Engineering antibodies beyond affinity: correlation of key biophysical characteristics with antibody specificity and pharmacokinetic properties

Tris Vaughan

Senior Director,

Antibody Discovery and Protein Engineering

WCBP 2015

27th-29th January 2015

2

Overview

1 Why antibody developability is important

2 Engineering beyond affinity: anti-NGF mAb case study

3

Overview

1 Why antibody developability is important

2 Engineering beyond affinity: anti-NGF mAb case study

4

Kd for targetantigen

Target specificity, species

cross-reactivity

Target clones

In vitro / in vivo potency

Lead Selection with Functional Focus

5

ChemicalStability

SolutionProperties

Physical Stability

Target clones

Lead Selection with Developability Focus

6

ChemicalStability

SolutionProperties

Physical Stability

• Conformational stability• Colloidal stability• pI • Aggregation

• Reversible self-association

• Solubility• Viscosity

• Fragmentation • Deamidation • Oxidation• Asp isomerization• Sequence variants• O-glycosylation• Glycation

Target clones

Lead Selection with Developability Focus

7

ChemicalStability

SolutionProperties

Physical Stability

Target clones

Lead Selection with Developability Focus

In Silico aggregationprediction tool

StarGazer 384-wellAggregation analysis

(Tagg)

Nanopro HT cIEF

45oC, T=2wks

T=0

T=0

Size exclusion Chromatography (SEC)

8

Overview

1 Why antibody developability is important

2 Engineering beyond affinity: anti-NGF mAb case study

9

Anti-NGF Antibody – Medi578

� Target

– Nerve growth factor

� Mechanism of action

– Selectively antagonises binding of NGF to its receptors TrkA and p75

� Characteristics

– Human, derived from phage display library– Selective over other neurotrophins– Cross-reactivity with cyno and rat NGF– KD = 69pM– Active in NGF driven in vitro assays

� Modelling suggested that a higher affinity/potency antibody would suppress serum NGF in vivo >90%

Goal to improve the affinity and potency of Antibody 10-foldGoal to improve the affinity and potency of Antibody 10-fold

Anti-NGF suppression of serum NGF (KD = 6.9pM)

X-ray co-crystal solved

MEDI578 MEDI578

NGF

Affinity Maturation of MEDI578 → MEDI1912

10

MEDI-578 Epitope competition assay

0 1 2 3 40

50

100

150

MEDI-1912MEDI-578

log antibody conc pM

surv

ival

% o

f max

Anti-NGF antibody MEDI1912

� Most improved variant, MEDI1912, exhibited ~10 fold improvement

� MEDI1912 contains 12 mutations (compared to MEDI578)

11

10-fold improved cellular potency

VHCDR1 VHCDR2 VHCDR3 VLCDR1 VLCDR2 VLCDR3

Antibody 1 GGTFSTYGIS GIIP-----IFDTGNSAQSFQG SSRIYDYAGGDHYYYDM------DV SGSSSN------IGNNYVS DN-----NKRPS GTWDS S LSA-------WV

MEDI578 GGTFSTYGIS GIIP-----IFDTGNSAQSFQG SSRIYDLNPSLTAYYDM------DV SGSSSN------IGNNYVS DN-----NKRPS GTWDS S LSA-------WV

MEDI1912 GGTFWFGAFT GIIP-----IFGLTNLAQNFQG SSRIYDLNPSLTAYYDM------DV SGSSSD------IGNNYVS DN-----NKRPS GTWDS S LSA-------WV

Kon (1/Ms) Koff (1/s) KD (pM)

2.0 x 107 (20 – 3 x 10-5) (1.6 – 9.8)

0 1 2 3 40

50

100

150MEDI-578MEDI-1912

log antibody conc pM

surv

ival

% o

f m

axDorsal root ganglion cell survival PC12 cell survival

MEDI1912 is specific for NGF

12

NGF epitope competition assay

MEDI1912 reduces mechanical hypersensitivity in a mouse intra-plantar FCA model

13

MFCA032: Effect of MEDI1912 on Mechanical HypersensitivityIpsilateral/Contralateral Ratio

NaivePre-dose 4h 1 2 3 70

20

40

60

80

100

120

PBS i.v. Isotype control i.v.

***

MEDI1912 3 mg/kg i.v.

MEDI1912 0.3 mg/kg i.v.

MEDI1912 1 mg/kg i.v.N=12 per group. Data analysed using 2 w ay ANOVA w ith time and treatment as dependant factors.

Subsequent statistical signif icnace obtained using Bonferroni's Post Hoc test

Time Post Dose

Ipsi

/Co

ntra

Rat

io

MEDI1912 exhibited significant problematic biophysical and DMPK characteristics

14

HPLC-SEC Analytical Ultracentrifugation

Visible OpalescenceReversible self-association

Monomer-dimer-trimer equilibrium Adsorption to surfaces

e.g. SEC column matrix, filter membranesLow purification yieldPoor solubilityPhase-separation & opalescence issuesHigh viscosity (when >30mg/mL)Poor PK in both rat (t½ = 4 days) & NHP

MEDI1912 exhibited significant problematic biophysical and DMPK characteristics

15

Visible OpalescenceReversible self-association

Monomer-dimer-trimer equilibrium Adsorption to surfaces

e.g. SEC column matrix, filter membranesLow purification yieldPoor solubilityPhase-separation & opalescence issuesHigh viscosity (when >30mg/mL)Poor PK in both rat (t½ = 4 days) & NHP

Rat PK NHP PK

16

In silico prediction of protein aggregation propensity

� Spatial aggregation propensity (SAP) algorithm developed by Professor Bernhardt Trout, MIT

– Predicts aggregation propensity based upon atomistic analysis

� SAP algorithm incorporated into Accelrys Discovery Studio

Chennamsetty et al., 2009

Engineering MEDI1912 biophysical properties via a structure-based methodology

� Spatial aggregation propensity (SAP) software used to predict aggregation prone regions on IgG surface

– Three potential positions of interest identified: W30S, F31T (both in VH CDR1) and L56T (VH CDR2)

17

MEDI578 MEDI1912

Engineering MEDI1912 biophysical properties via a structure-based methodology

� Spatial aggregation propensity (SAP) software used to predict aggregation prone regions on IgG surface

– Three potential positions of interest identified: W30S, F31T (both in VH CDR1) and L56T (VH CDR2)

18

MEDI578 MEDI1912

W30

F31

L56

VH CDR2

VH CDR1

Creation & screening of variant IgGs by HPLC-SEC

19

ControlsControls

MEDI1912

MEDI578

Reference WFT

WTL

WTT

SFL

SFT

STL

STT

Panel of variants

Mutant STT ameliorates aberrant biophysical properties

20

Analytical Ultracentrifugation

0.000 0.002 0.004 0.0061e-007

2e-007

3e-007

4e-007

5e-007

Conc (g/ml)

D (

cm2 /s

)

Dynamic Light Scattering

min2 4 6 8 10 12 14 16 18

mAU

0

50

100

150

200

250

VWD1 A, Wavelength=280 nm (APR2014\CL MICA AND TS 22NDAPR 2014 2014-04-22 10-33-53\MEDI1578.D)

8.6

89

VWD1 A, Wavelength=280 nm (APR2014\CL MICA AND TS 22NDAPR 2014 2014-04-22 10-33-53\MEDI1912.D)

Area : 5046 .48

15

.047

MEDI1912_STT

HPLC-SEC

AUC

AUC

Mutant STT ameliorates aggregation of MEDI1912

21

MEDI1912 MEDI1912 - STT

Mutant STT retains affinity and potency of MEDI1912

� STT demonstrates identical affinity for NGF and equipotency in relevant cellular assays

– The three mutations were not associated with decreased affinity

Affinity (BIAcore) for NGF

MEDI1912 1.6 - 9.8 pM

STT 1.8 – 8.3 pM

pERK Potency Assay

-12 -11 -10 -9 -8

0

50

100

NIP228 IgG1TM YTE

MED1912 IgG1TM YTE

MEDI1912_STT IgG1TM

IgG Log [M]

% N

GF

-in

du

ce

d p

ER

K

Mutant STT has reinstated favourable DMPK properties

MEDI1912 STTIsotype

LUNG

LIVER

KIDNEY

STT shows no discernible non-specific binding by IHC

Summary

� With higher affinity mAbs can come other challenges

– MEDI1912 exhibited aberrant CMC/manufacturability and non-linear, impaired PK profile that endangered product development

� In silico prediction can be used to enhance developability characteristics

� Mutant STT retains all of the affinity / potency gains of MEDI1912, but with optimal biophysical characteristics and PK profile

� Clear link between biophysical – DMPK – tissue specificity

� Generic engineering approach could be used to improve the biophysical properties of any therapeutic protein

24

Acknowledgements

� MEDI578 and MEDI1912 project teams

� Claire Dobson

� Andrew Buchanan

� Bojana Popovic

� Chris Lloyd

� Daniel Higazi

� Arthur Lewis

� David Lowe

� Tristan Vaughan

� Biopharmaceutical Development– Chris van der Walle

– David Hayes

– Catherine Galy

– Leanne Amery

– Sofia Ekizoglou

– Richard Turner

� BSU650

� DMPK Team– Jo Goodman

� AstraZeneca Discovery Sciences:– Lise-Lotte Olsson

– Anna aagaard

– Linda Cederblad

– Niek Dekker

– Paul Wan

– Tomas Akerud

� Biologics Expression Team– Neil Brikett

– Anna Czyz

– Sonia Raithatha

– Melanie Medcalf

– Carolina Casado

– Nathan Hudson

– Richard Porter

– Nicola Forrest-Owen

– Robin Butler

25