Embed Size (px)
Citation preview
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.
DEVELOPMENT OF A SCALABLE PLATFORM FOR AAV MANUFACTURINGJuline Guenat*, Adrien Soula*, Florian Leseigneur, Quentin Bazot, Nishanti Weeratunge, Nishant Pawa, Lily Li, Tony Bou Kheir, Gregory Berger, Tristan Thwaites, Mike Delahaye, Julie Kerby, Hadi Mirmalek-Sani and Jonathan Appleby *equally contributedPresenting author contact: [email protected], [email protected]
Conclusion: Here is presented a scalable AAV manufacturing platform using STRs bioreactors for
production. Production process was defined and optimised at small scale and is now being scaled up at
production scale. Purification and analytics are already enabling efficient end to end process but will be
further optimised.
Introduction: AAV vectors are an appealing tool for both ex-vivo and in-vivo gene therapy. The number of AAV gene products entering early and late phase clinical trials is significantly on the increase. High cost of
goods, low process yield, and poor product characterisation are all metrics that require substantial development and improvement within this emerging field. Currently most AAV vectors are manufactured using an adherent process
and the demand currently outstrips capacity. Whilst switching from a classical 2D approach to a suspension process using single use bioreactors might be appealing to meet the requirement in term of doses and patient number for
late stage of development (e.g Phase III study), this can come at the expense of laborious and costly comparability study. Hence, a reliable, low risk manufacturing platform delivering at the desired scale should be identified early on
during development. Therefore, it becomes clear that there is a need to develop the next generation of upstream platform processes using a suspension cell line in STRs. Following Quality by design principles, we sought to develop
this platform for AAV manufacturing. Using a scale down model, we investigated the impact of a broad range of process parameters using a design of experiment approach on AAV productivity. Scalability of the newly designed
process as well as the impact of our USP on full capsid enrichment during our purification process has been investigated. Overall, our latest efforts in developing an end to end scalable suspension platform for AAV manufacturing
will be presented.
Manufacturing challenges
USP OptimisationStatistical design of experiments (DoE) was applied to determine the relationshipbetween factors affecting the process and the output of that process. Transfectionefficiency and AAV titre were selected as outputs of the process.
Design-Expert® SoftwareFactor Coding: ActualOriginal ScaleTitre 1 (vg/mL)
Design Points1.04785E+011
3.30537E+009
X1 = A: PEI: DNA w/w ratioX2 = B: Total DNA plasmid per E6 cells
0.5 1 1.5 2 2.5 3
0.5
1
1.5
2
2.5
3
Titre 1 (vg/mL)
A: PEI: DNA w/w ratio (w/w)
B:
To
tal D
NA
pla
sm
id p
er
E6
ce
lls (
µg
)
5E+010
1E+011
1.5E+011
3
3 4
Design-Expert® SoftwareFactor Coding: ActualOriginal ScaleTitre 2 (vg/cell)
Design Points16311.8
1459.85
X1 = A: PEI: DNA w/w ratioX2 = B: Total DNA plasmid per E6 cells
0.5 1 1.5 2 2.5 3
0.5
1
1.5
2
2.5
3
Titre 2 (vg/cell)
A: PEI: DNA w/w ratio (w/w)
B:
To
tal D
NA
pla
sm
id p
er
E6
ce
lls (
µg
)
5000
5000
10000
15000
3
3 4
Design-Expert® Software
Factor Coding: Actual
Original Scale
Transfection efficiency (%)
Design Points
44.5
0.46
X1 = A: PEI: DNA w/w ratio
X2 = B: Total DNA plasmid per E6 cells
0.5 1 1.5 2 2.5 3
0.5
1
1.5
2
2.5
3
Transfection efficiency (%)
A: PEI: DNA w/w ratio (w/w)
B: T
ota
l D
NA
pla
sm
id p
er
E6
ce
lls (
µg
)
20
40
60
80
100
3
DoE #23D surface response of modelledoutputs. Y axis represent input(1) and X axis input (2). Redcolour indicates higher responseand Blue lower. The stars showpredicted best combination.
Design-Expert® Software
Factor Coding: Actual
transfection efficiency (%)
Design Points
90.4
67.5
X1 = A: repcap
X2 = B: gfp
2 3 4 5 6
2
3
4
5
6
Transfection efficiency (%)
A: repcap (ratio)
B: g
fp (
ratio
)
75
80
85
2 2
2 2
5
Design-Expert® Software
Factor Coding: Actual
titre (vg/mL)
Design Points
3.66E+011
1.19E+011
X1 = A: repcap
X2 = B: gfp
2 3 4 5 6
2
3
4
5
6
Titre 1 (vg/mL)
A: repcap (ratio)
B:
gfp
(ra
tio)
2.7E+0112.7E+011
2.8E+011 2.8E+011
2.9E+011
3E+011
3.1E+011
2 2
2 2
5
Design-Expert® Software
Factor Coding: Actual
titre2 (vg/cell)
Design Points
64550.3
21364.5
X1 = A: repcap
X2 = B: gfp
2 3 4 5 6
2
3
4
5
6
Titre 2 (vg/cell)
A: repcap (ratio)
B:
gfp
(ra
tio)
44000
4400046000
46000
48000
50000
2 2
2 2
5
DoE #53D surface response of modelledoutputs. Y axis represent input(1) and X axis input (2).
rpm #1 rpm #2 rpm #1 rpm #2
0
20
40
60
80
100
Transfection efficiency
GF
P p
osit
ive
cells (
%)
Baseline process Optimised process
rpm #1 rpm #2 rpm #1 rpm #2
b*0
b*1
b*2
b*3
b*4
b*5
0
25
50
75
100
Viable Cell Density at harvest
Via
ble
Cell D
en
sit
y
(vc
/ m
L)
Baseline process Optimised process
Via
bility
(%)
Viability pH #1
Viability pH #2
rpm #1 rpm #2 rpm #1 rpm #2
0*10^c
5*10^c
10*10^c
15*10^c
20*10^c
25*10^c
30*10^c
35*10^c
Titer at harvest
Tit
er
(vg
/mL
)
Baseline process Optimised process
Optimised process led to a eight times titer increase while decreasing the cost.rpm #1 rpm #2 rpm #1 rpm #2
0*10^b
5*10^b
10*10^b
15*10^b
20*10^b
25*10^b
Titer at harvest
Tit
er
(vg
/mL
)
Baseline process Optimised process
pH #1
pH #2
Despite all efforts…
’Viral vector manufacturing
capacity is a barrier that is
limiting the development of
therapies and places the
industry at risk.
We’ve seen an improvement in
capacity but demand has
increased’
Our Approach for process optimisation
US
PD
SP
Media 1 Media 2 Media 3 Media 1 Media 2 Media 30*10^c
5*10^c
10*10^c
15*10^c
20*10^c
25*10^c
30*10^c
35*10^c
Titer at harvest
Tit
er
(vg
/mL
)
HEK293-A HEK293-B
Media 1 Media 2 Media 3 Media 1 Media 2 Media 30
20
40
60
80
100
Transfection efficiency
%G
FP
po
sit
ive c
ells
HEK293-A HEK293-B
a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4b*0
b*1
b*2
b*3
b*4
b*5
b*6
Growth
Via
ble
Cell D
en
sit
y
(vc
/ m
L)
hrs post seeding
Media 1 Media 1Media 2 Media 2Media 3 Media 3
HEK293-A HEK293-B
Two cell lines were cultivated in three medias using historical culture parameters,on the Ambr® 15 system, transfection in triplicates.
Cell line and media selection
a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*4 a*1 a*2 a*3 a*40
20
40
60
80
100
Aggregation
Ag
gre
gate
s (
%)
hrs post seeding
Media 1 Media 1Media 2 Media 2Media 3 Media 3
HEK293-A HEK293-B
Screen
Optimisation
Scale up
Proving Run
Optimisation
Scale up
Scale-up from Ambr® 15to Ambr® 250• Maintenance of
geometric similarity• Equal specific energy
dissipation rates(volumetric powerinput P/V)
USP Scale-up
a*1 a*2 a*3 a*4b*0
b*2
b*4
b*6
b*8
Growth
Via
ble
Cell D
en
sit
y
(vc
/ m
L)
hrs post seeding Ambr 15 Ambr 250 VVM #1
Ambr 250 VVM #2
0*10^c
5*10^c
10*10^c
15*10^c
20*10^c
25*10^c
30*10^c
35*10^c
Titer at harvest
Tit
er
(vg
/mL
)
a*1 a*2 a*3 a*4b*0
b*2
b*4
b*6
b*8
Growth
Via
ble
Cell D
en
sit
y
(vc
/ m
L)
hrs post seeding
Ambr 15
Ambr 250VVM #1
Ambr 250VVM #2
Successful scale-up to 250mL using the optimised process developed at Ambr® 15 scale.Next steps: further optimisation at Ambr® 250 scale, and scale up to 2.0L UniVessels®.
DSP optimisation
Multi-angle dynamic light scattering (MADLS®) from Zetasizer®. X axis represent size (d.nm) and Y axis representintensity of the signal.
Two-steps purificationwith overall yield >75%With 60% full capsids.
Load Post-capture Post-polishing
Capture Polishing
Post capture (left) and postpolishing (right) materialpurified on AKTA®.
