrically binds
to the siRNA duplex with R2D2 bound to the thermodynamically stable end and Dicer-2
bound to the thermodynamically weaker end (Tomari et al., 2004). Argonaute2 is then
recruited to the ribonucleoprotein complex, recognizes and binds to the 5' end of the
guide strand (Schwarz et al., 2003). Once the guide strand is engaged with Argonaute2,
Argonaute2 cleaves the passenger strand facilitating loading into Argonaute2 (Matranga
et al., 2005). While cleavage of the siRNA passenger strand facilitates loading of the
guide strand into Argonaute2, microRNAs often contain internal bulges precluding
passenger strand cleavage.
19

Figure 5:
I
5' p Dicer
TRBP
i
JnQiQirDp"
i
SnQnQinF
i
Argonaute
S'
_ Dicer
Y- R2D2
Argonaute
20

Figure 5: Model for microRNA and siRNA guide-strand specific sorting into
Argonaute2. The pre-microRNA is recognized and cleaved by Dicer/TRBP. Following
Dicer cleavage, the Dicer/TRBP heterodimer asymmetrically bind to the microRNA
duplex. Argonaute is recruited to the ribonucleoprotein complex by TRBP and
recognizes the 5' end of the microRNA strand. Argonaute binds to the duplex and the
microRNA* strand is somehow removed, resulting in an Argonaute/microRNA complex.
The current model for sorting an siRNA guide strand into Argonaute. The Dicer/R2D2
heterodimer asymmetrically bind to the siRNA duplex. Argonaute is recruited to the
complex and binds the 5' end of
the
guide strand. The passenger strand is cleaved,
resulting in an Argonaute/siRNA complex
21

Interestingly, recent data has indicated that the 5' terminal uracil of let-7 is critical
for Argonaute2-catalyzed mRNA silencing (Felice et al., 2009). It has also been recently
shown that the 5' terminal nucleotide of plant microRNAs can dictate their assortment
into specific Argonaute family members (Mi et al., 2008). It is now appreciated that the
majority of microRNA/microRNA* strands have different 5' terminal nucleotides. In
particular the human let-7 microRNA is 22 nucleotides long, and that cleavage of the
hairpin by Dicer yields a let-7 with a 5' terminal uracil and a let-7* with a 5' terminal
cytosine (Landgraf et al., 2007; Zhang et al., 2004).
Therefore, we hypothesized that the preferential sorting of let-7 into Argonaute2
may also be dictated by the differing terminal 5' terminal nucleotides of the let-7 and let-
7*
strands. This material is discussed in Chapter 2. Chapter 3 of this dissertation focuses
on the identification of a novel unwinding activity in Hela cell, which we identified as
P68 RNA Helicase, and showed to be necessary for microRNA function, and sufficient to
unwind the let-7 microRNA duplex. Finally, chapter 4 will test the hypothesis that silent
mutations in the open reading frame of an mRNA can create novel, non-canonical
microRNA target sites.
Chapter
3:
Premise
This thesis addresses three main questions: 1) How does Argonaute know which
strand of
the
microRNA duplex to bind? 2) Is there an activity in cells that can unwind
the microRNA duplex? 3) Can silent mutations create novel microRNA target sites?
22

Together these three questions encompass a broader topic describing the regulation of
microRNA activity.
MicroRNAs arise from a transient duplex (the microRNA duplex) from which
only one strand (the microRNA strand) is specifically sorted into an Argonaute protein,
whereas the other strand (the microRNA* strand) is not. This observation provokes the
question; how is the microRNA specifically sorted into an Argonaute protein?
Figure 6:
microRNA
duplex
Argonaute
How Does Argonaute Know Which Strand of the MicroRNA Duplex to Bind?
Utilizing an in vivo reporter assay as well as an in vitro cleavage assay using
recombinant Argonaute2 we found that let-7 has superior silencing activity compared to
let-7*.
By studying the sub-reactions of the silencing pathway in vitro, we have
discovered that the let-7 and let-7* strands can be loaded into Argonaute2, however the
23

let-7*
strand binary complex is unable to recruit an mRNA. This result led us to studying
the structure of the let-7 and let-7* strands in the binary complex. Footprint analysis
showed that the 5' end of let-7 (also referred to as the microRNA 'seed' sequence) is
accessible whereas the 5' end of
let-7*
is not. Additionally, we have shown that deletion
of
the
5' terminal nucleotide of
the
let-7* strand (cytosine to uracil) is capable of rescuing
the let-7* deficiency. We have concluded from our results that the differing 5' terminal
nucleotides of
the
let-7 and let-7* strand indeed play an important role in strand specific
sorting into Argonaute2.
MicroRNAs arise from a precursor duplex termed the microRNA duplex. It is
clear that the, silencing activities of microRNAs are exerted by a class of RNA-binding
proteins called Argonautes. Argonaute proteins have been shown to have a much greater
affinity for a microRNA than the precursor duplex. Moreover, analysis of the microRNA
component of Argonaute containing silencing complexes from cells has shown the
association with only the microRNA strand of the duplex. Therefore, we hypothesized
that there must be an activity in cells that can unwind the microRNA duplex.
24

Figure 7:
microRNA
duplex
5'InnQniAiED["
Is There an Activity in Cells That Can Unwind the MicroRNA Duplex?
Using an in vitro unwinding assay, we were able to identify and purify a 68kDa,
ATP-dependant activity in Hela cell extracts that was capable of unwinding the let-7
microRNA duplex. Further purification of the unwinding activity suggested that P68
RNA Helicase was a likely candidate. Indeed, we found that recombinant P68 was
sufficient to unwind the let-7 duplex and using an in vivo reporter assay we found that
depletion of
P68
abrogated let-7 activity. From this we concluded that P68 RNA
Helicase is a critical component of the microRNA pathway.
Individuals harboring germ line mutations in the tumor suppressor gene BRCA1
have a much-increased risk of developing cancer. The majority of germ line mutations in
BRCA1 result in a truncated or mutated protein that lacks tumor suppressor function.
However, a number of single nucleotide sequence variants have been identified in the
open reading frame of BRCA1 mRNA, that still encode the same amino acid (silent
variants) yet are associated with an increased risk of cancer. It has been previously
25

established that the alteration of
a
single base pair is sufficient to modulate microRNA-
silencing activity. Thus, we hypothesized that silent variants might create a novel
microRNA target site and thereby suppress BRCA1 expression.
Figure 8:
DNA
mRNA
Norma] Individual Individual Harboring a Silent Variant
silent
H r variant
I
I
BHCA1 Tbfe
TTTTTTT
I miRNA
\
bttCAl
f
t
Can Silent Mutations Create Novel MicroRNA Target Sites?
After screening a number of silent variants using an in vivo reporter assay, we
have identified a silent variant (G4304A) that suppressed the expression of
the
mutant
mRNA compared to the wild type counterpart. Depletion of Dicer (a critical enzyme
required for microRNA biogenesis) abrogated the suppressive effect of
the
mutant
mRNA, implicating microRNA-mediated suppression. In silico analysis predicted that
microRNA-501 was a likely target for the mutant mRNA. Indeed, inhibition of
26

microRNA-501 activity inhibited the suppressive effect of the mutant mRNA, whereas
over-expression of microRNA-501 further attenuated the suppressive effect of the mutant
mRNA. From this we concluded that the G4304A silent variant could create a novel
microRNA target site.
Chapter 4: The Differing Termini of Let-7 and Let-7* Play a Critical Role in
Strand Specific Sorting into Argonaute2
4.1:
Introduction
The understanding of the mechanisms that control gene regulation has been
revolutionized by the discovery that small RNAs can modulate gene expression. There
are many classes of small RNAs including microRNAs (Lee et al., 1993; Reinhart et al.,
2000;
Wightman et al., 1993), piRNAs (Aravin et al., 2006; Brennecke et al., 2007;
Girard et al., 2006; Saito et al., 2006) and esiRNAs (Czech et al., 2008; Ghildiyal et al.,
2008).
These various classes of RNA likely modulate gene expression in different ways,
but it is believed that they all function by associating with a protein effector, typically a
member of the Argonaute family (Carmell et al., 2002; Farazi et al., 2008; Peters and
Meister, 2007; Tolia and Joshua-Tor, 2007).
MicroRNAs are derived from primary transcripts that adopt a hairpin structure
(Lee et al., 2003). This hairpin is recognized and cleaved by Drosha near the base of the
stem resulting in an approximately 70 nucleotide hairpin RNA with a 2 nucleotide
overhang on the 3' end and a recessed 5' phosphate (Basyuk et al.,
2003;
Lee et al.,
27

2002).
This product, the pre-microRNA, is the substrate for a second enzyme called
Dicer (Bernstein et al., 2001). Dicer recognizes the base of the hairpin, measures up
about 22 nucleotides and cleaves both strands of the pre-microRNA, generating a
symmetrical duplex, termed the microRNA duplex (Macrae et al., 2006; Zhang et al.,
2004).
Despite the apparent symmetrical structure of the microRNA duplex, there is a
remarkable level of specificity in the subsequent expression of
the
two strands. Deep
sequencing analysis of cellular cDNA has shown that only one strand (the microRNA) is
present in cells, whereas the other strand (the microRNA*) is not readily detectable
(Lagos-Quintana et al., 2001; Lagos-Quintana et al., 2002; Landgraf et al., 2007; Lau et
al.,
2001; Lee and Ambros,
2001;
Lim et al, 2003). Insight into this selectivity was
originally provided by the reconstitution of pre-microRNA hairpin directed silencing of
mRNA. In these experiments, incubation of the let-7 pre-microRNA with affinity
purified silencing complexes led to the robust and specific cleavage of a let-7 target
mRNA, with little effect upon a let-7* strand target mRNA (Gregory et al., 2005;
Maniataki and Mourelatos, 2005). This led to a model, which invoked the differing
thermodynamic stability of
the
duplex ends as an asymmetric feature and a critical
determinant of strand selectivity (Gregory et al., 2005). Indeed, subsequent experiments
suggested that the asymmetrical binding of a Dicer/TRBP complex was the key
determinant for the selective sorting of the let-7 strand into Argonaute2 (Chendrimada et
al.,
2005). The let-7* strand is not sorted into Argonaute2 and is believed to be degraded
by nonspecific nucleases (Bartel, 2004). Importantly, it was noted that let-7 or let-7*
provided as single strand species could both silence mRNA (Gregory et al., 2005;
28

Maniataki and Mourelatos, 2005). From this, it was concluded that the appropriate
sorting of the microRNA duplex strands into Argonaute2 was obligatorily coupled to the
processing of the pre-microRNA.
However, recent data has indicated that the 5' terminal uracil of let-7 is critical for
Argonaute2-catalyzed mRNA silencing (Felice et al., 2009). It has also been recently
appreciated that the 5' terminal nucleotide of plant microRNAs can dictate their
assortment into specific Argonaute family members (Mi et al., 2008). Most importantly,
it is now appreciated that human let-7 is 22 nucleotides long and that cleavage of the
hairpin by Dicer yields a let-7 with a uracil 5' terminus and a let-7* with a cytosine 5'
terminus (Landgraf et al., 2007; Zhang et al., 2004). Thus, we hypothesized that the,
preferential sorting of let-7 into Argonaute2 may also be dictated by the differing termini
of let-7 and let-7*.
4.2:
Results
4.2.1:
The let-7* Strand Does Not Silence Gene Expression In Vivo or In Vitro
To address our hypothesis we began by looking at let-7 and let-7* activity. In
previous studies, we have employed a luciferase reporter construct (let-7 sensor) to
measure the silencing activity of let-7 in Hela cells (Figure 9A, top panel). In those
studies, it was established that Argonaute2 was the principal effector of let-7 activity
(Felice et al., 2009). To determine the possible silencing activity of
let-7*
in Hela cells,
we designed a reporter construct containing a fully complementary target element for the
29

let-7*
strand (let-7* sensor) (Figure 9A, bottom panel). We found that endogenous let-7
exhibited a robust level of silencing activity whereas let-7* silencing activity was
undetectable (Figure 9B). Thus result confirms that the let-7* strand is not efficiently
sorted into an Argonaute protein in Hela cells.
Previous models have implied that strand specific sorting of microRNAs into
Argonaute2 is coupled to the cleavage of the pre-microRNA by Dicer (Gregory et al.,
2005;
Maniataki and Mourelatos, 2005). We decided to test the sufficiency of this model
by introducing let-7 and let-7* into Hela cells. If the obligate coupling to Dicer cleavage
was sufficient to impart specificity than we would expect that let-7 and let-7* would have
comparable silencing activities.
Although single strand let-7 was much less efficient than the duplex derivative it
was capable of silencing reporter expression (Figure 9C, top panel). On the other hand,
let-7*
was significantly impaired in its ability to silence the expression of its sensor
reporter. However, to conclude that this is the case, it is important to show that the
sensor reporter can indeed work. Previous work has shown that a
21
nucleotide
derivative of the let-7* strand (bearing a 5' terminal uracil) was capable, albeit with less
efficiency, of directing mRNA silencing in vitro. In the present studies, let-7* (5'-U) did
silence the sensor reporter and provided positive validation (Figure 9C, bottom panel).
Thus,
our data suggests that let-7* cannot efficiently direct mRNA silencing.
However these experiments were done in a cellular system and may be potentially
confounded by let-7* specific deactivating factors. To circumvent this problem we
assayed the ability of let-7 and let-7* to direct mRNA silencing in a purified system,
substantially free of confounding activities.
30

To do this, recombinant human Argonaute2 was pre-incubated with either let-7 or
let-7*
and then presented with a fully complementary target mRNA. Sorting into
Argonaute2 was measured by the specific cleavage of the labeled target mRNA. As
previously documented, let-7 directed the Argonaute2-catalyzed cleavage of a target
mRNA. However, let-7* directed Argonaute2-catalyzed cleavage of a target mRNA was
not detected. Quantitative analysis showed a 65 fold difference between the amount of
let-7 and let-7* directed cleavage using lOnM guide RNA. To validate that the target
mRNA used to assay let-7* can in fact work, we assayed let-7* (5'-U). Let-7* (5'-U) did
in fact direct mRNA cleavage, but not as efficiently as let-7 (Figure 9D). From this we
concluded that Argonaute2 itself might play a significant role in the preferential selection
for the strands of
the
microRNA duplex.
31

ure
9:
WT A6UAUACUUAACUAUACAACCUACUACCUCAAC6CGAU6UA
MUT A6UAUACX7UAACUAUACAACCUACUAGGAGAAC6C6AU6UA
WT A6UAUACCUU66AAAGACA6UAGAUU6UAUA6C6C6AU6UA
MUT AGUAUACCUU66AAA6ACA6UAGAUU6AUAU6C6CGAU6UA
---•'
32

CO
Normalized Luciferase (Renilla/Firefly)
H
3

c.
75-
P
js
§ 50-
e
a
%
X
0-i
1
RNA (uM) -
Fold
Repression
(MUTANT)
A.
RNA (uM) -
,.il
2 3 4 5
.1 1 .1 1
let-7 let-7
mutant
••llil
2 3 4 5 6 7
.1 1 .1 1 .1 1
let-7* let-7* let-7*
mutant (5'-U)
— u
<
.-s ^
> 3
•x o
w e
^ en
* o
60-
30-
0-
4-i
2 -
0-
0
-•-let-7
D 0.5 1
[Guide
RNA]
(nM)
-•-let-7* -•-let-7* (5'-U)
'•ill
0.5 1
[Guide RNA] (nM)
34

D.
let-7 target let-7* target
let-7
Guide RNA mutant
mutan^*-"! let-7 let-7"
mutant. let-7"
5'-lL
let-7"
5' cleavage
product
12 3 4 5 6 7 8
• «•»: MR
9 1011 12 131415161718 19 20
«•>' ^^ —— -U0 iff0
.Kr-J...
-. W *"* •
140-1 •— let-7
let-7"
let-7* 5*-U
0 4 6 8
[Guide RNA] (nM)
10
35

Figure 9: The let-7* strand does not silencing gene expression. A. A schematic
representation of the dual luciferase reporter used in the following experiments. The
sequences of the fully complementary let-7 and let-7* target elements that were
subcloned into the 3'UTR of the renilla luciferase gene are shown. For each target
element, mutations were made in the nucleotides complementary to the 'seed' sequence.
Indicated by the black bar. B. Hela cells were transfected with let-7 or let-7* sensor
(50ng) containing either the WT or MUT target elements. The cells were lysed 12 hours
after transfection and analyzed for dual luciferase activity. C. Hela cells were co-
transfected either the let-7 or let-7* sensor (50ng) containing either, the WT or MUT
element, and single strand RNA (as indicated). The cells were lysed 12 hours after
transfection and analyzed for dual luciferase activity. Data is expressed as the fold
repression of the MUT/WT type sensor. A graphical representation of the exogenous
RNA activity is shown in the lower panel, in which the values of the mutant RNA were
subtracted as background for each case. D. InM
5'-32P
end labeled let-7 or let-7* target
RNA was assay for cleavage by GST-Argonaute2 that had been pre-incubated with
.1,
.2,
.5,
or InM guide RNA as indicated. Following incubation the reactions were analyzed by
denaturing polyacrylamide gel electrophoresis. The position of the 5' cleavage product is
indicated. Mutant let-7 and let-7* RNAs are used as negative controls in this assay.
Quantification of cleavage activity is shown below.
36

4.2.2:
The let-7* Strand Does Not Readily Complex With Argonaute2
Our results have shown that let-7, but not let-7* can direct Argonaute2-catalyzed
cleavage of a target mRNA. Therefore, we postulated that either let-7* may not be
efficiently loaded into Argonaute2 or it may be loaded yet form a complex that cannot
recruit mRNA. To distinguish between these two possibilities, we first examined the
formation of
the
Argonaute2/microRNA binary complex. To do this,
5'end
labeled let-7
or let-7* was incubated with recombinant Argonaute2 and analyzed by native gel
electrophoresis for complex formation. Incubation of both let-7 and let-7* with
Argonaute2 resulted in the formation of slowly migrating complex that increased with the
concomitant addition of
protein.
No complex formation was observed upon incubation
with GST, which served as a negative control for this experiment. We observed that
Argonaute2 bound to let-7 approximately 5 fold greater than let-7* (Figure 10).
37

Figure 10:
5'-32P
lct-7
5'-32P
let-7*
Ago>i GSJ^ Ago>, GSJ^,
[Protein] (nM) - J^\ ^\ - ^\ ^\
1 2 3 4 5 6 7 8 9 10 11 12 1114
38

Figure 10: The let-7* strand does not readily complex with Argonaute2. A. InM 5'-
32P end labeled let-7 or
let-7*
was incubated with
20,
40 or 80nM GST or GST-
Argonaute2 D597A for 30 minutes at 37 degrees Celsius. Following incubation
50%
of
the reaction was electrophoresed in a
1%
agarose gel at 40V.
39

4.2.3:
The let-7* Strand Does Not Recruit mRNA
We have shown that let-7* is not efficiently loaded into Argonaute2, however this
difference does not quantitatively account for the inability of
let-7*
to direct Argonaute2-
catalyze mRNA cleavage. Thus, we investigated the ability of the Argonaute2/let-7 and
Argonaute2/let-7* binary complexes to recruit their target mRNAs. To do this in a
quantitative fashion, we determined the ability of the binary complexes to adsorb a
labeled target mRNA to nitrocellulose.
To establish and validate this assay, we tested the ability of Argonaute2 loaded
with let-7 and miR-16 microRNAs to recruit their respective target mRNAs. Argonaute2
loaded with let-7 was capable of recruiting its labeled target mRNA, whereas Argonaute2
loaded with miR-16 was not (Figure 11A). On the other hand, Argonaute2 loaded with
miR-16 recruited its target mRNA, whereas Argonaute2 loaded with let-7 did not (Figure
1
IB). From this we concluded that we had established a quantitative assay for mRNA
recruitment.
Using this assay we tested the ability of the Argonaute2/let-7* binary complex to
recruit its labeled target mRNA. The Argonaute2/let-7* binary complex displayed a 32
fold difference in its ability to recruit mRNA when compared to the Argonaute2/let-7
binary complex. However, Argonaute2 loaded with let-7* (5'-U) could recruit the same
target mRNA (Figure 11C). Thus, we concluded that the failure to recruit mRNA is
likely the major contributor in the inability of
let-7*
to direct Argonaute2-catalyzed
mRNA cleavage.
40

H
Recruited Target RNA (fmol)
o
9 *•
\
3
QfQ
C
•1
Recruited Target RNA (fmol)
VI <1
° S
B ©
a
n
JO
Z
>
2 5!
•la
3
?

c.
•let-7*
•let-7*
5'-U
miR-16
10 15
[Guide
RNA] (nM)
42

Figure 11: The Argonaute2-let-7* strand binary complex does not recruit mRNA.
Recombinant GST-Argonaute2 (80nM) was pre-incubated with increasing amounts of
guide RNA (as indicated) for 30 minutes. InM
5'-32P
let-7 target (A.), miR-16 target
(B.) or let-7* target (C.) RNA was added followed by a 15 minute incubation. The
reactions were filtered through nitrocellulose and bound radioactivity was measured by
scintillation counting. The values of tandem reactions containing GST were subtracted as
background. The results are expressed as the amount of target mRNA recruited.
43

4.2.4: Analysis of the let-7 and let-7* Strand Binary Complexes Reveals That the
let-7* Strand Does Not Adopt an Appropriate Structure for mRNA
Recruitment
So far we have determined that Argonaute2 can bind let-7*, albeit with a lower
affinity than let-7, however the resultant complex is remarkably impaired in its ability to
recruit a target mRNA. Thus, we hypothesized that there might be a difference in the
structure of the let-7 and let-7* RNAs within the binary complex. To test this we used a
footprinting strategy in which the dynamic structure of the RNA bound by Argonaute2
could be reported by the accessibility of
the
phosphodiester bonds to lead hydrolysis.
First, we analyzed the structure of let-7 in the binary complex. To do this 5' end-
labeled let-7 was pre-incubated with Argonaute2 or GST as a negative control. After
incubation the RNA was subjected to lead hydrolysis and subsequent analysis by
denaturing get electrophoresis. The extent of phosphodiester bond accessibility was
determined by densitometric quantification of
the
RNA bands. The results were
expressed as the fold difference of Argonaute2 over GST. Thus, a value less than 1
indicates a protected bond, whereas a value of
1
or greater than
1
indicates accessible and
hyper-accessible bonds respectively.
Footprinting analysis of let-7 in the binary complex showed that the 5' end of let-
7 displayed a remarkable level of accessibility and hyper-accessibility to lead hydrolysis
(bonds 2-8), whereas the middle and 3' end of let-7 were protected (bonds 12-21) (Figure
12A).
Interestingly, the accessible and hyper-accessible region of let-7 occurred between
bonds 2-8 in a region described as the microRNA 'seed' sequence. Previous data has
44

shown that complementarity between at least 6 nucleotides in the 'seed' sequence and a
target mRNA is required for microRNA-directed mRNA silencing in vivo and in vitro.
Now that we knew the structure of let-7 in the Argonaute2 binary complex, we
investigated the structure of
let-7*
in the Argonaute2 binary complex. Our results
showed that the let-7* displayed modest susceptibility to lead hydrolysis across most of
the RNA from bonds 2-10 and 12-14, bond 11 displayed hyper-accessibility. Whereas,
the 3' end of
let-7*
was protected from bonds 15-21 (Figure 12B).
Importantly, footprint analysis of
let-7*
(5'-U) in the Argonaute2 binary complex
displayed a similar pattern to let-7. Let-7* (5'-U) displayed accessibility at bonds 2-6
and 14 and hyper-accessibility at bonds 7-13 across the 5' and middle region of the RNA
and protection of bonds 15-20 at the 3' end. Interestingly, the region of hyper-
accessibility was shifted down 2 phosphodiester bonds (Figure 12C).
From this data we conclude that let-7 and let-7* are capable of being bound by
Argonaute2, yet adopt different conformations. This observation coupled with our data
from the mRNA recruitment assay provides a unique link between the structure and
function of the microRNA 'seed' sequence, and suggests that the hyper-accessible (and
likely base accessible) region of let-7 is a critical feature of
the
RNA in the Argonaute2
binary complex that is likely required for mRNA recruitment.
45

Figure
12:
A.
Iet-7
•
A G
22-
15
10-
5-m
mm
4»
2.Si
8
H
1.5-
I"
0
P
MP
..
A\ ,U G
l 4-i,Gp-t-.Gp—•—+-'i
p
• I
I—I—'—t—I—I
I I I I
u
UpA
G
G UPAP
UPUP
^A^GpUPUoH
5'P
1 2 3 4 S 6 7 8 9 10
11 12 13 14 15 16 17 18 19 20
21
Phosphodiester Bond
46

B.
let-7«
-AG
2.5
41
z
» 2 •
1.5
1 —
I-VA
UPAPUPAPCPAP:A^UPCP_^PCPUP
. + ... ,^_
v T ^^ u
GPUPCPUPUPUP^^
OH
0 P
5'P 1 2 3 4 S 6 7 8 9 10 11 12 13 14 IS 16 17 18 19 20 21
Phosphodiester Bond
47