BPI West 2017 Presentation, Leslie S. Wolfe, Ph.D. Principal Scientist, Downstream Process Development, KBI Biopharma & Abhinav A. Shukla, Ph.D. Senior Vice President, Process Development & Manufacturing, KBI Biopharma

1. Leslie S. Wolfe, Ph.D.
Principal Scientist, Downstream Process Development, KBI Biopharma
Abhinav A. Shukla, Ph.D.
Senior Vice President, Process Development & Manufacturing, KBI Biopharma

3. • Haynes group at Duke Human Vaccine Institute followed
patients infected with HIV‐1 from virus transmission
looking for individuals who developed broadly neutralizing
antibodies
• The mature broadly neutralizing antibody, CH103, was
identified and shown to neutralize ~55% of HIV‐1 isolates
• CH103 binds to HIV‐1 envelope protein gp120
• A vaccine strategy was devised with the intention of
inducing the evolution of CH103 in uninfected individuals

4. Haynes, B, Harrison, S, Kelsoe, G and Kepler T, Nature Biotech. 30: 423, 2012

5. • KBI joined the project team in 2013
• Scope of work included development of a platform for analytics, cell
line development, upstream and downstream process development
and manufacturing for 4 gp120 proteins selected along the bnAb
evolution pathway
H.X. Liao et al. Nature 496: 469; 2013
CH505TF CH505w53 CH505w78 CH505w100
CH103UCA
CH103 lineage intermediate antibodies
gp120

6. 6

7. • Platform Cell Line and Process Development:
• A generic approach to design and generate a cell line, express and
purify the target protein with a well‐defined sequence of unit
operations where most operating parameters are pre‐defined and
a subset of parameters require development efforts
• Platform approaches have been implemented for
monoclonal antibodies and Fc fusion proteins; however,
biochemical differences in molecules of the same class
warrant the need for some development efforts
7
Shukla, A., Hubbard, B., Tressel, T., Guhan, S., Low, D. Downstream processing of monoclonal antibodies
– Application of platform approaches, Journal of Chromatography B, 848, 28-39, 2007.

8. • Parameters shaded in gray are
defined across molecules.
Parameters shaded in yellow
require molecule specific
optimization
• The MTX amplification
concentration evolved across the
development of the 5 gp120 envs
worked on to date
• A 2nd round of limited dilution
cloning was employed for 2 of the 5
gp120 envs worked on to date
8

9. • Parameters shaded in gray are
defined across molecules.
Parameters shaded in yellow
require molecule specific
optimization
• For all Spero molecules the
operating parameters, basal
medium, feed type and some of
the supplement additions have
defined
• The need for additional
supplements is molecule specific
• Reasons for supplement
addition:
» Biocompatibility
» Increase in productivity
Scale
Temperature Set point 37.0 ± 0.5°C
Temperature Shift 33.0 ± 0.5°C on Day 6
DO Set point 30%
pH Set point 6.90 ± 0.1
Agitation (1-impeller) 50 rpm → 55 rpm
Air overlay 1.6 SLPM
Air Sparge 0.5 SLPM
Max. Oxygen Sparge 5 SLPM
Max. CO2 Sparge 5 SLPM
Medium CD OptiCHO + 8 mM Glutamine
Target VCD 0.50 x 106
cells/mL
Base 1M Sodium carbonate
Feed Type: LTI Feed A+B (1:1)
15% on Day 0, 10% current wv each
on Days 3, 6, and 9
Supplement 1 addition: HT Supplement 1X current wv each on Days 0 and 4
Supplement 2 addition: Cystine
Supplement 3 addition: Tyrosine
Supplement 4 addition: Soy:Yeastolate
Hydrolysate (2:3)
5g/L current wv each on Days 4 and 8
Supplement 5 addition: C1615
Harvest Add 10g/L Hydrolysate on harvest
9

10. 10
• Parameters shaded in gray are
defined across molecules.
Parameters shaded in yellow
require molecule specific
optimization
• Load and elution conditions for
three of the unit operations
require molecule specific
definition given the heterogeneity
of this class of molecules
• Env antigens structurally sensitive
to hydrophobic surfaces, hence
HIC not employed

11. • Vaccine candidate is monomeric gp120
• Heavily glycosylated with glycans comprising 50%
by mass
• High mannose glycans are the predominant
glycoform
• Charge profile of gp120 targets have a
heterogeneous charge profile
• Charge profile spans a broad pH range and is
molecule specific
• Molecule prone to proteolytic cleavage
1 2 3 4
Lane gp120
1 CH505TF
2 CH505w53
3 CH505w078
4 CH505w100
IEF Gel

12. • HIV env field routinely relied on lectin capture to isolate
gp120
• Some differences in glycan heterogeneity observed between
lectin and non‐lectin purified samples
• Not suitable approach for GMP manufacturing
• Implementation of a non‐affinity based capture step
• Focus on mixed mode capture for ability to fine tune selectivity
• Polishing chromatography steps rely heavily on charge
interactions
• Development supported by DHVI SPR assay and KBI octet
binding assay to ensure product activity was maintained

13. 13
rHCP levels
Capto MMC
Step Yield
SUB
Biocompatibility
CH505w100
SPR activity
concerns
RP-purity
%pre-main
peak levels
CH505w78CH505w53
SPR activity
concerns
Molecule
stability
SPR activity
concerns
CH505TF
High %HMW
by SEC-
HPLC

14. baseline
Incorporation of a modulator wash can improve HCP clearance for a mixed mode step
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
0.0% 20.0% 40.0% 60.0% 80.0% 100.0%
Normalized HCP
Recovery
HCP vs. Recovery after Intermediate Wash for Capto MMC Polishing
baseline
Wolfe, L., Journal of Chromatography A, 1340, 151-156, 2014.

15. • Material generated from platform assessment had weaker
affinity for CH103 intermediate antibodies
• Based on characterization experiments (SPR, ITC) it was
hypothesized that only a subpopulation of the BDS had the
desired activity
• Troubleshooting experiments confirmed that capture with Capto
MMC alters product quality of CH505w53
• An alternative CEX capture step was developed in
conjunction with high throughput binding assay to select
only the “active” fraction of product
• Ability for integrated real time analysis significantly reduced
timeline for improving CH505w53 capture step

16. 0 1 2 3 4 5 6 7 8 9
0-600mM
NaCl gradient
• Later eluting
fractions have better
binding for CH106
• An intermediate
NaCl wash was
developed to enrich
“active” species in
eluate fraction

17. • Consistent performance achieved with platform from
development to manufacturing
SEC‐HPLC % Main Peak
Sample ID
TF
Demo
TF
ENG
TF
GMP
w100
Demo
w100
ENG
w100
GMP
w78
Demo
w78
GMP
w53
Demo
w53
ENG
BDS 99.3% 98.9% 98.8% 98.9% 99.3% 99.2% 99.5% 99.6% 97.4% 99.0%

18. • Despite some differences in each
unit operation performance,
platform process had consistent
viral clearance
• 19.1‐23.9 logs XMuLV
• 9.6‐11.9 logs MMV
• Low pH and detergent viral
inactivation performed similarly
• Polish 1 is a robust unit operation
for XMuLV clearance
0
5
10
15
20
25
30
P1 P2 P3 P4 P5
Log Clearance
Glycoprotein Platform Process XMuLV
Clearance
VF
Polish 3
Polish 1
VI
Capture
0
2
4
6
8
10
12
14
P1 P2 P3 P4 P5
Log Clearance
Glycoprotein Platform Process MMV
Clearance
VF
Polish 3
Polish 1
Capture

19. CH505TF
(stable expression)
CH505w100
CH505w78
CH505w53
B63521
Clone Selection
Upstream/Downstream
Process Assessment &
confirmation of SPR activity
Verification of
Compatibility with
Single Use Bioreactor
Demonstration
Run
Engineering
Run
cGMP Run
CH505TF
(transient expression)

20. • Molecule specific challenges have led to evolution of the downstream
platform process
• Removal of intermediate urea washes
• Conversion of low pH VI to detergent inactivation
• Evaluation of alternative capture step where needed
• Despite minor optimization required, polishing unit operations, UF/DF
and viral filtration are largely unchanged from molecule to molecule
• Development and implementation of a platform downstream process
has reduced development time and streamlined tech transfer to
manufacturing
• Approach supports rapid development and manufacturing to enable
clinical entry for predicted iterative experimental Phase I trials

21. 21
Duke Human Vaccine Institute:
Dr. Barton Haynes,
Prof. Tom Denny and team
Tom VanCott, PhD and team
Dr. Mike Pensiero and team
Cell line development, Upstream &
Downstream PD, Analytical Development,
Formulation Development, cGMP
Manufacturing, QA/QC