Download Design considerations for highly specific and efficient

GE Healthcare
Design considerations for highly specific and efficient synthetic crRNA molecules
Anja van Brabant Smith, Emily M. Anderson, Shawn McClelland, Elena Maksimova, Tyler Reed, Steve Lenger, Žaklina Strezoska, Hidevaldo Machado
Dharmacon, part of GE Healthcare, 2650 Crescent Drive, Suite #100, Lafayette, CO 80026, US
Introduction
Off-target analysis should include gaps
In order to understand the parameters affecting CRISPR-Cas9 gene editing efficiency, we systematically transfected synthetic
CRISPR RNA (crRNA) and trans-activating CRISPR RNA (tracrRNA) reagents targeting components of the proteasome into
a reporter cell line in which knockout of proteasome function results in fluorescence of a ubiquitin-EGFP fusion protein
that is normally degraded by the proteasome pathway. We evaluated the functionality of > 1100 crRNA sequences in this
system; using these data, we developed and trained an algorithm to score crRNAs based on how likely they are to produce
functional knockout of targeted genes. We further tested our algorithm by designing synthetic crRNAs to genes unrelated to
the proteasome and examined their ability to knock out gene function using additional phenotypic assays. To augment our
functionality algorithm, we developed an optimized alignment program to perform rapid, flexible, and complete specificity
analysis of crRNAs, including detection of gapped alignments. We have combined this comprehensive specificity check with
our functionality algorithm to select and score highly specific and functional crRNAs for any given gene target.
DMSO-treated cells
B
MG132-treated cells
Transfect
synthetic crRNA:tracrRNA
Cas9 expression
plasmid
C
Cas9:crRNA:tracrRNA-treated
PPIB negative control PSMD7
siRNA-treated
PSMD14
Figure 5. Potential off-target sites in the genome
for any given crRNA include not just mismatches
but gaps as well. Gaps can exist in the crRNA
strand or in the DNA target strand. Many
commonly used web-based crRNA specificity
tools do not fully account for gaps when
performing alignments.
Gap in DNA
strand
NNNNNN N
NN
NNNNmNNNNNNN-NNNNNNN
|||| ||| ||| |||||||
NNNNNNNN-NNNNNNNNNNN
N
NN
NNNNNNNN
DNA
RNA
Gap in RNA
strand
Complete off-target analysis is important for
crRNA specificity
Functional assay for crRNAs
A
Flaws
m = mismatch
- = gap
VCP
VCP siGENOME pool
CRISPR target ‘A’
Figure 1. Recombinant U2OS cell line stably
expressing a mutant human ubiquitin fused
to EGFP. A Gly76Val mutation results in an
uncleavable ubiquitin moiety fused to EGFP,
which results in constitutive degradation
of the protein and little detectable EGFP
fluorescence. (A) Inhibition of proteasome
activity with chemical treatment (MG132)
results in EGFP fluorescence. (B) Schematic
of a CRISPR-Cas9 experiment using synthetic
crRNA and tracrRNA reagents. (C) Inhibition
of proteasome activity with CRISPR-Cas9 or
siRNAs targeting proteasome components
results in EGFP fluorescence.
P A M
CRISPR ‘A’
incorrectly rated
as GOOD Specificity
CRISPR ‘A’
correctly rated
as POOR Specificity
Incomplete alignment to the genome
Intended target
ALL alignments to the genome
P A M
+++
P A M
+
P A M
P A M
+
P A M
P A M
---
P A M
-
P A M
+
P A M
++
P A M
++
P A M
6
5
P A M
Intended target
Controls
Exon 1
4
Exon 2
mismatch in alignment
gap in alignment
Exon 3
3
Exon 4
Exon 5
2
Exon 6
Exon 7
1
0
0
6
6
5
5
100
200
300
400
500
600
700
800
Likelihood
of cleavage
crRNA functionality is position- and sequence-dependent
Controls
Exon 1
4
4
Exon 8
Exon 2
Exon 3
3
Exon 4
Exon 5
2
Exon 6
Exon 7
1
0
GFP signal, fold over negative control
0
6
5
100
200
300
400
500
600
700
Figure 6. Schematic demonstrating the effect of incomplete alignment during off-target analysis for a crRNA.
Exon 9
3
Exon 10
Exon 11
2
Exon 12
1
crRNA:
-
+
-
+
-
+
-
+
0
800
800
6
6
5
5
900
1000
1100
1200
1300
1400
1500
Target/Off-target
Sequence
PAM
Intended target
GGTCATCTGGGAGAAAAGCG
TGG
OT1
GGTCCTCTGGGAGAAAAGACG
CAG
OT2
GGT-ATCTGGGAGAAAAGCA
TGG
OT3
GGTC-TCTGGGAGAAAAG-G
AAG
Controls
Exon 1
4
4
Exon 8
Exon 2
Exon 3
3
Exon 9
3
Exon 10
Exon 4
Exon 5
2
Exon 6
Exon 7
1
0
100
200
300
400
500
600
700
Exon 13
Exon 14
3
Exon 15
Exon 11
2
Exon 12
1
800
800
6
6
5
5
Exon 16
2
Exon 17
1
0
0
4
900
1000
1100
1200
1300
1400
0
1500
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
Exon 8
Exon 9
3
Exon 10
Exon 11
2
Exon 12
4
Exon 13
Exon 14
3
Exon 15
Exon 16
2
Exon 17
1
1
0
1500
0
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
37
Target
cut site distance from CDS start in transcript (nts)
Figure 2. Ubiquitin[G76V]-EGFP U2OS cells stably expressing Cas9 were transfected with 266 synthetic crRNA:tracrRNA
complexes targeting the coding region of the VCP gene. EGFP fluorescence was measured 72 hours post- transfection;
an increase in EGFP fluorescence indicates functional knockout of the VCP gene resulting in disruption of proteasome
function. crRNAs in different exons are indicated by the different colors. The data indicate that crRNAs vary in their ability
to cause functional gene disruption.
4
Indel%
2500
6
0
16
OT1
0
OT2
OT3
Figure 7. Ubi[G76V]-EGFP U2OS cells stably expressing Cas9 were transfected with 25 nM synthetic crRNA:tracrRNA and
genomic DNA was isolated 72 hours after transfection. Potential off-target (OT) sites containing mismatches as well as
gaps in the RNA or the DNA strand were analyzed for off-target cleavage with a mismatch detection assay (using T7EI
endonuclease). Red nucleotides indicate mismatches, red dashes indicate gaps, and underlined red nucleotides indicate
insertions relative to the target DNA sequence. The off-targets were identified using the Dharmacon specificity tool and are
not identified by other common web-based crRNA specificity tools.
5
4
Exon 13
Exon 14
3
Exon 15
Exon 16
The optimal crRNAs balance functionality and
specificity
Machine learning for algorithm development
2
Exon 17
1
0
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
Figure 3. A training set consisting of 10
genes and 1115 crRNA target sites was
used to train our functionality algorithm.
Features examined include nucleotide
composition, nearest neighbor effects, PAM
sequence, position in exons, and distance
from the start codon. Receiver Operating
Characteristic (ROC) shows good fit of
training set data. The ROC measures the
area under the curve of True Positive Rate vs
False Positive Rate. The ROC for our test set
data is 0.78.
Gene
Exon 1
Exon 2
Exon 3
Exon 4
CDS region of exon
Exon 5
Transcript A
Exon 1
Exon 2
Exon 3
Transcript B
CRISPR ‘A’
CRISPR ‘B’
Specificity Rating
Functionality Rating
CRISPR targets
POOR
AVERAGE
GOOD
CRISPR target ‘A’ – May function, but poor specificity
CRISPR target ‘B’ – OPTIMAL PICK, scores well in both functionality & specificity
Functionality of algorithm-designed crRNAs in other
assays
Normalized ApoONE assay
BCL2L1
PLK1
Conclusions
• crRNAs vary in specificity and in efficiency for creating functional gene knockouts
WEE1
3
3
3
2.5
2.5
2.5
2
2
2
1.5
1.5
1.5
1
1
1
0.5
0.5
0.5
Figure 8. Schematic of how
any given crRNA may differ
with regard to its specificity
and functionality. Our
algorithm balances these two
attributes to pick the crRNAs
predicted to have the highest
specificity and functionality.
• We used a high-throughput fluorescence assay to develop and train an algorithm to score crRNAs for likelihood of
producing functional gene knockouts
• We developed an optimized alignment program to perform rapid and complete specificity analysis of crRNAs
0
0
H1
H2
• We demonstrate that targets with gaps in either the RNA or DNA strand can be cleaved and are therefore important to
identify during specificity checking
0
H1
H2
H1
H2
crRNAs
Figure 4. Box plot representation of the functionality of crRNAs targeting BCL2L1, PLK1 or WEE1 as determined by the
ApoONE homogeneous assay (Promega) 48 hours after transfection. For the box plots, crRNAs were divided into bottom
half (H1) and top half (H2) based on their functionality score. The medians as well as the distribution of data between the
lower and upper quartile demonstrate that high-scoring crRNAs have increased functionality.
gelifesciences.com/dharmacon
GE, imagination at work and GE monogram are trademarks of General Electric Company. Dharmacon is a trademark of GE Healthcare companies.
All other trademarks are the property of General Electric Company or one of its subsidiaries. ©2015 General Electric Company—All rights reserved.
Version published July 2015. GE Healthcare UK Limited, Amersham Place, Little Chalfont, Buckinghamshire, HP7 9NA, UK