Transcript A tale of mutated tails and overhangs 1

A tale of mutated tails and
overhangs
- rooting out possible sources of experimental error in RNAi studies
MNBTS 9000 - Oslo - October 22nd 2009
Torgeir Holen, CMBN and Institute for
Basal Medical Sciences (IMB), University of Oslo
1
02/06/2016
The aims of this lecture on
scientific logic and irrational
twists and turns



Scientific papers tell nice stories... but
the narrative is misleading...
"Is the Scientific paper a fraud?" Peter
Medewar (Nobel laureate 1960 on
tolerance to transplanted tissues)
A case study in confused thinking in the
middle of a confused study...
2
02/06/2016
3
02/06/2016
The usefulness of siRNA
- a short summary of the pitfalls
1.
siRNA exhibit position effects – some target positions superior to others
2.
siRNA activity fades out – in our case 3-4 days after transfection
3.
siRNA activity can be blocked by competition with less active siRNA
4.
Some siRNA have great tolerance for chemical modifications
5.
Some siRNA tolerate mutations
6.
1.
The common theme: if siRNAs tolerate mismatches in relation to the
mRNA target, then similar, random targets in other mRNA are at risk
2.
Early studies: 3’ overhang mismatch tolerance & central mutations
3.
Screening disruptive G:C mutations: some positions more vulnerable than
others...
4.
What are the general consequences for the thousands of siRNA now in
common use? Can off-target hits be avoided? Some conclusions from a
bioinformatic study… some musings on microRNA
Some recent data from three new studies on mutations
4
02/06/2016
Strength of repetition in different assays position effect consistent also in Northern
assays, protein assays and coagulation
assays
PSK739i
PSK566i
PSK546i
PSK314i
hTF562i
hTF478i
hTF372i
hTF167i
mock
- coagulation slowed by siRNA: an example of use of siRNA
to study phenotypic effect Inhibition of TF mRNA, procoagulant activity
Protein &andCoagulation
antigen by siRNA
TF
GAPDH
Northerns
Normalised TF expression
120
100
80
mRNA
procoag
antigen
60
40
20
0
mock
hTF167i
hTF372i
PSK314i
Legend: mRNA (filled bars), procoagulant activity
(dotted bars) and TF protein (hatched bars)
5
02/06/2016
A tale of tails and overhangs
- rooting out possible sources of experimental error



We discovered inactive siRNAs, but is this true?
Could it be that the Tuschl-signature (TT) was
actually deleterious in our assays?
The 3’ overhangs were not base-pairing (BO)
- animal studies very expensive: thorough basic research
is absolutely necessary before these are performed...
--> ribozyme-studies in mice...
hTF167i
5'-GCGCUUCAGGCACUACAAATT
TTCGCGAAGUCCGUGAUGUUU-5'
6
02/06/2016


Error source # 1: could
basepairing be vital?
But no: BO-versions
were just as active or
inactive, respectively, as
the TT-versions
Holen et al, Nucleic Acids Research, 2002
7
02/06/2016
Error source #2: Synthesis
failure...?




Could there be synthesis problems?
No. Yet other versions - after repeated
series of synthesis - also with RNAoverhangs (RO) - all had the same
pattern of activity
Conclusion must be that the inactive
siRNA were not caused by synthesis or
design failures...
...and also, that overhang mismatches
are tolerated
8
02/06/2016
The ribozyme failure...


My doctorate should have been on
ribozymes...
...unfortunately they don't work against
endogenous mRNA...
Lyngstadaas SP, Risnes S, Sproat BS, Thrane PS, Prydz
HP. A synthetic, chemically modified ribozyme eliminates
amelogenin, the major translation product in developing
mouse enamel in vivo.
EMBO J. 1995 Nov 1;14(21):5224-9.
9
02/06/2016
The antisense field has been
plagued with artifacts for 20 years



C. A. Stein, among others, has estimated
that thousands of scientific articles on
various antisense agents report better results
that they possibly could have obtained if
compared with larger studies (C.A. Stein,
Nature Biotechnology, 1999; Tu et al, JBC,
1998)
We decided to try inactive our most active
candidates by mutations
This in order to rule out sequenceindependent, unspecific silencing
10
02/06/2016
Control for specificity: A central
C:G-mutation

The M1-siRNA has a central
mutation in position 10
(relative to 5' end of sense
strand), still it loses only a
few percentage points of
activity, and retains over 80
% silencing
11
02/06/2016
Two central C:G mutations


The M2-siRNA has two central C:G
mutations, but still retains apprx.
70 % silencing
Note also that overhangs do not
basepair either, so this is really a
quadruple mismatch...
12
02/06/2016
A study mutating all G:C pairs show
tolerance in 5’end, but less so in 3’ end
**** * ** * *
5'-GCGCUUCAGGCACUACAAAUA
GCCGCGAAGUCCGUGAUGUUU-5'
Normalized
TF/GAPDH mRNA
mRNA
TF/GAPDH
Normalised
120
100
80
60
40
20
Amarzguioui, Holen et al, Nucleic
Acids Research, 2003
ds
7/
1
ds 0
10
/1
ds 1
10
/1
ds 3
10
/1
6
s1
6
s1
3
s1
1
s1
0
s7
s4
s3
s2
s1
t
w
m
oc
k
0
13
02/06/2016
Is the mismatch problem real?
- the spread of high capacity siRNAs...



Several other labs had reported that single mutations
inactivate their siRNA... Boutla et al, in Drosophila
(Curr. Biol, 2001), Yu et al, PNAS, 2002, Pusch et al,
Nucleic Acids Research, 2003, Jacque et al, Nature,
2002, Zeng & Cullen, RNA, 2003, Vickers et al, JBC,
2003, Saxena et al, JBC, 2003
Our candidates were the best in activity screens...
As more people start using high-activity, customdesigned siRNA, in research or in clinical studies, the
off-target problems might increase...
14
02/06/2016
What is the specificity relevance
of these experiments...?
Quantitative considerations



The siRNA that enters the cells, and is
incorporated into the RISC complex, has
to scan >10 000 different mRNAs, each
with an average of 2000 positions...
This means that the specificity must be
in the ball park of 1 : 20 000 000...
Surely some similar matches are in
danger of off-target silencing activity...
15
02/06/2016
An in silico risk assessment study


359 siRNAs from different studies were
collected and assessed against transcriptome
It was found that in 75 % of cases there
were matches with targets with three or less
mismatches
Snove & Holen, BBRC, 2004
16
02/06/2016
Many published
siRNAs have
similarity to
targets in other
mRNAs
Snove & Holen, BBRC, 2004
17
02/06/2016
An example of the dangers of
siRNA-use in functional
genomics:
An siRNA against laminB2 induced
apoptosis, leading to the conclusion
“…the two other lamins, B1 and B2,
are now identified as essential
proteins” (Harborth...Tuschl et al,
JCS, 2001)
But many possible off-target hits
exist... in addition to these doublemismatch hits, there are 63 triplemismatch hits... all of which might be
active silencers and might be the real
cause of the apoptotic reaction...
Other genes possibly
targeted by the
laminB2 siRNA
CLSTN2
5'-AAGAGGAGGAAGAAGCCGAGG-3'
|||||||||| |||||||||
3'-UUCUCCUCCUCCUUCGGCUCA-5'
HS6ST3
5'-AAGAGGAGGAGGAAGACGAGC-3'
||||||||||||||| ||||
3'-UUCUCCUCCUCCUUCGGCUCA-5'
NM_015897
5'-AAGAGGAGGAGGAAGACGAGG-3'
||||||||||||||| ||||
3'-UUCUCCUCCUCCUUCGGCUCA-5'
ATBF1
5'-AAGAGGAGGAGGAAGACGAGG-3'
||||||||||||||| ||||
3'-UUCUCCUCCUCCUUCGGCUCA-5'
SPTB
5'-AAGAGGAGGAGGAAACAGAGU-3'
|||||||||||||| | ||||
3'-UUCUCCUCCUCCUUCGGCUCA-5'
18
02/06/2016
With careful in silico screening sufficiently
unique candidates can be found in most genes

Using special search hardware (from Interagon Inc.), all possible
siRNA from the current transcriptome (58 million) were screened
against all other...
Snove & Holen, BBRC, 2004
19
02/06/2016
Other papers on off-target problem
using micro-arrays







Jackson et al, Nature Biotechnology, 2003 and Persengiev et al, RNA,
2004 find large, non-specific effects. Many genes affected by siRNA
transfection.
Thus siRNA has gone from "exquisitely specific" (Tuschl, 2001) to
generally be seen as wildly unspecific...
However, Semizarov et al, PNAS, 2003 and Chi et al, PNAS, 2003 find
siRNA to be very specific
Microarray studies are notoriously hard to perform. Who to believe...?
Who believes...? Jackson et al cited 203 times, Chi et al 90 times...
However Chi....Brown is the Brown who invented the microarray...
QUANTITATIVE MONITORING OF GENE-EXPRESSION PATTERNS
WITH A COMPLEMENTARY-DNA MICROARRAY SCHENA M, SHALON D,
DAVIS RW, BROWN PO SCIENCE 270 (5235): 467-470 OCT 20 1995 Cited
References: 18
Times Cited: 2439
20
02/06/2016
siRNA off-target effects might
be due to microRNA systems




briefly, microRNA produced by cells themselves
affect hundreds of targets each, but only by
weak action
hundreds of micoRNA means that >30 % of all
genes are influenced: 1000'ands of proteins
mechanism still unclear: see recent papers by
Selbach et al, Nature, 2008 and Baek et al,
Nature, 2008
21
02/06/2016
Intro to microRNA: David
Bartel, very popular study

The most cited miRNA review: Bartel DP
microRNAs: Genomics, biogenesis,
mechanism, and function CELL 116 (2): 281297 JAN 23 2004 Times Cited: 1157
22
02/06/2016
Part III: The possible link between RNA
and DNA silencing in higher organisms
1)
2)
3)
RNA silencing, more than siRNA
DNA & chromatin regulation: still unsolved...
Three cases of RNA-DNA silencing link
a)
b)
c)
d)
e)
dsRNA-induced DNA methylation in plants
transgene DNA silencing in Drosophila
centromere silencing in budding yeast (S. pombe)
X-chromatin silencing in humans...
recently: chromatin silencing induced by short RNA in
human cells
23
02/06/2016
More small RNA exist
• stRNA (short temporal RNA, also called micro-RNA,
miRNA) inhibit mRNA translation
• microRNA comes from microGenes, transcripted
as ~70 nt hairpins, and processed (as is shRNA) to
short ~21-mer RNAs...
• heterochromatic siRNA might affect centromeres in
budding yeast (S.pombe) (Reinhart, 2002; Volpe,
2002)
24
02/06/2016
Double-stranded RNA
cause silencing in plants

PTSVd viroids (Wassenegger, Cell, 1994)
• PTSVd can lead to methylation of a
target down to 30 bp of DNA (Pelissier
& Wassenegger, RNA, 2000)
HOW???
25
02/06/2016
Lehninger,
1993
X-chromosome inactivation centre (Xic)

Xic (450 kb region) can be transferred to chromosome 12, resulting in
gene silencing, hypoacetylation, histone methylation (His3mLys9),
delayed DNA replacation and RNA coating...
(Lee & Jaenisch, Nature, 1997)
Lee, Cell, 2000
26Lewin, 200002/06/2016
Summary: What is the
mechanism?
Is the proposed mechanism
right?
...and does it exist in
higher organisms?
Allshire, Science, 2002
Nobody knows... yet
27
02/06/2016
Two papers find RNA
chromatin modifications in
mammalian cells



Morris KV, Chan SW, Jacobsen SE, Looney DJ.
Small interfering RNA-induced transcriptional
gene silencing in human cells. Science. 2004 Aug
27;305(5688):1289-92. Epub 2004 Aug 05.
Kawasaki & Taira. Induction of DNA methylation
and gene silencing by short interfering RNAs in
human cells. Nature. 2004 Sep 9;431(7005):2117. Epub 2004 Aug 15.
The Kawasaki-paper was withdrawn due to
probable scientific fraud... Taira lost his position…
28
02/06/2016
Summary





Science in action (as opposed to
storytelling fairy tales) is confusing and
difficult
Controls and repetition in different
systems are crucial
Possible sources of error must be sought
out
Sometimes such searching for errors end
up in finding new surprising discoveries
Plans are useless, planning is essential",
General Dwight D. Eisenhower
29
02/06/2016
Acknowledgements




Mohammed Amarzguioui (now siRNAsense A/S)
Gunnar Nicolaysen and Prof. Ole Petter Ottersen,
IMB & CMBN
The Norwegian Cancer Society
FUGE, Norwegian Research Council
30
02/06/2016