single cell analysis of lentiviral integration to support ... · • single cell isolation and...
TRANSCRIPT
Cell and Gene Therapy Catapult
12th Floor, Tower Wing, Guy’s Hospital, Great Maze Pond, London SE1 9RT
+44 (0) 203 728 9500 | [email protected] | ct.catapult.org.uk
Cell and Gene Therapy Catapult is a trading name of Cell Therapy Catapult Limited, registered in England and Wales under company number 07964711.
PROPOSED SOLUTION
We developed a novel analytical workflow to analyse vector copy number at a single cell level and
demonstrated that this is representative of the population of transduced cells. The readout of this assay also
provides label-free, PCR-based transduction efficiency, overcoming the challenge of measuring this
specification when the standard flow cytometry approach is not applicable.
METHOD
• Lentiviral vector expressing ZsGreen used to transduce CD3+ T cells purified from healthy donors.
• Single cell isolation and processing performed on the Fluidigm C1 system.
• Droplet digital PCR (ddPCR) used as readout.
• An in-house analysis framework based on Bayesian statistics to convert measured data to integer copy
number predictions.
Conclusions
5. scVCN assay optimisation for gene-modified
cell therapy products
4. Accuracy of the scVCN assay is verified
on a larger sample set3. Development of the single cell VCN (scVCN) workflow
1. Multiplex ddPCR can be used on gDNA to measure
population Vector Copy Number (pVCN)
Single cell analysis of lentiviral integration to support ex-vivo gene modified cell therapy developmentIlaria Santeramo, Marta Bagnati, Enas Hassan, Beata Surmacz-Cordle, Damian Marshall, Vincenzo Di Cerbo*
Cell and Gene Therapy Catapult, London – *presenting and corresponding contact: [email protected]
0 1 2 3 4 5 6 7 8
Vector copies
Dis
trib
ution
0 5000 10000150002000025000
RG2
RG1
VG3
VG2
VG1
Non-amplified gDNA
Absolute number of copies
Targ
et gene
0 50000 100000 150000
RG2
RG1
VG3
VG2
VG1
Targeted pre-amplification
Absolute number of copies
Targ
et gene
0 20000 40000 60000
RG2
RG1
VG3
VG2
VG1
Whole Genome Amplification
Absolute number of copies
Targ
et gene
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
0
2
4
6
8
10
Single Cells
Mean s
ingle
cell
VC
N
95% confidence interval
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
0
2
4
6
8
10
Single Cells
Mean s
ingle
cell
VC
N
95% confidence interval
Vector copies
40%
18%
5%27%
9%
Mean of scVCN predictions= 2.35Bulk gDNA VCN/cell= 2.93 ± 0.12
N= 77 cells
1234567
Fluidigm C1 system
Single cellisolation
gDNAextraction
Targetspre-amplification
ddPCRanalysis
In-housestatistical
model
Quality control of cells
isolation and number
Single cell isolation in a
closed automated system
Automated droplet
digital PCR analysis
Absolute quantification of
viral and human targets
0 20 40 60 80 100
Flow cytometry
Imaging
PCR-based
Number of cells (%)
Non-transducedTransduced
0 1 2 3 4 5 6 7 8 9
0
10
20
30
Vector copies
Num
ber
of cells Chip2
Chip1
Human Reference
Gene 1Vector Gene 1
Vector Gene 2
Vector Gene 3
Multiplex TaqMan ddPCR reactions
Human Reference
Gene 2
Ch1 A
mplit
ude (
VG
)
5000
0
5000
10000
RG1+/VG1-
RG1+/VG1+RG1-/VG1+
RG1-/VG1-
Ch2 Amplitude (RG)
9000
Exemplar 2D ddPCR amplitude plot
pVCN1
pVCN2
pVCN3
pVCN4
pVCN5
pVCN6
0
1
2
3
Popula
tion V
CN
(bulk
gD
NA
)
MOI 1MOI 0.3
pVCN measured on two samples transduced
at differing MOI levels
• All six combinations of multiplex ddPCR reactions
provide consistent vector copy number results.
• pVCN analysis by ddPCR is precise and robust on
gDNA from transduced cells.
2. A targeted approach is required for linear pre-
amplification of the target gene sequences
• Three commercial whole genome amplification (WGA) kits and two pre-amplification master mixes tested.
• High variability in targets’ amplification observed in all WGA kits.
• Uniform amplification of targets observed with a targeted approach, comparable to non-amplified gDNA
quantification.
Representative graphs comparing pre-amplification strategies on gDNA as analysed by ddPCR
scVCN analysis on “low” pVCN sample scVCN analysis on “high” pVCN sample
• Single cell VCN analysis can estimate the number of vector units in each analysed single cell.
• Bayesian statistics is applied to convert measured scVCN values by ddPCR into integer values
corresponding to vector units per cell.
• Comparison of scVCN mean with the population VCN from gDNA analysis indicates that the assay can be
predictive of the cell population specifications.
• Comparison of scVCN predictions mean with pVCN is used to estimate the accuracy of the assay.
• The absolute difference between these values ranges from 0 to 0.6 copies, corresponding to an accuracy ≥ 80%.
• Linear regression test shows positive correlation between population and single cell VCN.
scVCN distribution across nine samples
compared to the population VCN and assay biasLinear regression comparing
scVCN mean with population VCN
1
N= 161 cells
0
234
18%
8%
22%
6%
17%
Viral copies
Mean of scVCN predictions= 1.93
Bulk gDNA VCN/cell= 2.31 ± 0.24
5679
29%
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
0
2
4
6
8
10
12
Single Cells
Me
an
sin
gle
ce
ll V
CN 95% confidence interval
• The scVCN assay can be applied to non-sorted cell sample used as a
proxy for a typical cell therapy product.
• The sample size can be increased by analysing multiple microfluidics chips
for the same population. No statistical difference between scVCN
distributions is observed here (p = 0.4297, two-tail Wilcoxon test).
• PCR-based transduction efficiency can be measured with this assay and is
comparable to conventional flow cytometry or imaging analysis.
• We designed a novel analytical method for advanced analysis of vector copy number and label-free transduction
efficiency at a single cell level to improve the characterisation of gene-modified cell therapy products.
• The scVCN assay can improve the control over the safety of gene-modified products by indicating the proportion of
single cell with a high number of vector integrations, which are generally associated with higher risk of insertional
mutagenesis, before therapy delivery and during follow-up studies.
• We envisage a broad applicability of the method to monitor product’s variability due to process, donor-to-donor, and
operational variations.
• Although it was developed on a lentiviral-based cell therapy model system, we anticipate this to be amenable to VCN
analysis of any transgene from viral or non-viral vectors, by simple modification of the primer sequences.
BACKGROUND
Ex-vivo gene-modified cell therapies are increasingly being developed for the treatment of monogenic
diseases and cancers. Product safety and consistency of manufacturing are key to the success of the cell
therapy field. The generation of genetically-modified cells often entails the use of lentiviral or retroviral vectors,
which randomly integrate into the host genome posing a risk for insertional mutagenesis, while product
variability is one of the main burdens to successful product generation.
CHALLENGE
Analytical characterisation is fundamental to monitor product variability
and safety, but current analytical methods are often inadequate or
laborious and only partially aid the evaluation of product’s attributes. In
particular, population analysis is the standard approach, yet this does
not provides insights on the underlying cell-to-cell variability.
Vector copies
64%
23%
5%
Mean of scVCN predictions= 1.36Bulk gDNA VCN/cell= 1.46 ± 0.32
N= 78 cells
3%
01234
5%
1.0 1.5 2.0 2.5 3.0
1.0
1.5
2.0
2.5
3.0
pVCN (bulk gDNA)M
ean s
cV
CN
pre
dic
tions
R2 = 0.7156
1 2 3 4 5 6 7
Vector copies
Analy
sed S
am
ple
s
0%20%40%60%80%
VCN
0
0.2
0.4
0.6
Bias
0%
5%
10%
15%
20%pVCN
1.0
1.5
2.0
2.5
3.0
scVCN analysis on a non-FACS sorted sample
Brightfield ZsGreen