RNA interference offers researchers a
relatively straightforward tool for investigating biological systems by
selectively reducing expression of mammalian genes. Intelligent siRNA
design, carefully selected expression systems, and optimized transfection
strategies can speed up the process and increase production of valid
results.
| Aug
1, 2005 |
| By:
Subrahmanyam
Yerramilli, Eric
Lader, Dirk
Loeffert, Friederike
Wilmer, Peter
Hahn, Elizabeth
Scanlan |
| Pharmaceutical
Discovery |
|
The application of RNA interference (RNAi) to mammalian cells has
significantly accelerated research in functional genomics and drug
discovery. RNAi allows simple, effective, and specific downregulation of
mammalian gene expression, making it a powerful and accessible technique
with enormous scientific, commercial, and potential therapeutic value.
Although a variety of methods can be used for the study of gene
function, many of these, such as gene knockouts, transgenic animal models,
and antisense RNA, are time-consuming, costly, and not amenable to
high-throughput studies. siRNA-mediated RNAi presents many advantages. It
is faster and less laborious than creating knockouts or transgenic
animals. In addition, design of potent siRNAs is easier than design of
effective antisense oligonucleotides. Studies have found that the
biological activity of siRNA is approximately 100-fold higher than
antisense oligonucleotides, meaning that siRNAs can be effective at much
lower concentrations (1). The robustness of siRNA-mediated RNAi has
resulted in its widespread use in high-throughput drug discovery research.
Critical factors for successful RNAi
Several factors in the RNAi workflow need to be optimized to ensure the
success of a high-throughput experiment. siRNA design can significantly
affect the outcome of the experiment, with suboptimal design leading to
insufficient knockdown and possibly nonspecific effects. siRNA can be
designed using one of many public or proprietary algorithms and the siRNA
sequence chosen can significantly influence siRNA potency (level of
knockdown achieved) and specificity (knockdown of the target gene only).
The choice of cell type should also be carefully considered. The cell
type used should be suitable for efficient, high-throughput siRNA
transfection. In addition, some cell types may be more biologically
responsive than others to the particular RNAi effect under study.
The efficiency of siRNA delivery into the cell must be as high as
possible, as limits in the effectiveness of delivery inevitably lead to a
decrease in knockdown. The transfection protocol should be robust and
straightforward and adaptable to high-throughput formats and automation,
if required.
When setting up the experiment, it is vital to include adequate
positive and negative controls to ensure that the high volume of data can
be correctly interpreted and to account for variation in the different
parameters of the procedure.
A reliable method for confirmation of gene knockdown is necessary to
verify that observed phenotypes correspond with target gene knockdown.
Quantitative, real-time RT-PCR is often chosen for this purpose due to its
ease of use.
The assay used for screening will reflect the ultimate aim of the
experiment, be it the study of genes in a particular pathway or the
identification of genes involved in a disease process. High-throughput
experiments yield a large amount of data, so methods to analyze these data
must be established. Identification of interesting target genes must be
confirmed in follow-up experiments using siRNAs targeted to different
areas of the mRNA to confirm the specificity of the result.
Potent siRNA design
Theoretically, any 21 nucleotide region of mRNA can be used as an siRNA
target sequence. In reality, only one out of every 4 or 5 randomly
selected target sequences will be functional, making this an inefficient
siRNA design strategy (2, 3). Target site selection has a large effect on
siRNA functionality, and it is advisable to use an algorithm for siRNA
design that selects target sites that will result in potent and specific
siRNA.
Regions of repetitive sequences should be avoided, as the siRNA could
potentially cause off-target effects. Many researchers prefer to design
siRNAs specifically to target a region within the open reading frame (ORF)
(4). The rationale behind this is that publicly available sequences for
coding regions are more reliable. Targeting the ORF also allows the option
of targeting an exon unique to a specific splice variant or an exon common
to all splice variants, depending on the purpose of the experiment.
Alternatively, some researchers prefer to target 3' UTR regions (5, 6).
Targeting the 3' UTR can be useful for validation of specificity in
follow-up experiments in which the target gene function is restored by
vector-based expression of a mutated form of the gene.
siRNAs designed taking into account the mechanism of RNAi are far more
likely to be functional and specific (7, 8). Mechanistic rules have been
used as the basis for the development of algorithms for siRNA design.
Figure 1. Twenty kinase genes were
silenced using 2 siRNAs for each target gene. siRNAs were designed
using the HiPerformance siRNA Design Algorithm (developed by
Novartis Pharma incorporating data from a very large study of
siRNA functionality). Knockdown efficiency was determined by
quantitative, real-time RT-PCR.
|
Artificial intelligence, trained by
large databases of actual siRNA performance, can significantly improve new
siRNA sequence selection. The authors use one such system, the
HiPerformance siRNA Design Algorithm developed by Novartis Pharma.
Performance data from 3300 siRNAs for 33 genes were used to train the
algorithm to accurately predict functional siRNA sequences (Figure 1). A
full human genomewide library has been successfully designed using this
algorithm and the knockdown efficiency for over 2000 targets has been
verified by quantitative, real-time RT-PCR (20).
Specific siRNA design
Avoidance of off-target effects is
critical in RNAi experiments, as they can produce misleading results.
Genomewide profiling using microarrays has been used to assess siRNA
specificity and the results showed that expression profiles obtained with
different siRNAs for the same target closely corresponded when using
optimal design and transfection conditions (9, 10). These results
demonstrated that, when optimized, siRNA-mediated RNAi provides specific,
reliable results.
However, it has also been observed
that siRNAs with 3–4 mismatches to an mRNA sequence can act as micro
RNAs (miRNAs), resulting in translational repression (11, 12, 13). This
means that siRNAs with partial homology to an unintended target could
cause nonspecific effects by acting as miRNAs. For this reason, accurate
homology analysis of target sequences is important. Potential target
sequences should be analyzed using highly sensitive algorithms, such as
the Smith-Waterman algorithm (14). The BLAST® algorithm has the advantage
of being much quicker than the Smith-Waterman algorithm. However, BLAST is
not sensitive enough to detect short regions of homology. In one study, it
was predicted that BLAST missed up to 20% of alignments to sequences that
could potentially lead to off-target effects (15). The authors use a
proprietary homology analysis tool similar to the Smith-Waterman algorithm
and an up-to-date, internally curated, nonredundant sequence database, for
siRNA design.
Choosing a model cell system
The cell type chosen for RNAi
experiments should be easy to transfect at high efficiencies and
compatible with the downstream screening assay. The cells should be easy
to handle when working in high-throughput formats. It is advisable to
perform initial RNAi experiments with more than one cell type, as
different cell types may vary in their biological responsiveness to
knockdown and the level of phenotypic effects. Research carried out by
QIAGEN and Affymetrix scientists has shown highly significant differences
between the level of responsiveness of HeLa S3 and MCF-7 cells to
knockdown of genes involved in the cell cycle (16). Following transfection
of CDC2 siRNA, cell cycle progression was analyzed and the genomewide
effect of CDC2 knockdown was assessed using GeneChip® arrays from
Affymetrix. Although quantitative, real-time RT-PCR and western blot
analysis showed that CDC2 was silenced by >80% in both cell lines, the
responses to knockdown were very different. CDC2 knockdown resulted in
accumulation of cells in the G2 phase of the cell cycle, but this effect
was much more pronounced in MCF-7 cells. Statistical analysis of
genomewide expression profiles showed that siRNA transfection had only a
marginal effect on global gene expression levels in HeLa S3 cells. Apart
from CDC2, no genes showed changes in expression at all the time points
tested. In contrast, the expression of 33 genes, in addition to CDC2, was
affected at all time points in MCF-7 cells. Data from MCF-7 cells revealed
interesting details of regulatory networks involving CDC2. Accordingly,
MCF-7 was chosen as the cell type for further analysis due to its
increased biological response to CDC2 knockdown.
siRNA delivery
High levels of gene knockdown are
necessary for downstream analysis. This means that high transfection
efficiency is also necessary. A reduction in transfection efficiency will
reduce knockdown and will also reduce the level of phenotypic effects.
This reduction may make phenotypic effects difficult to detect and reduce
reproducibility. For these reasons, optimal transfection conditions must
be determined to ensure the success of the experiment. A transfection
reagent and protocols that allow effective knockdown using low siRNA
concentrations will lead to more accurate results. Research suggests that
using low siRNA concentrations in RNAi experiments lowers the risk of
nonspecific effects (10, 17). In addition, using less siRNA in each
experiment reduces costs.
Figure 2. MCF-7 cells were
transfected with a range of concentrations of siRNA targeted
against lamin A/C using HiPerFect Transfection Reagent. siRNA was
transfected immediately or spotted in plate wells and stored at
room temperature (RT) for 48 hours prior to transfection. Gene
silencing was assessed by quantitative, real-time RT-PCR.
|
Traditionally, transfection procedures
involve seeding cells the day before transfection. On the day of
transfection, siRNAs are mixed with reagent for complex formation and then
complexes are added to cells. However, reverse transfection, where cells
are seeded and transfected on the same day, has become more widely used
for high-throughput experiments (21). In reverse transfection, siRNAs are
spotted into plates or onto glass slides. Next, transfection reagent is
added and complexes are formed. Finally, cells are added to the siRNA/reagent
complexes. Prespotted siRNAs can be stored prior to transfection, allowing
more flexibility in the experimental procedure (Figure 2).
Control experiments
Without appropriate control
experiments, data cannot be properly analyzed and results will be
unreliable. Positive and negative (nonsilencing) control siRNAs should be
transfected in each experiment. Positive control siRNAs are siRNAs that
are known to provide high knockdown of a target gene that produces the
desired phenotype. Routine transfection of positive control siRNAs shows
that transfection and assay conditions remain optimal.
Small molecules or bioactive
compounds that produce the phenotype under study can also be used as
positive controls for assay conditions. For example, an apoptosis-inducing
drug could be used as a positive control for an apoptosis screening assay,
or an inhibitory compound could be used to inhibit upstream components of
the pathway under study, causing translocation of the protein examined in
the screening assay.
Nonsilencing control siRNAs can be
siRNAs with no homology to any known mammalian gene or siRNAs that are
homologous to a gene that is not present in the cells under study (e.g.,
green fluorescent protein). Data from transfection of nonsilencing siRNAs
can be used to analyze the extent of nonspecific effects that may have
occurred as a result of siRNA transfection. Untransfected cells should
also be analyzed as a negative control. In addition, replicate experiments
should be performed to ensure reproducibility of results and to allow for
small variations in the experimental procedure. Phenotypic effects
observed after knockdown must always be confirmed by one or more
additional siRNAs targeted to different areas of the mRNA.
Confirmation of gene knockdown
Gene knockdown can be validated by
various methods. The most widely used methods are quantitative, real-time
RT-PCR analysis of knockdown at the mRNA level and western blot analysis
of knockdown at the protein level. Western blot analysis is the most
comprehensive way of showing that expression of the target gene has been
downregulated. However, it is restricted in its application to
high-throughput analysis because antibodies for the protein of interest
are not always available. In contrast, quantitative, real-time RT-PCR is
routinely used and is easily adaptable to high-throughput studies.
Figure 3. Real-time, two-step RT-PCR
analysis of β-actin was carried out using samples prepared
with the QuantiTect Reverse Transcription Kit. Samples were
prepared from 100 ng total RNA with genomic DNA removal and
reverse transcription (+RT) or with genomic DNA removal and
without reverse transcription (–RT). Quantitative, real-time PCR
was performed in duplicate. The β-actin-specific primers
could detect both mRNA and genomic DNA sequences. Control
reactions with no template were also performed (green). The red,
flat –RT plot indicates efficient removal of residual genomic
DNA.
|
High-quality template RNA is essential
for accurate real-time RT-PCR analysis. It is important that any residual
contaminating genomic DNA in the RNA sample is not amplified and detected,
otherwise knockdown efficacy will be underestimated. Genomic DNA
contamination can be eliminated by DNase digestion during or after the RNA
purification procedure (Figure 3). Alternatively, primers can be designed
to span exon–exon boundaries to ensure detection of RNA only. This is
not always possible, however. For example, genes that consist of only a
single exon or the existence of pseudogenes with identical or
near-identical sequence to the target cDNA may present problems. In
addition, genomic DNA can be detected by primers for housekeeping genes (e.g.,
GAPDH, 18S rRNA, or β-actin), which are often used to normalize the
expression level of a gene to the RNA content of the sample. For these
reasons, a genomic DNA removal step should be incorporated after RNA
preparation.
cDNA synthesis by reverse
transcription is the first step in two-step RT-PCR and is followed by PCR
with gene-specific primers. In two-step RT-PCR, cDNA synthesis is carried
out with nonspecific primers (oligo-dT and/or random primers), which
allows the same cDNA sample to be used to confirm knockdown of the gene of
interest and to analyze expression levels of several other genes.
Alternatively, one-step RT-PCR, in
which reverse transcription and PCR steps are carried out in a single
tube, uses gene-specific primers for both the reverse transcription and
PCR steps. This method may be chosen if many gene knockdowns need to be
confirmed in parallel because reaction setup is easier and more convenient
for high-throughput analysis.
In real-time RT-PCR, gene-specific
primers do not necessarily have to flank the siRNA binding site, since
siRNA hybridization to its mRNA target results in degradation of the
entire mRNA transcript that contains the siRNA binding site. This means
that primers can be located anywhere on the mRNA and the same primers can
be used for analysis of knockdown using multiple siRNAs designed to target
different areas of the mRNA.
Real-time RT-PCR analysis can be
performed using gene-specific primer pairs with detection using SYBR®
Green, which detects double-stranded DNA. Commercially available primer
pairs can provide specific and sensitive results and are an economical
solution for high-throughput work.
The combination of transfection of
siRNA specific for a gene target, validation of knockdown by real-time
RT-PCR, and a phenotypic change observed in a screening assay provides
strong evidence of a role for the gene in the pathway or process under
study. Validation of knockdown is essential to confirm the connection
between knockdown and phenotype.
Screening assays
The assays used for RNAi screens vary
depending on the purpose of the experiment. Experiments can range from
looking at a small group of gene targets for pathway analysis to screening
the whole human genome for drug discovery. Assays may be carried out
manually or can be partially or completely automated. The assay can range
from a simple homogeneous cell-based assay (18) to an assay which looks at
changes in the subcellular distribution of a protein using a high-content,
automated, confocal microscope (19). One of the great advantages of this
type of research is that the assays do not have to be highly complex to
yield valuable information about cellular pathways and responses.
Combining optimized parameters
accelerates drug discovery
Once the parameters described have
been optimized, integration of the different stages of the workflow will
result in successful RNAi screenings and faster research and discovery.
Highly informative RNAi screens are used for cancer research at the
Translational Genomics Research Institute (TGen) in Maryland, USA (www.tgen.org).
Screening at TGen involves siRNA-mediated knockdown of approximately 5000
genes in various cancer cell lines. Cell growth and survival after
knockdown are examined to identify etiologically relevant genes. Isogenic
cell lines, which vary only in the expression of a tumor suppressor gene,
are used to find synthetic lethal knockdowns that kill specific tumor
cells. In addition, the combination of RNAi screens with exposure to
anticancer drugs allows the researchers to pinpoint genes whose knockdown
enhances or suppresses responses to the drugs. These RNAi screens provide
a greater understanding of cancer development and drug action. They also
identify potential targets of novel drugs for patients with defined
genetic alterations in their tumors and targets for combination therapy to
improve the response to existing cancer drugs.
The future of RNAi in drug discovery
Huge advances have occurred in the
design, synthesis, and purification of siRNA over the last few years.
High-throughput analysis of thousands of gene targets using RNAi is now
possible and is amenable to all researchers, as it can be carried out at
different levels of throughput and with or without automation. The
advantages of efficient, economical knockdown offered by RNAi and the
large amount of data it provides will ensure that it remains a technology
of choice for functional genomics and drug discovery research.
Subrahmanyam Yerramilli and Eric
Lader are with QIAGEN Sciences, Germantown, MD, USA. Dirk Loeffert,
Friederike Wilmer, Peter Hahn, and Elizabeth Scanlan
are with QIAGEN GmbH, Hilden, Germany. Eric Lader can be reached at
QIAGEN Sciences, 19300 Germantown Rd., Germantown, MD 20874, USA. E-mail Eric.Lader@qiagen.com
References
1. R. Kretschmer-Kazemi Far and G.
Sczakiel, Nucleic Acids Res. 31, 4417-4424 (2003).
2. T. Holen, M. Amarzguioui, M.T.
Wiiger, E. Babaie, and H. Prydz, Nucleic Acids Res. 30,
1757-1766 (2002).
3. R. Kumar, D.S. Conklin, and V.
Mittal, Genome Res. 13, 2333-2340 (2003).
4. S.M.
Elbashir, J. Harborth, K.
Weber, and T. Tuschl, Methods 26, 199-213 (2002).
5. D.M.
Dykxhoorn, C.D. Novina, and
P.A. Sharp, Nat. Rev. Mol. Cell Biol. 4, 457-467 (2003).
6. S.M.
Elbashir, J. Harborth, W.
Lendeckel, A. Yalcin, K. Weber, and T. Tuschl, Nature 411,
494-498 (2001).
7. A. Khvorova, A. Reynolds, and S.D.
Jayasena, Cell 115, 209-216 (2003).
8. D.S. Schwarz, G. Hutvanger, T. Du,
Z. Xu, N. Aronin, and P.D. Zamore, Cell 115, 199-208 (2003).
9. J-T. Chi, H.Y. Chang, N.N. Wang,
D.S. Chang, N. Dunphy, and P.O. Brown, Proc. Natl. Acad. Sci. USA 100,
6343-6346 (2003).
10. D. Semizarov, L. Frost, A. Sarthy,
P. Kroeger, D.N. Halbert, and S.W. Fesik, Proc. Natl. Acad. Sci. USA
100, 6347-6352 (2003).
11. J.G.
Doench, C.P. Petersen, and
P.A. Sharp, Genes Dev. 17, 438-442 (2003).
12. S. Saxena, Z.O. Jonsson, and A.
Dutta, J. Biol. Chem. 278, 44312-44319 (2003).
13. Y. Zeng, R. Yi, and B.R. Cullen, Proc.
Natl Acad. Sci. USA 100, 9779-9784 (2003).
14. O. Snove Jr, M. Nedland, S.H.
Fjeldstad, H. Humberset, O.R. Birkeland, T. Grunfeld, and P. Saetrom, Biochem.
Biophys. Res. Commun. 325, 769-773 (2004).
15. O. Snove Jr and T. Holen, Biophys.
Res. Commun. 319, 256-263 (2004).
16. P. Hahn, T.A. Awad, F. Wilmer, Y.
Turpaz, A. Grewe, and W. Bielke QIAGEN News 2005 e8 (2005) (www.qiagen.com/literature/qiagennews/online_archive.aspx).
17. S.P.
Persengiev, X. Zhu, and M.R.
Green, RNA 10, 12-18 (2004).
18. J.P. MacKeigan, L.O. Murphy, and
J. Blenis, Nat. Cell Biol. 7, 591-600 (2005).
19. M. Bertelsen and A. Sanfridson, Assay
Drug Dev. Technol. 3, 261-71 (2005).
20. U. Krueger and S.Yerramilli,
QIAGEN unpublished results
21. A reverse transfection protocol
can be found at www.qiagen.com/goto/HiPerFect.
Disclaimers
Trademarks:
QIAGEN®, QuantiTect® (QIAGEN
Group); Affymetrix®, GeneChip® (Affymetrix, Inc.); SYBR® (Molecular
Probes, Inc.), BLAST® (US National Library of Medicine).
siRNA technology licensed to QIAGEN
is covered by various patent applications, owned by the Massachusetts
Institute of Technology, Cambridge, MA, USA and others.
The PCR process is covered by the
foreign counterparts of U.S. Patents Nos. 4,683,202 and 4,683,195 owned by
F. Hoffmann-La Roche Ltd.
QuantiTect Primer Assays are
optimized for use in the polymerase chain reaction (PCR) process covered
by patents outside the U.S. owned by F. Hoffmann-La Roche Ltd. No license
under these patents to use the PCR process is conveyed expressly or by
implication to the purchaser by the purchase of this product. Where the
PCR process is covered by patents, a license to use PCR for certain
research and development activities accompanies the purchase of certain
reagents from licensed suppliers such as QIAGEN when used in conjunction
with an authorized thermal cycler, or is available from Applied Biosystems.
Further information on purchasing licenses to practice the PCR process
where the process is covered by patents may be obtained by contacting the
Director of Licensing, Applied Biosystems, 850 Lincoln Centre Drive,
Foster City, California 94404 or the Licensing Department, Roche Molecular
Systems, Inc., 1145 Atlantic Avenue, Alameda, California 94501.
|
